Maternal anthropometric measurements as predictors of low birth weight in developing and developed countries.

Abstract

To evaluate the possibility of using maternal anthropometric measurements for prediction of low birth weight in developing and developed countries. Bivariate diagnostic meta-analysis was performed with hierarchical summary receiver operating characteristic curves. Ten databases, i.e., PubMed, EMBASE, CINAHL, PsycINFO, Wiley InterScience, ProQuest Health and Medical Complete™, ProQuest Dissertations and Theses Database, the entire Cochrane Library (e.g., CENTRAL), Google Scholar, and Scopus were searched. The references and PubMed-related citations of potentially eligible articles and reviews were also investigated. With regard to maternal height, weight, arm circumference, abdominal circumference, body mass index (BMI), and weight gain, 111, 126, 25, 4, 131, and 59 studies extracted from 49, 38, 12, 1, 50, and 23 data sources, respectively, were finally included in the analysis. The sensitivity [=0.46, 95% confidence interval (CI) 0.35-0.52 to 0.64, 95% CI 0.56-0.71], specificity (=0.46, 95% CI 0.38-0.54 to 0.72, 95% CI 0.64-0.79) and diagnostic odds ratios (=2, 95% CI 1-2 to 3, 95% CI 3-4) for maternal height, weight, arm circumferences, BMI, and weight gain were low and 95% confidence and prediction regions were too large for practical use. The findings for abdominal circumference were not generalizable because only one data source was available. There is no evidence that maternal anthropometric measurements are good predictors of low birth weight.