Community Health Worker Home Visiting, Birth Outcomes, Maternal Care, and Disparities Among Birthing Individuals With Medicaid Insurance

This study assesses the association of participation in a home visiting program with adverse birth outcomes and maternal care vs usual care among birthing individuals with Medicaid.

Key Points

Question

Is participation in a community health worker home visiting program associated with reduced risk of adverse birth outcomes and improved maternal health care?

Findings

In this propensity score matching cohort study (N = 125 252), participation in a home visiting program significantly associated with reduced risk of preterm birth (13.3% vs 15.5%), very preterm birth (1.8% vs 3.0%), very low birth weight (1.0% vs 1.8%), improved adequate prenatal care (74.5% vs 71.4%), and 3 weeks’ postpartum care (43.4% vs 22.4%), with larger birth outcome risk reductions among Black participants.

Meaning

Community health workers, in a combined home visiting model, may be a solution to improve maternal care, infant health, and reduce disparities.

Abstract

Importance

Home visiting is recommended to address maternal and infant health disparities but is underused with mixed impacts on birth outcomes. Community health workers, working with nurses and social workers in a combined model, may be a strategy to reach high-risk individuals, improve care and outcomes, and address inequities.

Objective

To assess the association of participation in a home visiting program provided by community health workers working with nurses and social workers (Strong Beginnings) with adverse birth outcomes and maternal care vs usual care among birthing individuals with Medicaid.

Design, Setting, and Participants

This retrospective, population-based, propensity score matching cohort study used an administrative linked database, including birth records and Medicaid claims, linked to program participation. The Strong Beginnings program exposure took place in 1 county that includes the second largest metropolitan area in Michigan. Study participants included primarily non-Hispanic Black and Hispanic Strong Beginnings participants and all mother-infant dyads with a Medicaid-insured birth in the other Michigan counties (2016 through 2019) as potential matching nonparticipants. The data were analyzed between 2021 and 2023.

Exposure

Participation in Strong Beginnings or usual care.

Main Outcomes and Measures

Preterm birth (less than 37 weeks’ gestation at birth), very preterm birth (less than 32 weeks’ gestation), low birth weight (less than 2500 g at birth), very low birth weight (less than 1500 g), adequate prenatal care, and postnatal care (3 weeks and 60 days).

Results

A total of 125 252 linked Medicaid-eligible mother-infant dyads (mean age [SD], 26.6 [5.6] years; 27.1% non-Hispanic Black) were included in the analytical sample (1086 in Strong Beginnings [mean age (SD), 25.5 (5.8) years]; 124 166 in usual prenatal care [mean age (SD), 26.6 (5.5) years]). Of the participants, 144 of 1086 (13.3%) in the SB group and 14 984 of 124 166 (12.1%) in the usual care group had a preterm birth. Compared with usual prenatal care, participation in the Strong Beginnings program was significantly associated with reduced risk of preterm birth (−2.2%; 95% CI, −4.1 to −0.3), very preterm birth (−1.2%; 95% CI, −2.0 to −0.4), very low birth weight (−0.8%; 95% CI, −1.3 to −0.3), and more prevalent adequate prenatal care (3.1%; 95% CI, 0.6-5.6), postpartum care in the first 3 weeks after birth (21%; 95% CI, 8.5-33.5]), and the first 60 days after birth (23.8%; 95% CI, 9.7-37.9]).

Conclusions and Relevance

Participation in a home visiting program provided by community health workers working with nurses and social workers, compared with usual care, was associated with reduced risk for adverse birth outcomes, improved prenatal and postnatal care, and reductions in disparities, among birthing individuals with Medicaid. The risk reductions in adverse birth outcomes were greater among Black individuals.

Introduction

Significant racial, ethnic, and socioeconomic disparities in maternal and infant morbidity and mortality exist in the US, largest between non-Hispanic Black (hereafter, Black) and non-Hispanic White (hereafter White) individuals.^1,2,3,4,5 To reverse the trend of maternal and infant health inequities, strategies are needed that are acceptable to populations most impacted by disparities with the potential to mitigate the effects of racism, improve care access and quality, address social determinants of health, expand access to resources, and, ultimately, improve health outcomes.^6,7,8 Increasing access to home visiting (HV) has been widely recommended to improve maternal and infant health and service utilization to reduce disparities.⁹ However, reaching, engaging, and maintaining Black and Hispanic Medicaid-insured pregnant individuals in HV, who are at greater health and social risk, remains a challenge¹⁰ and studies of well-established evidence-based HV models have shown mixed effects, if any, for birth outcomes.^11,12 Therefore, it is critical to test innovative, community-informed HV models with culturally relevant approaches that may advance birth equity.^13,14

Strong Beginnings (SB) is an intensive HV program designed to reach and serve mainly Black and Hispanic families during pregnancy and infancy with interventions provided by certified community health workers (CHWs) partnering with licensed nurses and social workers from a state Medicaid-sponsored home visiting program (Michigan Maternal Infant Health Program [MIHP]).¹⁵ Black and Hispanic birthing individuals are referred to SB for care and, by design, dually enrolled in MIHP and SB.

SB CHWs are race, ethnicity, and/or language concordant, frontline public health workers who use their shared community experiences and cultural knowledge to engage and build trusting relationships with pregnant and postpartum individuals.¹⁶ The SB-trained CHWs empower birthing individuals to actively participate in care and navigate resources, provide peer support, and deliver standardized health interventions, complementing nurse and social worker MIHP care.¹⁷ The development and implementation of SB was supported by a federal Healthy Start award, the W.K. Kellogg Foundation, and recently, through a Michigan Department of Health and Human Services (MDHHS) Pay for Success initiative.¹⁸

CHW-provided HV, partnering with licensed professionals, could increase reach to high-risk populations and improve maternal and infant health and service utilization. However, there is limited evidence that the combination of these health care and licensed professionals can improve birth outcomes or address disparities.^{19,20,21,22,23,24} This retrospective population-based cohort study compared risks of adverse birth outcomes and the prenatal and postnatal maternal care of Medicaid-eligible individuals who received prenatal HV by CHWs working with nurses and social workers vs usual care.

Methods

Data Sources

This study used statewide linked data from the MDHHS Health Services data warehouse under an MDHHS data use agreement. The data consisted of routinely collected birth records, Medicaid claims (all Medicaid maternal medical claims from 12 months before conception through birth and 60 days postpartum), Medicaid monthly eligibility, publicly available neighborhood data from the American Community Survey and Child Opportunity Index 2.0,²⁵ and administrative SB program data. The Michigan State University institutional review board determined the study did not involve human subjects. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.

Study Design

We conducted a retrospective cohort study with propensity score matching to compare birth outcomes and maternal care of participants in the SB program with HV services provided by CHWs working with nurses and social workers vs usual care. Participation in SB was recorded in a program database and matched to vital records birth data. The participation indicator was coded yes if a pregnant individual had a first SB CHW encounter, which is a screening encounter, and completed at least 1 additional CHW visit providing SB interventions; otherwise, participation was coded as no. The purpose of the screening encounter is to assess participants on a broad set of risk factors to build a plan of care. Usual care for most Medicaid-eligible individuals in Michigan includes clinical prenatal care and access to other community resources. The purpose of this study was to compare SB with usual care. Most Medicaid birthing individuals in Michigan and nationwide do not receive home visiting. For these reasons, participants in MIHP, the largest prenatal HV program in Michigan serving approximately 20% of all Medicaid-eligible pregnancies, were not included in the usual care group.

All individuals who had a Medicaid-insured singleton birth between January 1, 2016, and December 31, 2019, in Michigan were included in the study, after excluding births with gestational age less than 20 weeks, birth weight below 400 g, congenital abnormalities, or no valid geocoded census tracts. We conducted complete case analysis by excluding births with missing outcome (0.05%) and at least 1 baseline covariate (25%), an approach that meets HomVEE review standards,²⁶ which resulted in 1086 births among SB participants and 124 166 births among individuals in usual care. A sensitivity analysis with imputed baseline covariates for all births was also conducted.

Exposure

SB is a culturally responsive, intensive HV program where certified race, ethnicity, and/or language concordant CHWs, working with nurses or social workers, serve Black and Hispanic Medicaid-insured individuals considered at high medical and social risk. The CHWs partner with their clients to develop and achieve specific health and life-course goals. Using peer support, CHWs coach and empower individuals to access care, communicate with health care professionals, make choices about health care, and overcome individual and environmental barriers to improve their own health. The CHWs provided supportive referrals, helping clients engage and successfully use health and community-based resources to address social determinants of health. CHWs use a specified curriculum and interventions designed to align with the interventions of nurses and social workers who oversee the overall care of the participants. The program provides services, with dosage and frequency tailored to need, from early pregnancy up to 18 months postbirth. The median number of prenatal encounters was 12, of which 7 were provided by CHWs and 5 by nurses or social workers.

Baseline Characteristics

Demographic and socioeconomic characteristics at baseline included maternal prepregnancy Medicaid coverage (12 months before conception), maternal age, marital status, race and ethnicity (Black, Hispanic, White, and other, as reported on the birth record), and maternal education (coded less than high school, high school, or at least some college-level education). Pregnancy history covariates included indicators for prior births, prior preterm births, prior rapid repeat pregnancy within 18 months, and prepregnancy body mass index. Three binary indicators for maternal chronic conditions were also included, coded 1 if related claims 12 months before conception or during pregnancy were present based on diagnostics and procedure codes, and coded 0 otherwise. To minimize the possibility of measurement error, including potential disease onset after program enrollment during pregnancy, we considered the most prevalent conditions defined using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes and International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes,^27,28,29 as appropriate: asthma, diabetes, and hypertension. In addition, 2 binary indicators were included: 1 for the presence of substance, alcohol, or mental health disorders coded 1 if any related claims 12 months before conception or during the first 3 months of pregnancy were present based on diagnostic and procedure codes, and coded 0 otherwise; and another if any claims with a diagnostic of high-risk pregnancy were present during the first 3 months of pregnancy. We considered only the first 3 months of pregnancy to minimize the likelihood of endogeneity with the SB program. Two gestational-related risk indicators—gestational hypertension and gestational diabetes—were defined as binary indicators, coded 1 if related claims during pregnancy were present based on diagnostics and procedure codes and coded 0 otherwise, and were used as regression adjustments after propensity score matching.

Demographic and socioeconomic characteristics and pregnancy history were obtained from the birth certificate. Health characteristics were obtained from birth records or Medicaid claims and these risks are coded as missing if there were no claims or Medicaid eligibility data during the relevant time window. In addition to individual-level characteristics, we included neighborhood characteristics by merging 5-year census tract data from the American Community Survey (2016 to 2019) and Child Opportunity Index 2.0 (2015) to an individual’s geocoded residential address. These covariates include multidimensional neighborhood characteristics, such as residential racialized economic segregation (measured with the index of concentration at the extremes),³⁰ standardized scores in education, health and environment, and social and economic domains (higher scores indicate more opportunities).²⁵

Study Outcomes

Birth outcomes, defined as binary indicators, included: (1) preterm birth (PTB), defined as delivery before 37 completed weeks’ gestation based on the self-reported last menstrual period on the birth certificate; (2) very preterm birth (VPTB) or delivery before 32 completed weeks’ gestation; (3) low birth weight (LBW), defined as less than 2500 g reported on the birth certificate; and (4) very low birth weight (VLBW), defined as less than 1500 g reported on the birth certificate.

Maternal prenatal and postpartum health care utilization included 3 binary outcomes. Adequacy of prenatal care was defined based on the Adequacy of Prenatal Care Utilization (Kotelchuck) Index³¹ reported on the birth certificate (inadequate, intermediate, adequate, or adequate plus). Consistent with previous research and with state and federal reporting,³² we coded a binary outcome 1 if the Kotelchuck Index was adequate or adequate plus and 0 if it was intermediate or inadequate. Another binary indicator was coded 1 if any postpartum visit, based on qualifying Current Procedural Terminology and ICD-9 codes, was identified on maternal Medicaid claims within 3 weeks after delivery and coded 0 otherwise, consistent with American College of Obstetricians and Gynecologists recommendations.^{33,34,35,36,37,38} Lastly, to account for the fact that pregnancy-related Medicaid eligibility ended after approximately 60 days postpartum for many eligible mothers in Michigan at the time of the analyzed births, we also defined a binary indicator coded 1 if any postpartum visit was identified on maternal Medicaid claims in the first 60 days after delivery and coded 0 otherwise.

Statistical Matching and Analysis

Baseline characteristics of SB program participants and individuals receiving usual care before and after matching were compared. The probability of participation in the prenatal SB program (the propensity score) was estimated as a function of all baseline characteristics using logistic regression. Based on the estimated propensity score, we weighed individuals in the comparison group with a kernel algorithm using cross-validation as the optimal bandwidth selection and bootstrapping to produce unbiased standard errors and 95% CIs, with exact matching on Black race and Hispanic ethnicity.

Covariate balance was assessed using the absolute standardized mean difference (ASMD). Adequate matching was indicated by an ASMD less than 0.1.³⁹ Given that gestational diabetes and gestational hypertension occur during pregnancy and may have contributed to the outcomes of interest, we used these 2 indicators as postmatching regression adjustments. Because of the potential for type I error due to multiple comparisons, we controlled for a false discovery rate (FDR) of 5% across all 7 outcomes by applying the Benjamini-Hochberg linear step-up procedure to produce adjusted P values, called FDR-sharpened Q values.^40,41 Results with an FDR-sharpened Q value less than 0.05 were interpreted as statistically significant. FDR-adjusted standard errors were calculated to create 95% CIs following Altman and Bland.⁴² Stata version 15 (StataCorp) and SAS version 9.4 (SAS Institute) were used to perform these analyses.

Subgroup and Sensitivity Analyses

Several supplemental analyses were performed to assess the robustness of the study findings. First, alternative matching algorithms, including 1:1 match without replacement, 1:3 nearest neighbors with 0.1 caliper, kernel matching with pair-match bandwidth selection, and inverse probability weighting were, used (eTable 1 in Supplement 1). Second, a sensitivity analysis was conducted in a larger analytical sample with imputed missing baseline covariates using multiple imputation (eTable 2 in Supplement 1). Because Black families are facing the largest disparities in maternal and infant morbidity and mortality, we performed matching analyses separately for births among Black individuals (eTables 3 and 4 in Supplement 1) and individuals of other races.

Results

Study Population and Baseline Characteristics

Of the 125 252 linked Medicaid-eligible mother-infant dyads in the analytical sample, 1086 mothers participated in prenatal SB home visiting (with 108 more being screened into SB and not receiving additional SB services) and 124 166 mothers received usual prenatal care. Compared with usual care (mean [SD] age, 26.6 [5.5] years), the SB participants (mean [SD] age, 25.5 [5.8] years) were more likely to be Black or Hispanic (609 [56.1%] vs 33 276 [26.8%]), have had education less than high school, prior preterm birth (62 [5.7%] vs 4967 [4%]), have mental or substance-related disorders, and have had gestational hypertension (Table 1). The baseline characteristics after kernel matching are well balanced, indicated by the ASMD less than 0.05 in all comparisons. A similar balance of baseline covariates is also achieved for the subsample of Black individuals (eTable 3 in Supplement 1).

Table 1. Baseline Characteristics of Participants in Strong Beginnings (SB) and Usual Care Before and After Propensity Score Matching.

Characteristic	Before matching, %			After matching, %a
	SB (n = 1086)	Usual care (n = 124 166)	ASMD	SB (n = 1081)	Usual care (n = 123 922)	ASMD
Prior Medicaid coverageb	80.9	81.8	−0.02	80.9	80.8	<0.01
Maternal age, mean (SD), y	25.5 (5.8)	26.6 (5.5)	−0.20	25.6 (5.8)	25.6 (5.9)	−0.01
Married	19.4	32.3	−0.30	19.5	19.8	−0.01
Race and ethnicityc
Hispanic	31.2	9.1	0.57	31.4	31.4	<0.01
Non-Hispanic Black	56.1	26.8	0.62	56.1	56.1	<0.01
Non-Hispanic White	8.9	59.0	−1.25	9.0	9.0	<0.01
Otherd	3.8	5.0	−0.06	3.6	3.6	<0.01
Maternal education
<High school	38.1	17.5	0.47	37.9	37.1	0.02
High school	39.0	39.1	0.00	39.0	39.6	−0.01
Some college or higher	22.9	43.3	−0.44	23.0	23.3	−0.01
Have prior births	71.9	74.6	−0.06	72.0	72.3	−0.01
Have previous preterm	5.7	4.0	0.08	5.7	5.7	<0.01
Have previous rapid repeat pregnancye	26.8	27.5	−0.02	26.7	27.0	−0.01
Maternal BMI,f mean (SD)	29.8 (8.3)	28.2 (7.5)	0.21	29.8 (8.3)	29.8 (8.6)	<0.01
Medical risks
Diabetesg	6.2	6.8	−0.03	6.2	6.3	−0.01
Hypertensiong	5.8	4.4	0.06	5.8	5.8	<0.01
Asthmah	5.6	4.6	0.05	5.6	5.6	<0.01
Substance/alcohol/mental disordersi	36.2	25.8	0.23	36.0	35.7	0.01
High-risk pregnancyj	58.4	30.3	0.59	58.2	57.1	0.02
Neighborhood characteristics, tract, mean (SD)
Racialized economic segregationk	−0.06 (0.1)	−0.02 (0.2)	−0.23	−0.07 (0.1)	−0.06 (0.2)	−0.02
COI education score	27.9 (19.8)	36.8 (25.3)	−0.39	27.8 (19.6)	28.4 (24.2)	−0.03
COI health score	21.1 (16.4)	37.6 (27.5)	−0.73	21.1 (16.4)	21.8 (19.0)	−0.03
COI socioeconomic score	29.6 (21.6)	35.1 (24.9)	−0.23	29.4 (21.4)	29.8 (23.8)	−0.02

Abbreviations: ASMD, absolute standardized mean difference; BMI, body mass index; COI, Child Opportunity Index 2.0; ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification; ICD-10-CM, International Classification of Diseases, Tenth Revision, Clinical Modification .

^a Participants in the Strong Beginnings program were matched to those in usual care by their propensity score using kernel matching with exact match on Black and Hispanic participants and cross-validated bandwidth selection. The balance between the 2 groups was assessed with ASMD less than 0.05.

^b Medicaid coverage at any point up to 12 months before conception (excluding the month of conception).

^c As reported on the birth record.

^d Includes American Indian/Alaskan Native, Asian, Native Hawaiian/other Pacific Islander, multiple races.

^e Prior rapid repeat pregnancy if prior pregnancy is less than or equal to 18 months from prior birth to conception.

^f Calculated as weight in kilograms divided by height in meters squared.

^g Chronic diabetes (excludes gestational diabetes) using ICD-9-CM in 249.x, 250.x, 790.2x, V4585, V5491, V6546 and ICD-10-CM in E08.x, E09.x, E10.x, E11.x, E13.x or chronic hypertension (excludes gestational hypertension) using ICD-9-CM in 401.X and ICD-10-CM in I10.X, I67.4, N26.2 from birth records or claims up to 12 months prior to conception through pregnancy.

^h Chronic asthma from claims up to 12 months prior to conception through pregnancy using ICD-9-CM in 493.x and ICD-10-CM in J45.2x and J45.9x.

ⁱ Mental, alcohol-related, and substance-related disorders are from claims up to 12 months prior to conception through the first 3 months of pregnancy, using codes as a combination from the Agency for Healthcare Research and Quality’s clinical classifications software for ICD-9-CM and clinical classification software (beta) for ICD-10-CM as defined by category 5 in the multilevel category and ICD-10-CM codes from the following default categories in the CCSR: EXT021, FAC021, INJ075, MBD001, MBD002, MBD003, MBD005, MBD007, MBD008, MBD009, MBD012, MBD013, MBD014, MBD017, MBD018, MBD019, MBD020, MBD021, MBD022, MBD023, MBD025, MBD027, MBD034, NVS011, SYM008, and SYM009.

^j Claims with ICD-9-CM in V23.x or ICD-10-CM in O09.x in the first 3 months of pregnancy.

^k Measured with index of concentration at the extremes using both race and ethnicity and income to jointly measure racial and economic segregation at census tract following Krieger et al.³⁰

Association of Strong Beginnings Participation With Adverse Birth Outcomes and Maternal Health

Adverse Birth Outcomes

Individuals who participated in prenatal SB had reduced risk of PTB (−2.2%; 95% CI, −4.1 to −0.3), VPTB (−1.2%; 95% CI, −2.0 to −0.4), and VLBW (−0.8%; 95% CI, −1.3 to −0.3) compared with matched individuals who received usual prenatal care (Table 2). Black individuals who participated in the SB prenatal program had reduced risk of PTB (−3.8%; 95% CI, −6.5 to −1.1), VPTB (−2.2%; 95% CI, −3.5 to −0.9), and VLBW (−1.5%; 95% CI, −2.5 to −0.5) compared with matched individuals who received usual prenatal care (Table 3). Estimates from kernel within a pair-matching bandwidth, from inverse probability weighting, and in the larger sample with regression-imputed baseline covariates were similar (eTables 1 and 2 in Supplement 1). The risk reductions in PTB, VPTB, and VLBW were statistically significant among Black SB participants with no evidence of risk reductions among SB participants of other races and significantly larger for Black SB participants for VPTB and VLBW (eTable 4 in Supplement 1).

Table 2. Birth Outcomes and Maternal Care Before and After Propensity Score Matching.

Outcomes	Prevalence of events, No./total No.a		Between-group difference, percentage point (95% CI)
	SB	Usual care	Before matchingb	After matchingc
Preterm (<37 wk)	144/1086 (13.3)	14 984/124 166 (12.1)	1.2 (−0.9 to 3.3)	−2.2 (−4.1 to −0.3)
Very preterm (<32 wk)	19/1086 (1.8)	2542/124 166 (2.1)	−0.3 (−1.0 to 0.4)	−1.2 (−2.0 to −0.4)
Low birth weight (<2500 g)	140/1086 (12.9)	10 247/124 151 (8.3)	4.6 (2.5-6.8)	1.5 (−0.03 to 3.0)
Very low birth weight (<1500 g)	12/1086 (1.0)	1451/124 151 (1.2)	−0.1 (−0.6 to 0.5)	−0.8 (−1.3 to −0.3)
Adequate care and plus	786/1055 (74.5)	89 789/119 332(75.2)	−0.7 (−3.3 to 1.8)	3.1 (0.6-5.6)
Postpartum care (in 3 wk)	471/1086 (43.4)	22 887/124 166 (18.4)	24.9 (21.5-28.3)	21.0 (8.5-33.5)
Postpartum care (in 60 d)	938/1086 (86.4)	75 250/124 166 (60.6)	25.8 (23.7-27.8)	23.8 (9.7-37.9)

Abbreviation: SB, Strong Beginnings

^a Observed prevalence before matching.

^b The difference between the control mean and the intervention mean may not equal the difference column before matching because of rounding.

^c The difference is based on kernel matching on baseline covariates with exact match on Black and Hispanic participants and regression adjustments for gestational diabetes and gestational hypertension using cross-validated bandwidth selection and bootstrapped standard errors. False discovery rate–sharpened Q values and standard errors are calculated to report 95% CIs.

Table 3. Birth Outcomes and Maternal Care Before and After Propensity Score Matching Among Non-Hispanic Black Women.

Outcomes	Prevalence of events, No./total No.a		Between-group difference, percentage point (95% CI)
	SB	Usual care	Before matchingb	After matchingc
Preterm (<37 wk)	84/609 (13.8)	5591/33 304 (16.8)	−3.0 (−5.7 to −0.3)	−3.8 (−6.5 to −1.1)
Very preterm (<32 wk)	9/609 (1.5)	1113/33 304 (3.3)	−1.9 (−2.7 to −1.0)	−2.2 (−3.5 to −0.9)
Low birth weight (<2500 g)	100/609 (16.4)	4140/33 300 (12.4)	4.0 (0.9-7.0)	2.8 (0.1-5.5)
Very low birth weight (<1500 g)	6/609 (1.0)	701/33 300 (2.1)	−1.1 (−2.1 to −0.2)	−1.5 (−2.5 to −0.5)
Adequate care and plus	439/593 (74.0)	22 528/31 458 (71.6)	2.4 (−1.4 to 6.3)	3.1 (0.1-6.1)
Postpartum care (in 3 wk)	306/609 (50.3)	6496/33 304 (19.5)	30.7 (26.5-35.0)	25.2 (10.2-40.2)
Postpartum care (in 60 d)	528/609 (86.7)	19 077/33 304 (57.3)	29.4 (26.4-32.4)	23.9 (9.7-38.1)

Abbreviation: SB, Strong Beginnings.

^a Observed prevalence before matching.

^b The difference between the control mean and the intervention mean may not equal the difference column before matching because of rounding.

^c The difference is based on kernel matching on baseline covariates with regression adjustments for gestational diabetes and gestational hypertension using cross-validated bandwidth selection and bootstrapped standard errors. False discovery rate–sharpened Q values and standard errors are calculated to report 95% CIs.

Adequate Prenatal Care

Individuals who participated in prenatal SB were more likely to receive appropriate prenatal care (3.1%; 95% CI, 0.6-5.6) when compared with matched individuals who received usual prenatal care (Table 2). The result was similar among Black individuals (3.1%; 95% CI, 0.1-6.1). Estimates from kernel-within-a-pair matching bandwidth from inverse probability weighting and in the larger sample with regression-imputed baseline covariates were similar (eTables 1 and 2 in Supplement 1).

Timely Postnatal Care

Individuals who participated in prenatal SB were more likely to receive maternal care in the first 3 weeks postpartum, as recommended by American College of Obstetricians and Gynecologists (21.0%; 95% CI, 8.5-33.5) and in the first 60 days postpartum (23.8%; 95% CI, 9.7-37.9) when compared with matched individuals who received usual prenatal care (Table 2). The result was similar among Black individuals (3 weeks: 25.2%; 95% CI, 10.2-40.2 and 60 days: 23.9%; 95% CI, 9.7-38.1) (Table 3). Estimates from kernel-within-a-pair matching bandwidth, from inverse probability weighting, and in the larger sample with regression-imputed baseline covariates were similar (eTables 1 and 2 in Supplement 1).

Discussion

In this retrospective population-based cohort study, participation in an HV model teaming race, ethnicity, and/or language concordant CHWs with nurses and social workers was significantly associated with reduced risks of adverse birth outcomes and improved prenatal and postpartum maternal care when compared with usual prenatal care. The risk reductions in birth outcomes associated with program participation were significantly larger among Black individuals compared with others. The results remained consistent and robust across sensitivity analyses. The large improvements in timely postpartum care may be attributed to CHWs partnering with mothers to create a reproductive life plan during pregnancy and use an interactive education tool, Empower Yourself to be Healthy Mom, to engage in planning for postnatal care.

This study adds to the evidence supporting the role of CHWs facilitating access to care for underserved populations facing disparities and contributing to improved care, health, and equity.^19,20,21,22 Three HV studies of CHWs plus licensed professionals in combined models have shown improvements in selected birth outcomes for all^19,23 or subgroup participants,²⁴ although these studies have somewhat limited comparability with our findings due to participant characteristics, exclusion criteria, and methodology shortcomings, including limited baseline covariates used for addressing selection bias.

To address health inequities, evidence-based solutions are needed that can be rapidly deployed and build on existing HV infrastructures and Medicaid-sponsored programs, common in 60% of states, and that are already embedded in communities.⁴³ This work expands the knowledge base focused on the Medicaid-eligible population and on an HV program that, by design, reached participants of racial and ethnic minorities with higher health and social risks. The value of race, ethnicity, and/or language concordant CHWs working with nurses and social workers, key to this program and to the positive findings in this study, has long-term implications not only for birth outcomes and maternal care but also for opportunities to address disparities in maternal morbidity and mortality.⁴⁴ Better birth outcomes and improved prenatal and postpartum care have been linked to reduced risk of maternal morbidity⁴⁵ and mortality.^46,47 Last, the CHW combined HV model expands the utilization of certified CHWs, an actionable strategy to strengthen and diversify the health care workforce.^48,49

Limitations

Several limitations are noted. First, the propensity score matching analysis balanced on characteristics was expected to influence participation in SB, yet, as with other observational studies, our matching was limited to observable characteristics documented in the available data. Unmeasured confounders, such as environmental or geographic characteristics that may have differed between the SB county and controls in other counties, may have affected our findings. Second, individuals in the usual care group may have participated in HV programs, other than MIHP available in Michigan at the time of the study, which may have led to underestimating the associations with birth outcomes and maternal care. Third, SB was implemented in a single county, including the second largest city in Michigan, that may limit its generalizability. However, the findings reported here led to increased confidence in the model and to planning future studies to test the implementation and scalability of the CHWs working with nurses and social workers HV model across different locations and health systems.

Conclusions

In this population-based cohort study, participation in an HV intervention with services offered by CHWs working with nurses and social workers compared with usual care, was significantly associated with reduced risk of adverse birth outcomes, improved prenatal and postnatal maternal health care, and reductions in racial disparities in adverse birth outcomes among Medicaid-eligible birthing individuals. The risk reductions in adverse birth outcomes were greater among Black individuals, who are facing the largest disparities in infant and maternal morbidity and mortality compared with White individuals. The results are relevant as there are increased calls to expand culturally relevant, community-informed, models of maternal care to improve maternal and infant health equity.

Supplement 1.

eTable 1. Alternative Matching Algorithms

eTable 2. Birth Outcomes and Maternal Care Before and After Propensity Score Matching With Imputed Baseline Covariates

eTable 3. Baseline Characteristics of Participants in SB and Usual Care Before and After Propensity Score Matching, Non-Hispanic Black

eTable 4. Birth Outcomes and Maternal Care After Propensity Score Matching between Non-Hispanic Black and Other Races

Supplement 2.

Data sharing statement