Screening tools assessing mental illness in primary care: A systematic review

Abstract

Background

To better manage patients with a wide range of mental health problems, general practitioners would benefit from diagnostically accurate and time-efficient screening tools that comprehensively assess mental illness. Therefore, the aim of this systematic review was to identify screening tools that either take a multiple-mental disorder or a transdiagnostic approach. As primary and secondary outcomes, diagnostic accuracy and time efficiency were investigated.

Methods

The data bases MEDLINE, Embase, Cochrane Library, Psyndex and PsycINFO were searched. Studies reporting on multiple-mental disorder or transdiagnostic screening tools used in primary care with adult patients were included. Sensitivity, specificity, positive and negative predictive value served as measures of diagnostic accuracy. Time efficiency was evaluated by the number of items of a screening tool and the time required for its completion and evaluation.

Results

Eleven studies met the inclusion criteria. The majority of screening tools assessed multiple mental disorders separately. A sub-group of screening tools took a transdiagnostic approach by examining the spectrum of mood, anxiety and stress-related disorders. One screening tool used internalised, cognitive/somatic and externalised dysfunction as transdiagnostic domains of mental illness. Mostly, a sufficient sensitivity and specificity was reported. All screening tools were found to be time efficient.

Conclusion

The eleven identified screening tools can support general practitioners to identify patients with mental health problems. However, there was great heterogeneity concerning their diagnostic scope of psychopathology. Further screening tools for primary care are needed that target broad constructs of mental illness, such as transdiagnostic factors or personality dysfunction.

KEY MESSAGES

• Eleven screening tools assessing multiple mental health disorders or taking a transdiagnostic approach in primary care were identified.

• The tools were time efficient, and offer a satisfactory diagnostic accuracy.

• Future research should focus on screening tools that target transdiagnostic factors or maladaptive personality traits as informative constructs.

Introduction

Mental disorders are among the leading causes of the global health-related burden [1]. In 2023, almost half (46%) of respondents of a representative European survey reported an emotional or psychosocial problem [2]. Individuals with mental health problems tend to initially consult their general practitioner (GP) before seeking specialised psychotherapeutic services [3]. Thus, GPs play a crucial role in the management of mental disorders [4,5]. However, recognising mental disorders in primary care is challenging [6]. Potential difficulties may be physician-related (e.g. lack of experience in managing mental disorders [7,8]), patient-related (e.g. difficulties in acknowledging psychological distress [9]) and time-related (e.g. internationally, most GPs spend on average less than 10 minutes with a patient [10]). Another barrier is the high comorbidity rate between mental disorders in primary care [11]. Rather than using diagnostic criteria, the assessment of a patient’s mental health is usually guided by a GP’s general impression of a patient [12].

Due to these barriers, GPs could benefit from time efficient screening tools that reliably complement their diagnostic procedure and account for common mental disorders [13]. Instead of administering several single-disorder questionnaires, screening tools that assess multiple mental disorders could be promising [14]. Particularly useful might be multiple-mental disorder screening tools assessing depression, anxiety and somatoform disorders, as those are the most frequently observed mental disorders in primary care [11,15].

Although addressing comorbidity, the non-specific patterns rather than specific characteristics of mental illness in primary care remain a challenge for multiple-mental disorder screening tools [12,16]. Therefore, a transdiagnostic approach to screening for mental illness may be promising. Instead of defining symptomatic differences between mental disorders, it focuses on their commonalities, operating across or beyond single-disorder categories [17]. Transdiagnostic instruments may be employed in one of two ways: One possibility is to examine a broader spectrum of mental disorder symptoms, however, without entirely neglecting the criteria of conventional diagnostic frameworks (i.e. ICD [18] and DSM [19]) [20]. For instance, instead of screening for anxiety and depression separately, the spectrum of mood and anxiety-related disorders could be assessed. The second possibility is to screen for transdiagnostic factors, operating detached from standard diagnostic taxonomies [20]. Transdiagnostic factors are mechanisms that are hypothesised to underlie a range of mental disorders, contributing to their onset and persistence [21]. Several transdiagnostic factors have been proposed in previous research [22–24]. In terms of screening in primary care, transdiagnostic factors underlying ‘emotion-based disorders’ [25], such as anxiety and depression, may be particularly relevant [26,27]. Emotion-based disorders share an amplified experience of negative emotions (i.e. neuroticism), followed by an aversive reactivity to these emotions and the use of behavioural or cognitive coping strategies aimed at avoiding them [28]. Hence, screening tools assessing neuroticism, emotion regulation and/or emotion-based avoidance may serve GPs as transdiagnostic markers of their patients’ psychological distress.

To date, no other review has systematically searched for primary care screening tools that either assess multiple mental disorders separately or screen across or beyond conventional mental disorder classifications. Therefore, the aim of the current review was to systematically identify screening tools used in primary care which either (1) take a multiple-mental disorder approach or (2) target a transdiagnostic spectrum of symptoms or a transdiagnostic factor underlying mental illness. The primary outcome was the diagnostic accuracy of screening tools. As a secondary outcome, their time efficiency was assessed.

Methods

This systematic review and the following narrative synthesis adhere to the PRISMA guidelines [29]. The study protocol was registered at the PROSPERO international prospective register of systematic reviews (ID: CRD42022382572) [30].

Search strategy

A comprehensive literature search using the databases Medline, Embase, Cochrane Library, Psyndex and PsycINFO was conducted. Further, the reference lists of relevant studies were hand-searched. The last search was performed on the 18 July 2024.

A block-building method was used to create the search string. The four blocks ‘primary care setting’, ‘questionnaire or screening tool’, ‘disorders or transdiagnostic factors’ and ‘diagnostic accuracy measures’ were defined (Appendix 1, Supplementary material).

Eligibility criteria

Studies were included if (1) they reported on a screening tool, i.e. an index test, assessing at least two mental disorders in the categories of depression, anxiety or somatoform disorders, or following a transdiagnostic approach; (2) the index test was compared with a well-established reference standard; (3) diagnostic accuracy measures of the index test were provided (i.e. sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) or values allowing their calculation); (4) the study population consisted of adult patients (5) attending primary care services administered by a GP.

Studies were excluded if (1) the sample consisted of a special sub-group of the general population, e.g. military veterans, (2) the publication was not peer-reviewed and (3) not written in German or English. No restrictions were made with regard to the publication date or study design.

Study selection and data extraction process

Initially, all search results were exported to EndNote [31], with duplicates being automatically removed. The remaining data were exported to Rayyan [32] and deduplicated a second time (automatic and manually by BN). Screening of titles/abstracts and full-texts was performed independently by two authors (BN, KLu) in blind mode. Remaining titles and abstracts were assessed for eligibility [32]. Then, inclusion and exclusion criteria were applied to the screening of full-texts. Disagreements were resolved at each stage through discussion among authors (BN, KLu, CE, KLo). Data extraction was conducted independently by three authors (CE, BN, KLu) using Microsoft Excel (Microsoft Cooperation, 2018). The data extraction form included: study design, characteristics of the study sample (sample size, country, age), type of index test and reference standard used as well as diagnostic scope, diagnostic accuracy and time efficiency (i.e. number of items, time needed for completion and evaluation) of an index test. If full-text articles were unavailable or in case of missing data, the authors of a study were contacted.

Outcome measures

Primary outcome

Diagnostic accuracy was assessed by the sensitivity, specificity, PPV and NPV of the index tests.

As PPV and NPV are influenced by the prevalence (i.e. pre-test probabilities) of a disorder in the study population [33] and the samples of the included studies were not representative, those measures are only reported in Table 1. Also, due to the heterogeneity in the study samples’ characteristics as well as in the reference standards and diagnostic criteria used, no comparison of the diagnostic accuracy scores of the different index tests was performed.

Table 1.

Study characteristics, diagnostic accuracy and time efficiency measures of index tests.

Index test	Study	Population	Sample size	Reference standard	Dimensions of the index test	Corresponding dimension(s) of the reference standard	Cut-off points	Diagnostic accuracy (%)				Time efficiency number of items completion time evaluation time
								SE	SP	PPV	NPV
SDSS-PC	Broadhead et al. (1995)	Primary care patients, USA, 18 - 70 years (M = 38.5)	n = 388	SCID-P (DSM-III)	Major depression, 5 items	Matching diagnosis according to DSM-III	Any 2 items with yes	90.4	77.2	39.7	98.0	16 items not reported not reported
					Generalised anxiety disorder, 2 items		Any 1 item with yes	89.8	54.0	5.4	99.5
					Panic disorder, 3 items		Any 1 item with yes	78.3	80.0	20.8	98.2
					Obsessive compulsive disorder, 2 items		Any 1 item with yes	64.5	72.5	5.1	98.9
					Suicidal ideation, 2 items		Any 1 item with yes	43.6	90.6	50.8	87.8
					Alcohol abuse or dependence, 2 items		Any 1 item with yes	61.8	98.2	53.8	98.7
HADS	Bunevicius et al. (2007)	Primary care patients, Lithuania, 18 - 89 years (M = 52)	n = 503	MINI (DSM-IV /ICD-10)	Depression, 7 items	Major depressive episode	≥ 6	80.0	69.0	80.0	92.0	14 items 2-5 min not reported
					Anxiety, 7 items1	Any anxiety disorder	≥ 9	77.0	75.0	53.0	90.0
					Generalised anxiety disorder	Generalised anxiety disorder	≥ 9	76.0	73.0	49.0	90.0
					Panic disorder	Panic disorder	≥ 11	100.0	77.0	11.0	100.0
					Social phobia	Social phobia	≥ 9	95.0	63.0	9.0	100.0
CMDQ	Christensen et al. (2005)	Primary care patients, Denmark, 18 - 65 years (M = 38.8)	n = 1785	SCAN (ICD-10)	Any mental disorder (i.e. depression and/or anxiety), 8 items	Matching diagnosis according to ICD-10	≥ 3	72.0	72.0	50.0	88.0	37 items 2-5 min <1 min
					Depressive disorder, 6 items		≥ 3	78.0	86.0	39.0	97.0
					Any anxiety disorder, 4 items		≥ 3	39.0	87.0	38.0	88.0
					Any somatoform disorder2, 19 items
					1st subscale, 12 items		≥ 5	65.0	63.0	52.0	74.0
					2nd subscale, 7 items		≥ 2	74.0	65.0	51.0	77.0
					Alcohol abuse or dependence, 4 items		≥ 2	78.0	97.0	34.0	99.0
M-3	Gaynes et al. (2010)	Primary care patients, USA, 18 - 92 years (M = 45.7)	n = 647	MINI (DSM-IV /ICD-10)	Any mood or anxiety disorder (i.e. depression, anxiety, bipolar, post-traumatic stress disorder), 23 items	Matching diagnosis according to DSM-IV/ICD-10	Screening positive for any of the disorder subscales	83.0	76.0	65.0	89.0	27 items 5 min <30 s
					Major depressive disorder, 7 items		≥ 5	84.0	80.0	54.0	95.0
					Any anxiety disorder		≥ 3	82.0	78.0	59.0	92.0
					Generalised anxiety disorder, 2 items
					Panic disorder, 2 items
					Social anxiety disorder, 1 item
					Obsessive compulsive disorder, 3 items
					Post-traumatic stress disorder, 4 items		≥ 2	88.0	76.0	20.0	99.0
					Bipolar disorder, 4 items		≥ 2	88.0	70.0	23.0	98.0
				Not reported	Functional impairment, 4 items	Not reported	≥ 2					Not reported
MBHS	McCord (2020)	Primary care patients, USA, 18 - 79 years (M = 47.6)	n = 166	MMPI-2-RF	Internalising dysfunction							27 items 2.5 min <1 min
					Demoralisation, 3 items	RCd demorali-sation	≥ 3	73.3	78.3	55.9	88.6
					Anhedonia, 3 items	RCd anhedonia	≥ 3	81.0	72.4	50.0	91.7
					Anxiety, 3 items	RC7 dysfunc-tional negative emotions	≥ 5	72.1	75.4	50.8	88.4
					Suicidal tendencies, 3 items	SUI suicide/death ideation	≥ 1	56.3	75.2	48.0	80.7
					Somatic/Cognitive dysfunction
					Somatisation, 3 items	RC1 somatic complaints	≥ 4	75.0	73.0	70.3	77.3
					Cognitive issues, 3 items	COG cognitive complaints	≥ 5	80.8	83.2	68.8	68.4
					Externalising dysfunction
					Activation, 3 items	ACT activation	≥ 3	65.0	50.3	15.2	91.3
					Disconstraint, 3 items	DISC-r disconstraint	≥ 4	76.7	85.2	53.4	94.2
					Substance misuse, 3 items	SUB substance abuse	≥ 4	77.3	82.5	40.5	95.9
ADD	Means-Christensen et al. 2006	Primary care patients, USA, 18 - 70 years, (M = 41.49)	n = 801	CIDI-Auto (DSM-IV)	Major depression, 1 item	Matching diagnosis according to DSM-IV	Item with yes	85.0	73.0	63.6	90.0*	5 items not reported not reported
					Generalised anxiety disorder, 1 item		Item with yes	100.0	56.0	44.8	100.0*
					Panic disorder, 1 item		Item with yes	92.0	74.0	67.0	93.7*
					Social phobia, 1 item		Item with yes	69.0	76.0	51.8	86.7*
					Post-traumatic stress disorder, 1 item		Item with yes	62.0	83.0	47.5	89.4*
					Individual items as indicators for the presence of any of the five disorders		Respective item with yes
					Major depression item			78.0	80.0	78.2*	80.0*
					Generalised anxiety disorder item			87.0	68.0	71.6*	85.4*
					Panic disorder item			72.0	92.0	90.2*	76.1*
					Social phobia item			51.0	93.0	89.1*	63.4*
					Post-traumatic stress disorder item			33.0	91.0	80.0*	55.1*
DUKE-AD	Parkerson et al. 1996	Primary care patients, USA, 18 - 65 years (M = 40.4)	n = 413	CES-D	Depression3	Depressive symptomatology	Raw score of ≥ 5; final score of ≥ 304	73.9	78.2	72.6	79.3	7 items not reported not reported
				SAI	Anxiety	State anxiety		78.7	74.6	62.5	86.7
PRIME-MD	Rickels et al. 2009	Primary care patients, USA, > 18 years (M = 45)	n = 211	PRIME MD clinical evaluation guide (DSM-IV)	Anxious and depressive symptomatology (Anxiety, 2 items; Depression, 2 items)	Matching diagnosis according to DSM-IV	≥ 3	97.5*	64.5*	78.0*	95.2*	4 items 2 min not reported
MHI-5	Means-Christensen et al. 2005	Primary care patients, USA, 18 - 81 years (M = 41)	n = 246	PHQ	Total score, 5 items	Panic and/or major depressive disorder	≤ 23	90.9	57.6	17.4	98.5	5 items 2 min <1 min
				PHQ (Depression subscale)	Depression, 1 item	Major depressive disorder	≤ 4	88.2	61.6	14.6	98.6
				PHQ (Panic subscale)	Anxiety, 1 item	Panic disorder	≤ 4	100.0	65.4	9.9	100.0
				Not reported	Loss of behavioural or emotional control, 2 items	Not reported
					Psychological wellbeing, 1 item
CMFC	Rogers et al. 2021	Primary care patients, USA, 18 - 82 years (M = 47.10)	n = 234	SCID-5-RV (DSM-V)	Major depressive disorder,	Matching diagnosis according to DSM-V						8 initial items + additional items to be completed for diagnostic dimensions screened positively with the initial items5 completion time depends on patients’ responses evaluation is computerised
					2 initial items		Any 1 item with yes	94.0	65.0	41.0	98.0
					27 further items		not reported	45.0	93.0	63.0	0.87
					Generalised anxiety disorder,
					2 initial items		Any 1 item with yes	93.0	63.0	47.0	96.0
					11 additional items		not reported	73.0	89.0	63.0	93.0
					Bipolar disorder,
					1 initial item		Item with yes	63.0	79.0	42.0	89.0
					further items		not reported	50.0	97.0	81.0	89.0
					Attention deficit hyperactivity disorder,
					1 initial item		Item with yes	94.0	61.0	38.0	98.0
					further items		not reported	69.0	86.0	54.0	92.0
					Somatic symptom disorder,
					1 initial item		Item with yes	100.0	78.0	53.0	100.0
					further items		not reported	93.0	86.0	63.0	98.0
					Substance and alcohol use disorder,
					1 initial item		Item with yes	80.0	92.0	70.0	95.0
					further items		not reported	67.0	96.0	79.0	92.0
PDI-4	Houston et al. 2011	Primary care patients, USA, > 18 years (M = 50)	n = 343	SCID (DSM-IV)	Major depressive episode, 4 items	Matching diagnosis according to DSM-IV	At least 3 of 4 X’s must be in grey-shaded regions6	80.0	80.0	58.0	92.0	17 items not reported not reported
					Generalised anxiety disorder, 4 items			83.0	75.0		20.0		98.0
					Mania, 4 items			83.0	82.0	26.0	98.0
				ACDS	Attention-deficit/ hyperactivity disorder (ADHD), 4 items	Matching diagnosis according to ACDS		82.0	73.0	42.0	97.0
					Daily functioning, 1 item7		ADHD diagnosis: “often” other diagnoses:“sometimes”	not reported

Note:.

SE = Sensitivity; SP = Specificity; PPV = Positive predictive value; NPV = Negative predictive value; M = Mean.

SDSS-PC = Symptom-Driven Diagnostic System for Primary Care; HADS = Hospital Anxiety and Depression Scale; CMDQ = Common Mental Disorder Questionnaire; M-3 = My Mood Monitor; MBHS = Multidimensional Behavioural Health Screen 1.0; ADD = Anxiety and Depression Detector; DUKE-AD = Duke Anxiety-Depression Scale; PRIME-MD = Primary Care Evaluation of Mental Disorders; MHI-5 = Mental Health Index-5; CMFC = Connected Mind Fast Check; PDI-4 = Provisional Diagnostic Instrument-4.

SCID-P = Structured Clinical Interview for the DSM-III-R, version P; MINI = Mini International Neuropsychiatric Interview; SCAN = Schedules for Clinical Assessment in Neuropsychiatry; MMPI-2- RF = Minnesota Multiphasic Personality Inventory-2-Restructured-Form; CIDI-Auto = Composite International Diagnostic Interview-Auto; CES-D = Centre of Epidemiologic Studies Depression Scale; SAI = State Anxiety Inventory; PHQ = Patient Health Questionnaire;, SCID-5-RV = Structured Clinical Interview for the DSM-V – Research Version, ACDS = Adult ADHD Clinician Diagnostic Scale version 1.2.

¹ = the diagnostic accuracy of the HADS anxiety subscale was also tested for generalised anxiety disorder, panic disorder and social phobia.

² = the CMDQ comprises two subscales, scored separately, to assess a somatic symptom disorder.

³ = the exact number of DUKE-AD items corresponding to depression and anxiety was not reported.

⁴ = the DUKE-AD provides an overall score for anxiety and depression, but the diagnostic accuracy was calculated separately for both disorders.

⁵ = the exact number of additional items used for the second screening stage of the CMFC was mostly not reported.

⁶ = scoring of patients’ responses is done by using a transparent overlay on the PDI-4 that shows the required frequency level for each symptom to contribute to a provisional diagnosis.

⁷ = to avoid over-diagnosis, the item “often” must be marked on the daily functioning Likert-scale for a provisional diagnosis of ADHD and the item “sometimes” for all other diagnoses.

^*= calculated manually.

Secondary outcome

The time efficiency of the index tests was assessed using the number of items and the time required for their completion by patients and their evaluation by medical staff.

Quality assessment

The QUADAS-2 [34] framework was applied to determine the quality of included studies. Using the four domains ‘patient selection’, ‘index test’, ‘reference standard’ as well as ‘flow and timing’, their risk of bias was determined. For the first three domains, concerns about applicability were additionally examined. The overall risk of bias of a study was judged as ‘low’, ‘high’, or ‘unclear’ (Table 2). Quality assessment was carried out independently by two authors (BN, KLu) and disagreements were solved through discussion.

Table 2.

Quality assessment of included studies using QUADAS-2 framework.

Study	RISK OF BIAS				APPLICABILITY CONCERNS
	PATIENT SELECTION	INDEX TEST	REFERENCE STANDARD	FLOW AND TIMING	PATIENT SELECTION	INDEX TEST	REFERENCE STANDARD
Broadhead 1995	LOW	LOW	LOW	LOW	LOW	LOW	LOW
Bunevicius 2007	LOW	UNCLEAR	UNCLEAR	LOW	LOW	LOW	LOW
Christensen 2005	HIGH	UNCLEAR	LOW	HIGH	LOW	LOW	LOW
Gaynes 2010	LOW	LOW	LOW	LOW	LOW	LOW	LOW
Houston 2011	LOW	UNCLEAR	LOW	LOW	LOW	LOW	LOW
McCord 2020	LOW	UNCLEAR	LOW	LOW	LOW	LOW	LOW
Means-Christensen 2006	LOW	HIGH	UNCLEAR	HIGH	LOW	LOW	LOW
Means-Christensen 2005	LOW	UNCLEAR	UNCLEAR	LOW	LOW	LOW	LOW
Parkerson 1996	LOW	UNCLEAR	LOW	LOW	LOW	LOW	LOW
Rickels 2009	LOW	UNCLEAR	UNCLEAR	LOW	LOW	LOW	LOW
Rogers 2021	LOWS	UNCLEAR	LOW	LOW	LOW	LOW	LOW

Results

Study selection

A total of 5814 studies were identified. After deduplication, the titles and abstracts of 4540 studies were screened. Of those, 45 studies remained for full-text screening. Finally, 11 studies were included (Figure 1).

Figure 1.

PRISMA flow diagram.

Study characteristics

Of the eleven included studies, nine studies were conducted in the USA [35–43], one in Lithuania [44] and one in Denmark [45]. The oldest study was from 1995 [35], the most recent one was published in 2021 [43]. All studies had a cross-sectional design. The sample size ranged between 166 and 1785 primary care patients (median = 388), with an overall mean age of 44.3 years.

The following 11 index tests were identified: Symptom-Driven Diagnostic System for Primary Care (SDDS-PC) [35], Hospital Anxiety and Depression Scale (HADS) [44], Common Mental Disorder Questionnaire (CMDQ) [45], My Mood Monitor (M-3) [36], Multidimensional Behavioural Health Screen 1.0 (MBHS) [38], Anxiety and Depression Detector (ADD) [40], Duke Anxiety-Depression Scale (DUKE-AD) [41], Primary Care Evaluation of Mental Disorders (PRIME-MD) [42], Mental Health Index-5 (MHI-5) [39], Connected Mind Fast Check (CMFC) [43] and Provisional Diagnostic Instrument-4 (PDI-4) [37].

Eight of the identified index tests assess at least two mental disorders separately [35–37,39,40,43–45]. Four of those additionally [36,39,40,45] and two exclusively [41,42] provide an overall score along the spectrum of mood, anxiety and stress-related disorders. One index test [38] adopts an alternative classification of mental illness by assessing internalising, externalising and somatic/cognitive dysfunction.

A variety of reference standards was used. Psychiatric interviews were conducted in eight studies [35–37, 40, 42–45]. In the three other studies [38,39,41], validated questionnaires based on self-reports served for comparison.

Diagnostic accuracy and time efficiency of index tests

Diagnostic accuracy

Multiple-mental disorder index tests

The sensitivity and specificity scores of all identified index tests, together with their PPV and NPV are presented in Table 1.

Depression can be assessed with the SDSS-PC [35], HADS [44], CMDQ [45], M-3 [36], ADD [40], MHI-5 [39], CMFC [43] and PDI-4 [37]. Their sensitivity and specificity scores ranged from 78.0% (CMDQ) to 94.0% (CMFC, initial items) and from 61.6% (MHI-5) to 86.0% (CMDQ), respectively.

For index tests screening for any anxiety disorder (i.e. HADS [44], CMDQ [45] and M-3 [36]) sensitivity scores between 39.0% (CMDQ) and 82.0% (M-3) and specificity scores from 75.0% (HADS) to 87% (CMDQ) were found. Generalised anxiety disorder can be screened with the SDSS-PC [35], HADS [44], ADD [40], CMFC [43] and PDI-4 [37]. These index tests demonstrated sensitivity scores from 76.0% (HADS) to 100.0% (ADD) and specificity scores from 54.0% (SDSS-PC) to 75.0% (PDI-4). The presence of a panic disorder can be assessed with the SDSS-PC [35], HADS [44], ADD [40] and the MHI-5 [39], with sensitivity scores ranging from 78.3% (SDSS-PC) to 100.0% (HADS, MHI-5) and specificity scores between 65.4% (MHI-5) and 80.0% (SDSS-PC). Social phobia can be determined using the HADS [44] and the ADD [40]. The former reached a sensitivity of 95.0% and a specificity of 63.0%, while for the latter a sensitivity of 69.0% and a specificity of 76.0% were observed.

Somatic symptom disorder can be assessed by the CMDQ [45] and the CMFC [43]. The two somatisation-related sub-scales of the CMDQ showed a sensitivity of 65.0% and 74.0% and a specificity of 63.0% and 65.0%, respectively. For the initial item of the CMFC, a sensitivity of 100.0% and a specificity of 78.0% were stated.

Transdiagnostic index tests

The CMDQ [45], M-3 [36], ADD [40] and MHI-5 [39] additionally as well as the PRIME-MD [42] and the DUKE-AD [41] exclusively screen across the spectrum of mood, anxiety and stress-related disorders. With a reported sensitivity and specificity of 72.0% each, the CMDQ contains a subscale for detecting any mental disorder related to depression and anxiety. A positive screening for one of the M-3 subscales (i.e. depression, anxiety, post-traumatic stress disorder, bipolar disorder) was found with a sensitivity of 83.0% and a specificity of 76.0% to indicate the presence of at least one of those disorders. Similarly, any item of the ADD can be used as an indicator for any of the assessed mental disorders (i.e. depression, generalised anxiety disorder, panic disorder, social phobia, post-traumatic stress disorder). The best sensitivity and specificity ratio is obtained with the panic disorder item (72.0%; 92.0%, respectively). The total score of the MHI-5 resulted in a sensitivity of 90.9% and a specificity of 57.6% to recognise a depression and/or panic disorder. For the overall score of the PRIME-MD, a sensitivity of 97.5% and a specificity of 64.5% was calculated to identify a depressive and anxious symptomatology. The diagnostic accuracy of the total score of the DUKE-AD covering both depression and anxiety was provided for both disorders separately, with sensitivity and specificity scores of 73.9% and 78.2% as well as 78.7% and 74.6%, respectively.

In contrast to the previous index tests, the MBHS [38] functions detached from the ICD and DSM to enquire mental illness. It contains a domain on internalising dysfunction (sub-scales: demoralisation, anhedonia, anxiety, suicidal tendencies), somatic/cognitive dysfunction (sub-scales: somatisation, cognitive issues) and externalising dysfunction (sub-scales: activation, disconstraint, substance misuse). The following sensitivity and specificity scores were derived: demoralisation: 73.3%/78.3%; anhedonia: 81.0%/72.4%; anxiety: 72.1%/75.4%; suicidal tendencies: 56.3%/75.2%; somatisation: 75.0%/73.0%; cognitive issues: 80.8%/83.2%; activation: 65.0%/50.3%; disconstraint: 76.7%/85.2%; substance misuse: 77.3%/82.5%.

Time efficiency

The index tests comprise between 4 and 37 items (median = 14 items). In the CMFC [43], the total number of items depends on the answers of a patient. Initially consisting of eight items, additional questions - ranging from 11 to 27 - are asked if a patient answers ‘yes’ to any of the eight questions.

The HADS [44], CMDQ [45], M-3 [36], MBHS [38], PRIME-MD [42] and MHI-5 [39] were reported to take less than 5 minutes to complete. The completion time of the CMFC [43] depends on the number of positively answered items in the first screening stage.

A timeframe of less than one minute was specified to evaluate the results for the CMDQ [45], M-3 [36], MBHS [38] and MHI-5 [39]. For the CMFC [43], the evaluation is carried out automatically as part of its digital application (Table 1).

Quality assessment

The quality of the eleven identified studies varied in terms of risk of bias. Only two studies were judged as ‘low’ on all domains [35,36] (Table 2).

Discussion

This systematic review aimed to identify multiple-mental disorder and transdiagnostic index tests used in primary care. As outcomes, diagnostic accuracy and time efficiency were assessed. Eleven index tests were identified. Most had a sensitivity and specificity above 70% for correctly identifying a depression, any or a specific anxiety disorder or a somatic symptom disorder. This is comparable to the diagnostic accuracy reported for other commonly used screening tools in primary care, such as the Patient Health Questionnaire-9 (PHQ-9), Generalised Anxiety Disorder Screener-7 (GAD-7) and Patient Health Questionnaire-15 (PHQ-15) [46]. The index tests tended to have a lower specificity than sensitivity. Although acceptable for screening tools [47], the associated risk of misclassifying patients as positive cases [33] could lead to unnecessary stigmatisation [16] and increased workload for GPs [36]. It should be considered, however, that the results of index tests require confirmation by thorough testing using diagnostic tools [36,39,42,44]. A minority of index tests were reported to have a PPV below 10% (i.e. SDSS-PC [35], HADS [44], MHI-5 [39]). One reason could be the lower pre-test probability of mental illness in primary care. Compared with the psychiatric setting, primary care patients usually have not undergone any diagnostic pre-selection, reducing the likelihood that patients tested as positive actually have the disorder [48].

Regarding time efficiency, half of the index tests contain fewer than 14 items. With a maximum completion time of 5 minutes, these index tests can be considered short [49]. Notably, even studies of index tests with more than 14 items did not report a completion time above 5 minutes and an evaluation time exceeding one minute. Given the average primary care consultation time of less than 10 minutes per patient [10], all index tests appear to be administrable. Particularly feasible may be those index tests that take less than 3 minutes to complete [47].

To screen for multiple-mental disorders, the CMFC [43] may be the most promising index test for GPs. It covers a broad range of common mental disorders, i.e. depression, anxiety, somatoform disorder, substance abuse and ADHD [11,50]. In comparison - except of the CMDQ [45] - all other multiple-mental disorder index tests are limited by not including items on somatoform disorders [15,51]. Further, its computerised two-step diagnostic process enables a quick identification of distressed patients, followed by a more in-depth diagnostic investigation of those who screen positive. This may enhance diagnostic accuracy.

Suitable as overall markers for mental illness may be index tests that assess psychopathology dimensionally rather than categorically using a transdiagnostic approach [17]. The CMDQ [45], M-3 [36], MHI-5 [39], ADD [40], PRIME-MD [42] and DUKE-AD [41] each provide a subscale or total score to examine symptoms along the spectrum of mood, anxiety and stress-related disorders. Mostly, sensitivity and specificity scores above 70% was found. By examining a psychopathological spectrum but without neglecting the diagnostic criteria of conventional taxonomies, these index tests can be considered as ‘soft’ transdiagnostic index tests [20]. Overall, transdiagnostic index tests may be more compatible with common screening procedures in primary care, i.e. assessing patients’ mental health condition based on the general impression of GPs rather than on distinct diagnostic criteria [12]. However, the identified soft transdiagnostic index tests may be limited by their heterogeneous scope of psychopathology, e.g. PRIME-MD [42]: anxiety and depression-related pattern; M-3 [36]: dimension of depression, anxiety, bipolar and post-traumatic stress disorder. Consequently, GPs have to decide which screening scope is the best fit for their patients’ symptomatic profile.

The MBHS [38] might be a more practical approach to transdiagnostic screening for GPs. By assessing internalising, somatic/cognitive and externalising dysfunction, the MBHS functions detached from conventional diagnostic classifications and can therefore be labelled as a ‘hard’ transdiagnostic index test [20]. GPs may use the higher-order domains of dysfunction to anticipate a broad range of lower-order mental health problems in their patients (e.g. depression, anxiety, somatoform, substance abuse and ADHD-related symptoms). This approach is similar to other hierarchical taxonomies of psychopathology, such as the Hierarchical Taxonomy of Psychopathology (HiTOP) [52] and the Research Domain Criteria (RDoC) [53]. Notably, the theoretical basis of the MBHS, the Minnesota Multiphasic Personality Inventory (MMPI) [54], suggests another possibility for screening in primary care, i.e. the assessment of maladaptive personality traits and dysfunction. Similar to domains of psychopathological dysfunction, personality traits such as negative affectivity may serve as vulnerability markers of mental illness [55–58].

A functional bridge between higher-order domains of dysfunction and lower-order domains of mental health problems could be established by transdiagnostic factors [59]. Although included in the search strategy, no index tests specifically targeting transdiagnostic factors (e.g. emotion-based avoidance [60,61]) were identified. This indicates an additional area for further research. There is a growing number of transdiagnostic factors [22], many of which overlap in their conceptualisation [62]. Thus, index tests screening for the spectrum of mood, anxiety and stress-related disorders and those targeting transdiagnostic factors should cover broad and meaningful symptoms or mechanisms, e.g. underlying emotion-based disorders [23–25]. Otherwise, transdiagnostic index tests might not offer a substantial benefit to GPs compared to a conventional classification of mental disorders [5,8,12].

Strengths and limitations

This review is the first to systematically search for multiple-mental disorder and transdiagnostic index tests used in primary care, providing a comprehensive overview of the diagnostic scope, diagnostic accuracy and time efficiency of the index tests. Some limitations need to be acknowledged. First, the validity and reliability of the index tests were not investigated in detail. Second, the included studies were cross-sectional in design, had rather small sample sizes, collected data in only one or a few primary care practices and used heterogeneous reference standards. Third, the strict inclusion criteria may have excluded potentially valuable studies that, for example, targeted a different setting [63,64] or did not provide data on diagnostic accuracy [65]. Lastly, all but four of the studies [36–38,43] were published before 2010, which may affect the relevance of the index tests for current practice.

Conclusion

Eleven index tests were identified that take a multiple-mental disorder and/or transdiagnostic approach to screen for mental illness in primary care. All index tests were found to be time efficient and mostly offer a satisfactory diagnostic accuracy. The CMFC [43] may be the most promising multiple-mental disorder index test. Among transdiagnostic index tests, the MBHS [38] can be regarded as an appropriate tool. Future research should focus on screening tools that target transdiagnostic factors or maladaptive personality traits as informative constructs for identifying mental illness in primary care.

Funding Statement

Research reported in this publication was funded by the German Research Foundation (DFG-GrK 2621/POKAL-Kolleg) and endorsed by the German Centre for Mental Health (Deutsches Zentrum für Psychische Gesundheit [DZPG], grant: 01EE2303A).

Disclosure statement

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Data availability

The data supporting this study’s findings are not publicly available due to licencing restrictions. However, data are available upon reasonable request. Data requests may be directed at “Stiftung Allgemeinmedizin – The Primary Health Care Foundation” (www.stiftung-allgemeinmedizin.de). Mail: office@stiftung-allgemeinmedizin.de