Identification of Preschool Children with Mental Health Problems in Primary Care: Systematic Review and Meta-analysis

Abstract

Objective

Primary care practitioners determine access to care for many preschool children with mental health (MH) problems. This study examined rates of mental health (MH) problem identification in preschoolers within primary healthcare settings, related service use, and MH status at follow-up. The findings may inform evidence-based policy and practice development for preschool MH.

Method

For this systematic review, MEDLINE^®, EMBASE^®, PsycInfo^®, and ERIC ^® were searched from inception to March 7, 2018 for reports in which a screening measure was used to identify MH problems in children aged 24–72 months, seen in primary and community health care settings. Meta-analyses, using random effects models to provide pooled estimates, were used when three or more studies examined identification rates. Findings on service use and persistence of disorders are summarized.

Results

Thirty-five publications representing 21 studies met the inclusion criteria. MH problems were identified in 17.6% of preschoolers (95% Confidence Interval (CI): 11.1–24.1), Q = 4.9, p > 0.1 by primary/community healthcare practitioners. Psychiatric diagnoses were identified in 18.4% of preschoolers (95% CI: 12.3 – 24.4), Q= 1.6, p > 0.1. Based on three studies, parents of 67–72% of identified children received advice and 26–42% received specialist referrals. In the subset of studies examining persistence of MH disorders, 25–67% of identified children had MH disorders after one to three years.

Conclusion

While the identification rate by primary/community practitioners is similar to the diagnostic rate, these may not consistently be the same children. Substantial variability in management and outcomes indicate need for more rigorous evaluation of primary care services for this population.

Background

Mental Health (MH) disorders, as defined here, include manifestations of psychiatric disorders, specifically behavioural, emotional and psychosocial problems that cause distress or functional impairment, and often interfere with important family and social relationships (Gleason et al., 2016). MH disorders can be identified early in childhood, before children attend school (Egger et al., 2006; Franz et al., 2013; Wakschlag et al., 2008; Weitzman et al., 2015; Willoughby, Angold, & Egger, 2008). Many preschool children with MH disorders continue to have difficulties when they enter primary school (Bufferd, Dougherty, Carlson, Rose, & Klein, 2012; Bunte, Schoemaker, Hessen, van der Heijden, & Matthys, 2014; Keenan et al., 2011). Studies done on community samples suggest that symptom presentations, comorbidities and prevalence of common psychiatric disorders in preschool children are similar to those in older children, at rates between 10 and 15% (Egger & Angold, 2006; Kato, Yanagawa, Fujiwara, & Morawska, 2015).

The years between birth and age six include sensitive periods for brain development associated with regulation of emotions and behaviors, and development of social relationships (McCain, Mustard, & Shanker, 2007). Evidence suggests that interventions during this time period for young children at high risk can be cost effective (Karoly, Kilburn, & Canon, 2005). Advocates recommend widespread early identification and intervention in order to ameliorate poor educational, occupational, health and MH outcomes (Center of the Developing Child at Harvard University University, 2010; Weitzman et al., 2015; Whiteford et al., 2013). However, controversy remains with regards to the most effective approach to provide early identification of MH disorders in preschool children in primary and community health care settings.

Public health approaches to early identification may incorporate standardized measures and include universal screening for all young children, case identification among populations at high risk (e.g., those with low birthweight, with depressed mothers or living in high risk neighborhoods) and periodic monitoring or developmental surveillance at health supervision visits. Each approach results in a multi-step process whereby identified children are likely to require additional evaluation prior to referral for specialized assessment and interventions (Foy & AAP Task Force on Mental Health, 2010). Some jurisdictions, such as the Netherlands, Norway, and Hong Kong have public health clinics for young children that provide surveillance for MH problems alongside that for communication and physical development concerns at regular intervals from birth to school entry (Leung, Leung, Chan, Tso, & Ip, 2005; Reijneveld, de Meer, Wiefferink, & Crone, 2008; Wichstrom et al., 2012).

Alternatively, where such infrastructure does not yet exist, primary health care settings offer a promising opportunity for early identification at periodic well child visits (Sawyer et al., 2001). Primary care providers (e.g., community pediatricians, family physicians) are among the first professionals with whom parents discuss concerns regarding their child’s behaviors (Sawyer et al., 2001). As well, in many jurisdictions primary care providers are the ‘gate-keepers’ who refer children for specialized assessments and services (Foy & AAP Task Force on Mental Health, 2010). However, barriers to implementation in these settings exist. The health behaviour screening measures commonly used by physicians have variable accuracy (Sheldrick, Merchant, & Perrin, 2011) and access to evidence-based interventions is often limited (Gleason et al., 2016). In addition, community-based and primary care physicians may lack confidence in their knowledge or skills and have concerns regarding lack of time and resources (Foy & AAP Task Force on Mental Health, 2010).

Three recent systematic reviews focussed on universal screening at a single time point for developmental problems in young children in order to update preventive health care recommendations. The United States Preventive Services Task Force (USTSPF) examined screening for speech and language delay in children under five years (Wallace et al., 2015) and screening for Autism Spectrum Disorder in children 18 to 30 months (Siu et al. 2016). The Canadian Task Force on Preventive Health Care (CTFPHC) examined screening for developmental delay in children ages one to four years (CTFPHC, 2016). Both organizations documented lack of evidence that one time screening with standardized tools improves health outcomes among healthy infants and toddlers. Importantly, no randomized controlled trials examined child health outcomes following use of a screening measure. Furthermore, the questions examined in these systematic reviews did not include screening for behavioural, emotional or psychosocial problems, MH problems that are important potential signals for MH disorders.

Although rigorous reviews of evidence do not support universal use of screening tools for MH problems in young children at this time, at the policy level there is growing agreement about the importance of prevention and early intervention internationally (WHO, 2004; Marklund et al., 2012; Commissioner for Children and Young People WA, 2013). For example, in the United States, the American Academy of Pediatrics and the Society for Developmental and Behavioural Pediatrics produced a clinical report that provides guidance on implementation of early identification and access to care for children with behavioural and emotional problems in pediatric primary care (Weitzman et al., 2015).

Further development of evidence-informed policy and practice could be informed by a systematic examination of relevant literature. This meta-analysis aimed to determine rates of MH problem identification in preschoolers within primary healthcare settings, related service use, and MH status at follow-up. Implications for research and clinical practice are discussed.

Methods

Research Questions

The objectives of this study are to determine rates of identification of MH problems, service use and MH outcomes among preschool children in primary or community health care settings. The study aims are investigated in three stages, using the following research questions.

1. Among preschool children attending primary or community health care settings, aged 24–72 months, what is the prevalence rate of MH problems, defined as emotional, behavioural or psychosocial problems, identified by one of the following methods: 1.1) parent-report screening measures; 1.2) primary care and community practitioners using clinical assessments and 1.3) psychiatric disorders, using Diagnostic and Statistical Manual, third edition, revised, fourth edition, or fifth edition (DSM IIIR, DSM IV, DSM 5) (American Psychiatric Association (APA) 1994, 1987, 2013).

2. What is the prevalence of children identified by one of the above methods, who receive interventions, including consultation from a health clinician, or referral to specialty MH care?

3. What is the prevalence of persistent MH problems, including psychiatric disorders, 12 months or more following identification?

Search Strategy

MEDLINE^® (including to be indexed), EMBASE^®, PsycInfo^®, ERIC (Education Resources Information Center)^® databases were searched from inception (MEDLINE^®, 1946, EMBASE^®, 1947, PsycInfo^®, 1967, ERIC^®, 1966) through March 7, 2018. Strategies developed by a research librarian used combinations of controlled vocabulary (MeSH - medical subject headings) and text words (e.g., behavioral problems, psycho-social problems, disruptive behavior, externalizing / internalizing disorder, total problems or total behavior problems). Articles were identified and selected using standardized forms; data management was conducted using DistillerSR (Evidence Partners Inc., Ottawa, Ontario, Canada). Three investigators (F.M., A.C., A.E.) independently examined titles and abstracts for inclusion, and retained all potentially relevant publications for examination of full text. Two authors (F.M, A.E. or F.M., A.C.) independently reviewed full text of retained articles to determine eligibility. Secondary searches involved hand-searching reference lists of included studies, previous systematic reviews, and guidelines and grey literature e.g. information on websites, and published abstracts of unpublished theses and conference presentations. Disagreements were resolved by discussion. Three primary authors were contacted for clarification regarding inclusion criteria; two replied. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Moher, Liberati, Tetzlaff, Altman, & Group, 2009) reporting guideline was followed. For detailed search strategy see Appendix A.

Selection Criteria

Included articles had the following characteristics i) study design: a cross-sectional and/or longitudinal cohort, a controlled trial, a descriptive survey, or an evaluation of a screening and/or diagnostic instrument, ii) a published abstract available in English, iii) conducted in a primary or community health care setting, iv) used a standardized parent report screening measure for behavioral, emotional or psycho-social problems to identify the study population, v) children 24 to 72 months of age, and vii) a sample size > 50. In addition, study results had to include viii) rates of identification using any one of: standardized screening measure, practitioner assessment, or formal diagnostic criteria, or, to document ix) child or family outcomes, such as service use, presence of MH disorders, or adaptive functioning. In order to examine preschool children newly identified with MH problems in community or primary health care practices, articles were excluded in which children were recruited 1) as infants, 2) with developmental delays or autism spectrum disorder, 3) through school, daycare, and non-health community settings, or 4) by probability sampling from population registries.

Data Extraction

Key study elements, based on the research questions, were extracted from the published reports using standardized forms, and reviewed by a second author to confirm accuracy.

Assessment of Methodological Quality of Included Studies

Studies identified fell into two broad categories. Those describing prevalence rates were evaluated using the Quality Assessment of Diagnostic Accuracy Studies – 2 (QUADAS-2) tool (Whiting et al., 2011; See Appendix B, Table A). Specified domains (patient selection, index test, reference test, flow and timing) were evaluated for risk of bias and applicability to the current research questions (Whiting et al., 2011). Cross-sectional and longitudinal observational studies were evaluated using the Modified Newcastle-Ottawa Scales (Deeks et al., 2003; See Appendix B, Tables B & C). Domains evaluated were sample selection, comparability of groups and ascertainment of outcome. Two authors (F.M., A.C.) independently assessed the risk of bias. Disagreements were resolved by discussion.

No publication was excluded from potential inclusion in meta-analysis or narrative synthesis due to high risk of bias. All papers were included in tables. The authors placed somewhat greater emphasis on articles with low risk of bias when drawing conclusions in the narrative synthesis.

Data Synthesis

Meta-analytic techniques, using random effects models, were used to pool data and generate summary estimates when three or more similar studies provided identification rates by 1) primary care practitioners or 2) psychiatric disorders using diagnostic criteria. Forest plots were constructed using Microsoft Excel (Neyeloff, Fuchs, & Moreira, 2012). Between-study heterogeneity and consistency were evaluated through Cochran’s Q and I² statistics. A high probability of between-study heterogeneity is indicated by Q test when p < 0.10, and a moderate to high level of between-study heterogeneity is indicated by I² values greater than 50% (Higgins, Thompson, Deeks, & Altman, 2003). When indicated, the statistical stability of the resulting estimates was evaluated by removing one study at a time and re-calculating Q and I² statistics.

Role of the funding source

The study funder had no role in design, data collection, analysis, data interpretation or report writing. The corresponding author had full access to all study data and final responsibility for the decision to submit for publication.

Results

Reviewers screened 16,039 titles and abstracts, and assessed 1254 full text articles. Thirty five reports, representing 21 studies, met inclusion criteria and were included in the systematic review (see Figure 1) (Briggs et al., 2012; Brown, Copeland, Sucharew, & Kahn, 2012; Crone, Zeijl, & Reijneveld, 2016; Egger et al., 2006; Fallucco, Robertson-Blackmore, et al., 2017; Fallucco, Wysocki, et al., 2017; Franz et al., 2013; Harwood, O’Brien, Carter, & Eyberg, 2009; Husby & Wichstrom, 2017; Kruizinga, Jansen, van Sprang, Carter, & Raat, 2015; Lavigne, Arend, Rosenbaum, Binns, Christoffel, Burns, et al., 1998; Lavigne, Arend, et al., 1998a, 1998b; Lavigne et al., 1993; Lavigne et al., 1999; Lavigne et al., 1996; Leung et al., 2005; Rai, Malik, & Sharma, 1993; Reijneveld, Brugman, Verhulst, & Verloove-Vanhorick, 2004; Sim et al., 2013; Sourander, 2001; Sveen, Berg-Nielsen, Lydersen, & Wichstrom, 2013, 2016; Takayanagi et al., 2016; Theunissen, Vogels, & Reijneveld, 2012; Theunissen, Vogels, de Wolff, Crone, & Reijneveld, 2015; Theunissen, Vogels, de Wolff, & Reijneveld, 2013; Thompson et al., 1996; Wakschlag et al., 2015; Weitzman, Edmonds, Davagnino, & Briggs-Gowan, 2014; Wichstrom, Belsky, & Berg-Nielsen, 2013; Wichstrom, Belsky, Jozefiak, Sourander, & Berg-Nielsen, 2014; Wichstrom et al., 2012; Wichstrom, Penelo, Rensvik Viddal, de la Osa, & Ezpeleta, 2018; Wiggins et al., 2018). The 21 studies were set in community public health centers, well-child clinics, public health visitor programs in Europe, Great Britain, Hong Kong and Japan, and community-based paediatric practices in the United States (Briggs et al., 2012; Brown et al., 2012; Crone et al., 2016; Egger et al., 2006; Fallucco, Robertson-Blackmore, et al., 2017; Franz et al., 2013; Harwood et al., 2009; Kruizinga et al., 2015; Lavigne et al., 1993; Leung et al., 2005; Rai et al., 1993; Reijneveld et al., 2004; Sim et al., 2013; Sourander, 2001; Takayanagi et al., 2016; Theunissen et al., 2013;Theunissen, Vogels, & Reijneveld, 2012; Thompson et al., 1996; Wakschlag et al., 2015; Weitzman et al., 2014; Wichstrom et al., 2012). Two studies (Lavigne et al., 1996; Wichstrom et al., 2012) each had several companion articles that documented service use and MH outcomes one to three years later (Husby & Wichstrom, 2017; Lavigne, Arend, Rosenbaum, Binns, Christoffel, Burns, et al., 1998; Lavigne, Arend, et al., 1998a, 1998b; Lavigne et al., 1993; Lavigne et al., 1999; Sveen et al., 2013, 2016; Wichstrom et al., 2014; Wichstrom et al., 2018). One article (Theunissen et al., 2012) was a secondary analysis of three distinct samples of five to six year old children. These were drawn from existing datasets, only one of which examined preschool children (Brugman, Reijneveld, Verhulst, & Verloove-Vanhorick, 2001; Reijneveld et al., 2008; Wiefferink et al., 2006). A single pre-post study examined implementation of a screening program in primary care settings (Fallucco, Robertson-Blackmore, et al., 2017; Fallucco, Wysocki, et al., 2017). Study characteristics are described in Table 1.

Figure 1.

Flow Chart

Table 1.

Characteristics of included studies

Author	Date	Country	Type of study	Population (N, Age at BL, % Male)	Setting	Problems Studied/Outcome
Briggs et al.	2012	USA	Longitudinal	3169 screened, 541 enrolled 6–36 months 52% f/u: Attrition 66% at 6–12 months	Urban children’s hospital pediatric clinic	Social-emotional problems
Brown et al.	2012	USA	Cross-sectional	254 3–4 years 48%	Urban primary care physicians office	Social-emotional problems
Crone et al.	2016	Netherlands	Cross-sectional	4776 screened 3870 enrolled, 48.9 % Of these, 2032 were 3–6 years 3–4 years 704 5–6 years 1327	Community child health clinics	Psychosocial problems ∞
Egger et al.	2006	USA	Cross-sectional	1073 screened, 307 in-depth assessments 2–5 years 54%	Pediatric office	DSM IV disorders
Fallucco et al.	2017	USA	longitudinal	1469 3–5 years 53%	Urban pediatric primary care clinics	Behavior and emotional problems
Fallucco et al.	2017a	USA	Cross-sectional	2467* 3–5 years 54%	Urban pediatric primary care clinics	Behavior and emotional problems
Franz et al.	2013	USA	Cross-sectional	3433 screened, 917 in-depth assessments 2–5 years 48.2%	Pediatric office	DSM IV anxiety disorders
Harwood et al.	2009	USA	Cross-sectional	110 3–6 years 64%	Pediatric office	Externalizing behavior problems
Kruizinga et al.	2015	Netherlands	Longitudinal	4073 2 years 51% f/u: Attrition 36% 1 year after BL	Child health care centers	Psychosocial problems ∞
Husby and Wichstrøm, related to Wichstrøm et al., 2012	2017	Norway	Longitudinal	997 4.6 + 0.25 years 50% f/u: Attrition 30% 6 years after BL	Well-child clinic	DSM IV symptoms of ODD and CD
Lavigne et al.	1993	USA	Cross-sectional	3876 screened, 495 in-depth assessments 2–5 years 61%	Community Pediatric offices	DSM IIIR disorders
Lavigne et al.	1996	USA	Cross-sectional	3860 screened, 510 in-depth assessments 2–5 years 60%	Community Pediatric offices	DSM IIIR disorders
Lavigne et al.	1998a 1998b	USA	Longitudinal	510 2–5 years 60% f/u: Attrition 23% 1–3 years after BL	Community Pediatric offices	Outcome: DSM IIIR disorders
Lavigne et al., (1998)	1998	USA	Longitudinal	510 2–5 years 60% f/u: Attrition 24%, 1–3 years after BL	Community Pediatric offices	MH service visits
Lavigne et al.	1999	USA	Longitudinal	510 2–5 years 60% f/u: Attrition 23% 1–3 years after BL	Community Pediatric offices	DSM IIIR disorders, Externalizing problems, Internalizing problems
Leung et al.	2005	Hong Kong	Cross-sectional	942 4 + 0.5 years 54%	Maternal and child health centers	Disruptive behavior problems
Rai et al.	1993	India	Cross-sectional	200 3–6 years 58%	Pediatric outpatient department	Behavior problems
Reijneveld et al.	2004	Netherlands	Cross-sectional	2229 2–3 years 52%	Community child health clinics	Psychosocial problems ∞
Sim et al.	2013	Scotland	Cross-sectional	486 30 months N/A	General practice Home health visitors	Social/emotional/behavioral problems
Sourander, A.	2001	Finland	Cross-sectional	374 3 + 0.3 years 50%	Well-baby clinics	Total problems Externalizing, problems Internalizing problems
Sveen et al., related to Wichstrøm et al., 2012	2013	Norway	Cross-sectional	2475 screened 995 in-depth assessments 4.4 + 0.18 years 49%	Well-child clinic	DSM IV disorders
Sveen et al., related to Wichstrøm et al., 2012	2016	Norway	Longitudinal	2475 screened 1038 in-depth assessments 4.4 + 0.2 years 50% f/u: Attrition 23.5% 2 yrs after BL	Well-child clinic	Outcome: persistent DSM IV disorders
Takayanagi et al.	2016	Japan	Cross-sectional	954 screened, 159 in-depth assessments 5 years 54%	Community health check-up	Attention deficit/hyperactivity disorder symptoms
Theunissen et al. Datasets obtained from A: Brugman et al., 2001 B: Wiefferink et al., 2012 C: Reijnveld et al., 2008	2012	Netherlands	Cross-sectional	Dataset A: 1153 5–6 years 52% Dataset B: 3186 5–6 years 51% Dataset C: 1049 5–6 years 50%	Community child health clinics	Psychosocial problems ∞
Theunissen et al.	2013	Netherlands	Cross-sectional	839 3–4 years 51%	Community child health clinics	Psychosocial problems ∞
Theunissen et al.	2015	Netherlands	Cross-sectional	1650 3–4 years 52%	Community child health clinics	Psychosocial problems ∞
Thompson et al.	1996	UK	Cross-sectional	1047 3 years N/A	Home health visits/community health team	Total behavior problems
Wakschlag et al.	2015	USA	Longitudinal	1857 screened, 1857 screened, 497 enrolled 425 in-depth assessments 48.9% f/u: Attrition 19% 15 months after BL	Urban pediatric primary care clinics	Irritability, impairment
Weitzman et al.	2014	USA	Cross-sectional	378 1–4 years 56%	Pediatric primary care center	Socio-emotional/behavior problems
Wichstrøm et al.	2012	Norway	Cross-sectional	2475 screened 995 in-depth assessments 4.4 ± 0.18 years 49%	Well-child clinic	DSM IV disorders
Wichstrøm et al.	2013	Norway	Longitudinal	1000 4.4 ± 0.18 years 49% f/u: Attrition 20% 2 years after BL	Well-child clinic	Outcome: DSM IV anxiety disorders
Wichstrøm et al.	2014	Norway	Longitudinal	995 4.4 ± 0.18 years 49% f/u: Attrition 20% 2 years after BL	Well-child clinic	Outcome: MH Service use DSM IV disorders, impairment
Wichstrøm et al.	2018	Norway	Longitudinal	995* 4.7 ± 0.30 years 49% f/u: 4 and 6 years after BL	Well-child clinic	Outcome: DSM IV disorders
Wiggins et al.	2018	USA	Longitudinal	1857 screened, 497 enrolled 425 in-depth assessments 4.66 ± 0.85 years 48.9% f/u: Attrition 27% 3–5 years after BL	Urban pediatric primary care clinics	Outcome: Irritability, impairment, DSM 5 disorders

ADHD: Attention deficit hyperactivity disorder; BL: Baseline; ODD: Oppositional defiant disorder; CD: Conduct disorder; DSM IIIR: Diagnostic and Statistical Manual third edition, revised; DSM IV: Diagnostic and Statistical Manual fourth edition; DSM 5: Diagnostic and Statistical Manual of Mental Disorders, fifth edition; f/u: follow up; m: months; MH: Mental Health; N: number in sample; N/A: not available; SAD: Separation anxiety disorder

^∞“Psychosocial problems” incorporates behavioral and emotional problems

^#The letter next to number indicates publication is one of several associated studies from same dataset and research group

^*Fallucco 2017a only Cohort A included in review; Wichstrøm et al. 2018 only Norwegian cohort included in review

Sixteen studies examined identification of preschool MH problems broadly, representing both externalizing and internalizing symptoms and early delays in socio-emotional development (Briggs et al., 2012; Brown et al., 2012; Crone et al., 2016; Egger et al., 2006; Fallucco, Robertson-Blackmore, et al., 2017; Kruizinga et al., 2015; Lavigne et al., 1996; Rai et al., 1993; Reijneveld et al., 2004; Sim et al., 2013; Sourander, 2001; Theunissen et al., 2013; Theunissen et al., 2012; Thompson et al., 1996; Weitzman et al., 2014; Wichstrom et al., 2012). See Table 1. Four studies examined identification of externalizing behaviors only (Harwood et al., 2009; Leung et al., 2005; Takayanagi et al., 2016; Wakschlag et al., 2015) and one specifically examined anxiety disorders (Franz et al., 2013).

Quality of Studies

Twelve of 35 articles representing nine different studies, reported identification rates of problems or diagnoses or both and were evaluated using the QUADAS -2 (Egger et al., 2006; Fallucco, Wysocki, et al., 2017; Franz et al., 2013; Lavigne et al., 1993; Lavigne et al., 1996; Reijneveld et al., 2004; Sveen et al., 2013; Takayanagi et al., 2016; Theunissen et al., 2012; Theunissen et al., 2015; Theunissen et al., 2013; Wichstrom et al., 2012). Of these, eight articles included sufficient details in three or more of the four domains to evaluate risk of bias (Egger et al., 2006; Fallucco, Wysocki, et al., 2017; Lavigne et al., 1996; Reijneveld et al., 2004; Sveen et al., 2013; Theunissen et al., 2015; Theunissen et al., 2013; Wichstrom et al., 2012). Seven articles provided sufficient details to evaluate applicability to the current review questions (Lavigne et al., 1993; Lavigne et al., 1996; Reijneveld et al., 2004; Sveen et al., 2013; Theunissen et al., 2015; Theunissen et al., 2013; Wichstrom et al., 2012). Six of the articles included sufficient details to evaluate both risk of bias and applicability to the current review questions; (Lavigne et al., 1996; Reijneveld et al., 2004; Sveen et al., 2013; Theunissen et al., 2015; Theunissen et al., 2013; Wichstrom et al., 2012) one article showed high risk of bias in more than one domain evaluated (Takayanagi et al., 2016). See Appendix B, Table A.

Nine articles were cross-sectional observational designs and were evaluated using the Newcastle – Ottawa Scale modified for cross-sectional studies (Brown et al., 2012; Crone et al., 2016; Harwood et al., 2009; Leung et al., 2005; Rai et al., 1993; Sim et al., 2013; Sourander, 2001; Thompson et al., 1996; Weitzman et al., 2014). Six articles provided sufficient details regarding all areas evaluated and met criteria for low risk of bias (Brown et al., 2012; Crone et al., 2016; Leung et al., 2005; Sourander, 2001; Thompson et al., 1996; Weitzman et al., 2014). See Appendix B, Table B. Fourteen articles described longitudinal analyses, and represented six studies; (Briggs et al., 2012; Fallucco, Robertson-Blackmore, et al., 2017; Husby & Wichstrom, 2017; Kruizinga et al., 2015; Lavigne, Arend, Rosenbaum, Binns, Christoffel, Burns, et al., 1998; Lavigne, Arend, et al., 1998a, 1998b; Lavigne et al., 1999; Sveen et al., 2016; Wakschlag et al., 2015; Wichstrom et al., 2013; Wichstrom et al., 2014; Wichstrom et al., 2018; Wiggins et al., 2018) twelve of these articles met quality criteria for low risk of bias (Briggs et al., 2012; Husby & Wichstrom, 2017; Kruizinga et al., 2015; Lavigne, Arend, et al., 1998a, 1998b; Lavigne et al., 1999; Sveen et al., 2016; Wakschlag et al., 2015; Wichstrom et al., 2013; Wichstrom et al., 2014; Wichstrom et al., 2018; Wiggins et al., 2018). See Appendix B, Table C.

Research Question 1.1: Identification with standardized parent report measures

Prevalence rates of MH problems in preschool children identified by parent report measures ranged from 5.2% to 35.0%. See Table 2. The broad range represents heterogeneous methods, including use of different standardized screening measures, different subscales of these measures, and different thresholds on the same scale depending on the intent of the study. The most commonly used measure was the Child Behavior Checklist (CBCL) (Achenbach, 2010). The preschool version is a valid and reliable 100 item parent report scale for differentiating children with behavior and emotional problems from those without, normed for a non-clinical population. The measure has two broad-band scales, externalizing and internalizing, as well as a total problems scale. The CBCL was used in nine of 21 studies (Crone et al., 2016; Egger et al., 2006; Franz et al., 2013; Kruizinga et al., 2015; Lavigne et al., 1996; Reijneveld et al., 2004; Sourander, 2001; Theunissen et al., 2012; Theunissen et al., 2013). Seven of these used the CBCL total problems scale, (Crone et al., 2016; Egger et al., 2006; Kruizinga et al., 2015; Lavigne et al., 1996; Reijneveld et al., 2004; Theunissen et al., 2012; Theunissen et al., 2015) and four (Lavigne et al., 1996; Reijneveld et al., 2004; Theunissen et al., 2012; Theunissen et al., 2013) used the 90^th percentile as a threshold. Consistent use of the 90^th percentile as a threshold on the CBCL total problems scale resulted in reduced variability of parent identified preschool MH problems, ranging from 6.4% – 9.9%, values that might be expected at this threshold. See Table 2.

Table 2.

Identification of preschool children with mental health problems by parent-report screening measures, primary or community health care practitioners, and psychiatric diagnoses

Author/Date	Problem investigated	A- Parent-Report Screening Measures B- Primary or Community Health Care Practitioners C- Psychiatric Diagnoses		Outcome	Findings
		Method	Rate
Briggs et al., 2012	Social-emotional problems!	A- ASQ: SE Threshold: N/A	24 months: 23% 30 months: 28% 36 months: 29%	Normalization of at-risk status	Of children referred to co-located specialist, 24% were rescreened. Of these: 24% received intervention 22% received monitoring 14% referred out to specialists 40% declined service At-risk status improved for those who received intervention vs. declined service
Brown et al., 2012	Social-emotional problemsXdd Xmi	A- ASQ: SE Threshold: N/A	24.0%	Parent acceptance of referral to specialty care	Of parents with children scoring high on socio-emotional problems, 79% would welcome or would not mind a referral for mental health services
Crone et al., 2016	Psychosocial problems!	A- CBCL 1½-5 (total problems scale) Threshold: standardized T score = 60, 84%tile	3 years 9 months: 21.3% 5/6 years: 35.4%	CHP and parent agreement about presence of problems	Agreement about problems: 3 years 9 months: 8.8% 5/6 years: 13.5%
		B- CHP assessment*	3 years 9 months: 14.6% 5/6 years: 48.2%		Predictors of agreement: CBCL score in clinical range, child history of problems
Egger et al., 2006	Behavioral or emotional problemsXdd	A- CBCL 1½-5 (total problems scale) Threshold: 70%tile	28.6%
	DSM IV diagnoses Xdd Xmi	C- PAPA Any disorder, excluding elimination disorders	20.9% Serious Emotional Disorder 13.6%
Fallucco et al., 2017 See also Fallucco et al., 2017a	Behavioral and emotional problems!	A- ECSA Threshold >18	12.4%	Service use 6 months following provider training in screening intervention	Providers who participated in training program to implement screening described their practice PCPs reported Counseling Parents with concerns at most well visits: Before training: 67% 6m after training: 85% Change P < 0.06 PCPs reported Referring to Specialist at most well visits: Before training: 26% 6m after training: 52% Change P < 0.02
Fallucco et al., 2017a	Behavioral and emotional problems!	A- ECSA for cohort A	14.0%
Franz et al., 2013	Anxiety disorders DSM IV diagnoses Xdd Xmi	A- CBCL 1½ -5 (10 item anxious/ depressed scale) Threshold: 75%tile	27.5%
		C- PAPA Any Anxiety disorder	19.4%
Harwood et al., 2009	Disruptive behavior problems	A- ECBI disruptive behavior scale: intensity Threshold: T score = 60	34.0%	Parent acceptance of mental health services: stigma as barrier	Of mothers who identified their child with disruptive behavior, 75% reported stigma was NOT a barrier to accepting mental health services when recommended by physician
Husby and Wichstrøm, 2017 See also Wichstrøm et al., 2012, Sveen et al., 2013, 2016, Wichstrøm et al., 2013, 2014, 2018		C- PAPA, CAPA		DSM IV ODD and CD symptoms at ages 6, 8, 10 years (ODD, CD, ADHD, Anxiety/Depression)	ODD symptoms predicted CD symptoms 2 years later across age range, with small effect size Modest stability in ODD and CD symptoms, adjusting for comorbidity
Kruizinga et al., 2015	Parent reported Psychosocial problems!	B- Intervention: CHP used BITSEA Comparison: CHP used KIPPI		Cluster RCT: CBCL 1½ -5 (total problems scale) raw score at 12 months following implementation of new screening tool, as part of CHP assessment	Intervention: no change in mean score Comparison: mean score worsened slightly Small effect size favoring intervention Secondary Outcome: Intervention: Referrals to specialist, 5.7% of children, in Comparison: 7.9% of children (p = 0.042).
Lavigne et al., 1993, related to Lavigne et al., 1996 See also Lavigne et al., 1998, 1998a,1998b, 1999	Behavioral or emotional problems!!	A- CBCL 2–3 or 4–16 (total problems scale) Threshold: 90%tile	8.7%	MH service use	Of those identified by practitioner: 69.4% received advice/ ounseling, 41.9 % received referral
		B- Physician Report Form (clinical opinion about presence of emotional/ behavioral or develop-mental problem)			Of those with diagnoses: 25.9% received counseling, 19% received referral
		C- Agreement between two psychologist assessments for probable diagnoses; DSM IIIR	13.0% no V-codes 14.7% including V-codes
Lavigne et al., 1996, primary study for Lavigne et al., 1998, 1998a, 1998b, 1999 See also Lavigne et al., 1993	Behavioral or emotional problems DSM IIIR diagnoses!!	A- CBCL 2–3 or 4–16 (total problems scale) Threshold: 90% tile	8.3%
		C- Agreement between two independent psychologist assessment; probable DD diagnoses; all severity levels for ED: DSM IIIR	21.4% Severe casesb 9.1%
Lavigne et al., 1998 See also Lavigne et al., 1996				MH service use at 1–3 years	Attended > 1 mental health visit: Physician case: 39% Diagnosed case: 28% Predictors of mental health service use were older age and greater impairment when identified
Lavigne et al., 1998a, 1998b See also Lavigne et al., 1996, 1999				Stability of diagnoses and predictors of stable case status over 3.5–4 years	Stable diagnoses found for Children age 2–3 years: 56% Children age 4–5 years: 67% Predictors of stable case status: low family cohesion, low SES, older age, high maternal negative affect, negative life events
Lavigne et al., 1999 See also Lavigne et al., 1996, 1998a, 1998b				Trajectories of behavioral and emotional problems over 3.5–4 years	Total problems; children age 2–3 years: increased children age 4–5 years; decreased Predictors of increased problems: families have more conflict, less cohesion, more negative maternal affect, more negative life events
Leung et al., 2005	Disruptive behavior problems!!	A- ECBI disruptive behavior scale	29.9%
		A- ECBI-Intensity scale Threshold 90%tile	20.7%
		A- ECBI-problem scale Thresholds: 90%tile	13.7%
Rai et al., 1993	Behavior problems Xdd Xch	A- PBCL (behavior problems scale) Scale threshold ≥ 12	22.0%
Reijneveld et al., 2004	Psychosocial problems!	A- CBCL 2–3 (total problems scale) Threshold: 90%tile	6.4%	MH service use	Of identified children: 72.4% received advice 40.7% received referral to another professional
		B- CHP assessment*	Total problems, 9.4% Clinically significant problemsa, 4.6%		24.1% consult with daycare, colleagues or authorities 23.6% follow up
Sim et al., 2013	Social/ emotional/ behavioral problems!	A- SDQ (Total difficulties, 2–4) Threshold: 90%tile	8.8%
Sourander, A., 2001	Behavioral emotional problems!	A- CBCL/2–3 (any syndrome scales) Threshold: 98%tile	7.9%
Sveen et al., 2013 See also Wichstrøm et al., 2012, 2013, 2014, 2018, Sveen et al., 2016, Husby and Wichstrøm 2017	Emotional and behavioral disorders!	A- SDQ (total difficulties, 4–16) Threshold associated with PAPA Diagnoses	5.2%
Sveen et al., 2016 See also Sveen et al., 2013, Wichstrøm et al., 2012, 2013, 2014, 2018 Husby and Wichstrøm, 2017		A- SDQ (total difficulties, 4–16) Threshold associated with PAPA Diagnoses C- PAPA Any DSM IV psychiatric disorder		Persistent DSM IV emotional and behavioral disorders from 4 years to 6 years identified by PAPA	Of those retained in cohort, prevalence: at 4 years 5.8% (4.5–7.6) at 6 years 7.7% (6.1–9.7) Proportion of children with persistent disorders from age 4 to age 6: 26.5% (16.5–39.7)
Takayanagi et al., 2016	ADHD!	A- P-ADHD-RS-IV, Japanese version Threshold: 90%tile	5.2%
		C- Child psychiatrist: Diagnostic parent and child interviews	5.8%
Theunissen et al., 2012 Dataset A: Brugman et al., 2001	Psychosocial problems!	A- CBCL 4–16 (total problems scale) Threshold: 90%tile	9.4%
		B- CHP assessment*	-Total problems, 21.8% -Clinically significant problemsa, 9.5%
Theunissen et al., 2012 Dataset C: Reijnveld et al., 2008	Psychosocial problems!	A- CBCL 4–16 (total problems scale) threshold: 90%tile	8.6%
		B-CHP assessment*	-Total problems, 26.0% -Clinically significant
Theunissen et al., 2013 related to Theunissen et al., 2015	Psychosocial problems!	A- SDQ 3–4 (total difficulties) Threshold associated with > 0.90 specificity against high CBCL 1½ -5 (total problems scale) Threshold: 90%tile	SDQ: 13.8%
Theunissen et al., 2015 See also Theunissen, 2013	Psychosocial problems!	A- SDQ 3–4 (total difficulties) KIPPI, 1–4 (total difficulties) ASQ:SE 36months ASQ:SE 48months (total score) CBCL 1 ½ -5 (total problems scale) Threshold: 90%tile B- CHP assessment	SDQ: 13.8%KIPPI: 13.6% ASQ:SE 36 months: 15.3% ASQ:SE 48 months: 13.2%		SDQ total difficulties scale and ASQ: SE both provide significant added value to CHP assessment for prediction of elevated CBCL 1½ -5 total problems scale.
Thompson et al., 1996	Behavior problems!	A- BCL (behavior problems) Threshold: >10	13.2%
Wakschlag et al., 2015 See Wiggins et al. 2018	Irritability at baseline Xdd	A- MAP - DB Temper Loss Scale	N/A	Symptoms of DSM 5 disorders after 6 months and 15 months	Temper Loss scale scores predicted symptoms of ODD, ADHD, anxiety, depression at 6 and 15 months
		C- PAPA	N/A
Weitzman et al., 2014	Socio-emotional/behavior problems!	A- BITSEA (behavior/ socio-emotional problems) Threshold: 85%tile	19.8%
Wichstrøm et al., 2012 See also Sveen et al., 2013, 2016, primary study for Wichstrøm et al., 2014, Husby & Wichstrøm 2017 related to Wichstrøm et al., 2013, 2018	Emotional /behavioral disorders DSM IV diagnoses!	A- SDQ (total difficulties) Cut offs used to stratify samples: 0–4, 5–8, 9–11, 12–40)	N/A
		C- PAPA Any disorder, excluding encopresis	All severity levels, 13.0% With impairment, 7.1%
Wichstrøm et al., 2013 See also Wichstrøm et al., 2012, Sveen et al., 2013, 2016, Wichstrøm et al., 2014, 2018, Husby & Wichstrøm 2017		C- PAPA Anxiety disorders at 6 years	All severity levels, 7.5 %	DSM IV anxiety disorders at 6 years	Behavioral inhibition, ADHD, parent anxiety, peer victimization and poor social skills were predictors of anxiety at age 6, controlling for initial anxiety
Wichstrøm et al., 2014 See Wichstrøm et al., 2012, 2013, 2018, Sveen et al., 2013, 2016, Husby & Wichstrøm 2017			MH service use at 4 years and at 7 years	DSM IV anxiety disorders at 6 years	Three month rate of use among children with emotional and behavioral disorders, at age 4: 10.7%; at age 7: 25.2%; Predictors of service use at age 7: use of services at age 4, low SES, parental burden, identified need by teacher
Wichstrøm et al., 2018 See Wichstrøm et al., 2012, 2013, 2014, Sveen et al., 2013, 2016, Husby & Wichstrøm 2017		C- PAPA, CAPA DSM IV symptoms at age 8, 10 years (ODD, CD, ADHD, Anxiety/ Depression)		DSM IV disorder symptoms at 8 years and at 10 yearsN	Temperament at age 4 and 6 predicted disorder symptoms at ages 8 and 10 N
Wiggins et al. 2018 See Wakschlag et al., 2015		A- Two items derived from MAP-DB temper loss scale C- K-SADS-PL For Symptoms of DSM 5 disorders		Symptoms of DSM 5 disorders at early school age, mean age 7 years	High irritability at mean age 4, predicts persistent irritability, continued impairment at age 5.5 and 7years, and ODD, DMDD age 7

ADHD: Attention Deficit/Hyperactivity Disorder, ASQ: SE: Ages and Stages: Social-emotional; BASC: Behavior assessment system for children-parent report scale; BCL: Behavior checklist; BITSEA: Brief Infant-Toddler Social and Emotional Assessment; CAPA: the Child and Adolescent Psychiatric Assessment, CBCL: Child behavior checklist; CD: Conduct disorder; CHP: Child healthcare professionals (specialist and nurses); DD: disruptive disorders; DMDD: Disruptive Mood Dysregulation Disorder; DSM IIIR: Diagnostic and Statistical Manual of Mental Disorders, third edition, revised; DSM IV: Diagnostic and Statistical Manual of Mental Disorders, fourth edition; DSM 5: Diagnostic and Statistical Manual of Mental Disorders, fifth edition; ECBI: Eyberg Child Behavior Inventory; ED: emotional disorders; ECSA: Early Childhood Screening Assessment; ITSEA: Infant-Toddler Social and Emotional Assessment; KIPPI: Dutch acronym for Brief Instrument Psychological and Pedagogical Problem Inventory; K-SADS-PL: Kiddie-Schedule for Affective Disorders- Present and Lifetime version; m: months; MAP-DB: Multidimensional Assessment Profile of Disruptive Behavior; MH: mental health; N: Norwegian cohort reported here; N/A: not available; ODD: Oppositional defiant disorder; P-ADHD-RS-IV: parent report ADHD rating scale, version for DSM IV; PAPA: the Preschool Age Psychiatric Assessment; PBCL: Preschool behavior checklist; RCT: Randomized controlled trial SDQ: Strengths & Difficulties Questionnaire; SES: socioeconomic status

^aModerate/ severe problems

^bCGAS < 60

^*Child Health Professional (CHP) preventive health assessment: following routine history and physical, CHP answered question; Does the child have a psycho social problem at this moment? Y/N “Psychosocial problems” incorporates behavioral and emotional problems

^!Included children seen for well-child/preventive health visits. “Psychosocial problems” incorporates behavioral and emotional problems

^!!Unselected sample

^XmiExcluded children too medically ill to recruit

^XddExcluded children with known global developmental delay, autism, pervasive developmental disorder

^XchExcluded children with chronic physical illness, epilepsy

^Nincluded only Norwegian cohort

Two studies identified thresholds for the Strengths and Difficulties Questionnaire (SDQ) against different criteria, another potential source of heterogeneity (Goodman, 2001). The SDQ, is a valid and reliable parent report scale similar in purpose and design to the CBCL although much briefer with only 25 items. A Dutch study (Theunissen et al., 2013) developed the threshold for SDQ against criterion of specificity > 0.90 of 90^th percentile on CBCL total problems scale, and found parent identified MH problems in three to four year old children of 13.8%. A Norwegian study (Sveen et al., 2013) developed the threshold for SDQ against DSM IV diagnoses generated using semi-structured interviews (Egger et al., 2006) and found parents identified 5.2% of four year old children with MH problems. The divergence in results may represent differences in population prevalence from country to country, (Achenbach, Rescorla, & Ivanova, 2012) or reflect use of different criteria to develop the threshold. Overall, the studies using standardized parent report measures were too dissimilar in design, measurement tools, and purpose to pool data and generate a meaningful summary estimate for parent identification.

Research Question 1.2 Identification by primary care and community practitioners using clinical assessments

A summary rate of 17.6% (95% CI: 11.1, 24.1) was found for the identification of MH problems by practitioners. This is based on a combined total sample size of 11,946, drawn from one North American sample and five Dutch samples, three of which were independent datasets reported in Theunissen et al, 2012 (Crone et al., 2016; Lavigne et al., 1993; Reijneveld et al., 2004; Theunissen et al., 2012). A Cochran Q = 4.9, df = 4, p > 0.1 and an I²= 0% indicate low probability of between-study heterogeneity. See Figure 2A.

Figure 2.

Summary Estimates for Mental Health Problems in Preschool Children in Primary or Community Health Care Settings.

There were two methods described for obtaining the practitioner’s clinical assessment of MH problems. In the earliest study, community paediatricians were asked to report their clinical global impression “whether the child had emotional/behavioral problems or developmental problems” following each child’s appointment (Lavigne et al., 1993). Visits included those for health supervision (42.4% of visits) and those for acute physical problems (50.5% of visits) (Lavigne et al., 1993). In the Dutch studies, completed in preventive health care settings for the purpose of developmental monitoring, the child health professionals performed routine history and physical, and then reported whether the child had a psychosocial problem (Crone et al., 2016; Reijneveld et al., 2004; Theunissen et al., 2012). See Table 2.

Sample characteristics that influenced identification included age and level of functional impairment. Controlling for disadvantaged socio-economic status, practitioners identified fewer children age two to three years than five to six years, and fewer children with high levels of functional impairment (Reijneveld et al., 2004; Theunissen et al., 2012).

Research Question 1.3: Identification by standardized psychiatric diagnostic criteria

A summary rate of 18.4% (95% CI: 12.3, 24.4) was found for psychiatric disorders using DSM III-R (American Psychiatric Association (APA), 1987) obtained using agreement between two independent psychologist assessments or DSM IV (American Psychiatric Association (APA), 1994) criteria using the semi-structured interview, Preschool Age Psychiatric Assessment (PAPA) (Egger et al., 2006; Lavigne et al., 1996; Wichstrom et al., 2012). This is based on a combined sample of 7,408 drawn from three studies. A Cochran Q= 1.6, df =2, with p > 0.1 and an I²= 0%, indicate low probability of between-study heterogeneity. Given the study publication dates, no study used the updated DSM 5 criteria for child disorders. See Figure 2B.

One study evaluated the concordance of community pediatricians’ global clinical impressions (see section 1.2) with a MH specialist (independent agreement of two psychologists) diagnostic assessment (Lavigne et al., 1993). Pediatricians identified fewer children with MH problems (8.7%) compared with psychologists (13.0%), with a sensitivity of 20.5%, a specificity of 92.7% and a positive predictive value of 64.5% compared to DSM III-R diagnoses (Lavigne et al., 1993). The degree to which the physician’s identification of MH problems matched the DSM III-R diagnoses did not vary with type of clinic visit or how well physician knew the patient (Lavigne et al., 1993). As only one study compared community practitioner identification with formal diagnoses it remains unclear how consistently community practitioners identify the same children as formal diagnostic assessments.

As noted in section 1.2 for practitioner identification, sample characteristics that influence prevalence rates include age of child and level of functional impairment. Disorders were identified less often among younger children, and those with greater impairment (Lavigne et al., 1993; Lavigne et al., 1996; Wichstrom et al., 2012). See Table 2.

Research Question 2: Mental health service use following identification

Findings on MH service use outcomes (e.g., management provided by practitioners or referrals to specialist care) following use of parent report screening measure or clinical identification of MH problems in preschool children were identified including four observational studies, (Briggs et al., 2012; Harwood et al., 2009; Lavigne, Arend, Rosenbaum, Binns, Christoffel, Burns, et al., 1998; Wichstrom et al., 2014) two diagnostic studies, (Lavigne et al., 1993; Reijneveld et al., 2004) and two clinical trials (Fallucco, Robertson-Blackmore, et al., 2017; Kruizinga et al., 2015). These studies used a range of study designs and represented varying health care contexts limiting ability to synthesize the findings. See Tables 1, 2.

Similar patterns of results were reported by two studies from the United States and one Dutch study in regards to clinical management at the time of assessment (Fallucco, Robertson-Blackmore, et al., 2017; Lavigne, Arend, Rosenbaum, Binns, Christoffel, Burns, et al., 1998; Lavigne et al., 1993; Reijneveld et al., 2004). Of those children identified by primary or community health care practitioners, a majority of parents (67–72%) consulted with, or received advice from, the child’s physician directly, whereas less than half (26%–42%) were referred to MH specialists (Fallucco, Robertson-Blackmore, et al., 2017; Lavigne et al., 1993; Reijneveld et al., 2004). In another study, increased MH service use followed a six month implementation of a program to improve MH care for three to five year olds (Fallucco, Robertson-Blackmore, et al., 2017). For identified children, consultations with practitioners increased from 67% to 85% and referral to specialists increased from 26% to 52% following training in use of a brief screening tool for MH problems at well child visits (Fallucco, Robertson-Blackmore, et al., 2017). Within a Norwegian cohort, only 10.9% of four year olds with diagnosed disorders received MH services in the three months prior to baseline evaluation; the rate increased to 25% at seven years (Wichstrom et al., 2014). See Table 2.

One trial examined effectiveness of the Brief-Infant Toddler Social and Emotional Assessment (BITSEA) compared with a Dutch instrument in use by child health professionals for two year olds (treatment as usual condition) (Kruizinga et al., 2015). There was a small but significant increase in parent-reported symptoms and referrals to specialists in the treatment as usual condition compared with using the BITSEA tool (Kruizinga et al., 2015). Another study examined effectiveness of referral to a co-located specialist following use of the Ages and Stages Questionnaire: Social-Emotional scale, in two to three year olds (Briggs et al., 2012). The children whose parents participated in an intervention improved relative to those who declined referral to service (Briggs et al., 2012). See Table 2.

Research Question 3: Persistence of mental health problems, 12 months or more following identification

Three cohorts examined psychiatric disorder outcomes two to four years post initial identification (Husby & Wichstrom, 2017; Lavigne, Arend, et al., 1998a, 1998b; Sveen et al., 2016; Wichstrom et al., 2013; Wiggins et al., 2018). In the earlier of two United States cohorts, 56% of children age two to three years and 67% of children age four to five years who had previously identified disorders continued to have psychiatric disorders one to three years later; intra-class correlation (ICC) for within subject stability of disruptive disorder was 0.72 and for emotional disorder was 0.50 (Lavigne, Arend, et al., 1998a). The second US cohort followed children, initially evaluated at age three to five years until age seven; high levels of parent-reported irritability predicted Oppositional Defiant Disorder and Disruptive Mood Dysregulation Disorder, by DSM 5 (American Psychiatric Association (APA), 2013) criteria (Wiggins et al., 2018). In a Norwegian cohort, attention deficit hyperactivity disorder (ADHD), but not anxiety disorders, in children at age four predicted anxiety disorder at age six; (Wichstrom et al., 2013) while oppositional symptoms at age four predicted continued oppositional behaviours and new onset conduct disorder behaviors (Husby & Wichstrom, 2017). Overall, 26.5% of children with disorders at age four had disorders at age six (Sveen et al., 2016). Data were not available to evaluate impact of MH service use on these outcomes. See Table 2.

No trials were found that evaluated use of a standardized screening tool for identification of MH problems in preschool children as a health intervention in primary or community care. Therefore, no data were available to document measurable benefits or adverse effects of screening. However, when surveyed, parents anticipate few negative consequences. Seventy nine percent of parents of children with MH problems would ‘welcome’ or would ‘not mind’ a referral to specialty services, and 75% of parents of children with disruptive behavior report stigma is not a barrier to accepting physician recommended services (Brown et al., 2012; Harwood et al., 2009). See Table 2.

Discussion

The literature examining identification of MH problems in preschool children in primary or community health care settings is sparse and the methods used variable in quality. To date, no clinical trials have evaluated the effectiveness of routine use of screening tools by primary or community health care practitioners for MH problems to determine if child health outcomes improve. Of the 35 studies included in our analysis, four studies (six datasets) described rates of community practitioner identified MH problems, (Crone et al., 2016; Lavigne et al., 1993; Reijneveld et al., 2004; Theunissen et al., 2012) and three described rates of psychiatric diagnoses in preschool children seen in community health care settings (Egger et al., 2006; Lavigne et al., 1996; Wichstrom et al., 2012). Although some identified studies examined additional applied questions (e.g., comparing two screening tools used by community health care practitioner (Kruizinga et al., 2015) examined referral to a co-located specialist for a brief intervention (Briggs et al., 2012), assessed feasibility and uptake of routine MH screening (Fallucco et al., 2017)), there is not yet sufficient evidence to support widespread routine use of screening tools for MH problems in healthy preschool children.

Our prevalence analyses indicate MH problems are frequent, affecting about one in every six children, age two to six years, who see their health care provider. Based on the limited available data, we estimate that practitioners clinically identify MH problems in 17.6% (95% CI: 11.1, 24.1) of preschool children, an estimate that is similar to that for psychiatric diagnoses, 18.4% (95% CI: 12.3, 24.4), and rates similar to those in older children (Kato et al., 2015). Information to investigate concordance between practitioner identification and psychiatric diagnoses is available for one study, reflecting clinical practice from nearly three decades ago. Therefore, it is premature to draw conclusions regarding concordance of practitioner-identified children with those meeting diagnostic criteria.

Sources of heterogeneity in prevalence estimates include sample characteristics such as age and level of impairment, clinical identification methods, and practice contexts. The earliest study included in the pooled summary rates, used practitioner clinical impression, likely reflecting standard practice at the time, and resulting in a relatively low rate of identification (Lavigne et al., 1993). Since early 1990s, temporal changes to pediatric MH and behavioral health care have included refinements to formal diagnostic schema, with emphasis on assessment of adaptive functioning in DSM IV, (American Psychiatric Association (APA), 1994) and consideration of neurodevelopmental disorders in DSM 5 (American Psychiatric Association (APA), 2013). In addition, awareness has increased regarding the need to address behavioural health (Boat, 2015). Also included in the pooled estimate for practitioner identification, are studies from the Netherlands where a robust public health infrastructure exists for developmental surveillance (Crone et al., 2016; Reijneveld et al., 2004; Theunissen et al., 2012). Such an established preventive child health care system may increase early identification (Fleuren, van Dommelen, & Dunnink, 2015).

Information was also sparse regarding service use outcomes following identification of MH problems in preschool children. Although data are limited, it appears that when clinicians recognize early presentations of MH disorders, the majority provide advice to parents but less than half provide referrals to specialty care (see description under Results, section 2) (Fallucco, Robertson-Blackmore, et al., 2017; Lavigne et al., 1993; Reijneveld et al., 2004). Such health education as an initial intervention may represent a conservative management approach that avoids over-diagnosis of young children, and the cost of potentially unnecessary referrals. In addition, practitioners in many communities may adjust their practice due to lack of available resources for MH supports for preschool children and their families. Interestingly, such rates are consistent with observations of the threshold probability properties of widely used parent report MH screening tools for school age children (Sheldrick et al., 2015). As threshold values increase, the probability of a positive diagnosis for an individual child increases, however sensitivity decreases (Sheldrick et al., 2015). Practitioners may set their clinical opinion about a child’s need for formal intervention at a higher value than published thresholds on standardized measures; thus, fewer children are referred for specialty care than are identified (Sheldrick et al., 2015). Such low sensitivity associated with practitioner clinical opinion supports the rationale for a screening tool as an initial step in identification protocols.

As described in the results section, reported rates of MH problems identified by parent report measures were highly variable, reflecting the purpose of the research studies rather than thresholds set for identifying clinical concern. An important aspect of using standardized measures in a clinical context rather than a research one, however, is improved parent-practitioner communication. When asked explicitly on surveys, parents often raise concerns about their children’s behavioral, emotional and social health (Brothers, Glascoe, & Robertshaw, 2008; Glascoe, 1999; Reijneveld et al., 2008). Not surprisingly, practitioner use of standardized parent report tools improves agreement between practitioners and parents regarding presence of problems for the child (Brothers et al., 2008; Crone et al., 2016; Theunissen et al., 2012). Screening tools may lead to increased discussions, and more investigations in response to concerns, but may or may not lead to increased referrals and use of specialty services (Berger-Jenkins, McCord, Gallagher, & Olfson, 2012; Jonovich & Alpert-Gillis, 2014). Indeed, practitioners in U.S. jurisdictions where behavioral health screening has been mandated describe the primary functional benefit in terms of improved discussions with parents about their children rather than in use of the screening tool to identify children at risk (Van Cleave, Morales, & Perrin, 2013).

After a child has been identified, some families may decline referral and an opportunity for intervention even when easily accessible through co-location in pediatric practices (Briggs et al., 2012; Perrin, Sheldrick, McMenamy, Henson, & Carter, 2014). Based on models of health behavior, parental acceptance of intervention may often follow a lengthy process prior to recognition and acknowledgement of a problem, one that can be difficult for parents (Charach & Fernandez, 2013). Physician- parent discussions during the child’s preschool years may encourage parents to accept mental health interventions more easily at a future date.

Based on earlier experiences from implementation of developmental surveillance programs for infants and toddlers, routine use of parent report MH screening tools will require changes to health care practice infrastructure and procedures, and establishment of professional relationships to provide patient access to specialized care (King et al., 2010; Weitzman et al., 2015). Integrated primary care-behavioral health care programs may provide better outcomes for children and youth requiring MH interventions than usual primary care (Asarnow, Rozenman, Wiblin, & Zeltzer, 2015). Such integrated health care programs generally include routine use of brief standardized parent report tools as a basic component (Kolko & Perrin, 2014). The screening implementation study described earlier included changes to infrastructure and procedures, as well as a defined pathway to specialist care (Fallucco, Robertson-Blackmore, et al., 2017). Six months after training, primary care providers increased their use of standardized tools “most of the time” from 4% to 67% as well as increasing counselling and referrals (Fallucco, Robertson-Blackmore, et al., 2017). Interestingly, providers did not institute routine use of tools for all well child visits, but rather for some children, perhaps targeting those at high risk for problems. This clinical approach deserves further examination.

One way to conceptualize how to measure the utility of early identification in primary care would be to estimate the number needed to screen, NNS, an estimate that is developed from measurements of the effect of a screening strategy, such as the absolute risk reduction following interventions for risk factors (Rembold, 1998). This could be helpful for those developing public health policy regarding implementation of a mental health screening program. Several pieces of evidence are required to develop these estimates: the population base rate of disorder, the rate of identification, and child health outcomes following identification, such as receipt of treatment and rate of positive outcome following intervention (Rembold, 1998). This review addresses two of the rates required for calculating these measures: the base rate for diagnosed mental health disorders in preschool children who are seen in primary or community health care settings, 18.4%, and the rate of identification of problems by primary care or community practitioners, 17.6%. The chain of evidence required after this requires further investigation. Especially notable is the limited documentation about immediate clinical management and its outcomes, access to evidence-based interventions and subsequent impact on the child’s adaptive functioning and quality of life.

To address this lack of evidence, it will be necessary to develop, implement and evaluate programs that include service pathways from primary care identification to evidence-based interventions for preschool children with MH problems. Promising approaches appear to be 1) incorporation of brief standardized parent report tools for MH problems into developmental surveillance, 2) protocols to ensure timely access to MH services, and 3) methods to provide regular feedback to health care providers regarding the child’s MH outcomes so the clinical approach can be adjusted as needed. Effective methods will require adjustments to practice infrastructure and procedures, skill training for practitioners, and integrated clinical pathways to access MH specialists who provide evidence based interventions (King et al., 2010; Weitzman et al., 2015). Ideally each of these components would be rigorously evaluated to determine the extent to which practices and policies are evidence-based and leading to improved child outcomes.

An important limitation is the small number of articles that met inclusion criteria and hence the extent of the empirical database to inform practice and policy. The heterogeneity across studies also limited the ability to conduct qualitative and quantitative synthesis, including comparisons of cross-cultural practices. An additional limitation was the restriction to those studies that had an English abstract. Nevertheless, our results are consistent with those of other recent systematic reviews, (CTFPHC, 2016; Siu et al., 2016; Wallace et al., 2015) indicating little evidence exists to inform practice about early identification of MH problems.

In summary, we have documented that a substantial number, almost one in six, of preschool children seen in primary or community health care settings may have MH problems. Overall, primary and community health care practitioners identify essentially the same percentages of preschool children as MH specialists, however concordance with diagnoses has not yet been widely tested and remains to be defined. The evidence does not yet justify routine screening for MH problems for all preschool children in primary health care settings. Models of care that ensure access to effective interventions require further development and evaluation. Similar to care models for other chronic health conditions, best practice models generally include routine use of standardized screening tools, and embedded procedures for monitoring service use and outcomes (Kolko & Perrin, 2014). More generally, targeted surveillance during the preschool years could provide the foundation for an evidence-informed, outcomes-based practice to ensure that young children and their families in need receive timely and appropriate MH care. Effective early interventions exist for some MH problems (Gleason et al., 2016) and young children at high risk have shown improved long term outcomes following some early interventions (Karoly et al., 2005). However, substantial work is required to develop and evaluate methods that match preschool children in need with accessible evidence-based treatments that work.

Appendix A. Search Strategies

MEDLINE

The search strategy for OvidSP MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations (1946 to March 7, 2018) using a combination of MeSH terms and textwords.

Set	History
1	«attention deficit and disruptive behavior disorders»/ or attention deficit disorder with hyperactivity/ or conduct disorder/
2	«minimal brain d?sfunction*».mp.
3	((attention adj2 deficit*) or adhd).ti,ab.
4	Hyperkinesis/
5	Child Behavior Disorders/
6	aggression/ or agonistic behavior/
7	inattent*.ti,ab.
8	(disruptive adj4 disorder*).ti,ab.
9	(disruptive adj4 behavio*).ti,ab.
10	«non-complian*».ti,ab.
11	(temper adj2 tantrum*).ti,ab.
12	((opposition* or defian*) adj4 (disorder* or behavio*)).ti,ab
13	(dysregulat* or disregulat*).ti,ab.
14	Or/1–13
15	«sensitivity and specificity»/ or «predictive value of tests»/ or roc curve/ or signal-to-noise ratio/
16	evaluation studies.pt. or evaluation studies as topic/
17	validation studies.pt. or validation studies as topic/
18	diagnostic errors/ or false negative reactions/ or false positive reactions/ or observer variation/
19	(likelihood or likelihood ratio:).ti,ab.
20	likelihood functions/
21	mass screening/ or screen*.ti,ab.
22	questionnaire/ or self report/ or (questionnaire* or survey*).ti,ab.
23	(parent* adj2 report*).ti,ab.
24	(tool* or instrument* or scale*).mp.
25	exp Psychiatric Status Rating Scales/
26	Interview, Psychological/
27	(screen* or diagnos* or detect* or determine or casefinding* or (case adj2 finding*)).ti,ab.
28	Or/15–27
29	14 and 28
30	limit 29 to «preschool child (2 to 5 years)»
31	(preschool* or pre-school* or «2 year* old*» or «3 year* old*» or «4 year* old*» or «5 year* old*» or «two year* old*» or «three year* old*» or «four year* old*» or «five year* old*» or toddler* or tot or tots or kindergarten*).ti,ab.
32	30 or (29 and 31)
32	30 or (29 and 31)
33	(harm or harms or «adverse event*» or «adverse effect*»).ti,ab. or ae.fs. or/57–61 [****Harm terms****] (2053557)
34	((over or excess* or unnecessary or unjustified) and (treat* or diagnos*)).ti,ab.
35	label?ing.ti,ab.
36	Stress, Psychological/
37	social stigma/ or stereotyping/ or «denial (psychology)»/ or anger/ or anxiety/ or guilt/
38	Or/33–38
39	14 and 38
40	limit 39 to «preschool child (2 to 5 years)»
41	(preschool* or pre-school* or «2 year* old*» or «3 year* old*» or «4 year* old*» or «5 year* old*» or «two year* old*» or «three year* old*» or «four year* old*» or «five year* old*» or toddler* or tot or tots or kindergarten*).ti,ab.
42	40 or (39 and 41)
43	32 and 42

EMBASE

The search strategy for OvidSP EMBASE Classic + EMBASE (1947 to March 7, 2018) using a combination of EMBASE terms and textwords.

Set	History
1	attention deficit disorder/ or hyperactivity/ or conduct disorder
2	minimal brain dysfunction/ or «minimal brain d?sfunction*».mp
3	((attention adj2 deficit*) or adhd).ti,ab.
4	Behavior Disorders/
5	exp aggression/ or «agnostic behavi?r*».ti,ab.
6	inattent*.ti,ab.
7	exp impulse control disorder/
8	(disruptive adj4 behavio*).ti,ab.
9	«non-complian*».ti,ab.
10	(temper adj2 tantrum*).ti,ab.
11	oppositional defiant disorder/ or ((opposition* or defian*) adj4 (disorder* or behavio*)).ti,ab.
12	(dysregulat* or disregulat*).ti,ab.
13	Or/1–12
14	sensitivity analysis/ or «sensitivity and specificity»/ or signal noise ratio/
15	«prediction and forecasting»/ or prediction/
16	receiver operating characteristic/ or («roc curve*» or (roc adj2 curve*)).mp. or reproducibility/ or reliability/ or cronbach alpha coefficient/ or internal consistency/ or interrater reliability/ or intrarater reliability/ or item total correlation/ or kuder richardson coefficient/ or split half correlation/ or test retest reliability/
17	diagnostic error/ or false negative result/ or false positive result/ or ((diagnostic adj5 error*) or (false adj5 negative*) or (false adj5 positive*)).mp. or laboratory diagnosis/ or abnormal laboratory result/
18	likelihood functions/ or (likelihood or (likelihood adj2 ratio*)).mp
19	evaluation/ or validation study/ or ((evaluation or validation) adj2 (study or studies)).ti,ab.
20	Exp mass screening/ or screen*.ti,ab.
21	exp «named inventories, questionnaires and rating scales»/ or exp questionnaires/ or self report/ or (questionnaire* or survey*).ti,ab.
22	(parent* adj2 report*).ti,ab.
23	(tool* or instrument* or scale*).mp.
24	psychological rating scale/
25	exp psychologic test/
26	(screen* or diagnos* or detect* or determine or casefinding* or (case adj2 finding*)).ti,ab.
27	Or/14–26
28	13 and 27
29	limit 28 to preschool child <1 to 6 years
30	(preschool* or pre-school* or «2 year* old*» or «3 year* old*» or «4 year* old*» or «5 year* old*» or «two year* old*» or «three year* old*» or «four year* old*» or «five year* old*» or toddler* or tot or tots or kindergarten*).ti,ab.
31	29 or (28 and 30)
32	(harm or harms or «adverse event*» or «adverse effect*»).ti,ab. or ae.fs.
33	((over or excess* or unnecessary or unjustified) and (treat* or diagnos*)).ti,ab.
34	Mental Stress/
35	Stress, Psychological/
36	social stigma/ or social psychology/ or stigma/ or denial/ or anger/ or rage/ or anxiety/ or guilt/ or stereotyp*.ti,ab.
37	Or/32–36
38	13 and 37
39	limit 38 to preschool child <1 to 6 years
40	(preschool* or pre-school* or «2 year* old*» or «3 year* old*» or «4 year* old*» or «5 year* old*» or «two year* old*» or «three year* old*» or «four year* old*» or «five year* old*» or toddler* or tot or tots or kindergarten*).ti,ab.
41	39 or (38 and 40)
42	31 and 41

PsycINFO

The search strategy for OvidSP PsycINFO (1967 to March 7, 2018) using a combination of PsycINFO terms and textwords.

Set	History
1	attention deficit disorder/ or attention deficit disorder with hyperactivity/ or conduct disorder/ or hyperkinesis/
2	«minimal brain d?sfunction*».mp
3	((attention adj2 deficit*) or adhd).ti,ab.
4	Exp Behavior Disorders/
5	aggressiveness/ or aggressive behavior/.
6	inattent*.ti,ab.
7	exp impulse control disorders /
8	(disruptive adj4 (disorder* or behavio*)).ti,ab.
9	«non-complian*».ti,ab.
10	Tantrums/ or (temper adj2 tantrum*).ti,ab.
11	oppositional defiant disorder/ or ((opposition* or defian*) adj4 (disorder* or behavio*)).ti,ab.
12	(dysregulat* or disregulat*).ti,ab.
13	Or/1–12
14	(sensitivity or specificity).ti,ab.
15	statistical validity/ or statistical analysis/ or «consistency (measurement)»/ or exp prediction errors/ or exp statistical correlation/ or statistical reliability/ or exp statistical variables/
16	(«roc curve*» or (roc adj2 curve*)).mp. or evaluation/ or treatment effectiveness evaluation/ or exp errors/
17	maximum likelihood/
18	(likelihood or likelihood ratio:).ti,ab.
19	evaluation criteria/ or test reliability/ or test validity/
20	screening/ or exp psychiatric evaluation/ or exp screening tests/ or screen*.ti,ab.
21	questionnaires/ or mail surveys/ or exp surveys/ or telephone surveys/ or self report/ or (questionnaire* or survey*).ti,ab.
22	(parent* adj2 report*).ti,ab.
23	(tool* or instrument* or scale*).mp.
24	rating scales/ or psychodiagnostic interview/ or diagnostic interview schedule/ or structured clinical interview/ or intake interview/ or exp psychiatric evaluation/
25	exp psychological assessment/
26	(screen* or diagnos* or detect* or determine or casefinding* or (case adj2 finding*)).ti,ab.
27	Or/14–26
28	13 and 27
29	limit 28 to 160 preschool age
30	(preschool* or pre-school* or «2 year* old*» or «3 year* old*» or «4 year* old*» or «5 year* old*» or «two year* old*» or «three year* old*» or «four year* old*» or «five year* old*» or toddler* or tot or tots or kindergarten*).ti,ab.
30	(preschool* or pre-school* or «2 year* old*» or «3 year* old*» or «4 year* old*» or «5 year* old*» or «two year* old*» or «three year* old*» or «four year* old*» or «five year* old*» or toddler* or tot or tots or kindergarten*).ti,ab.
31	29 or (28 and 30)
32	(harm or harms or «adverse event*» or «adverse effect*»).ti,ab.
33	((over or excess* or unnecessary or unjustified) and (treat* or diagnos*)).ti,ab. or
34	labeling/ or label?ing.ti,ab.
35	stress/ or psychological stress/ or social stress/ or stress reactions/
36	stereotyped attitudes/ or stigma/ or denial/ or anger/ or anger control/ or anxiety/ or guilt/
37	Or/32–36
38	13 and 37
39	limit 38 to 160 preschool age
40	(preschool* or pre-school* or «2 year* old*» or «3 year* old*» or «4 year* old*» or «5 year* old*» or «two year* old*» or «three year* old*» or «four year* old*» or «five year* old*» or toddler* or tot or tots or kindergarten*).ti,ab.
41	39 or (38 and 40)
42	31 and 41

ERIC

The search strategy for OvidSP ERIC (1965 to December 15, 2015) using a combination of ERIC terms and textwords. (note: the license for OvidSP ERIC license expired prior to update in March 2017. Negligible unique articles were found using this database, therefore we chose not to seek another source for ERIC.)

Set	History
1	attention deficit disorders/ or attention deficit hyperactivity disorder/ or (conduct adj2 disorder*).ti,ab. or hyperactivity/
2	minimal brain dysfunction/ or «minimal brain d?sfunction*».mp
3	((attention adj2 deficit*) or adhd).ti,ab.
4	behavior problems/ or behavior disorders/
5	self control/ or (Impulse adj2 Control).mp.
6	inattent*.ti,ab.
7	((opposition* or defian*) adj4 (disorder* or behavio*)).ti,ab.
8	(disruptive adj4 (disorder* or behavio*)).ti,ab.
9	«non-complian*».ti,ab.
10	(temper adj2 tantrum*).ti,ab.
11	(dysregulat* or disregulat*).ti,ab.
12	Or/1–11
13	(sensitivity or specificity).ti,ab.
14	evaluation criteria/ or reliability/ or validity/ or evaluation/ or evaluation methods/ or prediction/
15	(«roc curve*» or (roc adj2 curve*)).mp. or «error of measurement»/ or scoring/ or true scores/
16	maximum likelihood statistics/
17	(likelihood or likelihood ratio:).ti,ab.
18	test reliability/ or test interpretation/ or test validity/
19	screening tests/ or psychological evaluation/ or psychological testing/ or psychometrics/ or screen*.ti,ab.
20	questionnaires/ or mail surveys/ or online surveys/ or semi structured interviews/ or surveys/ or telephone surveys/ or (self adj2 report*).ti,ab. or (questionnaire* or survey*).ti,ab.
21	(parent* adj2 report*).ti,ab.
22	(tool* or instrument* or scale*).mp.
23	exp rating scales/ or interrater reliability/
24	interviews/ or semi structured interviews/ or structured interviews/
25	(screen* or diagnos* or detect* or determine or casefinding* or (case adj2 finding*)).ti,ab.
26	Or/13–25
27	12 and 26
28	preschool children/ or toddlers/ or kindergarten/ or preschool education/ or (preschool* or pre-school* or «2 year* old*» or «3 year* old*» or «4 year* old*» or «5 year* old*» or «two year* old*» or «three year* old*» or «four year* old*» or «five year* old*» or toddler* or tot or tots or kindergarten*).ti,ab.
29	27 and 28
30	harm or harms or «adverse event*» or «adverse effect*»).ti,ab.
31	((over or excess* or unnecessary or unjustified) and (treat* or diagnos*)).ti,ab. or
32	«labeling (of persons)»/ or label?ing.ti,ab
33	anxiety/ or stress variables/
34	stereotypes/ or stigma*.ti,ab. or denial*.ti,ab. or aggression/ or anger.ti,ab. or guilt*.ti,ab.
35	Or/30–34
36	12 and 35
37	preschool children/ or toddlers/ or kindergarten/ or preschool education/ or (preschool* or pre-school* or «2 year* old*» or «3 year* old*» or «4 year* old*» or «5 year* old*» or «two year* old*» or «three year* old*» or «four year* old*» or «five year* old*» or toddler* or tot or tots or kindergarten*).ti,ab.
38	36 and 37
39	29 and 38

APPENDIX B. Quality of Studies

Table A.

Quality of Diagnostic Studies about Preschool Children with Mental Health Problems in Primary or Community Health Care Setting Assessed by QUADAS-2 Scale (Whiting et al., 2011).

Risk of bias assessment criteria	Egger, 2006	Fallucco, 2017b	Franz, 2013	Lavigne, 1993	Lavigne, 1996	Reijneveld, 2004	Takayanagi, 2016	Theunissen, 2012a,b,c	#Theunissen,2013, 2015	#Wichstrom,2012” Sveen, 2013
Risk of Bias
Patient Selection	*	*	?	?	?	*	H	*	*	*
Index Test	*	*	?	?	*	*	H	?	*	*
Reference Test	*	?	*	*	*	?	H	?	?	*
Flow and Timing	*	*	*	*	*	*	H	*	*	*
Applicability
Patient Selection	*	*	*	*	*	*	*	*	*	*
Index test	H	*	H	*	*	*	*	*	*	*
Reference Standard	*	?	*	*	*	*	?	?	*	*

^*Represents low risk of bias

? Represents unclear risk of bias

H Represents high risk of bias

a,b,c Datasets A: Brugman, 2001; B: Wiefferink 2006; C: Reijneveld, 2008

# Represents 2 reports, same dataset

Table B.

Quality of Cross-Sectional Studies about Preschool Children with Mental Health Problems in Primary or Community Health Care Setting as Assessed by the Modified New-Castle Ottawa Scale

Quality assessment criteria	Criterion	Brown 201244	Crone 201630	Harwood 200951	Leung 200540	Rai 199341	Sim 201352	Sourander 200138	Thompson 199653	Weitzman 201457
Representativeness of the exposed cohort	Truly representative of the general population visiting clinics		*		*			*		*
	Somewhat representative of the general population vising clinics	*					*		*
Selection of the non-exposed cohort	Drawn from the same community as exposed cohort	*	*	*	*	*	*	*	*	*
Ascertainment of exposure	Validated questionnaire/Structured interview	*	*	*	*		*	*	*	*
Comparability of groups on basis of design or analysis	Study controls for age	*	*	*	*	*	*	*	*	*
	Study controls for any gender, race and family psychosocial status	*	*	*	*	*		*	*	*
Adequacy of follow up of cohorts	Complete follow up: all subjects accounted for	*	*
	Subjects lost to follow up unlikely to introduce bias: >75% follow up, or description provided of those lost				*		*	*	*	*
Summary score (maximum of 6 stars)		6	6	4	6	3	5	6	6	6

^*Criterion fulfilled. Each criterion can be awarded a maximum of one star (representing “yes”) for each numbered item within the Representativeness of the exposed cohort’ and ‘Adequacy of follow up of cohort’ categories. A maximum of two stars can be given for Comparability of groups.

Table C.

Quality of Longitudinal Studies about Preschool Children with Mental Health Problems in Primary or Community Health Care Setting as Assessed by the Modified New-Castle Ottawa Scale

Quality assessment criteria	Criterion	Briggs 2012	Kruizinga 2015	Fallucco 2017	Lavigne 1998a	Lavigne 1998b 1998c, 1999	Wakschlag 2015, Wiggins 2018	Wichstrom 2013, 2014, Sveen 2016, Husby & Wichstrom 2017, Wichstrom 2018
Representativeness of the exposed cohort	Truly representative of the general population visiting clinics				*	*	*	*
	Somewhat representative of the general population vising clinics	*	*	*
Selection of the non-exposed cohort	Drawn from the same community as exposed cohort	*	*		*	*	*	*
Ascertainment of exposure	Validated questionnaire/Structured interview	*	*			*	*	*
Documentation that outcome of interest is present/absent at baseline	Yes	*	*	*		*	*	*
Comparability of cohorts on basis of design or analysis	Study controls for age	*	*		*	*	*	*
	Study controls for any gender, race and family psychosocial status	*	*		*	*	*	*
Assessment of outcome	Independent blind assessment			*
	Validated questionnaire/Structured interview	*	*			*	*	*
Adequacy of follow-up of cohort	Complete follow up: all subjects accounted for
	Subjects lost to follow up unlikely to introduce bias: >75% followed up, or description provided of those lost	*	*	*	*	*	*	*
Summary score (maximum of 8 stars)		CT8	CT8	CT4	5	##8	#8	### 8

^*Criterion fulfilled. A star can be awarded a maximum of one star (representing “yes”) for each numbered item within the ‘Representativeness of exposed cohort’, ‘Assessment of outcome’, and ‘Adequacy of follow-up of cohorts’. A maximum of two stars can be given for Comparability.

^#Represents 2 reports, same dataset;

^##Represents 3 reports, same dataset;

^###Represents 5 reports, same dataset; CT Clinical trial with follow up