Low birth weight and intelligence in adolescence and early adulthood: a meta-analysis.

BACKGROUND Research has demonstrated an association between low birth weight (LBW; <2500 g) and adverse intelligence quotient (IQ) outcomes in childhood and early adolescence. We systematically evaluated whether this association persists into late adolescence and early adulthood and also assessed the influence of age of IQ assessment on effect size. METHODS During Stage 1 (meta-analysis of data on adolescents/adults), we searched for relevant articles in PsychINFO, PubMed, Ovid, CINAHL, ProQuest and ERIC until February 2011 (no lower limit). Studies which assessed full-scale IQ among LBW individuals (<2500 g), aged 13 years and older, with a normal birth weight (NBW; ≥2500 g) comparison group were eligible. A random-effects meta-analysis provided a pooled estimate of the difference in IQ scores between LBW and NBW individuals. Publication bias was assessed using Rosenthal's classic fail-safe N and Duval and Tweedie's Trim and Fill. During Stage 2, we added data from the Kerr-Wilson et al. meta-analysis (which included data from children; in Meta-analysis of the association between preterm delivery and intelligence. Journal Public Health 2011;33:1-8) to our sample from Stage 1 and conducted a meta-regression to evaluate the effect of age of IQ assessment. RESULTS Using a total of 15 studies in Stage 1, it was demonstrated that NBW individuals scored an average of 7.63 IQ points higher than LBW individuals, CI = 5.95-9.31. After adjusting for publication bias, NBW samples demonstrated an IQ of 4.98 points higher than LBW samples, CI = 3.20-6.77. Furthermore, age at IQ assessment was a significant moderator of the association between birth weight and IQ, in that the effect size decreased from childhood into young adulthood. CONCLUSIONS Cognitive impairments associated with LBW persist into adolescence and early adulthood; however, the influence of LBW on IQ decreases from childhood to young adulthood. These conclusions must be interpreted with caution due to unmeasured variables and possible influence from publication bias.