Tanzanian women’s odds of maternal death due to direct causes, such as hemorrhage and eclampsia, rise sharply with the distance of their home from a hospital, according to a cross-sectional study conducted in a disadvantaged rural part of the country.1 Women living more than 35 kilometers from a hospital had nearly four times the odds of dying from such causes as did peers living within five kilometers. In contrast, distance was not associated with the odds of maternal death from indirect causes, such as malaria, HIV and anemia. Geographic access to a hospital was no guarantee of good outcomes, however: Even though nearly three-quarters of women who lived within five kilometers of a hospital gave birth in such a facility, this group still had a high rate of death during pregnancy, childbirth or the postpartum period.

Investigators performed a secondary analysis of cross-sectional census data obtained by a 2007 household survey in five districts of Tanzania’s Lindi and Mtwara regions. The survey collected sociodemographic information about all household members, and women aged 13–49 provided information about any live births they had had in the past five years, including the place and mode of delivery for births in the past year. The head of the household indicated whether any deaths had occurred in the household in recent years, and whether any such deaths among women of childbearing age had happened during pregnancy, childbirth or the postpartum period. Verbal autopsies were performed for pregnancy-related deaths that had occurred between June 2004 and May 2007, and the cause of death was classified as direct (due to obstetric factors), indirect (due to conditions aggravated during pregnancy), coincidental (due to injuries or accidents) or unknown. The investigators performed univariate analyses assessing the relationship between the pregnancy-related mortality ratio (number of pregnancy-related deaths divided by number of live births) and the straight-line distance between households and the nearest hospital and the nearest health facility of any type that provided delivery care. They used multilevel logistic regression analysis to calculate odds ratios for these associations, adjusting for household wealth, ethnicity, district and the head of household’s education level. The researchers also calculated the proportions of live births during the year before the survey that had occurred in health facilities and that were cesarean deliveries, according to distance to facilities.

Analyses were based on data from more than 818,000 people living in 226,000 households. In all, 194,000 women of childbearing age in these households had more than 64,000 live births during the study period. More than three-fourths of births were to women who lived within 25 kilometers of a hospital; nearly a fifth were to women who lived within 10 kilometers.

About a third of the live births occurred in the year before the survey, of which 29% took place in a hospital, 12% in a primary care facility (dispensary or health center) and 59% at home. Four percent were cesarean deliveries. The farther women lived from a hospital, the lower the proportion of live births that took place there and the proportion that were cesarean deliveries. For example, 72% of women living within five kilometers of a hospital gave birth there, compared with only 21% of women living more than 35 kilometers away; the proportions that were cesarean deliveries were 8% and 3%, respectively. In addition, the farther women lived from the nearest facilities of other types, the lower the proportions of live births occurring there. Some 48% and 66% of women living within a kilometer of a dispensary or health center, respectively, gave birth in those facilities.

Verbal autopsies indicated that 376 women had a pregnancy-related death during the study period. Overall, 43% of these deaths were due to direct causes, 25% to indirect causes, 1% to coincidental causes and 31% to unknown causes. Hemorrhage accounted for the largest share of death due to direct causes. Half of the deaths occurred in a facility, and a third occurred at home. Notably, 28% of the women who died had delivered in a hospital.

The pregnancy-related mortality ratio in the full cohort was 712 deaths per 100,000 live births, and ranged from 592 per 100,000 among women who lived 15–25 kilometers from a hospital to 976 per 100,000 among those who lived more than 35 kilometers away. In adjusted multivariate analysis, the odds of mortality did not differ significantly by distance to a hospital, district, or women’s ethnicity or household wealth. Compared with women who lived in a household whose head had no education, those whose head of household had completed primary school were less likely to die from pregnancy-related causes (odds ratio, 0.7); overall, the odds fell by 5% with each one-year increase in education.

The direct maternal mortality ratio increased sharply with the distance to the nearest hospital, from 111 deaths per 100,000 live births among women living within five kilometers of a hospital to 422 per 100,000 among those living more than 35 kilometers away. Relative to peers who lived within five kilometers, women who lived 25–35 kilometers away or more than 35 kilometers away were markedly more likely to experience maternal death due to direct causes (odds ratios, 2.6 and 3.7, respectively). Overall, the odds of such mortality increased by 1% with each additional kilometer of distance.

In contrast, distance to a hospital was not associated with indirect maternal mortality. Furthermore, analyses that considered all health facilities that provided delivery care (rather than just hospitals) revealed no significant associations between distance to the facility and direct or indirect maternal mortality.

The study’s findings suggest that hospitals are providing life-saving obstetric care to women in labor, but also that there is room for improvement in this care, according to the investigators, as mortality was high even among women who lived close to a hospital. Limitations of the research included an inability to interview some households and some women, lack of a verbal autopsy in about a quarter of cases of pregnancy-related death, the possibility of recall and selection biases, and lack of information on obstetric risk factors and complications. Nonetheless, taken together, the study’s findings “[underpin] the need to invest into improved access to maternal care“ and ”suggest that quality of care is an important issue,” they conclude.—S. London


1. Hanson C et al., Maternal mortality and distance to facility-based obstetric care in rural southern Tanzania: a secondary analysis of cross- sectional census data in 226,000 households, Lancet Global Health, 2015, 3(7):e387–e395, doi: 10.1016/S2214-109X(15)00048-0.