Identifying multiple risks of low birth weight using person-centered modeling

Objectives: Low birth weight outcomes result from multiple potential risks. The present study used latent class analysis to identify subgroups of women with multiple co-occurring risks and to examine the relationship of these risk classes to low birth weight outcome.

Methods: Data were analyzed on all live singleton births in 2010 and 2011 in West Virginia (N = 28,820). Ten risks were examined including marital status, stress, mother's age, parity status, reported smoking and drug use during pregnancy, delayed prenatal care, Medicaid coverage, uninsurance, and low education.

Results: Six latent classes were identified that ranged from a low-risk referent group to higher risk classes characterized by unique constellations of risk factors. Compared with the low-risk referent, all of the remaining five latent classes were significantly associated with increased odds of low birth weight. However, one class was at especially high risk; this class was characterized by unmarried women in the Medicaid program who reported drug use, smoking, stress, and late prenatal care (odds ratio, 4.78; 95% confidence interval, 4.07-5.61).

Conclusions: The person-centered approach identified subgroups of women with unique risk profiles. The results suggest that eliminating a single risk would not resolve the low birth weight problem. Smoking, for example, co-occurs with higher stress and higher levels of drug use among a Medicaid population. It may be beneficial to develop and test tailored interventions to groups with specific co-occurring risks to reduce low birth weight outcomes. Programs targeted to women in the Medicaid program who also engage in substance use and experience stress are especially indicated.