Birth weight and subsequent cholesterol levels: Exploration of the "fetal origins" hypothesis

Context: Inverse associations between birth weight and subsequent blood cholesterol levels have been used to support the "fetal origins" hypothesis of the relevance of fetal nutrition to adult disease.
Objectives: To perform a systematic review of the association between birth weight and total blood cholesterol levels, and to explore the impact of including unpublished results, adjusting for potential confounders.
Data Sources and Study Selection: Relevant studies published by September 30, 2004, were identified through literature searches using EMBASE and MEDLINE and MeSH heading search strategy (using terms such as birth weight, intrauterine growth retardation, fetal growth retardation and cholesterol, lipoprotein, lipid). Studies that reported qualitative or quantitative estimates of the association between birth weight and total blood cholesterol, or had recorded both measures but not reported on their associations, were included.
Data Extraction: A total of 79 relevant studies involving a total of 74122 individuals were identified; 65 had reported on the direction of the association between birth weight and total blood cholesterol. Although regression coefficients were published for only 11 studies and other quantitative estimates for 3 other studies, regression coefficients (published or unpublished) were obtained for 58 studies among 68974 individuals.
Data Synthesis: Inverse associations were observed in 11 of 14 studies that had previously published quantitative estimates but in only 18 of the remaining 51 that had reported on the direction of this association (heterogeneity P=.004). Similarly, the weighted estimate for the 11 studies was -1.89 mg/dL (-0.049 mmol/L) total cholesterol per kilogram birth weight compared with -0.69 mg/dL (-0.018 mmol/L) per kilogram for 47 studies that provided unpublished regression coefficients (heterogeneity P=.009). Overall, the weighted estimate from the 58 contributing studies was -1.39 mg/dL (-0.036 mmol/L) per kilogram (95% confidence interval, -1.81 to -0.97 mg/dL [-0.047 to -0.025 mmol/L]), but there was significant heterogeneity between their separate results (P<.001). Part of this heterogeneity appears to reflect stronger associations reported from smaller studies and studies of cholesterol levels in infants.
Conclusion: These findings suggest that impaired fetal growth does not have effects on blood cholesterol levels that would have a material impact on vascular disease risk.