Effect of an Intensive Nurse Home Visiting Program on Adverse Birth Outcomes in a Medicaid-Eligible Population: A Randomized Clinical Trial

Key Points

Question  What is the effect of an intensive nurse home visiting program on a composite outcome of preterm birth, low birth weight, small for gestational age, or perinatal death?

Findings  In this randomized clinical trial that enrolled 5670 Medicaid-eligible nulliparous pregnant individuals, assignment to the Nurse Family Partnership program of intensive nurse home visits compared with usual care resulted in a composite adverse birth outcome incidence of 26.9% and 26.1%, respectively. This difference was not statistically significant.

Meaning  Compared with usual care, the program did not reduce the risk of a composite of adverse birth outcomes; however, evaluation of the effectiveness of this program is incomplete pending assessment of early childhood and birth spacing outcomes.

Importance  Improving birth outcomes for low-income mothers is a public health priority. Intensive nurse home visiting has been proposed as an intervention to improve these outcomes.

Objective  To determine the effect of an intensive nurse home visiting program on a composite outcome of preterm birth, low birth weight, small for gestational age, or perinatal mortality.

Design, Setting, and Participants  This was a randomized clinical trial that included 5670 Medicaid-eligible, nulliparous pregnant individuals at less than 28 weeks’ gestation, enrolled between April 1, 2016, and March 17, 2020, with follow-up through February 2021.

Interventions  Participants were randomized 2:1 to Nurse Family Partnership program (n = 3806) or control (n = 1864). The program is an established model of nurse home visiting; regular visits begin prenatally and continue through 2 postnatal years. Nurses provide education, assessments, and goal-setting related to prenatal health, child health and development, and maternal life course. The control group received usual care services and a list of community resources. Neither staff nor participants were blinded to intervention group.

Main Outcomes and Measures  There were 3 primary outcomes. This article reports on a composite of adverse birth outcomes: preterm birth, low birth weight, small for gestational age, or perinatal mortality based on vital records, Medicaid claims, and hospital discharge records through February 2021. The other primary outcomes of interbirth intervals of less than 21 months and major injury or concern for abuse or neglect in the child’s first 24 months have not yet completed measurement. There were 54 secondary outcomes; those related to maternal and newborn health that have completed measurement included all elements of the composite plus birth weight, gestational length, large for gestational age, extremely preterm, very low birth weight, overnight neonatal intensive care unit admission, severe maternal morbidity, and cesarean delivery.

Results  Among 5670 participants enrolled, 4966 (3319 intervention; 1647 control) were analyzed for the primary maternal and neonatal health outcome (median age, 21 years [1.2% non-Hispanic Asian, Indigenous, or Native Hawaiian and Pacific Islander; 5.7% Hispanic; 55.2% non-Hispanic Black; 34.8% non-Hispanic White; and 3.0% more than 1 race reported [non-Hispanic]). The incidence of the composite adverse birth outcome was 26.9% in the intervention group and 26.1% in the control group (adjusted between-group difference, 0.5% [95% CI, −2.1% to 3.1%]). Outcomes for the intervention group were not significantly better for any of the maternal and newborn health primary or secondary outcomes in the overall sample or in either of the prespecified subgroups.

Conclusions and Relevance  In this South Carolina–based trial of Medicaid-eligible pregnant individuals, assignment to participate in an intensive nurse home visiting program did not significantly reduce the incidence of a composite of adverse birth outcomes. Evaluation of the overall effectiveness of this program is incomplete, pending assessment of early childhood and birth spacing outcomes.

Trial Registration  ClinicalTrials.gov Identifier: NCT03360539

Adverse birth outcomes can lead to mortality, morbidity, and childhood developmental challenges.^1-3 In the US, there are substantial racial and socioeconomic inequities in these outcomes.^4,5 More evidence is needed on the effects of interventions targeting low-income pregnant people, especially when the interventions are delivered at scale.^6,7

Expanding intensive home visiting programs has been recommended for improving maternal and newborn outcomes; these programs receive substantial federal funding through the Maternal, Infant and Early Childhood Home Visiting (MIECHV) program.⁸ The Nurse-Family Partnership program is an established model of nurse home visiting services designed to reach nulliparous low-income families during pregnancy and early childhood. The program was evaluated in modestly sized randomized trials in the 1970s and 1990s in New York, Tennessee, and Colorado. These early trials suggested some benefits for birth outcomes, including lower rates of pregnancy-induced hypertension⁹ and, for some subgroups, higher birth weights and fewer preterm deliveries.¹⁰ Subsequent observational studies suggested better birth outcomes among program participants compared with matched counterparts.¹¹

Motivated in part by this evidence and by South Carolina’s preterm birth rate, which in 2016 was the sixth highest in the US,¹² the state began to offer program services to Medicaid-eligible nulliparous women through a Medicaid 1915(b) waiver.¹³ Philanthropic funding supported the scale-up of the project through a “Pay for Success” model that embedded a randomized clinical trial of program effectiveness. The objective of this study was to determine the effect of an intensive nurse home visiting program on a composite outcome of any of preterm birth, low birth weight, small for gestational age, or perinatal mortality.

The study was approved by the Harvard T.H. Chan School of Public Health institutional review board (IRB) (IRB15-2939). Permissions were also obtained from cooperating institutions (eMethods in Supplement 2, section 1). The study’s consent form informed participants about randomization, participation in the program, and that researchers would track their data and their children’s data across a variety of administrative data records for up to 30 years. Electronic signatures were obtained from consenting study participants.

We conducted an individually randomized clinical trial. The original IRB protocol (Supplement 1, section 1) and pre-analysis plan (Supplement 1, section 2) were supplemented with a detailed published study protocol.^13,14 The timing of trial registration and selection of outcomes are detailed in the eMethods (Supplement 2, section 2). Study participants were randomly assigned either to an intervention group that was offered access to the program or to a control group (Figure 1). Control group members received usual care in South Carolina, which included access to all other community and medical services. All study participants were provided with a list of available community resources (eMethods in Supplement 2, section 3).

Study eligibility mirrored the program’s eligibility criteria: pregnancy with less than 28 weeks’ gestation, nulliparous, income-eligible for Medicaid during pregnancy, and residence in a program-served county (32 of 46 South Carolina counties). Individuals 14 years and younger, or who were incarcerated or in a lockdown facility, were excluded from the trial. Pregnant individuals either self-referred to the program or were referred through channels including clinicians, schools, or Medicaid to 1 of 9 program-implementing sites embedded in government agencies and hospital systems throughout South Carolina (eMethods in Supplement 2, section 3).¹³

Program staff assessed potential participants’ eligibility. Eligible participants provided written informed consent and completed a baseline survey that included information on demographic characteristics, socioeconomics, health behaviors, and physical and mental health. Investigators included questions about participants’ race and ethnicity to assess the program’s potential influence on racial disparities in maternal and child outcomes; participants were asked to self-identify their race and ethnicity from a list of prespecified options (Hispanic or Latina or not Hispanic or Latina for ethnicity and American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, or White). Respondents were able to report more than 1 option.

Participants were randomly assigned to the intervention or control group in a 2:1 ratio. Trained program staff conducted on-the-spot randomization using computer-generated random numbers on encrypted tablets and computer-assisted software.

The intervention was a prenatal and early childhood home visiting program. Nurses conducted home visits with participants during pregnancy and through the first 2 years of the child’s life. Nurses tailored activities to clients’ strengths, risks, and preferences using motivational interviews, educational tools, health assessments, and goal-setting related to prenatal health, child health and development, and maternal life course. They encouraged health care utilization when needed^10,15 and made referrals to health and social services. According to the program model, visits should last 60 to 90 minutes and occur every week during the first 4 weeks after enrollment and then every other week until delivery. Agencies were compensated on a per-visit basis under the Medicaid waiver (eMethods in Supplement 2, section 3), which covered up to 15 visits during pregnancy. Nurses were trained to flexibly support clients if additional or fewer visits were necessary. Services were provided in English and Spanish where bilingual nurse home visitors were available, and translation services were available for participants speaking other languages.

We matched study participants to vital records, Medicaid claims data, and hospital discharge records via a probabilistic match based on name, race, Social Security number, birth date, and Medicaid ID. To assess receipt of program services, we matched participants to intervention programmatic data. We matched to Maternal, Infant, and Early Childhood Home Visiting data to track participants’ participation in other federally funded home visiting programs, including Healthy Families America, Parents as Teachers, and Healthy Steps.

The analyses in this article assessed 1 of the 3 primary outcomes for the trial. The primary maternal and newborn health outcome was a composite indicating whether a participant experienced any of the following adverse birth events: preterm birth, low birth weight, small for gestational age, or perinatal death. The analysis of the 2 remaining primary outcomes, interbirth intervals of less than 21 months and major injury or concern for abuse or neglect in the child’s first 24 months, will be reported when data become available.

Thirteen of the 54 prespecified secondary outcomes were included with this analysis to more broadly understand maternal and newborn health outcomes around childbirth. Secondary newborn outcomes included each individual outcome in the composite outcome, continuous measures of birth weight and gestational age, large for gestational age, extremely preterm birth, very low birth weight, overnight neonatal intensive care unit (NICU) admission, and neonatal morbidity. Secondary maternal outcomes included cesarean delivery and severe maternal morbidity at delivery. Because maternal mortality is designed to be measured over a longer postpartum period and is reported with a significant delay, it is not included in these analyses.

All outcomes were obtained from vital records except overnight NICU admission and severe maternal morbidity, which were obtained from Medicaid and hospital discharge records. For participants with multiple births, binary outcomes were defined based on experiencing any adverse outcome for any infant and continuous outcomes were averaged across infants. See the eMethods in Supplement 2, section 4, for detailed information on outcomes. Additional secondary outcomes not reported in this article include health care utilization, maternal mental health, postpartum maternal health, child health, utilization of family planning, and utilization of social services.¹³

The analytical sample for the primary outcome was participants with an “index” live birth or fetal death in matched vital records within 120 days of the expected delivery date reported on the baseline survey (eMethods in Supplement 2, section 5). Outcomes for participants who did not meet this criterion or who withdrew from the study were excluded. Except for fetal death, infant outcomes were observed only for infants with a live index birth. For overnight NICU admission, we observed outcomes for participants who matched to an index birth and had a record of delivery in Medicaid and hospital discharge records. For severe maternal morbidity, we observed outcomes for participants who had an indication of delivery in either Medicaid or hospital discharge records (eMethods in Supplement 2, section 5).

We estimated the minimum detectable effect size for the sample size that was available within constraints of the Pay for Success project (eMethods in Supplement 2, section 2), which included planned enrollment of 6000 participants (control, 2000; intervention, 4000). For the primary composite birth outcome, the trial was powered to detect a 3.5–percentage point (14% relative) reduction in incidence in the full sample and a 5.2–percentage point (21% relative) reduction in incidence in a predefined subgroup of vulnerable participants discussed below.¹³ These minimum detectable effects within the prespecified subgroup are less than the 24% relative reduction in preterm birth observed through midwife continuity of care, which is one of the only effective interventions to reduce adverse birth outcomes based on evidence from systematic reviews.^2,6,7,16

We used ordinary least squares linear regression models to compare participants by randomization group using 2-sided hypothesis tests. For binary outcomes, regressions represent linear probability models. We estimated unadjusted models and models adjusted for prespecified covariates (eMethods in Supplement 2, section 6) to assess the robustness of the statistical model. Covariates hypothesized to have a strong association with study outcomes or that captured the heterogeneity in the sample were chosen to improve the precision of estimates of treatment effects. We used the dummy-variable adjustment method to account for missing baseline covariates (eTable 1 in Supplement 2) in our analysis as prespecified.¹⁷ While our original IRB protocol (Supplement 1, section 1) indicated we would use an instrumental variable approach to estimate local average treatment effects, we specified in our full study protocol¹³ prior to analysis that we would analyze participants according to their randomization group for ease of interpretation and comparison with prior studies. We also estimated local average treatment effects. Planned interim analysis of outcomes for the Pay for Success program occurred in February 2021.¹⁸

We conducted 2 prespecified subgroup analyses.¹³ First, we analyzed program outcomes among a prespecified group of participants who may be more vulnerable to challenges during pregnancy and early childhood based on prior studies and the targeting criteria of other home visiting programs. This included participants who were younger than 19 years old, had not finished high school, or had challenges with mental health (Patient Health Questionnaire 2 (PHQ-2)¹⁹ score 3 or higher at baseline or reported receiving mental health treatment in the year before enrollment). Second, because of the persistent racial disparities in birth outcomes, we analyzed program outcomes among participants who self-identified as non-Hispanic Black. We used ordinary least squares linear regression models to compare subgroup participants by randomization group using 2-sided hypothesis tests. We provide a figure of adjusted estimates and their confidence intervals for illustrative secondary outcomes. Because of the potential for type I error due to multiple comparisons, findings for analyses of secondary end points should be interpreted as exploratory. We also used the Benjamini-Hochberg linear step-up procedure to control for a false discovery rate (FDR) of 5% across all prespecified outcomes for each subgroup.²⁰ This procedure produces adjusted P values called FDR-sharpened Q values: findings with an FDR-sharpened Q value less than .05 were interpreted as statistically significant. Analyses were performed using Stata version 14.2 (StataCorp) and R version 1.4.1717 (R Foundation for Statistical Computing).

Because of safety concerns due to the COVID-19 pandemic, in consultation with the IRB, the investigators stopped enrollment on March 17, 2020. While the investigators considered converting to telephone enrollment, 94.5% of the target sample had already been reached and this deviation was determined to undermine the integrity of the evaluation, as it would have required nurses to begin their therapeutic relationship over the telephone. The modification had a small influence on statistical power (eMethods in Supplement 2, section 7). Among randomized participants, 87% had passed their expected due dates before this modification. After March 23, 2020, the program conducted 93% of remaining prenatal home visits via telehealth.

Study participants were enrolled between April 1, 2016, and March 17, 2020; outcomes were assessed through February 2021. The program screened 18 994 potential participants for eligibility; 12 189 (64%) were eligible, and 5670 (47%) consented to participate and were randomized (Figure 1). Subsequently, 15 people opted to withdraw from the study. Of the 3806 participants assigned to receive the program, 3319 matched to an index birth or fetal death record. Of the 1864 participants assigned to the control group, 1647 matched to an index birth or fetal death record. The remaining participants could not be matched; rates of missing outcomes were not statistically different between treatment and control groups (eTable 2 in Supplement 2).

Participants were enrolled at a median of 14 weeks’ gestation; 84.7% had already received at least 1 prenatal care visit (Table 1). At enrollment, 18% of participants were younger than 19 years; 54.8% were 19 to 24 years old (median age, 21 years). Among participants, 55.2% reported their race and ethnicity as non-Hispanic Black and 22.4% had not completed high school. At enrollment, 16.9% reported housing insecurity, 19.1% had depressive symptoms, and 66.0% reported high stress. Among participants, 34.5% had a body mass index greater than 30 (calculated as weight in kilograms divided by height in meters squared), 50.9% reported drinking alcohol, and 25.8% reported smoking in the 3 months before pregnancy. Participant characteristics and geography (eTable 3 in Supplement 2) were balanced between treatment and control groups in the analytical sample and the randomized sample (eTable 4 in Supplement 2).

Among the intervention and control groups, 0.8% and 1.3% participated in federally funded nonintervention home visiting programs, respectively (Table 2). In the intervention group, nearly all participants received at least 1 program home visit (98.2%), with participants receiving a median of 9 visits during pregnancy. Among those with an index birth, 78.3% continued to receive visits up to the birth of their child. The median in-person home visit lasted 65 minutes, with nurses spending a median of 38.5% of each prenatal home visit on overall personal health and 21.5% on maternal role. Among intervention group participants, 23.4% received a referral for any health care service and 14.1% received a referral for prenatal care. Implementation metrics are similar for the 2 subgroups (eTable 5 in Supplement 2).

The incidence of adverse birth outcomes was 26.9% in the intervention group and 26.1% in the control group (Table 3). The adjusted difference was 0.5 percentage points higher in the intervention group than the control group (95% CI, −2.1 to 3.1). Estimates from adjusted and unadjusted models, and from estimates of local average treatment effects from instrumental variables estimation (eTable 6 in Supplement 2), were similar.

There were no statistically significant differences between the control and intervention groups for any of the 13 maternal and neonatal health-related secondary outcomes assessed in this analysis.

The incidence of the composite adverse birth outcome within the control group among the vulnerable (n = 2304) and non-Hispanic Black (n = 2565) subgroups was 26.9% and 31.6%, respectively (Figure 2). eTable 7 in Supplement 2 displays subgroup analysis for all of the 13 maternal and neonatal health related secondary outcomes assessed in this analysis. Outcomes were not significantly improved for the intervention group vs the control group for the primary composite adverse birth outcome or any of the 13 maternal and neonatal health-related secondary outcomes assessed in this analysis in either subgroup (eTable 8 in Supplement 2). For non-Hispanic Black participants, the adjusted share of infants born large for gestational age was 1.9 percentage points higher in the intervention group than the control group (95% CI, 0.4-3.4). However, this difference was not statistically significant after accounting for multiple hypothesis testing (eTable 9 in Supplement 2).

In this randomized trial, assignment to participate in an intensive nurse home visiting program did not significantly reduce the incidence of adverse birth outcomes (preterm birth, low birth weight, small for gestational age, or perinatal mortality) or improve secondary maternal and newborn outcomes among Medicaid-eligible participants. These results are consistent with evidence from other recent evaluations of nurse home visiting during pregnancy.^23-25 The weight of the evidence increasingly suggests that intensive nurse home visiting is not an effective intervention for reducing adverse birth outcomes.

By design, the content covered by intensive nurse home visiting is driven by patients’ needs, interests, and concerns. The program could influence outcomes by changing patients’ knowledge and behaviors or through referrals. The results of this trial reinforce the challenge of relying on these mechanisms to reduce adverse birth outcomes.^7,16 The causes of adverse birth outcomes, particularly preterm birth,²⁶ are not well understood. Preconception health is increasingly recognized to play a key role in determining birth outcomes²⁷; interventions delivered during pregnancy may be too late to address many of the factors that contribute to adverse birth outcomes. More evidence is needed to understand which interventions are effective in improving preconception health.²⁸

While non-Hispanic Black participants experienced substantially higher rates of adverse birth outcomes in this trial population, assignment to program services did not improve their outcomes. Evidence has shown that interrelated structural factors, such as poverty, racism, environmental exposures, and neighborhood characteristics, influence both preconception and prenatal health and ultimately affect the risk of adverse birth outcomes.^29-32 Home visiting programs may not be adequate to address these long-standing structural challenges.

In the several decades since the publication of the original trials of the program, there have been changes in trends related to pregnancy risk such as lower rates of smoking during pregnancy,³³ substantial decreases in births to adolescents,³⁴ and increases in Medicaid coverage during pregnancy.³⁵ However, these changes in trends cannot fully explain the results. There were no significant effects of the program on birth outcomes within a subgroup of participants with characteristics identified by previous program trials to experience larger program effects. In addition, the incidence of adverse birth outcomes was not systematically lower in this study’s control group compared with earlier program trials. For example, the incidence of preterm birth observed in the control group is similar to that reported in previous trials of the program (11.6% in this trial, vs 7.3% in the New York trial and 13% in the Tennessee trial). The high incidence of adverse birth outcomes in this trial mirrors state trends; South Carolina’s rate of preterm birth is consistently among the highest in the nation,³⁶ despite recent clinical quality improvement efforts.³⁷

The potential for intensive nurse home visiting to facilitate earlier and more complete access to clinical care in pregnancy may be limited by who chooses to enroll. In this study setting, 53% of potentially eligible participants elected not to participate; somewhat lower but still significant rates of nonparticipation in intensive nurse home visiting have also been highlighted in other research.³⁸ Home-based services from medical professionals beginning early in pregnancy may be more attractive to pregnant people who are already well connected to other clinical services. In the current trial, 84.7% of participants initiated prenatal care prior to enrollment. More research is needed to understand what programs can be acceptable to and effective for pregnant individuals who are not already well integrated into clinical care or who may not be comfortable inviting a clinician into their home.

This study has several limitations. First, the birth outcome assessments presented in this evaluation represent an incomplete picture of the potential effects of the intervention program. The effect of the program on other important domains with longer assessment periods, including additional primary outcomes related to child development and the maternal life course, will be analyzed when these data are available. These important outcomes are the primary focus of many established home visiting programs, particularly those that enroll at the time of birth.³⁹ Second, because the trial overlapped with the COVID-19 pandemic, some visits were delivered virtually. However, 87% of participants’ expected due dates preceded the pandemic. The number of nurse home visits was similar to patterns reported in other settings that predate the COVID-19 pandemic,⁴⁰ and the median length of a home visit in this sample meets the minimum program requirement. Third, control group participants were able to access other similar home visiting services available in South Carolina during the study period, which may underestimate the possible effects of the intervention program. However, this study found that less than 2% of the control group participated in other federally funded home visiting programs during pregnancy. Nonetheless, it is possible that study participants benefited from other local programs. Fourth, the trial was designed to observe outcomes in administrative data only. While administrative data allow for tracking participants without the need for additional in-person follow-up and the attrition that inevitably entails, these records were missing for participants whose data could not be matched. However, there were no statistically significant differences in the overall rate of matching to the analytical sample across treatment and control groups. Fifth, administrative records cannot comprehensively assess how the program affected maternal and newborn well-being or the maternal perception of the program services.

In this South Carolina–based trial of Medicaid-eligible pregnant individuals, assignment to participate in an intensive nurse home visiting program did not significantly reduce the incidence of a composite of adverse birth outcomes. Evaluation of the overall effectiveness of this program is incomplete, pending assessment of early childhood and birth spacing outcomes.

Corresponding Author: Margaret A. McConnell, PhD, Department of Global Health and Population, Harvard T.H. Chan School of Public Health, 655 Huntington Ave, Room 1217, Boston, MA 02115 (mmcconne@hsph.harvard.edu).

Accepted for Publication: May 24, 2022.

Correction: This article was corrected on February 28, 2023, for incorrect percentages in the last section of Table 2.

Author Contributions: Dr McConnell had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: McConnell, Ayers, Martin, Zhou, Bates, Baicker.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: McConnell, Rokicki, Ayers, Allouch, Perreault, Martin, Chien, Bates.

Critical revision of the manuscript for important intellectual content: McConnell, Allouch, Perreault, Gourevitch, Martin, Zhou, Zera, Hacker, Chien, Bates, Baicker.

Statistical analysis: McConnell, Rokicki, Ayers, Allouch, Perreault, Gourevitch, Martin, Baicker.

Obtained funding: McConnell, Ayers, Bates, Baicker.

Administrative, technical, or material support: McConnell, Ayers, Allouch, Martin, Zhou, Hacker, Chien, Bates.

Supervision: McConnell, Zhou, Bates, Baicker.

Conflict of Interest Disclosures: Dr Baicker reported serving on the board of directors of Eli Lilly and Mayo Clinic and serving on advisory panels for the Congressional Budget Office and National Institute for Health Care Management. Dr Hacker reported serving on the Medical Advisory Board of Renovia Inc. No other disclosures were reported.

Funding/Support: This study was supported by the Children’s Trust of South Carolina, Arnold Ventures, The Duke Endowment, BlueCross BlueShield Foundation of South Carolina, and J-PAL North America Health Care Delivery Initiative.

Role of the Funder/Sponsor: The research team received feedback on the proposed research from funders that informed the design and conduct of the study. The study funders had no role in the collection, management, analysis or interpretation of the data; the preparation or approval of the manuscript; or the decision to submit the manuscript for publication. Funders involved in the Pay for Success project were provided with the opportunity to review the manuscript.

Disclaimers: Ms Bates is currently employed by the State of California’s Office of Cradle-to-Career Data. However, this article was conceived and drafted while Ms Bates was employed at the Massachusetts Institute of Technology, and the findings and views in this article do not reflect the official views or policy of the Office of Cradle-to-Career Data, State of California. The findings and conclusions of this article are those of the authors and do not necessarily represent the official positions of the South Carolina agencies and programs from which the data originated.

Data Sharing Statement: See Supplement 3.

Additional Contributions: The Abdul Latif Jameel Poverty Action Lab led the implementation of the trial. Adam Baybutt, MS (University of California Los Angeles), Kim Gannon, BA (Yale University), Noreen Giga, MPH (Massachusetts Institute of Technology), Elisabeth O’Toole, BA, and Pauline Shoemaker, BA, contributed to the development, implementation and management of the trial and received compensation for their work through their role at the Abdul Latif Jameel Poverty Action Lab (J-PAL) at the Massachusetts Institute of Technology. Anna Nachbor, MPH (Harvard Chan School), contributed to manuscript editing and organizational support of the research and was compensated for her work. We thank current and former leadership and staff at South Carolina Department of Health and Human Services, program nurse-home visitors, and current and former program leadership and staff who were instrumental to implementing the study.