Assessing the economic impact of paternal involvement: A comparison of traditional statistical models and decision analysis trees

Introduction: The U.S. continually struggles with poor health outcomes despite enormous healthcare expenditures. Underutilized, novel methods in cost effectiveness analysis and comparative effectiveness research have been emphasized by the U.S. government to reduce healthcare spending and improve health outcomes. Using both traditional statistical models and decision analysis trees, we assessed the economic impact of paternal involvement during pregnancy. Methods: We linked birth and death certificates to inpatient hospital discharge records for all singleton births in Florida from 1998-2005. Paternal involvement status was based on presence/absence of paternal first and/or last name on the birth certificate. Our primary outcome was mean inpatient hospitalization costs over the first year of life. Using cost-to-charge ratios and adjustment factors, we converted hospital record charges to refined cost estimates taking into account hospital and departmental variation in markup. We first compared costs using a generalized linear model (GLM) with a gamma distribution and log link. We then constructed a decision analysis tree, following infants through the first year of life and focusing on major outcomes, including method of delivery, gestational age, infant morbidity and mortality. Results: Overall, 10.7% of our cohort had a father-absent pregnancy. Decision analysis identified that paternal involvement results in a $1,139 savings per pregnancy. The GLM approach resulted in the same conclusion, although the estimated saving ($704 per pregnancy) was slightly lower. Discussion: When modeling cost, GLM and decision analysis approaches yield similar results and demonstrate that increasing paternal involvement can result in a significant reduction in infant hospitalization costs.