Maternal Mortality in Herat Province, Afghanistan, in 2002: Title and subTitle BreakAn Indicator of Women's Human Rights

Maternal mortality may be an important indicator of the health and human rights status of women, their access to health care, and the adequacy and ability of the health care system to respond to their needs.^{1 - 2} Disparities in maternal mortality rates also serve as important indicators of health inequality.^{1 - 2} Every year, more than 515 000 women worldwide die of complications of pregnancy and childbirth,¹ and 50 million women experience adverse health complications after childbirth, which could be prevented by cost-effective health interventions.

In Afghanistan, the combined effects of more than 20 years of war, persistent human rights violations, and Taliban restrictions on women's human rights have had devastating health consequences for women.^{3 - 5} Afghanistan is one of the poorest countries in the world and in 1997 was reported to have one of the highest infant (152/1000) and child (257/1000) mortality rates of all countries.⁶ Life expectancy of women was reported to be 44 years in 1997.⁷ Only 17% of rural residents and 38% of urban residents had access to safe drinking water in 1997,^{8 - 10} and it was estimated that annually 42% of all deaths¹¹ and the deaths of 85 000 children younger than 5 years were due to diarrheal diseases.¹² Malnutrition affected up to 52% of children younger than 5 years.^{8 - 11} More than 70% of the health care system in Afghanistan is reportedly dependent on external assistance.¹³

In 1997, the maternal mortality rate (No. of maternal deaths/100 000 live births per year) in Afghanistan was reported to be one of the worst in the world: 820 maternal deaths/100 000 live births per year.¹⁴ This rate was determined by a statistical modeling method developed by the World Health Organization (WHO) and the United Nations Children's Fund (UNICEF),¹⁴ and no additional assessments of maternal mortality have been available since that time. The objectives of this study were to (1) provide a rapid and accurate estimate of maternal mortality in Herat, Afghanistan; (2) assess women's human rights that may contribute to maternal mortality; and (3) assess maternal health services in the region.

This study was conducted by Physicians for Human Rights (PHR) with the assistance of local researchers from Herat Province.

At the time of the study, 1.1 million persons, of whom approximately 500 000 were women, were living in the 13 districts of Herat Province, which is located in northwest Afghanistan (WHO, unpublished data, 2002; Figure 1). The Herat Province was chosen for this survey because its population has remained relatively stable during the past 10 years. To obtain a representative sample of women, we randomly selected villages located within 7 districts, which had populations greater than 200 households and were within a 4-hour drive from Herat City. A total of 34 of 573 villages from 7 of 13 districts in Herat Province, Afghanistan, were included in the study. The 7 districts from which villages were sampled represented 72% (793 214/1.1 million) of the population in Herat Province (WHO, unpublished data, 2002). The villages (n = 864) located in the 6 other districts were excluded because of inaccessibility (length of drive time) and safety and cultural concerns of having female researchers staying in villages that are unknown to them. Two villages in Adraskan were excluded because they are nomadic.

For mortality estimates, we determined the sample size based on methods developed for the indirect Sisterhood Method.^{15 - 16} We assumed a maternal mortality rate of 820/100 000 as indicated by WHO/UNICEF in 1997 and assumed this rate to be only 50% accurate (410/100 000 to 1230/100 000) and used an estimate of 300/100 000, which is the lower limit of the confidence intervals (CIs) for this 1997 estimate.¹⁵ The number of respondents (according to the indirect Sisterhood Method) needed to establish a maternal mortality rate of 300 maternal deaths/100 000 live births per year within 20% was 4000 household surveys.^{15 - 16} To account for cluster sampling in Herat City, we planned to include 5000 households in the study. A total of 5014 households were selected from 7 districts in 34 established (non-nomadic) villages.

All study participants were selected using systematic random sampling (n = 4261; 87%) or a combination of systematic random sampling and cluster sampling (n = 625; 13%).¹⁷ In each village, we first asked a village elder to walk us around the entire village, and then we conducted a systematic random sample of the entire village. Urban areas in Herat City required cluster sampling due to size and difficulty in mapping. Seventy-five percent of the sample was collected in rural areas and 25% in urban areas to reflect the urban/rural representation of Herat Province.

The maternal mortality survey used the indirect Sisterhood Method¹⁵ because of the lack of national vital statistics, high fertility rate, and low migration levels in Herat Province. The indirect Sisterhood Method consists of 4 questions. (1) How many sisters (born to the same mother) have you had who were ever married (including those who are now dead)? (2) How many of these ever-married sisters are still alive? (3) How many of these ever-married sisters are dead? (4) How many of these sisters died while pregnant, or during childbirth, or during the 6 weeks after the end of the pregnancy?

The survey also contained 38 questions about respondent demographics; demographics of marriage, family, reproductive health, and pregnancy-related health care access; and beliefs about marriage, family, and reproductive health. Health care access and decisions of timing and spacing of children were assessed using Likert-type scales (eg, all of the time, some of the time, never). Opinions were assessed by a response of "agree" or "disagree" with statements concerning marriage, family, and women's roles in society.

The questionnaire was written in English, translated into Dari, the lingua franca of Afghanistan, and back-translated into English. Members of the PHR and the Afghan research team⁵ checked these translations for accuracy. Four regional, human rights, and medical experts reviewed the questionnaire for content validity. Researchers administered the survey in Dari in which they all were fluent. The survey was pilot tested among 20 women in Herat City.

The health facility survey was written in English and administered only by the PHR field supervisor (L.L.A.) with the assistance of a translator. This survey had 30 questions to assess facility characteristics, supply of medicines and equipment, and facility and staff treatment capabilities for essential and comprehensive obstetric-related problems.

The survey interviews were conducted by 48 Afghan women who were trained and supervised by the PHR field supervisor and 4 trained Afghan research team leaders. Researcher training consisted of 3 days of classroom teaching and role-play, followed by several days of field observation and continuous supervision.

All interviews were conducted over a 10-day period in March 2002. Interviews with participants lasted approximately 20 to 30 minutes and were conducted in a private setting with no one else present. All questionnaires were reviewed for completeness and for correctness of recording after the interview by the researchers themselves, and then by the Afghan research team leaders, and the PHR field supervisor at the end of each day.

This research was reviewed and approved by an independent group with expertise in clinical medicine, public health, bioethics, and international human rights research. In addition, permission for the study was granted by UN officials, representatives of international and national nongovernmental organizations, and local community leaders in each area surveyed. The research was conducted in accord with the revised Declaration of Helsinki.¹⁸ All data were kept anonymous. Oral informed consent was obtained from all participants and parental consent was obtained for all participants younger than 18 years unless they were emancipated minors (married and living out of a parental household). Participants did not receive any material compensation. There were no limitations placed on investigator movement or on surveying.

The data were analyzed using STATA statistical software.¹⁹ To control for clustering and design effect, the sample was weighted by the number of samples per location. All errors are nominal errors (not random) due to the inability to randomly sample every district. For 2 × 2 cross-tabulations containing cells with expected frequencies of less than 5, statistical significance was determined using the Fisher exact test. The Yates corrected χ² test was used for all other analyses. For cross-tabulations with greater than 2 rows, statistical significance was determined using the Pearson χ² test. Analysis of variance was used for statistical comparison of means and the Kruskal-Wallis test was used for comparison of medians. For all statistical determinations, significance levels were established at P<.05. The CIs calculated for the indirect Sisterhood Method estimation of the maternal mortality rate were based on methods developed by Hanley et al.²⁰

A household was defined as those people sleeping and eating under the same roof. Maternal mortality included deaths that occurred while pregnant, during childbirth, or during the 6 weeks after the end of the pregnancy.²¹ Reproductive age was defined as women between the ages of 15 and 49 years. An essential obstetric care (EOC) facility is one that can provide parenteral antibiotics, oxytocic drugs, anticonvulsants for preeclampsia; manually remove the placenta or retained products via manual vacuum aspiration; and can perform assisted vaginal delivery.²² A comprehensive EOC facility must be able to administer all of the basic services and perform surgery (ie, cesarean delivery) and blood transfusions.²²

Of the 5014 households sampled, 4886 females between the ages of 15 and 49 years participated in the survey (97.4% response rate). There were 105 women who were not eligible due to age exclusions; 18 were not available after 2 attempts at the time of sampling; and 5 refused to participate.

Demographics of the respondents are presented in Table 1. The mean (SE) age was 31 (0.23) years (range, 15-49 years). Half of the women surveyed were between the ages of 24 and 40 years. Women had 0.35 (0.11) years of formal education (range, 0-16 years). A total of 4589 (94%) women had less than 1 year of formal education; 128 (3%) had between 1 and 4 years; 125 (3%) had between 5 and 10 years; and 38 (0.8%) had more than 10 years. Of 4810 women, 4233 (88%) were married. Women had lived in the areas surveyed for a mean (SE) of 17 (1.1) years (range, 0.25-50 years); 50% of respondents had lived in the area between 7 and 25 years. Women were asked to rank their primary problems. The top primary problems (n = 4824) were lack of food (41%), lack of adequate shelter (18%), and lack of clean water (14%).

The 4886 household respondents reported on 14 085 ever-married sisters (Table 2). Overall, there were 276 pregnancy or childbirth-related deaths among the sisters. Of all deaths reported among the participants' sisters, 15% (276/1877) were attributed to maternal causes. The maternal mortality ratio point estimate was 593 deaths/100 000 live births (95% CI, 557-630).

Among respondents, the mean (SE) age at marriage was 15 (0.3) years (range, 5-39 years). However, the mean (SE) age reported as the most appropriate to get married was 18 (0.2) years (range, 5-30 years; P<.001 for average age of marriage vs average desired age of marriage; Table 3). Of 4721 women, 4008 (85%) stated they wanted to get married at the time of their wedding. Of 4729 women, 957 (20%) reported that they were pressured by their family to get married.

Women reported a mean (SE) number of pregnancies of 5.0 (0.08) (range, 0-20) and 4.6 (0.2) live births (range, 0-18). When asked the most appropriate age to have children, women reported 19 (0.3) years (range, 10-31 years). They chose 5.6 (0.3) children (range, 0-24) as the most appropriate number to conceive.

Of 4637 women, only 519 (11%) reported receiving prenatal care. A total of 3946 women provided reasons for lack of prenatal care. Of these women, 2939 (74%) reported financial reasons; 520 stated no health care facility was available (13%); 279 reported restrictions on movement (7%); 131 reported prenatal care was not necessary (3%); 54 reported spouse or male relative prohibited care (1%); and 23 reported work obligations conflicted with ability to get care (0.6%). Of 4703 respondents, 4117 (87%) reported that permission from a husband or male relative was required all of the time prior to seeing a trained health professional. Eight percent reported permission was required only some of the time. Five percent never had to ask for permission.

Of 4624 women, only 40 (0.9%) reported that births were attended by a trained health care worker (physician, nurse, midwife, or formally trained traditional birth attendant). A total of 4475 (97%) women reported presence of untrained traditional birth attendants.

Of 4881 women, 597 (12%) reported use of birth control methods. Oral contraceptives were used by 465 (10%); 108 used injectable medroxyprogesterone acetate (2%); 18 used barrier contraception (0.4%); 3 used natural or homeopathic methods (0.06%), and 3 had received a bilateral tubal ligation (0.06%). Of 4294 respondents, 1013 (23%) indicated that they would like to use birth control. Of 4306 women, 3189 (74%) stated that the number and spacing of children was decided equally among husband and wife; 417 (10%) by the woman only; 334 (8%) by the husband only; 202 (5%) mostly by the husband; and 164 (4%) mostly by the woman.

Eighty-six percent (4065/4748) of respondents indicated that women should have the right to freely choose a husband and enter into marriage (Figure 2). Ninety-one percent (4196/4601) also indicated that a woman should have the same right as her husband to decide the number and spacing of children. Seventy-nine percent (3786/4769) indicated that the "bride price" is a custom that should continue in Afghanistan. Eighty percent (3778/4716) expressed the view that it is a wife's duty/obligation to have sex with her husband even if she does not want to, and 45% (2168/4781) responded that a man has the right to beat his wife if she disobeys.

Table 4 shows the results of the health facility survey in 7 districts. At the time of the survey, only 63% (17/27) of the facilities listed by WHO (WHO, unpublished data, 2002) as functional were found to be operating. There were 1 comprehensive and 4 basic EOC facilities, all located in Guzara district. These facilities were within a 10- to 30-minute drive from the center of Herat City. However, they were not accessible from any of the rural districts except by more than a half a day's walk. Only 35 physicians, 59 nurses, 25 midwives, and 253 traditional birth attendants served the 793 214 people in these 7 districts. Of the 35 physicians recorded in the 7 districts, 19 were women and 15 of these were working at the provincial hospital in Herat City. Five districts had only 1 or 2 physicians. Medical records, which were available in most centers, generally consisted of only a diagnosis and prescription with no details or physical examination results. Prenatal care was offered at less than half of the facilities. The maternity hospital in Herat City, which is considered a comprehensive EOC, could handle complicated deliveries. However, this facility did not have the supplies to handle complications of pregnancy, such as eclampsia or preeclampsia, so it did not meet WHO guidelines as a comprehensive EOC. Surgery was possible in Herat City and Guzara district hospital, which are both in the urban center and are a 10 minutes' drive apart. Only the 5 Guzara EOCs had adequate drug supplies since these facilities were supplied by an international nongovernmental organization and did not rely on the health ministry to supply medications and equipment. All refrigeration systems were gas operated and most facilities listing electricity relied on generators because the provincial/district power supply was only available for 2 hours a day. Most facilities have the equipment for heat sterilization of instruments, but only 2 had chemical means as well.

The findings of this study indicate that in 7 of 13 districts in Herat Province, Afghanistan, annual maternal mortality during the last 10 to 12 years is between 557 and 630 maternal deaths/100 000 live births, or a point estimate of 593/100 000, despite having a provincial and district hospital capable of handling complicated deliveries. This estimate is more accurate and precise, but not significantly different from the 1997 modeled point estimate of 820/100 000 because it falls within the 95% CI of 300 to 1700 reported by WHO.^{15 ,20} Only 18 countries in Africa have a maternal mortality rate that exceeds the upper limit of the 95% CI for Herat Province, Afghanistan.¹⁵ This does not exclude the possibility that some regions within a particular non-African country may not exceed that of Herat Province. Data are not readily available for such comparisons. The maternal mortality ratio for Herat Province also exceeds that of all 6 countries bordering Afghanistan: Pakistan (200/100 000) and Afghan refugees in Pakistan (291/100 000), Iran (60/100 000), Turkmenistan (65/100 000), China (60/100 000), and Tajikistan (120/100 000).^{15 ,23} In contrast, the United States has an estimated rate of 12/100 000.¹⁵

Conditions for individual and community health often depend on the protection and promotion of human rights.^{24 - 26} The findings of this study identify a number of human rights factors that may contribute to preventable maternal deaths in Herat. These include access to and quality of health care services; adequate food, shelter, and clean water; and individual freedom concerns, such as marriage at an early age, inability to negotiate terms of sex, and access to birth control methods.

In 1997, WHO published guidelines that stated for every 500 000 population, there should be at least 4 basic and 1 comprehensive EOC facility.²² According to these guidelines, there should be slightly more than 4 basic and 1 comprehensive EOC facilities in the 7 of 13 districts surveyed in Herat Province. In the case of Herat hospital, considered a comprehensive EOC facility for 4 of Afghanistan's provinces (Herat, Baghdis, Farah, Faryab), this facility can only perform blood transfusions that are direct from staff to patient. Major surgery is performed without sterile procedure, and instrument sterilization consists of only boiling. Furthermore, this major maternity hospital has inadequate, outdated equipment and inadequate supplies of essential medicines and materials for surgical procedures. Consequently, the facility does not meet WHO guidelines for a comprehensive or basic EOC because its ability to care for mothers and newborns is severely compromised. Of the 7 districts surveyed in Herat, only 1 district had EOCs that met guideline standards and was in an urban area. Minimal acceptable levels of care also require that 15% of all births in the population take place in a comprehensive EOC and that 100% of women with obstetric complications are treated in EOC facilities.²² This level of care exceeds the current capacity of Herat's maternity hospital.

In this study, less than 1% of participants reported having a trained health care professional and 97% indicated the presence of an untrained traditional birth attendant at their deliveries. Although few studies support the training of traditional birth attendants alone as a way of decreasing maternal mortality,^{27 - 32} this may represent a means of preventing maternal deaths in Afghanistan in the short-term. Upgrading facilities is essential for long-term improvements, but traditional birth attendant training programs appear to warrant urgent consideration given the lack of health care facilities in Herat Province, prevailing cultural norms that generally preclude women from being examined by male physicians, and the time it will take to establish and staff facilities with qualified female midwives, physicians, and nurses.

There are other basic human rights including the right to food, shelter, and clean water that have an effect on maternal mortality. Lack of food (malnutrition with subsequent anemia) contributes to increased risk of complications in pregnant women.^{1 ,33 - 34} Chronic malnutrition in childhood frequently results in stunted growth and the development of a smaller than normal pelvic outlet.^{1 ,32} This increases the risk of obstructed births with poor outcomes for both mother and infant, especially because some marriages occur when the female is a young adolescent.^{1 ,35 - 36} In our study, 41% of women identified lack of food, 18% reported lack of shelter, and 14% reported lack of clean water as their primary problems. Lack of appropriate shelter and absence of clean water or sanitation may contribute to infection and diseases, which can complicate or be complicated by pregnancy and delivery. Lack of literacy and of educational and or employment opportunities deny women alternatives to early marriage and early childbearing, which may contribute to maternal mortality.^{1 ,37} Education is one of the best predictors of health status^{38 - 39} and information about birth spacing, contraception, health care, immunizations, and safe pregnancy is essential to the health of mothers and their children. In this study, the average number of years of formal education was 0.35 years. The apparent lack of education for women will delay the training of adequate numbers of female midwives, nurses, and physicians.

Early marriage and an inability to negotiate terms of sex including the use of contraception may contribute to maternal mortality by leading to a high number of pregnancies starting at an early age.^{1 - 2 ,37} Women in this study reported a desired age of marriage higher than actual the age (18 vs 15 years; P<.001), and although 85% indicated they did not feel forced to forced to get married by their family, there were 42 women who reported getting married when younger than 15 years. The majority of women stated that they agreed that women should choose to enter into marriage, which is similar to responses obtained in previous studies of Afghan women's opinions about women's human rights.^{3 - 5}

Restriction of movement was reported as a reason for no prenatal care by 7% of the respondents. Ninety-five percent of women stated they needed permission of a male relative to see a physician, nurse, or trained midwife, but only 1% reported that husbands or male relatives had forbidden access to prenatal care.

The majority of women (88%) stated that they had equal or primary control over number and spacing of children. This may represent an important factor in mitigating maternal deaths in Afghanistan. However, more than half of the women reported that their husbands had the right to beat them and that it was a wife's duty to have sex with her husband even if she did not want to. These stated beliefs appear to be in conflict with their assertions regarding their stated role in controlling the number and spacing of their children.

The indirect Sisterhood Method may underestimate maternal mortality.^{15 - 16} First, the pregnancy status of sisters who died may not have been known by some of the respondents. Second, abortion-related deaths may not have been revealed by sisters to protect the reputation of the family. Third, women may have died from obstetric complications after 42 days postpartum. This method provides a retrospective estimate of the annual maternal mortality rate over a period of 10 to 12 years and cannot be used to analyze trends in maternal mortality. Our sample size does not permit subgroup analysis. The exclusion of the most rural districts within Herat Province where access to and quality of services are likely to be worse than in the districts surveyed, suggests that our point estimate for maternal mortality if applied to all of Herat Province is most likely underestimated. Although it is possible that respondent's precision regarding accuracy of reported event dates may be limited, religious milestones (eg, sabbath and ramazan) were used to minimize this. While the attitudes and experiences of respondents may represent women living in Herat Province, they do not necessarily relate directly to the causes of maternal mortality among the sisters reported in this study. Although interviewers were careful to explain there would be no material or other gain by participating in the survey, women may have exaggerated deaths if they felt it was in their interest to do so.

Despite these limitations, the findings in this study indicate that women in most of Herat Province, Afghanistan, have a high risk of dying during pregnancy and childbirth. The study also identifies a number of human rights factors that may contribute to preventable maternal deaths in the region. These include access to and quality of health services, adequate food, shelter, and clean water, and denial of personal freedoms such as deciding when to marry, ability to negotiate the terms of sex, access to birth control, and beliefs that husbands have the right to beat wives who disobey them. This study suggests that high rates of maternal mortality may be an indicator of violations of women's human rights and that prevention of maternal deaths requires the protection and promotion of a wide range of women's rights over a sustained period.