Accounting for adverse events and outcomes related to maternity is fundamental to measuring the health of populations and the systems for delivering medical care. Traditionally, maternal mortality has been such a measure. The best information we have in the United States, from the Centers for Disease Control and Prevention’s (CDC) Division of Reproductive Health’s Pregnancy Mortality Surveillance System, suggests a steady and high rate of maternal death of approximately 17 per 100,000 births.1 Still, maternal deaths at state and local levels are generally rare, and, although it is critical to identify and review each death for the purpose of improving the quality of care, the utility of such reviews is limited by the relatively small number of events. Severe maternal morbidity (SMM), broadly defined as unexpected and potentially life-threatening events associated with labor and delivery, have significant long-term and short-term effects on women, are much more common than deaths, and can represent antecedents to mortality. As such, SMM surveillance presents a promising candidate to further our understanding of maternity care and how we might improve it. Currently, widely available administrative discharge data developed at the CDC based on International Classification of Diseases (ICD) codes serve as the basis for identifying cases of SMM at the population level.2 Real-time identification based on admission to intensive care units or transfusion of 4 or more units of blood has been recommended for facility-based reviews of potential SMM cases.3 These constructs are generally confined to events and procedures that occur during the delivery hospitalization.
In this issue of Obstetrics & Gynecology, Declercq et al4 (see page 165) extend the traditional concept of SMM to include both antepartum and postpartum admissions in addition to delivery admissions in Massachusetts. They used contemporary longitudinally linked ICD-based discharge files and natality data to create a woman-based set of hospital admissions, with the advantage that the unit of analysis became the woman across the perinatal time period from conception through 42 days postpartum. These investigators reported a 22% increase in detection of SMM compared with delivery-only hospitalizations, with the majority of the increase attributed to the 42 days postpartum and the largest contribution attributed to sepsis. Expanding surveillance to include prenatal and postpartum indicators of SMM and updating the period of surveillance to include recent years resulted in an SMM rate in 2018 of 259 instances of SMM per 10,000 deliveries, making this measure 150 times more common than pregnancy-related deaths.
The importance of going beyond the delivery hospitalization to account for the full range of severe complications cannot be overemphasized. Optimally, the care of a woman for pregnancy begins before conception and continues through delivery and beyond. This is especially important given the well-deserved attention now being given to the postpartum period.5 This report adds further credence to the notion that vigilance is required across the entire perinatal period; care does not end with delivery. As state maternal mortality review committees begin to consider identifying SMM cases for review, it becomes important to include SMM during and after pregnancy so that the entire spectrum of complications is recognized. Given the number of cases, most committees will not have the resources to review every instance of SMM. However, enhanced identification has the potential to target specific conditions. For example, the large contribution of sepsis to antenatal and postpartum SMM would be a reason for targeted review of such cases in an effort to direct prevention efforts. Precedence for targeted morbidity reviews in addition to complete death reviews can be found in the example of the United Kingdom’s Confidential Enquiry into Maternal Deaths and Morbidity.6
Declercq et al4 also demonstrate the power of data linkages for extending public health surveillance activities. Much discharge data, such as the National Inpatient Sample, can identify only the event of the discharge, without linking that event to a person who is receiving hospital care across time. Moreover, dates of events, such as the date of delivery, are usually not available in administrative discharge data sets. Hence, although the event of delivery can be identified, where that occurs relative to admission and discharge is unknown. By anonymously linking a person with events across time, we have a powerful tool to enhance surveillance. Ultimately, this gets us past a mere accounting and allows for analysis of care that can provide insights into quality—both successes and areas in need of improvement.
There is no recognized gold standard for SMM. Morbidity occurs across a spectrum of severity, and indicators include a wide variety of conditions, procedures, and events. Hence, it will always be difficult to account for SMM. Such accounting must start somewhere, and because we improve our understanding of how to use data to improve our care, we will also improve the data. Continuous improvement in identification, and instituting state and local processes to take thoughtful, data-based action to address severe complications, contributes to fulfilling our duty to the women who entrust us with the honor of providing care for them.