On a global basis, complications of pregnancy are a major source of mortality.1,2 Although the risk of dying from pregnancy in the United States decreased approximately 99% during the 20th century3 and is now at a level many consider irreducible, several facts indicate that further reductions are possible. Twenty-nine other developed countries have a maternal mortality ratio (maternal deaths per 100,000 livebirths), as measured by vital statistics, lower than that of the United States.1 In addition, we know that certain racial and ethnic groups have higher risks of pregnancy-related death—African-American women almost 4 times greater than white women and foreign-born Hispanic and Asian/Pacific Islander women 60–70% greater than non-Hispanic white women.4–5 Finally, several studies from Europe and the United States have found 1 of 3 to 2 of 3 pregnancy-related deaths to be preventable.6–12
To reduce the number of women who die from complications of pregnancy, one must first understand why the women died and how the deaths could have been prevented The availability of data from the state of North Carolina and the work of the North Carolina Pregnancy-Related Mortality Review Committee offered an opportunity to undertake such a project in a well-defined population. We first identified and then used all available pertinent information to review all pregnancy-related deaths in the state during a 5-year period. For each death we then determined the cause of death, whether the death had been potentially preventable, and if so, whether it could have been prevented through changes in preconception care, patient actions, the health care system, or quality of care.
MATERIALS AND METHODS
To identify all pregnancy-related deaths for 1995–1999, the North Carolina State Center for Health Statistics used 4 computerized data sets—the death, birth, and fetal death certificate files and the state-wide computerized hospital discharge data base. It used 3 methods to ascertain possible pregnancy-related deaths. First, it found all deaths with an International Classification of Diseases, 9th Revision, Clinical Modification pregnancy-related cause of death code (630–676). Second, it electronically linked the state's birth, fetal death, and death files to find all deaths that occurred during or within 1 year of delivery of a livebirth or stillbirth. Finally, the Center used its hospital discharge database to identify all discharges in 1995–1999 that included any International Classification of Diseases, 9th Revision, Clinical Modification diagnostic or procedure code indicating pregnancy and a discharge status of deceased.13
We used the definition of pregnancy-related death developed by the American College of Obstetricians and Gynecologists (ACOG)/Centers for Disease Control and Prevention (CDC) Maternal Mortality Study Group14: a death during or within 1 year of the end of pregnancy that was caused by a pregnancy complication or its treatment. All deaths to women of reproductive age identified as occurring during or within 1 year of the end of pregnancy were independently reviewed and classified as pregnancy related or not pregnancy related by 3 of the authors (C.J.B., W.M.C., M.A.H.) according to previously published guidelines.14 Discrepancies in classification among reviewers were discussed, and unanimous agreement was reached for all cases.
The North Carolina Pregnancy-Related Mortality Review Committee consists of physicians actively involved in the care of pregnant women and representing 7 of the 8 perinatal regions of the state, an obstetrical anesthesiologist who also represents the North Carolina State Medical Society, members of the North Carolina Division of Public Health Women's and Children's Health Section, and the cochairs of the ACOG/CDC Maternal Mortality Study Group. The committee is responsible for reviewing all pregnancy-related deaths and determining their preventability. For this review, the definition of “preventable death” used was adopted from that used by the Maternal Mortality and Morbidity Review Committee of the Massachusetts Department of Public Health, Bureau of Family and Community Health.15 A death was considered preventable “if the death may have been averted by 1 or more changes in the health care system related to clinical care, facility infrastructure, public health infrastructure and/ or patient factors.” For each case classified as potentially preventable, a determination was made regarding the way(s) in which the death might have been averted.
We identified 4 areas in which different actions might have prevented the deaths. We considered the death potentially preventable through preconception care and counseling (“preconception care”) when the woman had any of selected serious medical conditions diagnosed before becoming pregnant, but there was no evidence she had received counseling about the risk of pregnancy before she conceived. We considered the death preventable through changes in patient behavior (“patient actions”) when the patient had not adhered to medical advice, such as refusal of necessary treatment, including the use of blood products; failure to follow up with care or recommended therapies; or failure to seek care in a timely fashion for whatever reason. We considered deaths preventable through changes in the structure and functioning of the health care system (“system factors”) in those cases where overall functioning of the health care system, such as lack of a plan for patient follow-up or transfer, led to her receiving nonoptimal care. We considered deaths preventable through an improvement in clinical care (“quality of care”) when, in the opinion of the review committee members, the care provided did not meet the expected standards that they felt should be available at a facility of that given level.
The review of deaths proceeded in 2 steps. First, using data available from the death certificates, matching birth or fetal death certificates and, where readily available, autopsy or medical examiner reports and medical records, an initial deidentified clinical summary for each pregnancy-related death was prepared by 1 of the authors (M.A.H.). At the committee's first meeting, all cases underwent a primary level of review. Those cases determined to have sufficient information at that time were classified as preventable or not preventable. Cases whose preventability was still undetermined underwent a second level of review. Further information was obtained from the Office of Chief Medical Examiner for the State and from review of medical records from the health care facilities involved. In some cases, committee members obtained information from treating clinicians. Using this additional information, an enhanced case summary was prepared, and at the second review committee meeting a final determination of the potential preventability of each case was made.
We grouped causes of death into 12 categories: hemorrhage, pregnancy-induced hypertension (including hemolysis, elevated liver enzymes, low platelets), infection, microangiopathic hemolytic syndromes (including thrombotic thrombocytopenia purpura and hemolytic uremic syndrome), cardiovascular disease, preexisting chronic medical conditions, amniotic fluid embolism, pulmonary embolism, embolic and hemorrhagic cerebrovascular accident, cardiomyopathy, choriocarcinoma, and other conditions. The committee agreed upon rules regarding the preventability of deaths for 3 scenarios. Because of their rapid course and lack of uniformly effective therapy, we considered all deaths from amniotic fluid embolism and microangiopathic hemolytic syndrome to be not preventable. In cases where patients or their families refused the use of blood products, we considered deaths from hemorrhage to be potentially preventable because of patient actions.
For 52% of the deaths, the death certificate indicated that autopsy results were used to determine the cause of death. Full state medical examiner or hospital autopsy reports were obtained on 30% of cases. In 43% of cases, hospital records for the case were reviewed by a member of the committee.
We calculated 95% confidence intervals for the pregnancy-related mortality ratios using the Poisson distribution.16 We used the χ2 test for proportions to assess the difference in the proportions of preventable deaths between white and African-American women. The study was approved by the institutional review boards of Wake Forest University Health Sciences and the CDC.
The 108 pregnancy-related deaths identified for the study period were divided fairly equally by year (Table 1). Slightly more than one half of the deaths were among African-American women, and almost one half of the women who died were aged 20–29 years. Almost three quarters of the deaths followed delivery of a liveborn infant.
Of the 105 deaths with an identifiable cause, the most common cause of death was cardiomyopathy, which was responsible for 21% of the deaths (Table 2). The next most common causes of death were hemorrhage (14%), pregnancy-induced hypertension (10%), and cerebrovascular accidents and complications of chronic medical conditions (both 9%).
The committee was able to determine the preventability of 102 of the 105 deaths with a known cause. Of these, 41 (40%), were determined to be preventable through changes in 1 or more areas. In 8 cases, lack of preconception care (“preconception care”) contributed to the death; in 14, patient actions or lack of action contributed (“patient actions”); in 4 cases nonoptimal organization of the health care system was implicated (“system factors”); and in 22 cases the medical care received was felt to be below the standard expected for a facility of the level involved (“quality of care”).
The percentage of deaths that were determined to be preventable varied by the cause of death (Table 2). Approximately 90% of deaths due to hemorrhage or due to complications of chronic diseases were judged to be potentially preventable, as opposed to none of the deaths from cerebrovascular accident, amniotic fluid emboli, or microangiopathic hemolytic syndromes (the latter 2 having been categorized a priori as nonpreventable). In addition, the ways in which deaths could have been prevented varied by cause. Improved medical care (“quality of care”) was most important in preventing deaths from hemorrhage and infection, whereas preconception care could have potentially prevented more than one half of the deaths from chronic medical conditions. Similarly, changes in patient actions could have prevented one half of the deaths from pregnancy-induced hypertension.
For the study period, the pregnancy-mortality ratio was 42.0 (95% confidence interval [CI] 32.0–54.2) per 100,000 livebirths for African-American women in North Carolina and 12.3 (95% CI 9.0–16.4) for white women. If all preventable pregnancy-related deaths had not occurred, the pregnancy-related mortality ratio for African-American women would have been 21.7 (95% CI 14.6–31.0) and for white women 7.6 (95% CI 5.1–10.9). For those deaths for which we were able to determine preventability, the percentage of deaths that were considered to be potentially preventable was 46% for African-American women and 33% for white women, a difference that was statistically significantly different (P < .05).
Using multiple methods to identify pregnancy-related deaths in North Carolina in 1995–1999, we found that 40% of these deaths could potentially have been prevented by changes in 1 or more of the following areas: counseling women with certain serious chronic medical conditions about the risk of pregnancy, better adherence to medical advice by patients, a better-organized system of health care, and improved quality of medical care. The need for higher quality care was a factor in more than one half of the preventable deaths; in one third of the preventable deaths, patient actions were thought to have contributed to the death. Approximately 90% of deaths due to hemorrhage and chronic diseases were deemed potentially preventable. The percent of pregnancy-related deaths among African-American women that were preventable was greater than the percent among white women.
Our findings that 40% of pregnancy-related deaths in North Carolina were potentially preventable and that nonoptimal quality of care was a factor in many of these deaths were very similar to studies performed in other states and other developed countries with low rates of maternal mortality.6–12,17 Using data from the European Concerted Action on Mothers' Mortality and Severe Morbidity Study, researchers found that differences in maternal mortality among the European countries with low maternal mortality rates reflected differences in obstetric care.18 Perhaps the best-known review of maternal deaths, the Confidential Enquiry into Maternal Deaths of the United Kingdom, found “major substandard care” in 50% of direct obstetric deaths between 1997 and 1999.10
To a large extent, the North Carolina committee viewed deaths from hemorrhage as potentially preventable. In the past 2 decades, significant technologic progress has been made in materials and techniques to control acute, excess bleeding, such as the use of prostaglandin analogues, the B Lynch suture, and radiographically directed arterial embolism,19 as well as the early detection of ectopic pregnancy using quantitative β-hCG tests and improved ultrasound imaging. During this same time period, between 1979–1986 and 1991–1997, on a national level we experienced a 33% decrease in the percent of pregnancy-related deaths from bleeding.4 Although we cannot infer direct causality, it is not unreasonable to attribute the decline in deaths from hemorrhage to this improvement in care. Thus, with few exceptions, the committee believed that obstetrics and gynecology practitioners should be prepared to use these techniques to manage hemorrhage. Clinicians should also be attentive to situations in which a patient might be at an increased risk of hemorrhage or problems from blood loss, such as when there is placenta previa in the presence of a previous uterine incision or the patient has stated that she would refuse blood products.
The most frequent cause of pregnancy-related death in North Carolina was cardiomyopathy, a condition increasingly recognized as more states use methods such as computerized data linkages to identify pregnancy-related deaths.20,21 We classified deaths due to cardiomyopathy as preventable or not based on current knowledge and treatment, including the risk of recurrence based on functional cardiac status22 and compliance with current standard treatment (ie, β blockers, diuretics, and other therapies for congestive heart failure) Three quarters of the cardiomyopathy deaths in this study were considered nonpreventable. Of this group, 75% were among women who had previously been diagnosed with a cardiomyopathy who then died suddenly, usually at home or in the emergency room, a clinical picture consistent with arrhythmia. Greater awareness of this entity among clinicians is important for improved identification and treatment, as is clinical research into interventions to prevent these sudden deaths. The latter may include the use of implantable cardioverter-defibrillators, which have recently been shown to decrease mortality among nonpregnant patients with congestive heart failure23 with an ejection fraction similar to that found in peripartum of cardiomyopathy.24 Epidemiologic research into the risk factors for peripartum cardiomyopathy can to help focus these efforts.
Few guidelines are available for preconception counseling for women with serious medical conditions, counseling that might include a recommendation to avoid pregnancy. Maternal–fetal medicine specialists may find themselves in conflict, feeling that their expertise can provide for safe passage through pregnancy for almost any woman with a serious medical problem. In addition, women who receive advice about the risks of pregnancy may choose to accept these risks. However, as the number of older women becoming pregnant increases,25 the number of women with serious medical conditions who desire to conceive will grow. Both preconception care guidelines and the involvement of primary care and specialist physicians can help provide these women with a realistic appraisal of the risks of pregnancy and the information they need to make an informed decision.
Deaths from some causes were considered essentially not to be preventable by virtue of their catastrophic nature and lack of uniformly effective therapy. In addition to deaths from amniotic fluid embolus and microangiopathic hemolytic syndromes, which we had determined a priori not to be preventable, all the deaths in the study from cerebrovascular accident were also found not to be preventable. As medical knowledge and technology advance, however, new therapeutic advances in the prevention, diagnosis and treatment of these conditions will need to be integrated into clinical practice.
The purpose of the maternal mortality review process is to provide information to develop strategies to reduce the number of deaths. As such, the work in North Carolina has many strengths but was also subject to limitations. The case ascertainment performed by the state of North Carolina allowed us to identify all pregnancy-related deaths in a defined population. North Carolina's statewide system of medical examiners reports are public records; these usually contain thorough histories of the events leading up to the deaths. Although North Carolina does not have legislation mandating the release of medical records for pregnancy-related mortality review, a letter of support sent by the Chief of the Women's and Children's Health Section to all hospitals from which we requested records without doubt greatly increased our ability to obtain access to these source of data. However, even with medical records, in-depth information on many patient factors, such as the intendedness of a pregnancy, or reasons that a particular intervention did not take place, would require interviews with family and staff. Currently, this level of inquiry is only beginning to be explored in the United States.26
In our review we looked at potential preventability from a variety of perspectives, including care before, during and after pregnancy and issues around patients, providers, and system, which allowed us to approach prevention from a broader perspective. The review committee included practicing clinicians who were familiar with health care in the state. Although the use of expert opinion is subjective, in this review it represents the consensus of the committee in the framework of the local situation. Without some type of evaluation as to what could have been done differently, we would not be able to identify areas for change.26
Finally, even using all pregnancy-related deaths in the state during a 5-year period, we reviewed a relatively small number of deaths, and our findings are subject to random variation. In addition, to preserve the confidentiality of the women and care providers, we were not able to publish information on individual cases. However, because the goal of the review process is to apply the knowledge gained to improve maternal health, the information we obtained was appropriate for that purpose. The finding that cardiomyopathy is the leading cause of pregnancy-related death in North Carolina has led to the development of a state-wide surveillance system for this condition, as well as an educational program to raise awareness of this entity among the spectrum of health care providers who may encounter patients with it.
Ideally, pregnancy-related mortality review is a function based at the state level, both to facilitate access to necessary information and to institute the needed changes. Considered expert review committees, the goal of these review committees is not to assign blame to individuals or institutions but rather to identify broader opportunities for changes or improvement. Groups considering instituting or expanding their review of maternal deaths should 1) collaborate with the departments of vital statistics and of maternal and child health and medical examiner to optimize the amount of information available for the review process; 2) consider potential preventability from a broad perspective; and 3) seek input and support from their Office of General Counsel in terms of access to needed records, the confidentiality of records and deliberations and protection from discovery, and committee immunity.27
In all of this, the involvement of clinicians, as individuals and representatives of state medical and obstetric and gynecologic societies and ACOG sections, is crucial. The monograph Strategies to Reduce Pregnancy-Related Deaths: From Identification and Review to Action, developed by CDC, ACOG, and other groups, provides guidance on the steps involved in the review process (a pdf file is available from www.cdc.gov/reproductivehealth).14 The oldest and most famous comprehensive system of maternal death review is the Confidential Enquiries into Maternal Deaths in the United Kingdom, whose findings are often used to develop recommendations. An audit of the effect of these recommendations, made by the Royal College of Obstetricians and Gynaecologists, found that clinicians made many of the suggested changes.28 The process of in-depth review of pregnancy-related deaths may provide guidance to help reduce the number of these events.
1. Abou Zahr C. Maternal Mortality in 2000: Estimates Developed by WHO, UNICEF and UNFPA. Geneva (Switzerland): Department of Reproductive Health and Research, World Health Organization; 2004.
2. Royston E, Armstrong S. Preventing Maternal Deaths. Geneva (Switzerland): World Health Organization; 1989.
3. Centers for Disease Control and Prevention (CDC). Healthier mothers and babies [published erratum appears in MMWR Morb Mortal Wkly Rep 1999;48:892]. MMWR Morb Mortal Wkly Rep. 1999;48:849–58.
4. Berg CJ, Chang J, Callaghan WM, Whitehead SJ. Pregnancy-related mortality in the United States, 1991-1997. Obstet Gynecol 2003;101:289–96.
5. Centers for Disease Control and Prevention (CDC). Pregnancy-related deaths among Hispanic, Asian/Pacific Islander, and American Indian/Alaska Native Women—United States, 1991-1997. MMWR Morb Mortal Wkly Rep 2001;50:361–4.
6. Mertz KJ, Parker AL, Halpin GJ. Pregnancy-related mortality in New Jersey, 1975 to1989. Am J Public Health 1992;82:1085–8.
7. Hogberg U, Innala E, Sandstrom A. Maternal mortality in Sweden, 1980-1988. Obstet Gynecol 1994;84:240–4.
8. Schuitemaker NW, Gravenhorst JB, Van Geijn HP, Dekker GA, Van Dongen PW. Maternal mortality and its prevention. Eur J Obstet Gynecol Reprod Biol 1991;42:S31–5.
9. Bouvier-Colle MH, Varnoux N, Breart G. Maternal deaths and substandard care: the results of a confidential survey in France: Medical Expert Committee. Eur J Obstet Gynecol Reprod Biol. 1995;58:3–7.
10. Lewis G, Drise J, Botting B, Carson C, Cooper G, Hall M, et al. Why Mothers Die: Report on Confidential Enquiries into Maternal Deaths in the United Kingdom, 1997-1999. London (UK): Department of Health on behalf of the Controller of Her Majesty's Stationery Office; 2001.
11. Kilpatrick SJ, Crabtree KE, Kemp A, Geller S. Preventability of maternal deaths: comparison between Zambian and American referral hospitals. Obstet Gynecol 2002;100:321–6.
12. Jacob S, Bloebaum L, Shah G, Varner MW. Maternal mortality in Utah. Obstet Gynecol 1998;91:187–91.
13. Harper MA, Byington RP, Espeland MA, Naughton M, Meyer R, Lane K. Pregnancy-related death and health care services. Obstet Gynecol 2003;102:273–8.
14. Berg C, Danel I, Atrash H, Zane S, Bartlett L, editors. Strategies to reduce pregnancy-related deaths: from identification and review to action. Atlanta (GA): Centers for Disease Control and Prevention; 2001.
15. Pregnancy-associated mortality: medical causes of death, 1995–1998. In: Nannini A, Weiss J. Maternal Mortality and Morbidity Review in Massachusetts: A Bulletin for Health Care Professionals (no. 1, May 2000). Available at http://www.mass.gov/dph/fch/safemoms/preg2000.pdf
. Accessed March 18, 2000.
16. Matthews TJ, Menacker F, MacDorman MF. Infant mortality statistics from the 2002 period linked birth/infant linked data set. National vital statistics reports. Vol 53, No 10. Hyattsville (MD): National Center for Health Statistics; 2004.
17. Geller SE, Rosenburg D, Cox SM, Brown ML, Simonson L, Driscoll CA, et al. The continuum of maternal morbidity and mortality: factors associated with severity. Am J Obstet Gynecol 2004;191:939–44.
18. Wildman K, Bouvier-Colle MH, MOMS Group. Maternal mortality as an indicator of obstetric care in Europe. BJOG 2004;111:164–9.
19. Pahlavan P, Nezhat C, Nezhat C. Hemorrhage in obstetrics and gynecology. Curr Opin Obstet Gynecol 2001;13:419–24.
20. Whitehead SJ, Berg CJ, Chang J. Pregnancy-related mortality due to cardiomyopathy: United States, 1991-1997. Obstet Gynecol 2003;102:1326–31.
21. Nannini A, Weiss J, Goldstein R, Fogerty S. Pregnancy-associated mortality at the end of the twentieth century: Massachusetts, 1990-1999. J Am Med Womens Assoc 2002;57:140–3.
22. Elkayam U, Tummala PP, Rao K, Akhter MW, Karaalp IS, Wani OR, et al. Maternal and fetal outcomes of subsequent pregnancies in women with peripartum cardiomyopathy. N Engl J Med 2001;344:1567–71.
23. Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R, et al. Amiodarone or an implantable cardioverter-defibrillator for congestive heart failure [published erratum appears in N Engl J Med 2005;352:2146]. N Engl J Med 2005;352:225–37.
24. Pearson GD, Veille J-C, Rahimtoola S, Hsia J, Oakley CM, Hosenpud JD, et al. Peripartum cardiomyopathy: National Heart, Lung, and Blood Institute and Office of Rare Diseases (National Institutes of Health) workshop recommendations and review. JAMA 2000;283:1183–8.
25. Callaghan WM, Berg CJ. Pregnancy-related mortality among women aged 35 years and older, United States, 1991-1997. Obstet Gynecol 2003;102:1015–21.
26. Sachs BP. A 38-year-old woman with fetal loss and hysterectomy. JAMA 2005;294:833–40.
27. Wright RF, Smith JC. State level expert review committees—are they protected? Public Health Rep 1990;105:13–23.
28. Benbow A, Maresh M. Reducing maternal mortality: reaudit of recommendations in reports of confidential inquiries into maternal deaths. BMJ 1998;317:1431–2.
Figure. No caption available.