Hospital-Based Severe Maternal Morbidity Surveillance and Review Authors: Andreea A. Creanga, Carrie Wolfson, Elizabeth Stierman, Briana Kramer, Robert Atlas, Jeanne Sheffield, Irina Burd, Monica Jones, Elizabeth Zadzielski, Khyzer Aziz Published: 2026 ABSTRACT Background:Learning from promising maternal health interventions is limited by the lack of standardized outcome measures needed for their evaluation. We describe the development and utilization of Maryland’s severe maternal morbidity (SMM) surveillance by the Maryland Maternal Health Innovation (MDMOM) program.Methods:To ascertain the development, attributes, and data collected by SMM surveillance, we use 2020–2024 MDMOM program records and 2023 surveillance data. Hospital reports and MDMOM’s 2023 maternal health care provider survey data are used to document policy and practice changes attributable to participation in SMM surveillance. Data collected from MDMOM’s “Learning from Adverse Events in Maryland” trainees are analyzed to explore knowledge score changes (Wilcoxon signed-rank test) and training satisfaction.Results:SMM surveillance is a simple, flexible, stable, high-quality system for identifying areas of care in need of improvement. Before October 2024, when state legislation mandated all hospitals to participate, 24 of 32 hospitals (80% of the state’s annual deliveries) engaged in SMM surveillance. The data are used to assess SMM levels and contributors, disparities (eg, intensity of care practices and outcome differences between patient groups), and preventability. Knowledge scores improved after the training on SMM surveillance findings (P<0.001). Data can be used for quality improvement and to generate PCOR hypotheses.Conclusions:Maryland established hospital-based SMM surveillance, demonstrated its value, and gained legislative support for statewide scale-up. This model surveillance system can be replicated in other states, aiming to better understand and address SMM and SMM disparities. Adverse maternal and perinatal health indicators are disproportionately high in the United States compared with other high-income countries, making maternal and perinatal health priorities at the state and national levels.1–4 In 2022, the US maternal mortality (MM) ratio was 22.3 per 100,000 live births,5 3 times higher than in most other high-income countries.1 Severe maternal morbidity (SMM), defined as unexpected and potentially life-threatening events associated with labor and delivery, has more than doubled over the past 20 years.6 Marked and unacceptable racial-ethnic disparities persist in maternal health throughout the country, with rates of MM and SMM 2–3 times higher in Black than White women.3,5–10 The reasons behind these disparities are multifactorial and include socioeconomic, clinical, and structural factors in health care.7–11 MM and SMM are contributed by “traditional” causes (eg, obstetric hemorrhage, hypertensive disorders of pregnancy, sepsis) and, increasingly, by chronic medical conditions.12 In addition to having prolonged and higher cost hospital stays, patients with SMM are at higher risk of stillbirth, preterm delivery, and long-term consequences on physical and mental health.13–15 A large proportion of these adverse outcomes are preventable.8,10,16,17 However, widespread learning from interventions to improve maternal and perinatal health has been limited by the lack of standardized process and outcome measures needed for their evaluation.18 SMM is considerably more prevalent than MM but difficult to measure.17 In 2012, the Centers for Disease Control and Prevention (CDC) developed an algorithm that identifies SMM in administrative hospital records using International Classification of Diseases-Clinical Modification, 9th revision (ICD-9-CM) diagnosis and procedure codes for 25 indicators of organ-system dysfunction.19 Since then, the algorithm has been validated20 and revised to include only 21 indicators of SMM following the transition to ICD-10 for medical coding in October 2015.3 Subsequent revisions of the CDC-SMM ICD-10 algorithm have been proposed to simplify and improve SMM identification.3,6,18 Notably, it was recommended that the blood-transfusion indicator be excluded, given its lack of specificity to the number of units of blood products transfused, which alone may not indicate SMM.6,18 Unsurprisingly, SMM prevalence varies considerably based on the ICD-10 codes used for identification. For example, in 2019, the national SMM prevalence using the top-ranked 15 of 21 CDC-SMM indicators was 55 per 10,000 delivery hospitalizations, while the use of all 21 indicators yielded a 3 times higher prevalence.6 Federal agencies, their grantees, and other organizations monitor SMM, often using ICD-10 codes, but also showing an increased interest in quality indicators related to SMM events.21 For example, the Joint Commission established standards covering emergency preparedness, hemorrhage, and severe hypertension management to improve maternal health outcomes, requiring birthing hospitals to implement related safety protocols and review SMM events.22 Similarly, CMS’s “birthing-friendly” designation recognizes hospitals that meet specific quality standards.23 The CDC-SMM algorithm was not designed and is not suitable for use as a quality indicator to compare hospitals or regions without adjustment for patient-mix,18,21 and it captures both preventable and unpreventable SMM. To address these limitations, a modified version called “performance SMM” (pSMM) has been recently proposed to consider only conditions that are hospital-acquired with a method for patient-mix adjustment.21 Despite its conceptual and practical limitations, the pSMM measure is clinically more meaningful and likely more responsive to quality improvement (QI) activities than the unadjusted CDC-SMM measure. While ICD-based algorithms are useful for population-level monitoring, it is largely recognized that clinical reviews are best suited to guide the design of QI initiatives to eliminate preventable SMM and MM.24–26 The American College of Obstetricians and Gynecologists (ACOG) and the Society for Maternal-Fetal Medicine (SMFM) recommend and provide guidance for birthing facilities to routinely and systematically identify and review SMM and MM events using a case definition that addresses limitations with ICD-based algorithms: ICU admission and/or transfusion of 4+ units of blood products during pregnancy or within 42-day postpartum.24,25 To gain a deep understanding of maternity care improvement opportunities in Maryland, the Maternal Health Innovation Program (MDMOM)27,28 used the ACOG/SMFM guidance and initiated hospital-based SMM Surveillance and Review (hereafter SMM surveillance) in July 2020.10,29 The MDMOM program implements additional activities aiming to improve maternal health outcomes in Maryland, including workforce trainings, a telehealth initiative to reduce severe hypertension in pregnancy, an evidence-based intervention (ie, mothers and babies) to prevent postpartum depression, and provider surveys to evaluate the program.27,28 This article describes the development and utilization of hospital-based SMM surveillance in Maryland. METHODS We use multiple data sources: MDMOM program records; 2023 SMM surveillance data; hospital reports of policy and practice changes implemented as a result of participation in SMM surveillance; MDMOM’s 2023 maternal health care provider survey; and data collected from MDMOM’s “Learning from Adverse Events in Maryland” trainees between March 2022 and August 2023. MDMOM program records are used to develop an implementation and evaluation diagram for SMM surveillance that depicts hospitals’ participation over time; the types of activities implemented during the planning, pilot, and expansion phases; and the logic model used for program evaluation. We describe the attributes of this surveillance using CDC’s surveillance system evaluation framework,30 and discuss key evaluation indicators and the available data sources used for their derivation. Our SMM surveillance database has been described in detail elsewhere.10,27 Briefly, the MDMOM program maintains a standardized REDCap database used by trained data abstractors in participating hospitals to enter data from all SMM events that meet the ACOG/SMFM-recommended case definition: ICU admission and/or transfusion of 4+ units of blood products during pregnancy or within 42-day postpartum.10,27,29 The database includes structured data elements capturing the type and timing of SMM; patient sociodemographic characteristics and medical history; delivery information; treatment and management; follow-up after the SMM event; pregnancy, fetal, and newborn outcomes. This information and a case narrative with a timeline of key events that led to the SMM event are completed by data abstractors before hospital review meetings. During the meetings, review committees discuss the underlying and contributing causes of SMM and conduct a preventability assessment to capture opportunities to alter the SMM outcome and make recommendations for each SMM event reviewed. This information is subsequently entered in the database in open-ended text fields. We analyzed data from the 279 SMM events that were identified, reviewed, and analyzed in 2023 across all participating hospitals using frequency analysis and cross-tabulations to characterize SMM, examine SMM disparities, and assess preventability. Denominators for hospital SMM rate calculations are based on the number of births in participating hospitals from the Maryland Department of Health. Between November 2023 and February 2024, MDMOM conducted a cross-sectional, midline survey of maternal health care professionals in Maryland’s birthing hospitals, using the same methodology as the baseline survey conducted between June and August 2021 and described in detail elsewhere.9 Obstetric unit leaders distributed the web-based survey to all maternal health care professionals in their hospital; respondents received a $25 gift card for survey completion (response rate 57.4%). The survey assessed satisfaction with the SMM surveillance process, whether SMM surveillance findings were discussed, and awareness of policy or practice changes resulting from participation in SMM surveillance. These data were tabulated and stratified by key respondent and hospital characteristics. To encourage timely identification and reviews of SMM events, in October 2022, we instituted modest quarterly compensation to hospitals that enter data within 3 months of event occurrence and/or implement policy or practice changes following SMM reviews. For this article, we conducted a thematic analysis of policy or practice changes implemented and reported by hospitals between January 2023 and December 2024. MDMOM’s Hospital Initiative31 developed and offered a series of trainings to address maternal health inequities in birthing hospitals. One of these trainings, titled “Learning from Adverse Maternal Events in Maryland,” was a 1-hour, CME/CEU-approved training that provided an overview of national and state MM and SMM trends and presented representative SMM events from SMM surveillance. Hospitals offered this online (asynchronous), as a grand rounds presentation, or by both modalities. Between March and October 2022, the asynchronous version was offered on a custom learning management system (LMS) on the MDMOM website, and between October 2022 and August 2023 on a lower-cost, generic LMS maintained by our institution. This second LMS allowed inclusion of pre- and post-tests with 8 knowledge questions covering maternal health epidemiology and clinical practice topics. Participants could receive CME credit if they answered 6 of 8 (75%) post-training questions correctly. We used the training registration data to describe our trainees, and the information from pre- and post-training tests to examine knowledge changes on key topics covered by this training, comparing the mean test scores (standardized to a 0–100 scale) using the Wilcoxon signed-rank test. The training also included an optional, anonymous, 5-question post-training survey, results of which were tabulated to assess training satisfaction on relevance, content organization, delivery effectiveness, planned use of the material, and willingness to recommend the training to colleagues. All statistical analyses noted above were performed using Stata 18. The SMM surveillance program was deemed public health surveillance and not research involving human subjects; thus, it was exempt from review by our Institutional Review Board. The MDMOM program data, provider survey data, and training evaluation data used for analyses were collected as part of the larger MDMOM program evaluation, for which we obtained IRB approval (IRB00013694). RESULTS Development of Severe Maternal Morbidity Surveillance in Maryland The overarching goal of SMM surveillance is to identify SMM events as close to real time as possible and act upon review findings. The program grew continuously over time (Fig. 1). During a 6-month planning period, we consulted with experts with experience conducting SMM surveillance; developed the data abstraction form (Supplemental File 1, Supplemental Digital Content 1, http://links.lww.com/MLR/D144), data abstraction manual, and preventability assessment guideline (Supplemental File 2, Supplemental Digital Content 2, http://links.lww.com/MLR/D145) using ACOG/SMFM guidance and the abstraction form employed for hospital-based SMM reviews in Illinois; and obtained IRB approval to conduct this surveillance work. With support from state legislation (HB-837/2020),32 we approached leaders of Obstetrics and Gynecology Departments in 6 hospitals chosen for the pilot program to represent various care delivery models, levels of care, and geography. Following data abstractor training, data collection started in all 6 hospitals on July 1, 2020.10 We evaluated the pilot SMM surveillance 15 months after its start and revised the data abstraction form and database in line with the evaluation findings. In January 2022 and annually thereafter, we invited nonparticipating birthing hospitals in the state to participate in SMM surveillance. By October 2024, when new state legislation (HB-1051/2024)33 required all hospitals in the state to participate in MDMOM’s SMM surveillance, 24 of 32 hospitals were either actively contributing their data or negotiating terms on the standard service agreement that offers additional data confidentiality protections and allows hospital payments for participation. Notably, before the state mandate, both Level IV and 8 of 11 Level III hospitals were fully engaged in SMM surveillance; Level I–II hospitals had a small number of events that met the case definition for SMM surveillance and were less engaged with the program. Overall, before hospital participation was required by law, SMM surveillance covered about 80% of annual deliveries in Maryland. Severe MM surveillance and review in Maryland: program implementation and evaluation activities and timeline.​​​​ 1HB-837/2020, Public Health—Maternal Mortality and Morbidity—Implicit Bias Training and Study. 2HB-1051/2024, Maternal Health—Assessments, Referrals, and Reporting (Maryland Maternal Health Act of 2024). ACOG indicates American College of Obstetricians and Gynecologists; IRB, institutional review board; SMFM, Society for Maternal-Fetal Medicine; SMM, severe maternal morbidity. MDMOM offers training for data abstractors and regular learning and sharing sessions for data abstractors and members of hospital review committees. The surveillance data are checked on a continuous basis for missing or improbable data values and to assess the need for revisions. Reports and communications around data findings occur at least twice annually with each hospital, through an annual report and an annual data brief. A final SMM surveillance evaluation is planned 3 months before the end of the MDMOM program (July 2028). The program evaluation will aim to inform state efforts to maintain the SMM surveillance program either in its current form or adapted based on need. Our logic model allows assessment of outputs at the MDMOM program, hospital, and state levels, and ascertainment of changes in SMM prevalence rates and disparities in SMM over the course of 8 years (Fig. 1). Attributes of Severe Maternal Morbidity Surveillance in Maryland The SMM surveillance system is relatively simple, especially when engaging the same data abstractors and benefiting from stable leadership at the hospital level (Table 1). The system has proven to be flexible and to allow for timely changes, while preserving high acceptability and stability. Maintaining data quality starts with the data abstractors and ends with MDMOM scientists conducting regular data checks and providing individual and group feedback to hospitals. Benefiting from high positive predictive value,25 actionable data points, and the October 2024 legislative mandate for all hospitals to participate, the perceived and actual usefulness of SMM surveillance is high. Key evaluation indicators for each of the surveillance system’s attributes have been tested during the pilot evaluation, and the MDMOM surveys with health care providers offer additional assessment opportunities.10,28,29 TABLE 1SMM Surveillance and Review Program: Implementation Attributes and Evaluation Indicators and Data SourcesImplementationEvaluationAttributesDescriptionIndicatorsSource* SimplicitySMM events identified by trained abstractors using multiple data sources (eg, EHR; automated maternal complications reports, blood bank reports; staff alerts) or through coordination with case identification for other perinatal adverse events reviewedDetailed information on patient demographics, medical and obstetric history, and SMM abstracted and entered in REDCap database maintained by MDMOMHospital and group learning opportunities offered by MDMOMReal-time data entry tracking allows for continuous data checking and cleaning, export and analysis by MDMOMMethod used to identify SMM events Amount and type of data collected Ease of abstracting and entering data Training and learning requirements Ease of managing, cleaning, and analyzing data FlexibilityAccommodates EHR and surveillance case definition changesData entry system allows for changes and updates as neededHospital-based review committees typically include lead obstetrician, nurse or nurse-midwife, QI specialist(s), and data abstractor(s), with composition, size, and frequency of review meetings tailored to individual hospital needsAdaptability of the data entry system Composition of the abstraction team and review committees at the hospital level Tracking emerging public health threats AcceptabilityHospitals participate in case abstraction and regular check-in meetings, increasingly so with state legislation mandates in placeIncreasing recognition of the value of SMM surveillance for QI, and strong physician and nursing leadership engagementHospital engagement Perceived value of SMM surveillance and review TimelinessData entry required within 1 mo of the event occurring, but some hospitals take up to 3 mo due to delays in SMM event identification reports and limited staff capacityMonthly to quarterly review meetings, depending on case loadComplete data entry within days of review meetingsData analyses at the hospital- and state-level biannually, with data briefs, reports, and publications disseminated annuallyTime between the SMM event and entry in the data system Time between SMM event and hospital review meetings Time between hospital reviews and data entry completion for reviewed cases Time between completion of data entry and data analysis StabilityThe system experienced no outages and limited staff turnoverTime when the system is fully operating Data qualityData are from structured data elements available in EHR, yet missing for certain patient demographic (eg, education, nativity), obstetric history (eg, prenatal care)Unstructured event summary and preventability assessmentAbstractors trained to verify data in external sources (eg, birth certificates, CRISP, CareEverywhere)Monitoring and data quality checks incorporated in data entry system and completed by MDMOMCompleteness Data validity Predictive value positiveSMM events identified are determined to be true SMM cases during hospital reviewsProportion of reported SMM events that are true severe morbidity RepresentativenessPilot program with 6 hospitals expanded gradually to 27 hospitals; became statewide following state legislation mandateCause-specific morbidity UsefulnessThe systematic review of SMM events leads to actionable recommendations at hospital- and state-levelHospitals develop and implement policy and practice changes following SMM reviewsActual and perceived usefulness of the SMM surveillance data * Data source: program implementation monitoring, qualitative interviews, MDMOM investigators, SMM surveillance data, and provider surveys.CRISP indicates Chesapeake Regional Information System for our Patients in Maryland; EHR, electronic health records; ICD, International Classification of Diseases; QI, quality improvement; SMM, severe maternal morbidity. SMM Surveillance and Review Program: Implementation Attributes and Evaluation Indicators and Data Sources Data source: program implementation monitoring, qualitative interviews, MDMOM investigators, SMM surveillance data, and provider surveys. CRISP indicates Chesapeake Regional Information System for our Patients in Maryland; EHR, electronic health records; ICD, International Classification of Diseases; QI, quality improvement; SMM, severe maternal morbidity. Availability of Severe Maternal Morbidity Surveillance Data At the hospital and state levels, SMM surveillance data can be used to assess levels and contributors (eg, in 2023, the SMM rate in participating hospitals was 105.7 per 10,000 deliveries; 55.5% of SMM cases were contributed by obstetric hemorrhage; of patients with SMM, 38.9% had prepregnancy obesity, 30.4% had an emergency cesarean, and 32.9% delivered preterm); health disparities (eg, in 2023, in participating hospitals, there was a 2:1 ratio in SMM rate between non-Hispanic Black and White patients; 6.5% of non-Hispanic White vs. 9.2% of non-Hispanic Black patients had an exploratory laparotomy postpartum); and SMM preventability (eg, in 2023, in participating hospitals, 27.6% of patients with SMM were appropriately transferred to receive a higher level of care; 59.1% of SMM reviews resulted in specific QI recommendations). Figure 2 shows additional specific 2023 SMM indicators. Trends in these SMM indicators can also be assessed, and hospitals and the state can document progress toward meeting certain SMM targets. Examples of key indicators from Maryland’s Severe MM surveillance and review: Maryland, January–December 2023. CVD indicates cardiovascular disease; QI, quality improvement; SMM, severe maternal morbidity. Utilization of Severe Maternal Morbidity Surveillance Data Sharing and Dissemination of Findings With Key Stakeholders MDMOM shares data through individual hospital reports that summarize SMM events in each hospital over the course of the previous calendar year. Structured data are presented in tables, while unstructured data are organized using content analysis techniques and coded according to the 5Rs framework34 proposed by ACOG’s Alliance for Innovation on Maternal Health and commonly used for QI initiatives: readiness; recognition; response; reporting and systems learning; respectful, equitable, and supportive care. Using pooled, deidentified data from all participating hospitals, we develop annual comprehensive data briefs27 with key information from SMM reviews, including results from preventability analyses and recommendations. These are shared with all 32 birthing hospitals in the state and additional stakeholders, including the Maryland Department of Health, the Maternal Health Improvement Task Force, and state legislators. Developing Learning Opportunities for Health Care Providers The “Learning from Adverse Maternal Events in Maryland” training was launched in March 2022. Over an 18-month implementation period, 823 perinatal health care providers across the 32 birthing hospitals in the state received the training (Table 2). Most (73.4%) took the training online, and 95.4% received CME credits. Of all trainees, 73.5% were nurses; 80% worked at level III or IV hospitals; 8% and 21% worked in hospitals with <1000 and >3000 annual deliveries, respectively. Pre- and post-training knowledge tests were completed by the 274 trainees who used our institution’s LMS. Among them, the mean knowledge score increased significantly between pretraining (54) and post-training (90; P<0.001). Pretraining, the largest knowledge deficits were related to substance use disorder treatment recommendations and recognition of overdose as the main cause of pregnancy-associated mortality in the state (Supplemental Table 1, Supplemental Digital Content 3, http://links.lww.com/MLR/D146). Notably, >97% of trainees answered a question related to racial/ethnic disparities in maternal health outcomes correctly pre- and post-training. Trainees were generally satisfied with the training (Supplemental Table 2, Supplemental Digital Content 4, http://links.lww.com/MLR/D147)—they agreed or strongly agreed that the topics covered were relevant to their work (96%), the content was well organized (92%) and delivered (85%), reporting that they will use the information obtained (92%) and recommend the training to colleagues (86%). TABLE 2Learning From Adverse Maternal Events in Maryland: Training and Trainee Characteristics and Knowledge Changes (n=823)Characteristicsn (%)Training modality (timing) mdmom.org (March 2022–October 2022)331 (40.2) JHU-CoursePlus (October 2022–August 2023)274 (33.2) Grand rounds (March 2022–August 2023)218 (26.5)Completion date March–June 202281 (9.8) July–September 2022146 (17.7) October–December 2022274 (33.3) January–March 2023156 (19.0) April–June 2023131 (15.9) July–September 20235 (0.6) October–December 202330 (3.6)Trainees’ professional role Physician124 (15.1) Nurse605 (73.5) Midwife or other APP* 44 (5.3) Technician† 25 (3.0) Other/unknown‡ 25 (3.0)Trainees’ hospital maternity care level I/II162 (19.7) III/IV661 (80.3)Trainees’ hospital annual delivery volume <100062 (7.5) 1000–3000589 (71.6) >3000172 (20.9)Trainees obtained a CME certificate Yes785 (95.4) No38 (0.5)Knowledge mean score: pretest (54) and post-test (90).Data available only for trainees who took the training on the JHU-CoursePlus learning management platform (n=274). * APP (advanced practice provider) includes nurse practitioner and physician assistant. † Technicians include certified nursing assistants, patient care technicians, and operating room technicians. ‡ Other/unknown includes administrator, genetic counselor, and role not stated.z-score (P-value)=−13.856 (P<0.0001). Learning From Adverse Maternal Events in Maryland: Training and Trainee Characteristics and Knowledge Changes (n=823) Knowledge mean score: pretest (54) and post-test (90). Data available only for trainees who took the training on the JHU-CoursePlus learning management platform (n=274). APP (advanced practice provider) includes nurse practitioner and physician assistant. Technicians include certified nursing assistants, patient care technicians, and operating room technicians. Other/unknown includes administrator, genetic counselor, and role not stated. z-score (P-value)=−13.856 (P<0.0001). Policy and Practice Changes in Hospitals Of 2202 maternal health care professionals used by or affiliated with hospitals participating in SMM surveillance who responded to the midline survey in November 2023, 850 (38.6%) were aware of their hospital’s participation in SMM surveillance, with greater awareness reported by physicians (64.4%) and advanced practice nurses (51.1%) compared with RNs (36.2%) and CNAs or medical technicians (33.7%; data not shown). Among those aware of their hospital’s participation, 41.4% reviewed or discussed findings from SMM surveillance within their unit, and 46.8% were aware of policy or practice changes that resulted from participation in SMM surveillance. The most frequently reported changes were related to staff education (13.4%), availability of management protocols (10.5%), timing of treatment (10.0%), screening (9.5%), patient monitoring (9.4%), documentation (7.0%), admission triage (6.4%), and laboratory testing (5.2%). Ten hospitals submitted short descriptions of specific policy or practice changes attributable to their participation in SMM surveillance at least once since July 2023 (Table 3). These changes were organized into 6 themes: cross-disciplinary collaboration and communication enhancements (5 examples); education, training, and competency development (3 examples); process and policy updates for patient safety and QI (2 examples); data monitoring and early detection systems improvements (3 examples); clinical intervention protocols for high-risk patients (4 examples); and scheduling and resource allocation for patient-centered care (3 examples). Most of these changes were implemented to address the main causes of SMM, specifically obstetric hemorrhage (9 change examples), severe hypertension (3 examples), and sepsis (2 examples). Changes in the recognition and response domains of the 5Rs framework were most prevalent. TABLE 3Implemented Changes in Hospitals Participating in Severe Maternal Morbidity Surveillance and Review in Maryland(1) Cross-disciplinary collaboration and communication enhancementsCollaboration with CVIR physicians to expedite care for patients with PPH needing specialized interventionsQuality Management Collaborative that engages multidisciplinary teams (eg nursing, anesthesia, blood bank) to improve cesarean and severe maternal morbidity ratesEnhanced communication pathways for rapid response, involving ICU and L&D management for critically complex patientsRe-education on escalation protocols for clinical care concerns across departmentsExpanded SMM review committee with multidisciplinary membership to address complex cases(2) Education, training, and competency developmentResident and attending training in FAST exams to improve recognition of intra-abdominal bleedingHands-on training with hysterectomy instrument sets for OB technicians during cesarean sections to improve preparedness for emergenciesComprehensive simulation training for insulin administration on the Mother-Baby unit, with enhanced staff education on insulin drip errors and protocol adherence for IV line management to minimize errors(3) Process and policy updates for patient safety and quality improvementUpdated policies for OB-Stat code activation to empower early intervention for excessive bleedingEnhanced policy for transfers between L&D and Mother-Baby unit to avoid unnecessary movements and manage hemorrhage risk more effectively(4) Data monitoring and early detection systems improvementsEnhanced hypertension protocol in triage requiring Q1h vitals and Q15m BP monitoring for high BPQuantitative blood loss monitoring in ICU-transferred OB patientsEHR integration of scoring for early detection of sepsis(5) Clinical intervention protocols for high-risk patientsInstitution of the Jada System for PPH managementUse of a standardized antibiotic prophylaxis for cesarean sections and added vaginal preps to reduce infection risksImplementation of a “500 Moment” to prompt early PPH interventions and decision-makingStandardization of procedures for pulse oximetry, IV iron access, insulin administration, and management of placenta accreta spectrum(6) Scheduling and resource allocation for patient-centered careOn-call gynecologic oncologist addition to the care team for immediate availability in relevant casesDaily weight monitoring added to order sets for fluid management in preeclampsiaIOL request screening by L&D nurses to prepare for high-risk cases, reducing unnecessary cesarean sections and improving patient flowCVIR indicates cardiovascular and interventional radiology; EHR, electronic health records; FAST, focused assessment with sonography in trauma; ICU, intensive care unit; IOL, intraocular lens; IV, intravenous; L&D, labor and delivery; OB, obstetrics; PPH, postpartum hemorrhage. Implemented Changes in Hospitals Participating in Severe Maternal Morbidity Surveillance and Review in Maryland CVIR indicates cardiovascular and interventional radiology; EHR, electronic health records; FAST, focused assessment with sonography in trauma; ICU, intensive care unit; IOL, intraocular lens; IV, intravenous; L&D, labor and delivery; OB, obstetrics; PPH, postpartum hemorrhage. Development of Patient-Centered Outcome Research Hypotheses SMM surveillance data include the clinical nuances needed to generate new research hypotheses.10,35 For example, examining blood loss and comparing outcomes with intervention after a specific blood loss volume (eg, 500 mL), depending on medical history and the rate of blood loss. The recommendations from SMM surveillance can serve as a starting point for identifying specific research themes beyond comparison of treatment and management alternatives, such as shared decision-making, care coordination, and respect for cultural and personal beliefs. DISCUSSION Maryland has a Systematic and Robust Severe Maternal Morbidity Surveillance System Given its demonstrated value, it gained state legislative support requiring all birthing hospitals in the state to participate and contribute to shared, statewide learning on SMM events. Key facilitators of implementation were the buy-in from leaders of Obstetrics and Gynecology Departments, the simplicity and flexibility of the REDCap database and overall surveillance system, and the dedication of data abstractors and MDMOM scientists to collecting, protecting, and sharing deidentified, high-quality information. Barriers to program implementation included the timing of program implementation during the COVID-19 pandemic, when there were significant competing priorities on hospitals and staff, the high staff turnover during and in the first year after the pandemic, which included data abstractors in several hospitals, and the limited involvement by physicians in data collection and review in some hospitals. Despite our efforts to engage all hospitals, achieving statewide participation would have taken much longer without legislation, for which we advocated heavily. We will now be able to provide SMM rates at the state level and stratified by key sociodemographic and medical characteristics of patients. As the number of SMM events reviewed increases and with 4 years of data available for some hospitals at the end of 2024, we will be able to demonstrate that a reduction in SMM is possible with changes in key health system factors. Case studies will be developed for hospitals that achieve significant reductions in SMM and narrow disparities by race and ethnicity. This article shows that SMM surveillance is a powerful means for identifying and addressing areas of maternity care in need of improvement, including health disparities, and ways to ensure all pregnant and postpartum patients receive the care they need. SMM surveillance promotes health equity in maternity care by documenting differences in care processes and outcomes; translating findings into QI, policy, and practice changes; and supporting professional development activities. While limited to data from electronic health records and discussions during hospital review committee meetings, the planned addition of a new module to the database will enhance these data’s ability to highlight health disparities. This module is under development and is based on in-depth interviews with patients who experienced SMM in Maryland in the past 2 years. These interviews are underway and are not linked with the SMM surveillance. We documented a wide variety of ways SMM surveillance data are used, including a range of interventions and QI initiatives in participating hospitals. However, our provider survey shows that sharing of learnings from SMM surveillance with unit staff can be improved, which, in turn, will provide the rationale for additional practice changes to be implemented in the future. Many birthing hospitals in Maryland have a research mission to advance medical knowledge and improve patient care. PCOR research often uses clinical outcomes assessments, and, therefore, SMM surveillance data can be used to generate PCOR hypotheses to compare different medical treatments and interventions to help patients and families make informed health care decisions. In conclusion, Maryland’s SMM surveillance provides information that no other data source can provide. This model surveillance system can be replicated in other states, aiming to better understand and address SMM. Supplementary Material