The Sex-Specific Risk of Autism Spectrum Disorders Following Low Birth Weight

Background: In large epidemiological studies preterm birth, low birth weight (<2500g, LBW), and very low birth weight (<1500g, VLBW) have been associated with increased risks of autism. Recently, the risks of autism associated with preterm birth and LBW were shown to differ in autism subgroups defined by sex and the presence of other developmental disabilities. Objectives: To study the sex-specific risk of ASD in LBW groups in a population-based cohort study. Methods: From the Danish Medical Birth Register all live born children in Denmark from 1990 through 2003 were identified as the study population comprising a total of 931,770 children. Through linkage with the Danish Civil Registration System information on autism spectrum disorder (ASD) diagnoses (F84.0, F84.1, F84.5, F84.8, and F84.9 according to ICD-10) was retrieved from The Danish Psychiatric Central Register. By December 31, 2008, 7,536 children (6,138 boys and 1,398 girls) from the study population were diagnosed with ASD. Information on birth weight (BW) and important covariates were retrieved from the Danish Medical Birth Register. We assessed the sex-specific hazard ratios (HRs) of ASD in a low-normal BW group (2500g-2999g) and LBW groups (2000g-2499g,1500g-1999g, and <1500g) when compared to a reference BW group (3000g-3999g) using Cox regression. The HRs can be interpreted as relative risks, and is abbreviated RRs in the following. Results: Preliminary analyses stratified on sex show statistically significant increased RRs of ASD for the low-normal BW group (2500g-2999g) and the LBW groups (2000g-2499g,1500g-1999g, and <1500g) when compared to the reference BW group (3000g-3999g) for girls, with estimated RRs of 1.53 (95%CI 1.20, 1.94), 1.96 (95%CI 1.31, 2.91), 2.27 (95%CI 1.17, 4.44), and 2,51 (95%CI 1.11, 5.67) respectively. The trend of RRs increasing with decreasing BW for girls is highly statistically significant (p<0.001). Corresponding RRs of ASD for the low-normal BW group (2500g-2999g) and the LBW groups (2000g-2499g,1500g-1999g, and <1500g) when compared to the reference BW group (3000g-3999g) for boys were lower than for girls: 1.19 (95%CI 1.03, 1.36), 1.19 (95%CI 0.91, 1.55), 1.35 (95%CI 0.87, 2.10), and 1.93 (95%CI 1.10, 3.37), respectively. There was no statistically significant trend of increasing RRs with decreasing BW for boys (p=0.189). The provided estimates are adjusted for gestational age, apgar score, parity, multiples and birth year. Conclusions: VLBW is documented as a risk factor for ASD in both sexes, but for the other LBW groups (2000g-2499g,1500g-1999g) the estimated RRs for ASD are statistically significant for girls only. Further, a highly significant trend of increasing RRs of ASD with decreasing BW is documented in girls but not in boys.