Worldwide prevalence of fetal alcohol spectrum disorders: a systematic literature review including meta-analysis

BACKGROUND: Although fetal alcohol spectrum disorders (FASD) affect communities worldwide, little is known about its prevalence. The objective of this study was to provide an overview of the global FASD prevalence. METHODS: We performed a search in multiple electronic bibliographic databases up to August 2015, supplemented with the ascendancy and descendancy approach. Studies were considered when published in English, included human participants, and reported empirical data on prevalence or incidence estimates of FASD. Raw prevalence estimates were transformed using the Freeman-Tukey double arcsine transformation so that the data followed an approximately normal distribution. Once the pooled prevalence estimates, 95% confidence intervals and prediction intervals were calculated based on multiple meta-analyses with transformed proportions using random effects models, these estimates were transformed back to regular prevalence rates. Heterogeneity was tested using Cochran's Q and described using the I(2) statistic. RESULTS: Among studies that estimated prevalence in general population samples, considerable differences in prevalence rates between countries were found and therefore separate meta-analyses for country were conducted. Particularly high-prevalence rates were observed in South Africa for fetal alcohol syndrome (55.42 per 1,000), for alcohol-related neurodevelopmental disorder (20.25 per 1,000), and FASD (113.22 per 1,000), For partial fetal alcohol syndrome high rates were found in Croatia (43.01 per 1,000), Italy (36.89 per 1,000), and South Africa (28.29 per 1,000). In the case of alcohol-related birth defects, a prevalence of 10.82 per 1,000 was found in Australia. However, studies into FASD exhibited substantial heterogeneity, which could only partly be explained by moderators, most notably geography and descent, in meta-regressions. In addition, the moderators were confounded, making conclusions as to each moderator's relevance tentative at best. CONCLUSIONS: The worldwide pooled prevalence estimates are higher than assumed so far, but this was largely explained by geography and descent. Furthermore, prevalence studies varied considerably in terms of used methodology and methodological quality. The pooled estimates must therefore be interpreted with caution and for future research it is highly recommended to report methodology in a more comprehensive way. Finally, clear guidelines on assessing FASD prevalence are urgently needed, and a first step toward these guidelines is presented.