Provider’s node attributes contribute to maternal health and birth outcomes

Abstract

Abstract Background: Birth outcomes are deteriorating in the United States. These deteriorating birth outcomes can be partially explained by the medical needs during pregnancy. Meanwhile, there is an ongoing decline in the supply of obstetricians/gynecologists in the American workforce. Despite this shortage, providers can coordinate care through referral networks for pregnant women to ensure that necessary care can be accessed. This study aims to investigate the association between the network characteristics of prenatal care (PNC) providers and measures of maternal health and birth outcomes. Methods: This unbalanced time-series study utilised a South Carolina Medicaid claims dataset of live births in facilities during 2015-2018. The sample included providers (nodes) who served pregnancies with continuous Medicaid enrollment. Node attributes include patient volume, the percentage of shared patients, the number of peer connections (degree), and community size. Four dummy variables were created as independent variables to indicate an increase or not in each node attribute during two successive years. Interested outcomes were the annual PNC frequency, cost, and the percentages of C-section, preterm birth and low birth weight (LBW, < 2,500 grams). Univariate and bivariate analyses were conducted. Linear regressions with repeated measures were utilised due to the nature of repeated measures for the same provider in multiple years. Covariates controlled for patient, provider and area characteristics. Findings: The probability of an increase in the four node characteristics ranged from 41.3% to 56.0%. Increased patient volume and increased degree were associated with utilisation and birth outcomes in bivariate analysis without or with repeated measure. In the adjusted model, however, increased patient volume was only associated with decreased risks of preterm birth (coefficient: -0.92, p < 0.01) and increased degree was only associated with visit numbers (coefficient: .10, p < .001). In both unadjusted and adjusted analyses, increased community size was significantly associated with decreased risks of preterm birth (coefficient: -2.57, p < .001) and LBW (coefficient: -1.09, p < .001). Conclusions: Providers with increased node attributes may deliver more services and improve birth outcomes. The current study suggests that policy should encourage and support necessary referrals.