Applying measures of discriminative accuracy to revisit traditional risk factors for small-for-gestational-age

INTRODUCTION:  Small-for-gestational-age (SGA) is considered an indicator of intrauterine growth restriction. Multiple maternal and new-born characteristics have been identified as ‘risk factors’ for SGA. This knowledge is mainly based on measures of average association (i.e., differences in average risk between exposed and unexposed groups) like odds ratio (OR). Nevertheless, average associations do not assess the discriminatory accuracy of the risk factors (i.e., its ability to discriminate those babies who will develop SGA from those who will not). Therefore, applying measures of discriminatory accuracy, we aim to revisit the role of a number of traditional risk factors for being born SGA. 

METHODS:  Using the Swedish Medical Birth Register we investigated 731,989 babies born during 1987-1993. We measure maternal (smoking, hypertension, age, marital status, education) and child (sex, gestational age, birth order) characteristics and performed logistic regression models to estimate both ORs and measures of discriminatory accuracy like area under the ROC curve (AU-ROC), Net Reclassification Improvement (NRI), Integrated Discrimination Improvement (IDI) and risk assessment plots.

RESULTS:  We replicated the expected associations. For instance, smoking (OR= 2.57) and hypertension (OR=4.02) were strongly associated to SGA. However, their discriminative accuracy was minor (the AU-ROC for smoking was 59% and for hypertension 51%) The inclusion of all variables studied into a model considerably improved the AU-ROC= 69% but discriminatory accuracy was unsatisfactorily low.  

CONCLUSIONS:  Applying measures of discriminatory accuracy rather than measures of association only, our study revisits known risk factors of SGA and discusses their role from a public health perspective. We found that neither models including simple variables nor models including several variables at the same time have a good discriminatory accuracy to discriminate babies with SGA from those without SGA. This finding is of fundamental relevance in order to address future research and to design policymaking recommendation in a more informed way