Preconception Care for Women with Medicaid: Self-report vs. Claims-based Utilization Measures

Background Preconception care may improve perinatal outcomes and reduce disparities, but there is no standard population measure of preconception care utilization (PCU). Objective We compared claims-based PCU from Medicaid Analytic Extract (MAX) data to self-report in the Pregnancy Risk Assessment and Monitoring System (PRAMS) survey. Methods Among Medicaid-enrolled women ages 15-45 with births during 2012, we identified preconception services in MAX using 55 ICD9 codes published by Health and Human Services. We estimated the proportion reporting preconception care from 26 PRAMS states and compared this to the states’ proportion who received services in MAX. We fit mixed-effects logistic regression models of the probability of PCU on demographic factors (age, race/ethnicity) and diagnoses (depression, diabetes, or hypertension), separately for each dataset. Finally, we computed the population proportions receiving care by state (MAX) and the empirical Bayes means of the state-level effects (MAX and PRAMS). Results Among 652,929 deliveries in MAX from the included states, 28.1% received at least one preconception service. In PRAMS, 23.6% (95% CI [22.1, 25.3]) of Medicaid-covered respondents reported preconception care. In both datasets, PCU rates were higher for Black non-Hispanic vs. White non-Hispanic women (PRAMS OR 2.05 [1.60, 1.62]; MAX OR 1.51 [1.49, 1.54]) and for those with diabetes (PRAMS OR 1.82 [1.16, 2.85]; MAX OR 1.34 [1.29, 1.40]) or hypertension (OR 1.85 [1.41, 2.44]; MAX OR 1.22 [1.18, 1.27]). In PRAMS, Asian (OR 3.37 [2.28, 4.98]) and Hispanic women (OR 2.07 [1.5, 2.80]) were more likely to report PCU than White non-Hispanic women, but in MAX they were less likely to receive services. The correlation between the PRAMS state-specific effects and those from MAX was 0.31 (p = 0.124). Conclusions Claims-based estimates of PCU are moderately concordant with self-reported rates at the state level; however, rates measured through Medicaid claims vs. self-report diverge in some groups. Synopsis Study Question How do Medicaid claims-based measures of preconception care utilization compare to self-reported receipt of preconception counseling among Medicaid-covered respondents in the Pregnancy Risk Assessment and Monitoring System (PRAMS) survey? What’s already known PRAMS provides population-level estimates of preconception care utilization, while claims-based measures quantify specific services received. What this study adds Claims-based preconception care utilization among the Medicaid population varies by race (Black/White) and diagnosis of diabetes or hypertension in similar patterns as self-report in PRAMS, but ethnicity (Hispanic/non-Hispanic) and depression demonstrate divergent patterns between the two data sources. State-level variation in preconception care utilization is greater in claims data. Both data sources can be used by researchers with an understanding of their methodological benefits and limitations.