Chapter Fourteen by xl771209




Maternal mortality continues to be a serious problem in Tanzania, as it is characterised by relatively high fertility, high incidence of infectious diseases, poverty, and poor health services. Such characteristics have led to high mortality in general. The 2004-05 TDHS is only the second nationally representative household survey to collect data on maternal mortality. The first was the 1996 TDHS. Little is known about maternal and adult mortality in Tanzania when compared to infant and child mortality, for a number of reasons. First, while early childhood mortality can be estimated through the birth history approach, there is no equivalent in adult mortality measurement. Second, death rates are much lower at adult ages than at childhood, and hence estimates for particular age groups are more likely to be distorted by sampling errors. Third, there is usually very limited information about the characteristics of those who have died. While the same can be said about data on childhood mortality, it is reasonable to expect the characteristics of parents to influence directly their children’s chances of survival. This chapter presents information on overall adult mortality and maternal mortality in Tanzania from the 2004-05 TDHS. Mortality levels and trends provide a good measure of the health status of the population and thus an indicator for national development. It should be noted that routine data are collected through the Ministry of Health using a system known as the Health Management Information System (HMIS). Unfortunately, because the system does not include deaths that occur outside health facilities, it does not reflect the mortality picture from a population perspective. On the other hand, although Tanzania’s Demographic Surveillance Sentinel System does collect information that is community-based, it covers only a few districts. Thus, although the data collected are of high quality, like the HMIS it does not reflect the entire population of the country. Thus, the data from the 2004-05 TDHS are critical to understanding adult mortality across the entire population.



To estimate adult mortality, the 2004-05 TDHS included a sibling history in the Women’s Questionnaire. A series of questions was asked about all of the respondent’s biological brothers and sisters and their survival status. These data allow direct estimation of overall adult mortality (by age and sex) and maternal mortality. Survival of siblings (i.e., biological brothers and sisters) is a useful method for collecting information on adult mortality. Each female respondent was asked to list all children born to her biological mother, including herself. These included all siblings who were still alive and those who had died. For brothers and sisters who were alive, only the age at the last birthday was asked. For brothers who had died, only the number of years since death and age at death were asked. For sisters who had died at age 12 years or older, three questions were asked to determine whether the death was maternity related: “Was [name of sister] pregnant when she died?” and, if negative, “Did she die during childbirth?” and, if negative, “Did she die within two months after the end of a pregnancy or childbirth?” It is intended that this information will not only give an estimate of maternal risk but a complete profile of exposure to the risk of mortality for the adult population of Tanzania. Adult and maternal mortality estimation requires accurate reporting of the number of siblings the respondent ever had, the number who died and the number of sisters who have died of maternal causes (for maternal mortality). Although there is no definitive procedure for establishing the completeness of retrospective data on sibling survivorship, Table 14.1 presents several indicators that can be used to measure the quality of sibling survivorship data.

Adult and Maternal Mortality | 257

Table 14.1 Data on siblings Number of siblings reported by survey respondents and completeness of the reported data on age, age at death (AD) and years since death (YSD), Tanzania 2004-05 Sibling status and completeness of reporting All siblings Surviving Deceased Missing information Surviving siblings Age reported Age missing Deceased siblings AD and YSD reported Missing only AD Missing only YSD Missing both Females Number 32,014 25,675 6,329 11 25,675 25,664 11 6,329 6,261 21 12 35 Percentage 100.0 80.2 19.8 0.0 100.0 100.0 0.0 100.0 98.9 0.3 0.2 0.6 Males Number 32,004 25,164 6,830 10 25,164 25,145 19 6,830 6,737 20 12 61 Percentage 100.0 78.6 21.3 0.0 100.0 99.9 0.1 100.0 98.6 0.3 0.2 0.9 64,019 50,839 13,159 21 50,839 50,810 29 13,159 12,998 41 24 96 Total Number Percentage 100.0 79.4 20.6 0.0 100.0 99.9 0.1 100.0 98.8 0.3 0.2 0.7

The data do not show any obvious defects that would indicate poor data quality or significant underreporting. A total of 64,019 siblings were recorded in the maternal mortality section of the 200405 TDHS questionnaires. The sex ratio of the enumerated siblings (the ratio of brothers to sisters) is around 1.00, possibly indicating a slight underenumeration of brothers. The survival status for only 21 (less than 0.1 percent) of the siblings was not reported. For the surviving siblings, current age was not reported for only 29 (less than 0.1 percent). Among deceased siblings, both the age at death and years since death were missing for less than 1 percent. Rather than exclude the siblings with missing data from further analysis, information on the birth order of siblings in conjunction with other information was used to impute the missing data.1 The sibling survivorship data, including cases with imputed values, have been used in the direct estimation of adult and maternal mortality.



One way to assess the quality of data used to estimate maternal mortality is to evaluate the plausibility and stability of overall adult mortality. It is reasoned that if rates of overall adult mortality are implausible, rates based of a subset on deaths—maternal mortality in particular—are likely to have serious problems. Also, levels and trends in overall adult mortality have important implications in their own right for health and social programmes in Tanzania, especially with regard to the potential impact of the AIDS epidemic. The direct estimation of adult mortality uses the reported ages at death and years since death of respondents’ brothers and sisters. Because of the differentials in exposure to the risk of dying, ageand sex-specific death rates are presented in this report. The results are also compared with rates obtained from the 1996 TDHS. Since the number of deaths on which the rates are based is not very large (893 female deaths and 807 male deaths in the 2004-05 TDHS and 501 female deaths and 601

The imputation procedure is based on the assumption that the reported birth order of siblings in the history is correct. The first step is to calculate birth dates. For each living sibling with a reported age and each dead sibling with complete information on both age at death and years since death, the birth date was calculated. For a sibling missing these data, a birth date was imputed within the range defined by the birth dates of the bracketing siblings. In the case of living siblings, an age at the time of the survey was then calculated from the imputed birth date. In the case of dead siblings, if either the age at death or years since death was reported, that information was combined with the birth date to produce the missing information. If both pieces of information were missing, the distribution of the ages at death for siblings for whom the years since death was unreported, but age at death was reported, was used as a basis for imputing the age at death.


258 | Adult and Maternal Mortality

male deaths in the 1996 TDHS), the estimated age-specific rates are subject to considerable sampling variation. Table 14.2 presents age-specific mortality rates for women and men age 15-49 for the seven-year period preceding the survey. Generally, the rates show the expected increases for both sexes with increasing age, although women age 45-49 have a slightly lower mortality rate than women age 40-44. Female mortality exceeds male mortality among those younger than 30 years of age, with a greater difference being observed at age group 20-24 and 25-29; the rates are nearly the same at age group 30-34. Above age 35, male mortality exceeds female mortality by greater margins as age advances. Overall, mortality rates are slightly higher among females than males (6.6 and 6.2 deaths per 1,000 years of exposure, respectively), which is unusual since male mortality typically exceeds female mortality during these ages. However, AIDS is now a significant cause of death in Tanzania, and its emergence has altered the age and sex pattern of mortality.
Table 14.2 Adult mortality rates Age-specific mortality rates for women and men age 15-49 based on the survivorship of sisters and brothers of survey respondents for the seven-year period preceding the survey, Tanzania 2004-05 Age Deaths Exposure Mortality rates

WOMEN 15-19 20-24 25-29 30-34 35-39 40-44 45-49 15-49 66 130 193 194 153 106 51 893 26,599 29,609 27,048 21,726 15,515 9,408 5,202 135,106 MEN 2.470 4.400 7.152 8.913 9.868 11.223 9.852 6.610

Figure 14.1 shows the age-specific mortality rates for 41 24,288 1.708 males and females age 15-49 for the seven-year period 15-19 20-24 86 27,991 3.073 preceding the 2004-05 TDHS and the nine-year period 25-29 135 26,554 5.099 preceding the 1996 TDHS. Although the 2004-05 TDHS rates 30-34 187 21,469 8.714 159 15,304 10.362 at older ages appear somewhat erratic, that is most probably 35-39 127 9,707 13.124 due to sampling variability. A comparison of the 2004-05 40-44 45-49 71 5,333 13.261 TDHS and the 1996 TDHS rates indicates substantially higher a adult mortality rates for both males and females at all ages in 15-49 6.174 807 130,646 the later survey, with the exception of men age 15-24. The summary measure of mortality for age group 15-49 shows an a Age-standardised increase of 68 percent in female mortality rates and 24 percent in male mortality rates from the 1996 TDHS rates. It should be noted, however, that the 1996 TDHS report indicates the possibility of underreporting of deceased siblings. Thus, it is not possible to conclude that adult mortality has increased.

Adult and Maternal Mortality | 259

Figure 14.1 Trends in Adult Mortality, Tanzania 1988-1996 and 1998-2004
12 10 8 6 4 2 0 15-19 20-24 25-29 30-34 Age in Years 35-39 40-44 45-49 . Females Deaths per 1,000

, , , +

Figure 14.1 Trends in Adult Mortality, , Kenya 1991-1997 and 1996-2002
+ , , + + + + +


14 12 10 8 6 4 2 0

Deaths per 1,000


, +

, , + + +

+ ,

, +

, +



30-34 Age in Years





+ Tanzania 1996 (1988-1996)

, Tanzania 2004-05 (1998-2004)

Note: Data refer to the seven-year period preceding the 2004-05 TDHS and the nine-year period preceding the 1996 TDHS.



Two survey methods are generally used to estimate maternal mortality in developing countries: the sisterhood method (Graham et al., 1989) and a direct variant of the sisterhood method (Rutenberg and Sullivan, 1991). In this report, the direct estimation procedure is applied. Age-specific mortality rates are calculated by dividing the number of maternal deaths by woman-years of exposure. To remove the effect of truncation bias (the upper boundary for eligibility for women interviewed in the survey is 49 years), the report standardised the overall rate for women age 15-49 by the age distribution of the survey respondents. Maternal deaths are defined as any death that occurred during pregnancy, childbirth, or within two months after the birth or termination of a pregnancy.2 Estimates of maternal mortality are therefore based solely on the timing of the death in relationship with pregnancy.

This time-dependent definition includes all deaths that occurred during pregnancy and two months after pregnancy, even if the death was due to nonmaternal causes. However, this definition is unlikely to result in overreporting of maternal deaths because most deaths to women during the two-month period are due to maternal causes, and maternal deaths are more likely to be underreported than overreported.


260 | Adult and Maternal Mortality

Table 14.3 presents direct estimates of maternal mortality for the ten-year period preceding the survey. The data indicate that the rate of mortality associated with pregnancy and childbearing is 1.1 maternal deaths per 1,000 woman-years of exposure. The estimated age-specific mortality rates display a plausible pattern, being generally higher during the peak childbearing ages than at the younger and older age groups (although there is a slight fluctuation at age 40-44). However, the agespecific pattern should be interpreted with caution because of the small number of events—only 203 maternal deaths for women of all ages. Maternal deaths represent 18 percent of all deaths to women age 15-49 during the ten-year period preceding the survey, (203 maternal deaths /1,109 female deaths), a figure that is two-thirds of the proportion found in the 1996 TDHS (27 percent). The low proportion of maternal deaths could be due to an increase in nonmaternal deaths (e.g., AIDS-related deaths), to underreporting of maternal deaths in the 2004-05 TDHS, or to an underreporting of all female deaths in the 1996 TDHS.
Table 14.3 Maternal mortality Maternal mortality rates for the ten-year period preceding the survey, based on the survivorship of sisters of survey respondents, Tanzania 2004-05 Maternal deaths 15 50 48 47 23 15 5 203 Exposure (years) 39,664 41,586 36,755 28,670 19,736 11,755 6,140 184,305 Mortality rates (1,000) 0.382 1.194 1.298 1.646 1.181 1.311 0.829 1.104a 0.198 578

Age 15-19 20-24 25-29 30-34 35-39 40-44 45-49 Total 15-49

General fertility rate1 Maternal mortality ratio2
a 1

Age-standardised Expressed per 1,000 woman-years of exposure 2 Expressed per 100,000 live births; calculated as maternal mortality rate divided by the general fertility rate

The maternal mortality rate can be converted to a maternal mortality ratio and expressed per 100,000 live births by dividing the rate by the general fertility rate of 0.198, which prevailed during the same time period. The advantage of this conversion is that it highlights the obstetric risk, which has great prograrammatic significance. Thus, for Tanzania between 1995-2004, the maternal mortality ratio is estimated as 578 maternal deaths per 100,000 live births. In other words, for every 1,000 live births in Tanzania during this period, almost 6 women died of pregnancy-related causes. It should be noted that maternal mortality is a difficult indicator to measure because of the large sample sizes required to calculate an accurate estimate. (This is evidenced by the fact that the maternal mortality ratio is expressed per 100,000 live births, demonstrating that it is a relatively rare event.) The maternal mortality estimates are subject to large sampling errors.3 Thus, although the 2004-05 TDHS ratio of 578 is higher than the 1996 estimate of 529, the difference between the two figures is not statistically significant. Thus, it is not possible to conclude that there has been any change in maternal mortality in Tanzania.


The 95 percent confidence interval for the 2004-05 rate of 578 is 466-690.

Adult and Maternal Mortality | 261

To top