Autism Spectrum Disorder Is Associated with an Increased Risk of Development of Underweight in Children: A Systematic Review and Meta-Analysis

Background: Underweight, defined as having a BMI of ≤ 5^th percentile for age and sex (Kuczmarski et al., 2000), is associated with multiple macro- and micro-nutrient deficiencies, a high risk of acquiring infections (e.g. pneumonia, gastrointestinal infection) (Dobner & Kaser, 2018) and a higher mortality rate (Wake et al., 2013). Possibly due to restricted dietary patterns such as food selectivity (Bandini et al., 2010; Sharp et al., 2018) and abnormal meal time behaviors (Malhi et al., 2017), children with autism spectrum disorders (ASD) are often thought to be at a higher risk of being underweight, compared to typically developing children. However, evidence on the association between ASD and underweight are equivocal.

Objectives: To conduct a systematic review and meta-analysis of the literature to examine the prevalence and relative risk of underweight in children with ASD.

Methods: ProQuest, PubMed, Scopus and Web of Science databases were systematically searched per the PRISMA guidelines (see Figure 1). Results were screened to identify studies that reported the prevalence of underweight in children with ASD and when available, in typically developing control groups. DerSimonian–Laird random‐effects meta‐analyses were performed using the ‘meta’ package in R software to determine the pooled prevalence and the relative risk of underweight among children with ASD.

Results: Underweight had a prevalence of 7.2% (95% CI, 5.5-9.4) among children with ASD overall (24 study arms) and a prevalence of 4.8% (95% CI, 4.3-5.4) among children with ASD in the United States (9 study arms). These estimates were significantly greater than the prevalence of childhood underweight in the United States in 2013-2014 (i.e. 3.8%) (Fryar, Carroll, & Ogden, 2016). A subsequent exploratory univariate meta-regression analysis revealed that the proportion of non-Caucasian children in a given sample had a significant positive moderator effect (β = 0.470; SE = 0.185; 95% CI = 0.108, 0.832; p = 0.011) on the prevalence on underweight on children with ASD. Similarly, mean age of the study sample (β = 0.109; SE = 0.054; 95% CI = 0.004, 0.214; p = 0.042) was an independent positive moderator, while the year of publication was an independent negative moderator (β = -0.081; SE = 0.032; 95% CI = -0.145, -0.018; p = 0.012). The risk of being underweight for children with ASD overall and for children with ASD in the United States were 76.4% ((95% CI, -0.4-212.3; 8 studies) and 59.8% (95%CI, -4.1, 62.8; 4 studies) had a trend of being greater than typically developing controls (p = 0.0515 and p = 0.066 respectively). Year of publication negatively moderated this association (β = -0.176; SE = 0.066; 95% CI = -0.305, -0.047; p = 0.007).

Conclusions: Children with ASD seem to have a greater likelihood of becoming underweight. Non-Caucasian and older children with ASD appear to have a greater risk of developing underweight. Future research studies should be vigilant about exploring risk factors that may lead to development of underweight in this special population of children and explicating the role of these risk factors in interventions.