Cesarean Delivery Rates Associated With a State Quality Collaborative in Maryland

Abstract

Objective:

To assess the extent to which hospitals participating in Maryland’s perinatal quality collaborative (MDPQC) to reduce primary cesarean deliveries adopted policy and practice changes and the association of this adoption with state-level cesarean delivery rates.

Methods:

This prospective evaluation of the MDPQC includes 31 (97%) of the birthing hospitals in the state, which all voluntarily participated in the 30-month collaborative from June 2016 to December 2018. Hospital teams agreed to implement practices from the “Safe Reduction of Primary Cesarean Births” patient safety bundle, developed by the Council on Patient Safety in Women’s Health Care. Each hospital’s implementation of practices in the bundle was measured through surveys of team leaders at 12 and 30 months. Half-yearly cesarean delivery rates were calculated from aggregate birth certificate data for each hospital and differences in rates between the 6 months prior to the collaborative (baseline) and the 6 months afterwards (endline) were tested for statistical significance.

Results:

Among the 26 bundle practices that were assessed, participating hospitals reported having a median of 7 practices (range: 0–23) already in place prior to the collaborative and implementing a median of 4 (range: 0–17) new practices during the collaborative. Across the collaborative, the cesarean delivery rates decreased from 28.5% to 26.9% (p=0.011) for all nulliparous term singleton vertex births and from 36.1% to 31.3% (p<0.001) for nulliparous, term, singleton, vertex inductions. Five hospitals had a statistically significant decrease in nulliparous, term, singleton, vertex cesarean delivery rates and four had a significant increase. Nulliparous, term, singleton, vertex cesarean delivery rates were significantly lower across hospitals that implemented more practices in the “Response” domain of the bundle.

Conclusions:

The MDPQC was associated with a statewide reduction in cesarean delivery rates for nulliparous, term, singleton, vertex births.

Precis

The Maryland perinatal quality collaborative was associated with reductions in cesarean delivery for nulliparous, term, singleton, vertex births

Introduction

Although cesarean delivery can be lifesaving, evidence suggests that there is no benefit to maternal health when national cesarean rates are above 20 per 100 live births.¹ In the United States, the cesarean delivery rate underwent what has been characterized as an “epidemic” increase² from 20.7% in 1996 to a peak of 32.9% in 2006,³ with only slight decrease to 31.7% by 2019.⁴ The Department of Health and Human Services,⁵ the Joint Commission,² the American College of Obstetrics and Gynecologists (ACOG), and the Society for Maternal-Fetal Medicine (SMFM)⁶ have all prioritized reducing cesarean delivery among nulliparous term singleton vertex pregnancies. To this end, the Council on Patient Safety in Women’s Health Care (the Council) compiled evidence-based practices in the obstetric patient safety bundle, “Safe Reduction of Primary Cesarean Births” (the cesarean bundle).⁷

California is the first state to report results of a quality collaborative to reduce primary cesarean rates. Through three cohorts between 2016 and 2018, California’s collaborative was associated with a 3.2% reduction in the nulliparous, term, singleton, vertex cesarean delivery rate.⁸ A corresponding assessment of the safety of the collaborative reported no adverse effects on maternal or neonatal outcomes.⁹ It is not yet known whether other state perinatal collaboratives will have similar success when implementing the cesarean bundle. Birthing hospitals in Maryland undertook a similar quality collaborative to implement the cesarean bundle in 2016, when the state’s nulliparous, term, singleton, vertex cesarean rate was 28.5%, 3 percentage points above the national average.¹⁰ The primary aim of this study was to evaluate the effects of Maryland’s collaborative on state-level nulliparous, term, singleton, vertex cesarean rates. The secondary aims were to assess the extent to which hospitals participating in Maryland’s collaborative adopted policy and practice changes in the cesarean bundle, and whether adoption of more practices was associated with lower cesarean delivery rates.

Methods

Maryland has 32 birthing hospitals where about 70,000 deliveries occur every year.⁴ The Maryland Perinatal Quality Improvement Collaborative (MDPQC) was initiated in 2009 and completed one prior collaborative to reduce early elective deliveries between 2009 and 2013. In 2016, under the coordination of the Maryland Patient Safety Center, the MDPQC recruited hospitals to voluntarily participate in a collaborative to reduce cesarean delivery among nulliparous women with term, singleton, vertex fetuses. The MDPQC selected the Council’s cesarean bundle as the focus of the collaborative. A letter from the Governor was sent to hospital administrators encouraging their participation, and 31 hospitals signed a participation agreement.

The MDPQC followed an adapted “Breakthrough Series” model for the cesarean collaborative (Appendix 1, available online at http://links.lww.com/xxx). As a requirement of participation, hospital teams sent at least one team member to each collaborative event. Activities included a June 2016 half-day, in-person kick off meeting for two to three representatives from each hospital to familiarize them with the cesarean bundle and the requirements of participation, followed by conference calls that occurred every month in the first year and every two months in the second year. Additional in-person meetings for all hospital teams took place at 12 months and at the end of the collaborative (November 2018); nine webinars on related clinical topics were presented throughout the 30-month period (Appendix 1, Webinar Topics [http://links.lww.com/xxx]). The collaborative director provided facilitation support to site teams through calls and visits when requested by the site team or when site participation lapsed.

ACOG’s Alliance for Innovation in Maternal Health (AIM) program, that supports implementation of all of the Council’s patient safety bundles nationally, provided a data portal for monitoring implementation and outcomes across the collaborative. Each hospital’s nulliparous, term, singleton, vertex and overall cesarean rates, as well as severe maternal morbidity rates, were entered on behalf of hospitals by the collaborative management and Maryland Department of Health epidemiologists. The portal displayed each hospital’s nulliparous, term, singleton, vertex and overall cesarean delivery rates benchmarked anonymously against other hospitals in the collaborative. Hospital teams also self-reported process and structure measures on a quarterly basis to the data portal. The required measures track a limited set of practices recommended in the bundle, including healthcare professional training, policies on freedom of movement in labor, protocols for responding to labor challenges, establishment of internal case reviews, and bundle compliance as assessed through case reviews (see Appendix 2, available online at http://links.lww.com/xxx, for process and structure measure definitions at the time of Maryland’s collaborative). In order to incentivize greater adoption of practice changes during the second year, the MDPQC also offered a banner of excellence to hospitals that completed 7 of 8 practice changes that could be documented by policy documents (Appendix 1, Panel 1 [http://links.lww.com/xxx]). Hospitals that qualified received a printed banner in a ceremony at their facility.

This uncontrolled before-and-after evaluation assesses both implementation and health outcomes¹² of the Maryland cesarean collaborative and was approved by the Institutional Review Board of the University of Maryland, Baltimore County. Implementation of practices from the cesarean bundle was assessed through web-based surveys in Qualtrics at 12 months and 30 months after the start of the collaborative. An email invitation to complete the survey was sent to the hospital-designated lead of collaborative activities at each participating hospital. The survey expanded on the limited set of process and structure measures in the portal, to include 26 discrete clinical practices, policies, and strategies recommended within the bundle’s four domains of “Readiness,” “Recognition and Prevention,” “Response,” and “Reporting and Systems Learning”¹³ (see Appendix 3, available online at http://links.lww.com/xxx, for the list of practices assessed). For each practice, the survey asked respondents to characterize their hospital’s implementation progress as follows: not started, in the planning phase, partially implemented, fully implemented during the collaborative, or fully implemented before the collaborative. The surveys were first distributed in September 2018, and hospital leads were asked to respond within two months, with up to five e-mail reminders sent to encourage participation. All respondents provided informed consent and were offered a $50 incentive for completion of the survey. Follow up queries were sent to survey respondents to correct missing or discrepant data.

We supplemented data on hospital characteristics from secondary sources including the American Hospital Association hospital profiles¹⁴ (for-profit status, teaching status, and annual number of births), the Maryland Hospital Community Benefit Report (percentage of patients covered by Medicaid),¹⁵ the Office of Rural Health Policy (rural status),¹⁶ and the Maryland Department of Health (ACOG/SMFM level of maternal care¹⁷). We also extracted data on training coverage of physicians, midwives, and nurses from the AIM data portal, which hospital teams were required to enter quarterly from the 2^nd quarter of 2016 through the 2^nd quarter of 2018. The indicator definition provided by the AIM program (Appendix 2, http://links.lww.com/xxx) asked hospital teams to estimate, in 10% increments, the proportion of nurses and physicians who received training on the latest ACOG/SMFM guidelines on labor management and support for intended vaginal birth in the past two years. The Maryland Vital Statistics Agency provided quarterly aggregate birth data by hospital, extracted from the medical portion of each birth certificate, for 30 hospitals that signed a supplemental data use agreement. These data included the number of births (denominator) and number of cesarean births (numerator) overall, for all nulliparous, term, singleton, vertex births, and nulliparous, term, singleton, vertex inductions. Of note, one Level II hospital responded to the implementation survey but declined to provide vital statistics and portal data for this evaluation.

We calculated descriptive statistics for hospitals and survey respondents, the implementation status of each of the 26 practices in the cesarean bundle, and the coverage of physician and nurse training. Half-year rates were calculated for three outcomes: cesarean deliveries among all births and among nulliparous, term, singleton, vertex births, and cesarean deliveries among nulliparous, term, singleton, vertex births that were induced. The first six-months of 2016 were considered the baseline period (prior to collaborative activities) and the first half of 2019 was considered the endline period. The statistical significance of differences in baseline and endline rates was assessed using the chi-squared test for state-level rates and the one-sided Fisher’s exact test for hospital-level rates, with p<0.05 as the threshold for statistical significance. To assess whether greater adoption of bundle practices was associated with nulliparous, term, singleton, vertex cesarean rates at endline, we calculated the relative risk of cesarean delivery for births at hospitals that had implemented the median number or more practices for each bundle domain, compared with births at hospitals that had implemented fewer practices. All analyses were conducted in Stata/SE, version 15.1 (StataCorp LLC).

Results

Each of the 31 hospitals in the Maryland cesarean collaborative completed the endline implementation survey (100% response rate; Table 1). Among the 26 bundle practices that we assessed, participating hospitals reported having a median of 7 practices (range: 0–23) already in place prior to the collaborative (Table 2). During the collaborative, hospitals implemented a median of 4 new practices (range: 0–17). Although some hospitals did not fully implement any bundle practice, all hospitals reported at least planning to implement or partially implementing one or more practices (Appendix 3, http://links.lww.com/xxx).

Table 1.

Characteristics of survey respondents and hospitals (n=31)

Characteristic	Number (%)
Respondents
Position
Nursing director or manager	19 (61)
Nurse, other	4 (13)
Safety or quality manager	4 (13)
Medical director	3 (10)
Midwife	1 (3)
Gender
Female	30 (97)
Male	1 (3)
Years of experience working in labor & delivery (median; range)	20 (5–45)
Years of experience at index hospital (median; range)*	11.5 (0–43)
Respondent was leading collaborative activities at hospital
Yes	24 (77)
No	7 (23)
Hospitals
Type
Not-for-profit	31 (100)
Level of maternity care
I	6 (19)
II	11 (35)
III/IV	14 (45)
Health system membership
Yes	22 (71)
No	9 (29)
Location
Rural county	2 (6)
Nonrural county	29 (94)
Teaching status†‡
Major	3 (10)
Minor	19 (63)
Nonteaching	8 (27)
Continuous availability of in-hospital anesthesia care
Yes	25 (81)
No	6 (19)
No. of obstetricians delivering at hospital‡§
1–9	10 (33)
10–19	7 (23)
20–29	7 (23)
30–59	4 (13)
60 or more	2 (7)
No. of nurses or midwives delivering at hospital
0	10 (32)
1–4	11 (35)
5–9	7 (23)
10 or more	3 (10)
Annual no. of births (median; range)	1312 (238–9597)
Proportion of patients covered by Medicaid (median; range)	23.5 (8.7–48.4)

Notes:

^*Information missing for 3 hospitals.

^†Major teaching hospitals are members of the Council of Teaching Hospitals; minor teaching hospitals have an accredited residency program but are not members.

^‡Information missing for 1 hospital.

^§Includes all attending, fellows, and resident physicians, as reported by hospitals.

Table 2.

Hospital implementation of cesarean bundle practices by domain (n=31)

Bundle Domain	No. practices assessed	Median No. Practices Fully Implemented Before Collaborative (Range)	Median No. Practices Fully Implemented During Collaborative (Range)	Median Total No. Practices Fully Implemented (Range)
R1: Readiness	5	0 (0–4)*	1 (0–5)	2 (0–5)
R2: Recognition and Prevention	8	3 (0–8)	1 (0–8)	5 (0–8)
R3: Response	7	3 (0–6)	0 (0–5)	4 (0–7)
R4: Reporting and Systems Learning	6	0 (0–6)	1 (0–5)	3 (0–6)
All Domains	26	7 (0–23)	4 (0–17)	13 (0–24)

Notes:

^*Only four of five readiness practices were assessed before the collaborative. The creation of a team to lead bundle implementation (R1.1 in Appendix 3 [http://links.lww.com/xxx]) was not assessed before the collaborative.

The practices that were implemented by at least two-thirds of hospitals by the end of the collaborative were standardized assessment and documentation of fetal heart rate (25, 81%), standardized induction scheduling (25, 81%), in-house maternity care provider or alternative coverage for response to labor problems (23, 74%), establishment of a provider team to lead the bundle implementation (23, 74%), comfort measures for labor dysfunction (21, 68%), and tracking provider-level cesarean rates (21, 68%). The clinical practices most commonly implemented during the collaborative were policies and protocols to encourage freedom of movement in labor (10, 32%), staff training on labor support (9, 29%), and use of standard criteria for diagnosis and treatment of labor dystocia, arrest disorders, and failed induction (9, 29%). Hospitals also commonly reported establishing case reviews (14, 45%) and starting to track provider-level cesarean rates (10, 32%) during the collaborative. The practices with the lowest overall adoption include implementing a policy to integrate doulas in the birth care team (3, 10%) and integrating new tools or guidelines in the electronic health record system (4, 13%).

Physician, midwife, and nurse training on labor and support techniques per ACOG/SMFM guidelines was a key process indicator tracked on the AIM portal. Although 23 hospitals reported complete or partial implementation of training, the training coverage was not high at all hospitals. Training coverage across the collaborative increased gradually over time and was slightly higher for nurses (Figure 1) than physicians and midwives (Figure 2). Seventeen (57%) hospitals reported 60% training coverage or higher for nurses by Q2 2018, and 16 (53%) reported that level for physicians and midwives (Appendix 4, available online at http://links.lww.com/xxx).

Figure 1.

Change over time in the cumulative proportion of nurses that completed education on the American College of Obstetricians and Gynecologists and the Society for Maternal‒Fetal Medicine labor management guidelines at hospitals participating in the collaborative* (n=30). *Includes 30 out of 31 hospitals participating in the collaborative; training data not available for one hospital.

Figure 2.

Change over time in the cumulative proportion of physicians and midwives that completed education on the American College of Obstetricians and Gynecologists and the Society for Maternal‒Fetal Medicine labor management guidelines at hospitals participating in the collaborative* (n=30). *Includes 30 out of 31 hospitals participating in the collaborative; training data not available for one hospital.

Figure 3 charts the collaborative-wide half-yearly cesarean rates for nulliparous, term, singleton, vertex births, nulliparous, term, singleton, vertex induced births, and all births from Q1-Q2 2016 through Q1-Q2 2019. The largest change in cesarean rates was seen for nulliparous, term, singleton, vertex induced births, which decreased by five percentage points from 36.1% to 31.3% (p<0.001; see Appendix 5, available online at http://links.lww.com/xxx). Cesarean rates for all nulliparous, term, singleton, vertex births decreased from 28.5% to 26.9% (p=0.011) and for all births from 33.6% to 32.7% (p=0.012).

Figure 3.

Six-month cesarean delivery rates across hospitals participating in the collaborative, 2016‒2019* (n=30). *Includes 30 out of 31 hospitals participating in the collaborative; vital statistics data not available for one hospital. NTSV, nulliparous, term, singleton, vertex.

The change in individual hospital cesarean delivery rates for nulliparous, term, singleton, vertex births and nulliparous, term, singleton, vertex inductions demonstrates that although rates decreased from baseline to endline for the collaborative as a whole, the results were not consistent between hospitals. For nulliparous, term, singleton, vertex births (Appendix 6, available online at http://links.lww.com/xxx), 17 hospitals had a decrease in cesarean delivery rates (range: 0.1 to 11.9 percentage points lower) while 13 hospitals had an increase (range: 0.2 to 16.2 percentage points higher). For nulliparous, term, singleton, vertex inductions (Appendix 7, available online at http://links.lww.com/xxx), 19 hospitals had a decrease in cesarean delivery rates (range: 0.7 to 20.6 percentage points lower) and 11 hospitals had an increase (range: 2.8 to 31.1 percentage points higher). Decreases in rates were statistically significant at 5 hospitals for nulliparous, term, singleton, vertex births and 8 hospitals for nulliparous, term, singleton, vertex inductions. Statistically significant increases in cesarean delivery rates only occurred at Level I/II hospitals.

We also assessed whether the strength of implementation of the bundle was associated with lower cesarean delivery rates. Table 3 compares Q1-Q2 2019 cesarean rates for hospitals that implemented the median number of bundle practices or more versus fewer for each domain of the bundle. Implementation strength was associated with reduced cesarean delivery rates for only one of the four domains—Response. Across hospitals that implemented at least the median number of practices in the Response domain, the relative risk of cesarean delivery was 0.91 (95% CI: 0.85—0.97) for all nulliparous, term, singleton, vertex births and 0.82 (95% CI: 0.75—0.91) for induced nulliparous, term, singleton, vertex births. For all nulliparous, term, singleton, vertex births, the risk of cesarean delivery was also lower across hospitals that implemented at least the median number of practices in the Readiness and Recognition and Prevention domains, but these differences did not reach statistical significance.

Table 3.

Association between practice adoption by cesarean bundle domain and endline cesarean delivery rates among nulliparous, term, singleton, vertex births (n=30)^*

		NTSV Births			Induced NTSV Births
Implementation Strength	No. of Hospitals (%)	Cesarean Deliveries/Births	Cesarean Delivery Rate (%)	RR (95% Cl)	Cesarean Deliveries/Births	Cesarean Delivery Rate (%)	RR (95% Cl)
R1. Readiness
At least the median no. of practices	17 (57)	1,490/5,647	26.4	0.96 (0.90–1.03)	587/1,891	31.0	0.99 (0.90–1.08)
Less than the median no. of practices (ref)	13 (43)	1,153/4,195	27.5		581/1,845	31.5
R2. Recognition and prevention
At least the median no. of practices	17 (57)	1,245/4,736	26.3	0.96 (0.90–1.02)	590/1,844	32.0	1.05 (0.95–1.15)
Less than the median no. of practices (ref)	13 (43)	1,398/5,106	27.4		578/1,892	30.5
R3. Response
At least the median no. of practices	16 (53)	1,526/5,913	25.8	0.91 (0.85–0.97)	598/2,092	28.6	0.82 (0.75–0.91)
Less than the median no. of practices (ref)	14 (47)	1,117/3,929	28.4		570/1,644	34.7
R4. Reporting and systems learning
At least the median no. of practices	16 (53)	1,127/4,125	27.3	1.03 (0.96–1.10)	489/1,509	32.4	1.06 (0.97–1.17)
Less than the median no. of practices (ref)	14 (47)	1,516/5,717	26.5		679/2,227	30.5

NTSV, nulliparous, term, singleton, vertex; RR, risk ratio.

^*Includes 30 of 31 hospitals participating in the collaborative; vital statistics data not available for one hospital. The median number of practices adopted for R1 =2, R2=5, R3=4, R4=3.

Discussion

Based on a Pubmed search for articles in the last ten years with the terms “quality” and “collaborative” and “cesarean,” this study is only the second evaluation of a state perinatal quality improvement collaborative to reduce the nulliparous, term, singleton, vertex cesarean delivery rate. Most hospitals in Maryland’s primary cesarean collaborative reported implementing multiple practice changes during their participation. Following Maryland’s collaborative, the statewide nulliparous, term, singleton, vertex cesarean delivery rate was 1.6 percentage points lower than the baseline rate, a significant but smaller change than the corresponding 3.2 percentage point reduction reported by California’s collaborative.⁸ Maryland’s collaborative was also associated with a larger 4.8 percentage point reduction in the cesarean delivery rate for nulliparous, term, singleton, vertex births after induction. Overall, the reduction in cesarean delivery rates appears to be associated with improvements at Level III/IV hospitals with larger delivery volumes. It’s worth noting that at the end of the collaborative, Maryland’s nulliparous, term, singleton, vertex cesarean delivery rate of 26.9% was still above the national average of 25.9% and well above the recently set national goal of 19.4%.¹⁸

The large number of practices in the cesarean bundle and the variability in practice adoption patterns between hospitals—both before and during the collaborative—limits the ability to identify individual practices that independently reduce the cesarean rate. Therefore, we compared nulliparous, term, singleton, vertex rates by implementation strength for each of the four bundle domains. In this analysis, adopting more practices was only associated with lower rates for the Response domain. The Response domain has the largest number of practices that standardize clinical care, including induction scheduling, diagnosis and treatment of labor dystocia and failed induction, and interpretation of fetal heart rate patterns. The important role of standardizing care is consistent with a Cochrane review finding moderately strong evidence that implementation of clinical practice guidelines, alongside feedback to clinicians (e.g., second opinions, audit and feedback of rates), can reduce cesarean delivery.¹⁹ Indeed, multiple reports have highlighted variability in cesarean rates between hospitals^20,21 and physicians,²² suggesting that differences in clinical practice play a key role in cesarean rate patterns over time.

The majority of the implementation of practices in the Response domain at Maryland hospitals took place prior to the collaborative. The limited progress in the Response domain during the collaborative may help explain why the reduction in cesarean rates was smaller than that observed in California.⁸ California hospitals were also required to reduce nulliparous, term, singleton, vertex cesarean rates to the Healthy People 2020 goal of 24% in order to participate in health insurance exchange plans,^8,23 an external pressure to change that was not present in Maryland. It should also be noted that the results of the ARRIVE trial, which concluded that labor induction at 39 completed weeks resulted in lower cesarean delivery rates, were published in late 2018,²⁴ towards the end of both Maryland and California’s collaboratives.

The Readiness and Recognition and Prevention domains of the bundle include multiple practices to improve labor support and pain management, which have also been shown to reduce cesarean rates in some studies.^25–27 Although training on labor management and support did take place at most hospitals in the collaborative, only half of hospitals trained the majority (>60%) of physicians and nurses, and training coverage increased at some hospitals only towards the end of the collaborative. Delayed implementation of labor support practices may have prevented us from observing an association with cesarean rates during the study period. Another labor support practice associated with lower cesarean delivery rates, integration of doulas into birth care teams,^28,29 had very low adoption among participating hospitals.

An important finding from this study is the variability in implementation progress and cesarean rates between hospitals. On average, hospitals had implemented half of the practices in the bundle by the end of the collaborative, with some implementing more and others far fewer. Other evaluations of perinatal quality improvement collaboratives have also found that some hospitals get left behind in these efforts, without making considerable progress and improving outcomes.^9,30,31 Indeed, quality improvement collaboratives in other clinical areas are not always successful at all sites.³² The quality improvement capacity and motivation at individual sites are likely important determinant of success.³³ For example, labor and delivery units that have experienced interdisciplinary quality improvement teams, support from hospital-wide quality improvement offices, and strong patient safety cultures²³ are likely better prepared to implement maternal safety bundles. In the MDPQC, hospitals that used implementation strategies engaging clinicians made greater progress implementing early practice changes.³⁴ Given that work by state PQCs represents a primary national strategy for reducing maternal morbidity and mortality,³⁵ it is critically important to conduct further implementation research to identify determinants of success and strategies to support all participating hospitals to make improvements.

This evaluation study has strengths and limitations. The study includes 97% of birthing hospitals in the state. Although the before-and-after design limits our ability to examine causal associations, we have assessed intermediate outcomes—self-reported changes in clinical practices—at 2 time points in order to limit recall bias and strengthen the plausibility of the findings.³⁶ Bundle practice implementation was self-reported by hospital respondents, and we were not able to independently verify practice implementation or the degree to which individual healthcare professionals adhered to new practices and policies. Further studies are needed to assess the effectiveness of the individual practices in the cesarean bundle that are not already well studied.

The limited data available for this evaluation precluded some analyses. Outcome data were only available in aggregate, preventing patient-level analyses and adjustments for patient population characteristics or assessment of outcomes by such characteristics. In Maryland and many other states, stronger maternal data systems that link hospital discharge data with birth certificates are needed to accurately assess maternal morbidity among nulliparous, term, singleton, vertex births. While California’s cesarean collaborative reported no change in some maternal morbidities (e.g., transfusion, 3^rd and 4^th degree lacerations), strengthening the ability of all states to monitor maternal morbidities is important for ensuring maternal safety while working to reduce cesarean delivery rates.

Reducing cesarean delivery rates will continue to be a national priority for low-risk births, as was reaffirmed by the U.S. Department of Health and Human Services’ recent “Action Plan to Improve Maternal Health in America.”¹⁸ Maryland’s collaborative to implement the cesarean bundle was associated with a reduction in nulliparous, term, singleton, vertex cesarean rates, but the decrease was not as large as observed in California,⁸ and not for all hospitals that participated in the quality collaborative. Determining the best ways to support hospitals to implement practice changes is essential for maximizing the effects of perinatal quality improvement collaboratives.