Callaghan, William M. MD, MPH; Chu, Susan Y. PhD, MSPH; Jamieson, Denise J. MD, MPH
As of August 31, 2009, 28 pregnant women have been reported to have died from 2009 pandemic H1N1 influenza illness in the United States.1 These deaths represented 6% of the total reported 484 deaths among persons with confirmed or probable H1N1 during the same time period. Given that pregnant women represent about 1% of the U.S. population, this disproportion raised concerns that pregnant women were at increased risk of death from illness from this novel pandemic influenza compared with the general population.
Pregnancy has been associated with increased illness and death from influenza, especially during pandemics. Mortality rates among pregnant women in the pandemics of 1918 and 1957 appeared to be abnormally high, with case fatality rates reported between 20% and 51%.2–6 However, comparisons with these death rates are difficult given the ensuing changes in diagnostic practices and capabilities, and the current availability of antiviral agents, antibiotics, vasopressors, and mechanical ventilation that make it possible to save many patients who would not have survived earlier pandemics.7 Although the number of deaths among pregnant women due to pandemic H1N1 influenza illness has been presumed to be in excess of the number of deaths among pregnant women during a regular influenza season, no baseline rate of the expected number of deaths among pregnant women related to seasonal influenza exists. Hence, this analysis was conducted to provide an estimate of the average number of deaths related to influenza among pregnant women during recent seasonal influenza years in comparison with the current H1N1 pandemic.
MATERIALS AND METHODS
We used data from the Centers for Disease Control and Prevention (CDC) Pregnancy Mortality Surveillance System (PMSS) for the years 1998–2005. The PMSS was established in 1987 in CDC's Division of Reproductive Health in collaboration with state health departments and the American College of Obstetricians and Gynecologists in an effort to collect data on deaths causally related to pregnancy. Health departments in all 50 states, New York City, and the District of Columbia are asked to provide de-identified copies of death certificates for all women who died during or within 1 year of the end of pregnancy using whatever means available to identify them. For example, deaths may be identified at the state if the certificate lists a cause of death that contains a key word tied to the pregnancy chapter of the International Classification of Diseases (ICD), such as preeclampsia, postpartum hemorrhage, or placenta previa. Some state death certificates include check boxes that indicate that the woman who died was pregnant or had been pregnant at some defined interval before death; these de-identified death certificates also are sent to CDC. In addition, states are requested to provide death certificates for all women who die within 1 year of having a live birth or fetal death; these certificates are identified through linkages of death certificates of women of reproductive age with existing birth and fetal death certificates, and the matched sets of de-identified certificates are sent to CDC. The 1998–2005 data were chosen for this report because they are currently being analyzed, and this dataset represents the most current and complete set of pregnancy-related deaths since the comprehensive report on pregnancy-related mortality for the 1991–1997 period.8 Because death certificates are de-identified and all information on them concerned deceased persons, this study did not require review by an institutional review board.
Using a process developed by the American College of Obstetricians and Gynecologists/CDC Maternal Mortality Study group, the death, infant birth, and fetal death certificates are reviewed by medical epidemiologists at CDC to determine the cause and timing of death and conditions considered to be associated with the death; these causes and conditions are coded and entered into the PMSS database. Coding is specific to PMSS; ICD-9 or ICD-10 coding is not used. Deaths are considered to be pregnancy-related, and are coded as such in PMSS, if the woman died during pregnancy or within 1 year of the termination of pregnancy AND if the death resulted from complications of pregnancy itself, a chain of events initiated by pregnancy, or aggravation of an unrelated event or condition by the physiologic effects of pregnancy. Thus, for a death to be considered pregnancy-related, it must be both temporally and causally related to pregnancy. Respiratory conditions fit these criteria because they can be caused by complications of pregnancy itself (eg, postoperative pneumonia) or because the normal immunologic and/or physiologic changes during and shortly after pregnancy alters the response to a respiratory insult (eg, influenza and pneumonia).
In an effort to identify all women who potentially died from influenza during or shortly after pregnancy, we selected all pregnancy-related deaths in the PMSS database that listed a respiratory disorder as an associated condition leading to death. From this group, we retrieved the archived death certificates and, when available, the linked birth and fetal death certificates for women who had an associated respiratory condition related to death that was coded in the PMSS database as “acute upper respiratory infection,” “pneumonia (bacterial/viral)”, or “pneumonia not otherwise specified (NOS).” An obstetrician with extensive experience reviewing maternal death records (W.M.C.) reviewed all certificates and, based on text written on the death certificates, categorized each death as a “possible influenza-related maternal death” or not. Another obstetrician (D.J.J.) who independently reviewed all records agreed with the initial categorization in 94.6% (158 of 167) of cases. In the remaining nine cases, the two obstetricians discussed the record and reached consensus.
If the term “influenza” was written in any field on the death certificate, that death was included. For pneumonia deaths, the certificate was reviewed to determine whether the woman died from pneumonia without any mention of a predisposing condition that would have led to pneumonia as the terminal event. For example, if a certificate had pneumonia listed as the underlying cause of death, but the woman had evidence of hypoxic ischemic encephalopathy following eclampsia, that death was not considered as a possible death from pneumonia or influenza. Thus, deaths were included only when it was determined that pneumonia instigated the chain of events leading to death. Because influenza infection can be associated with secondary bacterial pneumonias, death certificates that listed specific bacterial pneumonias such as pneumococcal or staphylococcal pneumonias as an associated condition were considered potential influenza deaths.
To exclude deaths unrelated to influenza, pneumonia deaths were classified as occurring or not occurring during influenza season. We defined the influenza season for each year using the information on the weekly percentage of positive viral isolates generated by World Health Organization (WHO) and National Respiratory and Enteric Virus Surveillance System Collaborating Laboratories.9 The beginning of the influenza season for each year was considered as the first fall or winter month that included a week with at least 5% of isolates test positive for the circulating viruses and the end of the influenza season was considered as the last winter or spring month that included a week with at least 5% of isolates test positive for the circulating viruses (Table 1). If the date of death on the certificate occurred during an influenza season during 1998–2005, that pneumonia or influenza death was considered a possible influenza-related death in a pregnant woman. For these deaths, the certificates were further examined in an effort to identify comorbid conditions that might place them at higher risk of complications from influenza.
Cause-specific pregnancy-related mortality ratios (deaths per million live births) for possible influenza-related deaths were calculated for each year. US natality data served as the denominator (CDC WONDER http://wonder.cdc.gov/natality.html, accessed January 15, 2010). We apportioned the number of births to account for the months represented in the influenza seasons so that these ratios could be compared with the ratio for the first 4 months of data from of the 2009 pandemic.
During 1998–2005, there were 4,693 pregnancy-related deaths reported to the PMSS Of these, five had an associated condition leading to death classified as “acute upper respiratory infection,” 46 had an associated condition classified as “pneumonia (bacterial/viral),” and 116 had an associated condition “pneumonia not otherwise specified (NOS).” After careful review of the archived death, birth, and fetal death certificates for these 167 deaths, three of the five deaths classified as “acute upper respiratory infection,” 21 of the 46 deaths classified as “pneumonia (bacterial/viral),” and 54 of the 116 classified as “pneumonia not otherwise specified (NOS)” were determined to be pneumonia/influenza deaths (total=78). Among these 78 deaths, 40 occurred during an influenza season with a range of 2 to 14 per year (Table 2). Cause-specific mortality ratios, which account for both the number of deaths and the number of influenza season months in the year, ranged from 1.0 to 5.9. There were 14 deaths in 2003, a year in which the 2002–2003 influenza season extended into May 2003 and the 2003–2004 influenza season began in October 2003. These 40 deaths were considered possible pregnancy-related influenza deaths.
Among the 40 possible pregnancy-related influenza deaths, three had a notation of influenza (there were no deaths with a notation of influenza outside of an influenza season), two had an indication of viral syndrome, and the remaining 35 had an indication of pneumonia without any indication of influenza or viral syndrome leading to death on the death certificate. Most women did not have an indication of a comorbid condition. Asthma, the most commonly reported comorbid condition, was reported for five women, one of whom also had diabetes mellitus; one woman had diabetes mellitus and one woman had cerebral palsy. Characteristics of these women and their pregnancy outcomes are shown in Table 2. The interval between the end of pregnancy and death (defined as 0 for women who died undelivered) was 0–1 day for 14 women. An additional 15 women died 2–14 days after the end of pregnancy, and 10 women died 2 weeks to 3 months after the end of pregnancy; for one woman, the interval was not known, but a checkbox on the death certificate indicated that she had been pregnant within the previous 90 days (Table 2).
Over an 8-year period from 1998 to 2005, we identified 40 possible pregnancy-related influenza deaths in the United States. This average (mean) of five deaths per year is far fewer than the number of deaths (n=28) reported among U.S. women with confirmed pandemic H1N1 influenza in the early part of the 2009 pandemic (May through August). Using the most recent available natality data (2006) to provide a 4-month denominator for 2009 births, the mortality ratio of 20 per million live births for deaths due to confirmed pandemic H1N1 is more than six times the mean mortality ratio for seasonal influenza. So, even if underreporting was severe and only one half of seasonal influenza deaths were captured through PMSS, or if all of the deaths we excluded because they occurred outside of the influenza season were actually influenza deaths, the 28 deaths reported among pregnant women infected with 2009 H1N1 influenza during the early months of influenza circulation clearly exceeds the number of deaths that would be reasonably expected.
The increased mortality among pregnant women due to 2009 H1N1 influenza strain is consistent with reports from previous pandemics.10 During the 1918 pandemic, of 2,541 patients admitted to Chicago's Cook County Hospital with pneumonia, 101 (4%) were pregnant; 52 of these pregnant patients died, resulting in a mortality rate of 51.4%, a rate greater than the rates for nonpregnant women and for men of the same age.2 Similarly, among 1,350 cases of influenza among pregnant women identified during the 1918 pandemic, the overall case fatality rate was 27%.3 Reports from the 1957 Asian flu pandemic also suggest disproportionate mortality among pregnant women. Data from New York City and Minnesota found that pregnant women had 2–3 times the mortality compared with nonpregnant women, and that the pregnant women accounted for approximately 50% of the total mortality among healthy women aged 19–36 years.4,5 A similar doubling of risk of death was reported in 1957 in England and Wales, where 12 of the 103 women 15–44 years of age who died of influenza were pregnant.11
An increased mortality risk among pregnant women due to 2009 H1N1 influenza strains is consistent with several recent reports from the United States,12 Canada,13 and Australia and New Zealand14 demonstrating increased morbidity among pregnant women during the early months of this current pandemic. In all four countries, the rate of hospitalization and admission to intensive care units was higher among pregnant women than it was in nonpregnant women and in the general population. These reports, as well as the excess mortality due to pandemic H1N1 influenza compared with seasonal influenza in pregnant women suggested by our findings, clearly reinforce the recommendation that pregnant women should be considered a priority group for vaccination and treatment during pandemics of influenza.
It is possible that we did not identify all pregnant women who died from influenza/pneumonia between 1998 and 2005 in PMSS. This could occur if the deaths were not identified by the state and sent to CDC. For example, a death with only “pneumonia” or “influenza” written on the death certificate with no indication of pregnancy might not be identified as a maternal death. However, this limitation would not apply to states that link birth, fetal death, and death certificates. For 1998–2005, among all deaths in which the outcome of pregnancy was a live birth, a stillbirth, or an unknown outcome (unknown because there was no linkage to a birth or fetal death certificate), 78% of deaths had a matching certificate; among live births 96% had a matching birth certificate. Checkboxes were another way states could identify pregnancy-related deaths, by either adoption of the 2003 revised U.S. standard death certificate or the use of a state-developed question. For the years 1998–2005, 19 states had a pregnancy checkbox on their death certificate over the entire period; by 2005, an additional 16 states added checkboxes. However, it is unclear to what extent checkboxes facilitated identification of pregnancy-related influenza deaths given the variability in their use among states. Similarly, if “influenza,” “pneumonia,” or “other respiratory disorder” was not listed as a cause of death or associated condition on the death certificate, that death would be missed. However, it is notable that using PMSS, we detected an increase in the annual number of deaths due to influenza/pneumonia among pregnant women during 2003. This suggests that the system is sensitive to fluctuations in these events.
The reporting of maternal deaths due to the 2009 H1N1 influenza infection might be more complete than reporting during usual influenza seasons because influenza testing among pregnant women may have been increased or physicians and medical examiners might have been more likely to code deaths as clinically compatible with pneumonia caused by influenza because of the heightened awareness that an influenza pandemic had begun worldwide.15 In consideration of this phenomenon, we included pneumonia deaths during and shortly after pregnancy as possible influenza deaths during the nonpandemic years covered in this report. However, it is highly unlikely that reporting differences would explain the difference in the number of deaths in the United States among pregnant women attributed to laboratory-confirmed 2009 H1N1 influenza from May through August 2009 (28) compared with the expected number of influenza/pneumonia deaths among pregnant women during a typical influenza season, especially when considering the probability that many, if not most, of the pneumonia deaths identified in our report were not associated with influenza. For example, one pneumonia death occurred 6 weeks postpartum and there was no indication of maternal illness on the linked birth certificate. We still included that case in consideration that the course of pneumonia might have been protracted before death and hence, related to the pregnancy. Whereas our methods, if anything, would tend to overestimate the number of deaths among pregnant women potentially due to seasonal influenza, we acknowledge that real-time reporting during a time of heightened awareness is different than retrospective case-ascertainment.
The surveillance system we used, the PMSS, has served as the basis for national pregnancy-related mortality surveillance in the United States for more than 20 years and has identified more pregnancy-related deaths on a national level than vital statistics.8,16 All information on certificates is reviewed and the system does not rely on conventional ICD coding for cause of death. In the absence of a national registry for maternal deaths, this report represents the most complete count of pregnancy-related deaths due to pneumonia/influenza. Based on past trends, deaths due to influenza during and shortly after pregnancy are rare events, and the increased mortality among pregnant women in the early months of the 2009 H1N1 influenza pandemic was clearly cause for concern, and the recommendations for early empiric treatment for pregnant women with suspected pandemic H1N1 and establishing pregnant women as a priority group for immunization were important responses to this public health emergency. The baseline risk of pregnancy-related mortality from seasonal influenza needs to be considered in future planning for outbreaks of novel pandemic influenza.
2. Woolston WJ, Conley DO. Epidemic pneumonia (Spanish Influenza) in pregnancy: effect in one hundred and one cases. JAMA 1918;71:1898–9.
3. Harris JW. Influenza occurring in pregnant women: a statistical study of thirteen hundred and fifty cases. JAMA 1919;72:978–80.
4. Greenberg M, Jacobziner H, Pakter J, Weisl BAG. Maternal mortality in the epidemic of Asian influenza, New York City, 1957. Am J Obstet Gynecol 1958;76:897–902.
5. Freeman DW, Barno A. Deaths from Asian influenza associated with pregnancy. Am J Obstet Gynecol 1959;78:1172–5.
6. Rasmussen SA, Jamieson DJ, Bresee JS. Pandemic influenza and pregnant women. Emerg Infect Dis 2008;14:95–100.
7. White DB, Angus DC. Preparing for the sickest patients with 2009 influenza A (H1N1). JAMA 2009;302:1905–6.
8. Berg CJ, Chang J, Callaghan WM, Whitehead SJ. Pregnancy-related mortality in the United States, 1991–1997. Obstet Gynecol 2003;101:289–96.
10. Beigi RH. Pandemic influenza and pregnancy: a call for preparedness planning. Obstet Gynecol 2007;109:1193–6.
11. The Public Health Laboratory Service. Deaths from Asian influenza, 1957; a report by the Public Health Laboratory Service based on records from hospital and public health laboratories. BMJ 1958;1:915–9.
12. Jamieson DJ, Honein MA, Rasmussen SA, Williams JL, Swerdlow DL, Biggerstaff MS, et al. H1N1 2009 influenza virus infection during pregnancy in the USA. Lancet 2009;374:451–8.
13. Kumar A, Zarychanski R, Pinto R, Cook DJ, Marshall J, Lacroix J. Critically ill patients with 2009 influenza A (H1N1) infection in Canada. JAMA 2009;302:1872–9.
14. ANZIC Influenza Investigators. Critical care services and 2009 H1N1 influenza in Australia and New Zealand. N Engl J Med 2009;361:1925–34.
15. Thompson WW, Moore MR, Weintraub E, Cheng PY, Jin X, Bridges CB, et al. Estimating influenza-associated deaths in the United States. Am J Public Health 2009;99:S225–30.
16. Berg CJ, Atrash HK, Koonin LM, Tucker M. Pregnancy-related mortality in the United States, 1987–1990. Obstet Gynecol 1996;88:161–7.