Psychological therapies for depression in older adults residing in long‐term care settings

Abstract

Background

Depression is common amongst older people residing in long‐term care (LTC) facilities. Currently, most residents treated for depression are prescribed antidepressant medications, despite the potential availability of psychological therapies that are suitable for older people and a preference amongst many older people for non‐pharmacological treatment approaches.

Objectives

To assess the effect of psychological therapies for depression in older people living in LTC settings, in comparison with treatment as usual, waiting list control, and non‐specific attentional control; and to compare the effectiveness of different types of psychological therapies in this setting.

Search methods

We searched the Cochrane Common Mental Disorders Group Controlled Trials Register, CENTRAL, MEDLINE, Embase, five other databases, five grey literature sources, and two trial registers. We performed reference checking and citation searching, and contacted study authors to identify additional studies. The latest search was 31 October 2021.

Selection criteria

We included randomized controlled trials (RCTs) and cluster‐RCTs of any type of psychological therapy for the treatment of depression in adults aged 65 years and over residing in a LTC facility.

Data collection and analysis

Two review authors independently screened titles/abstracts and full‐text manuscripts for inclusion. Two review authors independently performed data extraction and risk of bias assessments using the Cochrane RoB 1 tool. We contacted study authors for additional information where required.

Primary outcomes were level of depressive symptomatology and treatment non‐acceptability; secondary outcomes included depression remission, quality of life or psychological well‐being, and level of anxious symptomatology. We used Review Manager 5 to conduct meta‐analyses, using pairwise random‐effects models. For continuous data, we calculated standardized mean differences and 95% confidence intervals (CIs), using endpoint data, and for dichotomous data, we used odds ratios and 95% CIs. We used GRADE to assess the certainty of the evidence.

Main results

We included 19 RCTs with 873 participants; 16 parallel group RCTs and three cluster‐RCTs. Most studies compared psychological therapy (typically including elements of cognitive behavioural therapy, behavioural therapy, reminiscence therapy, or a combination of these) to treatment as usual or to a condition controlling for the effects of attention.

We found very low‐certainty evidence that psychological therapies were more effective than non‐therapy control conditions in reducing symptoms of depression, with a large effect size at end‐of‐intervention (SMD −1.04, 95% CI −1.49 to −0.58; 18 RCTs, 644 participants) and at short‐term (up to three months) follow‐up (SMD −1.03, 95% CI −1.49 to −0.56; 16 RCTs, 512 participants). In addition, very low‐certainty evidence from a single study with 82 participants indicated that psychological therapy was associated with a greater reduction in the number of participants presenting with major depressive disorder compared to treatment as usual control, at end‐of‐intervention and short‐term follow‐up. However, given the limited data on the effect of psychological therapies on remission of major depressive disorder, caution is advised in interpreting this result.

Participants receiving psychological therapy were more likely to drop out of the trial than participants receiving a non‐therapy control (odds ratio 3.44, 95% CI 1.19 to 9.93), which may indicate higher treatment non‐acceptability. However, analyses were restricted due to limited dropout case data and imprecise reporting, and the finding should be interpreted with caution.

There was very low‐certainty evidence that psychological therapy was more effective than non‐therapy control conditions in improving quality of life and psychological well‐being at short‐term follow‐up, with a medium effect size (SMD 0.51, 95% CI 0.19 to 0.82; 5 RCTs, 170 participants), but the effect size was small at postintervention (SMD 0.40, 95% CI −0.02 to 0.82; 6 RCTs, 195 participants). There was very low‐certainty evidence of no effect of psychological therapy on anxiety symptoms postintervention (SMD −0.68, 95% CI −2.50 to 1.14; 2 RCTs, 115 participants), although results lacked precision, and there was insufficient data to determine short‐term outcomes.

Authors' conclusions

This systematic review suggests that cognitive behavioural therapy, behavioural therapy, and reminiscence therapy may reduce depressive symptoms compared with usual care for LTC residents, but the evidence is very uncertain. Psychological therapies may also improve quality of life and psychological well‐being amongst depressed LTC residents in the short term, but may have no effect on symptoms of anxiety in depressed LTC residents, compared to control conditions. However, the evidence for these effects is very uncertain, limiting our confidence in the findings.

The evidence could be strengthened by better reporting and higher‐quality RCTs of psychological therapies in LTC, including trials with larger samples, reporting results separately for those with and without cognitive impairment and dementia, and longer‐term outcomes to determine when effects wane.

Plain language summary

Are psychological therapies effective in reducing depression in older adults living in long‐term care?

Key messages

– Psychological therapies (sometimes known as talking therapies) may treat symptoms of depression better than non‐therapy approaches in older adults living in long‐term care (LTC) facilities (for example, nursing or residential homes and assisted‐living facilities).

– Psychological therapies may also be better than non‐therapy approaches for increasing quality of life and psychological well‐being (an individual's emotional health and overall functioning) in the short term.

– Due to insufficient evidence, the broader benefits of psychological therapies, and longer‐term effects, are unclear.

What did we want to find out?

We wanted to find out whether psychological therapies were beneficial for managing depression in LTC settings.

Why is this important?

Depression is common amongst older people living in LTC. This population is often prescribed medication to manage depression. Psychological therapies may be a viable alternative.

What psychological therapies may be used to treat depression in older people?

Psychological therapies to treat depression in people living in LTC include behavioural therapy, cognitive behavioural therapy, and reminiscence therapy.

What did we do?

We searched for studies that compared psychological therapies for older adults with depression living in LTC facilities with an alternative care approach. We examined outcomes following the completion of therapy, and at short‐term (up to three months), medium‐term (three to six months), and long‐term (more than six months) follow‐up.

We compared and summarized the results of the studies and rated our confidence in the evidence, based on factors such as study methods and sizes.

What did we find?

We found 19 studies with 873 participants with depression. Most studies compared cognitive behavioural therapy, behavioural therapy, or reminiscence therapy to usual care services or to a condition providing residents with a similar level of attention (for example, friendly visits from volunteers or current events discussion groups).

We found that psychological therapies may be better at reducing symptoms of depression than other approaches immediately following therapy and for up to six months. This effect was not as apparent in studies where psychological therapies were compared with a condition where older people were given increased attention in LTC.

Psychological therapies may also improve older peoples' quality of life and psychological well‐being for up to three months after therapy, but we did not find evidence that psychological therapies reduce symptoms of anxiety in older people with depression.

What are the limitations of the evidence?

Our confidence in the evidence is very limited because most studies included small numbers of participants and used unreliable methods.

Given the ageing population and projected increase in older people requiring LTC, high‐quality clinical trials on the effectiveness of treatments for depression are urgently required.

How up to date is this evidence?

The literature search was completed in October 2021.

Summary of findings

Summary of findings 1. Any psychological therapy versus any non‐therapy comparator: end‐of‐intervention and short‐term follow‐up.

Psychological therapies compared with non‐therapy comparators
Patient or population: older people with depression Setting: long‐term care facilities Intervention: psychological therapy Comparison: non‐therapy comparator (treatment as usual, non‐specific attentional control, or wait list)
Outcomes	Anticipated absolute effects (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
Depressive symptomatology (lower = less depressive symptomatology): end‐of‐intervention follow‐up (2 to 24 weeks)	The mean score for the psychological therapy group was 1.04 standard deviations lower (95% CI 1.49 to 0.58 lower) than for non‐therapy comparators	644 (18 RCTs)	⊕⊝⊝⊝ Very lowa,b	Measured with GDS‐12, GDS‐15, GDS‐30, SDS, BDI, HAM‐D, CSDD, and CES‐D. The effect observed was large.c At short‐term follow‐up (up to 3 months), the mean score for the psychological therapy group was 1.03 standard deviations lower (95% CI 1.49 lower to 0.56 lower) than for non‐therapy comparators, based on a sample of 512 participants (16 RCTs); very low‐certainty evidencea,b, large effect sizec.
Treatment non‐acceptability (higher = greater non‐acceptability): end‐of‐intervention follow‐up (4 to 13 weeks)	The odds of treatment non‐acceptability for psychological therapies was 3.44 times higher (95% CI 1.19 to 9.93 higher) than for non‐therapy comparators	313 (5 RCTs)	⊕⊝⊝⊝ Very lowb,d	While 12 RCTs reported attrition data, only 5 RCTs contributed to the effect estimate as 7 RCTs reported 0 events in both groups. Caution is advised in interpreting this finding. There was a paucity of attrition data at short‐term follow‐up.
Depression remission (presence or absence of MDD):end‐of‐intervention follow‐up (10 weeks)	The net difference in change in prevalence was 54.2% (i.e. psychological therapy led to a 4.2% reduction in prevalence of MDD, whereas the non‐therapy comparator led to a 50.0% increased prevalence)	82 (1 RCT)	⊕⊝⊝⊝ Very lowb,d	Assessed using the SCID‐IV. Data were from 1 RCT only. The same results occurred at short‐term follow‐up (up to 3 months).
Quality of life or psychological well‐being (higher = higher quality of life or psychological well‐being): end‐of‐intervention follow‐up (4 to 24 weeks)	The mean score for the psychological therapy group was 0.40 standard deviations higher (95% CI 0.02 lower to 0.82 higher) than for non‐therapy comparators	195 (6 RCTs)	⊕⊝⊝⊝ Very lowb,d	Measured with LSI, QOL‐AD, QOL‐AD‐NH, WHOQOL‐BREF, and PGCMS. The effect observed was small.c At short‐term follow‐up (up to 3 months), the mean score for the psychological therapy group was 0.51 standard deviations higher (95% CI 0.19 to 0.82 higher) than for non‐therapy comparators, based on a sample of 170 participants (5 RCTs); very low certaintyb,d, medium effect sizec.
Anxious symptomatology (lower = less anxious symptomatology): end‐of‐intervention follow‐up (4 to 13 weeks)	The mean score for the psychological therapy group was 0.68 standard deviations lower (95% CI 2.50 lower to 1.14 higher) than for non‐therapy comparators.	115 (2 RCTs)	⊕⊝⊝⊝ Very lowa,b,d	Measured with HADS‐A and DASS. The effect observed was medium.c There was a paucity of data at short‐term follow‐up.
BDI: Beck Depression Inventory; CES‐D: Center for Epidemiologic Studies Depression Scale; CI: confidence interval; CSDD: Cornell Scale for Depression in Dementia; DASS: Depression, Anxiety and Stress Scales; GDS‐12, GDS‐15, GDS‐30: Geriatric Depression Scale – 12, 15, and 30 items; HADS‐A: Hospital Anxiety and Depression Scale – Anxiety; HAM‐D: Hamilton Depression Rating Scale; LSI: Life Satisfaction Index; MDD: major depressive disorder; PGCMS: Philadelphia Geriatric Centre Morale Scale; QOL‐AD: Quality of Life in Alzheimer's Disease; QOL‐AD‐NH: Quality of Life in Alzheimer's Disease – Nursing Home version; RCT: randomized controlled trial, including cluster‐randomized controlled trial; SCID‐IV: Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; SDS: Zung Self‐Rating Depression Scale.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of the effect.

^aThe estimate indicated substantial‐to‐considerable statistical heterogeneity. Downgraded one level for inconsistency.
^bAll RCTs had a high risk of bias for at least one category. Downgraded two levels for risk of bias.
^cBased on the following interpretations of SMD per Cohen: 0.2 = small effect, 0.5 = medium effect, 0.8 = large effect.
^dThe estimate was based on a collectively small sample size. Downgraded one level for imprecision.

Background

Description of the condition

There is strong awareness of the importance of addressing depression in long‐term care (LTC) settings, with studies consistently documenting high rates of depression amongst residents. One systematic review of 26 studies reported that the median prevalence of major depressive disorder (MDD) was 10%, while the median prevalence of subthreshold depressive symptoms was 29% (Seitz 2010). One more‐recent systematic review of 32 observational studies of LTC residents without dementia reported a pooled prevalence rate of MDD of 18.9% (Fornaro 2020). Analyses of Australian population‐level data found continued increases in rates of depression in LTC residents over time (Amare 2020). One meta‐analysis comparing rates of depression amongst older people living in LTC facilities versus those living in the community reported significantly higher rates in people in LTC facilities (Xiu‐Ying 2012).

Depression appears to increase the risk of admission to long‐term residential care and is often present at the point of entry to care (Achterberg 2006; Boyle 2004; Davison 2021a; Dorenlot 2005; Hajek 2015; Ulbricht 2017). In one longitudinal study of recently admitted nursing home residents (with and without dementia) in Australia, 24% met criteria for MDD and 20% for a non‐MDD, with only 12% of cases remitting over a six‐month period (McSweeney 2008). The persistence of depressive symptoms has been reported in several longitudinal studies in LTC settings (Barca 2010; Davison 2021a; Smalbrugge 2006; Sutcliffe 2007). Established risk factors for late‐life depression, including medical comorbidity, disability and functional decline, and cognitive impairment are ubiquitous in LTC settings (Davison 2012). However, there is a significant association between residing in LTC and increased depressive symptoms, even when controlling for these risk factors (Anstey 2007; Pot 2005).

Associations between late‐life depression and increased mortality and medical morbidity, functional decline and disability, greater dependence on personal care and health services, and poorer well‐being and quality of life highlight the importance of establishing effective treatments for depressive disorders in LTC settings (Abrams 1992; Beekman 2002; Beerens 2013). Subthreshold depressive symptoms that do not meet criteria for a formal diagnosis of MDD (also known variously as subsyndromal, subclinical, or minor depression) are also associated with serious consequences in later life, including increased disability, greater healthcare utilization, and increased suicidal ideation (Meeks 2011), and therefore are also important targets for treatment providers.

Description of the intervention

Management strategies for the treatment of late‐life depression include both pharmacological and non‐pharmacological interventions. In general, older adults are more likely to be treated with pharmacological than non‐pharmacological interventions, despite indications that older people may respond to psychological therapy as well as, or better than, working‐aged adults (Saunders 2021). However, older adults living in LTC settings often have even more limited access to psychological treatment modalities than the broader older adult population, with treatment providers in LTC largely relying on antidepressant medications (George 2007). To illustrate, in one review of over one million newly admitted LTC residents in the USA, less than 1% received any amount or type of psychological therapy (Ulbricht 2017).

The range of psychological therapies that could be potentially used with older adults in LTC are similar to those that have been well established for use with younger adults, including:

• behavioural therapies (e.g. relaxation techniques, activity scheduling, and behaviour modification);

• cognitive behavioural therapies (CBT; e.g. cognitive restructuring, and skills training, such as stress management and problem‐solving);

• third‐wave CBT (e.g. acceptance and commitment therapy (ACT) and mindfulness‐based cognitive therapy (MBCT));

• psychodynamic therapies (e.g. brief psychotherapy, psychoanalytic therapy, and insight‐oriented therapy);

• humanistic therapies (e.g. existential therapy and non‐directive therapy);

• integrative therapies (e.g. counselling and interpersonal therapy); and

• systemic therapies (e.g. family therapy, narrative therapy, and solution‐focused brief therapy).

In addition, some interventions have been designed specifically for older adults, notably reminiscence therapies (e.g. life review).

Psychological therapies may be performed by practitioners from a range of professional backgrounds, including psychologists, counsellors, social workers, psychiatrists, psychiatric nurses, and occupational therapists. There is potential for LTC care staff (i.e. nurses, diversional therapists) to be trained to deliver structured psychological interventions, such as reminiscence programmes and activity scheduling.

How the intervention might work

Psychological therapies may be used as an adjunct to pharmacological treatment or alternative treatment approach for MDD, other depressive disorders, or depressive symptoms in older adults. The proposed mechanisms of actions vary with the underlying theoretical basis of individual psychological therapies.

Behavioural therapy (BT) is based on the assumption that behaviour is learned, and employs methods that focus on changing maladaptive behaviour patterns. Activity scheduling is often used for combating passivity and withdrawal in depression and assists a depressed person to gradually re‐engage in some of the routines of their daily life, with a focus on increasing activities that are pleasurable and associated with mastery (Lewinsohn 1974). Various forms of relaxation techniques are available, such as progressive muscle relaxation, distraction, and guided imagery, which have been described as helpful for managing depressive symptoms by younger adults in community surveys (Jorm 2008).

CBT programmes combine both behavioural and cognitive methods, and are future‐directed, goal‐oriented, systematic approaches. Cognitive restructuring is based on the theory that distressing emotions and maladaptive behaviours are due to unrealistic and unhelpful negative (or 'dysfunctional') patterns of thinking (Beck 1979). The therapy aims to identify and modify an individual's thoughts, and assists the individual to replace dysfunctional thoughts with more adaptive cognitions. A range of skills training techniques fit within the CBT approach, such as problem‐solving and stress management.

Interventions described as 'third‐wave CBT' incorporate concepts such as acceptance, mindfulness, and personal values. Two of the more commonly used approaches are MBCT (Segal 2012), which employs mindful meditation as a therapeutic technique to interrupt patterns of rumination and negative thinking, and ACT (Hayes 2012), which facilitates acceptance of those components of life that are beyond one's control. ACT teaches individuals to focus on their remaining resources and commit to courses of action that are in accordance with their core values.

In contrast to CBT, psychodynamic therapy focuses on an individual's past experiences and aims to understand the impact of early events and unconscious processes on the current mood and behaviour of the individual (Busch 2016).

Humanistic therapies focus on self‐development and personal growth, and aim to facilitate self‐awareness and enable an individual to recognise his or her strengths and develop towards their full potential. For example, non‐directive therapy provides a therapeutic relationship, whereby an individual feels valued, listened to, and understood, and works by enabling the individual to reflect on their problems and seek solutions (Rogers 1951).

Integrative therapies, such as counselling, combine components of different psychological therapy models. One such approach, interpersonal psychotherapy (Klerman 1984), assists the individual in understanding how their interpersonal problems, such as disputes, role transitions, grief, and interpersonal deficits, contribute to their well‐being, and provides strategies to improve interpersonal skills.

Systemic therapies, such as couple and family therapy, focus beyond the level of the individual, and aim to improve the functioning of a larger unit, such as a marital unit or a family, typically by altering interactions between or amongst members of these units. Narrative therapy is based on understanding the 'stories' that people use to describe their lives (White 1990). The therapist helps the individual to consider how these stories may restrict them from overcoming their present difficulties. It sees problems as being separate from the individuals themselves and assists the client to recognise the range of skills, beliefs, and abilities they already have (but may not recognise) and which can be applied to current problems in their lives. Solution‐focused brief therapy helps individuals find tools to manage symptoms and cope with challenges (de Shazer 1988). It works by assisting the individual to clarify their goals and develop a series of steps to achieve them, based on enhancing their internal abilities and experimenting with new approaches.

Reminiscence therapies are typically used with only older adults (Webster 2010), and are recognised evidence‐based treatments for late‐life depression (Scogin 2005). Therapies aim to evoke pleasant or shared memories and problem‐solving successes, and to develop an integrated account of one's life. Such treatments are variously referred to as simple reminiscence, life review, and life review therapy.

Why it is important to do this review

Despite the significance of the problem, there remains a lack of evidence regarding appropriate treatments for depression amongst older people living in LTC facilities. A number of experts have challenged the effectiveness of commonly used antidepressant medications in this setting, especially for people with dementia, and questioned their long‐term use in people with multiple comorbidities (Banerjee 2011; Snowdon 2010). One 2012 review of studies evaluating antidepressants in residents of nursing homes included two randomized controlled trials, which did not demonstrate a significant benefit for antidepressant pharmacotherapy over placebo (Boyce 2012). While six of the seven non‐randomized studies identified a response to an antidepressant, the authors explained, "their results must be interpreted with caution as they lacked a comparison group". Limited information is available on the efficacy of antidepressants amongst older people with cognitive impairment and dementia (Boyce 2012; Nelson 2011; Orgeta 2017), which is problematic given the high prevalence of these conditions in the LTC population. Surveys have indicated that many LTC residents continue to have high symptom counts, despite taking antidepressants (Davison 2007; O'Connor 2010).

As is the case in broader adult samples (Hanson 2016), older people typically report a preference for psychological treatments for depression over antidepressant medications (Gum 2006; Luck‐Sikorski 2017). Despite this preference, there continues to be a lack of clear evidence of the effectiveness of psychological therapies for depression amongst older people living in LTC facilities. A systematic review of the available evidence is required to inform appropriate service provision for the LTC population.

Several reviews have reported promising results of psychological therapies for older people in the general community, particularly using CBT, problem‐solving therapy, and reminiscence therapies (Cuijpers 2014; Jonsson 2016; Wilson 2008). However, the LTC population is substantially older, and typically has higher rates of cognitive and functional impairment than the population of older people who are able to remain in their own homes. Some trials in community settings recruited participants as young as 50 to 55 years, with mean ages of trial samples substantially below those found in LTC (e.g. the mean age of LTC residents is 84.9 years in Australia; Aged Care Financing Authority 2021). Generalizing findings from community studies to the LTC population is therefore problematic and a review of research specifically targeting LTC residents is required.

A number of previous reviews, both narrative and systematic, have indicated that psychological therapies may have an impact on depressive symptomatology in LTC residents (Bharucha 2006; Cody 2013; Hyer 2005; Powers 2008; Simning 2017; Snowden 2003). However, several previous reviews had a broader aim than the current review, such as 1. evaluating interventions designed to improve well‐being in general, rather than focusing specifically on treatments for depression (Bharucha 2006), 2. evaluating a broad range of non‐pharmacological or psychosocial interventions (e.g. including bright light therapy or exercise‐related therapies), rather than restricting the review to psychological therapies (Gramaglia 2021; Hyer 2005; Snowden 2003); or 3. including both non‐pharmacological and pharmacological interventions (Simning 2017). Some reviews excluded participants with dementia (Gramaglia 2021; Simning 2017), despite the high prevalence of dementia in LTC populations (Seitz 2010). In addition, previous reviews commonly included studies of convenience samples of LTC residents, rather than only including studies that recruited participants with depressive disorders or clearly defined subthreshold depressive symptoms. One review focused only on nursing homes and excluded studies of lower‐level dependency facilities (assisted‐living facilities; Simning 2017). The authors of one recently published systematic review of non‐pharmacological interventions were unable to draw clear conclusions about the effectiveness of the interventions (Gramaglia 2021). However, Cody and Drysdale reported a medium effect size of psychotherapy on symptoms of depression in LTC from a meta‐analysis of studies (Cody 2013). Cody and Drysdale's suggestion that those interventions that were implemented with the involvement of LTC facility staff (i.e. using an integrated care model) were more effective than therapies delivered without facility staff involvement is of particular interest.

Our review collected and examined up‐to‐date evidence on the effect of psychological therapies for depression in LTC settings. Given the impact of subthreshold depressive symptoms, we included treatment studies for subthreshold depressive symptoms (based on scores above a defined cut‐off on validated depression scales), as well as studies for MDD. This review aimed to address the gap in our knowledge of the relative effectiveness of different psychotherapeutic interventions in the LTC population, as well as synthesize evidence regarding the effectiveness of psychotherapy on residents with varying levels of cognitive impairment.

Objectives

To assess the effect of psychological therapies for depression in older people living in LTC settings, in comparison with treatment as usual, waiting list control, and non‐specific attentional control; and to compare the effectiveness of different types of psychological therapies in this setting.

Methods

Criteria for considering studies for this review

Types of studies

We included randomized controlled trials (RCTs) in this review. We included studies with cluster allocation, but excluded those with only one or two clusters in each intervention group because randomization was considered unlikely to achieve balance in factors (e.g. characteristics of residents in each facility, organizational climate, or care‐related factors) that could affect outcomes in this situation. We included trials with a cross‐over design, but only used data from the first treatment phase, due to the potential for maintenance of treatment outcomes and heterogeneity regarding wash‐out period methodology. The protocol stated that quasi‐RCTs, where allocation to an intervention condition is not strictly random (e.g. by resident record number or alternation), would be included only if we did not locate any eligible RCTs. Given we located eligible RCTs, quasi‐RCTs were excluded from the review.

Types of participants

Participant characteristics

We included studies with participants aged 65 years and older regardless of their gender, ethnicity, or religion. We considered studies if the age range began under 65 years but the mean age was over 65 years. We also considered studies that included a subset of participants aged 65 years and over, but only if the participants were randomized on the level of the subset and data for participants aged 65 years and over were reported separately.

Diagnosis

Participants were required to present with 1. MDD according to any version of the Diagnostic and Statistical Manual of Mental Disorders (DSM) or International Statistical Classification of Diseases and Related Health Problems (ICD) criteria (APA 2013; WHO 2019), or 2. subthreshold depressive symptoms, based on a score over a cut‐off on a validated depression instrument, as defined by the study authors. Given there is no established cut‐off score on a gold standard instrument to determine the presence of subthreshold depression symptoms amongst older adults in LTC settings, and the anticipated heterogeneity in both the instruments selected in studies and the cut‐off points employed, we did not prespecify the instruments (provided there was published evidence of the measure's validity with older adults) or cut‐off points that we would accept for inclusion in the review.

Comorbidities

Participants could have normal cognitive functioning or cognitive impairment of any type of severity. We aimed to include participants with or without a diagnosis of dementia. We accepted any definitions used to define cognitive impairment or dementia as long as the criteria used were included in the publication. Studies involving participants with comorbid physical conditions or other psychological disorders were eligible for inclusion as long as treatment of depression was a primary aim of the study.

Setting

The study setting was LTC facilities, including nursing homes, assisted‐living facilities, and residential aged care facilities, where staff employed in the facility provided some level of day‐to‐day care to older adults by (i.e. to assist in activities of daily living) in addition to provision of accommodation. We included studies with samples from multiple settings only if outcomes relevant for the review were reported separately for the LTC residents.

Types of interventions

Experimental interventions

We included studies assessing any forms of psychological therapy for the treatment of depression, where this was compared with an alternative condition (see list of comparator interventions below). We included both group and individual psychological therapies, with no restrictions on frequency, intensity, or duration of the intervention. We included studies in which the psychological therapy was delivered in combination with another intervention (co‐intervention, including pharmacotherapy and exercise) only if there was a separate comparison group that received the co‐intervention alone. We included interventions facilitated by a range of professionals, including psychologists, social workers, occupational therapists, nurses, therapists in supervised training, and other trained professionals. We also included interventions implemented by LTC facility staff, where they were trained to implement the intervention. Psychological therapies were grouped into the following categories.

• Behavioural therapies (e.g. relaxation techniques, activity scheduling, and behaviour modification).

• CBT (e.g. cognitive restructuring, and skills training, such as stress management and problem‐solving).

• Third‐wave CBTs (e.g. ACT and MBCT).

• Psychodynamic therapies (e.g. brief psychotherapy, psychoanalytic therapy, and insight‐oriented therapy).

• Humanistic therapies (e.g. existential therapy and non‐directive therapy).

• Integrative therapies (e.g. counselling and interpersonal therapy).

• Systemic therapies (e.g. family therapy, narrative therapy, and solution‐focused brief therapy).

• Reminiscence therapies (e.g. simple reminiscence, life review, and life review therapy).

We excluded treatments identified as:

• pharmacotherapy, pet therapy, or exercise;

• music, art, and drama therapies;

• other psychosocial interventions that were not clearly psychological therapies;

• self‐help interventions delivered without therapist involvement;

• interventions involving group discussions or social visits from peers or volunteers (e.g. befriending) without therapist involvement (although they could be included as a comparator intervention);

• interventions delivered by trained volunteers rather than professional therapists or trained LTC facility staff.

We considered stepped care and case management interventions, where psychological therapies were included as one component of a multicomponent treatment protocol, as co‐interventions and excluded them if there was no comparison group with the same intervention without inclusion of the psychological therapy.

Comparators

• Treatment as usual or standard care (any medical care during the course of the study, including monitoring, check‐ups, no treatment, or any type of treatment, as well as assistance with activities of daily living and access to scheduled activities within the facility).

• On a waiting list (the control group did not receive any treatment until the intervention group had completed the treatment).

• Non‐specific attentional control (e.g. friendly visits from volunteers or current events discussion group), to control for the effects of social interaction.

• An alternative active psychological therapy.

• Co‐intervention (if also used in the intervention arm of the study), including pharmacotherapy or exercise.

Types of outcome measures

We included studies that met the above inclusion criteria, regardless of whether they reported on the following outcomes.

Primary outcomes

• Depressive symptomatology: measured using a variety of validated instruments, including those administered through self‐report, structured clinical interview, or informant report, with the informant being a LTC facility staff or a family member. Commonly used scales in LTC settings include the self‐rated Geriatric Depression Scale (GDS) (Yesavage 1983), the Beck Depression Inventory (BDI; Beck 1961), Zung Self‐Rating Depression Scale (SDS; Zung 1965), and Hamilton Depression Rating Scale (HAM‐D; Hamilton 1960). The score used reflected the protocol employed to administer the instrument in the study. To illustrate, in the guidelines for Cornell Scale for Depression in Dementia (Alexopoulos 1988), the total score represents the clinical opinion of the rater, based on interviews with the participant and an informant.

• Treatment non‐acceptability: rate of participants who dropped out of therapy following allocation to a study arm was considered a proxy measure for treatment acceptability. We did not include in this measure dropouts that were clearly attributable to reasons unrelated to the study or intervention, such as decline in physical health, death, or relocation during the course of the study that impacted on participants' ability to continue in the study.

Secondary outcomes

• Depression remission: the presence or absence of MDD, according to DSM or ICD criteria. We included structured or semi‐structured diagnostic clinical interviews, such as the Structured Clinical Interview for DSM‐5 (SCID‐5) (First 2016); or the Mini International Neuropsychiatric Interview (MINI) (Sheehan 1998), administered either by a mental health clinician (e.g. psychiatrist, clinical psychologist) or a trained researcher. If the study did not employ a structured interview schedule, we included it if a mental health clinician conducted the diagnostic interview and assessed and reported appropriate psychometric data (e.g. the measure had acceptable inter‐rater reliability).

• Quality of life or psychological well‐being: measured using standardized scales, such as the Short Form 36 (SF‐36) (Ware 1992), the World Health Organization Quality of Life (WHOQOL) assessment (WHOQOL Group 1998), and the Satisfaction with Life Scale (SWLS) (Diener 1985). Instruments could be self‐rated or informant‐rated (or both), with the informant being a LTC facility staff or family member.

• Anxious symptomatology: measured using standardized measurement instruments, such as the Geriatric Anxiety Inventory (GAI) (Pachana 2007), the Rating Anxiety in Dementia (RAID) Scale (Shankar 1999), and the Hamilton Anxiety Scale (HAM‐A) (Hamilton 1959).

• Physical functioning: measured using scales of basic activities of daily living, such as the Katz Index of Independence in Activities of Daily Living (Katz ADL) (Katz 1963), and instrumental activities of daily living, such as the Lawton Instrumental Activities of Daily Living (IADL) Scale (Lawton 1969).

• Agitation: measured using standardized scales, such as the Agitation subscale of the Neuropsychiatric Inventory (NPI) (Cumming 1994), and the Cohen‐Mansfield Agitation Inventory (CMAI) (Cohen‐Mansfield 1989).

• Adverse effects: as defined by individual studies.

Timing of outcome assessment

We summarized end‐of‐intervention outcomes and outcomes at each reported follow‐up point. Where the data allowed, we categorized follow‐up outcomes as short‐term (up to three months), medium‐term (three to six months), or long‐term (more than six months). If a study reported follow‐up outcomes at more than one time point, within one of these time frames, we selected the outcome reported at the latest point within the time frame. We reported outcomes at the first assessment post‐treatment (end‐of‐intervention) and at short‐term follow‐up in the summary of findings table.

Hierarchy of outcome measures

We separately analysed depression symptoms (assessed using continuous measures) and MDD diagnosis (assessed using a dichotomous measure or rate recovered), with depressive symptoms considered the primary outcome. If a trial used more than one instrument to assess depression symptoms, then we prioritized these in the following order: clinician‐rated scale, informant‐rated scale, and self‐rated scale (given concerns regarding the reliability of self‐reported symptoms of depression in the LTC population; Davison 2009). If studies used multiple outcome measures to measure the same outcome (e.g. two self‐report measures), we chose the outcome measure that was most frequently used across studies (namely, availability across studies). If multiple outcome measures of the same type had equivalent availability across studies, we chose the one with the strongest psychometric evidence in previous research in LTC settings.

Search methods for identification of studies

Electronic searches

We searched the following databases in October 2021 (with a summary of results available in Appendix 1).

• Cochrane Common Mental Disorders Specialized Register (CCMDCTR‐Studies and CCMDCTR‐References) (archived) (all years to June 2016). The Cochrane Common Mental Disorders Group (CCMD) maintained two archived clinical trials registers at its editorial base in York, UK: a references register and a studies‐based register. The CCMDCTR is current to June 2016 only

• Cochrane Central Register of Controlled Trials (CENTRAL; Issue 10, 2021) in the Cochrane Library (searched 8 October 2021)

• MEDLINE Ovid (1946 to 8 October 2021)

• PsycINFO Ovid (1806 to 8 October 2021)

• Embase Ovid (1974 to 8 October 2021)

• Cochrane Library: Trials (via CRSWeb) (8 October 2021; crsweb.cochrane.org)

• CINAHL (Cumulative Index to Nursing & Allied Health) (EBSCOhost) (1980 to 31 October 2021)

• PubMed NOT MEDLINE (8 October 2021)

• Social Services Abstracts (ProQuest) (1980 to 8 October 2021)

• Sociological Abstracts (ProQuest) (1974 to 8 October 2021)

• AgeLine (EBSCO) (1978 to 31 October 2021)

There were no further restrictions on date, language, or publication status.

We also searched international trial registries, including the World Health Organization International Clinical Trials Registry Platform (ICTRP; apps.who.int/trialsearch; searched 8 October 2021), and US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; searched 8 October 2021) to identify additional unpublished or ongoing studies.

The CCMD Information Specialist designed and performed searches in the primary prespecified databases (with the exception of AgeLine) on 8 October 2021. One review author (YW) performed the AgeLine search (the search strategy of which is used for CINAHL) on 31 October 2021.

Searching other resources

Grey literature

We searched the following sources of grey literature (all available years to 18 October 2021).

• ProQuest Dissertations and Theses database

• Open Access Theses and Dissertations (oatd.org)

• DART‐Europe E‐theses Portal (www.dart-europe.eu)

• EThOS – the British Libraries e‐theses online service (ethos.bl.uk)

• Open Grey (www.opengrey.eu)

Annette Steere, Librarian at Swinburne University of Technology (Melbourne, Australia), designed and carried out these searches.

Reference lists

We checked the reference lists of all included studies and relevant systematic reviews to identify additional studies missed from the original electronic searches (e.g. unpublished or in‐press citations). We also conducted a cited reference search on the Web of Science Citation Index (SCI) for reports of included studies. One review author (YW) conducted the search of reference lists.

Correspondence

We contacted trialists and subject experts for information on unpublished or ongoing studies or to request additional trial data.

Data collection and analysis

Selection of studies

Several pairs of authors (YW, TD, EY, SBh, CD) independently screened titles, abstracts, or both, for inclusion of all potential studies identified as a result of the search, and coded them as 'include' (eligible or potentially eligible/unclear) or 'do not include' (ineligible). After retrieving the full‐text study reports/publications, pairs of review authors (YW, TD, EY, SBh, CD) independently screened the full text to identify studies for inclusion and recorded reasons for exclusion of ineligible studies. The two review authors resolved any disagreements through discussion or, if necessary, consultations with a third review author or the broader team. We contacted external experts or trial authors if necessary to reach agreement. Two review authors (PO and SBo) also screened included studies prior to commencement of meta‐analyses, and five review authors (TD, SBh, PO, YW, LF) completed a final review of included studies to confirm eligibility on the basis of study setting and experimental intervention.

We identified and excluded duplicate records, and collated multiple reports on the same study so that each study, rather than each report, was the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram and Characteristics of excluded studies table (Liberati 2009).

Data extraction and management

Two review authors (TD, SBh) independently extracted study characteristics and outcome data from included studies. These review authors discussed and resolved any disagreements. We contacted authors of the trials for further information where necessary to confirm details of the study method or to obtain additional data.

We extracted and entered information into Covidence data extraction forms, using the following categories.

• Study identification: source of funding; country; setting details; author's contact details

• Methods: study design; number of facilities

• Study population: inclusion and exclusion criteria; characteristics (age, gender, cognitive function) of participants reported for the whole sample and separately for each study arm; any differences in participant characteristics (e.g. demographics or baseline depression scores) between study arms reported by study authors

• Interventions and comparators: total duration and delivery mode; number, duration, and frequency of sessions; therapist characteristics; any co‐intervention provided

• Outcomes: description of primary and secondary outcomes; data reported on these outputs at different time points, including dropouts

We noted if outcome data were not reported in a usable way. Data were uploaded into Review Manager 5 (Review Manager 2014) directly from Covidence. We double‐checked that data were entered correctly by comparing the data presented in the systematic review with the study reports.

Assessment of risk of bias in included studies

Pairs of review authors (EY, CD, TD, SBh) independently assessed risk of bias for each study in duplicate using the Cochrane RoB 1 tool (Higgins 2017), with any disagreements resolved by discussion between the two review authors and with the broader team if required. We assessed risk of bias according to the following domains.

• Random sequence generation

• Allocation concealment

• Blinding of participants and personnel

• Blinding of outcome assessment

• Incomplete outcome data (e.g. due to loss to follow‐up)

• Selective outcome reporting

• Other bias

We rated each potential source of bias for outcomes as high, low, or unclear and provided a supporting quotation from the study report (if available), together with a justification for our judgement in the risk of bias table in Covidence forms. We summarized the risk of bias judgements across different studies for each of the domains listed. Where information on risk of bias related to unpublished data or correspondence with a trialist, we noted this in the risk of bias table. In cases where there was no information available to make a judgement, we explicitly stated this. We contacted authors of trials published from 2002 onwards for further information if missing.

When considering treatment effects, we considered the risk of bias for the studies that contributed to outcomes as a whole. For 'other bias', we considered treatment fidelity that could lead to bias, determined using a formal monitoring checklist, where researchers recorded the therapy against the protocol. We also considered potential researcher allegiance to the therapy. For cluster‐RCTs, we also considered recruitment bias and baseline imbalance.

Measures of treatment effect

Dichotomous data

We analyzed dichotomous data as odds ratios (ORs) and 95% confidence intervals (CIs). Where reported as risk ratios (RRs), we converted them to ORs (Deeks 2019).

Continuous data

We analyzed continuous data by calculating standardized mean difference (SMD) and 95% CIs, given studies used different outcome measures to assess the same outcome. We used endpoint data for all outcome time points. We did not use any change from baseline data.

For symptoms of depression and anxiety, a negative effect estimate indicated improvement. For quality of life and psychological well‐being, a positive effect estimate indicated improvement.

Unit of analysis issues

Cluster‐randomized controlled trials

We included cluster‐RCTs where LTC facilities were the unit of allocation if there were more than two clusters in each arm of the trial. We adjusted for the effect of clustering in meta‐analyses using an intracluster correlation coefficient (ICC). If an ICC was not reported and could not be obtained from the study authors, we used an ICC of 0.03, based on estimates from other cluster‐RCTs of interventions to reduce symptoms of depression in LTC (Davison 2021b; Underwood 2013).

Cross‐over trials

For cross‐over trials, we used data from the first treatment phase only, to avoid any carry‐over effects.

Studies with multiple treatment groups

For multiple‐arm studies (e.g. psychological intervention A, psychological intervention B, and control) included in the meta‐analysis, we combined the intervention groups to create a single pairwise comparison, in order to avoid possible bias caused by overlapping samples with multiple comparisons to one control group. Similarly, we pooled homogeneous control groups where applicable. We combined groups using appropriate formulae depending on whether the outcome was continuous (Higgins 2011a) or dichotomous (Higgins 2011b). However, in the subgroup analysis investigating the impact of each type of psychological therapy, for multiple‐arm studies we divided control groups evenly amongst the interventions to avoid analysis of overlapping samples.

Dealing with missing data

We contacted authors of 12 studies in order to verify key study characteristics and obtain missing numerical outcome data and information required to complete risk of bias assessments, with five responding with requested information and data (Dozeman 2011; Luo 2020; Meeks 2008; Meeks 2015; Reinhardt 2014). For the three cluster‐RCTs, we contacted study authors to request the ICC, as this information was not available in the trial reports (Abraham 1992; Luo 2020; Meeks 2015). No authors were able to provide the requested data.

For dichotomous data, we conducted an available‐case analysis only (i.e. no data imputations). For continuous data, we imputed endpoint mean and standard deviation (SD) where only alternative measures of spread and variance were reported. We imputed SDs from standard error (SE) for one study (Luo 2020), baseline for one study (Reinhardt 2014), P value and Z score for one study (Sales 2015), and most common trial (Hussian 1979) for one study (Zerhusen 1991). Moreover, we imputed mean and SD from median and interquartile range (IQR) for one study (Travers 2017). We were unable to obtain or impute mean or SD for one study (Rosen 1997).

Assessment of heterogeneity

We assessed heterogeneity with the I² statistic. Following the guidance in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2019), we used the following thresholds for the interpretation of the I² statistic.

• 0% to 40%: might not be important

• 30% to 60%: may represent moderate heterogeneity

• 50% to 90%: may represent substantial heterogeneity

• 75% to 100%: considerable heterogeneity

Assessment of reporting biases

We visually inspected funnel plots to assess possible publication bias when a meta‐analysis included at least 10 studies. However, we took into account the limitations of the funnel plots because they are prone to other causes of asymmetry (Forsman 2009).

We attempted to identify outcome reporting bias by comparing planned and reported outcomes. Where we found evidence of missing outcomes in recently published studies, we contacted the study authors for any available data. However, few protocols documenting planned outcomes were available, and the authors of only one study provided additional requested information (Meeks 2015).

Data synthesis

We conducted a meta‐analysis of included trials, using pairwise random‐effects models given the potential heterogeneity of the included trials. We used Review Manager 5 to conduct all analyses (Review Manager 2014).

Subgroup analysis and investigation of heterogeneity

Where data allowed, we conducted the following subgroup analyses for depressive symptoms at end‐of‐intervention.

• Baseline depression severity: this may impact on the primary outcome of interventions. We planned to test for differences between LTC residents with MDD and subthreshold depressive symptoms at baseline; however, the included studies did not allow for these analyses in practice. We also aimed to classify baseline levels of depressive symptoms as mild, moderate, and severe; however, insufficient data were available to enable this classification, with multiple scales used across studies and an absence of established criteria to indicate severity. Instead, we created two categories: high baseline depression and low baseline depression based on scores on a standardized 100‐point scale, by dichotomizing at the mid‐point (see 'Baseline depression severity' under Included studies for details).

• Types of psychological therapies: different therapies may have a different effect size and acceptability to participants. We tested for differences between different types of psychological therapy. There were sufficient data available to analyse the effect of CBT, BT, and reminiscence therapy (RT) on depressive symptoms in separate analyses.

• Types of comparators: different controls may yield differing effect sizes; therefore, we examined the effect of any psychological therapy compared to each distinct comparator type. There were sufficient data available to consider the effect compared with treatment as usual and non‐specific attentional control.

• Types of LTC residents: cognitive impairment may impact on the effect size and acceptability of interventions. We planned to test for differences in outcomes between LTC residents with and without dementia, and between residents with different levels of cognitive impairment. Given limitations in the information provided in study reports, we created two categories: high cognitive impairment (including where the majority of participants were reported to have dementia) and low cognitive impairment (see 'Cognitive function' under Included studies for details).

• LTC facility staff involvement in the psychological therapy: a previous review indicated that interventions implemented with the involvement of staff from the LTC facility, in addition to the therapist, were more effective than interventions delivered without staff involvement (Cody 2013). We tested for the effectiveness of facility staff involvement in the delivery of psychological therapies compared to no facility staff involvement, and therapies that were implemented by trained LTC facility staff only.

• Therapeutic contact: the duration and number of sessions may impact on the effectiveness and acceptability of psychological interventions. We tested for differences between therapies that vary in the total number of sessions, level of therapeutic contact (based on intervention duration), and dose (multiplying the number of sessions by the mean session duration). With an absence of established criteria for high versus low therapeutic contact, we categorized based on dichotomous data (see 'Format and therapeutic contact' under Included studies for details).

Subgroup analyses were only conducted for depressive symptomatology at end‐of‐intervention follow‐up between any psychological therapy versus any non‐therapy comparator. Given the limited number of studies for most subgroup analyses performed, as well as a loss of statistical power given the need to split control group samples to avoid overlapping samples, tests for subgroup differences should be considered exploratory in nature and interpreted with caution.

Sensitivity analysis

To test the robustness of the results of the primary outcome, we prespecified the following sensitivity analyses.

• Bias: excluding studies rated at high or unclear risk of bias for allocation concealment on the risk of bias assessment

• Attrition: excluding studies with dropout rate of 30% or above

• Missing data: excluding studies where we imputed missing data for meta‐analysis

• Treatment fidelity: excluding studies that did not measure treatment fidelity of the psychological models

Given the paucity of data, we only conducted sensitivity analyses for depressive symptomatology at end‐of‐intervention between any psychological therapy versus any control condition.

In addition to these planned sensitivity analyses, we performed additional sensitivity analyses to further explore findings of the review. We removed one study due to concerns regarding study design (Tsai 2008), and one study that used a depression outcome measure that required only a single positive item for study eligibility (Luo 2020). In addition, visual inspection of primary outcome data on forest plots indicated that one study may be an outlier (Daleo 1999), and it was removed in sensitivity analyses.

Summary of findings and assessment of the certainty of the evidence

Two review authors (SBo, PO) performed an assessment of the certainty of the evidence for each outcome using the GRADE approach described in the Cochrane Handbook for Systematic Reviews of Intervention (Schünemann 2017). Given all included studies were RCTs or cluster‐RCTs, our confidence in the evidence of each pairwise comparison was initially considered high. We considered five factors (limitations of detailed design and execution; inconsistency; indirectness; imprecision; publication bias) and, where appropriate, we downgraded our level of confidence in the available evidence.

We adopted the GRADE Working Group grades of evidence.

• High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.

• Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.

• Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect.

• Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

We constructed summary of findings tables to present the main findings of the review. The results of our analyses of any psychological therapy versus any non‐therapy comparator are reported in Table 1. This table reports the following primary and secondary outcomes, measured at end‐of‐intervention and short‐term follow‐up (where sufficient data were available), and presented standardized effect size estimates and 95% CIs.

• Depressive symptomatology

• Treatment non‐acceptability (number of participants who dropped out, not including those who dropped out for reasons clearly unrelated to the trial)

• Depression remission (presence or absence of MDD)

• Quality of life or psychological well‐being

• Anxious symptomatology

Additional summary of findings tables present results from key subgroup analyses examining the effect of level of baseline depression severity (Table 2), type of psychological therapy (Table 3), and type of non‐therapy comparator (Table 4).

1. Additional summary of findings. Baseline depression severity: end‐of‐intervention follow‐up.

Psychological therapies compared with non‐therapy comparators for high and low baseline depression severity
Patient or population: older people with depression Setting: long‐term care facilities Intervention: psychological therapy Comparison: non‐therapy comparator (treatment as usual, non‐specific attentional control, or wait list)
Outcomes	Anticipated absolute effects (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
HIGH BASELINE SEVERITY (> 54.8 points on a standardized 100‐point scale) Depressive symptomatology (lower = less depressive symptomatology): end‐of‐intervention follow‐up	The mean score for the psychological therapy group was 1.31 standard deviations lower (95% CI 2.08 to 0.53 lower) than for non‐therapy comparators	273 (9 RCTs)	⊕⊝⊝⊝ Very lowa,b,c	Measured with GDS‐15, GDS‐30, and BDI. The effect observed was large.d
LOW BASELINE SEVERITY (< 54.8 points on a standardized 100‐point scale) Depressive symptomatology (lower = less depressive symptomatology): end‐of‐intervention follow‐up	The mean score for the psychological therapy group was 0.81 standard deviations lower (95% CI 1.38 to 0.25 lower) than for non‐therapy comparators	371 (9 RCTs)	⊕⊝⊝⊝ Very lowa,b,c	Measured with GDS‐12, GDS‐15, GDS‐30, SDS, HAM‐D, CSDD, and CES‐D. The effect observed was large.d
BDI: Beck Depression Inventory; CES‐D: Center for Epidemiologic Studies Depression Scale; CI: confidence interval; CSDD: Cornell Scale for Depression in Dementia; GDS‐12, GDS‐15, GDS‐30: Geriatric Depression Scale – 12, 15, and 30 items; HAM‐D: Hamilton Depression Rating Scale; RCT: randomized controlled trials, including cluster‐randomized controlled trial; SDS: Zung Self‐Rating Depression Scale.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of the effect.

^aThe estimate indicated substantial‐to‐considerable statistical heterogeneity. Downgraded one level for inconsistency.
^bAll RCTs had a high risk of bias for at least one category. Downgraded two levels for risk of bias.
^cThe estimate was based on a collectively small sample size. Downgraded one level for imprecision.
^dBased on the following interpretations of SMD per Cohen: 0.2 = small effect, 0.5 = medium effect, 0.8 = large effect.

2. Additional summary of findings. Type of psychological therapy: end‐of‐intervention follow‐up.

Cognitive behavioural therapy, behavioural therapy, and reminiscence therapy compared with non‐therapy comparator
Patient or population: older people with depression Setting: low‐term care facilities Intervention: cognitive behavioural therapy, behavioural therapy, reminiscence therapy Comparison: non‐therapy comparator (treatment as usual, non‐specific attentional control, or wait list)
Outcomes	Anticipated absolute effects (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
COGNITIVE BEHAVIOURAL THERAPY Depressive symptomatology (lower = less depressive symptomatology): end‐of‐intervention follow‐up	The mean score for the cognitive behavioural therapy group was 1.02 standard deviations lower (95% CI 1.99 to 0.04 lower) than for non‐therapy comparators.	125 (5 RCTs)	⊕⊝⊝⊝ Very lowa,b,c	Measured with GDS‐30, BDI, and HAM‐D. The effect observed was large.d
BEHAVIOURAL THERAPY Depressive symptomatology (lower = less depressive symptomatology): end‐of‐intervention follow‐up	The mean score for the behavioural therapy group was 0.69 standard deviations lower (95% CI 1.21 to 0.18 lower) than for non‐therapy comparators.	303 (8 RCTs)	⊕⊝⊝⊝ Very lowa,b,c	Measured with GDS‐12, GDS‐15, GDS‐30, SDS, BDI, and CES‐D. The effect observed was medium.d
REMINISCENCE THERAPY Depressive symptomatology (lower = less depressive symptomatology): end‐of‐intervention follow‐up	The mean score for the reminiscence therapy group was 1.65 standard deviations lower (95% CI 2.83 to 0.47 lower) than for non‐therapy comparators.	203 (6 RCTs)	⊕⊝⊝⊝ Very lowa,b,c	Measured with GDS‐15, GDS‐30, SDS, and CSDD. The effect observed was large.d
BDI: Beck Depression Inventory; CES‐D: Center for Epidemiologic Studies Depression Scale; CI: confidence interval; CSDD: Cornell Scale for Depression in Dementia; GDS‐12, GDS‐15, GDS‐30: Geriatric Depression Scale – 12, 15, and 30 item; HAM‐D: Hamilton Depression Rating Scale; RCT: randomized controlled trial, including cluster‐randomized controlled trial; SDS: Zung Self‐Rating Depression Scale.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^aThe estimate indicated substantial to considerable statistical heterogeneity. Downgraded one level for inconsistency.
^bAll RCTs had a high risk of bias for at least one category. Downgraded two levels for risk of bias.
^cThe estimate was based on a collectively small sample size. Downgraded one level for imprecision.
^dBased on the following interpretations of SMD per Cohen: 0.2 = small effect, 0.5 = medium effect, 0.8 = large effect.

3. Additional summary of findings. Type of non‐therapy comparator: end‐of‐intervention follow‐up.

Psychological therapies compared with treatment as usual and with non‐specific attentional control comparators
Patient or population: older people with depression Setting: long‐term care facilities Intervention: psychological therapy Comparison: treatment as usual and non‐specific attentional control
Outcomes	Anticipated absolute effects (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
TREATMENT AS USUAL Depressive symptomatology (lower = less depressive symptomatology): end‐of‐intervention follow‐up	The mean score for the psychological therapy group was 1.47 standard deviations lower (95% CI 2.10 to 0.84 lower) than for the treatment as usual group	430 (12 RCTs)	⊕⊝⊝⊝ Very lowa,b	Measured with GDS‐12, GDS‐15, GDS‐30, BDI, and HAM‐D. The effect observed was large.c
NON‐SPECIFIC ATTENTIONAL CONTROL Depressive symptomatology (lower = less depressive symptomatology): end‐of‐intervention follow‐up	The mean score for the psychological therapy group was 0.30 standard deviations lower (95% CI 0.83 lower to 0.23 higher) than for non‐specific attentional control group	208 (6 RCTs)	⊕⊝⊝⊝ Very lowb,d	Measured with GDS‐15, GDS‐30, SDS, BDI, CSDD, and CES‐D. The effect observed was small.c
BDI: Beck Depression Inventory; CES‐D: Center for Epidemiologic Studies Depression Scale; CI: confidence interval; CSDD: Cornell Scale for Depression in Dementia; GDS‐12, GDS‐15, GDS‐30: Geriatric Depression Scale – 12, 15, and 30 items; HAM‐D: Hamilton Depression Rating Scale; RCT: randomized controlled trial, including cluster‐randomized controlled trial; SDS: Zung Self‐Rating Depression Scale.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^aThe estimate indicated substantial to considerable statistical heterogeneity. Downgraded one level for inconsistency.
^bAll RCTs had a high risk of bias for at least one category. Downgraded two levels for risk of bias.
^cBased on the following interpretations of SMD per Cohen: 0.2 = small effect, 0.5 = medium effect, 0.8 = large effect.
^dThe estimate was based on a collectively small sample size. Downgraded one level for imprecision.

The summary of findings tables were created before writing the discussion, abstract, and conclusions, which allowed all authors to consider the potential impact of the certainty of evidence for each estimate. Our confidence when interpreting results therefore informed the discussion, abstract, and conclusions of this review.

Results

Description of studies

Results of the search

The selection of studies through screening of study abstracts and full‐text records is reported in Figure 1. Overall, the database searches identified 11,192 records and searches of reference lists of review articles identified 114 records (total 11,306 records). A total of 9043 records remained after removal of duplicates. We screened the titles and abstracts of these 9043 records and excluded 8633 records at this stage.

1.

PRISMA flow diagram

We obtained and screened the full texts of 410 records. We excluded 375 full‐text records for the following reasons: ineligible participant population (171 records), ineligible design (91 records), ineligible setting (56 records), ineligible intervention (20 records), ineligible outcomes (two records), or ineligible comparator (one record). In addition, 26 records were duplicates and excluded, and eight records were review papers and excluded. We present a selection of excluded studies in the Characteristics of excluded studies table.

Seven records related to studies classified as 'awaiting classification' until further information is obtained (see Characteristics of studies awaiting classification table), and two records related to ongoing studies (see Characteristics of ongoing studies table).

A total of 26 records met the criteria for inclusion. After linking records belonging to the same study, we included 19 studies in the review (see Characteristics of included studies table).

Included studies

A total of 19 studies were included in the qualitative synthesis. Of these, 18 had sufficient data for meta‐analysis. We described the remaining study narratively (Rosen 1997).

Study design

Sixteen studies were parallel‐group RCTs and three were cluster‐RCTs (Abraham 1992; Luo 2020; Meeks 2015). There were no cross‐over RCTs identified. One study comprised two trials (Rosen 1997): the first was a parallel‐group RCT that was included in this systematic review (but there were insufficient data available for meta‐analysis). The second trial involved the same participants, who received another dose of the intervention (intervention group) or crossed over to receive the intervention (control group), and was not included in the review. As per the protocol, we did not include quasi‐RCTs, given that there were enough eligible RCTs for meta‐analysis. In one of the RCTs, each arm only received one intervention, but they were not delivered at the same time point (all outcome data from the treatment as usual arm were collected prior to commencement of the intervention) (Tsai 2008). We excluded this study from sensitivity analyses. In addition, one study was described as using a "quasi‐experimental interrupted time series design with three removed treatments", but nursing homes were randomly assigned to different arms of the trial, so it was considered a cluster‐RCT in our analyses (Abraham 1992). Where trial design was reported as a 'quasi‐randomized experiment' but the descriptions indicated that participants were randomly allocated to the condition, we considered the design to be an RCT for the purposes of this review.

Fifteen trials had two study arms, which compared a psychological therapy to treatment as usual (11 studies), a wait list control (Rosen 1997), or a condition that controlled for the effect of attention or social contact (Reinhardt 2014; Travers 2017; Tsai 2008). One study did not explicitly define the control condition (Sales 2015), but was assumed to be treatment as usual. One study had three arms (Karimi 2010), and one study had six arms (Hussian 1979). One additional study had three arms, comparing a psychological therapy with music therapy and a treatment as usual control (Zerhusen 1991), but we included only the psychological therapy and treatment as usual arms of the study in the systematic review and meta‐analysis. One other study had three arms (Abraham 1992), but focused visual imagery was not considered a clearly established psychological therapy and this arm was not included. If multiple study arms were determined to be variations of treatment or control, for example, two psychological therapy interventions, we pooled data for meta‐analysis. We did not compare psychological treatment arms due to paucity of data: Karimi 2010 compared two types of RT, and Hussian 1979 recruited only six participants per arm.

Sample size

The 19 studies included 873 participants at baseline (mean of 46 participants per study), with individual study sample sizes ranging from four participants per study arm (Sales 2015) to 62 and 67 participants in study arms (Dozeman 2011).

Setting

Most studies (11/19) were conducted in the USA. Two studies were conducted in Taiwan and two in Iran, while single studies were conducted in the Netherlands, Spain, Hong Kong, and Australia.

Studies were conducted in a variety of LTC facility settings, with types of facilities and naming of facility types reflecting the diversity in LTC funding and systems provided in different countries and across different states in the USA. Variation in setting descriptions also likely reflects changes that have occurred in LTC service provision over time, with studies published over a 44‐year period between 1976 (Hansen 1976) and 2020 (Luo 2020). The most common label describing the setting was 'nursing home', but other terms used (sometimes in combination) included LTC facility, skilled nursing facility, intermediate care facility, assisted living facility, rest home, residential home, and residence for the elderly. While all studies were set in a residential LTC facility, in general, the setting was not described in sufficient detail to determine the level and type of care that was provided to older adults by staff employed in the facility, such as the degree to which staff provided assistance with activities of daily living. An inclusive approach was adopted when determining eligibility for inclusion in this systematic review, including all studies set in a residential facility for older adults, regardless of the older adults' level of dependence in activities of daily living. The only exception was a study conducted exclusively in a "short‐term rehabilitation unit of an urban nursing home" (Sood 2003), which was considered ineligible and excluded from this systematic review (see Characteristics of excluded studies table).

Participants

The information available in study reports is described in Characteristics of included studies and briefly summarized below.

Age

The mean age of participants ranged from 70.5 years (Karimi 2010) to 85.2 years (Luo 2020). Four studies specified a minimum age of 60 years as an inclusion criterion (Bailey 2017; Hamzehzadeh 2018; Hussian 1979; Karimi 2010).

Gender

One study included only women (Hamzehzadeh 2018). All other studies included both men and women, although one study reported analyses for women only in a second paper (Abraham 1992). Most studies included more women than men; however, in a study set in a LTC facility for veterans, there were more men than women (16.0% women) (Hyer 2009), and another study included approximately equal numbers of men and women (Reinhardt 2014). In the remaining 10 studies that reported on the gender of participants, women represented between 58.6% and 90.2% of the sample. Due to poor reporting of gender in some studies, we were unable to report the gender distribution in the overall sample.

Cognitive function

Most studies excluded participants with a particular level of cognitive impairment or dementia, but the level of cognitive impairment and the instrument used to assess impairment varied between studies. Only one study did not report an exclusion criterion related to cognitive impairment or dementia (Hussian 1979). One study excluded participants with 'organic brain syndrome' (Zerhusen 1991), one excluded participants with 'senility' (Hansen 1976), and one excluded participants who were not 'oriented to self and other' (Daleo 1999).

The Mini‐Mental State Examination (MMSE; Folstein 1975) was the most commonly used instrument to assess level of cognitive function. Studies excluded older adults with MMSE less than 10 (Dhooper 1993), MMSE less than 14 (Meeks 2008; Meeks 2015), MMSE less than 16 (Tsai 2008), MMSE less than 18 (Hyer 2009; Reinhardt 2014; Rosen 1997), MMSE less than 21 (Dozeman 2011; Hamzehzadeh 2018; Karimi 2010), or MMSE less than 23 (Sales 2015). The remaining studies used other instruments to exclude older people with major cognitive impairment (Abraham 1992), or who were not cognitive intact or borderline intact (Luo 2020).

In contrast, two studies restricted recruitment to people with mild‐to‐moderate cognitive impairment (MMSE 10 to 24) or dementia, and excluded older adults with normal cognition (Bailey 2017; Travers 2017).

For the purposes of meta‐analysis, we created a subgroup of five studies where participants had high cognitive impairment, classified as studies with a mean participant MMSE of 23 or less (Bailey 2017; Luo 2020; Meeks 2008; Travers 2017) or where 'many' participants scored within the dementia range on the Modified Mini‐Mental State (3MS) Examination (Abraham 1992; Teng 1987). We also created a subgroup comprising seven studies where participants had low cognitive impairment, classified as studies with a mean participant MMSE greater than 23 (Dozeman 2011; Hyer 2009; Reinhardt 2014; Sales 2015; Tsai 2008), or where most participants were considered 'cognitively intact' or reported to have low cognitive impairment (Daleo 1999; Dhooper 1993). Other studies lacked sufficient information to enable classification into a high or low participant cognitive impairment subgroup, or appeared to include both subgroups but did not report results separately for participants with high versus low cognitive impairment.

Baseline depression severity

In 16 studies, inclusion criteria specified a minimum level of depressive symptoms using a screening tool, most commonly, the GDS‐30 (Yesavage 1983), GDS‐15 (Sheikh 1986), BDI (Beck 1961), or the SDS (Zung 1965). Studies typically used standard cut‐off scores indicative of a significant level of depressive symptoms, including a score of 10 or 11 on the GDS‐30 (Abraham 1992; Meeks 2008; Meeks 2015; Sales 2015; Tsai 2008), a score of 5, 6, or 8 on the GDS‐15 (Daleo 1999; Hamzehzadeh 2018; Hyer 2009; Karimi 2010), and a score of 50 on the SDS (Dhooper 1993; Hansen 1976).

One study recruited residents with "the presence of a mood problem indicated by the Resident Assessment Protocol (RAP), where one or more symptoms from 17 listed symptoms indicated a mood problem" (Luo 2020, p. 196). This instrument has been validated previously (Chou 2001), and so was included in the systematic review, but due to the low number of symptoms required for participant eligibility, we performed sensitivity analyses removing this study for relevant meta‐analyses. The authors indicated that "all of the participants had moderate depressive symptoms" (p. 200).

One study did not specify an instrument cut‐off but indicated that participants were moderately to severely depressed, based on the BDI score (Zerhusen 1991). Another study using the BDI similarly did not apply a cut‐off, but rather selected residents with the highest scores to participate in the study (Hussian 1979). Most participants in this latter study scored in the severely depressed range on the BDI (mean 35.64) (Hussian 1979). One study used a clinical diagnosis rather than a depression scale to define eligibility (Rosen 1997).

Five studies restricted eligibility to participants who met defined diagnostic criteria for a mood disorder, which typically included MDD, 'minor depression', and dysthymia (Meeks 2008; Meeks 2015; Reinhardt 2014; Rosen 1997), in addition to adjustment disorder with depression (Hyer 2009). No studies were restricted to residents with MDD at baseline.

In contrast, four studies excluded participants with more‐severe depression. One study focusing on subthreshold depression excluded participants who met DSM‐IV criteria for a depressive or anxiety disorder (Dozeman 2011), while three studies excluded participants with a severe level of symptoms (Dhooper 1993; Reinhardt 2014; Rosen 1997), as indicated by a high score on a depression rating scale or the results of a clinical interview.

Based on information from eligibility criteria and descriptions of study samples, we concluded that, in most studies, the majority of participants had mild‐to‐moderate levels of depressive symptoms. It was usually not possible to classify baseline levels of depressive symptoms of participants in individual studies as mild, moderate, and severe, due to the nature of the outcome measures used (i.e. an absence of an accepted cut‐off score to define severity), the limited use of upper limits of depression symptoms, and the limited information reported. Thus, subgroup analyses for baseline depression severity were conducted with two categories only: high versus low baseline symptoms of depression, based on scores on a standardized 100‐point scale (dichotomized at ± 54.8 points).

Intervention

Description of intervention

All interventions are described in the Characteristics of included studies table. We categorized interventions as CBT, BT, or RT on the basis of their content (i.e. the therapeutic techniques mainly used) and theoretical background (manual used).

BT referred to techniques such as pleasant events and activities, activity scheduling, positive reinforcement, behaviour modification strategies, exposure therapy, psycho‐education, and relaxation, which draw on social learning and behavioural models (e.g. operant conditioning; Lewinsohn 1974). Nine studies used BT. The dominant BT technique in seven of these studies was activity scheduling. The remaining two studies used psycho‐education (Hansen 1976), and positive reinforcement (Hussian 1979). Interventions were described as:

• guided self‐help activity scheduling (Dozeman 2011);

• didactic group therapy (Hansen 1976);

• group individual and staff therapy (GIST) (Hyer 2009);

• Positive Mood and Active Life (PMAL) programme (Luo 2020);

• behavioural activities intervention (BE‐ACTIV) (Meeks 2008; Meeks 2015; Travers 2017);

• control relevant intervention (Rosen 1997); and

• social reinforcement of activity (Hussian 1979).

CBT referred to interventions designed to help clients understand the link between thoughts, feelings, and behaviours and to modify unhelpful thinking patterns and underlying assumptions about self, others, and the world, as theorized by Beck 1979. Common CBT strategies included cognitive restructuring, problem‐solving training, and behavioural experiments. We categorized studies as using CBT when authors described a key aim of treatment as modification of beliefs and included cognitive modification or problem‐solving techniques as central aspects of the treatment. Five studies utilized CBT: three used cognitive restructuring (Abraham 1992; Sales 2015; Zerhusen 1991), and two used problem‐solving techniques (Hussian 1979; Reinhardt 2014). The interventions were labelled as:

• cognitive behavioural group therapy programme (Abraham 1992);

• problem‐solving training with or without social reinforcement of activity (Hussian 1979);

• problem‐solving treatment (Reinhardt 2014);

• cognitive restructuring therapy (Sales 2015); and

• cognitive therapy programme (Zerhusen 1991).

RT referred to interventions designed to help clients reflect on the past. RT techniques include life review, life review therapy, and simple reminiscence. Life review techniques are used to help clients understand meaning and continuous themes in their life‐history. Life review therapy techniques are used to help clients draw on experiences to reframe beliefs or cope with current problems. Simple reminiscence is used to help clients create social connections with others through sharing of memories. These approaches are based on models advanced by Butler 1963 and Webster 2010. We categorized studies as using RT when authors described a key component of their intervention as based on reminiscence. Seven RT interventions were featured in six studies: these interventions were categorized as life review (Karimi 2010; Tsai 2008), life review therapy (Bailey 2017; Hamzehzadeh 2018; Karimi 2010), and simple reminiscence (Daleo 1999; Dhooper 1993). The interventions were labelled as:

• question‐asking‐response (QAR) depression intervention (Bailey 2017);

• structured reminiscent group therapy (Daleo 1999);

• coping together group (Dhooper 1993);

• integrative and instrumental reminiscence groups (Karimi 2010);

• self‐worth therapy (Tsai 2008); and

• positive reminiscing therapy (Hamzehzadeh 2018).

One study described two interventions, one called 'Focussed visual imagery group therapy' (the second intervention was CBT) (Abraham 1992). The intervention involved visualization strategies to appreciate positive aspects of loss and change, to emphasise personal qualities and strengths, and to imagine desired results. We did not consider this approach to be consistent with protocols or theories of established psychotherapies, and hence omitted this treatment arm from the current review. However, the CBT arm of the trial was included.

Type of therapist and facility staff involvement in implementation

In nine studies, psychological therapies were delivered by specialists, including social workers (Dhooper 1993; Luo 2020; Reinhardt 2014; Travers 2017), a mental health consultant (Meeks 2008), a clinical nurse specialist (Abraham 1992), a Masters level therapist (Karimi 2010), a Masters‐prepared mental health nurse (Tsai 2008), and unspecified 'therapist' (Hyer 2009). In five studies, treatments were administered by graduate or doctoral students (Bailey 2017; Daleo 1999; Hansen 1976; Hussian 1979; Meeks 2015). Specialists and graduate students were both classified as 'professional therapists' in subgroup analyses. Of these 14 studies, nine recruited LTC personnel such as personal care workers, aged care volunteers, or recreational therapists to co‐deliver interventions, as assistants, or to encourage and support older people to use the agreed treatment strategies (Bailey 2017; Daleo 1999; Hussian 1979; Hyer 2009; Karimi 2010; Luo 2020; Meeks 2008; Meeks 2015; Travers 2017). In the remaining five studies, LTC facility staff were not involved in delivering or co‐delivering treatment activities.

In addition to the above 14 studies where professional therapists implemented the interventions, either with the involvement of LTC facility staff, three studies trained LTC personnel, such as personal care workers or recreational therapists to deliver the intervention activities as the primary facilitators (Dozeman 2011; Rosen 1997; Zerhusen 1991). In the remaining two studies, no details were provided about the personnel implementing the intervention (Hamzehzadeh 2018; Sales 2015).

We explored the effect of therapist type and involvement of LTC staff in interventions in subgroup analyses, comparing interventions led by professional therapists with and without LTC staff involvement, and comparing interventions where LTC staff were trained to implement the interventions as the primary facilitators with those delivered by professional therapists.

Format and therapeutic contact

All interventions involved in‐person contact. In nine studies, interventions were delivered in a group format (Abraham 1992; Bailey 2017; Daleo 1999; Dhooper 1993; Hamzehzadeh 2018; Hansen 1976; Karimi 2010; Sales 2015; Zerhusen 1991). In six studies, interventions were individually administered (Luo 2020; Meeks 2008; Meeks 2015; Reinhardt 2014; Travers 2017; Tsai 2008), and in two studies, interventions involved a combination of group and individual formats (Hussian 1979; Hyer 2009). In two studies, interventions were structured self‐directed activity with guidance from therapists (Dozeman 2011; Rosen 1997).

Most interventions were delivered in one or two sessions per week. In one study, interventions were delivered in five sessions per week (Hussian 1979). Most interventions involved between two and 10 sessions in total. No interventions were delivered in a single session; however, in one study, the same session was repeated 15 times, and participants could attend as many sessions as desired (Hyer 2009). We classified the total number of sessions as low (fewer than nine sessions) or high (more than nine sessions), based on dichotomized data.

Most interventions were delivered over a period of two to 12 weeks, with the longest total duration being 24 weeks (Abraham 1992). Treatment duration was classified as low (eight weeks or less) or high (nine weeks or more), based on dichotomized data.

Intervention sessions were usually up to one hour in duration, although in two studies, interventions lasted between 80 and 90 minutes (Hyer 2009; Karimi 2010). In six studies, session duration was not specified (Abraham 1992; Dozeman 2011; Luo 2020; Meeks 2015; Luo 2020; Rosen 1997; Zerhusen 1991). Total treatment dose was calculated by multiplying the number of sessions by the average session duration. Notably, we omitted one trial from the total hours estimate, given an identical session was provided on 15 occasions, which differed markedly to psychological therapies examined in other trials (Hyer 2009). Across studies, treatment dose ranged from a minimum of 120 minutes (two hours) (Tsai 2008) to a maximum of 600 minutes (10 hours) (Hansen 1976). We classified the dose as low (less than 6.5 total hours) or high (more than 6.5 total hours) based on dichotomized data.

We conducted separate subgroup analyses to explore the effect of therapeutic contact, considering treatment duration (in weeks) (Analysis 7.1), number of sessions (Analysis 7.2), and total treatment dose (session length multiplied by number of sessions) (Analysis 7.3). One trial was omitted from these estimates, given the use of a single repeated treatment session (Hyer 2009).

7.1. Analysis.

Comparison 7: Subgroup 6: therapeutic contact: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology, therapeutic contact (weeks)

7.2. Analysis.

Comparison 7: Subgroup 6: therapeutic contact: psychological therapies versus non‐therapy comparators, Outcome 2: Depressive symptomatology, therapeutic contact (number of sessions)

7.3. Analysis.

Comparison 7: Subgroup 6: therapeutic contact: psychological therapies versus non‐therapy comparators, Outcome 3: Depressive symptomatology, therapeutic contact (total treatment dose)

Comparators

Studies used various non‐therapy comparators.

Waiting list (Hussian 1979; Rosen 1997). Participants randomized to the waiting list were assigned to the active intervention arm after all participants in the intervention arm had received the intervention. During the course of the trial, people on the waiting list could receive any appropriate medical care.

Treatment as usual (Bailey 2017; Daleo 1999; Dhooper 1993; Dozeman 2011; Hamzehzadeh 2018; Hansen 1976; Hyer 2009; Luo 2020; Meeks 2008; Meeks 2015; Sales 2015; Zerhusen 1991). Treatment as usual, standard care, or usual care included any appropriate medical or nursing care provided routinely during the course of the study. This may, for example, have involved monitoring the person receiving treatment, regular check‐ups, no treatment, or any type of treatment, as well as access to a programme of planned activities within the facility. What constituted treatment as usual depended on the facility policies and resources in which the study was conducted. In one study, the comparator condition was unspecified; we assumed the condition involved treatment as usual (Sales 2015).

Non‐specific attentional control (Abraham 1992; Hussian 1979; Karimi 2010; Reinhardt 2014; Travers 2017; Tsai 2008). Non‐specific attentional control refers to conditions in which participants were provided with attention from researchers, facility staff, or volunteers, or were engaged to participate in activities. The duration and frequency of such activities matched those of the intervention condition. The control condition simulated the attention received by participants in the intervention condition, while excluding the techniques purported to be therapeutic in the intervention. Such control conditions included friendly visits by research staff (Reinhardt 2014; Tsai 2008), and social discussion or education groups (Abraham 1992; Hussian 1979; Karimi 2010; Travers 2017).

We did not identify any studies with a defined co‐intervention control group, where there was a separate non‐psychological treatment administered in both the control and intervention arms of the study.

Outcomes

All studies reported data on level of depressive symptomatology, with 11 studies using the GDS‐12 (Travers 2017), GDS‐15 (Daleo 1999; Hamzehzadeh 2018; Hyer 2009; Karimi 2010; Luo 2020), or GDS‐30 (Abraham 1992; Meeks 2008; Meeks 2015; Sales 2015; Tsai 2008). Other studies measured depressive symptomatology using the SDS (Dhooper 1993; Hansen 1976), BDI (Hussian 1979; Zerhusen 1991), HAM‐D (Reinhardt 2014; Rosen 1997), the Cornell Scale for Depression in Dementia (Bailey 2017), or the Centre for Epidemiological Studies – Depression (Dozeman 2011).

We used participant attrition to indicate treatment non‐acceptability. We did not include dropouts that were clearly attributable to reasons unrelated to the study or intervention, such as decline in physical health, death, or relocation during the course of the study that impacted on participants' ability to continue in the study in this measure. Twelve studies were judged as providing treatment non‐acceptability data based on this definition (Abraham 1992; Daleo 1999; Dhooper 1993; Dozeman 2011; Hamzehzadeh 2018; Hansen 1976; Hyer 2009; Luo 2020; Meeks 2015; Reinhardt 2014; Travers 2017; Tsai 2008). Seven studies reported zero attrition from both arms of the trial. Only five studies reported attrition from the psychological therapy (with a total of 18 dropouts from 312 participants) or non‐therapy comparator (with a total of four dropouts from 273 participants).

Two studies reported data on depression remission (Meeks 2008; Meeks 2015). One study reported number of participants with MDD, 'minor depression', 'intermittent depression', and no depression at end‐of‐treatment (total 14 participants) and 24‐week follow‐up (10 participants) (Meeks 2008). However, the sample included residents with minor depression and intermittent depression at baseline, and the study authors did not report outcomes separately for those with MDD at baseline. For a subsequent larger trial of the intervention (Meeks 2015), the study authors reported data on change in diagnostic outcomes for a mixed‐diagnostic group in the published paper, but provided additional data on the number of participants with MDD (including those with MDD in partial remission that did not reach criteria for full remission) at each time point upon request.

Six studies reported data on quality of life or psychological well‐being (Abraham 1992; Bailey 2017; Hyer 2009; Luo 2020; Sales 2015; Travers 2017), and two studies reported data on level of anxious symptomatology (Dozeman 2011; Sales 2015). No studies reported outcomes for physical functioning, agitation, or adverse events.

We conducted meta‐analyses for all outcomes with end of intervention as the primary end point. In addition, we considered short‐, medium‐, and long‐term follow‐up where there were data from two or more studies available. For depressive symptomatology, we meta‐analyzed short‐term (16 studies), medium‐term (eight studies), and long‐term (two studies) follow‐up. For quality of life and psychological well‐being, we meta‐analyzed only short‐term (five studies) and medium‐term (two studies) follow‐up, in addition to end‐of‐intervention data. Level of anxiety symptomatology was only considered at end‐of‐intervention given data paucity. Participant attrition was not described consistently across studies, although it was most commonly described over the total study period. However, due to limited data, treatment non‐acceptability was only considered in meta‐analyses at end‐of‐intervention.

One study had missing data and neither mean nor SD could be calculated; hence, this study was not included in meta‐analyses (Rosen 1997).

Excluded studies

After obtaining full‐text manuscripts, we excluded 375 studies (Figure 1).

We listed a selection of these studies in the Characteristics of excluded studies table if they were initially judged by review authors as eligible during the full‐text screen and identified as ineligible on further examination or if additional information obtained from study authors resulted in the study being reclassified as ineligible (Ahmadpanah 2017; Hsu 2019). Amongst the 10 studies listed in the Characteristics of excluded studies table, four were not RCTs or cluster‐RCTs, three were based in an ineligible setting (retirement flat, day care centre, rehabilitation unit), two did not focus on psychological therapies (stepped care, case management), and one did not use a validated tool to identify residents with depression.

Studies awaiting classification

Seven studies are awaiting classification (Bai 2018; Funaki 2011; Khezri Moghadam 2018; Lemos 2016; Lopes 2017; Pang 2020; Sturz 2015).

Ongoing studies

Two studies are ongoing (Bhar 2023; Velasquez Reyes 2019).

Risk of bias in included studies

All 19 included studies had at least one or more risk of bias domains rated as 'unclear' because information was missing from the study report or trial protocol/registration, or both (see Figure 2; Figure 3). Several papers provided an insufficient level of information about the methods used in the trials. We located only one published protocol paper (Dozeman 2011) and three trial registrations (Meeks 2008; Meeks 2015; Travers 2017). We contacted authors of trials published from 2002 onwards for additional information, with five authors responding to either provide this information or to notify reviewers that the information was not available.

2.

3.

Allocation

Four studies were at low risk of selection bias (random sequence generation and allocation concealment; Dozeman 2011; Reinhardt 2014; Travers 2017; Tsai 2008); for example, reporting the use of a random number generator to assign participants to conditions and central allocation by a biostatistician not otherwise involved in the trial (Figure 2; Figure 3). Most other studies failed to provide sufficient information about either the random sequence generation process or allocation concealment to allow a rating to be made. Two studies were at high risk of selection bias: one due to the use of additional participants to replace those who terminated the trial (Hussian 1979), and one study randomized participants based on the presence or absence of a diagnosis of MDD (Hyer 2009).

Blinding

Blinding of participants and personnel (performance bias) was likely not achieved in any of the included studies, and is typically not attempted in trials of psychological interventions (Uphoff 2020). One study comparing a psychological therapy to an active control condition implemented by different personnel was at unclear risk, based on the possibility that personnel and participants involved in an intervention that controlled for the effects of additional attention and time (participation in a discussion group), may have believed they were taking part in an active treatment (Abraham 1992). Hamzehzadeh 2018 provided insufficient information to determine blinding of participants and personnel. All other trials were rated at high risk.

In seven studies, authors limited the risk of detection bias by blinding personnel collecting outcome measures and were at low risk of detection bias (Abraham 1992; Dozeman 2011; Hyer 2009; Karimi 2010; Meeks 2008; Meeks 2015; Tsai 2008). In other trials, either outcome assessors were not blinded (high risk) or there was insufficient information provided (unclear risk).

Incomplete outcome data

Eight studies were at low risk of attrition bias (Daleo 1999; Dozeman 2011; Hansen 1976; Hyer 2009; Luo 2020; Meeks 2015; Rosen 1997; Travers 2017). Ten studies had incomplete outcome data (high risk). The key issues of concern were high rates of attrition overall, and uneven attrition from each study arm, including for reasons that could potentially affect trial outcomes (e.g. participants unwilling to continue). Incomplete data were typically handled by excluding participants from analyses, and only two studies imputed missing data (Dozeman 2011; Meeks 2015). One study did not provide sufficient information on participant attrition to allow a rating to be made (unclear risk; Sales 2015).

Selective reporting

Only two studies were at low risk of reporting bias, with all relevant outcomes listed in a prospective trial registration reported in the paper (Meeks 2015; Travers 2017). One study was at high risk of reporting bias, with the primary outcome from a prospectively published protocol paper not reported in the efficacy paper (Dozeman 2011). Risk of reporting bias could not be ascertained in the remaining 16 studies, due to the absence of an available published protocol paper or trial registration detailing outcomes. However, in most cases, all outcomes listed in the methods section of the paper were reported in the results.

Other potential sources of bias

Thirteen studies were at a high risk of other bias. Only four studies reported that independent assessments of treatment fidelity were conducted (Bailey 2017; Meeks 2008; Meeks 2015; Zerhusen 1991), although two of these studies had additional potential bias (Meeks 2008; Meeks 2015). Other identified issues included very small sample sizes or a small number of clusters relative to sample size and treatment arms, which may have impacted on the success of randomization efforts. Researcher or therapist allegiance was considered a potential risk in several trials, particularly for interventions that were developed by the researcher/author, and where the researcher was directly involved in key aspects of the implementation of the trial, including recruitment, randomization, outcome assessment, treatment implementation, and data analysis.

Effects of interventions

See: Table 1

See:Table 1; Table 2; Table 3; Table 4.

Psychological therapies versus non‐therapy comparators

Included studies compared any psychological therapy with any non‐therapy comparator (waiting list, treatment as usual, non‐specific attentional control).

Comparison 1: any psychological therapy versus any non‐therapy comparator

End‐of‐intervention outcomes

Depressive symptomatology

Psychological therapy may reduce depressive symptomatology at end‐of‐intervention follow‐up compared to any non‐therapy comparator, with a large effect estimate, but the evidence was very uncertain (SMD −1.04, 95% CI −1.49 to −0.58; 18 RCTs, 644 participants; Analysis 1.1). Sensitivity analyses (omitting Daleo 1999, Luo 2020, and Tsai 2008, due to concerns about study methods) had minimal impact on this overall conclusion (Analysis 12.1; Analysis 13.1; Analysis 14.1). A funnel plot revealed no indications of publication bias for depressive symptomatology (Figure 4).

1.1. Analysis.

Comparison 1: Any psychological therapy versus any non‐therapy comparator, Outcome 1: Depressive symptomatology

12.1. Analysis.

Comparison 12: Sensitivity analysis 5: omit Luo 2020: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology

13.1. Analysis.

Comparison 13: Sensitivity analysis 6: omit Tsai 2008: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology

14.1. Analysis.

Comparison 14: Sensitivity analysis 7: omit Daleo 1999: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology

4.

Figure 4. Funnel plot of comparison 1: Any psychological therapy versus any non‐therapy comparator, Outcome 1: Depressive Symptomatology, Follow‐up: End‐of‐intervention.

One study could not be included in the meta‐analysis due to incomplete data (Rosen 1997). The study authors reported that amongst those randomly allocated to psychological therapy, the improvement in depressive symptomatology was not greater than for those allocated to the non‐therapy comparator (Rosen 1997).

Treatment non‐acceptability

While 12 studies reported attrition data that could be used to infer treatment non‐acceptability, seven studies had zero dropouts from each treatment arm. There were dropouts in only five studies, which contributed to the effect estimate.

Treatment non‐acceptability may be higher at end‐of‐intervention follow‐up for any psychological therapy compared to any non‐therapy comparator, but the evidence was very uncertain (OR 3.44, 95% CI 1.19 to 9.93; 5 RCTs, 313 participants; Analysis 1.2). Interpretation of the result for treatment non‐acceptability requires particular caution due to inadequate reporting of this outcome and potential unreliability of the estimate, with very wide CIs for ORs (e.g. 0.47 to 202.37 for Hamzehzadeh 2018, and 0.30 to 130.07 for Reinhardt 2014). As Daleo 1999, Luo 2020, and Tsai 2008 did not contribute data to the effect estimate, sensitivity analyses omitting these RCTs were unnecessary.

1.2. Analysis.

Comparison 1: Any psychological therapy versus any non‐therapy comparator, Outcome 2: Treatment non‐acceptability

Depression remission

The authors of one study provided data on depression remission (including participants with partial remission) at end‐of‐intervention follow‐up (Meeks 2015). Amongst participants allocated to psychological therapy, the prevalence of MDD decreased from 47.6% at baseline to 43.4% at end‐of‐intervention follow‐up (4.2% decrease). Amongst participants allocated to the non‐therapy comparator, the prevalence of MDD increased from 25.0% at baseline to 75.0% at end‐of‐intervention follow‐up (50.0% increase). The net difference in change in prevalence between conditions was 54.2%.

Quality of life or psychological well‐being

Psychological therapy may have little to no effect on quality of life and psychological well‐being compared to any non‐therapy comparator at end‐of‐intervention follow‐up, but the evidence was very uncertain (SMD 0.40, 95% CI −0.02 to 0.82; 6 RCTs, 195 participants; Analysis 1.3). Omitting Luo 2020 due to concerns with the tool used for participant eligibility had minimal impact on this conclusion (SMD 0.25, 95% CI −0.17 to 0.68; Analysis 12.2).

1.3. Analysis.

Comparison 1: Any psychological therapy versus any non‐therapy comparator, Outcome 3: Quality of life and psychological well‐being

12.2. Analysis.

Comparison 12: Sensitivity analysis 5: omit Luo 2020: psychological therapies versus non‐therapy comparators, Outcome 2: Quality of life and psychological well‐being

Anxious symptomatology

Psychological therapy may have little to no effect on anxious symptomatology compared to any non‐therapy comparator at end‐of‐intervention follow‐up, but the evidence was very uncertain (SMD −0.68, 95% CI −2.50 to 1.14; 2 RCTs, 115 participants; Analysis 1.4).

1.4. Analysis.

Comparison 1: Any psychological therapy versus any non‐therapy comparator, Outcome 4: Anxious symptomatology

Due to data paucity, the following outcome measures were unable to be examined at end‐of‐intervention follow‐up: physical functioning, agitation, and adverse effects.

Short‐term outcomes

Depressive symptomatology

Depressive symptomatology was lower at short‐term follow‐up for any psychological therapy compared to any non‐therapy comparator, with a large effect estimate, but the evidence was very uncertain (SMD −1.03, 95% CI −1.49 to −0.56; 16 RCTs, 512 participants; Analysis 1.1). Sensitivity analyses (omitting Daleo 1999, Luo 2020, and Tsai 2008, due to concerns about study methods) had minimal impact on this overall conclusion (Analysis 12.1; Analysis 13.1; Analysis 14.1).

Depression remission

Data on depression remission (including those with partial remission) at short‐term follow‐up was provided separately by the author for the single study that examined this outcome (Meeks 2015). Results were as for end‐of‐intervention follow‐up. Amongst participants allocated to psychological therapy, the prevalence of MDD decreased from 47.6% at baseline to 43.4% at short‐term follow‐up (4.2% decrease). Amongst participants allocated to the non‐therapy comparator, the prevalence of MDD increased from 25.0% at baseline to 75.0% at short‐term follow‐up (50.0% increase). The net difference in change in prevalence between conditions was 54.2%.

Quality of life or psychological well‐being

Quality of life and psychological well‐being were higher at short‐term follow‐up for any psychological therapy compared to any non‐therapy comparator, with a medium effect estimate, but the evidence was very uncertain (SMD 0.51, 95% CI 0.19 to 0.82; 5 RCTs, 170 participants; Analysis 1.3). When Luo 2020 was omitted due to concerns with the tool used for participant eligibility, the effect estimate became small and crossed the line of no effect (SMD 0.35, 95% CI −0.02 to 0.72; Analysis 12.2).

Due to data paucity, the following outcome measures were unable to be examined at short‐term follow‐up: treatment non‐acceptability, anxious symptomatology, physical functioning, agitation, adverse effects.

Medium‐term outcomes

Depressive symptomatology

Psychological therapy may reduce depressive symptomatology at medium‐term follow‐up compared to any non‐therapy comparator, with a small effect estimate, but the evidence was very uncertain (SMD −0.43, 95% CI −0.81 to −0.06; 8 RCTs, 355 participants; Analysis 1.1). When Tsai 2008 was omitted due to concerns with study design, there was no evidence of an effect (SMD −0.44, 95% CI −0.90 to 0.02; Analysis 13.1).

Depression remission

Data on depression remission (including those with partial remission) at medium‐term follow‐up was provided separately by the author for the single study that examined this outcome (Meeks 2015). Amongst participants allocated to psychological therapy, the prevalence of MDD increased from 47.6% at baseline to 48.5% at medium‐term follow‐up (0.9% increase). Amongst participants allocated to the non‐therapy comparator, the prevalence of MDD at medium‐term follow‐up increased from 25.0% to 73.1% (48.1% increase). The net difference in change in prevalence between conditions was 47.2%.

Quality of life or psychological well‐being

Psychological therapy may have little to no effect on quality of life and psychological well‐being compared to any non‐therapy comparator at medium‐term follow‐up, but the evidence was very uncertain (SMD 0.26, 95% CI −0.95 to 1.48; 2 RCTs, 68 participants; Analysis 1.3).

Due to data paucity, the following outcome measures were unable to be examined at medium‐term follow‐up: treatment non‐acceptability, anxious symptomatology, physical functioning, agitation, adverse effects.

Long‐term outcomes

Depressive symptomatology

Psychological therapy may have little to no effect on depressive symptomatology compared to any non‐therapy comparator at long‐term follow‐up, but the evidence was very uncertain (SMD −0.16, 95% CI −0.58 to 0.27; 2 RCTs, 92 participants; Analysis 1.1).

Depression remission

Data on depression remission (including those with partial remission) at long‐term follow‐up was provided separately by the author for the only study that examined this outcome (Meeks 2015). Amongst participants allocated to psychological therapy, the prevalence of MDD increased from 47.6% at baseline to 55.5% at long‐term follow‐up (7.9% increase). Amongst participants allocated to the non‐therapy comparator, the prevalence of MDD increased from 25.0% to 65.2% (40.2% increase). The net difference in change in prevalence between conditions was 32.3%.

Due to data paucity, the following outcome measures were unable to be examined at long‐term follow‐up: treatment non‐acceptability, quality of life and psychological well‐being, anxious symptomatology, physical functioning, agitation, adverse effects.

Subgroup analyses

Preplanned subgroup analyses were limited by data paucity and only conducted for depressive symptomatology at end‐of‐intervention follow‐up between any psychological therapy versus any non‐therapy comparator by the following variables.

• Subgroup 1: baseline depression severity (dichotomized: ± 54.8 points on a standardized 100‐point scale)

• Subgroup 2: type of psychological therapy (CBT, BT, or RT)

• Subgroup 3: type of non‐therapy comparator (treatment as usual or non‐specific attentional control)

• Subgroup 4: level of cognitive function (high or low cognitive impairment)

• Subgroup 5: LTC facility staff involvement in therapy (professional therapist without staff involvement, professional therapist with staff involvement, or implemented by LTC facility staff only)

• Subgroup 6: therapeutic contact (dichotomized: ± 8.5 weeks (duration), ± 9.5 total sessions, ± 6.5 contact hours (dose))

Subgroup 1: baseline depression severity

Psychological therapy may reduce depressive symptomatology at end‐of‐intervention follow‐up for participants with baseline depression severity greater than 54.8/100 points (dichotomized at the median on a standardized 100‐point scale; see Subgroup analysis and investigation of heterogeneity), with a large effect estimate (SMD −1.31, 95% CI −2.08 to −0.53; 9 RCTs, 273 participants; Analysis 2.1); and for participants with baseline depression severity less than 54.8/100 points, with a large effect estimate (SMD −0.81, 95% CI −1.38 to −0.25; 9 RCTs, 371 participants; Analysis 2.1); however, the evidence was very uncertain. Subgroup analysis showed no difference in effect by baseline depression severity (Chi² = 1.01, P = 0.31; Analysis 2.1).

2.1. Analysis.

Comparison 2: Subgroup 1: baseline depression severity: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology (end‐of‐intervention)

Subgroup 2: type of psychological therapy

Depressive symptomatology may be lower at end‐of‐intervention follow‐up for CBT, with a large effect estimate (SMD −1.02, 95% CI −1.99 to −0.04; 5 RCTs, 125 participants); for BT, with a medium effect estimate (SMD −0.69, 95% CI −1.21 to −0.18; 8 RCTs, 303 participants); and for RT, with a large effect estimate (SMD −1.65, 95% CI −2.83 to −0.47; 6 RCTs, 203 participants) when compared to any non‐therapy comparator (Analysis 3.1); however, the evidence was very uncertain. Subgroup analysis showed no difference in effect by type of psychological therapy (Chi² = 2.23, P = 0.33; Analysis 3.1).

3.1. Analysis.

Comparison 3: Subgroup 2: type of psychological therapy: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology (end‐of‐intervention)

Subgroup 3: type of non‐therapy comparator

Psychological therapy may reduce depressive symptomatology at end‐of‐intervention follow‐up compared to treatment as usual comparator, with a large effect estimate (SMD −1.47, 95% CI −2.10 to −0.84; 12 RCTs, 430 participants), but not compared to non‐specific attentional control comparator (SMD −0.30, 95% CI −0.83 to 0.23; 6 RCTs, 208 participants) (Analysis 4.1); however, the evidence was very uncertain. Subgroup analysis showed a difference by non‐therapy comparator, whereby comparison to treatment as usual yielded greater effect estimates than comparison to non‐specific attentional control (Chi² = 7.87, P = 0.005; Analysis 4.1).

4.1. Analysis.

Comparison 4: Subgroup 3: type of non‐therapy comparator: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology (end‐of‐intervention)

Subgroup 4: level of cognitive function

Psychological therapy may reduce depressive symptomatology at end‐of‐intervention follow‐up compared to non‐therapy comparator for studies with participants with low cognitive impairment, with a large effect estimate (SMD −1.30, 95% CI −2.26 to −0.34; 7 RCTs, 267 participants), yet not for studies with participants with high cognitive impairment (SMD −0.41, 95% CI −0.89 to 0.07; 5 RCTs, 176 participants) (Analysis 5.1); however, the evidence was very uncertain. However, subgroup analysis found no evidence of a difference in effect estimates by participant level of cognitive function (Chi² = 2.65, P = 0.10; Analysis 5.1).

5.1. Analysis.

Comparison 5: Subgroup 4: level of cognitive function: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology (end‐of‐intervention)

Subgroup 5: long‐term care facility staff involvement in therapy

A subgroup analysis for any psychological therapy implemented by LTC facility staff without a professional therapist was not suitable, given the high degree of heterogeneity in the two included studies (Tau² = 2.69; I² = 96%). Therefore, subgroup analysis was conducted only comparing effects of any psychological therapy implemented by a professional therapist with and without the involvement of LTC facility staff.

Psychological therapy that was implemented by a professional therapist with involvement from LTC facility staff may reduce depressive symptomatology at end‐of‐intervention follow‐up compared to any non‐therapy comparator, with a large effect estimate, but the evidence was very uncertain (SMD −1.02, 95% CI −1.62 to −0.43; 9 RCTs, 303 participants; Analysis 6.1). In contrast, psychological therapy implemented by a therapist without involvement from LTC staff may have little to no effect on depressive symptomatology at end‐of‐intervention follow‐up compared to any non‐therapy comparator, but the evidence was very uncertain (SMD −0.76, 95% CI −1.68 to 0.16; 5 RCTs, 176 participants).

6.1. Analysis.

Comparison 6: Subgroup 5: long‐term care facility staff involvement in therapy: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology (end‐of‐intervention)

However, subgroup analysis showed no evidence of a difference in effect estimates by LTC facility staff involvement (Chi² = 0.22, P = 0.64; Analysis 6.1).

Subgroup 6: therapeutic contact

Depressive symptomatology may be lower at end‐of‐intervention follow‐up for any psychological therapy of duration less than 8.5 weeks (dichotomized at the median; see Subgroup analysis and investigation of heterogeneity) compared to any non‐therapy comparator, with a medium effect estimate (SMD −0.74, 95% CI −1.34 to −0.14; 8 RCTs, 251 participants), and lower for any psychological therapy of duration greater than 8.5 weeks, with a large effect estimate (SMD −1.25, 95% CI −1.99 to −0.51; 9 RCTs, 368 participants) (Analysis 7.1); however, the evidence was very uncertain.

Depressive symptomatology may be lower at end‐of‐intervention follow‐up for any psychological therapy fewer than 9.5 total sessions compared to any non‐therapy comparator, with a large effect estimate (SMD −1.08, 95% CI −1.86 to −0.30; 9 RCTs, 310 participants), and lower for any psychological therapy greater than 9.5 total sessions, with a medium effect estimate (SMD −0.72, 95% CI −1.15 to −0.28; 7 RCTs, 271 participants) (Analysis 7.2); however, the evidence was very uncertain.

Depressive symptomatology may be lower at end‐of‐intervention follow‐up for any psychological therapy of total dose greater than 6.5 hours compared to any non‐therapy comparator, with a large effect estimate (SMD −1.83, 95% CI −2.93 to −0.73; 6 RCTs, 134 participants), yet not for psychological therapy of dose less than 6.5 total hours (SMD −0.69, 95% CI −1.48 to 0.11; 6 RCTs, 192 participants) (Analysis 7.3); however, the evidence was very uncertain.

Despite the above findings, subgroup analysis showed no difference by therapeutic contact based on total therapy duration in weeks (Chi² = 1.12, P = 0.29; Analysis 7.1), number of sessions (Chi² = 0.63, P = 0.43; Analysis 7.2), or total dose in hours (Chi² = 2.72, P = 0.10; Analysis 7.3).

Sensitivity analyses

Preplanned sensitivity analyses were limited by data paucity and only conducted for depressive symptomatology at end‐of‐intervention follow‐up between any psychological therapy versus any non‐therapy comparator by the following variables.

• Sensitivity analysis 1: missing data (studies excluded if data were imputed for meta‐analysis)

• Sensitivity analysis 2: treatment fidelity (studies excluded if treatment fidelity was not measured)

• Sensitivity analysis 3: bias (studies excluded if rated at high or unclear risk of bias for allocation concealment)

• Sensitivity analysis 4: attrition (studies excluded if overall dropout rate was 30% or above)

Sensitivity analysis 1: missing data

When omitting RCTs with missing data imputed (Hussian 1979; Luo 2020; Reinhardt 2014; Sales 2015; Travers 2017; Zerhusen 1991), psychological therapy may reduce depressive symptomatology at end‐of‐intervention follow‐up for any psychological therapy compared to any non‐therapy comparator, with a large effect estimate, but the evidence was very uncertain (SMD −1.11, 95% CI −1.71 to −0.51; 12 RCTs, 469 participants; Analysis 8.1). This result is consistent with the main analysis (Analysis 1.1).

8.1. Analysis.

Comparison 8: Sensitivity analysis 1: missing data: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology (end‐of‐intervention)

Sensitivity analysis 2: treatment fidelity

When including only RCTs that measured treatment fidelity (Bailey 2017; Meeks 2008; Meeks 2015; Zerhusen 1991), psychological therapy may reduce depressive symptomatology at end‐of‐intervention follow‐up for any psychological therapy compared to any non‐therapy comparator, with a large effect estimate, but the evidence was very uncertain (SMD −1.06, 95% CI −1.91 to −0.21; 4 RCTs, 158 participants; Analysis 9.1). This result is consistent with the main analysis (Analysis 1.1).

9.1. Analysis.

Comparison 9: Sensitivity analysis 2: treatment fidelity: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology (end‐of‐intervention)

Sensitivity analysis 3: bias

When including only RCTs rated at low risk of bias due to allocation concealment (Dozeman 2011; Meeks 2008; Meeks 2015; Reinhardt 2014; Travers 2017; Tsai 2008), psychological therapy may have little to no effect on depressive symptomatology at end‐of‐intervention follow‐up compared to any non‐therapy comparator, but the evidence was very uncertain (SMD −0.16, 95% CI −0.54 to 0.23; 6 RCTs, 282 participants; Analysis 10.1). This result is in contrast to the main analysis (Analysis 1.1).

10.1. Analysis.

Comparison 10: Sensitivity analysis 3: bias: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology

Sensitivity analysis 4: attrition

When omitting RCTs with a dropout rate of 30% or higher (Abraham 1992; Meeks 2008; Reinhardt 2014; Tsai 2008), psychological therapy may reduce depressive symptomatology at end‐of‐intervention follow‐up compared to any non‐therapy comparator, with a large effect estimate, but the evidence was very uncertain (SMD −1.29, 95% CI −1.83 to −0.75; 14 RCTs, 499 participants; Analysis 11.1). This result is consistent with the main analysis (Analysis 1.1).

11.1. Analysis.

Comparison 11: Sensitivity analysis 4: attrition: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology

Discussion

Summary of main results

Results for each of the key comparisons are summarized in Table 1 and additional summary of findings tables: Table 2; Table 3; Table 4.

This review comprised 19 studies, including eight published since the systematic review and meta‐analysis by Cody and Drysdale was conducted (Cody 2013).

The objective of this review was to assess the effects of psychological therapies in comparison with treatment as usual, waiting list, non‐specific attentional control, other psychological therapies, or co‐intervention for depression in older people living in LTC settings. However, due to data paucity, we only considered the effects of psychological therapies in comparison with treatment of usual, waiting list, and attentional control.

Psychological therapies versus non‐therapy comparators

Trials in this review compared psychological therapies for the treatment of depression with non‐therapy comparators. We classified psychological therapies on the basis of intervention activities employed and theoretical alignment: BT, CBT, and RT. Eighteen trials reported outcomes at end‐of‐intervention, 16 trials at short‐term follow‐up (up to three months' postintervention), eight trials at medium‐term follow‐up (three to six months' postintervention), and two trials at long‐term follow‐up (six or more months' postintervention).

Primary outcomes

Psychological therapy may reduce depressive symptomatology at end‐of‐intervention (large effect), short‐term follow‐up (large effect) and medium‐term follow‐up (small effect), but have little to no effect at long‐term follow‐up compared to any non‐therapy comparators; however, the evidence was very uncertain.

Very low‐certainty evidence showed higher odds of treatment non‐acceptability for any psychological therapies compared to any non‐therapy comparator at end‐of‐intervention. However, analyses were restricted due to limited dropout case data and imprecise reporting, and should be interpreted with caution.

Planned subgroup analyses and sensitivity analyses

Subgroup analyses were limited by data paucity and only conducted for depressive symptomatology at end‐of‐intervention follow‐up between psychological therapies and non‐therapy comparators. Subgroup analyses for participant characteristics, including level of depressive symptoms at baseline (high versus low) and level of cognitive impairment (high versus low), showed no differences in treatment effectiveness.

We conducted subgroup analyses for therapy characteristics. Type of psychological therapy (BT, CBT, and RT), involvement of facility staff in therapy (staff involvement or no staff involvement), and volume of therapeutic contact (duration of therapy, number of sessions, or total treatment dose) showed no differences in treatment effectiveness.

However, subgroup analyses showed a difference by type of non‐therapy comparator, whereby comparison of psychological therapies versus treatment as usual yielded greater effect estimates than comparison with a non‐specific attentional control condition. Psychological therapy may have little to no effect on depressive symptomatology at end‐of‐intervention follow‐up compared to non‐specific attentional control comparators, but the evidence was very uncertain.

Sensitivity analyses were limited by data paucity and only conducted for depressive symptomatology at end‐of‐intervention follow‐up between any psychological therapy and any non‐therapy comparator. When omitting RCTs with missing data imputed for meta‐analysis (Hussian 1979; Luo 2020; Reinhardt 2014; Sales 2015; Travers 2017; Zerhusen 1991), when examining only the four RCTs that assessed treatment fidelity (Bailey 2017; Meeks 2008; Meeks 2015; Zerhusen 1991), and when omitting RCTs with a dropout rate of 30% or higher (Abraham 1992; Meeks 2008; Reinhardt 2014; Tsai 2008), depressive symptomatology continued to be lower at end‐of‐intervention for psychological therapies compared to non‐therapy comparators, with similarly high estimates of effect size. However, when examining only the six RCTs at a low risk of bias relating to the allocation concealment domain (Dozeman 2011; Meeks 2008; Meeks 2015; Reinhardt 2014; Travers 2017; Tsai 2008), depressive symptoms were no longer lower at end‐of‐intervention for psychological therapies compared to non‐therapy comparators.

Unplanned sensitivity analyses

We removed one study in analyses of psychological therapies versus non‐therapy comparators for depressive symptomatology, due to concerns about the tool used to determine participant eligibility (Luo 2020), one study due to concerns about study design (Tsai 2008), and one study that was a potential outlier (Daleo 1999). In all these cases, psychological therapy continued to reduce depressive symptomology compared to non‐therapy comparators at end‐of‐intervention and short‐term follow‐up, with a large effect size.

One of these studies was also removed from analysis of medium‐term follow‐up (Tsai 2008), which resulted in psychological therapy now showing little to no effect on depressive symptomatology at this end point, although the evidence was very uncertain.

Secondary outcomes

For depression remission, physical functioning, agitation, and adverse events, data paucity precluded meta‐analysis. A single study reporting remission of MDD indicated that BT may reduce the prevalence of MDD compared to treatment as usual at end‐of‐intervention, short‐term, medium‐term, and long‐term follow‐up, but the evidence was very uncertain.

Psychological therapies may improve quality of life and psychological well‐being compared to non‐therapy comparators at short‐term follow‐up (medium effect), but the evidence was very uncertain, and this effect was not found at end‐of‐intervention or medium‐term follow‐up.

Two trials of psychological therapy indicated no evidence that psychological therapies reduce anxiety symptomatology compared to non‐therapy comparators at end‐of‐intervention, but the evidence was very uncertain.

Unplanned sensitivity analyses

We removed one study from analyses of psychological therapies versus non‐therapy comparators for quality of life and psychological well‐being due to concerns about the tool used to determine participant eligibility (Luo 2020). With this study removed, psychological therapy appeared to have little to no effect on quality of life and psychological well‐being at end‐of‐intervention and short‐term follow‐up.

Overall completeness and applicability of evidence

Studies in this review were conducted in seven countries and published across a 45‐year period from 1976 to 2020. Eligibility for inclusion in the review was not based on geographic location, or language or date of publication, in order to capture the entire evidence base. While all studies were set in LTC residential facilities for older people, there was likely considerable heterogeneity in the settings across studies, in terms of size of facility, admission criteria, staffing number, and characteristics, and the level and type of care provided. However, this information was not typically provided in sufficient detail to determine the degree to which the studies reflected LTC facilities more broadly. Seven studies were published in the 20th century, and only seven studies were published in the 2010s prior to the literature search in October 2021. This small number of recently published studies is concerning, given the continuing high rate of depression symptoms in people living in LTC, and the potential increased complexity of LTC resident populations. As an example, Australian population data indicated that the proportion of LTC residents with high frailty scores increased from 19.7% to 49.7% between 2006 and 2015, with an increase in use of medicine and worse mortality over this time period, in addition to an increase in the prevalence of depression (Inacio 2021).

Changes may also have occurred in 'treatment as usual' approaches for management of depression in LTC settings, including in the use of pharmacological treatments. This may be important in considering the applicability of findings of the review, given the majority of studies compared psychological therapy to treatment as usual. While some studies excluded people who were prescribed antidepressant medications at baseline, few study authors reported data on the use of antidepressant medications by participants during the trials, and it is unclear if the findings of the review applied equally to residents taking and not taking the medications while participating in psychological therapy.

Nearly all studies included in this review were conducted in high‐income countries, in particular, in the USA. The limited evidence may reflect gaps in provision of LTC in many low‐ and middle‐income countries (see www.who.int/activities/providing-access-to-long-term-care-for-older-people). Several studies listed the ability to communicate in the country's primary language as a criterion for inclusion in the study, and there was no information in the remaining studies about provision of access to interpreters or other strategies to facilitate participation by individuals from culturally and linguistically diverse (CALD) communities. The evidence from this review may not be applicable to people who speak a non‐primary language in their country and the degree to which the findings are relevant to CALD LTC communities is unknown.

Trial participants appeared to be representative of the LTC population in terms of age and gender, with mean ages ranging from 70.5 to 85.2 years, and women representing the majority.

We included studies of participants with and without cognitive impairment or dementia, to reflect the high rates of these conditions in LTC settings. However, most studies in the review excluded people above a particular level of impairment, although the level (e.g. cut‐off score on a cognitive screening instrument, such as the MMSE (Folstein 1975)) varied markedly from study to study. Only two studies were designed to evaluate psychological therapy for depression in people with cognitive impairment (specifically, those with mild or moderate impairment) or dementia (Bailey 2017; Travers 2017). Thus, the evidence in this review may not apply to older people with more than mild‐to‐moderate cognitive impairment or dementia. Subgroup analysis found no difference in the efficacy of psychological therapy compared to no therapy in treating depressive symptoms for people with high versus low cognitive impairment. However, this finding should be interpreted with caution: only 12/19 studies could be classified with high versus low participant cognitive impairment, and when the five studies of participants with high cognitive impairment were analysed separately, psychological therapies were not superior to non‐therapy comparators.

Studies in this review typically recruited participants based on a score over a cut‐off on a validated depression screening tool, indicating at least mild or subthreshold depression to moderate levels of depressive symptoms. A few studies only included participants with a diagnosis of a mood disorder; however, others excluded participants with a mood disorder or who presented with severe conditions, focusing on those with mild‐to‐moderate symptoms. While there was variation in symptom severity across studies, the degree to which the studies represent the proportion of residents with different levels of symptoms of depression in LTC settings is unclear. Subgroup analyses suggested no difference in the efficacy of psychological therapies for those with lower versus higher levels of depressive symptoms at baseline. Only one study in this review provided data on remission of MDD, with positive findings. The failure of researchers to routinely assess the effect of psychological therapies in treating mood disorders using structured psychiatric interview schedules is a limitation.

There were a number of common participant exclusion criteria in studies in this review, including impaired hearing, vision, or communication difficulties; a range of comorbid conditions or physical inability to participate in therapy; and recent admission to the LTC facility. Thus, the findings of the effectiveness of psychological therapies may not apply equally to LTC residents with more complex presentations. No eligible studies focused on reducing depression in newly admitted residents (e.g. those who had been residing in a facility for less than six months), despite the challenges of relocation to an LTC facility (Brownie 2014), and high rates of depression in newly admitted residents (Davison 2021a), and the effectiveness of psychological therapies for this group remains unknown.

In most of the studies in this meta‐analysis, depressive symptoms were only monitored over the short‐term (up to three months; 16 studies) or medium‐term (three to six months; eight studies). Secondary outcomes were mostly only assessed at short‐term follow‐up. The longer‐term benefits of psychological therapies for depression are less clear.

While this review considered evidence for broad outcomes of psychological therapies, eligible studies did not include data on physical functioning (including basic and instrumental activities of daily living), agitation related to dementia, and adverse effects. Therefore, effects of psychological therapies on these broader outcomes remain unknown. Only two studies reported anxiety symptomatology as an outcome, which is concerning given indications in the literature of high levels of comorbid depression and anxiety (Creighton 2018).

Quality of the evidence

The certainty of the evidence was very low for all outcomes. As per the GRADE Working Group, this indicates very little confidence in the effect estimates and that the true effects are likely to be substantially different from estimates of the effects.

Some of our estimates included fewer than five trials and thus should be interpreted with caution as per prior recommendations (Saueressig 2021). No studies were considered to have an overall low risk of bias and all studies had at least one domain at a high risk of bias. Subsequently, our confidence in all estimates were downgraded two levels due to risk of bias. A domain of particular concern, allocation concealment, was determined to be at a low risk of bias for only six studies, with all other studies rated at unclear risk of bias due to insufficient reporting. When sensitivity analyses were conducted with only these six studies, the effect of psychological therapies was no longer observed.

The quality of the trials was also limited by risk of bias related to lack of blinding of trial personnel and participants, and the absence of a prospectively published protocol or trial registration, which made it difficult to assess the potential for selective outcome reporting. There were also potential conflicts of interest, with study authors commonly involved in developing the interventions and multiple other key aspects of the conduct of the study, and few studies assessed treatment fidelity. As a result, we rated the 'other bias' domain as high risk for most studies.

Owing to considerable amounts of statistical heterogeneity, the majority of estimates were downgraded due to inconsistency. Given the aforementioned concerns with risk of bias, our estimates may have been impacted by methodological heterogeneity. Similarly, a lack of in‐depth participant demographic data may have led to clinical heterogeneity.

Despite these concerns, sensitivity analyses indicated our initial estimates were robust given few changes in magnitude or statistical significance when studies with missing data imputed were omitted. Similarly, when only including studies that examined treatment fidelity in the meta‐analysis, evidence of a difference between outcomes from psychological therapy and non‐therapy comparators remained, with large estimates of effects.

Searches

We aimed to include the most important databases and international trial registries in our searches for trials of psychological therapies for depression in LTC settings, including key databases specific to gerontology. We searched for grey literature, including dissertations and theses, and checked the reference lists of included studies and previous reviews, but were unable to rule out the possibility that some relevant studies were missed from the review.

Missing data

We contacted authors of 12 studies for information required to complete the extraction of key data and the risk of bias assessment. Attempts were typically only made to contact authors of studies published between 2002 and 2021, given the low likelihood of obtaining data from older studies directly from authors. Eventually, authors of five studies responded. The difficulty in accessing missing data resulted in several studies with unclear risk of bias ratings for some domains.

For six studies, missing data (endpoint mean or SD) were estimated using other reported data available. Only one study could not be included in the meta‐analysis due to missing data that could not be obtained or estimated (Rosen 1997).

Publication bias

We located prospective trial registrations or protocol papers for only four studies in this review. Where they were available, there was limited evidence of bias in selection of reported results. Outcomes relevant to this review were typically reported consistently in the methods and results sections by study authors. However, the risk of this bias due to selective outcome reporting was unclear for most studies due to the absence of available prospectively published information.

We found no evidence of non‐reporting bias, whereby trials were selectively published (e.g. due to P values or magnitude of results). We only located two trial registrations or published protocol papers of eligible studies that had not yet been published; the authors confirmed that both studies were ongoing in October 2021.

We only evaluated one funnel plot for asymmetry, as there were fewer than 10 studies included in other meta‐analyses (Figure 4). For depressive symptomatology at end‐of‐intervention, the funnel plot did not indicate publication bias.

Outcomes

Our primary outcomes were level of depressive symptomatology and treatment non‐acceptability, operationalized as participant dropout rate, not including dropouts that were clearly attributable to reasons unrelated to the study or intervention. While depressive symptomatology was consistently well reported using scores on validated scales, the reporting of participant dropouts was poor in many studies. Authors commonly failed to identify the time point in the trial when individual participants were no longer involved. Of the 12 studies that provided data on participant attrition, only five reported non‐zero attrition. This raises the possibility that attrition may have been under‐reported in some studies. The results of the analysis were based on a total of only 22 events and should be interpreted cautiously. The operationalization of treatment non‐acceptability as participant dropout was problematic in this review, as several authors failed to provide sufficient details about the reason for dropout to enable us to determine whether the dropouts were potentially related to the intervention or trial procedures (e.g. the participant did not find the intervention acceptable) or was unrelated to the trial (e.g. the participant died, became acutely unwell, or relocated elsewhere). This is particularly problematic for trials of psychological therapies versus treatment as usual, which were the only comparisons included in our meta‐analysis of treatment non‐acceptability. Participants receiving psychological therapies typically receive a greater number of contacts with the research/treatment team – at each therapy session as well as at data collection points – compared to those receiving treatment as usual, who are only contacted for data collection. Thus, the difference in attrition rates may simply reflect more opportunities to communicate a desire to withdraw participation, rather than indicating poor acceptance of the intervention itself. Future trials should provide a greater level of detail on participant attrition throughout each stage of the study and seek to obtain clear data on reasons for dropout. The use of dropout data to indicate treatment non‐acceptability should ideally be restricted to trials with an attention‐control condition, where participants have a similar number of opportunities to indicate they no longer wish to participate in the trial.

In this review, we combined psychological well‐being and quality of life as a single multidimensional secondary outcome in our analyses. This outcome included varied measures of life satisfaction/dissatisfaction and quality of life. Several authors view quality of life and life satisfaction as conceptually overlapping or equivalent (e.g. Frisch 2006). Some authors have described life satisfaction as one of several subjective aspects of quality of life (e.g. Sirgy 2012; Veenhoven 1991), while others have considered both quality of life and life satisfaction as aspects of an over‐arching well‐being construct (Charlemagne‐Badal 2015). Future analyses examining the effectiveness of psychological therapy on specific aspects of psychological well‐being and quality of life may be of value, particularly where quality of life measures focus on physical health ('health‐related quality of life') and vary from broader subjective measures of well‐being.

Conflicts of interest

None of the review authors were involved in any of the included trials of psychological therapies in this review. TD, SBh, CD, and LF have conducted other trials of psychological therapy that did not meet eligibility criteria for inclusion. SBh, TD, and LF were involved in eligible trials that were classified as ongoing as of October 2021.

Agreements and disagreements with other studies or reviews

Conclusions regarding the effectiveness of psychological therapies for LTC residents with depression have been limited in previous narrative and systematic reviews due to different methods employed (Bharucha 2006; Cody 2013; Gramaglia 2021; Hyer 2005; Powers 2008; Simning 2017; Snowden 2003). For example, some reviews examined a broad range of non‐pharmacological interventions and were not specific to the effectiveness of psychological therapies (Gramaglia 2021; Hyer 2005; Snowden 2003). Moreover, one review focused on interventions designed to improve well‐being, rather than depression specifically (Bharucha 2006). Therefore, drawing conclusions from prior reviews regarding the effectiveness of psychological therapies for depression is difficult. Further limitations in prior reviews include the exclusion of LTC residents with dementia (Gramaglia 2021; Simning 2017), despite the high prevalence of dementia and cognitive impairment within the population of LTC residents (Seitz 2010), and the failure to restrict eligibility criteria to studies of participants with depressive symptoms or disorders, or both.

In addressing these limitations, our review observed a large effect estimate for psychological therapies in LTC residents on depressive symptomatology compared with no therapy. Our estimate was greater than the medium effect size estimated in the only known prior meta‐analysis (Cody 2013), which included 17 studies (666 participants), including five RCTs that were also eligible for our review. Similar to the prior meta‐analysis (Cody 2013), we found that interventions based on RT and CBT may reduce depressive symptoms compared with control conditions; however, we expanded this to include a finding that BT may also be effective (Analysis 3.1). We also similarly observed that psychological therapies had no effect on depressive symptoms when compared to active comparators (i.e. non‐specific attentional controls) (Analysis 4.1). However, it is important to note the very low certainty of evidence for all comparisons conducted in our review.

The prior meta‐analysis reported a difference in effect estimates based on the involvement or lack of involvement of LTC facility staff (Cody 2013). In the current meta‐analysis, only the subgroup of psychological therapies implemented by professional therapists (including graduate students) with involvement of LTC staff reduced depressive symptoms at end‐of‐intervention compared with non‐therapy comparators (Analysis 6.1). However, we found no overall difference in effect size between therapies implemented by professional therapists with and without staff input.

While we aimed to examine the effectiveness of training LTC facility staff to implement psychological therapies, we lacked sufficient data to enable us to compare the effect estimate for these studies compared to those where the intervention was implemented by a professional therapist (with or without some level of involvement from LTC facility staff). However, one of the trials that was ongoing at the time of the search (Velasquez Reyes 2019) has since been published, with relevant results (Almeida 2022). This trial examined the effectiveness of behavioural activation delivered by trained LTC facility staff on symptoms of depression, and found the intervention had no effect on symptoms of depression (Almeida 2022). A comparison between staff‐led and therapist‐led interventions may be possible in future meta‐analyses.

Authors' conclusions

Implications for practice.

Psychological therapies may be effective for reducing depressive symptoms in older adults who live in long‐term care (LTC) facilities, with large effect sizes at end‐of‐intervention and up to three months postintervention, and a small effect size at three to six months postintervention. In addition, psychological therapies were effective for improving quality of life and psychological well‐being for up to three months postintervention, with a medium effect size, although this difference was not evident at end‐of‐intervention or medium‐term follow‐up. However, it is important to note that the evidence about the effects of psychological therapies on all outcomes is very uncertain.

Currently, although there are a number of recommendations in the literature about the benefits of psychological treatments for LTC populations (Bhar 2020), there are no clinical guidelines that stipulate which psychological interventions are effective for depression in such populations. Older adults living in LTC typically have more complex and comorbid health and mental health profiles than those living in the community (Amare 2020), and hence many not adhere as easily to psychological treatments developed for community‐dwelling older adults. However, this systematic review suggests that cognitive behavioural therapy, behavioural therapy, and reminiscence therapy may all be more effective than non‐therapy control conditions in reducing depressive symptoms amongst depressed older LTC residents. There were no differences in the effectiveness of the three types of therapies in our subgroup analysis, but given this analysis was observational in nature, further research is required directly comparing the effectiveness of these psychological therapies in LTC, and to determine which therapies work best for which patient groups. In practice, the selection of psychological treatment types may also be informed by the preferences of individual older people.

Overall, psychological therapies did not appear to result in differential effects for individuals with higher or lower levels of depression at baseline. Also, the duration and number of sessions of treatment did not appear to moderate such effects. While it is important to note that the evidence is very uncertain about the effect of these treatment characteristics, these findings may suggest that clinicians have a range of options in selecting therapies when treating clients with low or high severity of depressive symptoms.

The effect of involving LTC facility staff in implementing the interventions is unclear. There were no subgroup differences found in the effectiveness of psychological therapies that were implemented by professional therapists with and without the involvement of LTC staff. However, only the subgroup of psychological therapies implemented by professional therapists with LTC staff involvement reduced symptoms of depression compared with the control. While the involvement of facility staff may be helpful in encouraging older people to engage in selected activities day‐to‐day, outside of formal therapy sessions, additional research directly comparing the effectiveness of different ways of delivering interventions is required to inform clinical practice.

The effectiveness of psychological therapies for reducing depression for residents with high versus low cognitive impairment remains equivocal. Subgroup analysis found no difference in treatment effects by type of LTC resident in terms of high versus low cognitive impairment. However, we found the presence of an effect when pooling results across studies with residents with low cognitive impairment, but not across studies with residents with high cognitive impairment. These results suggest that the effects of psychological therapy may not be as clear in residents with high cognitive impairment and dementia, and that clinicians may need to adapt such therapies to generate more potent effects in such residents.

Further, psychological therapies may have no effect on anxious symptomatology in depressed residents at end‐of‐intervention, but the evidence is very uncertain. Data on this outcome were limited, with few researchers assessing anxiety symptoms at follow‐up, precluding any guidance to clinicians about the use of psychological therapies to reduce symptoms of anxiety in depressed LTC residents.

The superiority of psychological therapies for depression compared to non‐therapy controls was not found at longer term follow‐up (more than six months' postintervention), although one study reported that the lower rates of major depressive disorder in the psychological therapy group compared with the control group continued through to long‐term follow‐up (Meeks 2015). Psychological therapies may have no effect on quality of life and psychological well‐being at three to six months' postintervention, compared to non‐therapy controls, although the evidence is very uncertain. Together, these findings suggest that while psychological therapies may be effective for depression, quality of life and well‐being in depressed older people in the shorter term, treatment gains may attenuate in the subsequent few months. It was notable that the psychological therapies trialled in these studies were relatively brief, with most involving between two and 10 sessions. Clinicians may need to arrange for continuous booster sessions and therapeutic follow‐up activities to foster longer term stability of treatment gains; however, these suggestions are speculative and require evaluation. It is also important to note that while we considered secondary outcomes in this review, including anxiety, quality of life, and well‐being, all participants had at least mild symptoms of depression. The effect of psychological therapies evaluated in these trials on these outcomes for older people without depression at baseline (e.g. people with anxiety but not depression) may differ from the current results.

Subgroup analyses suggested that psychological therapies may be superior to treatment as usual, which typically includes assistance with activities of daily living and an opportunity to participate in a programme of activities organized by the LTC facility, and may include prescription of antidepressant medications. However, psychological therapies may not be superior to providing residents with non‐specific attention, such as participation in a discussion or education group or a social visit from a volunteer or student. Although the evidence is very uncertain, this finding is provocative as it questions whether at least part of the effects associated with psychological therapies could be attributed to the companionship, attention, and social engagement that is offered to residents in treatment, rather than the purported mechanisms of change specific to the treatment model. Non‐psychological therapy approaches can be delivered by volunteers or family members (including remotely) and could be a scaleable, low‐cost approach for use in LTC facilities. However, in this review, we did not collect data of the effect of non‐specific attention on reducing depression compared with usual care, and we lack clear evidence of the effectiveness of non‐psychological interventions such as befriending (Doyle 2021) and video calls (Noone 2020) in LTC settings.

Treatment dropout rates may be higher in the psychological therapy condition than in treatment of usual control condition, which may indicate lower treatment acceptability amongst older people. However, this conclusion was based on only 5/19 studies; there was zero attrition in seven studies and no data on attrition in the remaining seven studies, and so the evidence is very uncertain. Further, as stated above, the higher attrition rate in therapy versus treatment as usual controls may reflect residents in treatment having more opportunity than those in non‐treatment control conditions to communicate their desire to drop out. Alternatively, it is possible that some residents do experience psychological therapy protocols as burdensome and do not want to invest in such activity. Given this perspective, clinicians may need to adapt their practices to offer more engaging treatments, for example, characterized by shorter sessions and creative tasks, but this suggestion is speculative only and requires further investigation.

Implications for research.

This review collated evidence from 19 randomized controlled trials (RCTs) conducted between 1976 and 2020. Despite the number of RCTs identified and the timespan over which questions posed in our review have been studied, our capacity to draw firm conclusions on the efficacy of psychological therapies in LTC settings is hindered by several methodological limitations that warrant consideration by future researchers. The conclusions reached by this review were of very low certainty, with subgroup comparisons particularly problematic. More research is needed; most importantly, research is needed with larger sample sizes, higher quality studies, better reporting, and attention to subpopulations. Future studies should also measure outcomes over a longer follow‐up period to determine the duration of effects.

Study design: samples

In general, research on psychological therapies is bedevilled with small sample sizes, particularly when strict inclusion criteria are used. Studies were included in the current review only if participants in the intervention and control groups had a diagnosis of major depressive disorder or subthreshold depressive symptoms, based on a score over a threshold on a validated depression measure. We excluded numerous trials that did not restrict inclusion to residents with confirmed symptoms of depression. With high rates of depressive symptoms in LTC settings (Amare 2020; Seitz 2010), and indications of other aspects of poor mental health and well‐being, including anxiety symptoms, poor quality of life, and low adjustment to the LTC facility (Davison 2021a), researchers may assume that psychological therapies will be effective over the broad LTC population. However, this limits the conclusions that can be drawn on the effectiveness of psychological therapies in treating threshold or subthreshold depressive conditions.

Most studies included in the review had small sample sizes and were conducted in a few LTC facilities. We recommend that future trials be sufficiently powered to address determination of key outcomes. Larger, well‐funded studies are required to overcome the considerable challenges of conducting psychological research in the LTC setting. Challenges may arise as a result of participants' advanced age, high levels of frailty, and multimorbidity (Inacio 2021), as well as the difficulty researchers commonly experience engaging busy LTC facility staff. We also recommend studies report results separately for older residents with and without cognitive impairment and dementia, given their high prevalence in LTC settings. Research on LTC residents from culturally and linguistically diverse backgrounds is also warranted. A better understanding of which people are most likely to benefit from which interventions, based on high‐quality research is critically important to clinicians attempting to provide evidence‐based treatment in LTC.

Study design: methods

Most of the studies included in the current review were prone to bias. Researcher involvement in the evaluation of interventions they developed created potential conflicts of interest for several of the included trials. More broadly, studies of psychological therapies are likely to be at risk of bias due to the difficulty of identifying a true placebo condition and the difficulty of blinding participants to the condition. Some studies employed a non‐specific attentional control as the comparator rather than treatment as usual, which helped differentiate treatment effects from attention effects. Effect sizes were smaller for the studies that employed an attentional control, highlighting the possible benefit of spending increased time with residents (who may be deprived of attention in LTC) and providing more opportunities for residents to interact with their peers. Further research exploring outcomes and costs of different types of non‐therapy approaches may have broad societal benefits, given limitations in access to clinicians to deliver psychological therapies, and evidence of high rates of loneliness (Gardiner 2020), as well as depression (Seitz 2010).

There have been limited attempts to directly compare different types of psychological therapies in LTC, with most studies in our review comparing a single psychological therapy with a single non‐therapy control condition. We hope that our findings of the effectiveness of psychological therapy for depressed LTC residents provide an impetus for new research to advance our understanding of which interventions, delivered by which types of personnel, work for which types of older people in LTC. We currently lack clear evidence of the effectiveness of interventions led by trained facility staff rather than professional therapists, with only two heterogeneous studies eligible for inclusion in the review. With a large proportion of older people living in LTC facilities experiencing symptoms of depression but limited access to mental health clinicians (Davison 2016), evidence around the impact of training facility staff to implement therapies directly may be valuable. However, given the risks of burdening stretched LTC facility staff with additional tasks, particularly within the context of substantial workforce shortages observed during and following the COVID‐19 pandemic (Xu 2020), efforts to restructure LTC systems to enable better access to specialists in psychological therapies may prove more beneficial.

Sample attrition is an additional source of bias that is difficult to address in LTC research, given that frail participants are at a high risk of dropping out of studies for a variety of reasons. Nevertheless, the absence of sufficient evidence on longer‐term impacts of psychological therapies is a clear gap in research. Only two studies were identified for this review with a follow‐up period of six months or more. There is a need to further our understanding of longer‐term outcomes and identify when the effects of participation in a psychological therapy might wane. There may be differential maintenance of therapeutic gains in subgroups of residents. Staff working in LTC settings require information as to whether and when further treatments might be needed for some residents.

The studies included in this review paid little attention to broader treatment outcomes such as quality of life and psychological well‐being. Future research should also consider multifactorial designs that enable direct assessment of effect modifiers (e.g. type of staff involvement, or amount of therapeutic contact) on a diverse range of clinically relevant outcome measures beyond depressive symptomatology alone.

In many trials, while a large proportion of the sample had been prescribed an antidepressant medication, and continued their use, authors did not typically attempt to control statistically for the effect of this medication use. Further research is required to determine the efficacy of psychological therapy as a standalone treatment versus as an adjunct to medications. Information on the relative benefits of psychological therapy versus non‐pharmacological treatment approaches in LTC requires attention and is critical to the development of clinical guidelines.

Finally, the assessment of treatment fidelity was often omitted and should be integrated into the design of future RCTs and reported in any associated publications.

Reporting

There was inconsistent use of terminology in many studies. Some studies were labelled as quasi‐experimental but reported randomized allocation of participants to treatment conditions. In contrast, others claimed to be RCTs but failed to provide sufficient details to determine adequacy of the randomization process. Further, it was often unclear whether participants, therapists, and researchers were blinded to treatment condition.

It was often difficult or impossible to disentangle accounts of continued participation over time to assess sample attrition. Clear information on sample attrition is important during the intervention phase because it may indicate the acceptability of the treatment condition, and important postintervention to determine the reliability of the assessment of longer‐term outcomes, since attrition is likely to bias the sample of participants remaining in the trial. It is recommended that authors use alternative methods to determine acceptability of psychological therapies, such as self‐report measures.

Importance of the topic

High‐quality research on the effectiveness of psychological therapies in LTC settings may be time‐consuming and costly but is clearly needed. However, lack of support for research evaluating psychological treatments for depression in LTC settings was evident. Few studies had been published in the previous 10 years (2010 to 2020) and the review identified only two ongoing studies. Depression in LTC is highly prevalent (Seitz 2010), and there are indications that rates may have increased in recent years with the ageing population (Amare 2020). However, despite the clear need, older people living in LTC facilities in many countries receive insufficient attention from psychologists and other mental health specialists (George 2007; Ulbricht 2017), possibly because of funding constraints or a misconception about the suitability of psychological therapies for the LTC population (Davison 2016). This review of 19 studies provides evidence that psychological therapies may be effective for depression in LTC. It also demonstrates the value of the evidence base that developed incrementally over 45 years and the importance of data synthesis, but higher quality research is required to increase confidence in these findings.

History

Protocol first published: Issue 6, 2018

Appendices

Appendix 1. Searches strategies

Summary:
CCMDCTR‐Studies/References, n=1005
Ovid MEDLINE, n=1917
Ovid PsycINFO, n=652
Ovid Embase, n=1566
CLib:Trials (via CRSWeb), n = 1956
PubMed‐not‐MEDLINE (to identify journal articles not yet indexed in MEDLINE), n=256
CINAHL (Cumulative Index to Nursing & Allied Health) (EBSCOhost) (1937 onwards), n=926
Social Services Abstracts (Proquest), n=118
Sociological Abstracts (Proquest), n=1346
AgeLine (EBSCOhost), n = 460
International Clinical Trials Registry Program (ICTRP), n = 14
ClinicalTrials.gov, n = 215
Proquest Dissertations and Theses 247
Open Access Theses and Dissertations (OATD), n = 42
Digital Access to Research Theses (DART), n = 19
E‐Theses Online Service (EThOS), n = 13
OpenGrey, n = 2
Web of Science Citation Index, n = 438
Reference lists of review articles, n = 113
Total=11305
Duplicates removed, n=2263
Records to screen, n=9042
Dates of search: 8, 18, and 31 October 2021


1a. CCMDCTR‐Studies Register
#1 ((depress* or dysthymi* or “affective disorder*” or “affective symptoms” or “mood disorder*” or "low mood”)):SCO,STC,SCC (6093)
AND
#2 (behavi* or care or cogni* or counsel* or educat* or group or manage* or psycho* or *therap* or train*)):SIN,SCI,CTI (3892)
AND
#3 (“nursing home” or “residential home” or "care home" or "residential care" or "long term care" or "aged care" or "nursing facility" or “care facility” or "assisted living" or “assisted‐living") [All Fields (1261)]
(#1 and #2 and #3) (n=17)

1b. CCMDCTR‐References Register
[Condition]
#1 (depress* or “affective disorder*” or “affective symptoms” or “mood disorder*” or "low mood"):ti,ab,kw,ky,emt,mh,mc AND INREGISTER
[Setting]
#2 (gerontopsych* or institutionali* or resident* or ((care or communit* or elder* or geriatri* or retirement or nursing) near2 (home or facilit* or “long term”)) or (assist* near3 (housing or living)) or ((day or daily) near3 care)):ti,ab,kw,ky,emt,mh,mc AND INREGISTER
#3 #1 AND #2
[Population by Age Group ]
#4 (geriatri* or geronto* or (old* near2 (people or adult* or men or women)) or elder* or retire* or seniors or "late* life" or "senior citizen*" or seniors or "old old" or "very old"):ti,ab,kw,ky,emt,mh,mc AND INREGISTER
#5 ("aged 80 and over" or "frail elderly" or "very elderly") [all fields] AND INREGISTER
#6 (aged):ky,emt,mh,mc AND INREGISTER
#7 #4 OR #5 OR #6
#8 #3 AND #7 (n=705)
[Psychological therapies for the elderly]
#9 (reminiscence or "life review" or "problem sol*" or "stress manage*" or "pleasant events" or "activity scheduling" or relaxation or psychoeducat* or ("daily living" and educat*)):ti,ab,kw,ky,emt,mh,mc AND INREGISTER
#10 (psychotherapy or "cognitive behavi*" or "behavi* therap*"):ti,emt,mh AND INREGISTER
#11 #9 OR #10
#12 (geriatri* or geronto* or (old* near2 (people or adult* or men or women)) or elder* or retire* or seniors or "late* life" or "senior citizen*" or seniors or "old old" or "very old"):ti AND INREGISTER
#13 #11 AND #12 (n=347)
#14 #8 OR #13 (988)
[Key to field tags. ti:title; ab:abstract; kw:keywords; ky:other keywords; mh:MeSH headings; mc:MeSH check words; emt:EMTREE headings]

2. Ovid MEDLINE Databases
Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations and Daily <1946 to October 8 2021>
Search Strategy:
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
1 Depression/ (137404)
2 depressive disorder/ or depressive disorder, major/ or dysthymic disorder/ (108467)
3 (depress* or dysthymi*).ti,ab,kf,ot. (506782)
4 (affective disorder* or affective symptoms or mood*).ti,kf. (25565)
5 or/1‐4 (558581)
6 Homes for the Aged/ (14581)
7 Housing for the Elderly/ (1652)
8 Senior Centers/ (100)
9 ((elder* or geriatri* or old* people* or psychogeri* or retirement or senior citizen? or seniors) adj3 (centre? or center? or home? or housing or facilit* or institution? or resident*)).ti,ab,kf. (13232)
10 (institutionali* adj (elder* or old* or aged or geriatri* or psychogeri* or resident* or retire* or senior or seniors)).ti,ab,kf. (2857)
11 ((elder* or older* or geriatri* or psychogeri* or retire* or senior or seniors) adj3 (community adj dwelling?)).ti,ab,kf. (15317)
12 "homes for old*".ti,ab,kf. (3226)
13 retirement commun*.ti,ab,kf. (775)
14 or/6‐13 (43536)
15 5 and 14 (4759)
16 Institutionalization/ (5467)
17 ((institutionali* or resident* or respite) adj3 care).ti,ab,kf. (13401)
18 institutionali*.ti,kf. or (institutionali* adj (adult* or inhabitant* or patient* or resident*)).ti,ab,kf. (5757)
19 ((resident* or respite) adj3 (centre? or center? or facilit* or home?)).ti,ab,kf. (17841)
20 ((assist* or commun*) adj (dwelling? or housing or living)).ti,ab,kf. (32553)
21 *Long‐Term Care/px or Residential Facilities/ or Respite Care/ or Assisted Living Facilities/ or Group Homes/ (9731)
22 ((long term or longterm) adj care adj3 (facilit* or home? or institut* or setting)).ti,ab,kf. (8591)
23 (communit* adj (care or healthcare*) adj (facilit* or home? or institut* or setting)).ti,ab,kf. (208)
24 ((day or daily) adj3 care* adj3 (cent* or facilit* or home? or institution* or setting)).ti,ab,kf. (4142)
25 ((care or nursing) adj home?).ti,ab,kf. (37225)
26 or/16‐25 (103458)
27 Aged/ or "Aged, 80 and over"/ or Frail Elderly/ or Health Services for the Aged/ (3373024)
28 geriatric assessment/ or geriatric nursing/ or geriatric psychiatry/ (45312)
29 (aging or ageing or elder* or frail or geriatri* or geronto* or psychoger* or geropsych* or seniors or (late* adj (life* or adulthood)) or (old* adj (adult? or age? or people? or person? or citizen? or men or women or male? or female? or patient? or population?)) or old old or very old or senior citizen? or pensioner? or retired or retirement or care home? or nursing home?).ti,kf. (415615)
30 (("65" or "69" or "70" or "75" or "79" or "80" or "85" or "90" or "95") adj years).ti,ab. (244417)
31 (("65" or "69" or "70" or "75" or "79" or "80" or "85" or "90" or "95") adj2 old*).ti,ab. (101872)
32 or/27‐31 (3591237)
33 5 and 26 and 32 (7422)
34 controlled clinical trial.pt. (94695)
35 randomized controlled trial.pt. (558755)
36 (randomi#ed or randomi#ation or randomi#ing).ti,ab,kf. (713085)
37 (RCT or "at random" or (random* adj3 (administ* or allocat* or assign* or class* or control* or determine* or divide* or division or distribut* or expose* or fashion or number* or place* or recruit* or split or subsitut* or treat*))).ti,ab,kf. (602041)
38 placebo*.ab,ti,kf. (230629)
39 trial.ab,ti,kf. (676009)
40 groups.ab. (2270836)
41 (control* and (trial or study or group*) and (placebo or waitlist* or wait* list* or ((treatment or care) adj2 usual))).ti,ab,kf,hw. (215304)
42 ((single or double or triple or treble) adj2 (blind* or mask* or dummy)).ti,ab,kf. (185001)
43 double‐blind method/ or random allocation/ or single‐blind method/ (296981)
44 or/34‐43 (3354610)
45 exp animals/ not humans.sh. (4961925)
46 44 not 45 (2857202)
47 33 and 46 (1904)
48 (depress* adj2 (elder* or geriatri* or geronto* or psychoger* or geropsych* or seniors or (late* adj (life* or adulthood)) or (old* adj (adult? or age? or people? or person? or citizen? or men or women or male? or female? or patient? or population?)) or old old or very old or senior citizen? or pensioner? or retired or retirement or resident*)).ab. (10052)
49 26 and 46 and 48 (483)
50 47 or 49 (1917)

***************************

3. Ovid PsycINFO <1806 to 8 October 2021>
Search Strategy:
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
1 late life depression/ (860)
2 major depression/ or anaclitic depression/ or dysthymic disorder/ or endogenous depression/ or reactive depression/ (141380)
3 "depression (emotion)"/ (26594)
4 (depress* or dysthymi*).ti,ab,id,ot. (336455)
5 (affective disorder* or affective symptoms or mood*).ti,id. (36277)
6 or/2‐5 (357512)
7 retirement communities/ (382)
8 ((elder* or geriatri* or old* people* or psychogeri* or retirement or senior citizen? or seniors) adj3 (centre? or center? or home? or housing or facilit* or institution? or resident*)).ti,ab,id. (7065)
9 (institutionali* adj (elder* or old* or aged or geriatri* or psychogeri* or resident* or retire* or senior or seniors)).ti,ab,id. (1255)
10 ((elder* or older* or geriatri* or psychogeri* or retire* or senior or seniors) adj3 (community adj dwelling?)).ti,ab,id. (6368)
11 "homes for old*".ti,ab,id. (207)
12 retirement commun*.ti,ab,id. (688)
13 or/7‐12 (14785)
14 residential care institutions/ or nursing homes/ or assisted living/ or group homes/ 21607)
15 institutionalization/ 4009)
16 ((institutionali* or resident* or respite) adj3 care).ti,ab,id. (9541)
17 institutionali*.ti,id. or (institutionali* adj (adult* or inhabitant* or patient* or resident*)).ti,ab,id. (7188)
18 ((resident* or respite) adj3 (centre? or center? or facilit* or home?)).ti,ab,id. (12625)
19 ((assist* or commun*) adj (dwelling? or housing or living)).ti,ab,id. (15091)
20 *long term care/ (4603)
21 ((long term or longterm) adj care adj3 (facilit* or home or institut* or setting)).ti,ab,id. (2703)
22 (communit* adj (care or healthcare*) adj (facilit* or home or institut* or setting)).ti,ab,id. (88)
23 ((care or nursing) adj home?).ti,ab,id. (14917)
24 or/14‐23 (57890)
25 geriatric*.hw. (28051)
26 gerontology/ (10004)
27 geropsychology/ or gerontological counseling/ (769)
28 elder care/ (5820)
29 (aging or ageing or elder* or frail or geriatri* or geronto* or psychoger* or geropsych* or seniors or (late* adj (life* or adulthood)) or (old* adj (adult? or age? or people? or person? or citizen? or men or women or male? or female? or patient? or population?)) or old old or very old or senior citizen? or pensioner? or retired or retirement or care home? or nursing home?).ti,id. (194100)
30 (("65" or "69" or "70" or "75" or "79" or "80" or "85" or "90" or "95") adj years).ti,ab. (28050)
31 (("65" or "69" or "70" or "75" or "79" or "80" or "85" or "90" or "95") adj2 old*).ti,ab. (36526)
32 aging/ (66586)
33 or/25‐32 (242058)
34 6 and 13 (2954)
35 6 and 24 and 33 (4685)
36 depress* resident*.ti,ab,id. (55)
37 34 or 35 or 36 (5358)
38 clinical trials.sh. (12086)
39 (randomi#ed or randomi#ation or randomi#ing).ti,ab,id. (99373)
40 (RCT or at random or (random* adj3 (administ* or allocat* or assign* or class* or control* or determine* or divide* or division or distribut* or expose* or fashion or number* or place* or recruit* or split or subsitut* or treat*))).ti,ab,id. (114424)
41 (control* and (trial or study or group) and (placebo or waitlist* or wait* list* or ((treatment or care) adj2 usual))).ti,ab,id,hw. (32295)
42 ((single or double or triple or treble) adj2 (blind* or mask* or dummy)).ti,ab,id. (28008)
43 trial.ti. (34985)
44 placebo.ti,ab,id,hw. (42574)
45 treatment outcome.md. (22640)
46 treatment effectiveness evaluation.sh. (26859)
47 mental health program evaluation.sh. (2295)
48 or/38‐47 (217581)
49 37 and 48 (523)
50 1 and 48 (126)
51 (depress* adj2 (elder* or geriatri* or geronto* or psychoger* or geropsych* or seniors or (late* adj (life* or adulthood)) or (old* adj (adult? or age? or people? or person? or citizen? or men or women or male? or female? or patient? or population?)) or old old or very old or senior citizen? or pensioner? or retired or retirement)).ab. (8699)
52 24 and 48 and 51 (141)
53 49 or 50 or 52 (652)

***************************

4. Ovid Embase <1980 to 8 October 202142>
Search Strategy:
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
1 depression/ or agitated depression/ or atypical depression/ or dysthymia/ or late life depression/ or major depression/ or masked depression/ or melancholia/ or minor depression/ or "mixed anxiety and depression"/ or mourning syndrome/ or post‐stroke depression/ or postoperative depression/ or reactive depression/ or recurrent brief depression/ or subsyndromal depression/ (491,598)
2 (depress* or dysthymi*).ti,ab,kw,ot. (733,067)
3 (affective disorder* or affective symptoms or mood*).ti,kw. (37,217)
4 or/1‐3 (888,625)
5 home for the aged/ (13,280)
6 senior center/ (415)
7 institutionalized elderly/ (302)
8 ((elder* or geriatri* or old* people* or psychogeri* or retirement or senior citizen? or seniors) adj3 (centre? or center? or home? or housing or facilit* or institution? or resident*)).ti,ab,kw. (18,580)
9 (institutionali* adj (elder* or old* or aged or geriatri* or psychogeri* or resident* or retire* or senior or seniors)).ti,ab,kw. (3891)
10 ((elder* or older* or geriatri* or psychogeri* or retire* or senior or seniors) adj3 (community adj dwelling?)).ti,ab,kw. (19,537)
11 "homes for old*".ti,ab,kw. (426)
12 retirement commun*.ti,ab,kw. (957)
13 or/5‐12 (52,536)
14 4 and 13 (6,167)
15 institutional care/ or institutionalization/ (15,600)
16 ((institutionali* or resident* or respite) adj3 care).ti,ab,kw. (18,127)
17 institutionali*.ti,kw. or (institutionali* adj (adult* or inhabitant* or patient* or resident*)).ti,ab,kw. (7,451)
18 ((resident* or respite) adj3 (centre? or center? or facilit* or home?)).ti,ab,kw. (23,569)
19 ((assist* or commun*) adj (dwelling? or housing or living)).ti,ab,kw. (42,551)
20 *long term care/ and (Disease Management or Prevention or Rehabilitation or Therapy).fs. (3,867)
21 residential home/ (7,989)
22 respite care/ (1,230)
23 assisted living facility/ (2,869)
24 ((long term or longterm) adj care adj3 (facilit* or home? or institut* or setting)).ti,ab,kw. (11,487)
25 (communit* adj (care or healthcare*) adj (facilit* or home? or institut* or setting)).ti,ab,kw. (287)
26 ((day or daily) adj3 care* adj3 (cent* or facilit* or home? or institution* or setting)).ti,ab,kw. (5,361)
27 nursing home/ (60,081)
28 home care/ and elderly care/ (2,643)
29 ((care or nursing) adj home?).ti,ab,kw. (49,072)
30 or/15‐29 (172,122)
31 aged/ or frail elderly/ or very elderly/ (3,510,242)
32 geriatrics/ or gerontopsychiatry/ (47,963)
33 geriatric care/ or elderly care/ or exp geriatric nursing/ (68,304)
34 geriatric assessment/ or geriatric depression scale/ or geriatric patient/ or geriatric rehabilitation/ (54,221)
35 gerontology/ (4,169)
36 (aging or ageing or elder* or frail or geriatri* or geronto* or psychoger* or geropsych* or seniors or (late* adj (life* or adulthood)) or (old* adj (adult? or age? or people? or person? or citizen? or men or women or male? or female? or patient? or population?)) or old old or very old or senior citizen? or pensioner? or retired or retirement or care home? or nursing home?).ti,kw. (591,067)
37 (depress* adj2 (elder* or geriatri* or geronto* or psychoger* or geropsych* or seniors or (late* adj (life* or adulthood)) or (old* adj (adult? or age? or people? or person? or citizen? or men or women or male? or female? or patient? or population?)) or old old or very old or senior citizen? or pensioner? or retired or retirement or resident*)).ab. (14,830)
38 (("65" or "69" or "70" or "75" or "79" or "80" or "85" or "90" or "95") adj years).ti,ab,kw. (409,606)
39 (("65" or "69" or "70" or "75" or "79" or "80" or "85" or "90" or "95") adj2 old*).ti,ab,kw. (171,513)
40 or/31‐39 (3,890,071)
41 4 and 30 and 40 (11,332)
42 14 or 41 (12,563)
43 randomized controlled trial/ (709,575)
44 randomization.de. (94,617)
45 controlled clinical trial/ and (Disease Management or Prevention or Rehabilitation or Therapy).fs. (117,180)
46 *clinical trial/ (19,152)
47 placebo.de. (391,943)
48 placebo.ti,ab. (345,726)
49 trial.ti. (365,081)
50 (randomi#ed or randomi#ation or randomi#ing).ti,ab,kw. (1,058,733)
51 (RCT or "at random" or (random* adj3 (administ* or allocat* or assign* or class* or control* or determine* or divide* or division or distribut* or expose* or fashion or number* or place* or recruit* or split or subsitut* or treat*))).ti,ab,kw. (854,426)
52 ((singl$ or doubl$ or trebl$ or tripl$) adj3 (blind$ or mask$ or dummy)).mp. (346,928)
53 (control* and (trial or study or group) and (placebo or waitlist* or wait* list* or ((treatment or care) adj2 usual))).ti,ab,kw,hw. (308049)
54 or/43‐53 (404,363)
55 ((animal or nonhuman) not human).de. (6,521,356)
56 54 not 55 (1,760,564)
57 42 and 56 (1566)

***************************
5. Cochrane Library: Trials (via CRSWeb) (8 October 2021)
#1 MESH DESCRIPTOR Depressive Disorder AND CENTRAL:TARGET
#2 MESH DESCRIPTOR Depressive Disorder, Major AND CENTRAL:TARGET
#3 MESH DESCRIPTOR Depression AND CENTRAL:TARGET
#4 MESH DESCRIPTOR Dysthymic Disorder AND CENTRAL:TARGET
#5 depress* or dysthymi* AND CENTRAL:TARGET
#6 ("affective disorder" or "affective disorders" or "affective symptoms" or mood*):ti AND CENTRAL:TARGET
#7 #1 OR #2 OR #3 OR #4 OR #5 OR #6
#8 MESH DESCRIPTOR Homes for the Aged AND CENTRAL:TARGET
#9 MESH DESCRIPTOR Housing for the Elderly AND CENTRAL:TARGET
#10 MESH DESCRIPTOR Senior Centers AND CENTRAL:TARGET
#11 (((elder* or geriatri* or (old* next people*) or psychogeri* or retirement or senior or seniors) near3 (centre* or center* or home* or housing or facilit* or institution? or resident*))) AND CENTRAL:TARGET
#12 ((institutionali* near2 (old* or aged or resident* or retire*))) AND CENTRAL:TARGET
#13 (((elder* or older* or geriatri* or psychogeri* or retire* or senior or seniors) AND (community and dwelling*))) AND CENTRAL:TARGET
#14 ("homes for depressed old" or "homes for old people" or "homes for older") AND CENTRAL:TARGET
#15 ("retirement community" or "retirement communities") AND CENTRAL:TARGET
#16 #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15
#17 #7 AND #16
#18 MESH DESCRIPTOR Institutionalization AND CENTRAL:TARGET
#19 (((institutionali* or resident* or respite) near3 care)) AND CENTRAL:TARGET
#20 (institutionali*):ti AND CENTRAL:TARGET
#21 ((institutionali* near3 (adult* or inhabitant* or patient* or resident*))) AND CENTRAL:TARGET
#22 ((institutionali* ADJ3 (adult* or inhabitant* or patient* or resident*))) AND CENTRAL:TARGET
#23 ((assist* or commun*) near2 (dwelling* or housing or living)) AND CENTRAL:TARGET
#24 MESH DESCRIPTOR Long‐Term Care AND CENTRAL:TARGET
#25 MESH DESCRIPTOR Residential Facilities AND CENTRAL:TARGET
#26 MESH DESCRIPTOR Respite Care AND CENTRAL:TARGET
#27 MESH DESCRIPTOR Assisted Living Facilities AND CENTRAL:TARGET
#28 (longterm or "long term" or "longer term"):ti AND CENTRAL:TARGET
#29 (((longterm next care) or (long next term next care) or (longer next term next care)) and (facilit* or home or homes or institut* or setting)) AND CENTRAL:TARGET
#30 ((communit* near (care* or healthcare*) near (facilit* or home or homes or institut* or setting))) AND CENTRAL:TARGET
#31 (((day or daily) near care* near (cent* or facilit* or home or homes or institution* or setting))) AND CENTRAL:TARGET
#32 (((care or nursing) near2 (home or homes))) AND CENTRAL:TARGET
#33 #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 OR #25 OR #26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32
#34 MESH DESCRIPTOR Aged EXPLODE ALL AND CENTRAL:TARGET
#35 MESH DESCRIPTOR Health Services for the Aged EXPLODE ALL AND CENTRAL:TARGET
#36 MESH DESCRIPTOR Geriatrics EXPLODE ALL AND CENTRAL:TARGET
#37 MESH DESCRIPTOR Geriatric Assessment EXPLODE ALL AND CENTRAL:TARGET
#38 MESH DESCRIPTOR Geriatric Nursing EXPLODE ALL AND CENTRAL:TARGET
#39 (aging or ageing or elder* or frail or geriatri* or geronto* or psychoger* or geropsych* or seniors or “late life” or “later life” or “late lifers” or “later lifers” or “late adulthood” or “later adulthood” or “old adult” or “old adults” or “older adult” or “older adults” or “old age” or “older age” or “old aged” or “older aged” or “old people” or “older people” or “old peoples” or “older peoples” “old person” or “older person” or “old persons” or “older persons” or “old citizen” or “older citizen” or “old citizens” or “older citizens” or “old men” or “older men” or “old women” or “older women” or “old male” or “older male” or “old males” or “older males” or “old female” or “older female” or “old females” or “older females” or “old patient” or “older patient” or “old patients” or “older patients” or “old population” or “older population” or “old populations” or “older populations” or “old old” or “very old” or “senior citizen” or “senior citizens” or pensioner* or retired or retirement or “care home” or “care homes” or “nursing home” or “nursing homes”):ti AND CENTRAL:TARGET
#40 ((“65 years" or “69 years" or “70 years" or “75 years" or “79 years" or “80 years" or “85 years" or “90 years" or “95 years” or “sixtyfive years" or “sixty‐five years” or “sixtynine years" or “sixty‐nine years” or “seventy years" or “seventyfive years" or “seventy‐five years" or “seventynine years" or “seventy‐nine years" or “eighty years" or “eightyfive years" or “eighty‐five years” or “ninety years" or “ninetyfive years” or “ninety‐five years”)):AB AND CENTRAL:TARGET
#41 ((("65" or sixtyfive or sixty‐five or “69" or sixtynine or sixty‐nine or "70" or seventy or "75" or “seventy five” or seventy‐five or “79" or seventynine or seventy‐nine or "80" or eighty or "85" or eightyfive or eighty‐five or “90" or ninety or “95" or ninetyfive or ninety‐five) near2 old*)):ab AND CENTRAL:TARGET
#42 #34 OR #35 OR #36 OR #37 OR #38 OR #39 OR #40 OR #41
#43 #7 AND #33 AND #42
#44 (depress* near3 (aging or ageing or elder* or frail or geriatri* or geronto* or psychoger* or geropsych* or seniors or (late* next life*) or (late* next adult*) or (old* next adult*) or (old* next age*) or (old* next people*) or (old* next person*) or (old* next citizen*) or (old* next men) or (old* next women) or (old* next male*) or (old* next female*) or (old* next patient*) or (old* next population*) or (old next old) or (very next old) or (senior next citizen*) or pensioner* or retired or retirement or (care next home*) or (nursing next home*) or resident*)) AND CENTRAL:TARGET
#45 #44 AND #33
#46 #17 OR #43 OR #45

6. Sociological Abstracts (Proquest)
((MAINSUBJECT.EXACT("Depression (Psychology)") OR MAINSUBJECT.EXACT("Affective Illness") OR (depress* OR dysthymi*) OR ("affective disorder*" OR "affective symptoms" OR mood*)) AND (MAINSUBJECT.EXACT("Nursing Homes") OR MAINSUBJECT.EXACT("Residential Institutions") OR MAINSUBJECT.EXACT("Retirement Communities") OR ("nursing home*" OR "care home*" OR "home nursing" OR "home care") OR ("assisted living" OR "residential home" OR "respite care") OR ("homes for the aged" OR "homes for the elderly" OR "homes for old*" OR "retirement communit*" OR "retirement home*" OR "senior center*" OR "senior centre*") OR (“community housing” OR “community dwelling” OR "community living") OR “longterm care”) AND ((MAINSUBJECT.EXACT("Geriatrics") OR MAINSUBJECT.EXACT("Elderly") OR MAINSUBJECT.EXACT("Gerontology")) OR (aging OR ageing OR elder* OR frail OR geriatri* OR geronto* OR psychoger* OR geropsych* OR seniors OR "late* life*" OR "late* adulthood" OR "old* adult*" OR "old* age*" OR "old* people*" OR "old* person*" OR "old* citizen*" OR "old* men" OR "old* women" OR "old* male*" OR "old* female*" OR "old* patient*" OR "old* population*" OR "old old" OR "very old" OR "senior citizen*" OR pensioner* OR retired OR retirement))) AND ((RCT OR randomized OR randomised OR randomization OR randomisation OR randomizing OR randomising OR "at random") OR MAINSUBJECT.EXACT("Epidemiology") OR ("random* administ*" OR "random* allocat*" OR "allocat* randomly" OR "random* assign*" OR "assigned randomly" OR "random* control*" OR "random* divide*" OR "divided randomly" OR "random* division" OR "random* distribut*" OR "distributed randomly" OR "random* fashion" OR "random* number*" OR "random* place*" OR "random recruit*" OR "recruited randomly" OR "random* split*" OR "split randomly" OR "random* subsitut*") OR ti(random* OR trial)) 1346

7.Social Services Abstracts (Proquest)
((MAINSUBJECT.EXACT("Depression (Psychology)") OR MAINSUBJECT.EXACT("Affective Illness") OR (depress* OR dysthymi*) OR ("affective disorder*" OR "affective symptoms" OR mood*)) AND (MAINSUBJECT.EXACT("Nursing Homes") OR MAINSUBJECT.EXACT("Residential Institutions") OR MAINSUBJECT.EXACT("Retirement Communities") OR ("nursing home*" OR "care home*" OR "home nursing" OR "home care") OR ("assisted living" OR "residential home" OR "respite care") OR ("homes for the aged" OR "homes for the elderly" OR "homes for old*" OR "retirement communit*" OR "retirement home*" OR "senior center*" OR "senior centre*") OR (“community housing” OR “community dwelling” OR "community living") OR “longterm care”) AND ((MAINSUBJECT.EXACT("Geriatrics") OR MAINSUBJECT.EXACT("Elderly") OR MAINSUBJECT.EXACT("Gerontology")) OR (aging OR ageing OR elder* OR frail OR geriatri* OR geronto* OR psychoger* OR geropsych* OR seniors OR "late* life*" OR "late* adulthood" OR "old* adult*" OR "old* age*" OR "old* people*" OR "old* person*" OR "old* citizen*" OR "old* men" OR "old* women" OR "old* male*" OR "old* female*" OR "old* patient*" OR "old* population*" OR "old old" OR "very old" OR "senior citizen*" OR pensioner* OR retired OR retirement))) AND ((RCT OR randomized OR randomised OR randomization OR randomisation OR randomizing OR randomising OR "at random") OR subject("Clinical trials”) OR ("random* administ*" OR "random* allocat*" OR "allocat* randomly" OR "random* assign*" OR "assigned randomly" OR "random* control*" OR "random* divide*" OR "divided randomly" OR "random* division" OR "random* distribut*" OR "distributed randomly" OR "random* fashion" OR "random* number*" OR "random* place*" OR "random recruit*" OR "recruited randomly" OR "random* split*" OR "split randomly" OR "random* subsitut*") OR ti(random* OR trial)) AND noft(random*) OR noft(trial) 116

8. PubMed NOT MEDLINE (8 October 2021)
(depress* AND (“Homes for the Aged” OR “housing for the elderly” OR “retirement community” OR “retirement communities” OR “community dwelling” OR “senior centre” OR “senior centres” OR “senior center” OR “senior centers” OR “care home” OR “care homes” or “nursing home” OR “nursing homes” OR "homes for old” OR "homes for older” OR "assisted living” OR “longterm care” OR “long term care” OR "aged care" OR "nursing facility" OR “care facility” OR “respite care” OR institutionali* OR resident*) AND (RCT OR random* OR trial) AND (pubmednotmedline[sb] OR inprocess[sb] or publisher[sb])) n=256

9.EBSCOhost CINAHL (31 October 2021)
S24 (S5 AND S22 AND S23) 926
S23 (S6 OR S7 OR S8 OR S9 OR S10 OR S11 OR S12 OR S13 OR S14 OR S15) 72,013
S22 (S16 OR S17 OR S18 OR S19 OR S20 OR S21) 559,438
S21 "Randomized Controlled Trials"152,559
S20 TI ( ((single or double or triple or treble) N2 (blind* or mask* or dummy)) ) OR AB ( ((single or double or triple or treble) N2 (blind* or mask* or dummy)) )54,139
S19 TI ( (control* and (trial or study or group*) and (placebo or waitlist* or wait* list* or ((treatment or care) N2 usual))) ) OR AB ( (control* and (trial or study or group*) and (placebo or waitlist* or wait* list* or ((treatment or care) N2 usual))) )58,961
S18 TI trial OR AB trial 381,630
S17 TI ( (RCT or "at random" or (random* N3 (administ* or allocat* or assign* or class* or control* or determine* or divide* or division or distribut* or expose* or fashion or number* or place* or recruit* or split or subsitut* or treat*))) ) OR AB ( (RCT or "at random" or (random* N3 (administ* or allocat* or assign* or class* or control* or determine* or divide* or division or distribut* or expose* or fashion or number* or place* or recruit* or split or subsitut* or treat*))) )178,882
S16 TI ( randomized or randomised or randomization or randomisation or randomizing or randomising ) OR AB ( randomized or randomised or randomization or randomisation or randomizing or randomising )278,937
S15 ("Institutionalization")4,284
S14 TI "retirement commun*" OR AB "retirement commun*"693
S13 TI "homes for old*" OR AB "homes for old*"303
S12 TI ( ((elder* or older* or geriatri* or psychogeri* or retire* or senior or seniors) N3 (community OR dwelling*)) ) OR AB ( ((elder* or older* or geriatri* or psychogeri* or retire* or senior or seniors) N3 (community OR dwelling*)) )18,098
S11 TI ( (institutionali* N1 (elder* or old* or aged or geriatri* or psychogeri* or resident* or retire* or senior or seniors)) ) OR AB ( (institutionali* N1 (elder* or old* or aged or geriatri* or psychogeri* or resident* or retire* or senior or seniors)) )2,058
S10 TI ( ((elder* or geriatri* or "old* people*" or psychogeri* or retirement or senior or seniors) N3 (centre* or center* or home* or housing or facilit* or institution* or resident*)) ) OR AB ( ((elder* or geriatri* or "old* people*" or psychogeri* or retirement or senior or seniors) N3 (centre* or center* or home* or housing or facilit* or institution* or resident*)) )14,702
S9 ("Senior Centers")508
S8 ("Housing for the Elderly")61
S7 ("Gerontologic Nursing")13,432
S6 ( "Nursing Homes")29,850
S5 (S1 OR S2 OR S3 OR S4) 209,856
S4 TI ( "affective disorder*" or "affective symptoms" or mood* ) OR TI ( "affective disorder*" or "affective symptoms" or mood* )8,039
S3 TI ( depress* or dysthymi* ) OR AB ( depress* or dysthymi* )163,454
S2 ("Affective Disorders")9,951
S1 ( "Hamilton Rating Scale for Depression") OR ("Geriatric Depression Scale") OR ("Self‐Rating Depression Scale") OR (MH "Depression, Reactive") OR ("Depression")185,654

10.EBSCO AgeLine (31 October 2021)
S24 (S5 AND S22 AND S23) 460 
S23 (S6 OR S7 OR S8 OR S9 OR S10 OR S11 OR S12 OR S13 OR S14 OR S15) 34,538
S22 (S16 OR S17 OR S18 OR S19 OR S20 OR S21) 11,948
S21 "Randomized Controlled Trials" 2,856
S20 TI ( ((single or double or triple or treble) N2 (blind* or mask* or dummy)) ) OR AB ( ((single or double or triple or treble) N2 (blind* or mask* or dummy)) ) 650
S19 TI ( (control* and (trial or study or group*) and (placebo or waitlist* or wait* list* or ((treatment or care) N2 usual))) ) OR AB ( (control* and (trial or study or group*) and (placebo or waitlist* or wait* list* or ((treatment or care) N2 usual))) )1,282
S18 TI trial OR AB trial 6,632
S17 TI ( (RCT or "at random" or (random* N3 (administ* or allocat* or assign* or class* or control* or determine* or divide* or division or distribut* or expose* or fashion or number* or place* or recruit* or split or subsitut* or treat*))) ) OR AB ( (RCT or "at random" or (random* N3 (administ* or allocat* or assign* or class* or control* or determine* or divide* or division or distribut* or expose* or fashion or number* or place* or recruit* or split or subsitut* or treat*))) ) 7,371
S16 TI ( randomized or randomised or randomization or randomisation or randomizing or randomising ) OR AB ( randomized or randomised or randomization or randomisation or randomizing or randomising ) 4,524
S15 "Institutionalization"2,829
S14 TI "retirement commun*" OR AB "retirement commun*" 1728
S13 TI "homes for old*" OR AB "homes for old*" 189
S12 TI ( ((elder* or older* or geriatri* or psychogeri* or retire* or senior or seniors) N3 (community OR dwelling*)) ) OR AB ( ((elder* or older* or geriatri* or psychogeri* or retire* or senior or seniors) N3 (community OR dwelling*)) ) 11,109
S11 TI ( (institutionali* N1 (elder* or old* or aged or geriatri* or psychogeri* or resident* or retire* or senior or seniors)) ) OR AB ( (institutionali* N1 (elder* or old* or aged or geriatri* or psychogeri* or resident* or retire* or senior or seniors)) )1,217
S10 TI ( ((elder* or geriatri* or "old* people*" or psychogeri* or retirement or senior or seniors) N3 (centre* or center* or home* or housing or facilit* or institution* or resident*)) ) OR AB ( ((elder* or geriatri* or "old* people*" or psychogeri* or retirement or senior or seniors) N3 (centre* or center* or home* or housing or facilit* or institution* or resident*)) ) 11,598
S9 "Senior Centers" 1228
S8 "Housing for the Elderly" 848
S7 "Gerontologic Nursing" 13
S6 "Nursing Homes" 15.040
S5 (S1 OR S2 OR S3 OR S4) 18,415
S4 TI ( "affective disorder*" or "affective symptoms" or mood* ) OR TI ( "affective disorder*" or "affective symptoms" or mood* ) 400
S3 TI ( depress* or dysthymi* ) OR AB ( depress* or dysthymi* )17,342
S2 "Affective Disorders" 385
S1 ("Hamilton Rating Scale for Depression") OR ("Geriatric Depression Scale") OR ("Self‐Rating Depression Scale") OR ("Depression, Reactive") OR ("Depression")16,962

11.ICTRP (18 October 2021)
1. depression AND “nursing home” [14]
2. depression AND “residential home” [0]
3. depression AND “care home” AND elderly [0]
4. depression AND “residential care” AND elderly [0]
5. depression AND “assisted living” AND elderly [0]
Total: 14

12. ClinicalTrials.gov (18 October 2021)
Advanced search
Condition or disease: Depression in Old Age
Total: 215

13. Proquest Dissertations and Theses (18 October 2021)
1. noft(“older adult*” OR elderly OR aged OR geriatric OR “older people”) 
2. noft(depression) 
3. noft(“long term care” OR “aged care” OR “retirement home” OR “nursing home” OR “residential home” OR “care home” OR “residential care” OR “care facility” OR “assisted living” OR “elder care” 
4. noft(intervention* OR treatment* OR therapy OR psychotherapy OR counsel* OR outcome* 
5. 1 AND 2 AND 3 AND 4 
Total: 247

14. Open Access Theses and Dissertations (18 October 2021)
1. depression OR depressive OR depressed
. home OR residential OR resident OR facility OR assisted
3. old OR elderly OR geriatric
4. randomized OR randomised OR RCT OR randomly
5. Add all to Search Builder with AND
Total: 42

15. DART (18 October 2021)
(depression or depressive or depressed) and (home or residential or resident or facility or assisted) and (old or elderly or geriatric) and (randomized or randomised or RCT or randomly)
Total: 19

16. EThOS (18 October 2021)
1. depression AND nursing home [9]
2. depression AND residential home [3]
3. depression AND care home AND elderly [1]
4. depression AND residential care AND elderly [0]
5. depression AND assisted living AND elderly [0]
Total: 13

17. OpenGrey (18 October 2021)
(“older adult” OR elderly OR aged OR geriatric OR “older people”) AND depression AND (“long term care” OR “aged care” OR “retirement home” OR “nursing home” OR “residential home” OR “care home” OR “residential care” OR “care facility” OR “assisted living” OR “elder care”) AND (intervention* OR treatment* OR therapy OR psychotherapy OR counsel* OR outcome*)
Total: 2

18. Web of Science Citation Index/Core Collection (18 October 2021)
((TS=(Depression)) AND ALL=("aged care" OR "residential care")) AND ALL=(Intervention OR program OR therapy) AND (DOP=1900‐01‐01/2021‐10‐18) (n=438)

Data and analyses

Comparison 1. Any psychological therapy versus any non‐therapy comparator.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Depressive symptomatology	18		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.1.1 End‐of‐intervention	18	644	Std. Mean Difference (IV, Random, 95% CI)	‐1.04 [‐1.49, ‐0.58]
1.1.2 Short‐term	16	512	Std. Mean Difference (IV, Random, 95% CI)	‐1.03 [‐1.49, ‐0.56]
1.1.3 Medium‐term	8	355	Std. Mean Difference (IV, Random, 95% CI)	‐0.43 [‐0.81, ‐0.06]
1.1.4 Long‐term	2	92	Std. Mean Difference (IV, Random, 95% CI)	‐0.16 [‐0.58, 0.27]
1.2 Treatment non‐acceptability	12	585	Odds Ratio (M‐H, Random, 95% CI)	3.44 [1.19, 9.93]
1.2.1 End‐of‐intervention	12	585	Odds Ratio (M‐H, Random, 95% CI)	3.44 [1.19, 9.93]
1.3 Quality of life and psychological well‐being	6		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.3.1 End‐of‐intervention	6	195	Std. Mean Difference (IV, Random, 95% CI)	0.40 [‐0.02, 0.82]
1.3.2 Short‐term	5	170	Std. Mean Difference (IV, Random, 95% CI)	0.51 [0.19, 0.82]
1.3.3 Medium‐term	2	68	Std. Mean Difference (IV, Random, 95% CI)	0.26 [‐0.95, 1.48]
1.4 Anxious symptomatology	2	115	Std. Mean Difference (IV, Random, 95% CI)	‐0.68 [‐2.50, 1.14]
1.4.1 End‐of‐intervention	2	115	Std. Mean Difference (IV, Random, 95% CI)	‐0.68 [‐2.50, 1.14]

Comparison 2. Subgroup 1: baseline depression severity: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Depressive symptomatology (end‐of‐intervention)	18	644	Std. Mean Difference (IV, Random, 95% CI)	‐1.04 [‐1.49, ‐0.58]
2.1.1 > 54.8 points on a standardised 100‐point scale	9	273	Std. Mean Difference (IV, Random, 95% CI)	‐1.31 [‐2.08, ‐0.53]
2.1.2 < 54.8 points on a standardised 100‐point scale	9	371	Std. Mean Difference (IV, Random, 95% CI)	‐0.81 [‐1.38, ‐0.25]

Comparison 3. Subgroup 2: type of psychological therapy: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Depressive symptomatology (end‐of‐intervention)	18	631	Std. Mean Difference (IV, Random, 95% CI)	‐1.02 [‐1.47, ‐0.57]
3.1.1 CBT	5	125	Std. Mean Difference (IV, Random, 95% CI)	‐1.02 [‐1.99, ‐0.04]
3.1.2 BT	8	303	Std. Mean Difference (IV, Random, 95% CI)	‐0.69 [‐1.21, ‐0.18]
3.1.3 RT	6	203	Std. Mean Difference (IV, Random, 95% CI)	‐1.65 [‐2.83, ‐0.47]

Comparison 4. Subgroup 3: type of non‐therapy comparator: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Depressive symptomatology (end‐of‐intervention)	18	638	Std. Mean Difference (IV, Random, 95% CI)	‐1.05 [‐1.51, ‐0.58]
4.1.1 Non‐specific attentional control	6	208	Std. Mean Difference (IV, Random, 95% CI)	‐0.30 [‐0.83, 0.23]
4.1.2 Treatment as usual	12	430	Std. Mean Difference (IV, Random, 95% CI)	‐1.47 [‐2.10, ‐0.84]

Comparison 5. Subgroup 4: level of cognitive function: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Depressive symptomatology (end‐of‐intervention)	12	443	Std. Mean Difference (IV, Random, 95% CI)	‐0.81 [‐1.35, ‐0.26]
5.1.1 High cognitive impairment	5	176	Std. Mean Difference (IV, Random, 95% CI)	‐0.41 [‐0.89, 0.07]
5.1.2 Low cognitive impairment	7	267	Std. Mean Difference (IV, Random, 95% CI)	‐1.30 [‐2.26, ‐0.34]

Comparison 6. Subgroup 5: long‐term care facility staff involvement in therapy: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
6.1 Depressive symptomatology (end‐of‐intervention)	14	470	Std. Mean Difference (IV, Random, 95% CI)	‐0.93 [‐1.43, ‐0.43]
6.1.1 Professional therapist without LTC staff involvement	5	167	Std. Mean Difference (IV, Random, 95% CI)	‐0.76 [‐1.68, 0.16]
6.1.2 Professional therapist with LTC staff involvement	9	303	Std. Mean Difference (IV, Random, 95% CI)	‐1.02 [‐1.62, ‐0.43]

Comparison 7. Subgroup 6: therapeutic contact: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
7.1 Depressive symptomatology, therapeutic contact (weeks)	17	619	Std. Mean Difference (IV, Random, 95% CI)	‐0.98 [‐1.45, ‐0.52]
7.1.1 < 8.5 weeks	8	251	Std. Mean Difference (IV, Random, 95% CI)	‐0.74 [‐1.34, ‐0.14]
7.1.2 > 8.5 weeks	9	368	Std. Mean Difference (IV, Random, 95% CI)	‐1.25 [‐1.99, ‐0.51]
7.2 Depressive symptomatology, therapeutic contact (number of sessions)	16	581	Std. Mean Difference (IV, Random, 95% CI)	‐0.87 [‐1.31, ‐0.42]
7.2.1 < 9.5 total sessions	9	310	Std. Mean Difference (IV, Random, 95% CI)	‐1.08 [‐1.86, ‐0.30]
7.2.2 > 9.5 total sessions	7	271	Std. Mean Difference (IV, Random, 95% CI)	‐0.72 [‐1.15, ‐0.28]
7.3 Depressive symptomatology, therapeutic contact (total treatment dose)	12	326	Std. Mean Difference (IV, Random, 95% CI)	‐1.20 [‐1.87, ‐0.54]
7.3.1 < 6.5 total hours	6	192	Std. Mean Difference (IV, Random, 95% CI)	‐0.69 [‐1.48, 0.11]
7.3.2 > 6.5 total hours	6	134	Std. Mean Difference (IV, Random, 95% CI)	‐1.83 [‐2.93, ‐0.73]

Comparison 8. Sensitivity analysis 1: missing data: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
8.1 Depressive symptomatology (end‐of‐intervention)	12	469	Std. Mean Difference (IV, Random, 95% CI)	‐1.11 [‐1.71, ‐0.51]

Comparison 9. Sensitivity analysis 2: treatment fidelity: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
9.1 Depressive symptomatology (end‐of‐intervention)	4	158	Std. Mean Difference (IV, Random, 95% CI)	‐1.06 [‐1.91, ‐0.21]

Comparison 10. Sensitivity analysis 3: bias: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
10.1 Depressive symptomatology	6	282	Std. Mean Difference (IV, Random, 95% CI)	‐0.16 [‐0.54, 0.23]

Comparison 11. Sensitivity analysis 4: attrition: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
11.1 Depressive symptomatology	14	499	Std. Mean Difference (IV, Random, 95% CI)	‐1.29 [‐1.83, ‐0.75]

Comparison 12. Sensitivity analysis 5: omit Luo 2020: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
12.1 Depressive symptomatology	17		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
12.1.1 End‐of‐intervention	17	594	Std. Mean Difference (IV, Random, 95% CI)	‐1.07 [‐1.57, ‐0.58]
12.1.2 Short‐term	15	462	Std. Mean Difference (IV, Random, 95% CI)	‐1.07 [‐1.58, ‐0.56]
12.2 Quality of life and psychological well‐being	5		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
12.2.1 End‐of‐intervention	5	145	Std. Mean Difference (IV, Random, 95% CI)	0.25 [‐0.17, 0.68]
12.2.2 Short‐term	4	120	Std. Mean Difference (IV, Random, 95% CI)	0.35 [‐0.02, 0.72]

Comparison 13. Sensitivity analysis 6: omit Tsai 2008: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
13.1 Depressive symptomatology	17		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
13.1.1 End‐of‐intervention	17	581	Std. Mean Difference (IV, Random, 95% CI)	‐1.13 [‐1.59, ‐0.67]
13.1.2 Short‐term	15	449	Std. Mean Difference (IV, Random, 95% CI)	‐1.12 [‐1.57, ‐0.68]
13.1.3 Medium‐term	7	292	Std. Mean Difference (IV, Random, 95% CI)	‐0.44 [‐0.90, 0.02]

Comparison 14. Sensitivity analysis 7: omit Daleo 1999: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
14.1 Depressive symptomatology	17		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
14.1.1 End‐of‐intervention	17	619	Std. Mean Difference (IV, Random, 95% CI)	‐0.86 [‐1.26, ‐0.46]
14.1.2 Short‐term	15	487	Std. Mean Difference (IV, Random, 95% CI)	‐0.84 [‐1.22, ‐0.45]

Comparison 15. Funnel plot: psychological therapies versus non‐therapy comparators.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
15.1 Depressive symptomatology	18	644	Std. Mean Difference (IV, Random, 95% CI)	‐1.04 [‐1.49, ‐0.58]

15.1. Analysis.

Comparison 15: Funnel plot: psychological therapies versus non‐therapy comparators, Outcome 1: Depressive symptomatology

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Abraham 1992.

Study characteristics
Methods	Study design: cluster randomized controlled trial Study grouping: parallel group Number of facilities: 7
Participants	Number of participants: 76 Baseline characteristics Cognitive behavioural therapy Age: not specified Gender: not specified Cognitive function (3MS): mean 73.57 (SD 15.15) Focused visual imagery Age: not specified Gender: not specified Cognitive function (3MS): mean 81.07 (SD 9.88) Education‐discussion Age: not specified Gender: not specified Cognitive function (3MS): mean 70.88 (SD 14.74) Overall Age: range 71–97 years; mean 83.38 (SD 6.13) years Gender: not specified Cognition (3MS): not specified Inclusion criteria: GDS ≥ 11 or ≥ 10 with clinical evidence of depressive symptomology; hearing sufficient to participate in group sessions; adequate vision; verbal and comprehension skills sufficient to participate in group sessions Exclusion criteria: major cognitive impairment that would make group participation impossible; use of antidepressant medication; history of endogenous depression Pretreatment: no group differences on depression, hopelessness, dissatisfaction with life, cognitive functioning, or demographic data
Interventions	Intervention characteristics Cognitive behavioural therapy Length of treatment: 24 weeks Mode of treatment: group therapy Number of sessions: 24 Therapist(s): clinical nurse specialist with credentials and experience in the treatment modality Co‐interventions: none (checked for medications) Frequency of sessions: weekly Session length: not specified Focused visual imagery Length of treatment: 24 weeks Mode of treatment: group therapy Number of sessions: 24 Therapist(s): clinical nurse specialist with credentials and experience in the treatment modality Co‐interventions: none (checked for medications) Frequency of sessions: weekly Session length: not specified Education‐discussion Length of treatment: 24 weeks Mode of treatment: group therapy Number of sessions: not specified Therapist(s): clinical nurse specialist, trained to not conduct formal group therapy Co‐interventions: none (checked for medications) Frequency of sessions: not specified Session length: not specified
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: GDS‐30 Range: 0–30 Direction: lower is better Data value: end point Life dissatisfaction Outcome type: continuous outcome Scale: Life Satisfaction Index – short version (inverted) Range: not reported Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness
Identification	Sponsorship source: funded by National Center for Nursing Research, National Institutes of Health, and National Institute of Mental Health. Country: USA Setting: nursing homes Comments: NA Author's name: Ivo L Abraham Institution: University of Virginia Email: not provided Address: Center of Aging and Health, the University of Virginia, Charlottesville, Virginia, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Treatment condition were randomly assigned to 7 nursing homes". Judgement comment: insufficient information about the sequence generation process to permit judgement.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: insufficient information about concealment method to permit judgement.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "Once subjects had been assigned to treatment conditions, they were contacted by project clinicians and group sessions were scheduled. Three clinical nurse specialists with credentials and experience in the respective treatment modalities conducted the 24‐week intervention". Quote (from Journal of Women & Aging publication: "… clinicians were blinded as to the interviewer assigned to the subject"). Judgement comment: steps were possibly taken to blind clinicians. No information provided on blinding of participants. However, given the control group consisted of an education discussion group, it is possible that participants considered this an active intervention, and were blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Four trained interviewers blinded to the treatment conditions to which subjects were assigned, were randomly assigned to the nursing homes". Judgement comment: blinding of outcome assessment was done.
Incomplete outcome data (attrition bias) All outcomes	High risk	Judgement comment: 76 participants randomized but only 42 completed the intervention and the follow‐up data collection. The cognitive behavioural condition lost 11/30 (37%) participants, and the comparison control condition 9/17 (53%). Imbalance in attrition amongst groups, although author suggested attrition was due to death and illness.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no published protocol identified. All outcome measures listed in methods were reported in results.
Other bias	High risk	Judgement comment: no information on assessment of treatment fidelity. A few clusters (7 clusters allocated to 3 conditions) may have introduced bias.

Bailey 2017.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 5
Participants	Number of participants: 51 Baseline characteristics QAR‐Depression (Question‐Asking‐Reading) Age: mean 84.35 (SD 7.68) years Gender: 92% women Cognitive function (MMSE): mean 16.56 (SD 4.59) Usual nursing home activities Age: mean 83.92 (SD 9.18) years Gender: 88% women Cognitive function (MMSE): mean 16.40 (SD 4.25) Overall Age: not specified Gender: 90% women Cognitive function (MMSE): not specified Inclusion criteria: aged ≥ 60 years; mild‐to‐moderate cognitive impairment (MMSE score 10–24); symptoms of depression (GDS score ≥ 8) Exclusion criteria: none Pretreatment: multivariate analysis (Hotelling's T) revealed no significant differences between the experimental (n = 26) and control conditions (n = 25) on any measures at baseline.
Interventions	Intervention characteristics QAR‐Depression Length of treatment: 6 weeks Mode of treatment: group therapy Session length: 30 minutes Number of sessions: 12 Therapists: 2 upper‐level graduate students in clinical psychology and 1 PhD psychologist Co‐interventions: not specified Frequency of sessions: twice per week Usual nursing home activities Length of treatment: 6 weeks Mode of treatment: group therapy Session length: not specified Number of sessions: 12 Therapists: none – usual staff co‐ordinating activities Co‐interventions: not specified Frequency of sessions: twice per week
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: Cornell Scale for Depression in Dementia Range: 0–38 Direction: lower is better Data value: end point Depressive symptomatology Outcome type: continuous outcome Scale: GDS‐30 items Range: 0–30 Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Direction: lower is better Notes: did not include attrition related to death, relocation, or illness Quality of life Outcome type: continuous outcome Scale: QOL‐AD Range: 13–52 Direction: higher is better Data value: end point
Identification	Sponsorship source: National Institute on Aging Country: USA Setting: nursing home Comments: brief report Author's name: Forrest Scogin Institution: The University of Alabama Email: fscogin@ua.edu Address: Department of Psychology, The University of Alabama, Tuscaloosa, Alabama, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Whole units or stations within each of the five nursing homes were randomly assigned to either the experimental or control condition (as opposed to randomly assigning individuals)". Judgement comment: insufficient information about the sequence generation process to permit judgement.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: insufficient information about concealment method to permit judgement.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "two upper‐level graduate students in clinical psychology and one PhD psychologist served as group leaders". Judgement comment: it is unlikely that the personnel were blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "Multiple methods of data collection (e.g., behavioral observation, self‐ and caregiver report, clinician interview) were used to assess treatment outcome". "Staff and research assistants were not blinded to the study design due to clear differences between experimental and control groups within each nursing home". Judgement comment: blinding of outcome assessment was not done. To complete the QOL‐AD and the Barthel scale, staff (not blinded) filled the role of caregivers.
Incomplete outcome data (attrition bias) All outcomes	High risk	Judgement comment: 51 participants randomized with 26 in intervention group and 25 in control group. 4 dropouts in intervention group including 1 death and 3 failing to attend group sessions. The latter was likely to have affected the effects of the treatment. Repeated measure analysis was conducted to examine treatment outcomes. Missing data were not addressed.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no published protocol identified. All outcome measures listed in methods were reported in results.
Other bias	Low risk	Quote: "Ten percent (three) of each group leader's sessions were randomly selected and evaluated by the first and second author using the Treatment Fidelity Monitoring Checklist". "Mean adherence for each group leader was well within acceptable limits". Judgement comment: treatment fidelity was assessed, with high adherence rates reported. The small number of clusters (5 clusters allocated to 2 conditions), and no information on differences between nursing homes may have introduced bias; however, there were no differences between conditions on scores on baseline measures.

Daleo 1999.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 2
Participants	Number of participants: 25 Baseline characteristics Group reminiscence therapy Age: 69–89 years Gender: 69.2% women Cognitive function: not specified Treatment as usual Age: 66–87 years Gender: 75.0% women Cognitive function: not specified Overall Age: 66–89 years Gender: 72.0% women Cognitive function: sample described as 'cognitively intact' but no measures reported. Inclusion criteria: mood disturbance on Resident Assessment Protocol from e minimum data set – symptoms of depression, anxiety, loneliness, or sad mood; GDS‐15 score ≥ 6; oriented to self and others; English speaking; admitted to the LTC facility for at least the next 9 weeks Exclusion criteria: residents receiving antipsychotic medications Pretreatment: no group differences in baseline measures
Interventions	Intervention characteristics Group reminiscence therapy Length of treatment: 9 weeks Mode of treatment: group therapy Length of sessions: 45 minutes Number of sessions: 9 Therapists: PhD student and facility social service director Frequency of sessions: once weekly Co‐interventions: not specified Treatment as usual control Length of treatment: 9 weeks Mode of treatment: N/A Length of sessions: N/A Number of sessions: N/A Therapists: Activity director's assistant Frequency of sessions: N/A Co‐interventions: N/A
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: GDS‐15 Range: 0–15 Direction: lower is better Data value: end point Depressive symptomatology Outcome type: continuous outcome Scale: RSQ Range: 12–36 Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness
Identification	Sponsorship source: not mentioned Country: USA Setting: nursing home Comments: student clinical dissertation Author's name: David V Daleo Institution: California School of Professional Psychology, San Diego, USA Email: not provided Address: not provided
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "All 40 referred residents were randomly assigned to one of two short‐term reminiscent therapy groups for the treatment of depression, or to one of two no‐treatment control groups prior to screening". "From the original forty referrals, twenty‐five residents met the screening criteria for inclusion in this study or chose to participate". Judgement comment: insufficient information about the sequence generation process to permit judgement.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: insufficient information about concealment method to permit judgement.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: interventions were implemented by the researcher and a co‐therapist. Each social service director served as group co‐therapist for their own facility. It is unlikely that participants and personnel were blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "At the conclusion of the initial individual interview, at the fourth session, and at the final group session, each resident was asked to fill out the Geriatric Depression Scale‐Short Form"; "The Residence Status Questionnaire … was administered at the beginning, middle (during the fourth session), and at the completion of treatment as a means of assessing whether or not each resident's nurse noticed any change in the resident's level of depression during the reminiscent therapy group". Judgement comment: outcome measures were administered within the group therapy sessions (or at approximately the same time for those in the control condition) by personnel who were unlikely blinded to condition.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "In the event that a resident missed more than three full sessions, his/her data would not have been used in the research. However, no residents missed more than two group sessions". Judgement comment: no participants were excluded from analyses. Of the 25 participants (13 in treatment, 12 in control), only 1 participant withdrew from a treatment condition, due to discharge from the LTC facility, which is unlikely to have affected the treatment effects.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no published protocol identified. All outcome measures listed in methods were reported in results.
Other bias	High risk	Judgement comment: some sessions were audio‐recorded "… to assess for qualitative change in participants' level of depression", which described the content of sessions. However, there was no formal, independent ratings of treatment fidelity conducted using these transcripts. The research was conducted for a student dissertation; the researcher conducted recruitment and outcome assessment, implemented the intervention, and analysed the data, suggesting potential allegiance to the intervention. Small sample size may have introduced bias.

Dhooper 1993.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 16 Baseline characteristics Coping together group Age: not specified Gender: 87.5% women Cognitive function (MMSE): 87.5% cognitively intact, 12.5% moderately impaired Usual care Age: not specified Gender: 62.5% women Cognitive function (MMSE): 87.5% cognitively intact, 12.5% slightly impaired Overall Age: 64–94 years (mean 77.6 years) Gender: 75.0% women Cognitive function (MMSE): 87.5% cognitively intact Inclusion criteria: presence of depression from 'minimal to mild' and 'moderate to marked' – score 50–69 on the SDS. Able to communicate verbally and sit for 1 hour Exclusion criteria: presence of severe‐to‐profound orientation impairment (score 25+ on the Orientation questionnaire); severe cognitive impairment (MMSE < 10); physical inability to participate in group activity; unwillingness to participate in the project; severe depression (scores 70+ on SDS) Pretreatment: no differences in mean scores on cognition, orientation, or depression at baseline.
Interventions	Intervention characteristics Coping together group Length of treatment: 9 weeks Mode of treatment: group therapy Number of sessions: 9 Therapist(s): 3 social workers with Masters degrees and experience working with older adults Co‐interventions: not specified Usual care Length of treatment: 9 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: not specified
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: SDS Range: 25–100 Unit of measure: index scores Direction: lower is better Data value: end point Notes: also reported the number of participants in each category of severity at each time point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness
Identification	Sponsorship source: funded by the Lexington Junior League. Project run by local LTC ombudsman program Country: USA Setting: nursing home (intermediate and skilled nursing facility) Comments: none Author's name: Surjit S Dhooper Institution: University of Kentucky Email: not provided Address: 649 Patterson Office Tower, University of Kentucky, Lexington, KY 40506‐0027, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "the study subjects were randomly assigned to either the treatment or control group". Judgement comment: insufficient information about the sequence generation process to permit judgement.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: insufficient information about concealment method to permit judgement.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Three professional social workers with master's degrees and years of experience were hired to demonstrate an effective group approach to reducing depression". Judgement comment: given the nature of the intervention, participants and personnel were unlikely to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Quote: "The researchers were aware of the possibility of question distortion in oral administration and they minimized this possibility by reading each question exactly as written on the sheet with no interpretation of clarification". Judgement comment: no information provided about these researchers and whether they were blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Judgement comment: 2/8 participants in the control group withdrew (1 declined; 1 was transferred to hospital). No participants withdrew from the treatment group (0/8). This amount of missing data, while small, could affect the treatment effects, given the very small sample sizes in each arm of the trial.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no published protocol was identified. The manual mentioned in the study was unavailable. All outcome measures listed in methods were reported in results.
Other bias	High risk	Quote: "The [social] workers devised an eclectic approach called the "Coping Together Group". Judgement comment: potential allegiance of study authors to the intervention. No information provided about treatment fidelity assessment. Sample size was very small, which may also have introduced bias.

Dozeman 2011.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 14
Participants	Number of participants: 129 Baseline characteristics Guided self‐help activity scheduling Age: mean 83.7 (SD 6.7) years Gender: 68.7% women Cognitive function (MMSE): mean 27.1 (SD 2.0) Usual care Age: mean 84.2 (SD 6.8) years Gender: 80.6% women Cognitive function (MMSE): mean 26.8 (SD 2.0) Overall Age: mean 84.0 (SD 6.7) years Gender: 74.0% women Cognitive function (MMSE): not specified Inclusion criteria: depressive symptoms (CES‐D score ≥ 8), which had not improved by ≥ 5 points at secondary screening 1 month later Exclusion criteria: met criteria for DSM‐IV anxiety or depressive disorder on MINI at initial screening or secondary screening 1 month later; cognitive impairment (MMSE < 21); insufficient Dutch language skills. Pretreatment: no differences in baseline scores for depressive or anxiety symptoms
Interventions	Intervention characteristics Guided self‐help activity scheduling Length of treatment: 3 months Mode of treatment: structured self‐help programme with guidance from coach Number of sessions: 2–5 visits from coaches Therapists: facility staff or volunteers with 1 hour of training Frequency of sessions: not specified Session length: not specified Co‐interventions: not specified Usual care Length of treatment: 3 months Mode of treatment: N/A Number of sessions: N/A Therapists: N/A Frequency of sessions: N/A Session length: N/A Co‐interventions: residents had unrestricted access to any form of health care deemed appropriate.
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: CES‐D Range: 0–60 Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness Anxiety symptomatology Outcome type: continuous outcome Scale: HADS‐A Range: 0–21 Direction: lower is better Data value: end point Engagement in activities Outcome type: continuous outcome Scale: Trimbos/iMTA Questionnaire for Costs Associated with Psychiatric Illness (Tic‐P) (Roijen 2002) Range: not specified Direction: higher is better Data value: change from baseline
Identification	Sponsorship source: financial support (grant no. 62200026) was received from the Netherlands Organization for Health Research and Development (ZonMw). Country: the Netherlands Setting: residential homes (lower level care than nursing homes) Comments: in the Netherlands, several types of facilities for sheltered accommodation for the elderly are available, the 2 most important being residential homes and nursing homes. Residential homes provide assisted‐living facilities, including daily care (e.g. meals and house cleaning) and, if needed, uncomplicated demand‐led medical care. Nursing homes provide more specialized medical care to dependent people of all ages. Author's name: Els Dozeman Institution: VU University Medical Centre Email: els.dozeman@vumc.nl Address: Department of General Practice, Institute for Health and Care Research (EMGO+), VU University Medical Centre, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The participants were randomized to the intervention group or to the usual care group, in blocks of four, by an independent research assistant using random number tables". Judgement comment: random sequence generation performed.
Allocation concealment (selection bias)	Low risk	Quote: "Randomization took place after the baseline measurements … by an independent research assistant". Judgement comment: allocation unknown until after recruitment.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "… because residents in the intervention group received the self‐help intervention and those in the control group did not, it was not possible to blind the participants. Care‐providers were allocated to individual residents in the intervention group to help them with the intervention when needed". Judgement comment: participants and personnel were not blinded due to nature of intervention.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The interviewers were blinded for the randomization status of the respondents during all measurements". Judgement comment: outcome assessment was blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Judgement comment: 16/67 participants missing from treatment group; 6/62 participants missing from control group. Attrition was high, and related to death (n = 3), illness (n = 10), and unwillingness (n = 9). Higher dropout rate due to unwillingness in the treatment group (n = 7) than in the control group (n = 2). Primary analyses were conducted on an intention‐to‐treat basis, using maximum likelihood estimation to impute missing data.
Selective reporting (reporting bias)	High risk	Judgement comment: low risk – having mean and SD but no exact P values.
Other bias	High risk	Quote: [Of the 51 eligible participants, only 44] "finally attended the self‐help course with guidance from a coach. Evaluation, by means of the process evaluation leaflets revealed that two residents were unwilling to participate in the self‐help course, while five coaches failed to contact the resident … due to lack of time themselves or illness of the residents … [Only 14 residents] … completed all four steps in the self‐help course". Judgement comment: process evaluation indicated poor compliance with the intervention. The intervention was drawn from a protocol developed previously by a consortium (Prevention of Anxiety and Depression in Late Life) that included 2 of the authors of the trial. Potential researcher allegiance to the intervention.

Hamzehzadeh 2018.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 27 Baseline characteristics Reminiscence group Age: mean 77.9 (SD 6.9) years Gender: 100% women Cognitive function: not reported Treatment as usual Age: mean 74.1 (SD 7.5) years Gender: 100% women Cognitive function: not reported Overall Age: not reported Gender: 100% women Cognitive function: not reported Inclusion criteria: at least a moderate depression (obtaining a minimum score of 8 on GDS‐15); having a moderate cognitive status (MMSE ≥ 21) Exclusion criteria: no mental disability; attending other therapeutic groups; being individually treated by another person psychologically Pretreatment: no differences between groups at baseline in age, duration of stay, psychological problems score, depression score, optimism score
Interventions	Intervention characteristics Reminiscence group Length of treatment: not specified Mode of treatment: group therapy Number of sessions: 8 Therapist(s): not specified Co‐interventions: none Frequency of sessions: twice a week Session length: 1 hour Treatment as usual Length of treatment: not specified Mode of treatment: greeting and regular care Number of sessions: N/A Therapist(s): N/A Co‐interventions: N/A Frequency of sessions: N/A Session length: N/A
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: GDS‐15 Range: 0–15 Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from outcome Notes: did not include attrition related to death, relocation, or illness Optimism Outcome type: continuous outcome Scale: Life‐Orientation Test (Hamzezade 2013) Direction: higher is better Data value: end point
Identification	Sponsorship source: none specified Country: Iran Setting: 1 nursing home Comments: women‐only nursing home Author's name: Mahshid Foroughan Institution: University of Social Welfare and Rehabilitation Email: M_ Foroughan@yahoo.com Address: Iranian Research Center on Ageing, University of Social Welfare and Rehabilitation, Tehran, Iran
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "15 subjects were randomly assigned to the intervention group and 12 subjects were selected as the control group". Judgement comment: insufficient information about the sequence generation process.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: no information provided.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "This was as if the two groups did not know how the examiner treats each of them". Judgement comment: insufficient information to determine blinding of participants and personnel. Unlikely, due to nature of intervention.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Judgement comment: no information provided on blinding of outcome assessment.
Incomplete outcome data (attrition bias) All outcomes	High risk	Judgement comment: 5/15 participants missing from treatment group (4 for study‐related aspects); 1/12 participants from control group (not study‐related). Attrition rate was high in the treatment group and reasons differed from the control group. Unable to determine if intention‐to‐treat or per‐protocol approach to analysis was employed. Unclear how missing data were managed.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no protocol or trial registration was located. Depression and anxiety outcomes listed in the methods section of the paper were reported in the results.
Other bias	Unclear risk	Judgement comment: small sample size. No information about treatment fidelity assessments. Researcher and clinician allegiance, and role in the study could not be assessed. Major limitations in the reporting of the methods used in the trial.

Hansen 1976.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 22 Baseline characteristics Didactic group therapy Age: 76.9 years Gender: 63.6% women Cognitive function: not specified Usual care Age: 78.0 years Gender: 72.7% women Cognitive function: not specified Overall Age: 77.5 years Gender: 68.2% women Cognitive function: not specified Inclusion criteria: all participants scores SDS ≥ 50 but this was not clearly reported as an inclusion criterion. Exclusion criteria: diagnosis of schizophrenia, mental retardation or senility, and debilitating physical problems Pretreatment: similar scores on depression at baseline between groups; similar age and educational level. Note: no difference testing reported by author
Interventions	Intervention characteristics Didactic group therapy Length of treatment: 10 weeks Mode of treatment: group therapy Number of sessions: 10 Therapist(s): Masters student researcher Co‐interventions: not specified Frequency of sessions: weekly Session length: 60 minutes Usual care Length of treatment: 10 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: not specified Frequency of sessions: N/A Session length: N/A
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: SDS Range: 25–100 Unit of measure: raw scores Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness
Identification	Sponsorship source: none reported Country: USA Setting: nursing home Comments: Student dissertation Author's name: John J Hansen Institution: Eastern Illinois University Email: not specified Address: not specified
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Twenty‐two subjects were selected for the experiment with eleven subjects randomly selected for the experimental group and eleven subjects randomly selected for the control group". Judgement comment: insufficient information about the sequence generation process.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: the primary outcome measure was likely administered prior to allocation. No information provided about who conducted the allocation.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: the researcher ("Experimenter") administered the assessments and intervention. Blinding of participants unlikely due to nature of intervention.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "[Participants were] interviewed by the Experimenter. The Zung Self‐Rating Depression Scale [SDS] was administered at the immediate onset of the interview session". Judgement comment: outcome assessment was likely not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Judgement comment: no missing data – all 22 participants allocated to a condition completed the study and contributed data to analyses.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no protocol or trial registration was located (Masters thesis). Depressive symptomatology and sample attrition (used in analyses of treatment non‐acceptability in this review) were the only reported outcomes.
Other bias	High risk	Judgement comment: small sample size. No information about treatment fidelity assessments. The research was conducted for a Masters degree; the researcher conducted recruitment and outcome assessment, and implemented the intervention, suggesting potential allegiance to the intervention.

Hussian 1979.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 36 Baseline characteristics Social reinforcement of activity Age: 69.33 years Gender: not specified Cognitive function: not specified Problem‐solving training Age: 74.67 years Gender: not specified Cognitive function: not specified Social reinforcement of activity then problem‐solving training Age: 74.50 years Gender: not specified Cognitive function: not specified Problem‐solving training then social reinforcement of activity Age: 73.17 years Gender: not specified Cognitive function: not specified Wait‐list control Age: 74.67 years Gender: not specified Cognitive function: not specified Wait‐list control then informational control Age: 75.33 years Gender: not specified Cognitive function: not specified Overall Age: 73.61 years Gender: not specified Cognitive function: not specified Inclusion criteria: at least mildly depressed on the Beck Depression Inventory (selected 36 participants with the highest scores) Exclusion criteria: aged < 60 years, deaf, blind, or prescribed antidepressant medication Pretreatment: no group differences in depressive symptoms at baseline
Interventions	Intervention characteristics Social reinforcement of activity Length of treatment: 2 weeks Mode of treatment: individual therapy plus group aspects Number of sessions: 10 Therapist(s): researcher (4th year graduate student). Assisted by 2 staff members from the activities department Co‐interventions: not specified Frequency of sessions: 5 per week Session length: 30 minutes Problem‐solving training Length of treatment: 2 weeks Mode of treatment: individual therapy Number of sessions: 10 Therapist(s): researcher (4th year graduate student). Assisted by 2 staff members from the activities department Co‐interventions: did not participate in activities Frequency of sessions: 5 per week Session length: 30 minutes Social reinforcement of activity then problem‐solving training Length of treatment: 2 weeks Mode of treatment: individual therapy plus group aspects Number of sessions: 10 Therapist(s): researcher (4th year graduate student). Assisted by 2 staff members from the activities department Co‐interventions: did not participate in activities during problem‐solving training phase Frequency of sessions: 5 per week Session length: 30 minutes Problem‐solving training then social reinforcement of activity Length of treatment: 2 weeks Mode of treatment: individual therapy plus group aspects Number of sessions: 10 Therapist(s): researcher (4th year graduate student). Assisted by 2 staff members from the activities department Co‐interventions: did not participate in activities during problem‐solving phase Frequency of sessions: 5 per week Session length: 30 minutes Wait‐list control Length of treatment: 2 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: not specified Frequency of sessions: N/A Session length: N/A Wait‐list control then informational control Length of treatment: 2 weeks Mode of treatment: not specified Number of sessions: 5 Therapist(s): researcher (4th year graduate student) Co‐interventions: not specified Frequency of sessions: 5 per week Session length: not specified
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: Beck Depression Inventory Range: 0–63 Direction: lower is better Data value: end point Notes: only reported mean difference scores (not mean scores at each time point). No measure of variance reported. However, raw data were reported in an appendix in the thesis, which was used to calculate means and standard deviations. Depressive symptomatology – difference scores Outcome type: continuous outcome Scale: Beck Depression Inventory – difference scores Range: −63 to 63 Direction: lower is better Data value: change from baseline Notes: data in paper reports change scores only Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness
Identification	Sponsorship source: not specified Country: USA Setting: nursing home Comments: study is reported in student thesis (Hussian 1979) and a journal article (Hussian 1981) (both under Hussian 1979) Author's name: Richard A Hussian Institution: University of North Carolina Email: not specified Address: Department of Psychology, University of North Carolina at Greensboro, Greensboro, North Carolina 27412, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote "… 36 subjects were randomly assigned to one of three conditions … A pool of 10 subjects with high scores was used to replace subjects who terminated". Judgement comment: limited information about the sequence generation process, with possible non‐random component.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: no information provided.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: participants and personnel were likely not blinded due to nature of intervention. The author was the main therapist, with staff members from the facility assisting.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "The HAS [Hospital Adjustment Scale] was administered by the nursing home personnel". Judgement comment: outcome assessment was likely not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Judgement comment: there was attrition throughout the study (during the baseline, post‐treatment, and follow‐up), but not reported separately by condition and reasons for attrition were unclear.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no protocol or trial registration was located (PhD thesis). Outcomes listed in the method were reported in the results.
Other bias	High risk	Judgement comment: no information about treatment fidelity assessments. The research was conducted for a PhD thesis; the researcher likely conducted recruitment and outcome assessment, and implemented the intervention, suggesting potential allegiance to the intervention. Very small sample size per condition (n = 6), and very brief intervention period.

Hyer 2009.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 25 Baseline characteristics Group, individual and staff therapy Age: 78 years Gender: not specified Cognitive function (MMSE): 23.6 (SD 4.3) Depression diagnosis: 8 MDD; 5 adjustment disorder Treatment as usual Age: 81 years Gender: not specified Cognitive function (MMSE): 23.6 (SD 1.8) Depression diagnosis: 7 MDD; 5 adjustment disorder Overall Age: not specified Gender: 16% women Cognitive function (MMSE): not specified Depression diagnosis: 15 MDD; 10 adjustment disorder Inclusion criteria: depression diagnosis (e.g. MDD, adjustment disorder with depression) based on DSM‐IV, determined by a geropsychiatrist and geropsychologist; GDS‐short form score ≥ 5 Exclusion criteria: greater than mild level of cognitive impairment (MMSE < 18) Pretreatment: no pretreatment differences between groups on the outcome measures
Interventions	Intervention characteristics Group, individual and staff therapy Length of treatment: 13 weeks Mode of treatment: a single, repeated‐session group (open format, with participants joining at any point), complemented with individual sessions, and staff and peer coaching. Number of sessions: up to 2 individual and 13 weekly sessions (but participants attended ≥ 5 in total) Therapist(s): not specified Co‐interventions: continued medications (no changes during trial) Frequency of sessions: weekly Session length: 75–90 minutes Treatment as usual Length of treatment: 13 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: continued medications (no changes during trial) Frequency of sessions: N/A Session length: N/A
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: GDS – Short Form Range: 0–15 Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness Life satisfaction Outcome type: continuous outcome Scale: 18‐item Life Satisfaction Index Range: not specified Direction: higher is better Data value: end point
Identification	Sponsorship source: not specified Country: USA Setting: Veteran's nursing home Comments: N/A Author's name: Catherine A Yeager Institution: Institute for Mental Health Policy, Research, and Treatment Email: yeager.catherine@gmail.com Address: 204 Grove Avenue, Cedar Grove, New Jersey 07009, USA
Notes	2 trials are reported in the paper – the 'initial trial' which is described in this review, and a second 'continuation trial' in which the intervention group remained for a second course of treatment and the treatment as usual comparator crossed over to receive the intervention. The 'continuation trial' is not included in this review.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "Participants were randomly assigned to the GIST [group, individual and staff therapy] group (N = 13) or treatment as usual (TAU, N = 12). Randomisation was by diagnosis so that each group had roughly the same number of MDD participants". Judgement comment: possible non‐random component in the sequence generation process.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: no information provided. Baseline measures were collected postallocation at the first intervention session.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: participants and personnel were likely not blinded due to nature of intervention. Staff members were involved in coaching.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The posttreatment evaluations were performed by geropsychology trainees blind to the treatment condition". Judgement comment: outcome assessment was blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "… some data from the postsessions were missing from both groups due to uncontrollable events (sickness and hospitalization)". Judgement comment: 1 participant was recorded as a dropout due to hospitalization, although this appeared to have been in the second trial not included in this review (see Notes section), and reasons for missing outcome data seemed unlikely related to true outcome.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no protocol or trial registration was identified. Outcomes listed in the methods were reported in the results.
Other bias	Unclear risk	Judgement comment: assessment of treatment fidelity was not reported. The authors designed the intervention, adapting a previous intervention that was developed by other authors, so some minor risk of potential researcher allegiance.

Karimi 2010.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 39 Baseline characteristics Integrative reminiscence Age: not reported Gender: 60.0% women Cognitive function (MMSE): not reported Instrumental reminiscence Age: not reported Gender: 55.6% women Cognitive function (MMSE): not reported Social discussion control Age: not reported Gender: 70.0% women Cognitive function (MMSE): not reported Overall Age: 64–87 years (mean 70.5 years) Gender: 58.6% women Cognitive function (MMSE): not reported Inclusion criteria: resided in facility for > 6 months; aged ≥ 60 years; not currently receiving antidepressant medication (or if taking, stabilized for ≥ 3 months); able to understand and speak Persian easily; GDS‐15 score ≥ 5 Exclusion criteria: significant cognitive impairment (MMSE < 21); physical impairment prohibiting participation in group sessions; current participation in another psychotherapeutic intervention Pretreatment: no evaluation of group differences at baseline
Interventions	Intervention characteristics Integrative reminiscence Length of treatment: 6 weeks Mode of treatment: group therapy Number of sessions: 6 Therapist(s): Masters level therapist, supervised by a registered clinical psychologist Co‐interventions: not reported Frequency of sessions: weekly Session length: 90 minutes Instrumental reminiscence Length of treatment: 6 weeks Mode of treatment: group therapy Number of sessions: 6 Therapist(s): Masters level therapist, supervised by a registered clinical psychologist Co‐interventions: not reported Frequency of sessions: weekly Session length: 90 minutes Social discussion control Length of treatment: 6 weeks Mode of treatment: group therapy Number of sessions: 6 Therapist(s): Masters level therapist, supervised by a registered clinical psychologist Co‐interventions: not reported Frequency of sessions: weekly Session length: 90 minutes
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: GDS‐15 Range: 0–15 Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness
Identification	Sponsorship source: no funding reported Country: Iran Setting: nursing home Comments: NA Author's name: H Karimi Institution: Razi University Email: akmkarimi@gmail.com Address: Faculty of Social Sciences, Department of Psychology, Razi University, Kermanshah, Iran
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote "39 participants were randomly selected. In order to form matched groups … were then randomly assigned to the three conditions of intervention". Judgement comment: participants were systematically divided into 3 groups and then randomly assigned to 3 treatments. No information on the generation of a randomized sequence.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: no information provided.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: the same therapist conducted the 2 active treatments and the active control (social discussion group), so was likely not blinded. It is plausible that participants in all 3 conditions believed they were receiving an active intervention; hence it is possible that participants were blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "All assessments of participants were conducted by a Master's level clinical psychologist who was not the therapist and was blind to the subjects' intervention group". Judgement comment: outcome assessment was blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "Ten participants were excluded from the study for different reasons, including suffering from an illness or not attending at least 60% of the sessions". Judgement comment: study attrition (10/39 participants) was not reported separately for each intervention and reasons were not specified per intervention. Participants with missing data or who did not complete a sufficient number of treatment sessions were excluded from analyses.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no protocol or trial registration was identified. Outcomes listed in the method were reported in the results.
Other bias	Unclear risk	Judgement comment: assessment of treatment fidelity was not reported. Authors followed a treatment manual developed by others, so lower risk of researcher allegiance.

Luo 2020.

Study characteristics
Methods	Study design: cluster randomized controlled trial Study grouping: parallel group Number of facilities: 7
Participants	Number of participants: 68 Baseline characteristics Care as usual Age: mean 84.37 (SD 9.51) years Gender: 23 women, 7 men (76.7% women) Cognitive function (MMSE): mean 21.87 (SD 4.38) Positive Mood and Active Life Age: mean 85.94 (SD 7.14) years Gender: 27 women, 5 men (84.4% women) Cognitive function (MMSE): mean 21.50 (SD 4.49) Overall Age: not reported Gender: 50 women, 12 men (80.6% women) Cognitive function (MMSE): not reported Inclusion criteria: presence of a mood problem indicated by the Resident Assessment Protocol, triggered by the Minimum Data Set 2.0; Cognitive Performance Scale score of 0 or 1 (intact or borderline intact); no other acute clinical variations; voluntary participation Exclusion criteria: none Pretreatment: at baseline, participants in the treatment group were significantly older than those in the control, with a significantly lower pain score, and a significantly lower level of aggressive behaviour. Other demographics and clinical characteristics did not differ, but the paper did not report whether there were significant differences in the depression outcome measure at baseline.
Interventions	Intervention characteristics Care as usual Length of treatment: 12 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: N/A Frequency of sessions: N/A Session length: N/A Positive Mood and Active Life Length of treatment: 12 weeks Mode of treatment: individual sessions, with support from LTC team in engaging in activities Number of sessions: 36 Therapist(s): led by a trained social worker in the facility, who engaged participants in the selected activities through psychological care and counselling. Programme workers and personal care workers helped participants engage in daily activities. Also input from other facility staff (nurses, occupational therapists, and physiotherapists) Co‐interventions: continued medications as part of usual care Frequency of sessions: not specified Session length: not specified
Outcomes	Depressive symptoms Outcome type: continuous outcome Scale: GDS‐15 Range: 0–15 Direction: lower is better Data value: end point Notes: reported as estimated marginal means Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness Quality of life Outcome type: continuous outcome Scale: WHOQOL‐BREF (Hong Kong version) Range: not reported Direction: lower is better Data value: end point
Identification	Sponsorship source: none reported Country: Hong Kong Setting: LTC facilities Comments: – Author's name: Vivian WQ Lou Institution: The University of Hong Kong Email: wlou@hku.hk Address: Department of Social Work and Social Administration, Rm 522, The Jockey Club Tower, Centennial Campus, The University of Hong Kong, Pokfulam Road, Hong Kong, China
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Seven long‐term care facilities managed by a single nongovernment organisation were randomly assigned as experimental and control sites". Judgement comment: author noted that assignment was conducted using a computer random number (personal correspondence).
Allocation concealment (selection bias)	Unclear risk	Judgement comment: no information provided.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: participants and personnel were likely not blinded due to nature of intervention.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "The items of the project‐tailored assessment were read to the residents by an experienced research assistant who was trained on the GDS‐15 and WHOQoL‐BREF by the principal investigator". Judgement comment: author confirmed outcome assessors were not blinded (personal correspondence).
Incomplete outcome data (attrition bias) All outcomes	Low risk	Judgement comment: 2/34 participants missing from treatment group; 4/34 participants missing from control group. Attrition was low, with similar reasons across the groups (frailty and death), unlikely related to the treatment.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no published protocol (confirmed by author in personal correspondence). All outcome measures listed in methods reported in results.
Other bias	High risk	Judgement comment: small number of clusters, all part of same organization. No information about differences between sites on outcome measures. Author designed treatment, so potential researcher allegiance. No assessment of treatment fidelity.

Meeks 2008.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 6
Participants	Number of participants: 20 Baseline characteristics Behavioural Activities Intervention (BE‐ACTIV) Age: mean 76.9 (SD 11.5) years Gender: not specified Cognitive function (MMSE): mean 18.3 (SD 5.5) Treatment as usual Age: mean 79.4 (SD 4.3) years Gender: not specified Cognitive function (MMSE): mean 22.9 (SD 5.5) Overall Age: not specified Gender: not specified Cognitive function (MMSE): not specified Inclusion criteria: GDS ≥ 11; met diagnostic criteria for MDD, minor depression, or intermittent depressive disorder on the SADS (Endicott 1978); MMSE score > 13 Exclusion criteria: receiving psychotherapy; terminally ill or under hospice care; or had an unstable medical condition Pretreatment: author reported that discussion of baseline group comparisons was beyond the scope of this article. The treatment group had more diseases but less impairment
Interventions	Intervention characteristics Behavioural activities intervention (BE‐ACTIV) Length of treatment: 10 weeks Mode of treatment: individual sessions with mental health consultant, plus facility staff training and consultations between facility staff and mental health consultant Number of sessions: 10 Therapist(s): the mental health consultant was an experienced clinician (and the primary author). Facility staff collaborated in treatment and assisted in implementing chosen activities Co‐interventions: none Frequency of sessions: weekly Session length: 30–40 minutes Treatment as usual control Length of treatment: 10 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: none Frequency of sessions: N/A Session length: N/A
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: GDS‐30 item Range: 0–30 Direction: lower is better Data value: end point Depressive symptomatology Outcome type: continuous outcome Scale: Hamilton Depression Rating Scale Range: not specified Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness Depression remission Outcome type: dichotomous outcome Scale: SADS Data value: end point Notes: reporting the number with any type of depression diagnosis at post‐treatment and follow‐up (not reported at pretreatment)
Identification	Sponsorship source: funded by National Institute of Mental Health Country: USA Setting: nursing homes Comments: N/A Author's name: Suzanne Meeks Institution: University of Louisville Email: smeeks@louisville.edu Address: Department of Psychological & Brain Sciences, University of Louisville, Louisville, Kentucky 40292, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Randomly assigned". Judgement comment: author noted possible use of a random number scheme by the biostatistician (personal correspondence). There were differences in baseline characteristics between the groups.
Allocation concealment (selection bias)	Low risk	Judgement comment: author stated a biostatistician generated a list of numbers that determined which group each participant went in (personal correspondence). Only the biostatistician and lead author (also pretreatment assessor and therapist) knew which participant was assigned to which group.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: participants and personnel were likely not blinded due to nature of intervention.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "A doctoral student blind to treatment condition and trained to be reliable on the SADS with the principal investigator and criterion training tapes conducted posttreatment interviews". Judgement comment: the person conducting post‐treatment outcome assessments was blinded, although pretreatment assessor was not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Judgement comment: 3/13 participants missing from treatment group; 3/7 participants from control group. Attrition was high, relative to sample size, with reasons related to medical morbidity and assessment burden.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: trial registration posted 2003 did not list outcome measures. All outcome measures listed in methods relevant to this review reported in results.
Other bias	High risk	Judgement comment: very small sample size may have hindered randomization creating balance across groups. The study author developed the intervention, was the lead therapist, conducted pretreatment outcome assessment, and supervised the intervention, so potential for researcher/therapist allegiance. Assessment of treatment fidelity by examining data on client and staff activity adherence, and therapist adherence to session content.

Meeks 2015.

Study characteristics
Methods	Study design: cluster randomized controlled trial Study grouping: parallel group Number of facilities: 23
Participants	Number of participants: 82 Baseline characteristics Behavioural activities intervention (BE‐ACTIV) Age: mean 77.31 (SD 12.07) years Gender: 66.7% women Cognitive function (MMSE): not specified Treatment as usual Age: mean 72.85 (SD 11.87) years Gender: 64.1% women Cognitive function (MMSE): not specified Overall Age: mean 75.16 (SD 12.11) years Gender: 65.4% women Cognitive function (MMSE): not specified Inclusion criteria: diagnosis, positive facility screen, or treatment of depression, reported by staff; met criteria for DSM‐IV depression disorder, using the SCID‐IV; GDS score ≥ 11; aged ≥ 55 years; expected stay ≥ 3 months; staff belief they were sufficiently cognitively intact to give consent Exclusion criteria: under the care of or referral to hospice for a terminal condition; medical condition deemed unstable or terminal; unable to participate in activities due to physical condition; receiving weekly psychotherapy (treatment group only); MMSE score < 14 Pretreatment: fewer participants with MDD and more with minor depression in the treatment group than in the control group. Treatment participants had fewer days on an antidepressant at baseline, more education, and a larger number of physician visits than the control participants
Interventions	Intervention characteristics Behavioural activities intervention (BE‐ACTIV) Length of treatment: 10 weeks Mode of treatment: individual sessions with a mental health therapist (with staff facilitators attending some sessions). Plus sessions between the therapist and staff facilitators Number of sessions: 10 Therapist(s): mental health therapists were clinical psychology doctoral students who completed a training seminar and practicum on the intervention. Staff facilitators received a 2‐hour training programme Co‐interventions: most participants were prescribed antidepressant medication (unclear if medication changes were made during the study period) Frequency of sessions: weekly Session length: not specified Treatment as usual Length of treatment: 10 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: most participants were prescribed antidepressant medication (unclear if medication changes were made during the study period) Frequency of sessions: N/A Session length: not specified
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: GDS‐30 (Brink 1982) Range: 0–30 Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness. For this study, just included data from the treatment phase (not follow‐up) Presence of MDD Outcome type: dichotomous outcome Scale: SCID‐IV Data value: end point Notes: data provided by study author upon request. Note that this variable simply records presence or absence of MDD at each time point, not whether the participant met DSM criteria for remission MDD non‐remission Outcome type: dichotomous outcome Scale: SCID‐IV Direction: lower is better Data value: change from baseline Notes: non‐remission included participants in partial remission (i.e. whose symptoms had resolved, but who had not been clear for ≥ 2 months (following DSM criteria))
Identification	Sponsorship source: supported by the National Institute of Mental Health. Country: USA Setting: nursing homes Comments: supplementary material included with article. Author's name: Suzanne Meeks Institution: University of Louisville Email: not provided Address: Department of Psychological and Brain Sciences, University of Louisville, Kentucky, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "The unit of randomization was the nursing home, with residents nested within nursing homes; nursing homes were blocked by size (greater or fewer than 100 beds)". Judgement comment: insufficient information about the sequence generation process.
Allocation concealment (selection bias)	Low risk	Judgement comment: author confirmed that randomization was performed by a biostatistician, who provided assignment to research staff unaware of allocation until after the facility was recruited (personal communication).
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: participants and personnel were likely not blinded due to nature of intervention.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The posttreatment and follow‐up interviews were completed by the third author, who remained blind to condition throughout the study". Judgement comment: the person conducting post‐treatment outcome assessments was blinded (with steps taken to maintain blindness), although pretreatment assessor was not blinded (confirmed by author in personal correspondence).
Incomplete outcome data (attrition bias) All outcomes	Low risk	Judgement comment: at post‐treatment, 9/42 participants missing from treatment group; 3/40 participants missing from control. At 3‐month follow‐up, 9/42 participants missing from treatment; 7/40 participants missing from control. At 6‐month follow‐up, 13/42 participants missing from treatment; 12/40 participants missing from control. Unequal attrition at post‐treatment, but not present by 3‐month follow‐up. Most attrition related to change in health status, although during the treatment phase, 4 participants voluntarily withdrew from the treatment group and 1 from the control group. Intention‐to‐treat analysis used, carrying forward result for missing data.
Selective reporting (reporting bias)	Low risk	Judgement comment: reported outcomes matched those in trial registration posted 2007. Activities of daily living were reported in the methods and not in the results due to limitations of this administrative data (personal correspondence).
Other bias	High risk	Judgement comment: the prevalence of MDD appeared to be unbalanced between groups (47.6% in psychological therapy group vs 25% in non‐therapy comparator group). Author developed the intervention, so potential researcher allegiance.

Reinhardt 2014.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 37 Baseline characteristics Problem‐solving therapy Age: mean 75.2 (SD 11.4) years Gender: 52.4% women Cognitive function (MMSE): mean 25.0 (SD 4.4) Social contact comparison Age: mean 75.9 (SD 8.3) years Gender: 50.0% women Cognitive function (MMSE): mean 24.4 (SD 4.4) Overall Age: 75 (SD not reported) years Gender: 51.4% Cognitive function (MMSE): not specified Inclusion criteria: significant depression symptoms – Patient Health Questionnaire 9 score ≥ 10; diagnosis of subthreshold or minor depression, dysthymia, or MDD using the SCID; HAM‐D score < 20 (to include older adults with less‐severe depressive disorders); intact/mildly impaired cognitive function (Cognitive Performance Scale score 0–2 or Brief Mental Status score 8–15 with institution of minimum data set 3.0, plus MMSE score ≥ 18); English speaking; able to communicate verbally; sufficient hearing for a personal interview Exclusion criteria: none Pretreatment: the 2 groups had similar sociodemographic characteristics; however, only 56% of the social contact group was white, compared to 86% of the problem‐solving therapy group. The groups were equivalent on major screening and outcome variables at baseline
Interventions	Intervention characteristics Problem‐solving therapy Length of treatment: 6 weeks Mode of treatment: individual sessions Number of sessions: 6 Therapist(s): licenced clinical social worker who was trained and supervised by a consulting clinical psychologist Co‐interventions: not specified Frequency of sessions: varied – offered weekly for nursing home residents, twice‐weekly for short‐stay residents Session length: 60 minutes Social contact comparison Length of treatment: varied – 6 weeks for nursing home residents, likely shorter for short‐stay residents Mode of treatment: friendly visits Number of sessions: 6 Therapist(s): intern Co‐interventions: not specified Frequency of sessions: varied – offered weekly for nursing home residents, twice‐weekly for short‐stay residents Session length: not specified
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: Hamilton Rating Scale for Depression (Williams 1988) Direction: lower is better Data value: change from baseline Notes: means and SDs at baseline were reported in a table. Means at other time points were in graphical form, with no reporting of SDs Depressive symptomatology Outcome type: continuous outcome Scale: HAM‐D Direction: higher is better Data value: change from baseline Notes: unadjusted net mean differences, reported in a table Depressive symptomatology Outcome type: continuous outcome Scale: PROMIS depression scale Range: not specified Direction: lower is better Notes: means and SDs at baseline were reported in a table. Means at other time points were in graphical form, with no reporting of SDs Depressive symptomatology Outcome type: continuous outcome Scale: PROMIS depression scale Range: not specified Direction: higher is better Data value: change from baseline Notes: unadjusted net mean differences, reported in a table Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness.
Identification	Sponsorship source: funded by the Mount Sinai School of Medicine Claude D. Pepper Older Americans Center on "Promoting Independence through Pain and Symptom Management" (1P30AG028741‐01A2). Country: USA Setting: nursing home of a major geriatric healthcare organization Comments: sample included 32 short‐stay rehabilitation residents and 5 long‐stay nursing home residents. Author's name: Joann P Reinhardt Institution: Research Institute on Aging Email: jreinhardt@jewishhome.org Address: Jewish Home Lifecare, 120 West 106th Street, New York, New York 10025, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "After the baseline interview, participants were randomized to the treatment group (receiving PST) or to a comparison group (social contact)". Personal correspondence from author: "treatment vs. comparison group designation was noted on paper placed in a sealed envelope and envelopes were picked at random following participants' baseline interview". Judgement comment: sequence generation appeared to have been random, with shuffled envelopes.
Allocation concealment (selection bias)	Low risk	Quote (personal correspondence from author): "treatment vs. comparison group designation was noted on paper placed in a sealed envelope and envelopes were picked at random following participants' baseline interview". Judgement comment: baseline measures were collected prior to randomization, and concealed using sealed envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote (personal correspondence from author): "There was no blinding of personnel, participants or assessors". Judgement comment: participants and personnel were not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote (personal correspondence from author): "There was no blinding of personnel, participants or assessors". Judgement comment: outcome assessment was not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "… only about half of the PST intervention group completed their treatment and followup interviews, relative to the social contact group". Judgement comment: 11/21 participants from the treatment group and 1/16 from the control group dropped out prior to data collection at the second time point. Substantial levels of attrition, predominantly from the treatment condition, including for reasons potentially related to the intervention. Missing data were not imputed.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no protocol or trial registration identified. 2 outcomes listed in the methods (HAM‐D and PROMIS depression scale) were reported in the results. The SCID was listed in the methods, but diagnostic data were not reported in the results. However, it is unclear whether the SCID was intended as a screening measure only.
Other bias	High risk	Quote: "… study sample … was composed of 2 potentially different groups of nursing home residents". Judgement comment: sample may not have been homogeneous. Unclear if treatment fidelity was assessed. Researcher allegiance to the intervention could not be determined.

Rosen 1997.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 22 Baseline characteristics Control‐relevant intervention Age: not reported Gender: not reported Cognitive function: not reported Wait‐list control Age: not reported Gender: not reported Cognitive function: not reported Overall Age: not reported Gender: not reported Cognitive function:not reported Inclusion criteria: psychiatric diagnosis of minor depressive episode or MDD, with mild or moderate severity, made using the SCID; MMSE ≥ 18; resided in the facility for ≥ 3 months Exclusion criteria: severe major depression Pretreatment: author reported the 2 groups were similar in age, gender, education, medical burden, cognitive ability, antidepressant use, and depression rating scale scores (but no data presented).
Interventions	Intervention characteristics Control‐relevant intervention Length of treatment: 8 weeks Mode of treatment: self‐directed activity programme, facilitated by therapist Number of sessions: not reported. Participants engaged in a variable number of activities (typically 2 per day, 5 days per week) Therapist(s): licenced recreational therapist Co‐interventions: monitored medications Frequency of sessions: not reported. Participants engaged in a variable number of activities (typically 2 per day, 5 days per week) Session length: not reported. Participants engaged in activities of variable duration (typically 1–2 hours) Wait‐list control Length of treatment: 8 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: monitored medications Frequency of sessions: N/A Session length: N/A
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: Hamilton Depression Scale Direction: lower is better Notes: data on change scores (t‐tests and an analysis of variance) were reported, but the authors did not report means and SDs separately for participants who were randomized. Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness.
Identification	Sponsorship source: supported by grants from the Jewish Healthcare Foundation of Pittsburgh and the National Institute of Mental Health Country: USA Setting: LTC facility Comments: N/A Author's name: Jules Rosen Institution: University of Pittsburgh School of Medicine Email: not provided Address: Mental Health Clinical Research Center for the Study of Late‐Life Mood Disorders, Department of Psychiatry, Western Psychiatric Institute and Clinic, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania 15213, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "After initial assessment, the first 22 participants were randomly assigned either to 2‐month intervention program or to a "wait‐list" control condition for 2 months before the intervention". Judgement comment: insufficient information about the sequence generation to permit judgement. An additional 9 participants were recruited later (non‐random allocation) to increase the sample size, but data from these additional participants were not included in the meta‐analysis.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: no information provided.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: participants and personnel were likely not blinded due to nature of intervention.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Quote: "assessments were completed by a single trained rater before the intervention (pre‐intervention), during the final week of the 2‐month intervention (post‐intervention) and 2 months after discontinuation of treatment (followup)". Judgement comment: while facility staff who completed an outcome not included in this review (a categorical rating of global improvement or functioning) by blinded staff, no information was provided about the blinding of the single rater who administered other measures.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Judgement comment: no missing outcome data were reported. Only data from the 22 participants who were randomly assigned were considered for this review.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no protocol or trial registration was identified. All outcomes listed in the methods were reported in the results; however, means and SDs were not reported separately for the 22 participants who were randomly allocated to condition.
Other bias	High risk	Quote: "… we created a psychosocial intervention for nursing home residents with depression". Judgement comment: the authors may have had a vested interest in the therapy provided. The author noted that the primary outcome (a rating of improvement by facility staff) lacked established validity. This outcome was not eligible for inclusion in the review. Assessment of treatment fidelity was not reported.

Sales 2015.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 8 Baseline characteristics Cognitive restructuring therapy Age: mean 89.5 (SD 2.6) years Gender: 75.0% women Cognitive function (MEC): 25.75 (SD 0.96) Control Age: mean 88.5 (SD 3.4) years Gender: 50% women Cognitive function (MEC): mean 26.25 (SD 1.7) Overall Age: 85–93 years (mean 89 years) Gender: 62.5% women Cognitive function (MEC): not specified Inclusion criteria: aged > 65 years; depressive symptoms (GDS‐30 score ≥ 11; DASS score ≥ 10) Exclusion criteria: cognitive impairment (Mini Cognitive Exam score ≤ 23); history of neurological disease, psychiatric disorder, or severe systemic disease (except depressive disorder); history of substance abuse according to DSM‐IV‐TR; diagnosis of dementia according to DSM‐IV‐TR; chronic use of psychoactive or hypnotic medication; illiteracy Pretreatment: the 2 groups did not differ in age, gender, marital status, education, number of children, or cognition
Interventions	Intervention characteristics Cognitive restructuring therapy Length of treatment: 4 weeks Mode of treatment: group therapy Number of sessions: 8 Therapist(s): not specified Co‐interventions: not specified Frequency of sessions: twice weekly Session length: 60 minutes Control Length of treatment: 4 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: not specified Frequency of sessions: N/A Session length: N/A
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: DASS – Depression Scale Range: 0–42 Direction: lower is better Data value: end point Notes: GDS was also administered, but scores were not reported in the paper Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness Anxiety symptomatology Outcome type: continuous outcome Scale: DASS – Anxiety Scale Range: 0–42 Direction: lower is better Data value: end point Life satisfaction Outcome type: continuous outcome Scale: Philadelphia Geriatric Center Morale Scale (note: named 'Escala de Satisfacción de Filadelfia [PGC]' in the original publication) Range: not specified Direction: higher is better Data value: end point
Identification	Sponsorship source: none reported Country: Spain Setting: residence for the elderly Comments: N/A Author's name: Alicia Sales Galan Institution: University of Valencia Email: alicia.sales@uv.es Address: Department of Evolutionary Psychology and Education, Faculty of Psychology, University of Valencia, AV. Blasco Ibanez 21, 46010 Valencia, Spain
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "A randomized experimental study was conducted with two analysis groups, an experimental group (n = 4) and a control group (n = 4), which were evaluated in two different time periods (pre and post)". Judgement comment: insufficient information about the sequence generation process.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: no information provided.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: participants and personnel were likely not blinded due to nature of intervention.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Judgement comment: no information provided regarding person who conducted the assessments and whether they were blinded.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Judgement comment: attrition rates were not reported for the treatment or control groups.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no published protocol or trial registration identified. All outcome measures listed in methods reported in results (although scores on the GDS were not included in the table of results).
Other bias	High risk	Judgement comment: very small sample size (4 participants in each group) may have hindered randomization and created imbalance across groups. Assessment of treatment fidelity was not conducted. Therapist allegiance was unclear.

Travers 2017.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 4
Participants	Number of participants: 19 Baseline characteristics Behavioural activation intervention – BE‐ACTIV Age: mean 87.2 (SD 7.7) years Gender: 80% women Cognitive function (sMMSE): 13–26, mean 19.5 (SD 3.9) Walking and Talking intervention Age: mean 85.5 (SD 10.9) years Gender: 100% women Cognitive function (sMMSE): 12–22, mean 16.13 (SD 3.7) Overall Age: mean 86.5 (SD 8.8) years Gender: 88.9% women Cognitive function (sMMSE): mean 18.0 (SD 4.1) Inclusion criteria: GDS‐12R score ≥ 4 (Sutcliffe 2000); mild‐to‐moderate dementia (standardized MMSE score ≥ 10); diagnosis of dementia; ability to communicate in English; resided in the facility for ≥ 3 months Exclusion criteria: receiving psychotherapy; dying Pretreatment: reported that the characteristics of the 2 groups were in general similar
Interventions	Intervention characteristics Behavioural activation intervention (BE‐ACTIV) Length of treatment: 8 weeks (reduced from 10 weeks) Mode of treatment: individual sessions with mental health therapist, plus therapist‐staff facilitator sessions Number of sessions: 8 Therapist(s): qualified social worker experienced in working with older people, who received training in the treatment and ongoing supervision Co‐interventions: not specified Frequency of sessions: weekly Session length: 45 minutes Walking and talking intervention Length of treatment: 8 weeks Mode of treatment: individual time spent with volunteer walking and talking Number of sessions: varied Therapist(s): facility volunteer Co‐interventions: not specified Frequency of sessions: varied Session length: 30 minutes per week in total
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: GDS‐12 Range: 0–12 Direction: lower is better Data value: end point Change in depressive symptomatology Outcome type: continuous outcome Scale: GDS‐12 Range: −12 to 12 Direction: higher is better Data value: change from baseline Notes: assumed that Table 3 reported medians and interquartile ranges Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness. Quality of life Outcome type: continuous outcome Scale: QOL‐AD‐NH Range: 15–60 Direction: higher is better Data value: change from baseline Change in quality of life Outcome type: continuous outcome Scale: QOL‐AD‐NH Range: −60 to 60 Direction: higher is better Data value: change from baseline
Identification	Sponsorship source: funded by the JO & JR Wicking Trust Country: Australia Setting: nursing homes Comments: N/A Author's name: Catherine Travers Institution: Queensland University of Technology Email: catherine.travers@qut.edu.au Address: Queensland Dementia Training Study Centre, Queensland University of Technology, Level 6, N Block, Victoria Park Road, Kelvin Grove, Queensland 4059, Australia
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The Project Coordinator assigned participants a unique study identification number and randomly allocated them to either the BE‐ACTIV or the Walking and Talking intervention using the SPSS randomization function". Judgement comment: random sequence generation performed.
Allocation concealment (selection bias)	Low risk	Quote (from trial registration): "The Research Assistant responsible for recruiting participants into the study will be a different person to the person who will be acting as the therapist for the study and will be unaware regarding group allocation". Judgement comment: allocation concealment likely performed.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote (from trial registration): "Open (masking not used)". Judgement comment: participants and personnel were not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "The measures of QOL (QOL‐AD‐NH) and depression (GDS‐12R) were re‐administered following completion of the intervention by the Project Coordinator only, who was not blinded regarding participants' group allocation". Judgement comment: outcome assessor was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Judgement comment: 0/10 participants missing from treatment group; 1/9 participants missing from control group. Single case of attrition due to death.
Selective reporting (reporting bias)	Low risk	Judgement comment: trial registration from 2013 (ACTRN12613000296730) listed agitation as an outcome measure, which was not reported. However, this outcome was not included in the meta‐analysis so did not impact on results. All other outcome measures were reported.
Other bias	Unclear risk	Quote: "Weekly contact between the [therapist] and the Project Coordinator was maintained throughout the intervention to ensure a high level of treatment fidelity and to address any difficulties that arose". Judgement comment: treatment fidelity was considered but unclear if formally assessed. Author did not develop the intervention, so low risk of researcher allegiance.

Tsai 2008.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 111 Baseline characteristics Self‐worth therapy Age: mean 78.5 (SD 5.5) years Gender: 61.3% women Cognitive function (MMSE): mean 27.5 (SD 2.3) Attention control group Age: mean 76.3 (SD 6.5) years Gender: 65.6% women Cognitive function (MMSE): mean 25.8 (SD 4.4) Overall Age: not reported Gender: 63.5% women Cognitive function (MMSE): not reported Inclusion criteria: GDS‐30 score > 10 (Brink 1982) Exclusion criteria: severe cognitive deficits (MMSE score < 16 or score < 20 for those who completed primary schooling or more); change in antidepressant medications during the trial or in the 3 months prior to the trial commencement Pretreatment: no demographic differences. However, the treatment group had significantly more symptoms of depression at baseline (higher GDS score) than the control group
Interventions	Intervention characteristics Self‐worth therapy Length of treatment: 4 weeks Mode of treatment: individual sessions Number of sessions: 4 Therapist(s): Masters‐prepared mental health nurse Co‐interventions: no changes in antidepressant medications Frequency of sessions: weekly Session length: 30 minutes Attention control group Length of treatment: 4 weeks Mode of treatment: individual sessions Number of sessions: 4 Therapist(s): Masters prepared mental health nurse Co‐interventions: no changes in antidepressant medications Frequency of sessions: weekly Session length: 30 minutes
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: GDS‐30 Range: 0–30 Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness
Identification	Sponsorship source: supported by the National Science Council (NSC 94‐2314‐B‐ 182‐015) Country: Taiwan Setting: nursing home Comments: not stated Author's name: Yun‐Fang Tsai Institution: Chang Gung University Email: yftsai@mail.cgu.edu.tw Address: School of Nursing, Chang Gung University, Tao‐Yuan, Taiwan
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The first author allocated each participant a number, which was used with a table of random numbers to assign them to the experimental or control group". Judgement comment: random sequence generation done.
Allocation concealment (selection bias)	Low risk	Quote: "The list of names and group assignments thus created was given only to RA1". Judgement comment: allocation appeared to have been generated by the first author of the study, and then communicated to RA1, a mental health nurse who administered the intervention.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "The control group received no treatment but RA1 visited and chatted with each participant once a week for 30 minutes". [Participants in the control group] "… were individually visited by the same research assistant, who chatted with them for 30 minutes on 1 day/week for 4 weeks". Judgement comment: participants were visited by the same mental health nurse, so the personnel were not blinded. Although there was no information about whether participants were blinded, it is plausible that control group participants may have believed that they were in an active intervention condition and were blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Immediately after the 4‐week therapy or control treatment and again 2 months later, RA2 assessed participants for depressive, cognitive and functional status using the GDS, MMSE and Barthel Index respectively. RA2 did not know which participant belonged to which group". Judgement comment: outcome assessment was blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Judgement comment: high attrition rate. At post‐treatment, 14/56 participants missing from treatment group (13 missing due to discharge, 1 due to death); 13/55 participants missing from control (due to discharge). At 2‐month follow‐up, 25/56 participants missing from treatment (24 due to discharge, 1 due to death); 23/55 participants missing from control (due to discharge). Most attrition related to discharge, with equal attrition rates across conditions. Participants who withdrew from the 2 groups were similar in demographic characteristics.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no published protocol or trial registration identified. All outcome measures listed in methods reported in results.
Other bias	High risk	Judgement comment: participants were recruited from only 1 nursing home. Assessment of treatment fidelity was not conducted. Treatment allegiance was likely, as self‐worth therapy intervention was developed by first author.

Zerhusen 1991.

Study characteristics
Methods	Study design: randomized controlled trial Study grouping: parallel group Number of facilities: 1
Participants	Number of participants: 60 Baseline characteristics Cognitive therapy Age: not reported Gender: not reported Cognitive function: not reported Music therapy Age: not reported Gender: not reported Cognitive function: not reported Treatment as usual control Age: not reported Gender: not reported Cognitive function: not reported Overall Age: 70–82 years (mean 77 years) Gender: not reported Cognitive function: not reported Inclusion criteria: moderately to severely depressed, according to the Beck Depression Inventory score; free from organic brain syndrome Exclusion criteria: none Pretreatment: no baseline group differences reported
Interventions	Intervention characteristics Cognitive therapy Length of treatment: 10 weeks Mode of treatment: group therapy Number of sessions: not reported Therapist(s): nursing home staff – 2 registered nurses and a social worker. All received training in delivering the cognitive therapy intervention Co‐interventions: none reported Frequency of sessions: not reported Session length: not reported Music therapy Length of treatment: 10 weeks Mode of treatment: group therapy Number of sessions: 20 Therapist(s): a trained professional Co‐interventions: none reported Frequency of sessions: twice weekly Session length: 60 minutes Treatment as usual Length of treatment: 10 weeks Mode of treatment: N/A Number of sessions: N/A Therapist(s): N/A Co‐interventions: N/A Frequency of sessions: N/A Session length: N/A
Outcomes	Depressive symptomatology Outcome type: continuous outcome Scale: Beck Depression Inventory Range: 0–63 Direction: lower is better Data value: end point Treatment non‐acceptability Outcome type: dichotomous outcome Scale: participant dropout Data value: change from baseline Notes: did not include attrition related to death, relocation, or illness
Identification	Sponsorship source: no funding reported Country: USA Setting: nursing home Comments: NA Author's name: Kathleen Boyle Institution: University of Cincinnati Email: not provided Address: College of Nursing and Health, University of Cincinnati, Cincinnati, OH 45221‐0038, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "All consenting residents were initially given the Beck Depression Inventory, then matched for age, sex and depression; those who qualified as moderately to severely depressed and free from organic brain syndrome were randomly assigned to one of three groups". Judgement comment: insufficient information about the sequence generation process.
Allocation concealment (selection bias)	Unclear risk	Judgement comment: no information provided.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Judgement comment: participants and personnel were likely not blinded due to nature of intervention.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Judgement comment: the participants completed the Beck Depression Inventory. It is not known if the inventory was completed as a self‐report measure or was administered by an assessor. Hence, it is unclear if outcome assessments were blinded. If completed by participants, blinding was not likely. If administered by an assessor, blinding is unclear. We have assumed the former, in the absence of details about the administration.
Incomplete outcome data (attrition bias) All outcomes	High risk	Judgement comment: attrition appeared minimal, but was managed by removing additional participants. 1/20 participants dropped out (reason unknown) from treatment group. The researchers then discarded 1/20 participants from the music therapy group (control; not included in this review), and 1/20 participants from the routine nursing care group (control).
Selective reporting (reporting bias)	Unclear risk	Judgement comment: no published protocol or trial registration identified. All outcome measures listed in methods reported in results.
Other bias	Low risk	Quote: "… five judges rated the group leaders on videotaped samples of the group activities using the Cognitive Therapy Scale developed by Young (1980)". Judgement comment: assessment of treatment fidelity was conducted, with all group leaders "performing well above the minimum standard". Participants were recruited from only 1 nursing home. Researchers used a well‐documented treatment, so researcher allegiance was not as high as it would have been if the treatment was developed by the researcher.

3MS: Modified Mini‐Mental State Examination; CES‐D: Centre for Epidemiologic Studies Depression Scale; DASS: Depression Anxiety and Stress Scales; DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders, 4th Edition; DSM‐IV‐TR: Diagnostic and Statistical Manual of Mental Disorders, 4th Edition Text Revision; GDS‐12R: 12‐item Geriatric Depression Scale – Residential edition; GDS‐15: 15‐item Geriatric Depression Scale; GDS‐30: 30‐item Geriatric Depression Scale; HADS‐A: Hospital Anxiety and Depression Scale; HAM‐D: Hamilton Rating Scale for Depression; LTC: long‐term care; MEC: mobile examination centre; MINI: Mini International Neuropsychiatric Interview; MMSE: Mini‐Mental State Examination; n: number; N/A: not applicable; QOL‐AD: Quality of Life – Alzheimer's Disease; QOL‐AD‐NH: Quality of Life – Alzheimer's Disease, for Nursing Homes; WHOQOL‐BREF: World Health Organization Quality‐of‐Life Scale – Short Form; RSQ: Resident Status Questionnaire; SADS: Schedule for Affective Disorders and Schizophrenia; SD: standard deviation; SCID‐IV: Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, 4th Edition; SDS: Zung Self‐Rating Depression Scale; sMMSE: Short Mini‐Mental State Examination.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Ahmadpanah 2017	Ineligible setting. Study set in a day centre (confirmed by study author via email)
Crespy 2016	Ineligible intervention – stepped care
Fry 1984	Ineligible setting – supervised retirement apartment
Hsu 2009	Ineligible design – quasi‐RCT
Hsu 2019	Ineligible design – quasi‐RCT (confirmed by author via email)
Kim 1999	Ineligible design – non‐equivalent group design; participants not randomly assigned to groups
Konnert 2009	Ineligible design – quasi‐RCT
Sood 2003	Ineligible setting – rehabilitation unit within a nursing home
Tesky 2019	Ineligible intervention – case management, with potential for medication use
Wilson 2010	Ineligible patient population – no diagnosis of depression or cut‐off on a validated depression scale used to select participants

RCT: randomized controlled trial.

Characteristics of studies awaiting classification [ordered by study ID]

Bai 2018.

Methods	RCT
Participants	Elderly in care homes
Interventions	Group nostalgia therapy vs individual nostalgia therapy
Outcomes	Depressive symptoms, loneliness
Notes	49 in each group Contacted authors to request more information

Funaki 2011.

Methods	Unclear
Participants	Unclear
Interventions	Group reminiscence therapy
Outcomes	15‐item Geriatric Depression Scale, Life Satisfaction Index A, Multidimensional Observation Scale for Elderly Subjects
Notes	Conference abstract. Contacted study authors for more details.

Khezri Moghadam 2018.

Methods	Appears to be a randomized controlled trial
Participants	Long‐term care residents with depression, but tool used and cut‐off scores to define depression were not specified.
Interventions	Group cognitive existential therapy. Only limited information about the intervention provided in the paper.
Outcomes	Appeared to be depressive symptoms and hope (although translated as 'life expectancy')
Notes	Published English translation does not appear to contain all information. Emailed author for additional information 14 March 2022.

Lemos 2016.

Methods	Randomized controlled trial
Participants	Unclear
Interventions	Neuropsychological group rehabilitation programme
Outcomes	Geriatric Depression Scale
Notes	Conference paper. Contacted author for more information

Lopes 2017.

Methods	Unclear; possibly randomized controlled trial
Participants	Unclear
Interventions	Reminiscence therapy
Outcomes	Geriatric Depression Scale, Geriatric Anxiety Inventory
Notes	Conference paper. Contacted author for more information.

Pang 2020.

Methods	Randomized controlled trial
Participants	Elderly people with depression living in a nursing home
Interventions	Supportive self‐management
Outcomes	Symptoms of depression
Notes	Contacted authors to request more information

Sturz 2015.

Methods	Unclear
Participants	Unclear
Interventions	Cognitive training vs relaxation training
Outcomes	Unclear
Notes	Conference paper. Contacted authors for more information.

Characteristics of ongoing studies [ordered by study ID]

Bhar 2023.

Study name	ELders AT Ease Program (ELATE): a cluster randomised controlled trial of a sustainable and scalable mental health service for Australian residential aged care facilities
Methods	Cluster‐randomized controlled trial
Participants	Long‐term care residents aged ≥ 65 years, with significant depressive symptoms and no significant cognitive impairment
Interventions	3 components: 1. counselling for resident (including cognitive behavioural therapy, behavioural activation, and reminiscence therapy); 2. family support group; and 3. staff consultative meetings.
Outcomes	Depressive symptoms (9‐item Patient Health Questionnaire); anxiety symptoms (Geriatric Anxiety Inventory), quality of life (Quality of Life – Alzheimer's Disease, for Nursing Homes and EuroQol 5‐Dimensions 5‐Levels), suicidal ideation.
Starting date	19 November 2019
Contact information	Prof Sunil Bhar, Department of Psychological Sciences, Faculty of Health, Arts and Design Swinburne University of Technology, John Street, Hawthorn, Victoria, 3122, Australia Email: sbhar@swin.edu.au
Notes	Trial still in progress

Velasquez Reyes 2019.

Study name	Behavioural activation in nursing homes to treat depression (BAN‐Dep)
Methods	Pragmatic 1‐arm parallel cluster‐randomized controlled trial
Participants	666 residents aged ≥ 60 years from 100 LTC facilities
Interventions	Behavioural activation (delivered by training facility staff) vs control staff training intervention
Outcomes	Primary outcome: depressive symptoms using 9‐item Patient Health Questionnaire
Starting date	Not reported
Contact information	Osvaldo P Almeida Email: osvaldo.almeida@uwa.edu.au
Notes	Author reported that the study was completed, but results were not available in October 2021. The study has subsequently been published (see Almeida 2022 under Velasquez Reyes 2019).

Differences between protocol and review

We made the following changes from the protocol (Davison 2018).

Two new authors joined the review team (PO, SBo). These authors were responsible for the meta‐analysis.

In the protocol, it was planned for two review authors to screen records, but due to the large number of records elicited by the search strategy, multiple review authors were involved in screening records. However, the screening was still completed in duplicate, with a third review author to resolve any discrepancies and involvement of other team members if required. In addition, the review team deliberated around the eligibility of studies prior to meta‐analyses, to confirm which studies could be classified as psychological therapies and which settings could be classified as LTC. A subgroup of five review authors was formed for this deliberation (TD, SBh, PO, YW, LF) for this deliberation.

During the screening of studies, other therapies that were not clearly psychological therapies were defined, and added to the list of excluded treatments. Excluded treatments added after the protocol was published included pet therapy, and stepped care or case management interventions where psychological therapy was included as only one component in a multicomponent treatment protocol. Interventions involving group discussions or social visits from peers or volunteers (e.g. befriending) without therapist involvement were not considered psychological therapies, but met criteria for comparator interventions. Interventions delivered by trained volunteers were similarly excluded.

The protocol stated a plan to use a piloted data extraction form, based on the Cochrane Effective Practice and Organisation of Care (EPOC) data extraction form (EPOC 2017), to extract study characteristics and outcome data. Instead, we extracted study data using standard data extraction forms in Covidence. Similarly, the risk of bias assessment for each study was conducted using the Cochrane RoB 1 tool on Covidence.

The protocol stated that the main comparisons would include psychological therapies versus 1. treatment as usual; 2. waiting list; 3. non‐specific attentional control; and 4. alternative active psychological therapy, as well as 5. psychological therapies plus co‐intervention versus co‐intervention alone. Instead, only one main comparison was conducted: psychological therapies versus all control conditions (either treatment as usual, waiting list, or non‐specific attentional control). Subgroup analyses were then conducted, comparing psychological therapies with each type of control group (treatment as usual, waiting list, and non‐specific attentional control) on the primary outcome. This approach is consistent with recent Cochrane reviews of psychological therapies (e.g. Uphoff 2020). We did not conduct the comparison of psychological therapies versus alternative active psychological therapy, due to the limited number of studies (two) which tested alternative psychological therapies. Karimi 2010 compared two types of reminiscence therapy, and Hussian 1979 recruited only six participants per arm, limiting the conclusions that could be drawn. However, subgroup analyses allowed for individual therapy types (cognitive behavioural therapy, reminiscence therapy, behavioural therapy) to be assessed against control conditions. We did not conduct the comparison of psychological therapies plus co‐intervention versus co‐intervention alone due to the absence of studies with this design.

The protocol stated that meta‐analyses of data from cluster randomized controlled trials would be adjusted with an intracluster correlation coefficient (ICC), using 0.1 if the ICC was not reported or was not available from the study authors. None of the three cluster‐RCTs in this review reported an ICC and we were unable to obtain this information from study authors. We decided to use an ICC of 0.03, based on estimates from other cluster‐RCTs of interventions to reduce symptoms of depression in LTC (Davison 2021b; Underwood 2013). This lower value is more consistent with the broader clinical trials literature, with Higgins 2019 noting that "values lower than 0.05 are typical" (p. 573).

Subgroup analysis of the effect of LTC facility staff involvement in the psychological therapy was planned, comparing interventions delivered with and without involvement of staff from the LTC facility as assistants. An additional category was identified in the included studies – interventions delivered by LTC staff who were trained to deliver the intervention as the primary facilitator, rather than the use of professional therapies. The effect of psychological therapies delivered by professional therapists or trainee therapists was expected to differ from the effect of psychological therapies delivered by LTC facility staff, who typically have limited training in mental health and psychotherapy. However, data paucity precluded inclusion of this subgroup in analyses.

The protocol stated that end‐of‐intervention outcomes would be reported in the summary of findings table. However, we also included reporting of outcomes at short‐term follow‐up (up to three months) in Table 1, where there were sufficient data. Outcomes beyond the immediate post‐treatment date, and potentially some weeks without engagement with a therapist may be critical for decision‐making.

Contributions of authors

TD and SBh conceptualized the review.

TD, EY, and CD drafted the protocol, with contributions from SBh, YW, and LF.

YW and TD co‐ordinated the searches and screening.

YW, TD, EY, SBh, and CD screened the records.

TD and SBh extracted the data and drafted the description of studies section of the results.

EY, CD, TD, and SBh conducted the risk of bias assessments and EY, CD, and TD drafted the risk of bias section of the results.

PO and SBo conducted the meta‐analysis.

PO drafted the effects of intervention section of the results and prepared the summary of findings table.

TD, PO, SBh, and YW drafted the discussion and conclusions.

TD drafted the abstract

YW drafted the plain language summary.

All authors contributed throughout the conduct of the review and meta‐analysis, edited the review, and approved the submission.

Sources of support

Internal sources

• None, Other

  N/A

External sources

• None, Other

  N/A

Declarations of interest

TD: reports publishing a literature review and opinion piece with relevance to the topic of the effectiveness of psychological therapies in long‐term care in Australian Psychologist which advocates for increased access to psychologists in this setting, and involvement in an ongoing trial that is eligible for this review (Bhar 2023).

SBh: reports involvement in an ongoing trial that is eligible for this review (Bhar 2023).

YW: none.

PO: none.

EY: none.

CD: receives funding from the National Health & Medical Research Council to investigate the effect of a social intervention on depression in nursing home residents but the trial is ongoing and not eligible for inclusion in this review.

SBo: none.

LF: reports being a Cochrane Editor (specifically in the Cochrane Dementia and Cognitive Improvement Group), with no involvement in the editorial process of the article, and also reports involvement in an ongoing trial that is eligible for this review (Velasquez Reyes 2019).

New