Menu
Menu
Menu
Menu
Background:Pregnancy-related (PR) deaths are often a result of direct obstetric complications occurring at childbirth.Methods and Findings:To estimate the burden of and characterize risk factors for PR mortality, we evaluated deaths that occurred between 2003 and 2008 among women of childbearing age (15 to 49 years) using Health and Demographic Surveillance System data in rural western Kenya. WHO ICD definition of PR mortality was used: “the death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the cause of death”. In addition, symptoms and events at the time of death were examined using the WHO verbal autopsy methodology. Deaths were categorized as either (i) directly PR: main cause of death was ascribed as obstetric, or (ii) indirectly PR: main cause of death was non-obstetric. Of 3,223 deaths in women 15 to 49 years, 249 (7.7%) were PR. One-third (34%) of these were due to direct obstetric causes, predominantly postpartum hemorrhage, abortion complications and puerperal sepsis. Two-thirds were indirect; three-quarters were attributable to human immunodeficiency virus (HIV/AIDS), malaria and tuberculosis. Significantly more women who died in lower socio-economic groups sought care from traditional birth attendants (p = 0.034), while less impoverished women were more likely to seek hospital care (p = 0.001). The PR mortality ratio over the six years was 740 (95% CI 651-838) per 100,000 live births, with no evidence of reduction over time (χ2 linear trend = 1.07; p = 0.3).Conclusions:These data supplement current scanty information on the relationship between infectious diseases and poor maternal outcomes in Africa. They indicate low uptake of maternal health interventions in women dying during pregnancy and postpartum, suggesting improved access to and increased uptake of skilled obstetric care, as well as preventive measures against HIV/AIDS, malaria and tuberculosis among all women of childbearing age may help to reduce pregnancy-related mortality. © 2013.
The study site is located in a rural part of Nyanza Province in western Kenya in the areas of Asembo (Rarieda District), Karemo (Siaya District) and Gem (Gem District) in Siaya County [9], [10]. The population comprises approximately 225,000 individuals living in 385 villages spread over 700 km2. It has a typical rural African population age distribution with 44.6% under 15 years of age, and only 5.5% over 65 years of age. By 2008, a total of 94,106 persons were aged 15–49 years, 41.7% of the population, of whom 50,820 (54%) were women of childbearing age. The population is culturally homogeneous; over 95% are members of the Luo ethnic community and live through subsistence farming and local trading. The society is polygynous, with males frequently having more than one wife, each of whom lives in a separate house with young children within a single compound. HDSS residents (defined as those residing in the study area for at least 4 consecutive months or infants born to residents) are visited every four months. Previous studies identified the population to be generally very poor [11]. Malaria is endemic in this area, and transmission occurs throughout the year. The prevalence of malaria among individuals over 15 years of age ranged between 10–20% in the period 2006 to 2008 (KEMRI/CDC, unpublished observations). HIV, tuberculosis (TB) and geohelminth prevalence are also some of the highest in the country. In the period between 2003 and 2008, in the HDSS, the prevalence of HIV among girls between 15–19 years of age was estimated at 8.6% [12], the prevalence of TB in individuals over 15 years of age was 600/100,000 [13], and geohelminth prevalence in pregnant women was recorded to be as high as 76.2% [14]. During this time period, HIV treatment and care centers expanded [15], the coverage of malaria interventions (insecticide-treated bednets and intermittent preventive treatment in pregnancy) increased [16], and training of healthcare workers to provide focused antenatal care was rolled out [17]. There was a gradual shift in Kenyan policy from allowing traditional birth attendants (TBAs) to conduct deliveries to redefining their role as referral agents and birth companions. There are 36 health facilities in the HDSS, including one district hospital, two privately owned hospitals, 11 health centers and 22 dispensaries. The entire population is registered and geo-spatially located within the HDSS [10]. A household census (“round”) is conducted three times per year to capture pregnancies, births, deaths, and internal migration. Socio-economic status (SES), educational and marriage status data are collected every two years from all HDSS residents. Demographic data are used to provide mid-year denominators per 5-year age group, stratified by gender and study area. Deaths are captured in two ways. First, village reporters report all deaths to HDSS field supervisors as they occur. Second, community interviewers record any deaths that occurred during the prior 4 months at each routine HDSS round. Field staff then visit the GPS-located coded households at least one month after the reported death to validate deaths and record events surrounding death using VA. VAs [18] are administered to the primary caregiver of the deceased. A standardized questionnaire is used [19], to cover demographic and personal history, pre-mortem illness signs and symptoms, and events surrounding the death. VA is conducted for all deaths. The adult questionnaire is restricted to persons aged 15 years and above. For this analysis, we included women of childbearing ages: 15 to 49 years. Deaths were linked to HDSS data including socio-demographic, educational, marital status, and occupational information. For all deaths, VA information was reviewed independently and conflicts resolved by at least two clinical officers (equivalent to physician assistants in the U.S.A.) and one underlying cause of death assigned. Further details of the VA methodology used in the HDSS have been provided elsewhere [20]. Through the year 2007, VA questionnaires asked for information on miscarriage related to both spontaneous and induced abortions. In 2008, the standardized WHO VA questionnaire which only asked women to report induced abortions was adopted. However, as abortion is illegal in Kenya, we assume that the data gathered in 2008 predominantly capture spontaneous miscarriage. VA data are limited in their ability to differentiate between miscarriage and abortions, thus we do not present data separately by these categories. Following cultural customs, compound heads provide written consent for all compound members to participate in the HDSS activities. Any individual can refuse to participate at any time. The HDSS protocol and consent procedures, including surveillance and VA, are approved by KEMRI and CDC Institutional Review Boards annually. Data analyzed included all deaths in the HDSS occurring between January 1, 2003 and December 31, 2008 among female residents aged 15–49 years at the time of death. Karemo area (Siaya County), the immediate catchment area of the Siaya District Hospital, was included in the HDSS in 2008 only. The following WHO ICD definition of pregnancy-related (PR) mortality was used: “the death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the cause of death”. Deaths were further categorized as either (i) directly PR, where the main cause of death determined by VA was obstetric, or (ii) indirectly PR, where the main cause of death ascribed through VA included any non-obstetric cause: infectious, non-infectious, or external causes. Data analyses were conducted using SPSS for Windows (Release v18.0), and EpiInfo Stat Calc (CDC Atlanta, USA). In the absence of comparative data among survivors, within-death comparisons were made between PR and non-PR deaths to explore differences in characteristics, subdividing analyses into the WHO grouping of died in pregnancy, died after miscarriage/abortion, and died within 42 days of pregnancy. Key social and demographic characteristics included marital status (ever married; divorced or widowed at time of death), education (attended and completed primary school; attended secondary school), SES, and place of death (home, health facility, hospital, on route to/from hospital/health facility). A hospital is a district level or above facility, and a health facility is a local lower level facility. Routinely collected SES indicators such as occupation of household head, primary source of drinking water, use of cooking fuel, in-house assets (e.g. lantern lamp, sofa, bicycle, radio and television) and livestock (poultry, pigs, donkey cattle, sheep and goats) were used to calculate a wealth index as a weighted average using multiple correspondence analysis [21]. This was used to rank households into wealth quintiles with the first quintile representing the poorest and the fifth representing the least poor; for some analyses we collapsed into most (quintiles 1–2) and least (quintiles 3–5) poor. The significance of changes in rates over time was examined using Mantel Haenszel χ2 for linear trend. Differences between groups were determined using Pearson’s χ2 test and Fisher’s Exact test for small numbers, and a p-value of <0.05 was considered statistically significant. The pregnancy-related mortality ratio (PRMR) was calculated as the number of deaths among women of childbearing years (15–49 years) over the total number of live births to women of the same age range per year. The HDSS data are stored securely and, through a formal process of data sharing established at KEMRI/CDC, are available for access to the scientific public two years after the data are cleaned and frozen.
N/A
