Prenatal Alcohol Exposure, FASD, and Child Behavior: A Meta-analysis
- Tracey W. Tsang, PhDa,b,
- Barbara R. Lucas, MPHa,b,c,d,
- Heather Carmichael Olson, PhDe,
- Rafael Z. Pinto, PhDf, and
- Elizabeth J. Elliott, MDa,b,g,h
- aDiscipline of Paediatrics and Child Health, Sydney Medical School, The University of Sydney, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia;
- bThe George Institute for Global Health, and
- cPoche Centre for Indigenous Health, Sydney Medical School, The University of Sydney, Sydney, New South Wales, Australia;
- dPhysiotherapy Department, Royal North Shore Hospital, St Leonards, New South Wales, Australia;
- eDepartment of Psychiatry and Behavioral Sciences, University of Washington School of Medicine, Seattle Children’s Research Institute, Seattle, Washington;
- fDepartamento de Fisioterapia, Faculdade de Ciências e Tecnologia, Universidade Estadual Paulista, Presidente Prudente, Sao Paulo, Brazil;
- gThe Sydney Children’s Hospital Networks (Westmead), Westmead, New South Wales, Australia;
- hThe Australian Paediatric Surveillance Unit, Kids’ Research Institute, Westmead, New South Wales, Australia.
Dr Tsang conceptualized and designed the study, undertook the electronic and manual searches of the literature, data extraction, analysis, and appraisal of the articles, designed and created the data collection forms, coordinated the data collection, and drafted the initial manuscript; Ms Lucas contributed to the study design and to the design of the data collection forms, was the second assessor for this review, which involved independently checking the selection of articles for the review, data extracted, and study appraisal, and reviewed and revised, the final manuscript; Prof Carmichael Olson assisted in original design of data review documents, reviewed and critically revised the manuscript; Dr Pinto provided guidance on meta-analysis procedures, assisted in discussions regarding eligibility of some articles, and reviewed and revised the final manuscript as submitted; Prof Elliott conceptualized the study, contributed to the design of the study and data collection forms, and reviewed and critically revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
CONTEXT: Fetal alcohol spectrum disorders (FASD) and prenatal alcohol exposure (PAE) are associated with behavioral difficulties, although there are no published systematic reviews that summarize and critique the literature.
OBJECTIVE: To describe the behavioral characteristics of children with PAE and/or FASD, assessed using the Achenbach System of Empirically Based Assessments (ASEBA) for school-aged children with parent, teacher, and youth (self-report) forms.
DATA SOURCES: Electronic literature databases, reference lists, hand-searches.
STUDY SELECTION: peer-reviewed observational studies.
DATA EXTRACTION: Study appraisal and data extraction were undertaken by 2 independent assessors. Meta-analyses were performed for parent-rated Internalizing, Externalizing, and Total problems scales. All other ASEBA scales were summarized qualitatively.
RESULTS: Included were 23 articles; 16 were used in meta-analyses. Pooled results showed higher Total (mean difference 12.1, 95% confidence interval [95% CI] 7.7–16.5), Internalizing (6.3, 95% CI 3.1–9.5), and Externalizing problems scores (12.5, 95% CI 7.9–17.0) in FASD than No FASD; and greater odds of scoring in the “Clinical” range in FASD. Pooled results demonstrated higher problem scores in children with PAE (P > .05). Qualitative summaries of other scales from parents, teachers, and self-report show poorer behavior ratings in children with FASD and PAE on composite Problem and Competence scores and many Syndrome subscales.
LIMITATIONS: Findings were restricted to behaviors assessed using the ASEBA. The published literature was limited, often with only 1 study reporting on a particular scale.
CONCLUSIONS: Meta-analysis reveals that FASD and PAE are associated with problematic behavior in many, but not all domains. This clearly affects families, and should be considered in clinical practice by providers.
- ADHD —
- attention deficit hyperactivity disorder
- ASEBA —
- Achenbach system of empirically based assessment
- CBCL —
- child behavior checklist
- CI —
- confidence interval
- FAS —
- fetal alcohol syndrome
- FASD —
- fetal alcohol spectrum disorders
- OR —
- odds ratio
- PAE —
- prenatal alcohol exposure
- TRF —
- teacher report form
- YSR —
- youth self-report
Behavior impairments (predominantly in attention, disruptive behavior and conduct disorders, academic performance, and social judgment) have been documented in children with fetal alcohol spectrum disorders (FASDs) and significant prenatal alcohol exposure (PAE).1,2 Secondary conditions found with high frequency among individuals with FASD and PAE include mental health problems, psychiatric illness,3,4 school difficulties, trouble with the law, placement in confined settings (eg, psychiatric hospitals), inappropriate sexual behavior, and substance abuse.5
The importance of screening for and assessing behavior problems in the field of FASD is well-recognized6,7 and used to guide clinical practice. The psychometrically sound Achenbach System of Empirically Based Assessment (ASEBA)8 has been the most commonly used system of behavior assessment in children and adolescents with FASD since the mid-1990s. ASEBA School-Age Forms include the parent/caregiver-rated Child Behavior Checklist (CBCL), the Teacher Report Form (TRF), and the Youth Self-Report (YSR).8 The School-Age forms assess a broad range of behaviors, including competencies, problems, and adaptive function. They also provide raw and, for clinical interpretation, standardized scores. We aimed to conduct a systematic review of the literature, with meta-analysis if appropriate, on behavioral ratings from multiple informants (parents/caregivers, teachers, self-report) in children with FASD and/or PAE, who were assessed by using the ASEBA School-Age Forms. To our knowledge, this is the first review of this type. We hypothesized that behavior ratings would be poorer in children with PAE and/or FASD than without.
This review was registered with PROSPERO (registration: CRD42014007040; Web link: http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42014007040#.VOJ7A8lI_MA), and conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses recommendations.9
Selection criteria are summarized in Table 1. If a research team reported the same outcomes from the same cohort in multiple articles, only the article with the larger sample size was included to prevent duplicate inclusion of results.
The primary variables of interest were the scores for Internalizing problems (comprising the syndrome subscales: Anxious/depressed, Withdrawn/depressed, and Somatic complaints), Externalizing problems (comprising Rule-breaking and Aggressive behavior syndrome subscales), Total problems (sum of Internalizing and Externalizing problems scales; and syndrome subscales: Social, Thought, Attention, and Other problems), and Total competence (sum of Activities, Social, and School competence subscales).8 All other scale scores and Critical items were considered secondary outcomes. Data extracted were restricted to T scores, raw scores, and/or proportions scoring within defined clinical ranges (Table 1).
The search was conducted by using electronic databases, and reference lists from eligible articles were inspected and hand-searched. Electronic databases used were Medline (Ovid; 1946 to February week 3, 2015), PsycINFO (1806 to February week 3, 2015), Maternity and Infant Care (1971 to January 2015), Embase (1974 to March 10, 2014), AMED (1985 to February 2015), CINAHL Plus (EBSCO Host), and EBM Reviews—Cochrane Database of Systematic Reviews (2005 to January 2015). The electronic literature search was conducted on March 11, 2014 (updated on February 17, 2015) and manual searches concluded in March 2015.
Our search strategy was formulated with the assistance of a medical librarian. Search terms used for all databases were “fetal alcohol,” “prenatal alcohol exposure,” “prenatal alcohol,” “ASEBA,” “child behavior checklist,” “Achenbach,” “teacher report form,” “youth self-report,” “child behavior,” and “psychosocial.” Due to changes in FASD diagnostic criteria over time, we selected broad search terms to maximize our yield. No limits were assigned for the electronic searches. As an example, our Medline (OVID) search strategy is provided in Supplemental Appendix 1.
Screening and study selection was a 2-stage process. Inspection of article titles, abstracts, and article type was conducted during the first stage (by TWT). The second stage involved review of full-text articles identified as potentially eligible (Table 1) by 2 independent reviewers (TWT and BRL), with disagreements resolved with EJE.
Data Extraction Process
Data extraction/interpretation was undertaken by TWT, independently checked by BRL, and disagreements resolved by EJE and RZP. Data were collated into an electronic database. Corresponding authors were contacted via e-mail for additional information when necessary.
Risk of Bias in Individual Studies
Data Synthesis and Analysis
Types of data extracted included T scores, raw scores, proportions (%), or odds ratios (OR) for scoring within defined clinical ranges. These different types of data could not be combined in subsequent meta-analyses or forest plots, so separate analyses were undertaken for each data type (for primary outcomes) if reported by at least 2 separate studies.11 T or raw score data were used to calculate the difference in means between “cases” and controls and 95% confidence intervals (95% CIs). Proportions within clinical ranges were used to calculate the OR and 95% CI where applicable. Because this review only included the School-Age Forms, no standardization of age and gender-normed T-score data were undertaken. T scores are more directly related to scores obtained from ASEBA forms than other standardized scores (eg, z scores). Also, T scores are typically used in clinical practice. Scores in the “Clinical” or “Borderline” range on ASEBA scales, defined by T scores, generally indicate need for therapeutic intervention.
Separate analyses were conducted based on the presence or absence of (1) an FASD diagnosis (including “mixed” groups containing children with PAE in which 100% of them had an FASD diagnosis), and (2) PAE; according to how results were reported in the articles. Data were grouped according to age category (mean age: <5 years, 5–10 years, and >10 years), outcome (ASEBA score), and PAE level where applicable. Effects of different PAE levels on primary outcomes were explored after attempted categorization of PAE levels, using definitions suggested by O’Leary et al,12 and also using the classification of PAE specified in each study (low/moderate/high).
Where data were stratified and presented in multiple groups, data from PAE/FASD and control groups were extracted according to (1) consistency with groups of interest for this review, (2) largest total sample size, and (3) longest duration of PAE. If individual group sample sizes were not reported or could not be calculated from results provided, information was sought from corresponding author(s). In accordance with Cochrane guidelines, if a study had no control/comparison group, data from another study’s comparison group (with comparable sample characteristics) was borrowed and the comparison group sample size divided by the number of studies by using the same comparison group.11
Comprehensive Meta Analysis (version 2.2.064; Biostat, Inc, Englewood, NJ) software was used for meta-analyses and forest plots, by using random effects models. A P < .05 denoted statistical significance. Heterogeneity of studies in meta-analyses was assessed by using the I2 statistic, where I2 <40% suggested homogeneity within pooled data.11 Funnel plot asymmetry was examined if at least 10 studies were included in the meta-analysis.11 Data not included in meta-analyses are presented descriptively in the Supplemental Appendices.
Database and manual searches yielded 491 articles (Fig 1). After screening and review, 23 articles were included in qualitative or quantitative analysis, with between 2 and 8 articles (total: 16) included in various meta-analyses.
Tables 3 and 4 show the characteristics of included studies, according to presence of FASD (10 studies)13–21 or PAE (14 studies), respectively22–34: one study reported data for both PAE and FASD.35 Most were from North America (n = 15; 65.2%), followed by Australia (n = 4) and Canada (n = 2), with 1 study each from Finland and Taiwan. Four North American studies assessed predominantly African American (83%–94%)24,30 or exclusively African American cohorts.32,33 In addition to PAE/FASD groups, 3 studies included subgroups with attention-deficit/hyperactivity disorder (ADHD),15,26,34 and 1 study included a subgroup with prenatal cocaine exposure (Supplemental Appendix 2).33
Additional data (eg, individual group sample size data) were requested for 3 articles,17,23,29 but authors were unable to access archives (n = 2) or did not respond (n = 1). Sample sizes for Robinson et al31 were estimated based on number and corresponding percentage of children scoring within “Clinical” ranges for the primary ASEBA outcomes.
Of 23 studies included, 12 (52.2%) were retrospective cohort studies. Other study types included prospective cohort studies (n = 5), cross-sectional studies (n = 2), follow-up analysis studies (n = 2), and historical prospective cohort studies (n = 2; Tables 3 and 4) . In 13 studies, participants were accessed from common sources/databases: 5 sources/databases provided cases for 2 studies each and a sixth provided cases for 3 studies, although there was no individual subject overlap for 2 of these 3 studies. All but 1 study used the CBCL form (95.7%), 4 studies reported TRF data, and 2 used the YSR. One reported both CBCL and YSR results and 4 reported both CBCL and TRF results. Tables 3 and 4 detail the ASEBA version(s) used in each study. Due to the small number of studies reporting consistent TRF and YSR results, meta-analysis was restricted to studies that used the CBCL and reported our primary outcomes of interest: Total problems, Internalizing problems, and Externalizing problems scores. The Total competence score from the CBCL was not used in meta-analysis, as it was reported in only 3 studies (2 FASD studies, 1 PAE study). Detailed data are reported in the Supplemental Appendices.
Results for a total of 21 different ASEBA scales were reported in studies of FASD and 44 in studies of PAE. The number of FASD studies reporting data for any 1 scale ranged from 1 to 8 (median of 3 studies per scale). In PAE studies, between 1 and 9 studies provided data for any 1 scale but 28 scales (63.6% of all scales reported in PAE studies) were each reported by only 1 study. Data on children with FASD or PAE were reported for 24 CBCL scales, 19 TRF scales, and 3 YSR scales (total of 46 scales).
Parent-Rated Behavior: Primary Outcomes (Meta-analysis)
Presence of an FASD Diagnosis: CBCL Total, Internalizing and Externalizing Problems, Total Competence
FASD diagnostic criteria used in each study are included in Table 5. For all FASD studies, diagnostic assessments were made by experienced dysmorphologists, specially trained professional interdisciplinary teams, or trained clinicians. In FASD studies, very few CBCLs were completed by the biological parent (3.0%–22.7%).
Ten studies reported T scores for Total problems, or Internalizing or Externalizing problems in children with FASD (Fig 2). One study was excluded from the Total problems component of this meta-analysis, as control group sample size was not available.17 Figure 2 indicates significantly higher Total problems scores (pooled mean difference [95% CI]: 12.1 [7.7–16.5]; P < .0001), Internalizing problems scores (pooled mean difference [95% CI]: 6.3 [3.1–9.5]; P < .0001), and Externalizing problems scores (pooled difference [95% CI]: 12.5 [7.9–17.0]; P < .0001) in children with a FASD diagnosis compared with those without.
Four FASD studies reported the number of children scoring within “Clinical” ranges for Total, Internalizing, or Externalizing problems,13,14,17,20 2 of these reporting data from a comparison group.13,20 Compared with children without FASD, those with FASD were more likely to have Total (OR 34.0, 95% CI 2.6–450.8]), Internalizing (OR 10.0, 95% CI 1.3–77.6), or Externalizing problems scores (OR 18.2, 1.8–186.6) within the “Clinical” range, and less likely to have scores within the “Normal” range (Fig 3). They were also more likely to score in the “Borderline” range on the Total problems scale. There were no significant differences between groups in the ORs for scoring within the “Borderline” range for Internalizing and Externalizing problems (P > .05). Critical items important to clinical intervention (eg, Talks suicide, Attacks others), were not reported in any studies. There were insufficient studies to compare age groups.
Only 2 studies reported Total competence outcomes according to FASD diagnoses.17,19 Both studies reported poorer (lower) parent-rated Total competence scores in children with a FASD diagnosis compared with those without, regardless of FASD diagnostic subgroup. Compared with children with similar IQs, mean Total competence scores among children with fetal alcohol syndrome (FAS) appeared lower in boys (FAS 31.3 [7.4]; No FAS IQ comparison 45.4 [9.6]) but not in girls (FAS 40.5 [12.5]; No FAS IQ comparison 42.7 [7.3]; P not calculated).17 Olson et al17 reported that 5 of 9 children with FAS scored within the “Clinical” range, but did not report the finding in children without FAS (Table 5).
Presence of PAE: CBCL Total, Internalizing and Externalizing Problems; Total Competence
Prenatal Alcohol Exposure: T Scores
Seven studies reported raw (n = 2), percentile (n = 1), or T scores (n = 4) for primary outcomes. In these studies, the CBCL was usually completed by a biological parent (range 82.8%–100%). Of 4 studies reporting T scores, 1 did not report or provide subject numbers for PAE groups and could not be included in the forest plot.23 Of 3 studies reporting raw or percentile scores,30,32,33 only 2 provided SDs30,33 and only 1 had control group data.33 Therefore, there were inadequate raw score data for meta-analysis, so a forest plot was generated for the 3 studies reporting both T scores and subject numbers. Due to limited data available for this analysis, PAE was stratified as either present (for any duration) or absent, rather than indicated by low/moderate/high levels. There was a trend for higher problems scores (poorer outcomes) among children with PAE than without, although the forest plot reveals heterogeneity in results of the 3 studies included (Fig 4). The largest difference in problem scores between PAE and control groups was observed in 2 studies in which children in the PAE group were selected based on high levels of PAE (>4 alcoholic drinks at least once per week, or ≥13 drinks per week throughout pregnancy).26,34 Levels of PAE were not defined or reported in Brown et al.24 Pooled effects for the 3 summary problem scales showed no statistically significant differences between groups (P > .05; Fig 4). Age group comparisons were not possible.
Prenatal Alcohol Exposure: ORs
Of 23 studies included in this review, 14 stratified groups based on presence or absence of PAE or level of PAE; however, levels of PAE were inconsistent across different studies. Only 3 studies reported OR for our primary outcomes.29,31,32 Exploratory forest plots were created, incorporating the 3 primary “problems” scores (Total, Externalizing, and Internalizing problems) to examine whether odds of having problematic behavior ratings were higher after PAE. These primary outcomes, grouped by standardized PAE categories based on O’Leary et al,12 indicated no increased odds for poorer behavior ratings with higher PAE levels (data not shown). However, when data were grouped by PAE categories designated by study authors, the odds of poorer behavior ratings tended to fall in favor of moderate/high PAE compared with no PAE (Fig 5), although results were not statistically significant (P > .05). There was no significant increase in odds of behavioral problems with low/light/occasional PAE compared with no PAE (Fig 5). All studies included children aged between 5 and 10 years.
CBCL Total competence score according to PAE was reported in only 1 study.24 Children with no PAE had significantly higher (favorable) Total competence ratings than children with PAE, even if drinking had stopped during the second trimester (T scores: No PAE 16.6 [2.8]; Stopped drinking 13.8 [2.7]; Continued drinking: 13.5 [3.4]; P < .008).
A detailed table displaying the primary outcomes for all subgroups in included studies is provided in Supplemental Appendix 3.
Secondary Outcomes: Other Forms: Other CBCL Scores, TRF, YSR
Secondary outcomes included (1) other ASEBA School-Age Form scales and Critical items reported in eligible studies (rated by parents/caregivers, teachers, and/or youth), and (2) primary outcome scales rated by teachers and youth. Data were summarized qualitatively and highlights reported below. Detailed summaries of secondary outcomes based on presence of FASD diagnosis or PAE can be found in Supplemental Appendices 4 (summary) and 5 (table of results).
Presence of an FASD Diagnosis
Parent and teacher ratings of Total, Externalizing, and Internalizing problems were higher among children with FASD than without, even when compared with children with PAE but not FASD. CBCL scales showed the most common problems among children with FASD were the following: Thought problems, Rule-breaking behavior/Delinquency, Aggressive behavior, Attention problems, and Social problems. Children with FASD generally received higher Externalizing behavior scores than children with ADHD. However, parent and teacher ratings were inconsistent for Internalizing and Externalizing problems when comparing groups with FASD with ADHD (Supplemental Appendices 3, 4, and 5).
Total competence (and School competence subscale) scores were poorer among children with FASD than without (only parent ratings available).17,19 Scores on other competence subscales were mixed: in 1 study, children with FASD had better Social competence scores than those with PAE (without FASD),35 although scores in both groups were generally in the “Clinical” range. The Activities competence score in FASD was similar to comparison peers and within the “Normal” range in most studies. Somatic complaints,14,19,35 Anxiety/depression,14,19,35 Withdrawn/depressed,14,19,35 Academic performance (1 study),18 and Adaptive functioning (1 study)18 subscale scores were similar in FASD and comparison groups, and usually within the “Normal” range (Supplemental Appendices 4 and 5).
In 1 large study, CBCL ratings showed no differences in Total, Externalizing, or Internalizing problems in FASD diagnostic subgroups (FAS/partial FAS; static encephalopathy, alcohol-exposed; neurobehavioral disorder, alcohol-exposed) using the 4-digit diagnostic code.19 Critical items were not reported in any of the studies in our review.
Presence of PAE
The literature on PAE was less consistent than that on FASD in terms of behavior ratings among groups with different exposure levels, and ratings from different informant types. Generally, children with PAE had poorer parent and teacher ratings for Total competence and Total problems than nonexposed children. Most scales and subscales in the PAE literature were reported in only 1 study (Supplemental Appendices 4 and 5), and no study reported Critical items.
Risk of Bias
According to our methodological appraisal criteria (Table 2), scores for the 23 included studies ranged from 1 to 5 (mean: 2.7 [1.1]; Table 6). As per eligibility criteria, all studies reported ASEBA scores of interest. Only 1 study met criterion on sample size calculation/justification19 (which is not standard practice in reporting of observational studies). Most (9) studies received a score of 3. In relation to statistical adjustment for other prenatal exposures or psychiatric/behavior diagnoses, studies were scored “No” (ie, 0) if adjustments were not applied to ASEBA results, even if adjustments were applied to other variables (given the current study focus on FASD/PAE effects on ASEBA scores). We were unable to examine funnel plot asymmetry, as fewer than 10 studies were included in the meta-analyses.
This systematic review and meta-analysis examining behavior in FASD and PAE (as assessed using the popular ASEBA School-Age Forms) yields several important findings. First, of clinical concern, children with FASD had much poorer parent ratings on a range of behavioral outcomes than children without FASD. Second, there were no significant differences in behavior ratings in children with and without PAE when data for the most commonly reported parent-rated problem scales were pooled. Third, the published body of literature lacks quality and consistency of reporting. Over time, diagnostic systems have become more systematized and study quality has improved.
Our meta-analysis confirms that children with FASD have significantly higher scores for Total problems, Internalizing problems, and Externalizing problems than children without FASD. Pooled mean group differences in T scores ranged from 6 for Internalizing problems to 12 for Total and Externalizing problems. Children with FASD were more likely to have scores within the “Clinical” range for these primary measures with pooled ORs ranging from 10 (95% CI 1.3–77.6) to 34 (95% CI 2.6–450.8). Children with “Clinical” scores usually require therapeutic intervention. Although a limited number of scales were reported by both parents and teachers, there was general agreement between informants that children with FASD had poorer behavior than children without, with some inconsistencies when FASD and ADHD (without PAE) groups were compared.
Few informants for FASD groups were biological parents, whereas most informants in PAE studies were biological parents. The effect of different informants and home placements on CBCL ratings is unclear. Fagerlund et al13 was the only author to compare CBCL ratings in different living situations: children in residential care had more Total and Internalizing problems than those in foster/adoptive homes or biological homes, and more Externalizing problems than children in foster/adoptive homes (P < .05). This may reflect the informants’ viewpoint or the impact of institutional care. Paley et al18 used the Parenting Stress Index and reported higher stress levels in foster/adoptive parents (77%) than biological parents (23%), but did not compare CBCL scores between these informants. Only 1 PAE study incorporated home placement as a covariate in their CBCL analyses and found no significant effect of place of residence on behavior.34
In our qualitative assessment, children with FASD had poorer scores on the following scales: Total competence, Total problems, Internalizing problems, Externalizing problems, School competence, Rule breaking behavior/delinquency, Aggressive behavior, Attention problems, Social problems, and Social competence. These findings are in agreement with previous research reporting that items from the Rule breaking, Attention problems, and Aggressive CBCL scales discriminated children with FASD from children with ADHD but no PAE, oppositional defiant/conduct disorder, and typically developing healthy control children, although competence items were not examined in those studies.7 These findings concur with studies that use different behavioral assessments.36,37 Congruent with our review findings, a major focus in FASD intervention research is to develop and validate interventions for challenging behaviors, social skills deficits, and problems with self-regulation.38
No significant problems were observed in children with FASD in Activities competence, Somatic complaints, Withdrawn/depressed, Anxiety/depression, Academic performance (1 TRF study), Adaptive functioning (1 TRF study), or Hyperactive scales. These findings do not concur with studies that use other assessment tools, which report impairments in most of these areas,37,39 although few studies in this review reported each of these outcomes. It is possible that ASEBA items assessing these constructs are insufficiently sensitive to these deficits in the FASD population.
We found no group differences in behavior among children with and without PAE. The number of studies included was limited and PAE levels were inconsistently defined, so it was difficult to establish a dose-response pattern by standardized definitions. However, when using classifications for PAE levels provided by authors, there was a tendency toward increasing behavior problems with higher levels of PAE. Most articles reported only a few of the possible 78 scales from the CBCL, TRF, and YSR inclusive, and 28 scales were reported in only 1 study, precluding generalization of results.
A potential limitation of this review was the focus on assessments using the ASEBA. However, in many included studies, other measures used concurrently identified similar behavior problems. The review unveiled limitations within the available literature. Study quality was generally low, perhaps due to inclusion of any observational study that reported outcomes of interest, regardless of whether ASEBA scores reported were of primary interest in that study (hence many studies did not calculate sample sizes or statistical adjustments for ASEBA results). Our focus on the ASEBA scales should be set in context for clinicians. We have documented significant behavior problems in FASD; however, challenging behavior is only 1 aspect of central nervous system dysfunction in FASD. Neurocognitive and communication deficits, for example, are also common.
Several studies drew data from the same databases, so some children may have been included more than once. Thus, subjects in different studies may not be independent. Different FASD diagnostic systems were used in included studies, as several different criteria are used internationally. However, the consistency of behavioral outcomes within our FASD meta-analyses suggests little effect of diagnostic criteria on behavior ratings assessed by using the ASEBA School-Age Forms.
The paucity of data made it difficult to assess several ASEBA scales or to stratify by age. Inconsistencies in reporting of results precluded assessment of publication bias because <10 studies could be included in any meta-analysis. Although risk of bias could not be addressed quantitatively in this review, it is worth mentioning that data from 2 different kinds of studies were included: Longitudinal cohort studies (based on PAE), and case-control studies (based on FASD diagnosis). Findings from separate PAE and FASD meta-analyses are intuitive, and possibly reflect a detection bias favoring greater effects in children with FASD than PAE. This is because women drinking alcohol during pregnancy do not necessarily give birth to a child with behavioral impairments or FASD, whereas children with an FASD diagnosis are more likely to show greater severity of impairments.
There are strengths in the current study, including the meta-analysis of results derived from the most commonly reported behavior assessment tool used to study FASD and PAE. Additionally, we summarized all scales and subscales reported in the literature using the ASEBA School-Age Forms in this systematic review.
Research and Clinical Implications
Overall, this review unveiled behaviors apparently less problematic in FASD, behaviors in need of additional study due to inconsistent findings, or an insufficient number of studies on particular scales. It also revealed consistency in behavior ratings between different studies and by different informants in studies of FASD, but not PAE.
Findings provide direction for future research on FASD and PAE. Descriptive data comprehensively detailing behavioral deficits are crucial to guide research on interventions for the surprisingly prevalent FASD. To improve quality of the knowledge base, there is clear need for consistency in (1) reporting of and stratification according to PAE levels and timing, (2) selection of ASEBA scales, and (3) reporting of ASEBA scores. When feasible, methodology changes are needed. For instance, in ASEBA research, it would be useful to use multiple informants, to blind assessors to PAE/FASD status, and to control for other psychiatric/behavior diagnoses with similar symptoms by using contrast groups (eg, ADHD or IQ comparison groups with no teratogenic exposure). More complete investigation and reporting of the often overlooked ASEBA subscales and Critical items are warranted, especially given their significance for clinical practice.
Behavior problems are almost universal in children with FASD and significantly affect family life and schooling. From a clinical perspective, a clear understanding of behaviors in children with FASD and after PAE is important to families and providers seeking to improve outcomes for affected individuals. This review makes clear that providers should ask about PAE, and refer for assessment/diagnosis of FASD when appropriate. Indeed, FASD should be included in the differential diagnosis of any child with behavioral difficulties. Children with FASD and clinically concerning behavior problems may benefit considerably from early intervention, and knowledge of behaviors associated with FASD will inform diagnosis and management.
The authors thank the librarian at The Children’s Hospital at Westmead, Ms Trish Bennett, for her help with the electronic search strategy.
- Accepted November 20, 2015.
- Address correspondence to Tracey W. Tsang, PhD, Kids’ Research Institute (APSU), The Children’s Hospital at Westmead, Locked Bag 4001, Westmead NSW 2145, Australia. E-mail:
This trial has been registered with PROSPERO (http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42014007040#.VOJ7A8lI_MA) (identifier CRD42014007040).
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: The Discipline of Paediatrics and Child Health, Sydney Medical School, The University of Sydney, New South Wales, Australia, is responsible for initiating and managing the review. Drs Tsang and Pinto are funded by a National Health and Medical Research Council project grant (1024474). Prof Elliott is supported by the National Health and Medical Research Council Practitioner Fellowships (457084 and 1021480). Ms Lucas was supported by a fellowship from the Poche Centre for Indigenous Health, Sydney Medical School, The University of Sydney, Sydney, Australia. Prof Carmichael Olson has ongoing salary support from the University of Washington School of Medicine and Seattle Children’s Research Institute.
POTENTIAL CONFLICT OF INTEREST: Prof Carmichael Olson was an author of one of the articles included in the review; the other authors have indicated they have no potential conflicts of interest to disclose.
- Copyright © 2016 by the American Academy of Pediatrics