Best Practice & Research Clinical Obstetrics and Gynaecology Vol. 22, No. 3, pp. 517–531, 2008 doi:10.1016/j.bpobgyn.2007.10.004 available online at http://www.sciencedirect.com
7 Maternal mortality in the United States Christopher T. Lang *
Fellow Maternal–Fetal Medicine, The Ohio State University College of Medicine, Columbus, USA
Jeffrey C. King
Clinical Professor Maternal–Fetal Medicine, The Ohio State University College of Medicine, Columbus, USA
Despite a significant improvement in the US maternal mortality ratio since the early 1900s, it still represents a substantial and frustrating burden, particularly given the fact that – essentially – no progress has been made in most US States since 1982. Additionally, the US Centers for Disease Control and Prevention has stated that most cases are probably preventable. Two disheartening issues within this topic include a gross underestimation of the magnitude of maternal mortality – particularly before 1987, but which likely persists to a lesser degree today – and the continued significant racial disparity in maternal mortality. Explanations for the plateau in maternal mortality include the recent trend of delayed childbearing, with the potential accompanying complications associated with older reproductive age (particularly over 35 years) and multiparity. The impressive increase in multifetal pregnancies related to delayed childbearing and assisted reproductive technology also plays a role. Finally, peripartum cardiomyopathy has become an increasingly recognized source of maternal mortality. Pregnancy-related mortality is largely accounted for by thromboembolic disease, hemorrhage, hypertension and its associated complications, and infection. However, since the inclusion of maternal deaths occurring after 42 days post-delivery as pregnancy related, traumatic injuries – including homicides and suicides – are an alarming source of maternal mortality. An especially important contemporary issue to consider within this topic is cesarean delivery ‘‘on maternal request’’, opponents of which cite concerns not only for immediate morbidity and mortality increased over that associated with a vaginal birth, but also for potential morbidity and mortality associated with future pregnancies.
* Corresponding author. Department of Obstetrics and Gynecology, The Ohio State University College of Medicine, 545 Means, 1654 Upham Drive, Columbus, OH 43210, USA. Tel.: þ1 614 293 8696; Fax: þ1 614 293 5712. E-mail address: [email protected]
(C.T. Lang). 1521-6934/$ - see front matter ª 2007 Elsevier Ltd. All rights reserved.
518 C. T. Lang and J. C. King
One particularly appealing opportunity to reduce maternal mortality is to recognize, examine, and learn from so-called ‘‘near-miss’’ cases. Key words: African–Americans; cardiomyopathies; cesarean section; homicide; maternal mortality; parity; pregnancy complications; suicide; twins; United States.
OVERVIEW Expectant couples, young nurses and medical students, and junior residents rarely view pregnancy as a potentially life-threatening period in a reproductive woman’s life. Unless one has firsthand experience with a maternal death, or takes a moment to review the maternal mortality literature, the burden it represents is easily neglected. This is understandable, as the absolute risk of a pregnancy-related death in the US is very low – estimated at 11.8 deaths per 100,000 live births during the period 1991–1999, which represents a 99% reduction since 1900 (Figure 1).1 However, this is a long way from the Healthy People 2000 goal of 3.3 deaths per 100,000 live births, which was readopted as a goal for Healthy People 2010. Also, when maternal mortality is viewed as an indicator of either national or women’s health care, one discovers some alarming findings: First, since 1982, the maternal mortality ratio has in essence remained unchanged, despite advances in diagnosis and acute critical care. Second, it is likely that, to this day, maternal mortality is underestimated, although the improved surveillance embodied by the Pregnancy-Related Mortality Surveillance System (PMSS) introduced in 1987 by the US Centers for Disease Control and Prevention (CDC) has increased detection. For example, when one relies solely on International Classification of Diseases (ICD) codes, maternal mortality might be underestimated by as much as 93%, as reported in one study.2 Third, a significant proportion of this burden is borne by African–Americans and other minorities; the risk of pregnancy-related mortality is four-fold higher in African–Americans.1
Figure 1. Maternal mortality rates in the United States 1915–2003 (from Ref. 49).
Maternal mortality 519
Older reproductive age represents a significant risk for maternal mortality, with women aged 40 years and older having a two-fold greater risk than women aged 35–39 years, and a four-fold greater risk than those aged 30–34 years.1 The four most common culprits with regards to pregnancy-related mortality are thromboembolic disease in 20% of cases, hemorrhage in 17% of cases, hypertension and the associated complications in 16% of cases, and, in cases ending in stillbirth, infection (19%).1 Peripartum cardiomyopathy and cardiovascular medical conditions exacerbated by pregnancy have become increasingly common, noted in 9% and 34% of cases, respectively.1 Not surprisingly, pregnancy-related mortality appears to be inversely related to extent of education (with the lowest ratio demonstrated by women with at least 12 years of education) and is three- to four-fold more common in women who sought no prenatal care as compared to those who sought at least some care.1 Ultimately, and for a variety of reasons, modern medicine has failed to meet the Healthy People 2000 goal, and the same 2010 goal at times seems insurmountable. Nonetheless, even if efforts fall short of this goal, each maternal death that is prevented potentially means one less child without a mother, one more contributing member of society, and perhaps many fewer cases leading to severe morbidity or ‘‘near-misses’’. HISTORICAL PERSPECTIVE ON SURVEILLANCE AND DEFINITIONS The first study reporting the feasibility of nationwide surveillance of maternal mortality was published by Kaunitz et al in 1985.3 Copies of death certificates for deaths categorized as maternal occurring in the period 1 January 1974 to 31 December 1978 were requested from the health departments of each US State and evaluated, as were additional cases identified by various State agencies but not formally recorded by the National Center for Health Statistics (NCHS) as a maternal death. The ICD-9 was the classification system used at the time, which defined maternal death as ‘‘the death of a woman while pregnant or within 42 days of the termination of the pregnancy, irrespective of the duration and the site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management but not from accidental or incidental causes.’’4 Nonetheless, these investigators were able to collect not only those cases as defined by the ICD-9 – and similarly by the World Health Organization (WHO), the American College of Obstetricians and Gynecologists (ACOG), and the NCHS – but also those that occurred up to 1 year after delivery and were fortunately categorized as ‘‘maternal’’ by individual State health departments, as was done in the UK through the system of triennial Confidential Enquires. Ultimately, the inclusion of these later maternal deaths allows one to account for those women who died several weeks to months after the initial insult, which was indeed either directly or indirectly associated with pregnancy. Also, this allowed mortality statistics to be reported as ‘‘rates’’3 (later to be referred to as a ‘‘ratio’’) or as maternal deaths per 100,000 live births, which proved to be a much more manageable denominator than 100,000 reproductive age women. Atrash et al later reported on maternal mortality during the period September 1979 to 1986, on behalf of the CDC in collaboration with ACOG.5 Once again, individual State health departments sent copies of maternal death certificates to the CDC. Maternal death reflected: ‘‘1) complications of the pregnancy itself, 2) a chain of events that was initiated by the pregnancy, or 3) aggravation of an unrelated condition by the
520 C. T. Lang and J. C. King
physiologic effects of the pregnancy.’’5 Later, these instances would be categorized as ‘‘pregnancy-related’’ by the CDC and ACOG, defined as the ‘‘death of a woman while pregnant or within one year of the termination of pregnancy, regardless of duration and site of pregnancy, from any cause related to or aggravated by her pregnancy or its management’’.6 ‘‘Pregnancy-associated’’ death would include those with the same temporal restriction but due to any cause. Atrash et al also made use of birth or fetal death certificates, which could be linked to maternal deaths.5 Later, this would be a particularly useful strategy employed by the CDC’s PMSS to capture additional maternal mortality cases. Beginning in 1987, the PMSS used experienced clinicians to review (including reading notes written in the margins of death certificates) and to code maternal mortality data obtained from individual State health departments. Data from 1987–1990 were published by Berg et al.7 As mentioned above, linkages were made with birth or fetal death certificates to establish temporal relationships, and further data and information were requested from State maternal mortality review committees. Additionally, all media were monitored to capture cases that would otherwise be missed. Of additional help in identifying cases during this period was the increasing presence and use of pregnancy check-boxes on death certificates. Later, in 1991, the PMSS requested that all States amend their death certificates such that all deaths occurring within 1 year of the conclusion of pregnancy are categorized as maternal deaths. Also, some States began using the Revised United States Standard Certificate of Death, which makes identification of maternal mortality much clearer. Data from 1991–1999, presented as a CDC Maternal Mortality Weekly Report (MMWR), were published by Chang et al.1 During this period, 4200 pregnancy-related deaths were recorded, providing the above-mentioned pregnancy-related mortality ratio of 11.8 deaths per 100,000 live births. Sixty per cent of these deaths followed a live birth, most commonly as a result of embolism or hypertensive disease. Again, African–American women and those aged 35 years and above demonstrated greater risk for a pregnancy-related death. More recently, the ICD-10 was introduced, which includes additional coding for ‘‘late maternal death’’ (i.e. deaths occurring more than 42 days after a pregnancy’s conclusion) and ‘‘pregnancy-related death,’’ which, in contrast to the CDC/ACOG definition, takes into account death due to any cause (i.e. ‘‘pregnancy-associated death’’ according to CDC/ACOG).8 Although this introduces some confusion, the ICD-10 coding for maternal deaths remote from the conclusion of pregnancy is very appropriate, as noted earlier by Kaunitz et al.3 UNDERREPORTING As might be expected from the preceding section, properly documenting a maternal death can be a confusing task, especially for those without training in completing a death certificate. In addition, even when a death certificate is completed fully and correctly, if a given situation fails to fulfill a particular definition of maternal death it is not ultimately labeled as such, and therefore goes unnoticed. Also, maternal deaths associated with a very early pregnancy, such as cases involving ectopic pregnancies or surgical abortions (i.e. those not ending in a live birth or stillbirth) can easily be categorized incorrectly. The magnitude of underreporting was alluded to earlier. In North Carolina in 1988 and 1989, 29 additional maternal death cases were discovered with enhanced surveillance involving matching death certificates with birth records.9 Deneux-Tharaux et al reported underestimation of mortality causally related to
Maternal mortality 521
pregnancy to be as much as 93% in Massachusetts during the period 1999–2000, when the physicians relied solely on ICD coding to capture maternal mortality as opposed to computer-based linkages and manual review of any death certificate involving a reproductive age woman.2 A particularly intriguing account of this issue was presented by Horon in 2005.10 Only 62% of 129 identified maternal death cases in Maryland in the period 1993– 2000 were captured by death records alone; 62% were captured via record linkages alone and 64% via medical examiner records alone. The longer the duration from pregnancy conclusion to maternal death, the more likely it was to go unreported; for example, 75% of deaths occurring 31–42 days after pregnancy termination were not categorized properly. Even in cases in which a woman died while pregnant, 12 of 28 cases were missed. Deaths due to cardiovascular disorders were the most likely to go unreported (56%). Deaths in teenage mothers and in mothers over the age of 40 were also significantly underestimated. This study ultimately provided further support for the use of multiple linked data sources to capture maternal deaths most efficiently. Underestimation is not the only problem. Various data-collection agencies report substantially different estimates of maternal mortality. This was illustrated by MacKay et al11, who compared the two national systems responsible for reporting maternal mortality: the CDC/NCHS and the PMSS. In the period 1995–1997, 898 ‘‘maternal deaths’’ were reported by the National Vital Statistics System (NVSS) on behalf of the CDC/NCHS (for a maternal mortality ratio of 7.5) and 1387 ‘‘pregnancy-related deaths’’ were reported by the PMSS (for a maternal mortality ratio of 11.9). Of the total, 6% were identified by the NVSS only (simply due to the fact that these death certificates were not submitted to the PMSS for review) and 40% by the PMSS only. There are two clear reasons for this discrepancy. First, as described earlier, the PMSS temporally defines a maternal death as up to 1 year following pregnancy termination, whereas the NVSS interval is 42 days. Second, consistent with the recurrent theme, the more sources that are used and the more time devoted to capturing and properly categorizing maternal deaths, the more likely one is to accurately portray real-life statistics. In an ideal world, maternal mortality statistics would be recorded quickly, accurately, and reliably by one national agency based on a standard, realistic, and useful definition, with cases submitted via a user-friendly electronic database asking the user very specific questions to determine if a pregnancy was involved and not requiring the user to fill in blank spaces. RACIAL DISPARITY One particularly disheartening fact stands out from the others – minority women (especially African–Americans) are much more likely than white women to die as a result of pregnancy. The 18 June 1999 MMWR reported that black-to-white maternal mortality ratios varied from 2.6 to 6.3 across the US between 1987 and 1996.12 From 1987 to 1991, the ratio among African–American women was 18.8, and from 1992–1996 it was 20; the corresponding ratios for white women were 5.5 and 5.0. Harper et al reported an adjusted odds ratio for African–American versus white women of 2.65 (95% confidence interval [CI] 1.73–4.07) in the period 1992–1998 in North Carolina following stepwise logistic regression taking into account multiple confounders including socioeconomic and medical risk factors.13 In terms of the causes of maternal mortality with respect to race, cardiomyopathy, hemorrhage, and pneumonia and adult
522 C. T. Lang and J. C. King
respiratory distress syndrome were more commonly noted in African–American women, with relative risks of 4.6 (95% CI 2.2–9.9), 4.9 (95% CI 1.2–19.4), and 6.1 (95% CI 1.2–31.3), respectively. The 11 May 2001 MMWR evaluated maternal mortality among other minorities nationally in the period 1991–1997.14 Among native Americans/native Alaskans, the maternal mortality ratio was 12.2; it was 11.3 for Asians/Pacific Islanders, and 10.3 for Hispanic women. All these are lower than the ratio seen in African–American women, but greater than the ratio for white women. Interestingly, Hispanic and Asian/Pacific Islander women born outside the US had higher ratios than those born in the US, by as much as 50%. A closer look at maternal mortality in Hispanic women was presented by Hopkins et al in 1999.15 In the period 1979–1992, the ratio for Hispanic women was exactly that reported in the aforementioned MMWR: 10.3.14 Subgroup analysis revealed ratios of 9.7 for Mexican women, 7.8 for Cuban women, and 13.4 for Puerto Rican women.15 The leading cause of death in Hispanic women following pregnancy conclusion was hypertensive disease; these women experienced a threefold increased risk of death due to this complication. Whether race and ethnicity are risk factors in and of themselves for maternal mortality is a matter of debate. Rather, they might simply be societal constructs that reflect more meaningful factors such as culture, economics, and baseline health, which might more directly relate to the availability of, seeking for, and receipt and benefit of health care. Minorities, particularly African–Americans, tend to be more often socioeconomically disadvantaged and therefore less likely to receive regular prenatal care (e.g. perhaps because of an inability to find or afford reliable transportation), and might even be skeptical of what professional healthcare has to offer, especially if previous experiences have been suboptimal. This, combined with the fact that African–Americans are more likely to develop various complications, such as hypertension or peripartum cardiomyopathy, clearly predisposes to pregnancy-related deaths. Unfortunately, as mentioned above, even when one controls for these potential confounders, African–American women still bear a significant proportion of the maternal mortality burden.13 The more troubling issue is to what degree racial inequality in health care comes into play. One fact is painfully clear – there is a tendency for minorities to receive substandard care, not just in obstetrics but in all other medical disciplines.16 Opponents of this statement might propose that, in actuality, African–Americans receive quality care but are more likely to present with more significant disease and at later stages of disease, as suggested by a greater likelihood of antepartum hospital admission.17 Therefore, one might conclude that more attention needs to be paid to promoting early and often prenatal care for the African–American population, particularly for those with concerning epidemiologic and medical risk factors, especially incorporating prepregnancy counseling. However, a particularly thought-provoking study presented in 2000 by Saftlas et al suggests that this is not the case.18 In the period 1979–1986, nationwide maternal mortality data were collected with particular attention paid to the white and African–American populations. Subjects were placed into one of four groups, depending on birth weight and live birth order – the two variables that most significantly interacted with race in terms of maternal mortality. Interestingly, the greatest disparity in the maternal mortality ratio was demonstrated by the lowest-risk group (i.e. those women of low-to-moderate parity delivering appropriate-for-gestational age infants), which persisted even after simultaneously adjusting for sociodemographic and reproductive risk factors. This finding perhaps speaks to differences between African–Americans and whites that have yet to gain our full attention, such as stress levels13,18, nutrition18, family structure18, and genetics19, which might make even
Maternal mortality 523
a ‘‘low-risk’’ pregnant African–American woman more susceptible to life-threatening complications. One final comment deserves mention and was briefly alluded to earlier. If there truly are inherent differences between minorities and whites, discovering these would require categorizing individuals on a purely objective basis (i.e. microbiologically19) and not simply by subjective reports of one’s ‘‘race’’ or ‘‘ethnicity’’, which are perhaps societal constructs as best. There is likely to be a great degree of genomic overlap among individuals of seemingly distinct races, and differences among those grouped into one particular race (e.g. African–American, Hispanic, or Asian/Pacific Islander). Nonetheless, such racial categories are each distinct enough to be potent predictors of outcome, as studies of maternal mortality attest, but to appreciate these differences enough to yield preventative and treatment strategies is yet to be realized. DELAYED CHILDBEARING As mentioned above, one of the reasons why there has been no apparent improvement in maternal mortality is the fact that older women are becoming increasingly represented in the pregnant population. In the period 1976–1997 there was a 74% increase in pregnancy rates in women aged 35–39 years, and a 38% increase in women aged 40 years and older.20 Clearly, older women are more likely to have disease prior to conception and/or to develop disease during pregnancy. In terms of excess risk for pregnancy-related mortality, conditions such as hemorrhage, cardiomyopathy, thromboembolic disease, and other medical conditions, especially cardiac, are to blame in older women.21 In the previously mentioned 21 February 2003 MMWR, women aged 35–39 years demonstrated a maternal mortality ratio of 21.6, consistent with a risk ratio of 2.5 (95% CI 2.0–3.2) when compared to younger pregnant women, and for women aged greater than 40 years, a ratio of 45.4 was reported, consistent with a risk ratio of 5.3 (95% CI 4.2–6.6).1 Again worth noting is the racial disparity noted in this analysis. When matched with women of similar older ages, African– American women continue to demonstrate a markedly increased risk for maternal mortality. A closer look at this particular issue was reported by Callaghan and Berg in 2003.21 Pregnancy-related deaths as collected by the PMSS among women aged 35 years and over across the US were analyzed and compared with a younger patient population in the period 1991–1997. Ultimately, regardless of parity, extent of prenatal care, and level of education, the pregnancy-related mortality ratio was still higher for older women, with women 40 years of age and older demonstrating a two-fold greater risk as compared to women aged 35–39 years. A particularly interesting finding in this study came to light when the investigators looked at the actual causes of pregnancy-related death in this population. Among older white women, hemorrhage was a large source of mortality, with more than half being secondary to ectopic pregnancy, uterine atony, and placental implantation abnormalities (e.g. placenta accreta). Most strikingly, white women aged 40 years and older were 7.9 times more likely to die from hemorrhage and 15 times more likely to die from a placental implantation abnormality; not surprisingly, the corresponding figures for African–American women were substantially higher. As mentioned earlier, older women are more likely to present for care already pregnant with chronic disease. Thirty-nine percent of deaths in this study were related to cardiovascular conditions, with women aged 35–39 years 3.9 times more likely to die from cardiovascular disease compared to a younger pregnant
524 C. T. Lang and J. C. King
population, and for women aged 40 years and older, 8.2 times more likely. Among older African–American women, hypertension and cerebrovascular accidents accounted for a substantial proportion of maternal mortality. The excess pregnancy-related mortality secondary to placental implantation disorders described above perhaps offers the most straightforward opportunity thus far in this chapter for making a substantial improvement in mortality rates. Prenatal diagnosis of abnormal placental implantation is becoming increasingly possible with a combination of ultrasound and MRI. Also, there are typically clear risk factors for this complication, which are easily recognized. Being prepared for such conditions, formulating operative strategies, and alerting the blood bank and interventional radiology ahead of time reduces mortality as compared to women in whom the diagnosis is not made until at the time of delivery.22 To some degree, a difficulty experienced in further reducing maternal mortality can be accounted for by the dramatic increase in multifetal gestations seen over the course of the last two decades. Twin births have increased by 55% and for triplet and higher-order gestations by 388% in the period 1979–2000, largely as a result of assisted reproduction technologies, but also due to advancing maternal age.23 It is well-known that multifetal pregnancies are associated with increased risks for complications over that seen in singleton gestations – preeclampsia and thromboembolic disease in particular. This issue was most recently explored by MacKay et al in 2006.24 The 1979–2000 PMSS data were reviewed and maternal mortality rates among women with multifetal pregnancies were compared with those with singleton gestations. Among multifetal pregnancies, the rate was 20.8 deaths per 100,000 multifetal pregnancies, representing a risk ratio as compared to singleton pregnancies of 3.6 (95% CI 3.1–4.1). The most common causes for pregnancy-related mortality included thromboembolic disease, hemorrhage, hypertensive disease, and infection, similar to singleton gestations. However, the likelihood of death from these conditions given a multifetal pregnancy was higher, with a relative risk of 4.2 for thromboembolic disease, 3.9 for hemorrhage, and 3.1 for hypertensive disease. Interestingly, as maternal age exceeded 30 years, relative risks for death decreased when compared with singleton gestations. Not surprisingly given the above discussion, African–American women carrying twins and higher-order multiples were almost three times as likely to die as compared to white women. PREGNANCY-RELATED MORTALITY DUE TO SPECIFIC CAUSES Three particular sources of pregnancy-related mortality have received attention in the recent literature – hemorrhage, preeclampsia, and peripartum cardiomyopathy. Hemorrhage was investigated in 1999 by Chichakli et al.25 In the period 1979– 1992, 763 maternal deaths due to hemorrhage associated with intrauterine pregnancies as collected by the PMSS were included for analysis. At the time, hemorrhage surpassed thromboembolic disease as the leading source of pregnancy-related mortality. Placental abruption proved to be the most common hemorrhagic cause of death in 18.5% of cases, with obstetrical lacerations and/or uterine rupture and uterine atony close behind at 16.4% and 15.1% of cases, respectively. African–American women were three times as likely to die as a result of hemorrhage than white women, and women aged 40–49 years were 13-fold more likely to die of a hemorrhagic complication than women aged less than 20 years. Uterine atony accounted for the majority of hemorrhagic deaths at the extremes of age (i.e. less than 20 years
Maternal mortality 525
and greater than 40 years), while placental abruption accounted for the majority in women aged 25–29 years and 35–39 years. Despite advances in ultrasonography to better predict hemorrhagic complications (e.g. as with placenta previa) and the increasing utility and safety of blood product transfusion, this data serves as a stark reminder that all episodes of obstetrical hemorrhage are potentially catastrophic and life-threatening. MacKay et al paid particular attention to preeclampsia and eclampsia as sources of pregnancy-related mortality in a 2001 publication.26 The pregnancy-related mortality ratio due to preeclampsia or eclampsia was 1.5 deaths/100,000 live births in the period 1979–1992, as determined by the PMSS. 19.6% of the 4024 pregnancy-related deaths reported in this time-frame were due to preeclampsia or eclampsia, with each representing an approximately equal proportion of these deaths. Not surprisingly, cerebrovascular hemorrhage was the most common cause of death among these women, with hepatic failure, renal failure, and/or disseminated intravascular coagulation playing significant roles. Across all age groups, African–American women were once again more likely to die as a result of preeclampsia or eclampsia, with the disparity being the greatest with white women in the 30- to 34-year-old age group (a four-fold increased risk). Eclampsia was associated with a case-fatality rate of 71.6, much higher than that seen with preeclampsia (3.4). As was mentioned earlier, the fact that many pregnancy-related deaths are avoidable makes this issue particularly troubling; this is perhaps best illustrated by preeclampsia and eclampsia, deaths from which seem almost inexcusable in this day and age when diagnosis is usually quite straightforward, when in essence a cure is available (i.e. delivery), and when neonatal intensive care allows for delivery and a reasonable expectation of favorable outcomes even at an early gestational age. Also mentioned earlier was the growing percentage of pregnancy-related deaths due to cardiomyopathy, particularly peripartum cardiomyopathy. Whitehead et al presented data in this regard in 2003.27 Once again, the PMSS was used to collect data regarding cardiomyopathy-related maternal deaths in the period 1991–1997. The pregnancy-related mortality ratio due to cardiomyopathy in general was 0.88 deaths/ 100,000 live births, with 70% of the total number of cardiomyopathy deaths specifically secondary to peripartum cardiomyopathy (ratio of 0.4 deaths/100,000 live births in singleton pregnancies). The National Institutes of Health (NIH) define peripartum cardiomyopathy as ‘‘the development of cardiac failure in the last month of pregnancy or within five months of delivery, in the absence of both an identifiable cause of cardiac failure and of recognizable heart disease before the last month of pregnancy.’’28 One especially striking finding concerning the epidemiology of peripartum and other cardiomyopathies is the dramatic increase seen in risk with advancing maternal age and, once again, with marked racial disparity demonstrated, as African–American women are 6.4-fold more likely to die as a result of a cardiomyopathy than white women. One additional risk factor is multiparity, particularly with a live-birth order of four or more. As was discussed in the previous section dealing with underreporting, so-called ‘‘delayed’’ maternal deaths (i.e. those occurring after 42 days postpartum) are no less significant than those occurring earlier and must be considered. This is clearly illustrated by deaths secondary to peripartum cardiomyopathy, as half of the women in Whitehead’s analysis who succumbed to this disease did so between 43 days and 1 year postpartum.27 Also in the light of underreporting, peripartum cardiomyopathy was not given a specific ICD-9 code, leaving categorization to the discretion of the obstetrician or – more commonly – the cardiologist, who might well not be aware that
526 C. T. Lang and J. C. King
the inciting event was in fact pregnancy. ICD-10 coding offers some improvement in that postpartum (not peripartum) cardiomyopathy is recognized as a distinct entity (O90.3), but omits specific coding for an instance of peripartum cardiomyopathy that occurs prior to delivery. Again this example illustrates the utility of more ‘‘user-friendly’’ death certificates, which include a pregnancy check-box to perhaps better remind the physician of this possibility (particularly if it is not the obstetrician), and of the need for a specific ICD code for peripartum cardiomyopathy such that the true incidence of this potentially devastating illness can be appreciated, particularly by distinguishing it from cardiomyopathies due to other causes. After all, peripartum cardiomyopathy is truly a pregnancy-related death, just as much as preeclampsia or an obstetrical hemorrhage. HOMICIDE AND SUICIDE Increasing the temporal relationship between pregnancy and death to 1 year postpartum, as well as the CDC’s recognition of ‘‘pregnancy-associated’’ deaths or those due to any cause while pregnant or thereafter, has led to a better appreciation for the unfortunate number of reproductive-age women who fall victim to fatal injuries, including intentional injuries. The most recent analysis of this issue was presented by Chang et al, in 2005, in which PMSS data was collected in the period 1991–1999.29 In all, 27.1% of all maternal deaths during this time were pregnancy-associated and injury related. Of these, 31% were as a result of homicide, providing a pregnancy-associated homicide ratio of 1.7 deaths/100,000 live births, surpassed only by motor vehicle accidents (44.1%). Suicide accounted for 10.3% of injury-related deaths. African–American women, particularly those aged less than 20 years, demonstrated an increased homicide ratio, with the greatest disparity as compared to white women being 11-fold. Women, regardless of race, who were victims of homicide were more likely to have received minimal or no prenatal care. The majority of these women (52.7%) were killed in the postpartum period; 20.6% occurred in the antepartum period. Firearms accounted for 56.6% of these deaths, followed by stabbing and strangulation. Harper and Parsons focused their attention on injury-related pregnancy-associated death in the period 1 January 1992 to 31 December 1994 in North Carolina.30 A total of 37% of maternal deaths during this time was secondary to injuries, with homicide being the most common cause (36% of these deaths), and most associated with firearms. Once again, a racial disparity was demonstrated, with the relative risk of injuryrelated death among African–American women being 1.7 (95% CI 1.4–2.2) compared with white women. Suicide accounted for 8% of injury-related deaths. This particular study also made it clear that homicide-related death among pregnant women is not just an urban issue. Data from New York City in the period 1987–1991 were analyzed in 1995 and presented by Dannenberg et al.31 Similar to the previously quoted study, injury-related deaths accounted for 39% of maternal deaths during this time, with an astonishing 63% due to homicide, more than half of which involved firearms. In all, 53% of injury-related deaths were among African–American women and 24% among Hispanic women. Suicide accounted for 13% of injury-related deaths. An often-quoted study presented by Fildes et al in 1992 analyzed data from the Cook County Medical Examiner in the period January 1986 to December 1989.32 ‘‘Nonmaternal’’ (i.e. pregnancy-associated) causes of death accounted for the majority of maternal mortality during this time, with 46.3% categorized as ‘‘traumatic’’. Of
Maternal mortality 527
these, 57% were due to homicide; 22.7% of these deaths involved firearms. Suicide accounted for 9% of injury-related deaths. At the crux of this issue is whether, by virtue of being pregnant or having recently been pregnant, a woman is at greater risk for an intentional injury-related death. In a study presented by Krulewitch et al in 2001, this proved true: Homicides among urban minorities were seen more frequently in pregnant or postpartum women, and in women younger than those nonpregnant.33 Another finding that might also support this unfortunate fact is that homicide is frequently associated with previously abusive relationships, such that the perpetrator is well-known to the woman.34 In such instances, given the added physical, emotional, and financial stress that a pregnancy might place on an already unstable relationship, it is perhaps not surprising that an abusive partner would become especially aggressive. Unfortunately, information regarding any relationship between the woman and the perpetrator is not available to the PMSS. This issue speaks to the importance of screening for domestic violence frequently during the course of a woman’s prenatal care, as well as inquiring about violent experiences with people not well known to her. The thought of a pregnant or recently postpartum woman taking her own life can be difficult to comprehend. Nonetheless, for some women in an abusive relationship or carrying an unwanted pregnancy, or giving birth out of wedlock, suicide sadly becomes an option, particularly in less industrialized countries where the social stigma associated with unplanned pregnancy is intense or where abortion is unavailable.35 Also, postpartum depression has become increasingly recognized as a source of self-inflicted injury and suicide. CESAREAN DELIVERY ON MATERNAL REQUEST Most commonly out of fear of a painful labor, neonatal injury, or the risk of urinary or fecal incontinence, some women request a primary cesarean delivery when in fact no medical indications exist to justify such a procedure. In such cases, a cesarean delivery is looked on as a ‘‘prophylactic’’ measure, as originally described by Feldman and Freiman in 1985.36 Opponents to a so-called ‘‘patient-choice cesarean’’ cite concerns for increased morbidity and cost over that seen with vaginal deliveries, as well as the possibility of increased mortality. Currently, perhaps as many as 22% of elective cesarean deliveries are by virtue of a woman’s request.37 Literature examining the mortality associated with cesarean delivery versus that seen with vaginal delivery is in large part limited by not distinguishing between cesarean deliveries performed under controlled, stable circumstances and those performed for emergent indications. Also, patient populations in these studies are sufficiently different from each other such that results are valid only within each study. Nonetheless, elective cesarean deliveries are associated with a maternal mortality rate of up to 3.4 deaths per 100,000 deliveries, whereas the more emergent procedure is associated with a rate approximately four-fold higher.38 One of the first studies regarding this issue and lending support to elective cesarean delivery was presented by Sachs et al in 1988.39 In Massachusetts over the period 1954–1985, the cesarean-related death rate was 5.8 per 100,000 deliveries, whereas for vaginal deliveries it was 10.8 per 100,000 deliveries. In the first study to distinguish elective from emergent cesarean deliveries, Lilford et al reported a specific elective cesarean-related death rate of 23 per 100,000 deliveries – almost four-fold higher than that seen in the vaginal delivery group, in the period 1975–1986 in South Africa.40
528 C. T. Lang and J. C. King
In 1990, in the UK, similar findings were reported by Hall, who reported a 5.5-fold increased risk for death with elective cesarean delivery as compared to a vaginal birth.41 A seven-fold increased risk was reported in the Netherlands by Schuitemaker et al in 1997, but did not stratify cesarean delivery type.42 Finally, Lydon-Rochelle et al concluded, based on the results of their 2001 retrospective cohort analysis in Washington State, that ‘‘cesarean delivery might be a marker for serious preexisting comorbidities associated with increased mortality rather than a risk factor for death in and of itself’’ after logistic regression analysis failed to support an increased risk of death with cesarean delivery by controlling for maternal age, underlying disease, and marital status.43 Ultimately, cesarean delivery, whether elective or emergent, is likely to involve an increase in risk of maternal mortality, although the absolute risks in this day and age remain low. Nonetheless, as the cesarean delivery rate continues to rise, a growing proportion of cases of pregnancy-related mortality will be accounted for by this procedure. How much of a role patient-choice cesarean delivery will play in this remains to be seen. LEARNING FROM ‘‘NEAR-MISSES’’ Maternal mortality is perhaps best viewed as the ‘‘tip of the iceberg’’ or the extreme end of a long spectrum following severe morbidity and life-threatening ‘‘near-misses’’. In either case, the implication is that maternal mortality is a culmination of events involving action or inaction on the part of the patient, physician, and/or State healthcare agencies, which ultimately pave the way for a maternal death. Cases in which lifethreatening events take place but the patient survives are underappreciated. Undoubtedly, lessons can and should be learned from these near-miss cases. There is certainly plenty of room for improvement: in 2000, the US ranked 30th in terms of worldwide maternal mortality.44 One particularly alarming finding when it comes to preventable maternal deaths is that in many cases the physician appears to be at fault. Support for this comes from two studies, one performed in the UK and the other from the University of Illinois at Chicago. The UK’s Confidential Enquiries into Maternal Deaths from 1997–1999 found that the provider was at fault in as many as 60% of pregnancy-related deaths.45 Geller et al presented a case-control study of the ‘‘morbidity/mortality continuum’’ specifically evaluating sources of error and opportunities for prevention.46 Even after controlling for sociodemographic characteristics, provider-related error (especially with regards to management and failure to identify high-risk women) accounted for 86.7%, 93.3%, and 88.2% of potentially preventable deaths, near-miss cases, and cases of severe morbidity, respectively. To look on this as a gateway to improving maternal mortality, one must first be able to reliably distinguish severe morbidity from life-threatening near-miss cases, and in particular must enhance epidemiologic studies and, clinically speaking, heighten awareness within the healthcare team. One such means of accomplishing this was developed by Geller et al in 2004, and involved a scoring system utilizing various clinical factors, including organ failure, extended intubation, admission to an intensive care unit, surgical intervention, and transfusion.47 Utilizing these five factors yielded a specificity of 93.9% in the identification of near-miss cases, and 78.1% specificity when only the latter four, more easily defined, factors were utilized. It was not uncommon for cases categorized as ‘‘severe morbidity’’ by clinicians to be categorized as ‘‘near-miss’’ by
Maternal mortality 529
this four-factor system, perhaps highlighting a clinical underappreciation for these particularly sick women with the downstream consequence of missing opportunities for intervention. The same authors later detailed ‘‘strategies for change’’ in a 2006 publication.48 Not surprisingly, the obstetrician plays the key role in mortality prevention, with the hospital and State health agencies providing support. Clearly, there is the potential for a strong relationship to develop between physician and patient, particularly in the context of a high-risk pregnancy necessitating frequent visits. Screening for domestic violence (the importance of which was detailed earlier) and encouraging healthy habits such as seatbelt use are examples of preventive strategies. Also, this contact provides the opportunity for the physician to gather a clear understanding of a woman’s past medical history and her associated fears and concerns, which might dictate seeking consultations from various other medical specialties and social services. Outside the examination room, peer-review committees allow for the critical review of cases and offer constructive criticism such that lessons can be learned and systems-related breakdowns identified and ameliorated. At the State level, maternal mortality review committees provide a broad perspective, encompassing not only medical factors but also sociodemographic factors that might, in fact, provide a ‘‘common denominator’’ for many deaths – something that might go unnoticed when a particular hospital, hospital system, city, or region is studied. The value of research must also not be underestimated, not only in terms of specific management strategies for a particular condition (e.g. placenta accreta) but also in terms of more general issues such as racial disparity. The multiple references made to governmental agencies in this chapter speak to their importance. Such influence gathers nationwide attention, reminding women – both pregnant and perhaps soon-to-be pregnant – and policymakers of the importance of regular prenatal care and safe behavior, particularly when sociodemographic and medical risk factors for death exist. Also, easier documentation and recording of maternal mortality, perhaps electronically as alluded to earlier, is ultimately dependent on governmental agencies. Research agenda Investigate the issue of racial disparity in maternal mortality, especially in terms of demographics, genetics, physiology, nutrition, and the healthcare system, such that specific interventions can be formulated. Establish ‘‘near miss’’ reviews at the hospital, State, and national levels to identify those clinical conditions that might contribute to a maternal death, such that more timely and effective interventions can be identified. Develop community involvement in the review process so that ‘‘system-related’’ barriers can be eliminated, and a patient-centered perspective obtained. Examine the factors associated with peripartum cardiomyopathy, such that the etiology of and modifiable risk factors for cardiac dysfunction can be impacted. Examine the impact of domestic abuse on the occurrence of maternal death. Evaluate methods of incorporating effective screening for domestic and sexual abuse, substance abuse, and depression into prenatal care. Establish a structured, multidisciplinary review using linkage analysis of all deaths of reproductive age women such that accurate reporting of pregnancy-associated and pregnancy-related deaths would be possible.
530 C. T. Lang and J. C. King
SUMMARY Maternal mortality in the US remains a pressing issue for multiple reasons, the simplest and perhaps most important of which is the reality of an infant and child without a mother, and the consequences for the family as a whole. Improving maternal mortality involves a committed and long-lasting effort on behalf of many individuals along with medical and social organizations to better appreciate the scope of and risks associated with maternal mortality and then to enact multiple, interwoven strategies to that end. REFERENCES 1. Chang J, Elam-Evans LD, Berg CJ et al. Pregnancy-related mortality surveillance – United States, 1991– 1999. MMWR Surveill Summ 2003; 52: 1–8. 2. Deneux-Tharaux C, Berg C, Bouvier-Colle MH et al. Underreporting of pregnancy-related mortality in the United States and Europe. Obstet Gynecol 2005; 106: 684–692. 3. Kaunitz AW, Hughes JM, Grimes DA et al. Causes of maternal mortality in the United States. Obstet Gynecol 1985; 65: 605–612. 4. World Health Organization (WHO). Manual of the international statistical classification of diseases, injuries, and causes of death. 9th revision. Geneva: WHO, 1977. p. 1. 5. Atrash HK, Koonin LM, Lawson HW et al. Maternal mortality in the United States, 1979–1986. Obstet Gynecol 1990; 76: 1055–1060. 6. Atrash HK, Rowley D & Hogue CJ. Maternal and perinatal mortality. Curr Opin Obstet Gynecol 1992; 4: 61–71. 7. Berg CJ, Atrash HK, Koonin LM et al. Pregnancy-related mortality in the United States, 1987–1990. Obstet Gynecol 1996; 88: 161–167. 8. World Health Organization (WHO). Manual of the international statistical classification of diseases, injuries, and causes of death. 10th revision. Geneva: WHO, 1993. p. 1. 9. May WJ, Buescher PA & Murray AL. Perspectives in disease prevention and health promotion enhanced maternal mortality surveillance – North Carolina, 1988 and 1989. MMWR Morb Mortal Wkly Rep 1991; 40: 469–471. 10. Horon IL. Underreporting of maternal deaths on death certificates and the magnitude of the problem of maternal mortality. Am J Public Health 2005; 95: 478–482. 11. MacKay AP, Berg CJ, Duran C et al. An assessment of pregnancy-related mortality in the United States. Paediatr Perinat Epidemiol 2005; 19: 206–214. 12. CDC. State-specific maternal mortality among black and white women – United States, 1987–1996. MMWR Morb Mortal Wkly Rep 1999; 48: 492–496. 13. Harper MA, Espeland MA, Dugan E et al. Racial disparity in pregnancy-related mortality following a live birth outcome. Ann Epidemiol 2004; 14: 274–279. 14. 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: 361–364. 15. Hopkins FW, MacKay AP, Koonin LM et al. Pregnancy-related mortality in Hispanic women in the United States. Obstet Gynecol 1999; 94: 747–752. 16. Freeman HP & Payne R. Racial injustice in health care. New Engl J Med 2000; 343: 1045–1047. 17. Brett KM, Schoendorf KC & Kiely JL. Differences between black and white women in the use of prenatal care technologies. Am J Obstet Gynecol 1994; 170: 41–46. 18. Saftlas AF, Koonin LM & Atrash HK. Racial disparity in pregnancy-related mortality associated with livebirth: can established risk factors explain it? Am J Epidemiol 2000; 152: 13–19. 19. Kaufman JS & Cooper RS. Commentary: considerations for use of racial/ethnic classification in etiologic research. Am J Epidemiol 2001; 154: 291–298. 20. Ventura SJ, Mosher WD, Curtin SC et al. Trends in pregnancies and pregnancy rates by outcome: Estimates for the United States, 1976-96. NCHS. Vital Health Stat 2000; 21: 56. 21. Callaghan WM & Berg CJ. Pregnancy-related mortality among women aged 35 years and older, United States, 1991–1997. Obstet Gynecol 2003; 102: 1015–1021.
Maternal mortality 531 22. Oyelese Y & Smulian JC. Placenta previa, placenta accreta, and vasa previa. Obstet Gynecol 2006; 107: 927–941. 23. Wright VC, Schieve LA, Reynolds MA et al. Assisted reproductive technology surveillance – United States, 2000. MMWR Surveill Summ 2003; 52: 1–16. 24. MacKay AP, Berg CJ, King JC et al. Pregnancy-related mortality among women with multifetal pregnancies. Obstet Gynecol 2006; 107: 563–568. 25. Chichakli LO, Atrash HK, MacKay AP et al. Pregnancy-related mortality in the United States due to hemorrhage: 1979–1992. Obstet Gynecol 1999; 94: 721–725. 26. MacKay AP, Berg CJ & Atrash HK. Pregnancy-related mortality from preeclampsia and eclampsia. Obstet Gynecol 2001; 97: 533–538. 27. Whitehead SJ, Berg CJ & Chang J. Pregnancy-related mortality due to cardiomyopathy: United States, 1991–1997. Obstet Gynecol 2003; 102: 1326–1331. 28. Pearson GD, Veille JC, Rahimtoola S et al. Peripartum cardiomyopathy: National Heart, Lung, and Blood Institute and Office of Rare Diseases (NIH) workshop recommendations and review. JAMA 2000; 283: 1183–1188. 29. Chang J, Berg CJ, Saltzman LE et al. Homicide: a leading cause of injury deaths among pregnant and postpartum women in the United States, 1991–1999. Am J Public Health 2005; 95: 471–477. 30. Harper M & Parsons L. Maternal deaths due to homicide and other injuries in North Carolina: 1992–1994. Obstet Gynecol 1997; 90: 920–923. 31. Dannenberg AL, Carter DM, Lawson HW et al. Obstetrics: homicide and other injuries as causes of maternal death in New York City, 1987 through 1991. Am J Obstet Gynecol 1995; 172: 1557–1564. 32. Fildes J, Reed L, Jones N et al. Trauma: the leading cause of maternal death. J Trauma 1992; 32: 643–645. 33. Krulewitch CJ, Pierre-Louis ML, de Leon-Gomez R et al. Hidden from view: violent deaths among pregnant women in the District of Columbia, 1988–1996. J Midwifery Womens Health 2001; 46: 4–10. 34. McFarlane J, Campbell JC, Sharps P et al. Abuse during pregnancy and femicide: urgent implications for women’s health. Obstet Gynecol 2002; 100: 27–36. 35. Frautschi S, Cerulli A & Maine D. Suicide during pregnancy and its neglect as a component of maternal mortality. Int J Gynaecol Obstet 1994; 47: 275–284. 36. Feldman GB & Freiman JA. Prophylactic cesarean at term? N Engl J Med 1985; 312: 1264–1267. 37. Wax JR, Cartin A, Pinette M et al. Patient choice cesarean: an evidence-based review. Obstet Gynecol Surv 2004; 59: 601–616. 38. Harper MA, Byington RP, Espeland MA et al. Pregnancy related death and health care services. Obstet Gynecol 2003; 102: 273–278. 39. Sachs BP, Yeh J, Acker D et al. Cesarean section-related maternal mortality in Massachusetts, 1954–1985. Obstet Gynecol 1988; 71: 385–388. 40. Lilford RJ, Van Coeverden de Groot HA, Moore PJ et al. The relative risks of caesarean section (intrapartum and elective) and vaginal delivery: a detailed analysis to exclude the effects of medical disorders and acute pre-existing physiological disturbances. BJOG 1990; 97: 883–892. 41. Hall MH. Commentary: Confidential enquiry into maternal death. BJOG 1990; 97: 752–753. 42. Schuitemaker N, van Roosmalen J, Dekker G et al. Maternal mortality after cesarean in the Netherlands. Acta Obstet Gynecol Scand 1997; 76: 332–334. 43. Lydon-Rochelle M, Holt VL, Easterling TR et al. Cesarean delivery and postpartum mortality among primiparas in Washington State, 1987–1996. Obstet Gynecol 2001; 97: 169–174. 44. World Health Organization (WHO). Maternal mortality in 2000: Estimates developed by WHO, UNICEF and UNFPA. Geneva: WHO, 2003. 45. Lewis G, Drife J & Central Assessors. Why mothers die: report on confidential enquiries into maternal deaths in the United Kingdom, 1997–1999. Royal College of Obstetricians and Gynaecologists, 2001. 46. Geller SE, Rosenberg D, Cox S et al. The continuum of maternal morbidity and mortality: factors associated with severity. Am J Obstet Gynecol 2004; 191: 939–944. 47. Geller SE, Rosenberg D, Cox S et al. A scoring system identified near-miss maternal morbidity during pregnancy. J Clin Epidemiol 2004; 57: 716–720. 48. Geller SE, Cox S, Callaghan WM et al. Morbidity and mortality in pregnancy: laying the groundwork for safe motherhood. Womens Health Issues 2006; 16: 176–188. 49. Hoyert DL. Maternal mortality and related concepts. National Center for Health Statistics. Vital Health Stat 2007; 3(33).