Background Intrauterine growth restriction has been linked with adverse outcomes including neonatal death, and diabetes and cardiovascular disease in adulthood. However, there is no consensus regarding definition of growth restriction.

Objective To determine which definition and threshold of growth restriction best predicts short and long term outcome in term (≥37 weeks gestation) born individuals.

Methods A systematic review was performed according to published guidance. Electronic searches were conducted across multiple databases (inception to April 2011) without language restrictions. Studies were selected by two reviewers and results extracted. Authors were contacted where necessary. Meta-analyses were performed according to birth weight parameter, threshold and outcomes.

Results 90 articles were included, with 23,637,684 individuals. Birth weight <2.5kg had a positive likelihood ratio (LR+ve) 5.26 (95% CI 3.57 to 7.76) to predict neonatal death. A threshold of 1.5kg had LR +ve 49.1 (95% CI 27.2 to 88.54). A threshold of < 10th percentile on population growth chart had LR +ve of 2.48 (95% CI 2.21 to 2.78) for the same outcome. Reducing the threshold to < 5^th centile gave LR +ve 4.67 (95 % CI 2.84 to 7.67) and < 3^rd centile 6.31 (95% CI 3.57 to 11.14). Birth weight <2.5kg had LR +ve 1.34 (95% CI 0.87 to 2.05) to predict morbidity in adulthood and <10^th centile LR+ve 1.23 (95%CI 1.01 to 1.50) for the same outcome.

Conclusion Absolute birth weight is a better predictor of neonatal mortality than centiles. Birth weight by any definition does not predict adult outcomes in term infants.