Issue of Data Imbalance on Low Birthweight Baby Outcomes Prediction and Associated Risk Factors Identification: Establishment of Benchmarking Key Machine Learning Models With Data Rebalancing Strategies.

BACKGROUND: Low birthweight (LBW) is a leading cause of neonatal mortality in the United States and a major causative factor of adverse health effects in newborns. Identifying high-risk patients early in prenatal care is crucial to preventing adverse outcomes. Previous studies have proposed various machine learning (ML) models for LBW prediction task, but they were limited by small and imbalanced data sets. Some authors attempted to address this through different data rebalancing methods. However, most of their reported performances did not reflect the models' actual performance in real-life scenarios. To date, few studies have successfully benchmarked the performance of ML models in maternal health; thus, it is critical to establish benchmarks to advance ML use to subsequently improve birth outcomes. OBJECTIVE: This study aimed to establish several key benchmarking ML models to predict LBW and systematically apply different rebalancing optimization methods to a large-scale and extremely imbalanced all-payer hospital record data set that connects mother and baby data at a state level in the United States. We also performed feature importance analysis to identify the most contributing features in the LBW classification task, which can aid in targeted intervention. METHODS: Our large data set consisted of 266,687 birth records across 6 years, and 8.63% (n=23,019) of records were labeled as LBW. To set up benchmarking ML models to predict LBW, we applied 7 classic ML models (ie, logistic regression, naive Bayes, random forest, extreme gradient boosting, adaptive boosting, multilayer perceptron, and sequential artificial neural network) while using 4 different data rebalancing methods: random undersampling, random oversampling, synthetic minority oversampling technique, and weight rebalancing. Owing to ethical considerations, in addition to ML evaluation metrics, we primarily used recall to evaluate model performance, indicating the number of correctly predicted LBW cases out of all actual LBW cases, as false negative health care outcomes could be fatal. We further analyzed feature importance to explore the degree to which each feature contributed to ML model prediction among our best-performing models. RESULTS: We found that extreme gradient boosting achieved the highest recall score-0.70-using the weight rebalancing method. Our results showed that various data rebalancing methods improved the prediction performance of the LBW group substantially. From the feature importance analysis, maternal race, age, payment source, sum of predelivery emergency department and inpatient hospitalizations, predelivery disease profile, and different social vulnerability index components were important risk factors associated with LBW. CONCLUSIONS: Our findings establish useful ML benchmarks to improve birth outcomes in the maternal health domain. They are informative to identify the minority class (ie, LBW) based on an extremely imbalanced data set, which may guide the development of personalized LBW early prevention, clinical interventions, and statewide maternal and infant health policy changes.

Disparities in postpartum contraceptive use among immigrant women with restricted Medicaid benefits.

BACKGROUND: The Emergency Medicaid program offers restricted Medicaid benefits for people who meet the same financial eligibility criteria as Traditional Medicaid recipients but do not meet the citizenship requirements for enrollment in Traditional Medicaid. By federal law, Emergency Medicaid covers care for life-threatening emergencies or a hospital admission for childbirth. No prenatal or postpartum care is covered. Most of the women enrolled in Emergency Medicaid are Latina. OBJECTIVE: We assessed postpartum visits and receipt of postpartum contraception and compared the outcomes for Emergency (restricted benefit) Medicaid recipients with those of Traditional (full-benefit) Medicaid recipients in Oregon and South Carolina, 2 states with similar-sized immigrant populations. STUDY DESIGN: We conducted a retrospective cohort study using linked Medicaid claims and birth certificate data of live births covered by Medicaid (Traditional and Emergency) between January 1, 2010 and September 30, 2017, in Oregon and South Carolina. Our analysis was at the individual level. Primary outcomes were postpartum visit attendance and receipt of postpartum contraception within 2 months. We examined differences in demographic and delivery characteristics by Medicaid type. If women received postpartum contraception, we compared the timing of receipt (immediate postpartum, ≤1 month, 1-2 months, and 2-6 months after delivery) by the type of Medicaid. Among women using contraception, we described the type of contraceptive received at each time point, stratified by Medicaid type. Associations between Medicaid type (Traditional vs Emergency) and postpartum visit attendance and contraception use were assessed using adjusted absolute predicted probabilities from logistic regression models. We ran models for the entire cohort and conducted a subanalysis restricted to only Latina women. RESULTS: Our study included 375,544 live births to 288,234 women, with 12.7% of births among Emergency Medicaid recipients. Women enrolled in Emergency Medicaid tended to be older (age >35 years; 18.1% vs 7.2%; P<.001) and were more likely to be multiparous (76.8% vs 60.8%; P<.001) and Latina (80.3% vs 9.5%; P<.001) than their Traditional Medicaid peers. Among women enrolled in Emergency Medicaid, the probability of having a postpartum visit was 6.1% (95% confidence interval, 5.9-6.4) compared with 58.8% (95% confidence interval, 58.6-58.9) for women covered by Traditional Medicaid. After 6 months following delivery, 97.6% of Emergency Medicaid recipients had no evidence of contraceptive use compared with 55.6% of Traditional Medicaid enrollees (P<.001). In our adjusted model, Emergency Medicaid recipients were also significantly less likely to receive postpartum contraception than Traditional Medicaid enrollees (1.9% vs 35.5%; 95% confidence interval, [1.8-2.1] vs [35.4-35.7]). We examined the role that race may play in postpartum contraceptive use by conducting a subanalysis restricted to Latina women only.Latinas with births covered by Emergency Medicaid had a 1.9% (95% confidence interval, 1.8-2.0) adjusted probability of postpartum contraception use within 2 months compared with 39.8% (95% confidence interval, 38.7-39.9) among Latinas enrolled in Traditional Medicaid. CONCLUSION: Women enrolled in Emergency Medicaid experience large disparities in postpartum care and contraceptive use. Policies that restrict Medicaid coverage following delivery exacerbate inequities in postpartum care, potentially leading to worse health outcomes for low-income immigrants and their children.

Analysis of Contraceptive Use Among Immigrant Women Following Expansion of Medicaid Coverage for Postpartum Care.

Importance: Access to postpartum care is restricted for low-income women who are recent or undocumented immigrants enrolled in Emergency Medicaid. Objective: To examine the association of a policy extending postpartum coverage to Emergency Medicaid recipients with attendance at postpartum visits and use of postpartum contraception. Design, Setting, and Participants: This cohort study linked Medicaid claims and birth certificate data from 2010 to 2019 to examine changes in postpartum care coverage on postpartum care and contraception use. A difference-in-difference design was used to compare the rollout of postpartum coverage in Oregon with a comparison state, South Carolina, which did not cover postpartum care. The study used 2 distinct assumptions to conduct the analyses: first, preintervention differences in postpartum visit attendance and contraceptive use would have remained constant if the policy expanding coverage had not been passed (parallel trends assumption), and second, differences in preintervention trends would have continued without the policy change (differential trend assumption). Data analysis was performed from September 2020 to October 2021. Exposures: Medicaid coverage of postpartum care. Main Outcomes and Measures: Attendance at postpartum visits and postpartum contraceptive use, defined as receipt of any contraceptive method within 60 days of delivery. Results: The study population consisted of 27 667 live births among 23 971 women (mean [SD] age, 29.4 [6.0] years) enrolled in Emergency Medicaid. The majority of all births were to multiparous women (21 289 women [76.9%]; standardized mean difference [SMD] = 0.08) and were delivered vaginally (20 042 births [72.4%]; SMD = 0.03) and at term (25 502 births [92.2%]; SMD = 0.01). Following Oregon's expansion of postpartum coverage to women in Emergency Medicaid, there was a large and significant increase in postpartum care visits and contraceptive use. Assuming parallel trends, postpartum care attendance increased by 40.6 percentage points (95% CI, 34.1-47.1 percentage points; P < .001) following the policy change. Under the differential trends assumption, postpartum visits increased by 47.9 percentage points (95% CI, 41.3-54.6 percentage points; P < .001). Postpartum contraception use increased similarly. Under the parallel trends assumption, postpartum contraception within 60 days increased by 33.2 percentage points (95% CI, 31.1-35.4 percentage points; P < .001). Assuming differential trends, postpartum contraception increased by 28.2 percentage points (95% CI, 25.8-30.6 percentage points; P < .001). Conclusions and Relevance: These findings suggest that expanding Emergency Medicaid benefits to include postpartum care is associated with significant improvements in receipt of postpartum care and contraceptive use.

Coverage of immediate postpartum long-acting reversible contraception has improved birth intervals for at-risk populations.

BACKGROUND: In 2012, South Carolina revised the Medicaid policy to cover reimbursement for immediate postpartum long-acting reversible contraception. Immediate postpartum long-acting reversible contraception may improve health outcomes for populations at risk with a subsequent short-interval pregnancy. OBJECTIVES: We examined the impact of the Medicaid policy change on the initiation of long-acting and reversible contraception (immediate postpartum and postpartum) within key populations. We determined whether immediate postpartum long-acting and reversible contraception use varied by adequate prenatal care (>7 visits), metropolitan location, and medical comorbidities. We also tested the association of immediate postpartum and postpartum long-acting, reversible contraception on interpregnancy interval of less than 18 months. STUDY DESIGN: We conducted a historical cohort study of live births among Medicaid recipients in South Carolina between 2010 and 2017, 2 years before and 5 years after the policy change. We used birth certificate data linked with Medicaid claims. Our primary outcome was immediate postpartum long-acting and reversible contraception, and our secondary outcome was short interpregnancy interval. We characterize trends in long-acting and reversible contraception use and interpregnancy interval over the study period. We used logistic regression models to test the association of key factors (rural, inadequate prenatal care, and medical comorbidities) with immediate and outpatient postpartum long-acting and reversible contraception following the policy change and to test the association of immediate postpartum and postpartum long-acting and reversible contraception with short interpregnancy interval. RESULTS: Our sample included 187,438 births to 145,973 women. Overall, 44.7% of the sample was white, with a mean age of 25.0 years. A majority of the sample (61.5%) was multiparous and resided in metropolitan areas (79.5%). The odds of receipt of immediate postpartum long-acting and reversible contraception use increased after the policy change (adjusted odds ratio, 1.39, 95% confidence interval, 1.34-1.43). Women with inadequate prenatal care (adjusted odds ratio, 1.50, 95% confidence interval, 1.31-1.71) and medically complex pregnancies had higher odds of receipt of immediate postpartum long-acting and reversible contraception following the policy change (adjusted odds ratio, 1.47, 95% confidence interval, 1.29-1.67) compared with women with adequate prenatal care and normal pregnancies. Women residing in rural areas were less likely to receive immediate postpartum long-acting and reversible contraception (adjusted odds ratio, 0.36, 95% confidence interval, 0.30-0.44) than women in metropolitan areas. Utilization of immediate postpartum long-acting and reversible contraception was associated with a decreased odds of a subsequent short interpregnancy interval (adjusted odds ratio, 0.62, 95% confidence interval, 0.44-0.89). CONCLUSION: Women at risk of a subsequent pregnancy and complications (inadequate prenatal care and medical comorbidities) are more likely to receive immediate postpartum long-acting and reversible contraception following the policy change. Efforts are needed to improve access in rural areas.

Precision of provider licensure data for mapping member accessibility to Medicaid managed care provider networks.

BACKGROUND: In July 2018, the Centers for Medicare and Medicaid Services (CMS) updated its Medicaid Managed Care (MMC) regulations that govern network and access standards for enrollees. There have been few published studies of whether there is accurate geographic information on primary care providers to monitor network adequacy. METHODS: We analyzed a sample of nurse practitioner (NP) and physician address data registered in the state labor, licensing, and regulation (LLR) boards and the National Provider Index (NPI) using employment location data contained in the patient-centered medical home (PCMH) data file. Our main outcome measures were address discordance (%) at the clinic-level, city, ZIP code, and county spatial extent and the distance, in miles, between employment location and the LLR/NPI address on file. RESULTS: Based on LLR records, address information provided by NPs corresponded to their place of employment in 5% of all cases. NP address information registered in the NPI corresponded to their place of employment in 64% of all cases. Among physicians, the address information provided in the LLR and NPI corresponded to the place of employment in 64 and 72% of all instances. For NPs, the average distance between the PCMH and the LLR address was 21.5 miles. Using the NPI, the distance decreased to 7.4 miles. For physicians, the average distance between the PCMH and the LLR and NPI addresses was 7.2 and 4.3 miles. CONCLUSIONS: Publicly available data to forecast state-wide distributions of the NP workforce for MMC members may not be reliable if done using state licensure board data. Meaningful improvements to correspond with MMC policy changes require collecting and releasing information on place of employment.

Small-area Variation in Hypertension Prevalence among Black and White Medicaid Enrollees.

OBJECTIVES: To examine within-state geographic heterogeneity in hypertension prevalence and evaluate associations between hypertension prevalence and small-area contextual characteristics for Black and White South Carolina Medicaid enrollees in urban vs rural areas. DESIGN: Ecological. SETTING: South Carolina, United States. MAIN OUTCOME MEASURES: Hypertension prevalence. METHODS: Data representing adult South Carolina Medicaid recipients enrolled in fiscal year 2013 (N=409,907) and ZIP Code Tabulation Area (ZCTA)-level contextual measures (racial segregation, rurality, poverty, educational attainment, unemployment and primary care physician adequacy) were linked in a spatially referenced database. Optimized Getis-Ord hotspot mapping was used to visualize geographic clustering of hypertension prevalence. Spatial regression was performed to examine the association between hypertension prevalence and small-area contextual indicators. RESULTS: Significant (alpha=.05) hotspot spatial clustering patterns were similar for Blacks and Whites. Black isolation was significantly associated with hypertension among Blacks and Whites in both urban (Black, b=1.34, P<.01; White, b=.66, P<.01) and rural settings (Black, b=.71, P=.02; White, b=.70, P<.01). Primary care physician adequacy was associated with hypertension among urban Blacks (b=-2.14, P<.01) and Whites (b=-1.74, P<.01). CONCLUSIONS: The significant geographic overlap of hypertension prevalence hotspots for Black and White Medicaid enrollees provides an opportunity for targeted health intervention. Provider adequacy findings suggest the value of ACA network adequacy standards for Medicaid managed care plans in ensuring health care accessibility for persons with hypertension and related chronic conditions.