Selection bias in prenatal care use by Medicaid recipients.

摘要: Objective: This study examines whether four types of selection bias in estimates the effectiveness prenatal care utilization for improving birthweight occur a population economically disadvantaged women. Methods: We categorized adequacy use using Adequacy Prenatal Care Utilization Index (APNCU) and Revised-GINDEX 142,381 Medicaid recipients who gave birth to live, singleton infant Washington State (1994–1998). Multinomial logistic regression was used model categories as functions variables chosen indicate high- or low-risk status. A series linear models were estimated quantify magnitude direction any effects on that could be attributed accounting each risk covariate. Results examined patterns risk, use, estimation equated with processes. Results: found modest evidence adverse, favorable, confidence, estrangement biases. The overriding effect, relative low overestimation adequate coefficient by 8.68 g APNCU, underestimation 3.36 R-GINDEX because competing confidence Relative intensive effect underestimated (17.58 APNCU; 13.34 R-GINDEX) adverse small countervailing favorable process. Conclusions: associated noted prior studies appears result from multiple processes working different directions. Understanding can help assessment contribution outcomes development appropriate programs policies.