Comprehensive Perinatal Safety Initiative to Reduce Adverse Obstetric Events

Abstract

A comprehensive perinatal safety initiative (PSI) was incrementally introduced from August 2007 to July 2009 at a large tertiary medical center to reduce adverse obstetrical outcomes. The PSI introduced: (1) evidence-based protocols, (2) formalized team training with emphasis on communication, (3) standardization of electronic fetal monitoring with required documentation of competence, (4) a high-risk obstetrical emergency simulation program, and (5) dissemination of an integrated educational program among all healthcare providers. Eleven adverse outcome measures were followed prospectively via modification of the Adverse Outcome Index (MAOI). Additionally, individual components were evaluated. The logistic regression model found that within the first year, the MAOI decreased significantly to 0.8% from 2% (p<.0004) and was maintained throughout the 2-year period. Significant decreases over time for rates of return to the operating room (p<.018) and birth trauma (p<.0022) were also found. Finally, significant improvements were found in staff perceptions of safety (p<.0001), in patient perceptions of whether staff worked together (p<.028), in the management (p<.002), and documentation (p<.0001) of abnormal fetal heart rate tracings, and the documentation of obstetric hemorrhage (p<.019). This study demonstrates that a comprehensive PSI can significantly reduce adverse obstetric outcomes, thereby improving patient safety and enhancing staff and patient experiences.

Purpose

Since the call to improve patient safety, progress has been generally slow, with notable exceptions in anesthesia, cardiology, and critical care. The reasons for the slow change are varied: the complexity of healthcare, medicine's tenacious commitment to the individual, professional autonomy, fear, and skepticism of broad changes (Leape & Berwick, 2005). To address the gap in obstetrics, we implemented a multistep, multicomponent perinatal safety initiative (PSI) to evaluate and subsequently decrease adverse obstetric outcomes.

Review of the Literature

In 1999, the Institute of Medicine (IOM) reported that medical errors may result in as many as 98,000 deaths a year, costing as much as 29 billion dollars, noting that ‘‘bad systems, not bad people’’ are responsible for the majority of error and injury (Kohn, Corrigan, & Donaldson, 2000). A key IOM recommendation to prevent errors is the establishment of an interdisciplinary ‘‘team approach’’ to patient care. The Joint Commission (JC) identifies ineffective provider communication as a contributing factor in up to 80% of adverse and near harm events (Scalise, 2006).

Creating a culture of safety is a complex process requiring a long-term approach that changes behaviors, practices, and relationships. This process involves changes for both the individual practitioner as well as the healthcare team.

We implemented a multicomponent PSI in a large, tertiary medical center, delivering approximately 5,300 infants per year. Designated as a Regional Perinatal Center, it provides services to a diverse group of patients across cultural, racial, and socioeconomic backgrounds.

Study Design and Methods

QI Diagnosis

The PSI was developed to incrementally introduce multiple safety plans from August 2007 through July 2009, following an internal and external review of sentinel events (root cause analysis). Each month, all obstetrical hemorrhage charts and 10 charts with abnormal fetal heart rate tracings were reviewed. A list of critical criteria was established to evaluate case management and documentation. A single assessor reviewed the charts applying the critical criteria. Charts were then deemed appropriate or not appropriate for management as well as documentation. As a result of this review, the following were identified as potential causes for adverse events: (1) poor communication between providers; (2) inadequate escalation policy; (3) lack of standard protocols; and (4) lack of standardization of interpretation of fetal heart rate tracing. A multidisciplinary perinatal safety committee, comprised of physicians, nurses, educators, and administrators, was charged with addressing audit findings and developing/implementing the PSI.

Interventions

The PSI utilized several safety initiatives designed to target potential contributing factors to adverse outcomes (see Table 1). To target communication, we implemented the Team STEPPS methodology for team training, which emphasizes communication techniques (Agency for Healthcare Research and Quality, 2010). To facilitate communication, daily multidisciplinary teaching rounds were introduced. The entire perinatal team (attending Maternal Fetal Medicine [MFM], attending obstetrical, resident, PA and RNs, Anesthesiologist, and Neonatologist) reviewed and discussed appropriate assessment/management of obstetrical service admissions.

Table 1. Safety Initiatives Designed to Target the Potential Contributing Factors to Adverse Outcomes

	Contributing Factors to Adverse Outcomes
Safety Initiatives	Communication	Escalation	Lack of Standardized Protocol	Lack of Standardized EFM
EFM, electronic fetal monitoring.
Team STEPPS	√	√
EFM course and exam	√		√	√
Multidisciplinary teaching rounds	√	√	√	√
Obstetrical emergency simulation	√	√	√
Introduction of evidence-based protocols	√	√	√

To standardize electronic fetal monitoring (EFM) terminology we instituted an EFM educational course and exam in the interpretation and management of fetal heart rate patterns. We mandated that every member of the perinatal team successfully complete the same online courses on Definition of FHR changes, Pathophysi-ology of FHR changes, the Relationship between FHR and fetal oxygenation, Fetal acid base values, Clinical relevance of FHR changes, and Management scenarios for specific FHR changes.

We also enhanced our electronic medical record (entitled Centricity Perinatal manufactured by General Electric) to require the use of correct National Institute of Child Health and Human Development terminology using prompts (Electronic Fetal Heart Rate Monitoring, 1997). Several evidence-based management protocols were developed and implemented, including: (1) the use of Pitocin augmentation of labor; (2) the use of antibiotics and thromboembolic prophylaxis for cesarean; (3) the use of magnesium for seizure prophylaxis; (4) hemorrhage protocol; (5) protocol for induction of labor; (6) management of intrapartum fetal heart rate abnormalities; and (7) obstetrical rapid response team.

We introduced and standardized a number of evidence-based protocols. To ensure that proper action is taken (decreasing or discontinuing the Pitocin infusion), we empowered all team members to intervene and follow the protocol when presented with an abnormal contraction pattern regardless of the FHR tracing.

Another intervention with a positive impact on the cesarean section and associated maternal morbidity rates was the protocol for elective inductions that confine interventions to patients with a favorable cervix at time of induction.

For situations where severe fetal heart rate abnormalities was noted (fetal bradycardia, prolonged decelerations) a specific protocol was introduced that outlined the intervention timetable and a rapid response team was formed to act in critical situations, particularly with regard to accomplishing delivery within the first 10–12 min from acute change in fetal status.

To eliminate elective deliveries (elective induction or repeat cesarean section) before 39 weeks gestation, we implemented a policy and clinical practice based on the American College of Obstetricians and Gynecologists (ACOG) recommendations (ACOG Practice Bulletin, 1999), instituting an educational process during case scheduling. In situations where gestational age at induction or the indication for intervention was unclear, the case was referred to the attending MFM physician for a final decision at the time of booking rather than after patient admission.

To address escalation policy, a hemorrhage protocol was instituted to enable the multidis-ciplinary team (anesthesia, surgery, critical care) to be quickly summoned in situations associated with massive obstetrical bleeding, thereby minimizing complications with severe hemorrhagic shock.

Finally, an obstetrical emergency simulation program in high-risk care scenarios, including shoulder dystocia, maternal hemorrhage, and seizure was undertaken. Multidisciplinary drills were performed. Nursing and physician staff was required to complete the same fetal monitoring competency, attend the same monthly educational activities, and participate in the same simulated obstetric emergencies together, as a team.

To measure the impact of the PSI, we utilized a modified version of the Adverse Outcome Index (AOI) published in 2006 (Mann et al. 2006). The authors noted their concern with two outcomes (term NICU admissions and third/fourth degree lacerations) because of definition variation. We also surveyed several large obstetrical/neonatal units regarding their indication for NICU admissions of term infants for >24 hr. We could find no specific protocols/guidelines regarding these admissions. Furthermore, admission rates for specific indications varied greatly by covering attending physician, as well as unit. We also found, through retrospective chart review, significant discrepancies between coding for third/fourth degree lacerations and physician documentation regarding this complication. Thus, these two outcomes were excluded from our analysis. Based on these observations, our Modified Adverse Outcome Index (MAOI) included the maternal and fetal/neonatal measures highlighted in Table 2.

Table 2. The Modified Adverse Outcome Index

Maternal Indicator	Fetal/Neonatal Indicators
HIE, hypoxic-ischemic encephalopathy.
Maternal death	Stillbirth
Admitted to higher level of care	Neonatal death
Uterine rupture	5 min APGAR <7
Peripartum hysterectomy	Iatrogenic prematurity
Return to OR	Birth trauma HIE

These 11 MAOI measures were followed prospectively, following the introduction of the PSI. The MAOI was calculated by dividing the number of pregnancies complicated by one or more of these measures by the total number of deliveries for that time period.

To compare our outcomes to other medical centers or national norms, we utilized published data, where available. In addition to these specific outcome measures, we chose to review the management and documentation of two high-risk situations with potential for poor outcome: obstetrical hemorrhage and fetal heart rate abnormalities. The reviews of both outcome and process measures were standardized utilizing a review tool to ensure objective evaluation of these cases.

Patient perceptions of teamwork and commitment to patient safety were continually evaluated by an anonymous questionnaire after discharge from hospital. To do this, two questions: ‘‘Would you recommend the institution?’’ and ‘‘Did the staff work together?’’ were tracked and analyzed over the time period utilizing a Press-Ganey survey instrument (Kaldenberg, Mylod, & Drain, 2003).

Staff perceptions of safety were assessed at the beginning and approximately 18 months after project initiation, using an item from the Safety Culture Climate Survey, publicly available on the Institute for Healthcare Improvement (IHI) website (Sexton & Thomas). It should be noted that only a single question was tracked, which inquired about the culture of safety on the unit.

Statistical Analysis

Incidence rates of MAOI were compared across time points using logistic regression (SAS Version 9.2, SAS Institute, Cary, NC). It is important to note that MAOI rates were computable for each calendar quarter of the years 2008 and 2009; however, quarterly data for 2006 and 2007 were not available without a laborious manual chart review of over 5,400 medical charts. For the purposes of applying the regression model, equal spacing of time-adjacent observations could not be assumed, because it did not apply to 2007. Accordingly, 2007 and each quarter were coded as dummy variables with odds ratios computed relative to the observed MAOI rate in 2007.

Exploratory analyses of the individual components that comprise the MAOI were carried out using exact logistic regression. A linear mono-tonic trend over time was assumed, because the frequencies of the individual components were very small, thus making the coding of calendar quarters as dummy variables infeasible. For comparability, the MAOI analysis was also rerun using the linear trend model.

Institutional Review Board Approval

This study was approved by the Institutional Review Board under exempt status (IRB #09-279).

Results

MAOI

The MAOI rate decreased from 1.95% to 0.89% (Figure 1 and Table 3); the logistic regression model demonstrated a significant decrease across calendar year quarters relative to 2007, p<.0004). Also shown in Figure 1, the odds ratios for an MAOI tended to decrease over time, with a slight ‘‘leveling off’’ in 2009. Specifically, the odds ratio for the first quarter in 2008 was 0.768 (i.e., a 23% decrease in MAOI risk) and decreased to 0.291 (i.e., a 71% decrease in risk) for the first quarter in 2009, leveling at about 0.460 (i.e., a 54% decrease) for the last three quarters of 2009.

Table 3. Incidence of Modified Adverse Outcomes and Odds Ratios^a

	2007	Q1 2008	Q2 2008	Q3 2008	Q4 2008	Q1 2009	Q2 2009	Q3 2009	Q4 2009
^aIf multiple adverse outcomes occurred with a delivery, then that delivery was counted only once in the determination of ‘‘number of deliveries with an adverse outcome’’
# of deliveries	5411	1281	1398	1340	1246	1235	1330	1258	1239
# of deliveries with an adverse outcome	104	19	15	15	10	7	12	11	11
Modified Adverse Outcome Index rates (%)	1.92	1.48	1.07	1.12	0.80	0.57	0.90	0.87	0.89
Odds ratios (relative to 2007)	1.00	0.77	0.55	0.58	0.41	0.29	0.47	0.45	0.46

Figure 1. Modified Adverse Outcome Index Rates and Odds Ratios Over Time

Individual Measures

Individual components of the MAOI were evaluated in an exploratory way. The incremental decrease in the MAOI odds ratio from one quarter to the next was 0.88. For example, at the end of the fourth quarter of 2009 (8 quarters after 2007), the risk of an adverse outcome was 36% of the risk in 2007 (0.88⁸=0.36). Utilizing exact logistic regression, declines in individual adverse outcomes and their 2009 Q4 odds ratios were computed for ICU (0.47), return to the operating room (0.19), stillbirth (0.47), neonatal death (0.39), iatro-genic premurity (0.27), and trauma (0.11). Of these declines in risk, only the rates for returning to the operating room (p<.018) and birth trauma (p<.0022) were statistically significant. Increases in individual  adverse outcomes and their 2009 Q4 odds ratios were computed for uterine rupture (2.48), hysterectomy (1.08), APGAR 5 (1.59), and hypoxic-ischemic encepha-lopathy (HIE; 1.27), none of which were statistically significant. Statistically significant trends could not be demonstrated for the remaining nine individual measures, although these measures tended to show a decrease over time.

Next, we compared individual outcome measures to benchmark data in the literature, although we did not risk adjust any of these incidence rates. The incidence of peripartum hysterectomy is reported as 0.8–5% of deliveries, compared with our rate of 1.1% (see Table 4; Briery et al., 2007; Forna, Miles, & Jamieson, 2003; Glaze et al., 2008). There is no benchmark data on maternal ICU admissions (unplanned transfer to a higher level of care) during the delivery hospitalization; however, the incidence of severe maternal morbidity during that hospital stay is reported to be 5.1%. As can be seen in Table 4, our unplanned transfer to a higher level of care rate of 1.38% compares favorably to the norm (Callaghan, MacKay, & Berg, 2008). The reported incidence of HIE in term or near-term infants ranges from 1% to 8% (American College of Obstetricians and Gynecologists, 2003). When confined to severe birth asphyxia (5min APGAR score <3), the reported incidence is 2.9%; whereas our incidence for the last year was 0.6% (Thorngren-Jerneck & Herbst, 2001). The normed incidence of 5min APGAR score <7 is reported as 0.7–1.6%; the incidence in our sample was 0.22% (Martin et al., 2009). Perinatal mortality rates compared favorably for our population at 2.2% for stillbirths and 0.4% for neonatal deaths, in comparison with the national data of 6.2% and 4.8%, respectively (Mazza et al., 2007). Our birth trauma rates were 0.2% compared with published data of 1.6–7.1% (MacDorman & Kirmeyer, 2009). We were able to virtually eliminate iatrogenic prematurity, with an incidence of 0.2%, compared with the JC normed rate of 5% (Scalise, 2006). Unfortunately, we could not identify normed rates for return to the operating room or uterine rupture.

Table 4. Individual Modified Adverse Outcome Indicators

	LIJ (per 1,000 births; rates for 2009)	Literature
^aThis rate is based on birth rates per 100 not 1,000.
Peripartum hysterectomy	1.0	0.8–5.0
Unplanned transfer to a higher level of care	1.38	5.1
HIE	0.6	2.9
5 min APGAR <7	0.22^a	0.7–1.6
Stillbirth	2.2	6.2
Neonatal death	0.4	4.8
Birth trauma	0.2	1.6–7.1
Iatrogenic prematurity	0.02^a	50.0
Return to OR	0.6	Not
		available
Uterine rupture	0.4	Not
		available

Additional Measures

It was important to evaluate our primary outcome measures (MAOI) regarding facility performance on multiple process indices. First, we examined staff perceptions of safety during the period corresponding to the PSI. Figure 2 shows that staff perceptions of patient safety increased following PSI implementation (p<.0001).

Figure 2. Staff Perceptions of Patient Safety, p<.0001

Next, we examined patient perceptions of safety: whether patients would recommend the institution and whether patients perceived that the staff worked together. As seen in Figure 3a and b, a larger percentage of patients, over time, indicated that they would both recommend the institution (not significant) and that staff worked together (p<.028).

Figure 3. (a) Patients Would Recommend the Institution, Not Significant. (b) Patients’ Perception of Whether the Staff Worked Together, p<.028

We also found that the cesarean delivery rate remained stable during the study period (Figure 4, not significant). In assessing the management of fetal heart rate abnormalities, the percentage of appropriately managed cases increased greatly, from 53% in January–March 2008 to 93% in September–November 2009 (Figure 5, p<.002). In terms of documenting fetal heart rate abnormalities, the total percentage of adequate documentation increased greatly, from 29% in January–March 2008 and to 100% in September–November 2009 (Figure 6, p<.0001).

Figure 4. Cesarean Delivery Rate, Not Significant

Figure 5. Process Measure Assessing Management of Abnormal Fetal Heart Rate Tracings, p<.002

Figure 6. Process Measure Assessing Documentation of Abnormal Fetal Heart Rate Tracings, p<.0001

As seen in Figure 7, the total percentage of obstetrical hemorrhages considered appropriately managed remained stable during the study period (not significant). Lastly, as shown in Figure 8, the total percentage of adequately documented obstetrical hemorrhage cases increased greatly during the study period, starting at 45% in January–March 2008 to 100% in September–November 2009 (p<.019).

Figure 7. Process Measure Assessing Management of Obstetric Hemorrhage, Not Significant

Figure 8. Process Measure Assessing Documentation of Obstetric Hemorrhage, p<.019

Limitations and Directions for Future Research

The study's major limitation was a methodological one: it was neither feasible nor practical to randomly assign practitioners to different protocols. A methodological limitation of not having a structured control group is the lack of a comparative cohort. It is for this reason that we compare our findings to benchmark data available in the literature.

Discussion

It is difficult to attribute the impact of the individual components of the PSI to the observed overall decline of the MAOI. This is mostly due to the rarity of adverse events. Maternal deaths and uterine ruptures, for example, are so rare that examining a trend requires thousands of cases, supporting the utility of the composite index.

With regard to management of fetal heart rate abnormalities, the PSI seems to have contributed to more consistent guideline adherence in 2009 than 2008. Also, the PSI appears to have less of an impact on hemorrhage management. This may be explained by the adoption of a hemorrhage protocol in 2006 suggested by the New York State Department of Health (2008), following the publication of a report by the Safe Motherhood Initiative indicating its contribution to maternal mortality (World Bank, 1993).

A critical component of this initiative was the educational process designed to improve recognition, appropriately document complications, and avoid interventions that increase risk of complications. At multidisciplinary teaching rounds, concerns, and suggestions made by the healthcare team resulted in actual changes to the management plan, facilitating greater understanding of the plan's importance.

Introducing a standardized approach to EFM interpretation was another important component. This educational process contributed not only to improved individual expertise and knowledge, but also greatly enhanced the ‘‘team approach’’ to patient care, resulting in a 100% pass rate on the competency exam. Improved communication and work interaction between various members of the team, was further enhanced through Team STEPPS (Agency for Healthcare Research and Quality, 2010), as evidenced by reported patient perceived increases in staff working together.

Implementing the ACOG-recommended policy regarding elective deliveries significantly increased compliance of attending staff and prevented patient and physician dissatisfaction associated with canceling a scheduled procedure at patient admission to the hospital.

Improved compliance with the Pitocin augmentation protocol is likely another reason for improved perinatal outcome. A common finding of the root cause analysis for perinatal injury was the presence of abnormal uterine contractions preceding the development of FHR abnormalities, fetal acidosis, and potential fetal brain injury. Therefore, the presence of a normal FHR tracing would no longer justify continued Pitocin augmentation in the presence of an abnormal uterine contraction pattern.

As reported in the literature, others have implemented comprehensive safety initiatives on perinatal outcomes with similar results. Mann et al. (2006) reported a 17.2% decline following the introduction of team training, multi-disciplinary rounds, and a process of cross-monitoring. Pettker et al. (2009) used a similar AOI to study the impact of their performance initiative consisting of: (1) appointment of a Perinatal Safety Nurse to collect and investigate adverse and centennial events; (2) organization of an obstetrical safety committee to address the need for protocols/policies; (3) team training; (4) standardized EFM interpretation through requiring training and certification. Following implementation, they reported a drop in the AOI from 3.4% in 2004 to 1.8% in 2007.

Mazza et al. (2007) introduced several initiatives in 2004 targeting improved communication, standardized interpretation of fetal heart rate, and the adoption of IHI perinatal bundles, reporting an 85% drop in birth trauma incidence.

Implications for Practice

Improving quality of care is a complex and difficult undertaking that should focus on various components of the healthcare system, including improving individual performance, fostering a team approach, better communication, adoption of best practices, standardized approaches to interpreting medical data (EFM), and the measurement of these initiatives upon patient outcomes. A critical requirement in maintaining success is the ability to constantly monitor the impact of the PSI utilizing the MAOI.

A major challenge in the field of obstetrics has been the lack of validated measures that reflect the quality of perinatal care. A comprehensive, multicomponent PSI can offer significant and lasting improvements in preventing adverse obstetric outcomes, thereby improving patient safety and enhancing staff and patient experiences.

Biographies

• Brian Wagner, MD, North Shore-Long Island Jewish Health System, Great Neck, NY and Bronx Municipal Hospital Center, Bronx, NY, had primary responsibility for conceiving of and designing the study, manuscript development, and review.

• Natalie Meirowitz, MD, North Shore-Long Island Jewish Health System, Great Neck, NY, was responsible for leading team training sessions, data collection, and manuscript review.

• Jalpa Shah, MD, North Shore-Long Island Jewish Health System, Great Neck, NY, was responsible for leading team training sessions, data collection, and manuscript review.

• Deepak Nanda, MD, North Shore-Long Island Jewish Health System, Great Neck, NY, was responsible for leading team training sessions, data collection, and manuscript review.

• Lori Reggio, RN, North Shore-Long Island Jewish Health System, Great Neck, NY, was responsible for leading team training sessions, data collection, and manuscript review.

• Phyllis Cohen, RN, North Shore-Long Island Jewish Health System, Great Neck, NY, was responsible for leading team training sessions, data collection, and manuscript review.

• Karen Britt, RN, North Shore-Long Island Jewish Health System, Great Neck, NY, was responsible for leading team training sessions, data collection, and manuscript review.

• Leah Kaufman, MD, North Shore-Long Island Jewish Health System, Great Neck, NY, was responsible for leading team training sessions, data collection, and manuscript review.

• Rajni Walia, MA, North Shore-Long Island Jewish Health System, Great Neck, NY, was responsible for analyzing and interpreting data, and writing and revising the manuscript.

• Corinne Bacote, RN, North Shore-Long Island Jewish Health System, Great Neck, NY, participated in the patient and provider survey process and reviewed the manuscript.

• Martin L. Lesser, PhD, North Shore-Long Island Jewish Health System, Great Neck, NY, the Feinstein Institute for Medical Research, Manhasset, NY, and Hofstra University School of Medicine in partnership with North Shore-Long Island Jewish Health System, Hempstead, NY, had primary responsibility for conceptualizing the statistical methods and analyses, and writing and revising the manuscript.

• Renee Pekmezaris, PhD, North Shore-Long Island Jewish Health System, Great Neck, NY, Albert Einstein College of Medicine, Bronx, NY, the Feinstein Institute for Medical Research, Manhasset, NY, and Hofstra University School of Medicine in partnership with North Shore-Long Island Jewish Health System, Hempstead, NY, was responsible for analyzing and interpreting data, and writing and revising the manuscript.

• Adiel Fleischer, MD, North Shore-Long Island Jewish Health System, Great Neck, NY, had responsibility for con-ceiving of and designing the study, writing the manuscript, and manuscript review.

• Kenneth J. Abrams, MD, North Shore-Long Island Jewish Health System, Great Neck, NY, and Hofstra University School of Medicine in partnership with North Shore-Long Island Jewish Health System, Hempstead, NY, participated in reviewing and editing the manuscript and was responsible for setting the quality improvement focus and the patient and provider survey process.