Impact of caregiver readiness on outcomes of a nonpharmacological intervention to address behavioral symptoms in persons with dementia

Abstract

Background

Previous research shows that nonpharmacological strategies may effectively manage behavioral symptoms (agitation, wandering) in persons with dementia and improve caregiver wellbeing. However, strategies depend upon caregivers for their implementation. We examine the impact of caregiver readiness to use nonpharmacological strategies on treatment outcomes.

Methods

Data were from a randomized trial involving 110 family caregivers in the treatment group which received nonpharmacologic strategies for managing behavioral symptoms. Interventionists rated caregiver readiness to use nonpharmacologic strategies as pre-action (precontemplation, contemplation, preparation) or action at treatment initiation and conclusion. Caregivers in pre-action and action stages by treatment conclusion (16 weeks) were compared on proximal (frequency of, and caregiver upset and confidence with targeted behaviors) and more distal (caregiver burden and wellbeing) outcomes at 16 and 24-week follow-ups.

Results

By treatment conclusion, 28.2% (n = 31) and 71.8% (n = 79) of caregivers were rated at pre-action and action respectively. Means for proximal outcomes differed between the groups at 16 and 24 weeks; those at action showed greater improvement on all proximal and distal outcomes. Hierarchical regressions showed significant relationships of readiness to targeted outcomes. By 24 weeks, caregiver readiness predicted lower frequency estimates of targeted behaviors (β = −.180, p = .041) and higher confidence (β = .27, p = .009). Readiness was not a significant predictor of caregiver burden and wellbeing at 16 or 24 weeks.

Conclusion

By treatment conclusion, >25% of participants were not activated to use nonpharmacologic strategies. Activated caregivers reported greater decline in distressing behavioral symptoms, and more confidence than non-activated participants. Activation is needed to impact behavioral management but not other caregiver outcomes.

Introduction

Most people with dementia are cared for at home by family members (Callahan et al., 2012). One of the most challenging aspects of caring for persons with dementia is behavioral symptoms such as agitation, rejection of care or aggression. Also referred to as neuropsychiatric symptoms or behavioral and psychological symptoms of dementia, behaviors are almost universal, occur throughout the disease trajectory and have negative sequelae for both persons with dementia and their family members (Lyketsos et al., 2011). For persons with dementia, behaviors are associated with rapid disease progression, poor quality of life, greater hospital utilization and nursing home placement (Beeri et al., 2002). Likewise, for caregivers, behaviors are associated with upset and depression and more time spent providing care. Given these personal, familial and societal costs, developing and testing ways to help caregivers manage behaviors is a critical public health priority (Fauth et al., 2006; Gitlin et al., 2012).

Most caregiver intervention studies addressing behavioral symptoms report positive outcomes including reductions in caregiver upset and behavioral occurrences (Brodaty and Arasaratnam, 2012). These interventions typically involve imparting to caregivers problem solving skills, cognitive reframing techniques and/or specific nonpharmacological strategies such as simplifying communications, daily tasks and physical environments to reduce demands on persons with dementia. However, the effectiveness of strategies mostly depends upon the willingness of caregivers for their implementation and continued use. Thus, the readiness of caregivers to employ new care techniques is an important consideration in such interventions.

To date, there is little research on the role of caregiver readiness in nonpharmacological interventions. Specifically, it is unclear whether caregiver readiness impacts treatment benefits, and if so, which outcomes are affected. It may be that caregivers with low readiness do not fully use intervention strategies which in turn may result in poor treatment outcomes concerning behavioral symptom management and caregiver wellbeing. In contrast, caregivers with high readiness may be more likely to use nonpharmacologic strategies, resulting in declines in behavioral occurrences and improvements in their own wellbeing.

To understand and evaluate readiness, we previously applied the Transtheoretical Model of Change framework (TTM: Prochaska et al., 1992) to caregiving. We suggested that caregivers with little knowledge of dementia and acceptance of using nonpharmacologic strategies would be at pre-action stages of readiness (e.g. precontemplation, contemplation or preparation). Alternately, those willing to try new strategies would be at an action stage (Gitlin and Rose, 2014). Using these classifications (described below), we showed that in a trial testing a nonpharmacologic approach to behavioral management (Project ACT; Gitlin et al., 2010), caregivers varied widely in readiness; in turn, readiness was associated with different caregiver and patient-related baseline factors. At the beginning of treatment, a higher level of caregiver readiness was associated with better caregiver mood, and less financial difficulty; and for persons with dementia, lower cognitive status and more behavioral symptoms. At treatment conclusion, caregivers at pre-action stages reported greater strain paying for basics than those at action (p < .001) and were more likely to be in spousal relationships (68% spousal; 32% non-spousal) than those at action (46% spousal; 54% non-spousal; p = .036). Also, caregivers in action were more therapeutically engaged (p = .030) and perceived greater intervention benefits (p = .003; Gitlin and Rose, 2014).

Our present study systematically extends this work by examining the relationship of caregiver readiness to specific treatment outcomes of Project ACT. Specifically, we examine the impact of caregiver readiness on five trial outcomes previously shown to improve for caregivers in the intervention group compared to those in the usual care control group (Gitlin et al., 2010). Outcomes include those proximal to intervention (frequency of occurrence of targeted behaviors, and caregiver upset with and confidence managing targeted behaviors); and more global or distal outcomes (caregiver burden and well-being). Outcomes were examined in the original trial and in this study at two time points, 16 weeks from baseline (study enrollment) which was immediately following treatment completion; and 24 weeks or four months from baseline (or two months following treatment completion). Those rated by interventionists as actively using intervention strategies were considered as in an action stage by the conclusion of the intervention and were compared to those who were rated by interventionists as not actively using strategies, reflecting pre-action stages. We reasoned that caregivers in action would be using intervention strategies to address behavioral symptoms and thus have better treatment outcomes (proximal and distal) at both time points compared to those in pre-action.

Methods

Sample and procedures

As reported elsewhere (Gitlin et al., 2007), recruitment for trial participation occurred between December 2003 and March 2007 and involved media announcements and mailings by social agencies. Caregivers were eligible who lived with persons with dementia having a physician diagnosis of dementia or Mini Mental State Examination score <24 (MMSE; Folstein et al., 1975), were ≥21 years, English speaking, planning to live in the area for at least 6 months, not actively seeking nursing home placement, managing behavioral symptoms and reporting upset (>5 on 10 point scale). Dyads were excluded if either had terminal illness with life expectancy <6 months; active treatments for cancer; and >3 acute hospitalizations in past year to assure a physically stable sample. Also excluded were those involved in other behavioral management trials and patients with schizophrenia or bi-polar disorders, dementia secondary to probable head trauma or MMSE = 0 and bed-bound.

Written informed caregiver consent was obtained at baseline using a form approved by the Institutional Review Board. Following baseline, caregivers were randomized to treatment or a no contact (usual care) control group.

Of 272 dyads, 137 were in intervention of whom110 were included in this study, five (3.6%) did not complete any sessions and 22 (15%) had incomplete data on variables of interest. No large or statistically significant differences in baseline characteristics were found for the 110 included and 27 excluded in this study. Data from two sources were used. First, interview data collected at baseline, 16 and 24 weeks by assessors (not interventionists) blinded to participants’ group allocation were used to capture characteristics of caregivers and persons with dementia and study outcomes. Second, interventionists, occupational therapists, rated caregiver readiness at initial and final (16-week) treatment sessions.

Intervention

The intervention involved up to 11 in-home 1-h sessions involving nurse and occupational therapy visits over 16 weeks. The nurse met with caregivers to provide education about common concerns in dementia care (dehydration, pain, taking care of self). Also, the nurse met with persons with dementia, conducted a visual inspection of skin integrity and brown bag medication review and also drew blood and urine to rule out underlying medical contributors to behaviors. The nurse also contacted caregivers by telephone to discuss laboratory results in addition to mailing results and conducting any necessary follow-up with physicians. Results of this aspect of the intervention have been reported elsewhere (Hodgson, Gitlin, Winter, and Czekanski, 2011).

Occupational therapists met with caregivers over nine in-home sessions to identify behavioral symptoms most distressing to families and potential modifying precipitating factors, and instruct caregivers in nonpharmacological strategies to address targeted behavioral symptoms. Strategies might include having caregivers simplify communications, tasks and the home environment and introducing and using activities of meaning to persons with dementia (Gitlin et al., 2010).

Measures

Background characteristics

Caregiver characteristics (age, sex, race [white/non-white], relationship to person with dementia [spouse/nonspouse] and education [<high school, high school, >high school]) were examined. For persons with dementia, we examined age, gender and cognitive status using the MMSE, a 30-item screen for cognitive impairment (Folstein et al., 1975). Correct responses are summed. Higher scores indicated higher cognitive functioning. Caregivers also rated difficulty paying for basics (food, housing, medical care, heating) as 1 = “not difficult at all”, 2 = “not very difficult”, 3 = “somewhat difficult” and 4 = “very difficult”.

Readiness measure

As readiness scales for nonpharmacological interventions do not exist, we developed a rating approach modeled on TTM (Prochaska et al., 1992). This involved a two-dimensional measure reflecting caregiver understanding of behaviors as dementia-related, and willingness to try strategies to manage them. We developed behavioral anchors for four readiness levels adapted from behavioral change studies with older adults (Burbank et al., 2000; McNulty et al., 2004): 1 = precontemplation indicating lack of knowledge of dementia and behavioral symptoms, a belief that nothing will help or unwillingness to change; 2 = contemplation indicating caregivers acknowledged behavioral symptoms were dementia-related, understood behaviors were not intentional manipulations, were thinking about change but not ready to take action; 3 = preparation reflecting awareness that behaviors could be better managed, and intent to try new strategies; 4 = action reflecting awareness and active efforts to use new behavioral management strategies. We did not include the final stage of readiness of the original conceptualization, maintenance of behavior change, as we were unable to evaluate long-term use of strategies.

Although dependent upon ratings from interventionists, we previously demonstrated construct validity of readiness, consistent with its application to other health behaviors (Prochaska et al., 1988; Prochaska et al., 1992; Armitage, 2006; Rose, Gitlin, and Dennis, 2010; Gitlin and Rose, 2014).

As few caregivers were at precontemplation (n = 1), contemplation (n = 7) and preparation stages (n = 23) by conclusion of the intervention, we collapsed these categories and labeled them as pre-action stages (n = 31) and compared them to those in action (n = 79) for all study analyses.

Although interventionists (occupational therapists) rated caregiver readiness at initial and final (16-weeks) treatment sessions, for this study, we used only final ratings. Ratings were based on clinical interviewing and direct observations of caregiver–patient interactions. Scores were reviewed at bi-monthly research team meetings. Interventionists presented cases and readiness scores for team discussion with unanimous agreement reached. Evidence supporting stability (reliability) is suggested by the correlation between readiness ratings at initiation and conclusion of treatment (r = .467; p < .001).

Outcome measures

Proximal or immediate outcomes of the intervention included frequency of occurrence of targeted behaviors, and level of caregiver upset and confidence. To identify targeted behaviors, we used a standard method employed in other trials (Battle et al., 1966). At baseline, caregivers indicated whether each of 23 behaviors occurred (yes/no) in the past month using standardized scales (Logsdon et al., 1999; Teri et al., 1992). For each occurring behavior, caregivers indicated frequency of occurrence in the past month and then identified up to four behaviors most distressful to them and for which they sought strategies to manage better. Percentages of those identifying one behavior was .9% (n = 1), two behaviors, 3.6% (n = 4), three behaviors, 10% (n = 11) and four behaviors 85.5% (n = 94). At follow-up (16 and 24weeks), caregivers were asked the frequency of occurrence of the targeted behaviors identified at baseline. A mean total frequency score was computed across targeted behaviors for baseline, 16 and 24weeks.

For each targeted behavior, caregivers also rated their level of upset with (0 = no upset to 10 = extreme upset with midpoint labeled “fairly upset”) and confidence in managing it (0 = not confident to 4 = extremely confident). A total mean upset score and confidence score was computed across the four targeted behaviors at baseline, 16 and 24-week assessments.

We reasoned that with reductions in behaviors and upset and improved confidence, caregivers would also experience improvements in burden and wellbeing, considered distal intervention outcomes. Burden was assessed using the Zarit Burden 12-item measure in which caregivers rate each item on 5-point scales (0 = “Never” to 4 = “Nearly always”; Bédard et al., 2001). A total score was derived by summing scores across 12 items with higher total scores indicating greater burden (α = .87, .85, .87 for baseline, 16 and 24 weeks, respectively).

Wellbeing was assessed using the 13-item Perceived Change Index in which caregivers rated change (1 = “got much worse” to 5 = “improved a lot”) over the past month in three domains: managing care challenges (5 items), affect including anger and stress (4 items) and somatic symptoms including energy and sleep quality (4 items) (Gitlin et al., 2006). A total score was computed representing an average response across 13 items with higher scores indicating improvement (α = .91, .94, .93 for baseline, 16 and 24 weeks, respectively).

Data analysis

To examine the relationship between readiness to use intervention strategies and treatment outcomes, we first computed means and standard deviations for all outcome measures by readiness stages (pre-action/action) and time of measurement (baseline, 16, 24 weeks).

To examine whether readiness at treatment conclusion predicted treatment outcomes, hierarchical regression analyses were used. In all analyses, relationship to person with dementia (spouse/nonspouse) and financial strain were entered as covariates given our previous research suggesting their relationship to readiness (Gitlin et al., 2014). In each analysis, baseline values of outcomes (frequency, upset, confidence, burden, wellbeing) were entered on the first step, followed by relationship and financial strain on the second step, and caregiver readiness at 16 or 24weeks on the third step. SPSS version 22.0 was used for all analyses.

Results

Sample characteristics

Caregivers had a mean age of 67.08 (SD = 12.53), were mostly female (78%), white (66.4%) and reported ≥high school (61.8%) education. There were slightly more spouse (51.8%) than non-spouse (48.2%) caregivers. Persons with dementia were mostly female (56.4%), with an average age of 82.22 (SD ± 8.72) and MMSE score of 12.91 (SD ± 8.21) (Table 1).

Table 1.

Characteristics of caregivers and persons with dementia by caregiver level of readiness^* (N = 110)

	Pre-action (contemplation and preparation)	Action	t	χ2	p Value	Total
	(N = 31)	(N = 79)
Caregivers
Age, mean ± SD	69.68 ± 14.52	66.06 ± 11.59	1.37		.175	67.08 ± 12.53
Gender, %
Male	20.8	79.2		.82	.365	21.8
Female	30.2	69.8				78.2
Race, %
White	30.1	69.9		.41	.522	66.4
Non-White	24.3	75.7				33.6
Relationship to patient, %
Spouse	36.8	63.2		4.38	.036	51.8
Non-spouse	18.9	81.1				48.2
Education level, %
<High school	40.0	60.0		3.32	.190	9.1
High school	37.5	62.5				29.1
>High school	22.1	77.9				61.8
Financial difficultya	2.71 ± 1.01	2.09 ± .99	2.95		.004	2.26 ± 1.03
Person with dementia
Age, mean ± SD	81.58 ± 7.33	82.47 ± 9.23	−.48		.633	82.22 ± 8.72
Gender, %
Male	31.2	68.8		.40	.529	43.6
Female	25.8	74.2				56.4
Cognitive status, mean ± SD	13.61 ± 8.00	12.65 ± 8.33	.27		.603	12.91 ± 8.21

Note: Items in bold indicate significance.

^aRated from 1 = not difficult at all to 4 = very difficult.

^*Refers to caregiver level of readiness rated by interventionists at the conclusion (16 weeks) of intervention.

Readiness measures showed improvement over the course of intervention. At baseline, 66.4% (n = 73) were rated at pre-action whereas only 33.6% (n = 37) were rated at action. By treatment conclusion, 28.2% (n = 31) and 71.8% (n = 79) of caregivers were rated at pre-action and action respectively.

Readiness and treatment outcomes

Proximal outcomes

Means for proximal outcomes differed for those at pre-action and action stages of readiness at 16 and 24 weeks. At 16 weeks, the most striking change occurred in upset; those at action showed larger decreases in upset relative to baseline (M = 6.66 to M = 4.44) than those at pre-action (M = 7.19 to M = 6.51). By 24 weeks, caregivers at action showed decreases in frequency of targeted behaviors relative to baseline measures (M = 34.58 to M = 25.63) whereas, caregivers at pre-action showed an increase (M = 42.71 to M = 47.32) (Figure 1).

Figure 1.

Proximal outcomes for caregivers in pre-action and action stages.

Hierarchical regressions showed significant relationships of readiness to targeted outcomes (Table 2). Caregiver readiness to use intervention strategies was associated with both upset (β = −.303, p = .001) and confidence (β = .220, p = .042) at 16 weeks such that higher readiness was associated with less upset with the targeted behavior and greater confidence managing it. At 24 weeks or two months following treatment conclusion, caregiver readiness also predicted frequency of targeted behaviors (β = −.180, p = .041) and confidence (β = .27, p = .009); higher readiness predicted lower frequency estimates of targeted behaviors and greater confidence in managing those behaviors.

Table 2.

Multiple regression analysis: readiness to use strategies at treatment conclusion as predictor of treatment outcomes at 16 and 24 weeks (N = 110)

	16 weeks		24 weeks
Outcomes	β	p Value	β	p Value
Proximal
Frequency	−.09	.276	−.18	.041
Upset	−.30	.001	−.06	.566
Confidence	.22	.042	.27	.009
Distal
Burden	.00	.982	−.09	.220
Wellbeing	.13	.188	.13	.196

Note: All regression analyses controlled for caregiver relationship (spouse/non-spouse) and financial difficulty level.

Distal outcomes

Distal outcomes produced relatively smaller changes, although means generally favored caregivers in action. However, relative to baseline measures, there were larger decreases in burden at 24 weeks for those at action (M = 20.54 to M = 18.32) compared to those at pre-action stages (M = 21.45 to M = 21.33). Nevertheless, readiness was not a significant predictor of distal outcomes (caregiver burden, wellbeing) at 16 or 24 weeks (Figure 2).

Figure 2.

Distal outcomes for caregivers in pre-action and action stages.

Across all hierarchical regressions, baseline outcome measures were predictive of all targeted outcomes (behavioral frequency, upset, confidence, p_s < .01) and distal outcomes (caregiver burden, wellbeing, p_s < .001), except confidence at 16 weeks (β = .09, p = .365). Relationship to caregiver and financial strain were not significant predictors of targeted outcomes at 16 or 24 weeks; however, relationship to caregiver was a significant predictor of burden at 16 (β = −.21, p = .007) and 24 weeks (β = −.18, p = .013). Spousal relationships were associated with higher burden than non-spousal relationships.

Discussion

This study extends our understanding of the impact of caregiver readiness to use nonpharmacological behavioral management strategies by examining its relationship to treatment outcomes (Rose et al., 2010; Gitlin et al., 2014). Caregivers may volunteer for study participation for a range of reasons which may or may not include learning new ways of providing dementia care. Taking action or enacting strategies provided in an intervention is not a given. Previous research has shown that nonpharmacological strategies are not uniformly adopted by caregivers, and various factors such as caregiver depression may affect study participation and derived benefits (Gitlin et al., 1999). As nonpharmacological interventions are dependent on caregivers to implement strategies, understanding the impact of their level of readiness on treatment outcomes is important for developing effective interventions.

This study shows that most caregivers entered this nonpharmacological intervention at a stage of pre-action and then moved to an action stage by treatment conclusion. However, 28.2% did not become activated, and this impacted some, but not all, treatment outcomes. We found that those at an action stage reported decreased frequency of occurrences of targeted behaviors and caregiver upset and improved confidence managing these behaviors. This suggests that caregivers activated to use intervention strategies may have had more success in addressing distressing behavioral symptoms and consequently better immediate outcomes including fewer occurrences of targeted behaviors, less distress concerning these behaviors and more confidence in their management. In contrast, those in pre-action may be considered in part as treatment failures as they were rated by interventionists as not actively utilizing nonpharmacological strategies by treatment conclusion. Subsequently, this group did not appear to derive the same level of benefit on targeted behavioral outcomes as those in an action stage. Our findings may partially explain why nonpharmacological interventions tend to have small to moderate effect sizes. There may be a subgroup who never or completely accept and use intervention strategies for behavioral management.

In contrast, readiness level did not impact the benefits derived from the intervention for distal outcomes. Caregivers at the highest level of readiness did not differ from those at lower levels of readiness on burden and wellbeing at 16 and 24 weeks. That is, those in action and pre-action at treatment conclusion reported similar benefits in these areas. We previously reported that the main trial resulted in improvements for caregivers in the intervention group in these two areas when compared to caregivers in usual care. Thus, it appears that the enactment of treatment strategies for reducing behavioral symptoms may not be necessary in order for caregivers to derive affective benefits from an intervention. It may be that the therapeutic relationship, validation and support offered through a nonpharmacological intervention suffice for such benefits to caregivers’ own well-being to occur.

The results of this study confirm the importance of considering caregiver readiness when introducing nonpharmacological therapeutic modalities to improve dementia care that are dependent upon caregivers for their implementation. In the present study, clear differences emerged between readiness groups immediately after the intervention and two months later for targeted, proximal outcomes concerning behavioral management and their consequences. As most nonpharmacological approaches depend upon caregivers for their implementation (e.g. removing clutter, using memory boards or signage, follow-up with health professionals, communicating differently), caregiver readiness to use strategies may impact their clinical viability.

As low readiness is a risk factor for failure to use strategies to address behavioral symptoms, identifying predictors of activation and ways to effectively move caregivers along a readiness pathway in future research is critical. In a previous study for example, we found that financial strain predicted readiness level at both treatment initiation and conclusion (Gitlin and Rose, 2014). This suggests that financial strain would be important to address prior to providing behavioral management strategies. As readiness level was not related to whether caregivers experience enhanced wellbeing, identifying the elements of the intervention (e.g. feeling supported by interventionists) that most contribute to this outcome would also be important to pursue in future research.

Several study limitations should be noted. Readiness was measured by interventionists which may have introduced bias in ratings. The desire to perceive low readiness initially and then gains because of the intervention may have influenced ratings. However, there might not be bias in the rating of less action, or a treatment failure. Also, bias may not be a concern as treatment outcomes were collected by independent raters masked to group allocation yet readiness levels differentiated level of benefit at least for proximal outcomes.

Another limitation is that readiness was measured using investigator-developed items. Although content validity was achieved with participating interventionists prior to study initiation, development of a reliable and valid measure and obtaining independent ratings should be a goal of future research. In previous work, we demonstrated preliminary construct validity; that is, readiness was associated with treatment process variables. Also, we have used the same measurement approach with other study populations and have shown its importance relative to the use of other types of nonpharmacological strategies (Rose et al., 2010). Finally, as caregivers are volunteers, their readiness levels may be higher than the caregiver population at-large, potentially limiting generalizability of results. Nevertheless, only 33.6% were rated at the highest readiness level initially or upon study entry, suggesting that participants may be ready to volunteer for a trial but not necessarily to use nonpharmacological strategies requiring behavioral change. Finally, as readiness level was collected on intervention participants only, we do not have comparable control group data.

Despite limitations, this study adds incrementally to previous studies pointing to the importance of considering readiness when examining caregiver participation in a nonpharmacological interventions. Findings also contribute to behavioral change research by examining the types of treatment outcomes most impacted by a caregivers’ level of readiness.

Given the importance of readiness, future research is warranted to determine if tailoring interventions to caregiver readiness enhances treatment effects as it concerns behavioral symptom outcomes and how best to help those in pre-action effectively become activated to use nonpharmacological strategies.

Key points.

• Caregiver readiness is an important construct for understanding caregiver’s willingness to learn nonpharmacologic management of behavioral symptoms in persons with dementia

• Most caregivers move to a high level of readiness at treatment conclusion

• Caregivers at a high level of readiness report reduced occurrences of targeted behaviors, and less upset and greater confidence managing those behaviors compared to caregivers at a low level of readiness at treatment conclusion

• Readiness level does not impact treatment benefits as it concerns improving caregiver wellbeing and reducing feelings of burden.