Causes of death among women aged 17-49 years between 2007 and 2010 in Maputo, Mozambique

Objectives To describe causes of death among young women and estimate the role of HIV/AIDS as a cause in Maputo City, based on the civil death register. Methods Death data of 17-49 year-old women were abstracted from January 2007-March 2010 from the civil death register in Maputo City, registering overall about 15 000 deaths per-year. Causes of death in the register were either based on physicians’ diagnoses on death certificates or determined by asking questions to deceased relatives. Causes of death were written in Portuguese; we translated them into English and classified them into 106 codes using ICD-9; these codes were then categorized into 10 groups. Estimated populations from the 2007 census were used to calculate annual mortality rates. An earlier study was used to compare deaths in 2001. Findings A total of 9640 deaths (6510 for residents of Maputo City) were registered and 77% had a specified cause of death reported. HIV- deaths represented 36% of all deaths and 40% among 25-39 year-olds. The death rate did not increase linearly by age, as there was a peak among women aged 30-34 years. The overall annual death rate was 6.7 deaths per 1000 population, with a notable decline by year. Death rates for HIV slightly declined by year. HIV-deaths explained most of the peak in death rate among 30-34-year-olds. The share of HIV- deaths among all deaths increased from 18% in 2001 to 35% in 2007- 2010. Sixty-eight percent of all and 92% of HIV-related deaths occurred in hospital, with no increase over time. Conclusions Routine death register was useful to study death rates, distribution of deaths, and change over time in the urban setting of Maputo during late 2000s. Over time, the death rate among 17-49 years old women seemed to have declined, but the relative contribution of HIV increased. Originally the death data were collected to complete the tracing of 4326 pregnant women recruited for a trial in Maputo [10]. In that trial, the effects of two iron administration policies (routine iron prophylaxis vs screening and treatment for anemia during pregnancy) on maternal and child health were compared. The results from the trial have been presented in a previous paper [10]. Ethical approval for the study was obtained from the Mozambique Ministry of Health Ethics Committee (CNBS [Ref. 84/CNBS/06]) and Eduardo Mondlane University Medical Faculty Ethics Board (Jan 25, 2006) [10]. A positive statement was obtained from the National Research and Development Centre for Welfare and Health (STAKES), Helsinki, Finland (Dno 2571/501/2007). Oral and written informed consent was obtained from participating women [10]. The deaths of all 17–49 year–old women registered in the Maputo City civil death register (Registo Civil) from January 2007 to March 2010 were included in the study. The death registration system in Maputo is described in Online Supplementary Document(Online Supplementary Document). In brief, there was only one civil death register (Registo Civil) in Maputo City, registering overall about 15 000 deaths per year. Deaths were registered by the location of death, and many deaths of people from nearby areas were registered in Maputo City. If the death occurred at home, relatives reported it to a civil administrative office, where a letter of declaration of death was written. The relatives took the letter to the Registo Civil office, where a death report was written and a death certificate issued. If the death occurred in a health facility, the death certificate was filled in by a physician. The death certificate was taken to the Registo Civil by the relatives or a person from the Registo Civil located in a mortuary or in hospital. In the Registo Civil death reports were first piled as loose papers and later compiled into books. All documents were handwritten (Online Supplementary Document(Online Supplementary Document)). The cause of death was reported in death reports (Online Supplementary Document(Online Supplementary Document)). In the case of a hospital death a physician wrote the cause on the death certificate, and the Registo Civil official copied it into the death report. If the deceased relatives had no death certificate, the Registo Civil official asked certain questions to elicit the causes of death. For this study we collected the following information from the death reports in Registo Civil: register number, age, date of birth, residence, date of death, place of death (hospital/out of hospital), date of death registration, basic cause of death (written in Portuguese or a local abbreviation code). Data were then entered by hand on data collection sheets, and later computerized into Microsoft Access 2000 database. We did not very the causes of death, but took the information as noted in the death register. Causes of deaths were hand written in Portuguese, either as diagnoses, lay terms or by local abbreviation codes. The different terms (about 1000 and 534 when similar terms were combined) were translated into English (a Finnish researcher with nursing background, fluent in both Portuguese and English, SP). The English terms and the local abbreviation codes were classified using the 9th version of the International Classification of Diseases and Related Health Problems (ICD–9 codes) into 106 codes. The coding was made by a physician (EH), discussing with other researchers. The most common abbreviations were clarified by the local researchers. The ICD codes were grouped into 18 and 10 groups (see Results) to reflect the likely cause of death, taking into account the inaccuracy at the registration stage. Mostly only one cause of death was given. In a few cases two causes were given, eg, HIV and tuberculosis. In that particular case, we classified the cause as HIV. All classifications were made solely on the basis of the causes of death and women’s background characteristics were not used in the coding process. For analysis, the data in Microsoft Access 2000 format was transformed into SPSS. Cross–tabulations of the grouped causes of deaths were made by age and year, and the proportions of HIV deaths were calculated. In the 2007 death register books, 12 women died in 2006: they were included in the total deaths, but excluded from yearly data. Maputo City residents were defined by the information in the Register. To calculate annual mortality rates for Maputo City residents, the projected population numbers from the Instituto Nacional de Estatistica (INE) were used [17]. The INE used 2007 census numbers to make projections for the coming 40 years. The projected numbers were given by 5–year age groups, sex and province, separating Maputo City, as Maputo City is informally treated as an independent province. We calculated the total death rates (per 1000 population) across age groups by dividing the total number of deaths by the population size of each age group. Similarly, HIV–related death rates (per 1000 population) were calculated across age groups by dividing the number of HIV–related deaths by the population size of each corresponding age group. We calculated the confidence interval for the death rates using the Wilson score method [18]. At the time of the data collection, death data for 2010 was available for the first three months of the year. For calculating the annual HIV–related death rates for 2010, we multiplied the death rate by four. As some deaths, particularly malaria, are sensitive to the time of year, the 2010 results are given with reservation (see Results). To compare our results to those reported in 2001 we relied on a table describing deaths of 15–44 year–old women, who had been registered in the Maputo City death register [9]. To make our study population more similar to the 2001 population, women aged 45–49 were excluded. The grouping of causes of deaths was modified so that it matches the grouping used in the 2001 study, which was based on the Burden of Disease Study categories [9].