The Journal – December 2012
Tex Med. 2012;108(12):e1.
By Larissa J. Estes, DrPH; Linda E. Lloyd, PhD; and Beatrice J. Selwyn, ScD
Larissa J. Estes, DrPH, Office of Surveillance and Public Health Preparedness, Houston Department of Health and Human Services, Houston, Texas; Linda E. Lloyd, PhD, and Beatrice J. Selwyn, ScD, School of Public Health, University of Texas Health Science Center at Houston, Houston, Texas. Send correspondence to Larissa J. Estes, DrPH, Office of Surveillance and Public Health Preparedness; Houston Department of Health and Human Service, 8000 North Stadium, Houston, Texas 77054; email: firstname.lastname@example.org.
Maternal mortality is often used to measure health and well-being for women. Improved surveillance efforts can improve maternal mortality estimates and inform the development of strategies to address the needs of maternal and child health populations. The purpose of this study was to provide better estimates of maternal mortality in Texas by using enhanced surveillance methods. Results from our analyses of fetal death and live birth records in Texas from 2000 through 2006 were then linked to pregnancy-related death records and death records of women of childbearing age (15-44 years) in Texas from 2001 through 2006. Enhanced surveillance identified almost 3.5 times as many deaths that might be associated with pregnancy than do current methods and confirmed a persistent race/ethnicity trend in maternal mortality. The leading cause of these 2001-2006 pregnancy-associated deaths was accidents. Enhanced surveillance allows the identification of additional deaths possibly associated with pregnancy and provides a stable foundation to investigate trends further and to review maternal mortality cases systematically.
Maternal mortality is at a great and unique risk for being misclassified because pregnancy may not be immediately considered as being related to the cause of death. Multiple sources are required to identify pregnancy-associated deaths and to understand contributing risk factors.1 When only death certificates were used, the numbers of pregnancy-associated deaths were considerably underestimated.2-4 As many as 30% to 60% of pregnancy-associated deaths, irrespective of cause, are underreported when traditional surveillance methods with International Classifications of Death, 10th revision (ICD-10) coding are used.5
The standard method used by the Texas Department of State Health Services (DSHS) Center for Health Statistics to identify maternal deaths, or pregnancy-related deaths, uses ICD-10 codes (G. Willard, telephone communication, September 2009) and 42-day temporal window post-childbirth. DSHS added a question to the Texas death certificate (from the national standard death certificate of 2003) identifying the temporal relationship of pregnancy to death and lengthened the temporal window to 1-year after childbirth. This takes advantage of other ICD-10 codes for deaths that may be associated with pregnancy (ie, pregnancy-associated deaths). Researchers in Maryland found that the pregnancy question on the death certificate was particularly effective in identifying pregnancy-related deaths.6 Other than the ICD-10 codes and the pregnancy question on the death certificate, Texas does not use any other surveillance methods to determine if there are additional deaths associated with pregnancy. Other states have demonstrated increased identification of maternal deaths by using enhanced surveillance methods. Enhanced surveillance methods include the linking of live birth and fetal death records to death records of women of childbearing age plus the expertise of and systematic review by maternal mortality review boards. The implementation of enhanced surveillance methods in Texas has the potential to identify additional cases of maternal deaths.
Despite mandates to complete vital records accurately and submit them to state and federal entities for analysis and evaluation, inherent issues exist regarding the completion of the primary data source for determining maternal mortality, namely death certificates. Maternal mortality ratios (MMRatios) ([number of maternal deaths / number of live births] x 100) for a particular time period can be incorrect because of incomplete death certificates, inaccurate information recorded on the death certificates, discordance in definitions used, or poor access to associated information needed to complete death certificates accurately. Researchers have found that failure to document the decedent's recent pregnancy on the death certificate is a recurrent problem, and underreported and misclassified maternal deaths were more likely for women aged 20-24 years and among women who died from nonobstetric causes.7,8
From 2001 through 2006, the MMRatio for Texas was higher than that of the United States. In 2006, the MMRatio was 22.5 deaths per 100,000 live births in Texas, compared with 13.3 deaths per 100,000 live births for the United States.9,10 Texas estimates also demonstrate a disparity among racial groups. According to the data derived from Texas' current surveillance methods, black women are 3 times more likely to die from pregnancy-related causes than white women.9
Prior research has demonstrated that enhanced surveillance (ie, the linking of death records with fetal death and live birth records) has repeatedly resulted in the identification of more pregnancy-associated deaths when compared with using pregnancy-related death records alone, where death is identified only through ICD-10 maternal death or Chapter O codes.2,10-16 While traditional causes of maternal mortality are identified through these codes, cases in which the pregnancy status of the decedent is unknown continue to escape traditional surveillance efforts. The purpose of this study was to provide better estimates of maternal mortality in Texas by using enhanced surveillance methods.
A cross-sectional analysis of maternal mortality in Texas from 2001 through 2006 used enhanced surveillance methods, by linking live birth and fetal death records in Texas from 2000-2006 to death records in Texas from 2001-2006, to confirm cases of pregnancy-related deaths (ICD-10 Chapter O) and to identify new cases of possible pregnancy-associated deaths among women of childbearing age (15-44 years) occurring within 1 year of the end of a pregnancy. An eligible case was a woman whose death certificate showed an ICD-10 Chapter O code as the single, main cause of death, or a woman aged 15-44 years who died within 1 year of giving birth, regardless of cause of death. A pregnancy-related death was defined as the death of a woman that occurred up to 1 year postpartum with an ICD-10 Chapter O code as the main or single cause of death.17 A pregnancy-associated death is defined as a new case of death (irrespective of cause) within 1 year postpartum, identified through enhanced surveillance.17 The maternal first and maiden name and the maternal date of birth variables were used to link death records in Texas from 2001 through 2006 to live birth and fetal death records in Texas from 2000 through 2006. Linked records were de-identified and assigned a record identification number. The enhanced surveillance variable was created through the linkage of vital records, and three categories were established:
- Confirmed Cases of Pregnancy-Related Death: identified deaths with an ICD-10 Chapter O code successfully linked to a live birth or fetal death record within 1 year of the death.
- New Cases of Possible Pregnancy-Associated Death: identified deaths of women of childbearing age (15-44 years) with an ICD-10 cause-of-death code (not from Chapter O) successfully linked to a live birth or fetal death record within 1 year prior to death.
- Unconfirmed Cases of Pregnancy-Related Death: identified deaths with an ICD-10 Chapter O code not successfully linked to a live birth or fetal death record within 1 year prior to death.
Variables abstracted from the vital records for this study are listed in Table 1. These variables have been examined in previous maternal mortality research using the same or similar methodology. These variables may be able to describe groups at risk and provide more information on how to address pregnancy-associated deaths, particularly those that are missed by Texas' current surveillance methods.
The pregnancy-associated mortality ratio (PAMRatio) ([the combined number of confirmed cases of pregnancy-related maternal death plus new cases of possible pregnancy-associated maternal death over a 1-year period after a birth / the number of live births during a 1-year period] x 100) was calculated for the enhanced surveillance variable for each year. The goal was to gain broader and more precise estimates of maternal mortality, regardless of cause or the length of time since the end of the pregnancy. PAMRatios trends were examined and compared over time. Frequency distributions for each variable and chi-squared statistics for cross tabulations of the demographic and prenatal care variables were used to explore whether a statistically significant association existed between how the case was identified and demographic and prenatal care variables. The enhanced surveillance PAMRatio was calculated for individual years and across the 6-year period by ethnicity.
In total, 1144 deaths were identified by using this enhanced surveillance method: 181 cases were confirmed pregnancy-related deaths, and 708 were new cases of possible pregnancy-associated deaths. Two hundred fifty-five deaths were cases of pregnancy-related death that were not successfully linked to a live birth or fetal death record within 1 year of the death. Table 2 compares the percentage of deaths identified by using the state's current methods with the percentage of deaths identified by using enhanced surveillance. From 2001 through 2006, nearly 64% of the deaths identified through enhanced surveillance deaths were previously unidentified. Table 3 describes the percentage of deaths by demographic variables from 2001 through 2006 for new cases of pregnancy-associated deaths and shows the frequency trends in age, education, and ethnicity. Overall, most deaths occurred in women aged 18-34 years, with high school or less education, and of white and Hispanic ethnicities.
Almost 27%18 (n = 308) of all deaths identified using enhanced surveillance had an ICD-10 Chapter O code. From 2001 through 2006, Late Maternal Death (Codes O96, O97) was the most frequent maternal cause of death (n = 94; 30.5%),18 followed by other direct and then indirect causes. Among single causes of death that were not related to ICD-10 Chapter O, the leading cause of death was accidents (n = 300; 35.88%).18 Accidents (Codes V01- X59) include deaths caused by transport (pedestrian, motor vehicle), falls, accidental drowning, exposure to smoke, fire and flames, and accidental poisoning. Accidents were followed as causes of death by malignant neoplasms (Codes C00-D48), assaults (Codes X85-Y09), and intentional self-harm (Codes X60-X84).18
Examining the PAMRatio over a 3-year period by using overlapping periods demonstrates a more stable view of the trend in the mortality ratios. Table 4 and Figure 1 compare the online Texas Health Data pregnancy-related mortality ratio (PRMRatio), determined by using state data available online, and the estimated PAMRatio, based on the enhanced surveillance methods used in this study. Both trend lines demonstrate an increasing mortality ratio over time. The estimated PAMRatio for all deaths was 3 to 4 times greater than the online Texas Health Data PRMRatio, which uses current surveillance methods, for each time period.
Enhanced surveillance methods identified black women as having a PAMRatio 4 times that of white women and almost 3 times that of Hispanic women during the same time. Figure 2 shows that for the period 2001-2006, black women suffered an estimated PAMRatio of 103 deaths per 100,000 live births, compared with 25.57 deaths per 100,000 live births for white women and 34.64 deaths per 100,000 live births for Hispanic women. Figure 2 demonstrates the disparity trend experienced by black women in the period 2001-2006 and the steeper rate of increase in PAMRatios over that time period for black women compared with white and Hispanic women.
Several demographic and prenatal care variables demonstrated, via chi-squared analysis, statistically significant associations (P≤.05) with the enhanced surveillance variable (Table 5). The race variable across all years was statistically significant (P≤.001). The prenatal care variables and insurance status or source of payment for labor and delivery were statistically significant (P≤.05) for both years that the data were collected. The number of prenatal visits did not demonstrate a statistically significant association during any of the years analyzed.
This study used state vital statistics to conduct enhanced surveillance and provide more complete estimates of maternal mortality in Texas for the years 2001 through 2006. Enhanced surveillance identified almost 3.5 times the number of possible pregnancy-associated deaths than the deaths identified as pregnancy-related by the online Texas Health Data system. This study's use of similar methods (enhanced surveillance) from previous published research yielded similar results of identifying additional deaths by broadening the scope of time examined after pregnancy and the cause of death.2,4,10,12,19,20
Enhanced surveillance confirmed persistent disparity trends in maternal mortality. These results further confirm previous research that black women suffer disproportionately the burden of maternal mortality. Researchers in Wisconsin found that black women were almost 7 times likelier to die from pregnancy-related causes than were white women.20 In Chicago, black women were at an elevated risk of death that persisted in both overall and condition-specific analyses.16 Among national data sets, black women experienced higher pregnancy-related mortality from preeclampsia, eclampsia, and postpartum hemorrhage.21
Accurate surveillance of maternal mortality facilitates the monitoring of progress toward nationally established goals. Findings from this study further confirm previous studies on enhanced surveillance: maternal deaths have the potential to be underreported, causes other than those coded as pregnancy-related are leading causes of deaths after pregnancy, and a disparity persists in maternal deaths experienced by black women compared with whites. This study suggests an inclusive surveillance methodology to discover as many pregnancy-associated deaths as possible and to investigative the underlying relationship to pregnancy that would allow for more accurate identification and classification of pregnancy-related deaths.22
The addition of late maternal death codes in ICD-10 and the inclusion of an additional pregnancy status question on the US standard death certificate were positive changes that improved ascertainment of pregnancy-related deaths.23 It is unclear how case ascertainment has been improved in Texas with these changes to the death certificate.23
Sources of Error and Study Limitations
Information bias is present in this study as the result of potential variation in definitions of pregnancy-related death and how pregnancy status was recorded on the death certificate.24 Why we were unable to confirm 58% of all pregnancy-related deaths identified through enhanced surveillance (ie, by linking them to live birth or fetal death records) is unclear. Some deaths may have been related to births occurring in the year 2000, which was outside the range of birth and fetal death certificates used in this study. Births occurring in 2006 may have been related to deaths of women of childbearing that occurred in 2007, outside the range of death certificates used in this study. Despite these limitations, the data available (at the time of analysis in 2010) remain valuable and are our only information when analyzing maternal mortality in Texas. Difficulties in linking certificates could be the result of inaccurate information on the death certificate. The inaccuracy and incompleteness of death certificates is well documented.2,13,25,26 Death certificates in Texas are the only resource for understanding maternal mortality on a population level. The use of vital records with incorrect or incomplete information could affect estimates of pregnancy-associated mortality ratios and the chi-squared test for association. The chi-squared test yielded some significant results, but most calculations suffered from small numbers and did not meet the rule that 80% of the expected frequencies exceed 5 and all expected frequencies exceed 1.27
Policy and Practice Recommendations
Based on this study and current literature, we have several policy recommendations to improve maternal mortality surveillance in Texas. Stakeholders and policymakers in Texas should examine potential enhanced surveillance methods that could be used to capture data better and improve surveillance. The Texas Department of State Health Services should examine its current surveillance efforts and determine if they are sufficient to identify maternal deaths accurately and should monitor trends in maternal death among geographic and sociodemographic populations. DSHS already provides data to the Center for Disease Control and Prevention's Pregnancy Mortality Surveillance System. Those data could be analyzed by DSHS to obtain a better understanding of maternal mortality in Texas, without costing additional state funds for surveillance.
This study demonstrated that the current maternal mortality surveillance methods in Texas do not accurately or completely identify deaths possibly associated with pregnancy. Enhanced surveillance identified almost 3.5 times as many deaths that may be associated with pregnancy as the deaths identified as pregnancy-related by the online Texas Health Data system. The persistent maternal mortality disparity experienced by black women was confirmed through the use of enhanced surveillance with Texas data. Accurately understanding this disparity and the sociodemographic context of maternal deaths is crucial to strategically address maternal and child health needs of Texas families.
The authors would like to acknowledge the work of Ms. Renke Zhou and Ms. Karyn Popham.
- Atrash HK, Alexander S, Berg CJ. Maternal mortality in developed countries: Not just a concern of the past. Obstet Gynecol. 1995;86(4):700-705.
- Horon IL. Underreporting of maternal deaths on death certificates and the magnitude of the problem of maternal mortality. Am J Public Health. 2005;95(3):478-482.
- Panting-Kemp A, Geller SE, Nguyen T, Simonson L, Nuwayhid B, Castro L. Maternal deaths in an urban perinatal network, 1992-1998. Am J Obstet Gynecol. 2000;183(5):1207-1212.
- Deneux-Tharaux C, Berg C, Bouvier-Colle M, et al. Underreporting of pregnancy-related mortality in the United States and Europe. Obstet Gynecol. 2005;106(4):684-692.
- Berg CJ, Atrash HK, Koonin LM, Tucker M. Pregnancy-related mortality in the United States, 1987-1990. Obstet Gynecol. 1996;88(2):161-167.
- Horon IL, Cheng D. Effectiveness of pregnancy check boxes on death certificates in identifying pregnancy-associated mortality. Public Health Rep. 2011;126(2):195-200.
- 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 Suppl:S31-S35.
- Kao S, Chen L, Shi L, Weinrich MC. Underreporting and misclassification of maternal mortality in Taiwan. Acta Obstet Gynecol Scand. 1997;76(7):629-636.
- Texas Vital Statistics. Texas health data. 2009. http://soupfin.tdh.state.tx.us.www5.sph.uth.tmc.edu:2048/deathdoc.htm. Accessed September 14, 2009.
- Heron MP, Hoyert DL, Murphy SL, Xu JQ, Kochanek KD, Tejada-Vera B. Deaths: Final Data for 2006. Hyattsville, MD: National Center for Health Statistics; 2009.
- Centers for Disease Control and Prevention. Misclassification of maternal deaths. MMWR Morb Mortal Wkly Rep. 1986;35(39):621.
- Centers for Disease Control and Prevention. Pregnancy-related mortality – Georgia, 1990-1992. MMWR Morb Mortal Wkly Rep. 1995;44(5):93.
- Buescher PA, Harper M, Meyer RE. Enhanced surveillance of maternal mortality in North Carolina. North Carolina Public Health, Center for Health Informatics and Statistics. CHIS Studies. No. 125. April 2001.
- Horon IL, Cheng D. Enhanced surveillance for pregnancy-associated mortality – Maryland, 1993-1998. JAMA. 2001;285(11):1455-1459.
- Chang J, Elam-Evans LD, Berg CJ, et al. Pregnancy-related mortality surveillance – United States, 1991-1999. MMWR CDC Surveill Summ. 2003;52(2):1-8.
- Rosenberg D, Geller SE, Studee L, Cox SM. Disparities in mortality among high risk pregnant women in Illinois: a population based study. Ann Epidemiol. 2006;16(1):26-32.
- Berg C, Danel I, Atrash H, Zane S, Bartlett L (eds). Strategies to Reduce Pregnancy-Related Deaths: From Identification and Review to Action. Atlanta, GA: Centers for Disease Control and Prevention; 2001.
- Estes LJ. Maternal mortality in Texas 2001-2006. Texas Medical Center Dissertations (via ProQuest). 2011. Paper AAI3464795. http://digitalcommons.library.tmc.edu/dissertations/AAI3464795. Accessed July 6, 2012.
- Dietz PM., Rochat RW, Thompson BL, Berg CJ, Griffin GW. Differences in the risk of homicide and other fatal injuries between postpartum women and other women of childbearing age: Implications for prevention. Am J Public Health. 1998;88(4):641-643.
- Mascola MA, Schellpfeffer MA, Kruse TK, Conway AE, Kvale KM, Katcher ML. Pregnancy-associated deaths and pregnancy-related deaths in Wisconsin, 1998-2001. WMJ. 2004;103(5):61-66.
- Tucker MJ, Berg CJ, Callaghan WM, Hsia J. The black-white disparity in pregnancy-related mortality from 5 conditions: Differences in prevalence and case-fatality rates. Am J Public Health. 2007;97(2):247-251.
- Callaghan WM. Overview of maternal mortality in the United States. Semin Perinatol. 2012;36:2-6.
- Hogan MC, Foreman KJ, Naghavi M. Maternal mortality for 181 countries, 1980–2008: A systematic analysis of progress towards millennium development goal 5. Lancet. 2010;375(9726):1609-1623.
- Szklo M, Nieto FJ. Epidemiology: Beyond the Basics. Boston, MA: Jones and Bartlett; 2007.
- Lahti RA, Penttilä A. The validity of death certificates: Routine validation of death certification and its effects on mortality statistics. Forensic Sci Int. 2001;115(1-2):15-32.
- Percy C, Stanek E III, Gloeckler L. Accuracy of cancer death certificates and its effect on cancer mortality statistics. Am J Public Health. 1981;71(3):242-250.
- Cochran WG. Some methods of strengthening the common χ2
test. Biometrics. 1957;10:417-451.
December 2012 Texas Medicine Contents
Texas Medicine Main Page