Demographic and Health Surveys: caesarean section rates in sub-Saharan AfricaBMJ 2003; 326 doi: http://dx.doi.org/10.1136/bmj.326.7381.136 (Published 18 January 2003) Cite this as: BMJ 2003;326:136
- Pierre Buekens (), professora,
- Siân Curtis, senior evaluation specialistb,
- Silvia Alayón, research associateb
- a University of North Carolina at Chapel Hill, Department of Maternal and Child Health, School of Public Health, Rosenau Hall, CB#7445, Chapel Hill, NC 27599-7445, USA
- b ORC Macro International, 11785 Beltsville Drive, Calverton, MD 20705, USA
- Correspondence to: P Buekens
- Accepted 15 August 2002
It is well documented that caesarean section rates have increased in many Latin American and Asian countries, and that many of these interventions are unnecessary. 1 2 Fewer data are available on trends in caesarean section rates in sub-Saharan Africa, where concern centres on lack of access to necessary interventions.3
Participants, methods, and results
We analysed Demographic and Health Surveys performed at two different times in eight sub-Saharan African countries—Burkina Faso (1992 and 1999), Cameroon (1991 and 1998), Ghana (1993 and 1998), Kenya (1993 and 1998), Madagascar (1992 and 1997), Niger (1992 and 1998), Tanzania (1992 and 1996), and Zambia (1992 and 1996).4 We calculated caesarean section rates for singleton live births that occurred during the three years before the interviews, weighted for the complex sample designs. We calculated urban and rural rates according to the residence of the mothers. We used a logistic regression of the pooled data from the eight countries to calculate the crude odds ratio of having a caesarean section during 1991–3 (reference group) and 1996-9. We calculated confidence intervals after taking the complex design into account. In a next step, we added mother's age (<20, 20-34, ≥35 years), parity (1, 2-4, ≥5), and residence (urban, rural) to the logistic regression. We also calculated the proportion of singleton live births delivered in health facilities and included place of delivery in the logistic regression.
The table shows that caesarean section rates were lower than 5% in all countries except Kenya, and that the rates were lower than 2% in Burkina Faso, Madagascar, Niger, and Zambia. Caesarean section rates decreased in five countries between 1991–3 and 1996-9. The proportion of singleton live births delivered in health facilities decreased between 1991–3 and 1996–9 in Burkina Faso (43.1% and 33.3%, respectively), Cameroon (60.9% and 54.0%), Kenya (42.5% and 42.3%), Madagascar (44.0% and 34.2%), Tanzania (52.4% and 46.6%), and Zambia (50.3% and 44.8%). It increased in Ghana (41.9% and 44.2%) and Niger (16.0% and 17.8%).
The overall odds ratio for the trend was 0.88 (95% confidence interval 0.77 to 1.00). After adjustment for age, parity, and urban or rural residence, the odds ratio was 0.89 (0.79 to 1.01). After including whether delivery was in a health facility in the logistic regression, we found the odds ratio increased to 0.94 (0.83 to 1.07).
Our results show that access to caesarean sections is not improving in sub-Saharan Africa and that it might be worsening. Including delivery in a health facility in the logistic regression rendered the odds ratio closer to 1, suggesting that the trend toward a decline of caesarean section rates is partly linked to a decline in access to health services. This shows the lack of progress of “Safe Motherhood” programmes in a region where maternal mortality is extremely high.
There is suggestive evidence that a caesarean section rate of 3.6% to 6.5% (median 5.4%) is needed to address obstetric complications in west Africa, and that a rate of 2% is a minimum. 3 5 Our results therefore point to an urgent need for better access to caesarean sections in sub-Saharan Africa. However, any support programmes should be carefully designed to avoid the risk of simultaneously increasing unnecessary caesarean sections and iatrogenic morbidity and mortality. Support should be part of a global effort toward appropriate use of caesarean sections, limiting its frequency where it is overused, and increasing its frequency where use is too low.
We thank José Belizán and Luc de Bernis for their useful comments and suggestions.
Contributors: PB and SC contributed to the conception and design of the study. PB, SC, and SA participated in the analysis and the interpretation of the data. PB took the lead role in writing the paper. SC and SA contributed to writing and revising the paper. PB and SC are guarantors.
Funding This work was supported in part by a cooperative agreement (HRN-A-00-97-0001-8-00) with the United States Agency for International Development.
Competing interests None declared.