Examining trends, outcomes and barriers to midwifery integration within Louisiana hospitals: a mixed-methods study protocol

Abstract

Background

Childbirth in the USA, including Louisiana, has become increasingly medicalised, with high rates of obstetrical interventions even among clinically low-risk pregnancies. Midwifery-led care has been associated with fewer interventions and comparable or improved outcomes for low-risk births, but midwives remain under-represented in Louisiana hospitals. To improve perinatal outcomes in the Gulf South through increasing midwifery care, data are needed on barriers and facilitators to integrating midwifery care, trends in midwifery care utilisation and birth outcomes by provider type.

Methods and analysis

This sequential explanatory mixed-methods study includes two phases. Phase 1 is a retrospective cohort analysis of Louisiana Vital Records birth certificates from 2012 to 2024, limited to in-hospital births. We will (1) describe temporal trends in the proportion of births attended by certified nurse–midwives (CNMs) overall and within key sociodemographic and geographical subgroups and (2) compare maternal characteristics, obstetrical intervention patterns and selected birth outcomes between CNM-attended and obstetric-attended births to low-risk women with singleton term vertex pregnancies. Study outcomes include caesarean delivery, labour induction, epidural use (where available), 5-minute Apgar score and birth weight. Log-binomial regression (or Poisson regression with robust variance if convergence fails) will be used to estimate adjusted relative risks controlling for sociodemographic and clinical covariates. Phase 2 consists of semi-structured interviews with approximately 20–30 key stakeholders (CNMs, obstetrical clinicians, labour and delivery nurses, hospital administrators and health system or policy advocates) to explore barriers and facilitators to integrating midwifery care into hospital-based maternity care. Interviews will be analysed using thematic analysis.

Ethics and dissemination

This study was approved by the Tulane University Institutional Review Board (Protocol #2026-008, approved 18 March 2026). Findings will be disseminated through peer-reviewed publication, the New Orleans Maternal and Child Health Coalition, Louisiana Perinatal Quality Collaborative and other state maternal health stakeholders.

Trial registration number

Not applicable (observational and qualitative study protocol).

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study uses 13 years (2012–2024) of statewide vital records to examine trends in certified nurse–midwife (CNM)-attended hospital births and compare interventions and outcomes by recorded birth attendant (CNM vs physician).

• The sequential explanatory mixed-methods design enables quantitative trends and outcomes to be interpreted in light of stakeholder-reported contextual and institutional factors, providing depth beyond what either method alone could yield.

• Purposive sampling across diverse hospital settings (urban/rural, academic/community, varying levels of midwifery integration) will capture a range of perspectives on barriers and facilitators.

• CNM-attended births are known to be under-reported on birth certificates, which may result in conservative estimates of midwifery utilisation; sensitivity analyses will assess the impact of this limitation.

• As a secondary analysis of vital records, the quantitative phase is limited by the variables captured on the birth certificate; unmeasured confounding cannot be fully excluded, and causal inference is not claimed.

Background

Compared with other high-income countries, childbirth in the USA is characterised by high and often rising rates of obstetrical interventions, including labour induction, augmentation, continuous electronic fetal monitoring and caesarean delivery,^1–5 even among clinically low-risk pregnancies.^{6 7} While these interventions can be lifesaving when medically indicated, routine use in low-risk births (ie, singleton gestation, term (37–41 weeks), cephalic/vertex presentation and absence of significant maternal comorbidities or obstetrical complications)⁶ has been associated with increased maternal morbidity, postpartum haemorrhage and neonatal respiratory complications.^7–9 National and international recommendations emphasise avoiding non-medically indicated interventions and supporting low-intervention birth where possible.^10–12

In Louisiana, these concerns are intensified by persistently high rates of severe maternal morbidity and mortality, and adverse pregnancy and birth outcomes, especially among black birthing people and those in rural or under-resourced areas.^13–15 The state has consistently ranked worst in maternal and infant health indicators in the nation,^16–18 driven in part by systemic inequities, including structural racism and place-based disadvantage (eg, poverty; maternity care deserts) that limit access to high-quality perinatal care.^{14 15 19 20} These patterns point to profound inequities in access to low-intervention, respectful perinatal care and highlight the importance of understanding how institutional context and birth provider may shape birth experiences and outcomes.

Midwifery-led care is a key evidence-informed strategy to promote safer, less medicalised childbirth for low-risk pregnancies.^21–23 Studies from the USA and other high-income countries suggest midwifery care is associated with lower rates of induction, episiotomy and other interventions,^21–26 along with higher satisfaction compared with obstetrician-led care for low-risk pregnant people.^27–30 Yet in Louisiana, certified nurse–midwives (CNMs), who are trained to provide comprehensive care for clinically low-risk pregnancies and most commonly practice in hospital settings, attend fewer births compared with states with better perinatal outcomes.^{31 32} Instead, obstetricians/physicians provide the vast majority of intrapartum care and institutional cultures, reimbursement mechanisms and regulatory frameworks have historically privileged intervention-intensive models of care.^{7 31 33}

Louisiana’s midwifery landscape is shaped by restrictive advanced practice registered nurse (APRN) regulations,³⁴ including a requirement that CNMs maintain a physician collaborative practice agreement as a condition of practice. These requirements can constrain practice autonomy, workforce deployment and integration within hospital systems and contribute to Louisiana’s low midwife-to-birth ratio.^{33 34} Louisiana has among the lowest rates of midwife-attended births nationally, and the midwifery workforce remains small relative to maternity care need.³⁵ CNMs are licensed APRNs and may be employed by hospitals, health systems or physician or midwifery practices. Employment status is distinct from practice setting, and CNMs attending hospital births practice within the credentialing, privileging and collaborative arrangements of the institutions in which they provide care.^33–35 Hospital-based maternity care is team based, with labour and delivery nurses, CNMs and physicians contributing to intrapartum care according to their respective roles and institutional practice models.^{35 36} CNMs and physicians may provide intrapartum management and attend births within their scope of practice and institutional privileges.^{7 26 36} These organisational and regulatory arrangements are therefore central to understanding midwifery integration in Louisiana.

There is limited population-level information on how CNMs have been incorporated into Louisiana’s hospital system over time. Specifically, little is known about which populations access midwifery care, how intervention patterns and outcomes differ between CNMs and obstetricians/physicians and what system-level factors facilitate or impede integration. Birth certificate data capture the recorded attendant at delivery (eg, CNM or physician) and key childbirth interventions and outcomes statewide, offering an opportunity to quantify trends in CNM-attended births and compare low-risk births by recorded attendant type.^{37 38} Qualitative inquiry with stakeholders is also needed to illuminate how regulatory, institutional and cultural contexts influence these patterns.^{39 40}

Study purpose and aims

The objective of this study is to examine how recorded attendant at delivery and hospital context are associated with childbirth interventions and outcomes in low-risk hospital births in Louisiana from 2012 to 2024 and identify barriers and opportunities for integrating CNM-led care within hospital settings.

Specific aim 1 (quantitative)

To examine the prevalence and temporal trends (2012–2024) in childbirth interventions and associated maternal and neonatal outcomes among low-risk hospital births in Louisiana, comparing births by recorded attendant at delivery (CNM vs physician).

Specific aim 2 (qualitative)

To explore barriers and facilitators to integrating midwifery-led practice within hospital settings through in-depth interviews with key stakeholders, including CNMs, obstetrical clinicians, labour and delivery nurses, hospital administrators and health system or policy advocates.

Research questions

• Among low-risk (ie, singleton gestation, term (about 37–41 weeks), cephalic/vertex presentation, without maternal comorbidities or obstetrical complications) hospital births in Louisiana, how do rates and trends of obstetrical interventions and perinatal outcomes (eg, labour induction, delivery complications, postpartum haemorrhage, neonatal intensive care unit (NICU) admission, low Apgar score) differ between births attended by CNMs and those attended by obstetricians?

• How do midwives, obstetrical clinicians, hospital administrators and other maternity care stakeholders describe barriers and facilitators that shape the integration of CNM-led care in hospital settings?

• How can integrated interpretation of quantitative and qualitative findings inform strategies to reduce unnecessary interventions and expand midwifery-integrated maternity care in Louisiana?

Methods and analysis

Study design

This study will use a sequential explanatory mixed-methods design with two phases (figure 1), drawing on a pragmatist orientation in which quantitative and qualitative approaches are combined to generate a more comprehensive understanding of complex health phenomena.^{41 42} This design was chosen because the research questions require population-level estimates of intervention patterns alongside in-depth exploration of contextual and institutional factors shaping midwifery integration.

Figure 1. Sequential explanatory mixed-methods design for examining trends, outcomes and barriers to midwifery integration in Louisiana hospitals. CNMs, certified nurse–midwives; L&D, labour and delivery; OB, obstetrical.

Phase 1 (quantitative) will involve a retrospective cohort analysis of Louisiana Vital Records birth certificates (2012–2024) to describe trends in midwifery use among hospital births and to compare maternal characteristics, obstetrical intervention patterns and selected birth outcomes between CNM-attended and obstetric-attended low-risk hospital births.

Phase 2 (qualitative) will consist of semi-structured interviews with key stakeholders, including CNMs, obstetrical clinicians and clinical leaders, labour and delivery nurses, hospital administrators and health system or policy actors and advocates, to explore barriers and facilitators to integrating CNMs into hospital-based maternity care and explain patterns observed in Phase 1.

Quantitative and qualitative findings will be integrated during interpretation using joint displays and narrative weaving, consistent with recommended strategies for mixed-methods integration in public health and health services research.^43–45

Study setting

The quantitative phase will include all low-risk hospital births in Louisiana recorded in the state’s Vital Records system between 2012 and 2024. The qualitative phase will focus on stakeholders involved in maternity care in Louisiana, including those practicing or making decisions in urban and rural hospitals, community practices with hospital affiliations and relevant agencies or organisations. The overall study is being conducted from July 2026 through June 2028. Phase 1 is a retrospective analysis of birth records from 2012 to 2024; Phase 2 stakeholder recruitment and interviews will follow the initial quantitative analyses, with integration of quantitative and qualitative findings occurring during the final study period.

Phase 1: quantitative component

Data source

Phase 1 will involve a secondary analysis of de-identified birth certificate data obtained from the Louisiana Department of Health Vital Records, including all live births from 1 January 2012 through 31 December 2024. Variables to be used include maternal sociodemographic characteristics (eg, age, race/ethnicity, education, insurance status), prepregnancy and pregnancy conditions (eg, prepregnancy body mass index, hypertension, diabetes), obstetrical procedures and interventions (eg, labour induction, epidural use, caesarean delivery), provider type (CNM vs physician) and key perinatal outcomes (eg, preterm birth, low birth weight, NICU admission). The analytical sample is expected to include approximately 780 000 in-hospital births, of which an estimated 25 000 were attended by CNMs.

Target population

The primary analytic population for provider type comparisons will be low-risk hospital births. Inclusion and exclusion criteria will be defined as follows:

Inclusion criteria: (1) births occurring in a hospital in Louisiana, (2) singleton gestation, (3) gestational age ≥37 completed weeks, (4) cephalic/vertex presentation and (5) absence of significant maternal comorbidities or obstetric complications.

Exclusion criteria (high-risk conditions where identifiable): (1) pre-existing (pregestational) diabetes, (2) chronic hypertension, (3) placenta previa, (4) multiple gestation, (5) known major congenital anomalies where recorded and (6) other conditions (eg, gestational hypertension, gestational diabetes, prior caesarean delivery) indicating at-risk pregnancy when reliably captured in the certificate.

These criteria align conceptually with prior work comparing midwifery and obstetrical care in low-risk hospital births^{22 23} and ensure that comparisons focus on clinical scenarios appropriate for either midwifery or obstetric management.

Variables

The primary exposure will be provider type, derived from the birth certificate field indicating the attendant at delivery. Provider categories will be constructed as CNM versus obstetrician or physician (including obstetrician–gynaecologists and other physicians such as family physicians listed as attending the birth). For trend analyses, CNM attendance will be operationalised as the proportion of hospital births recorded as attended by CNMs in each calendar year, overall and within selected maternal subgroups. Because CNM-attended births are known to be underreported on birth certificates,^{46 47} these estimates are likely to be conservative.

The primary trend outcomes for 2012–2024 will be the annual number and percentage of hospital births attended by CNMs, as well as the annual CNM share of births within strata of maternal age, race and ethnicity, payer (Medicaid vs private insurance) and geographical region. These annual measures will be used to estimate and describe temporal changes in CNM involvement in hospital births over time and across key sociodemographic subgroups.

Maternal characteristics recorded on the birth certificate will include age, race and ethnicity, parity, education, marital status, insurance type, smoking during pregnancy and prepregnancy and gestational conditions. Obstetrical intervention variables will include induction of labour and epidural use. Neonatal outcomes will include 5-minute Apgar score (<7 vs ≥7), birth weight and low birth weight (<2500 g), and NICU admission. In multivariable models, covariates will include maternal age (modelled as continuous or categorical), race and ethnicity, parity, education, insurance type, smoking during pregnancy and available pregnancy complications that may confound associations between provider type and study outcomes.

Sample size and statistical power

The expected analytical sample of approximately 25 000 CNM-attended low-risk births and a far larger number of obstetric-attended births provides substantial power to detect clinically meaningful differences in the primary outcome. Using Louisiana’s estimated baseline caesarean rate of approximately 35–38% for low-risk births, a two-proportion z-test indicates that this sample is adequately powered (>99%) to detect absolute differences of 3 percentage points or greater at α=0.05. The large sample size similarly provides adequate power for subgroup analyses by race/ethnicity and payer.

Statistical analyses

Stage 1: Descriptive and trend analyses will characterise the distribution of maternal characteristics, provider types, interventions and outcomes in the full birth cohort and among the low-risk hospital births. We will also estimate, for each year from 2012 to 2024, the proportion of hospital births attended by CNMs overall and stratified by maternal age, race/ethnicity, parity and payer, with temporal trends examined using graphical displays and regression models with year as a continuous variable (eg, log-binomial or logistic regression estimating annual changes in CNM attendance).

Stage 2: Among low-risk hospital births, comparative analyses by provider type will compare maternal characteristics between CNM-attended and obstetric-attended births using χ² tests for categorical variables and t-tests for continuous variables. We will then estimate unadjusted relative risks for each intervention and birth outcome, and fit multivariable log-binomial or Poisson regression models with robust variance to obtain adjusted relative risks and 95% CIs adjusting for maternal sociodemographic and clinical covariates. To account for potential clustering of births within hospitals, SEs will be clustered at the hospital level.

Stage 3: Subgroup analyses will assess effect modification by race/ethnicity, insurance type and geography (rural/urban) using interaction terms. Where interactions are statistically and substantively meaningful, we will present estimates for each stratum/subgroup. Sensitivity analyses will apply alternative definitions of low-risk status (eg, restricting to nulliparous pregnancies with no pre-existing or pregnancy complications) and, where feasible, restrict to hospitals and periods with stable, complete provider coding to evaluate robustness of findings given known under-reporting of CNM births. Because vital records do not identify the provider at admission or primary labour manager, CNM-to-physician transfers during labour cannot be reclassified in sensitivity analyses.^{36 48} Missing covariate data will be handled using complete case analysis, with a description of the extent and pattern of missingness reported; multiple imputation will be used as a sensitivity analysis if missingness exceeds 10% for key variables. All statistical analyses will be performed using SAS V.9.4 (SAS Institute, Cary, North Carolina, USA), and p < 0.05 will be considered statistically significant.

Phase 2: qualitative component

Design and sampling

Phase 2 is an exploratory qualitative study using semi-structured, in-depth interviews. The goal is to understand how stakeholders perceive and experience the integration of CNM-led care into Louisiana’s hospital maternity system and to identify barriers and facilitators at multiple levels.

Stakeholder groups will include: (1) CNMs practicing in Louisiana hospitals, (2) obstetrical clinicians (eg, obstetrician–gynaecologists and family physicians attending births), (3) labour and delivery nurses and (4) health system leaders, public health officials and policy stakeholders involved in organising, regulating or financing maternity care (eg, hospital administrators, state maternal health officials, and representatives of professional organisations).

Purposive sampling will be used to capture variation by practice setting (urban vs rural; academic vs community; public vs private), degree and model of midwifery integration (eg, hospitals with established CNM units vs those without, collaborative practices vs more siloed arrangements) and stakeholder role and perspective (direct care vs leadership/policy).

Recruitment

Participants for Phase 2 will be identified through existing professional networks and partnerships maintained by the research team with hospitals, maternity care providers and maternal health organisations across Louisiana (eg, the Louisiana Perinatal Quality Collaborative and the Maternal and Child Health Coalition). Potential stakeholders will be contacted via email and/or phone and invited to participate in a semi-structured interview. Eligibility will be based on having current or recent (within the past 5 years) professional experience in providing, overseeing or shaping maternity care in Louisiana (eg, as a CNM, obstetrical clinician, hospital administrator or maternal health policymaker); all participants will be aged 18 years or older and not recruited from vulnerable populations.

We anticipate conducting approximately 20–30 interviews, with final sample size determined by thematic saturation, defined as the point at which no new themes arise in successive interviews.⁴⁹ We will additionally apply an information power framework⁵⁰ to assess sample adequacy, considering sample specificity, study aims and analytic approach.

Data collection

Interviews will be conducted via video calls (eg, Zoom) or in person, according to participant preference and feasibility, and scheduled at a date and time convenient to the participant. Verbal informed consent will be obtained prior to the start of each interview. Interviews (expected to last approximately 45–60 min) will be audio-recorded and professionally transcribed. A semi-structured interview guide will include open-ended questions and probes such as: (1) how participants describe the current role of midwifery care in Louisiana’s hospitals; (2) factors that have supported or hindered integrating CNMs into specific institutions or regions; (3) perceived effects of hospital policies, credentialing, reimbursement or liability concerns on midwives’ ability to provide physiological, low-intervention care; (4) views on how provider type and institutional culture influence intervention use and outcomes for low-risk births; and (5) perceived system, institutional or policy changes that would most help expand equitable access to midwifery-integrated care. Questions will be refined in light of Phase 1 findings. The semi-structured interview guide is provided as online supplemental file 1. Field notes will be taken after each interview to capture contextual information, non-verbal cues where applicable and initial analytical impressions.⁵¹

Qualitative data analysis

Interview transcripts will be analysed using thematic analysis, drawing on the approach outlined by Braun and Clarke⁵² and combining both inductive and deductive coding logics.^{53 54} The analytical process will begin with familiarisation through repeated reading of transcripts and field notes to gain an overall sense of the data, followed by line-by-line coding of segments related to midwifery integration, barriers and facilitators, interprofessional dynamics, equity concerns and perceived effects on interventions and outcomes. A preliminary codebook will then be developed that integrates inductive codes emerging from participants’ accounts with a focused set of deductive codes linked to the study aims (eg, regulatory context, hospital culture and payment structures), and these codes will be grouped into candidate themes and subthemes.^{54 55} Themes will be iteratively reviewed and refined by comparing patterns across stakeholder groups and settings, examining points of convergence and divergence and refining definitions to ensure conceptual clarity and analytical depth.⁵⁵ To support credibility, a subset of participants will be invited to review preliminary themes and provide feedback on the accuracy and resonance of interpretations.^{56 57}

At least two analysts will code an initial subset of transcripts independently, compare coding and resolve discrepancies through discussion before finalising the codebook, which will then be applied to remaining transcripts with periodic checks to ensure consistency. NVivo or similar qualitative software will be used for data management and coding, and an audit trail documenting coding decisions, codebook revisions and theme development will be maintained to support transparency and rigour.⁵³

Positionality

The research team includes interdisciplinary social scientists, reproductive epidemiologists, CNMs and public health researchers with professional and scholarly commitments to improving maternity care in Louisiana. Several team members have direct clinical experience in midwifery or obstetrical care. We acknowledge that these perspectives may shape data collection and interpretation. To mitigate potential bias, we will document and critically reflect on our positionality throughout data collection and analysis, use multiple coders with divergent viewpoints and maintain an audit trail of analytic decisions.⁵⁸ Reflexive memos will be used to record and interrogate interpretive assumptions during thematic analysis.⁵⁹

Integration of quantitative and qualitative components

Integration of quantitative and qualitative components will follow an explanatory mixed-methods logic in which preliminary Phase 1 findings will guide Phase 2 sampling and data collection. Quantitative results (eg, regional patterns of midwifery use, hospital-level differences in labour interventions by provider type or subgroups with limited access to midwifery care) will be used to refine the interview guide and purposively recruit stakeholders from hospitals with contrasting patterns. During analysis, joint displays will juxtapose key quantitative patterns (such as temporal changes in CNM-attended births by payer or region) with qualitative themes (such as stakeholder accounts of regulatory, institutional or payer-driven practice changes) to develop integrated, contextually grounded explanations for observed associations.^{44 60} Integrated interpretation will then be used to construct a coherent narrative about how regulatory frameworks, institutional conditions, professional relationships and community demand shape trends in midwifery use, provider-type differences in intervention and outcome patterns and potential strategies as well as recommendations to expand midwifery-led maternity care.

Patient and public involvement

Patients and the public were not directly involved in the design of this study protocol. Findings will be shared with community partners (eg, the New Orleans Maternal and Child Health Coalition) and disseminated in accessible formats to relevant community organisations. For the qualitative phase, participants (including patient advocates where applicable) will have the opportunity to review themes for member checking.⁵⁷

Ethical considerations

Tulane University Institutional Review Board approval for this study has been obtained. All reporting of quantitative results from de-identified vital records will adhere to federal and state disclosure rules, including suppression or aggregation of small cell sizes where necessary to protect privacy. All qualitative participants will provide verbal informed consent after being informed about the study purpose, procedures, risks, potential benefits and their right to decline participation or withdraw at any time without penalty. Interviews will be scheduled at times and in formats that support confidentiality, and transcripts will be de-identified, with pseudonyms or role descriptors used in all reports and publications. Given potential professional sensitivities, care will be taken to avoid inadvertently identifying individual providers or institutions when presenting themes and illustrative quotes.

Rigor and quality assurance

The quantitative component will adhere to recommended reporting standards for observational studies (Strengthening the Reporting of Observational Studies in Epidemiology (STROBE)) and will include clear specification of inclusion and exclusion criteria, handling of missing data and assessment of model assumptions and diagnostics. For the qualitative component, rigour will be guided by established reporting standards for qualitative research (Standards for Reporting Qualitative Research (SRQR) and/or Consolidated Criteria for Reporting Qualitative Research (COREQ),^{61 62} as appropriate). Strategies to enhance credibility, dependability, confirmability and transferability will include prolonged engagement with and iterative review of the data, involvement of multiple analysts and maintenance of an audit trail and documented coding framework.⁶³ Purposive sampling will be used to capture diverse institutional and geographical contexts, supported by rich description of settings and participants. Mixed-methods rigour will be strengthened by explicitly specifying the purpose of integration, documenting how quantitative results informed qualitative sampling and interpretation, and using joint displays to present integrated inferences in a systematic and transparent way.^{42 60}

Discussion

This protocol describes a sequential explanatory mixed-methods study designed to advance understanding of midwifery care integration in Louisiana’s hospital-based maternity system. By combining statewide trend and outcome analyses using 2012–2024 birth records with in-depth interviews from diverse stakeholders, the study responds to calls for mixed-methods research linking quantitative patterns with qualitative system-level data.^{64 65}

The quantitative phase will provide the first comprehensive description of trends in CNM-attended hospital births in Louisiana over more than a decade, and of differences in intervention patterns and selected outcomes between CNMs and obstetricians among low-risk births. The qualitative phase will offer insight into regulatory, institutional and cultural factors that shape these patterns, including barriers and facilitators to midwifery integration, such as credentialing processes, liability concerns, interprofessional relationships, payment structures and community demand.^{26 32 33}

Key limitations include: (1) birth certificate under-reporting of CNM births, which will be addressed through sensitivity analyses; (2) unmeasured confounding in the quantitative phase, as self-selection into CNM-attended birth cannot be fully controlled; (3) potential social desirability bias in qualitative interviews, given professional stakes; and (4) generalisability of qualitative findings beyond Louisiana’s specific regulatory and institutional context. A further limitation is that provider type reflects the attendant at delivery rather than the individual or attendant-type who initiated or primarily managed labour. Patients who begin labour under CNM care but transition to physician-led care before giving birth may be classified as physician-attended. This crossover cannot be identified in the available vital records and could bias provider comparisons by shifting more complex CNM-initiated cases into the physician-attended group.

Our findings will inform state and institutional strategies to expand access to midwifery care in the Gulf South, with relevance for other US settings seeking to reduce unnecessary obstetrical interventions and improve maternal and infant health. Findings will be disseminated to the New Orleans Maternal and Child Health Coalition, the Louisiana Perinatal Quality Collaborative, the Louisiana Department of Health Office on Women’s Health and Community Health, hospital system leadership and other state maternal health stakeholders.

The funder had no role in study design, data collection, analysis, interpretation or the decision to submit for publication.