Abstract
Introduction:
Hawai‘i reports favorable perinatal health outcomes compared to national averages (preterm birth rate: 9.8%; infant mortality rate: 4.6 per 1,000 live births), yet disparities persist. Native Hawaiian and Pacific Islander women experience higher rates of preterm birth (12.6%) and lower early prenatal care (62.7%). Med-QUEST, Hawai‘i’s Medicaid program, supported implementation of a Perinatal Quality Collaborative (PQC) to expand access to evidence-based maternal safety practices across birthing facilities. Hospitals implemented protocols (e.g., training, drills, and equipment changes) aligned with the Alliance for Innovation on Maternal Health (AIM) Patient Safety Bundles, with cross-hospital learning facilitated through structured monthly meetings and peer exchange. This study examines how the PQC operationalized quality improvement (QI) among Medicaid-serving hospitals.
Methods:
We used a mixed-methods approach, including observational analysis of severe maternal morbidity (SMM) rates among Hawai‘i’s Medicaid population and 10 semi-structured interviews from participating birthing facilities. Interviews explored implementation processes, supports, and barriers. Hospitals’ progress was tracked through milestone completion.
Results:
Participants reported structured AIM bundle implementation, gap analyses, and sustained peer exchange within a cooperative, non-competitive culture. SMM rates increased from 2016 to 2022 and stabilized in 2023 to 2024; racial/ethnic disparities persisted throughout the PQC implementation. Qualitative findings emphasized infrastructure and context-specific adaptations rather than targeted disparity-reduction strategies.
Health Equity Implications:
Expanding access to structured implementation support across diverse and rural hospitals may strengthen statewide QI infrastructure; targeted strategies beyond inpatient bundle implementation are likely necessary to reduce disparities.
Conclusion:
The state-level Pay-for-Performance effectively facilitated statewide implementation of AIM bundles and hospital-level QI processes among Medicaid-serving facilities. Observed SMM trends underscore the importance of sustained implementation efforts and complementary strategies to address disparities.
Keywords
Introduction
Persistent inequities in maternal and infant health outcomes remain a critical challenge in the United States.1,2 This challenge is particularly acute in Hawai‘i, where Native Hawaiian and Pacific Islander women experience disproportionately high rates of adverse outcomes. Namely, their preterm birth rate is 12.6%, and 37.3% of this population has limited access to early prenatal care.3,4 Geographic barriers compound these disparities by limiting access to consistent obstetric care for those in rural areas. 5 Immigrant populations concurrently encounter systemic barriers. Recent large policy changes to Medicaid, which covers vulnerable populations who experience significantly higher severe maternal morbidity (SMM) rates than privately insured individuals, may have widespread negative effects on maternal and infant health.6–9 In response to these challenges, the state of Hawai‘i has established Perinatal Quality Collaboratives (PQCs) as a key strategy to implement systemic, evidence-based improvements by convening diverse stakeholders—including clinicians, public health experts, and community advocates—to accelerate the implementation of evidence-based practices.10–12 Three bundles were implemented in succession—Obstetric Hemorrhage, Severe Hypertension, and Substance Use—with the initial two selected in alignment with the 2019 Joint Commission PC Standards for Maternal Safety. Given that hemorrhage and hypertension are among the leading causes of SMM, 13 and that Native Hawaiian and Pacific Islander women in Hawai‘i experience SMM at disproportionate rates, 14 these bundles represent a targeted response to existing disparities.
National- and state-level PQCs have achieved significant gains, including reducing early elective deliveries, lowering cesarean section rates, preventing surgical site infections, increasing the administration of antenatal corticosteroids to eligible women, and improving screening for social determinants of health.15,16 Nevertheless, no previous studies on PQCs focused on identifying quality outcomes among Medicaid recipients, thus highlighting the need to specifically investigate the ability of collaboratives to effectively target quality outcomes for vulnerable groups, for the Medicaid population at large, and subpopulations specifically.
Understanding how PQCs operationalize evidence-based practices across diverse hospital contexts is valuable for replicating and scaling successful models in other complex care environments. This study, therefore, examines how the Hawai‘i PQC expanded access to evidence-based practice implementation across Medicaid-serving hospitals, with a focus on the Severe Hypertension and Obstetric Hemorrhage bundles, and the implementation processes that supported maternal and infant quality improvement (QI).
Methods
Intervention: Stepped Implementation of Alliance for Innovation on Maternal Health Bundles and Gap Analysis
Hawai‘i’s PQC is supported through several state and federal funding streams with overlapping goals: reduce maternal mortality and SMM; improve maternal health overall; address maternal health disparities; enhance workforce training, data capacity, and quality; and support innovative programming and evidence-based implementation.
The Hawai‘i PQC sought to improve maternal and perinatal health by enrolling in the Alliance for Innovation on Maternal Health (AIM) Patient Safety Bundles in 2021, a federally funded initiative supported by the Health Resources and Services Administration. All 10 participating hospitals implemented the Obstetric Hemorrhage and Severe Hypertension in Pregnancy bundles. The Severe Hypertension bundle launched in 2021, with hospitals transitioning to the Obstetric Hemorrhage bundle upon completion, resulting in varied timelines across facilities. Most hospitals began the hemorrhage bundle in 2022–2023. Progress was tracked through annual progress reports and monthly meetings in which hospital representatives reported on their current implementation stage.
As part of the collaborative, hospitals gained access to the AIM data center, providing hospital-specific and statewide outcome data. Hospitals also completed gap analyses to assess bundle readiness, workflows, and outcomes, with attention to populations at higher risk for maternal morbidity where data were available. Findings informed hospital-specific interventions and targeted technical assistance, allowing tailored support based on local capacity. Equity within the collaborative was conceptualized as expanding access to structured implementation support across diverse and rural hospitals rather than implementing targeted disparity-reduction interventions.
QI support
The collaborative supported hospitals through monthly webinars sharing best practices, access to AIM bundles and a data center, and clinical program manager guidance. Hospitals used gap analysis findings to revise protocols, adapt workflows, train staff, conduct simulations, and improve documentation as part of bundle implementation.
Participants
All 10 birthing hospitals in Hawai‘i (100% of eligible facilities), including facilities on the Island of Hawai‘i (n = 3), O‘ahu (n = 4), Kaua‘i (n = 2), and Maui (n = 1), participated in the collaborative through Med-QUEST’s Pay-for-Performance (P4P) Program. Participating facilities included a mix of system-affiliated, stand-alone, and state-operated hospitals. All 10 hospitals participated in both the Severe Hypertension and Obstetric Hemorrhage bundles. Implementation leads and meeting representatives varied by facility size and structure. Specifically, roles were typically filled by nursing managers, quality coordinators, or OB providers, with smaller hospitals often having individuals serving in multiple capacities. All hospitals successfully completed both bundles.
Data collection and analysis
We used a mixed-methods design to examine Hawai‘i PQC implementation processes and contextualize observed outcome trends. Qualitative data were collected through 14 participants across 10 facility-level interviews, including one representative from the coordinating organization. Participants were selected based on their roles within the collaborative and were primarily nurse managers or quality directors with clinical experience ranging from 2 to 28 years (M = 17.3).
All birthing facilities participated in the evaluation as part of the P4P Program. Representatives from all birthing facilities completed informed consent procedures (see Supplementary Data S1 and S2). Custom interview guides (see Supplementary Data S2) were developed guided by the Agency for Healthcare Research and Quality (AHRQ) Learning Collaboratives Taxonomy. 17 Interviews lasted 30–50 min and were conducted via Zoom videoconferencing software, recorded, and transcribed verbatim. Qualitative data analysis was conducted through directed content analysis using NVivo software v15.18,19 A priori codes were identified by iterating on themes adapted from the AHRQ framework proposed by Nix et al.. 20 Codes were differentiated by participant role category (e.g., “Director/Manager,” “Quality Coordinator,” “Care Provider”). Two coders double-coded three transcripts, collaboratively discussing and collating codes until interpretations aligned. After achieving inter-rater reliability (κ = 0.83), they independently coded the remaining transcripts, meeting weekly to address categorization challenges and ensure shared meaning. AHRQ themes were then combined to derive cross-cutting implementation themes.
Second, we identified focus points for quantitative insights. This included the selection of outcome measures in line with the AIM bundle implementation and a focus on specific race and ethnicity differences in outcomes.
Quantitative measures were aligned with the Obstetric Hemorrhage and Severe Hypertension AIM bundles and drawn from available Medicaid administrative data. Analyses described outcome trends during the implementation period rather than evaluating causal impact.
Medicaid administrative claims and encounter data from 2016 to 2024 were used for analysis and were retrieved on September 18, 2025. We calculated all birth admissions of the Hawai‘i Medicaid maternal population, excluding ectopics and miscarriages, including only births at the collaborating birthing facilities. We then identified incidents of SMM, obstetric hemorrhage, and preeclampsia using diagnosis and procedure codes from the 2024 AIM code list. 21 Rates of SMM, obstetric hemorrhage, and preeclampsia among all birth admissions were calculated. Additionally, SMM rates among participants who experienced obstetric hemorrhage or preeclampsia were calculated and stratified within racial groups. SMM was selected as the main outcome measure, and data on major contributing factors of obstetric hemorrhage and preeclampsia are provided for context. Measures were included according to the AIM bundle core data collection plan for Obstetric Hemorrhage in Patient Safety Bundle and Severe Hypertension in Pregnancy Patient Safety Bundle.22,23 Missing race/ethnicity was coded as unknown. Analyses were limited to available administrative race/ethnicity classifications. We conducted a chi-square test to identify statistically significant differences in rates between racial groups.
Results
Interviews
Key factors driving QI and program barriers and facilitators were identified and categorized under seven core themes: (1) program scope and design, (2) communication and engagement modalities, (3) culture of the collaborative and peer learning, (4) coordinator credibility and engagement, (5) shared vision and goal clarity, (6) sustainability, and (7) roles, processes, and capacity.
Quotations derived from the qualitative interviews can be referenced in Supplementary Data S3.
(1) Program scope and design
The learning collaborative operated statewide from 2022 across all 10 birthing facilities in Hawai‘i (O‘ahu: 4; Hawai‘i Island: 3; Kaua‘i: 2; Maui: 1). The primary focus was perinatal QI through adoption of AIM bundles to standardize evidence-based practices aimed at preventing maternal morbidity and mortality.
The initiative combined structured bundle-defined deliverables with site autonomy, allowing contextual adaptation for approaching the goals and facilitating a clear working structure. Birthing facilities were encouraged to implement their own tailored evidence-based improvements following the results of their gap analyses. Participants valued the provided freedom to develop targeted interventions and priorities, as it allowed projects to be individualized and highly relevant to their hospital.
To support consistent implementation across diverse settings, AIM bundle interventions were tailored to the specific infrastructure and resource constraints of each hospital. This collaborative process between the program manager and hospital teams focused on maintaining a high standard of care despite physical limitations. For instance, one Director/Manager details how they overcame space constraints by substituting a fixed hemorrhage cart with a mobile response team.
The resulting changes from the collaborative focused on internal policies and care procedures supported through staff education and simulation drills, including hypertension-specific (e.g., taking secondary blood pressure measurement within 15 min) and hemorrhage-focused interventions (e.g., creating postpartum hemorrhage carts, transitioning from estimated to quantitative blood loss measurement). Some facilities sought to strengthen the continuum of care; however, participants noted that factors beyond the inpatient setting (i.e., outpatient follow-up appointments and patient education) are beyond the manager’s direct control.
(2) Communication and engagement modalities
Recognizing that travel requirements disproportionately burden hospitals located on rural islands, the PQC prioritized equitable access through a multi-modal virtual engagement strategy. Specifically, monthly 60-min webinars and virtual office hours provided a low-barrier entry point for participation, ensuring that distance did not dictate the level of support or engagement received from the program manager. These efforts were supplemented by in-person training and ongoing training certification opportunities that supported onboarding and skill development. Key outcome measures were tracked through aggregated dashboards. Communication directionality ranged from expert-to-participant to mixed peer learning and exchange rather than competitive. Participants noted that the non-competitive nature of the birthing facilities facilitated peer-to-peer learning.
(3) Culture of collaborative and peer learning
Participation in the collaborative was mandatory to qualify for P4P reimbursement, yet the degree of internal and cross-facility exchange varied. Dialogue across birthing facilities occurred through monthly check-ins and touchpoints, which facilitated information dissemination and alternative approaches.
Collectively, sharing best practices across facilities reduced trial and error in implementation and enabled hospitals to apply or adapt strategies that were both relevant and effective for their local contexts, supporting more consistent QI across settings with varying resources and patient populations. Birthing facilities that reported strong executive endorsement and an established culture of collaboration described smoother alignment across departments and fewer barriers to implementing changes. Overall, the peer-learning structure was frequently cited as helpful for surfacing practical solutions, although the depth of engagement depended on local norms and leadership reinforcement.
(4) Coordinator credibility and engagement
Sites consistently described strong, credible coordination that was available, responsive, and knowledgeable. The program manager’s firsthand experience with clinical and administrative operations, approachable style, and ability to foster a safe environment for discussion were key characteristics to being effective.
(5) Shared vision and goal clarity
All birthing facilities articulated and shared a commitment to improving outcomes for their community in Hawai’i, indicating support for the collaborative. However, the specific goals and operationalization varied by site. For example, some sites prioritized substance-use-related perinatal complications. Hospitals that entered the collaborative with already low baseline rates reported needing to reinterpret or broaden their target outcomes to maintain relevance and momentum.
Furthermore, resource limitations influenced hospitals’ goals and scope of implementation, as capacity varied by geography, particularly between rural and urban settings. Many smaller, rural hospitals admitted difficulty in implementation due to available resources, including staffing, dedicated personnel, or additional financial support.
In an effort to address geographic inequities, the program manager facilitated resource-sharing by connecting hospitals facing similar limitations or challenges. This networking reframed individual hospital limitations into opportunities for exchanging resources and expertise, reducing isolation among smaller or rural facilities.
(6) Sustainability
The collaborative was initiated in 2021, started improvement activities in 2022, and is still ongoing. It was explicitly designed for integration beyond the program period. Routine cadence (webinars, office hours) and deliverable-linked governance created a backbone for continuity, helping teams transition from pilot activities to standard operating procedures.
This structure is especially salient for resource-constrained hospitals, where regular support helped mitigate drift and turnover-related disruption. This commitment to long-term sustainability was further demonstrated through workforce development efforts aimed at addressing challenges faced by rural hospitals. The PQC offered a free in-person patient safety training along with certification opportunities. Several hospital teams expressed strong appreciation for this opportunity, as it not only enhanced staff’s knowledge of best practices but also enabled rural hospitals to certify local staff as clinical instructors. This promoted further internal dissemination of evidence-based practices, especially in resource-limited settings. Rural hospital teams, in particular, emphasized that such an opportunity would not have been available to them without the collaborative’s support.
Participants additionally have shared their desire for the collaboration to continue.
(7) Roles, processes, and capacity
Successful bundle implementation required cross-departmental participation. The presence of experienced quality managers was pivotal in absorbing administrative burden (e.g., documentation, data pulls, deliverables), allowing clinicians to focus on care. Where these roles were thin or absent, internal processes sometimes limited implementation, and staff reported taking on additional tasks to meet program requirements. Hospital size, therefore, influenced the distribution of workload, with smaller, rural sites more likely to feel capacity strain, ultimately impacting pacing and support intensity.
Quantitative results
These findings describe observed trends during the implementation period and do not imply causal effects of bundle implementation. As detailed in Table 1 and illustrated in Figure 1, the SMM rates increased beginning in 2017, peaked in 2022 (3.76), and decreased between 2022 and 2023, remaining stable in 2024. The increase in SMM rates corresponded with a rise in the absolute number of SMM cases rather than substantial changes in birth admissions. The annual number of birth admissions remained relatively constant, averaging approximately 6,074 per year, while the count of SMM cases grew from a low of 174 in 2017 to a high of 236 in 2022.
Severe maternal morbidity by year and race.
Severe Maternal Morbidity Among Hawai‘i Medicaid Maternal Population from 2016 to 2024
SMM, severe maternal morbidity.
Analysis of two major SMM contributors—obstetric hemorrhage and preeclampsia—showed distinct trends from 2016 to 2024 (Table 2). Obstetric hemorrhage (only) peaked in 2022 at 15.80%. Preeclampsia (only) reached peaks in 2021 (2.20%) and 2024 (2.30%). The rate for patients with both conditions peaked in 2023 at 16.90%.
Prevalence of Obstetric Hemorrhage or Preeclampsia Among Hawai‘i Medicaid Maternal Population from 2016 to 2024
Values in the same row and subtable not sharing the same superscript are significantly different at p <0.05 in the two-sided test of equality for column proportions.
When stratifying the patient population by clinical condition, patients diagnosed with both obstetric hemorrhage and preeclampsia experienced the highest SMM rates, peaking at 49.30% in 2021 (Table 3). Despite fluctuations in rates of hemorrhage and preeclampsia among those with SMM, those with neither (hemorrhage or preeclampsia) accounted for the highest proportion (41.8%) in 2024.
Severe Maternal Morbidity Rates Among Hawai‘i Medicaid Maternal Population with Obstetric Hemorrhage or Preeclampsia from 2016 to 2024
—Indicates the cell size is <11. Bolded values indicate column proportions are statistically different, p < .05.
SMM rates showed racial/ethnic disparities from 2016 to 2024 (Table 4, Fig. 1), but fluctuated annually. Pacific Islanders and Native Hawaiians peaked highest at 5%; White recipients had the lowest average rate at 2.9% (confidence interval: 2.5–3.3%), significantly lower than the highest groups (p < 0.05). No consistent reduction in disparities occurred during the study period.
Severe Maternal Morbidity Among Hawai‘i Medicaid Maternal Population from 2016 to 2024 by Race/Ethnicity
Values in the same row and subtable not sharing the same superscript are significantly different at p <.05 in the two-sided test of equality for column proportions.
—Indicates the cell size is <11. Bolded values indicate column proportions are statistically different, p < .05.
Discussion
PQC participation was incentivized through reporting obligations in the state’s P4P structure, 24 resulting in participation by all birthing facilities. Participants widely recognized the collaborative’s P4P structure as effective for initiating QI within Medicaid facilities. However, quantitative data showed that SMM rates rose during parts of implementation before stabilizing, indicating participation and implementation alone did not correspond with immediate improvements in outcome measures. While the California Maternal Quality Care Collaborative achieved a 20% SMM reduction, it did not detail the effects on the Medicaid population. 25
Although participants shared the collaborative’s vision, effective implementation of targeted interventions for vulnerable groups was hindered by a lack of alignment among hospital-level P4P incentives, leadership, and financial support for implementation staff. This finding aligns with Schneider et al. 15 who noted the need for strong leadership engagement and the dedication of resources to drive equitable QI. PQC participants identified the need to address factors beyond the hospital setting, specifically those serving rural populations, and those lacking system support faced more barriers due to limited resources, such as referral options and shortages of key personnel (e.g., social workers/case managers).
These contextual barriers may explain why racial and ethnic disparities in SMM persisted over time. Quantitative findings showed no significant change in SMM rate over time across racial and ethnic groups within the Medicaid population. In contrast, national data show a 56% increase in SMM rates among non-Hispanic Asian/Pacific Islander populations from 2016 to 2021. 9 In Hawai‘i, Pacific Islanders and Native Hawaiians experienced some of the highest SMM rates, peaking at approximately 5% for each. While a subgoal of the Illinois PQC was to reduce outcome disparities, 15 Hawai‘i’s PQC did not require implementation of dedicated, race- or ethnicity-specific disparity-reduction initiatives during the study period. Accordingly, the persistence of disparities was not unexpected. Given that disparities remain concentrated in Medicaid-serving populations, PQCs may offer infrastructure to support equitable improvement efforts if paired with targeted resources and interventions.
Limitations
There was a risk of sample bias in the qualitative component of this study, as interviews only captured the perspectives of implementation leads rather than the full range of clinical staff involved in the PQC. Although implementation themes were identified, the perspectives shared may not represent all participating facilities. This study was observational and did not evaluate causal relationships between AIM bundle implementation and changes in SMM outcomes. Because hospitals were permitted to self-direct implementation activities, the specific actions taken varied across sites and were not systematically documented for comparative analysis, limiting cross-hospital comparisons. Observed outcome trends may have been influenced by unrelated confounding factors, such as the state’s 2022 implementation of the All Patient Refined Diagnosis Related Group system, 26 as well as the COVID-19 pandemic, which may have disrupted hospital operations and affected SMM rates in 2020 and 2021. Findings from this study apply to Hawai‘i, but may not be generalized to other states. Each PQC operates in unique ways, fitting to the local context. Results from this study may be transferable to other states intending to support the organization of collaboratives that drive equitable QI.
Health equity implications
Targeted recommendations that address equitable QI efforts through quality collaboratives include support for hospitals in tailoring implementation approaches to their most pressing needs or gaps. Designing state’s P4P programs to support collaborative work while providing resources to address disparities beyond the hospital setting may facilitate targeted interventions. Collaborative peer-to-peer learning supports, particularly, smaller rural communities and facilities serving vulnerable populations that do not have larger hospital systems to rely on. Existing state-based community organizations offer the opportunity to drive and connect leadership with practical and tailored innovations, freeing up resources to increase staffing that supports connecting vulnerable patients to social and community resources. 27
Additional recommendations by the AHRQ theme are provided in Supplementary Data S4.
Conclusion
The state P4P structure incentivized participation in statewide AIM bundle implementation and hospital-specific gap analyses, strengthening QI infrastructure among Medicaid-serving facilities. However, targeted disparity-reduction initiatives were not systematically targeted during the collaborative period, and racial and ethnic disparities in SMM remained present. These findings underscore the importance of PQC’s pairing implementation infrastructure with dedicated equity-focused strategies to achieve measurable improvements in disparities.
Ethics Statement
This project was a QI initiative and was not deemed to be research as defined by the U.S. federal regulation at 45 CFR 46.102(d). 28 The primary goal of this project was to evaluate changes intended for QI within Hawai‘i’s local institutional setting. Therefore, this activity did not require review by an Institutional Review Board under the Common Rule. All data were handled in accordance with the institution’s policies for QI activities to maintain patient privacy and confidentiality.
Authors’ Contributions
A.N.: Methodology, formal analysis, data curation, visualization, writing—original draft, and writing—review and editing. M.R.: Conceptualization, methodology, formal analysis, data curation, project administration, supervision, writing—original draft, and writing—review and editing. R.G.G.-G.: Conceptualization, methodology, formal analysis, project administration, supervision, writing—original draft, and writing—review and editing. F.Y.: Methodology, formal analysis, validation, and writing—review and editing. C.I.: Methodology, formal analysis, investigation, project administration, resources, and writing—review and editing. E.M.: Writing—review and editing. J.P.B.: Methodology, formal analysis, validation, supervision, funding acquisition, and writing—review and editing. Y.L.: Methodology, formal analysis, supervision, resources, and writing—review and editing. All authors have read and approved the final article.
Data Availability
This study used Hawai‘i’s Medicaid administrative data. The data were made available to the researchers by the Med-QUEST Division, Hawai‘i Department of Human Services, based on our role as evaluators. Data access was granted through a Business Associate Agreement between the University of Hawai‘i and the Hawai‘i Department of Human Services. Based on this agreement, the researchers are not authorized to reshare the data used in this study. Data requests may be made using 1,183 forms sent to the Department of Human Services Med-QUEST Division in Hawai‘i (
Footnotes
Author Disclosure Statement
No competing financial interests exist.
Funding Information
This work was supported by the Med-QUEST Division, Hawai‘i Department of Human Services, United States. The findings and conclusions in this article are those of the authors and do not necessarily represent Med-QUEST or members of the collaborative.
Supplemental Material
Supplemental Material
Supplemental Material
Supplemental Material
Abbreviations Used
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
