Prediction model for low birth weight and its validation

Indian J Pediatr. 2014 Jan;81(1):24-8. doi: 10.1007/s12098-013-1161-1. Epub 2013 Aug 16.

Abstract

Objective: To evaluate the factors associated with low birth weight (LBW) and to formulate a scale to predict the probability of having a LBW infant.

Methods: This hospital based case-control study was conducted in a tertiary care university hospital in North India. The study included 250 LBW neonates and 250 neonates with birth weight ≥2,500 g. Data were collected by interviewing mothers using pre-designed structured questionnaire and from hospital records.

Results: Factors significantly associated with LBW were inadequate weight gain by the mother during pregnancy (<8.9 kg), inadequate proteins in diet (<47 g/d), previous preterm baby, previous LBW baby, anemic mother and passive smoking. The prediction model made on these six variables has a sensitivity of 71.6 %, specificity 67.0 %, positive LR 2.17 and negative LR of 0.42 for a cut-off score of ≥29.25. On validation, it has a sensitivity of 72 % and specificity of 64 %.

Conclusions: It is possible to predict LBW using a prediction model based on significant risk factors associated with LBW.

Publication types

  • Validation Study

MeSH terms

  • Case-Control Studies
  • Female
  • Forecasting
  • Humans
  • Infant, Low Birth Weight*
  • Infant, Newborn
  • Male
  • Models, Statistical*
  • Risk Assessment