A Series of Anesthesia-related Maternal Deaths in Michigan, 1985–2003

This protocol was approved by the University of Michigan Health System Privacy Board and the Michigan Department of Community Health. Maternal death reporting is not mandatory in Michigan. Since its inception in 1950,11Michigan Maternal Mortality Surveillance (MMMS) relied on voluntary reports by hospitals, medical examiners, other health and social service providers, and newspaper obituaries.10–15Beginning in 1990, the Michigan Department of Community Health has matched death certificates for women of reproductive age (10–45 yr old) with live birth certificates for that year and the previous year.13,15,16Starting in 1999, the Michigan Department of Community Health electronically linked women's death certificates with live birth certificates and fetal death files.13,16§

A pregnancy-associated death is defined as the death of a woman while pregnant or within 1 yr of the termination of pregnancy, irrespective of cause.17A pregnancy-related death is defined as the death of a woman while pregnant or within 1 yr of termination of pregnancy from any cause related to or aggravated by her pregnancy or its management, but not from accidental or incidental causes.17The relevant postpartum interval used by the State of Michigan to define a pregnancy-associated death increased over the course of the study as illustrated in table 1. Once a pregnancy-associated death is identified, the MMMS coordinator requests available hospital records, autopsy reports, death certificates, and, when applicable, birth certificates. For deaths not attributed to cancer or injury, a senior obstetrician reviews all material and generates a standardized summary. These summaries are presented to the MMMS Medical Review Committee members, who categorize each death as to cause, preventability, and responsible factors (patient, physician, hospital, or community) and determine whether the death was pregnancy-associated or pregnancy-related.13The primary records, case summaries, and committee data sheets comprise the case files.

Records for all pregnancy-associated deaths were reviewed by two authors to identify 25 cases in which an anesthetic complication (table 2) seemed to take place during the time the patient was under the care of an anesthesia provider. For each of the 25 cases, the authors extracted maternal age, race, marital status, insurance, highest level of education, gravidity, parity, height, weight, cause and timing of death relative to the termination of pregnancy, contributing causes leading to death, autopsy findings, and conclusions of the MMMS Medical Review Committee. An abbreviated narrative of the course of events included the surgical or obstetric procedure, urgency of the procedure, anesthetic management techniques, airway management techniques, and complications of anesthesia.

Four obstetric anesthesiologists independently reviewed and classified the 25 cases as anesthesia-related, anesthesia-contributing, or not related to anesthesia. Cases were initially categorized with reference to definitions published by Endler et al.  10in a previous review of anesthesia-related maternal death in Michigan. According to these definitions, “Primary cause was defined as a death attributable to anesthesia, either as the result of the medications used, method chosen, or the technical maneuvers performed, whether iatrogenic in origin or resulting from an abnormal patient response. A contributory factor was defined as death to which anesthesia contributed, but one that would likely have occurred even in the absence of an anesthetic intervention.”10 

Disagreements among the four reviewers were resolved through an iterative process in which the reviewers sequentially refined the definitions of anesthesia-related and anesthesia-contributing maternal mortality and applied these definitions to the 25 cases as well as to examples in the literature.2,3,10,18,19Discussion continued until all cases could be categorized with unanimous agreement. This resulted in a formal procedure to determine whether a case qualified as an anesthesia-related or anesthesia-contributing maternal death (fig. 1). Two obstetricians reviewed the definition procedure illustrated in figure 1to ensure consistent terminology with the obstetric literature.

Last, anesthesia-related deaths were reviewed to identify the chain of medical errors or care management problems that contributed to each patient death. Medical errors were defined as the failure of a planned action to be completed as intended or the use of an inappropriate plan to achieve an aim.20Care management problems included actions or omissions by staff in the process of care, in which care deviated beyond safe limits of practice and in which the deviation had a direct or indirect effect on the eventual adverse outcome for the patient.21In cases where the impact on the patient was unclear, it was determined to be sufficient in our study that the care management problem had a potentially adverse effect.21 

The Health Data Development Section of the Michigan Department of Community Health provided data on the proportion of live births to African-American women.

From 1985 through 2003, 855 pregnancy-associated deaths were reported to the State of Michigan (fig. 2). Of these pregnancy-associated deaths, 349 were pregnancy-related. The procedure to categorize a case as an anesthesia-related maternal death is presented in figure 1. There were eight such fatalities, accounting for 2.3% of all pregnancy-related deaths.

Demographic data are presented in table 3. The mean age of the eight patients who died was 31.2 yr, and 75% were African-American. From 1985 through 2003, 17.8% of live births in the State of Michigan were to African-American women.

Details of each anesthesia-related maternal death are provided in table 4. There were no deaths associated with a labor epidural for successful vaginal delivery. Three cases were true surgical emergencies due to maternal hemorrhage or nonreassuring fetal heart tones.

Six women who died were obese (body mass index [BMI] 30–39.9 kg/m²) or morbidly obese (BMI ≥ 40 kg/m²).

These eight anesthesia-related deaths occurred at seven different care facilities. In two of the eight deaths, care was provided by a nurse anesthetist supervised by the operating obstetrician, and in the remaining six of eight deaths, care was provided by a nurse anesthetist supervised by an anesthesiologist. In two cases, an anesthesiologist was present at the time of the primary anesthetic event. Medical records revealed 19 specific medical errors or care management problems, listed in table 5.

Anesthesia was considered a contributing factor in seven additional deaths (table 6). Among this group, three were African-American and four were white. Mean age and BMI of this group were not different than mean age and BMI of the group with anesthesia-related death (mean age 30.3 yr [SD = 5.4]vs.  31.1 yr [SD = 7.6], unpaired t  test P = 0.81; mean BMI 32.1 kg/m²[SD = 5.2]vs.  37.9 kg/m²[SD = 7.6], unpaired t  test P = 0.22).

This review identified eight anesthesia-related and seven anesthesia-contributing maternal deaths in Michigan between 1985 and 2003. The pattern of deaths illustrates three key points. First, all anesthesia-related deaths from airway obstruction or hypoventilation took place during emergence and recovery, not during the induction of general anesthesia. Second, system errors contributed the majority of deaths. Of concern, lapses in standard postoperative monitoring and inadequate supervision by an anesthesiologist seemed to contribute to at least half of all deaths. Finally, this report confirms previous work that obesity and African-American race are important risk factors for anesthesia-related maternal mortality.1,3,10 

Airway disaster during the induction of general anesthesia remains one of the most recognized causes of anesthesia-related maternal mortality.2–5Yet in this study of more than 850 maternal deaths, we were unable to identify a single case of failed intubation during elective or emergent induction of general anesthesia. In contrast, there were five cases of hypoventilation or airway obstruction during emergence, extubation, and recovery. Judgment and vigilance in monitoring oxygenation and ventilation are important throughout the perioperative period. Continuing medical education courses on airway management should remind perioperative and peripartum providers of this basic standard of care.

In seven of eight anesthesia-related maternal deaths, 19 specific medical errors or care management problems could be identified from the medical record (table 5). The errors listed were identified by the authors and were determined based on the authors' clinical judgment without the use of a validated instrument. Furthermore, ideal care may not have prevented the deaths in this series, and similar errors or care management problems may not necessarily lead to deaths in future patients. Further research is needed to evaluate the epidemiology of medical errors in obstetric anesthesia and to measure the association between specific errors and a range of patient outcomes including adverse events, near-miss events, and completely benign patient outcomes.

Although some of the care management problems listed in table 5could be attributed solely to the anesthesia provider, most represent system-level failures of prenatal care, patient education, or peripartum patient safety. Organizational accidents such as patient deaths typically result from a concatenation of contributing factors, including latent failures in organizational processes and environmental conditions, as well as active failures such as individual unsafe acts.22,23Proposals to improve patient safety in obstetric anesthesiology must attend to the broader context of maternity care and not be limited to the specific actions performed by the individual anesthesiologist or nurse anesthetist.

For example, antenatal anesthesiology evaluation has the potential to address preexisting conditions before delivery, but requires a robust system to identify patients who may benefit. Maternal deaths are now more likely to result from exacerbation of preexisting disease (indirect deaths) than from conditions specific to pregnancy (direct deaths).3In the current series, half of the anesthesia-related deaths may have been related to unrecognized coexisting disease. In France, an antenatal anesthetic visit has been mandatory for all pregnant women since 199824and is usually completed in the eighth month of pregnancy. Mandatory evaluation is not currently required or recommended in the United States. Research is needed to define the cost effectiveness of mandatory anesthesiology evaluation. If mandatory evaluation is not cost effective, research should focus on establishing clear indications for selective antenatal anesthesiology evaluation and the method by which patients are best identified.

Standard postoperative monitoring may have prevented at least three of these deaths. Among these, two women died of airway obstruction or apnea in postanesthesia care units without continuous pulse oximetry, including the most recent death in 2003. The American Society of Anesthesiologists guideline for postoperative care suggests that pulse oximetry is associated with early detection of hypoxemia, and recommends periodic assessment of airway patency, respiratory rate, and oxygen saturation measured by pulse oximetry during emergence and recovery.25Continuous pulse oximetry may further improve detection of hypoxemia, particularly for African-American patients in whom cyanosis may be more difficult to detect by visual inspection alone.10 

Inadequate supervision of care contributed to at least three anesthesia-related deaths (table 5), including two cases in which the nurse anesthetist was supervised by the operating obstetrician. In addition, four women died of postcesarean intraperitoneal hemorrhage without the benefit of an anesthesiologist to suggest concealed bleeding or to ensure sufficient intravascular access and resuscitation (table 6). As perioperative physicians, anesthesiologists are uniquely qualified to supervise the anesthesia care team, to manage and minimize anesthesia-related maternal risk, to provide peripartum medical diagnosis and treatment, to facilitate life-saving surgical interventions with efficient and safe anesthesia, and to lead prompt, coordinated, and effective resuscitation efforts. An anesthesiologist may not always be immediately available. However, as much as possible, workforce policies should ensure the appropriate involvement of anesthesiologists in developing clinical protocols and in delivering patient care.

Consistent with previous literature,1,10the majority of deaths in this series were in African-American women. Although medical errors and care management problems seemed to contribute the majority of the deaths identified in this series, it is not clear whether African-American women are more likely to experience an error or whether such errors have more severe consequences for this population. African-Americans represent a high risk group with increased rates of chronic hypertension,26–28pregnancy induced hypertension,29preeclampsia,29–34obesity,35–39peripartum cardiomyopathy,40and overall pregnancy-related mortality.1,41In most42–44but not all studies,45African-American women are more likely to deliver by cesarean delivery and therefore are more likely to require a surgical anesthetic.

This review is also consistent with previous reports that anesthesia-related maternal death is associated with maternal obesity.3,10An association between obesity and anesthetic risk is concerning because the prevalence of obesity and morbid obesity among nonpregnant women of childbearing age in the United States has increased from 11.7% and 1.8%, respectively, in 197246to 28.9% and 8%, respectively, in 2003.47Strategies to maximize anesthetic safety for obese and morbidly obese pregnant women have been described.48,49 

Observations from this retrospective case review must be interpreted with caution for three reasons. First, approximately 40% of pregnancy-associated deaths may have been missed over the duration of the study. The major limitations to identifying every maternal death50were that (1) the MMMS system does not review hospital records and autopsy reports of all women of reproductive age; (2) death certificates were not matched to neonatal death certificates before 1990; (3) death certificates were not matched to fetal death records before 1999; (4) the pregnancy check box on the Michigan death certificate51was implemented in 2004, after the final year of this study; and (5) maternal death reporting is not mandatory in Michigan.

Second, the complete medical record may not capture everything known about the anesthetic event at the time it occurred. Analysis of the temporal relation between specific anesthetic maneuvers and hemodynamic outcomes is not always possible using a standard anesthetic record. In several cases, the anesthesiologists carefully documented the relevant sequence of events. In other cases, the patients arrested under the care of a nurse anesthetist or postanesthesia nurse, neither of whom recorded notes detailing their observations leading up to the arrest. Therefore, in some cases, the mechanism of death is unproven and represents a best estimate based on the records available.

Third, portions of the medical record for each case may be missing, leading to inaccurate judgments about cause of death. However, significant omissions are unlikely for the following reasons. The Michigan Department of Community Health is required by state statute to investigate pregnancy-associated deaths and receives a very high percentage of requested records. The MMMS coordinator confirmed that all birthing hospitals supply medical records when requested; however, prenatal care records are not always available. Redaction, when observed on the medical records, was limited to patient and professional identifiers.

Despite these limitations, this report is notable for two reasons. First, by reviewing primary medical records and case documents, we were able to ascertain cases that would not have been identified using the vital records alone. Primary records also allowed us to report the probable mechanism of death. Second, we developed a structured definition procedure to categorize a death as an anesthesia-related maternal mortality. A similar procedure is recommended by the Centers for Disease Control and Prevention to define a pregnancy-related death.17The definition of anesthesia-related maternal mortality presented in this study was generated through an iterative process with reference to 25 cases as well as additional cases in the published literature. Future work is needed to test the interrater reliability of this definition of anesthesia-related maternal death.

Over the past 50 yr, anesthesiologists have achieved a remarkable reduction in the number of anesthesia-related maternal deaths.3,4However, African-American and obese women continue to face unacceptably high risks for anesthesia-related maternal mortality.1,3,10The following actions might further reduce mortality: (1) recognize the relatively high potential for anesthetic disaster during emergence and recovery from general anesthesia; (2) ensure appropriate monitoring throughout the perioperative period, particularly for postoperative hypoventilation and airway obstruction; (3) use strategies to maximize safety for obese and morbidly obese women; and (4) advocate for workforce policies that ensure appropriate involvement of anesthesiologists in developing clinical protocols and in delivering patient care. Future research is needed to evaluate the epidemiology of care management problems in obstetric anesthesiology and the efficacy of systems solutions to improve peripartum maternal safety.

The authors thank Violanda Grigorescu, M.D., M.P.H. (State of Michigan Maternal and Child Health Epidemiologist), James Gell, M.D. (Head of the Michigan Maternal Mortality Surveillance Medical Review Committee), and Norma Killilea, R.N. (Michigan Maternal Mortality Surveillance Coordinator), for facilitating access to the primary case records; the Michigan Department of Community Health for supporting Maternal Mortality Surveillance efforts for the past 55 years; and Frank Anderson, M.D, M.P.H. (Assistant Professor), and Timothy R.B. Johnson, M.D. (Chair), from the University of Michigan Department of Obstetrics and Gynecology, Ann Arbor, Michigan, for reviewing the manuscript and the definition of anesthesia-related maternal mortality.