Menu
Menu
Menu
Menu
Background: Despite caesarean section (CS) being a lifesaving intervention, there is a noticeable gap in providing this service, when necessary, between different population groups within a country. In Burundi, there is little information about CS coverage inequality and the change in provision of this service over time. Using a high-quality equity analysis approach, we aimed to document both magnitude and change of inequality in CS coverage in Burundi over 7 years to investigate disparities. Methods: For this study, data were extracted from the 2010 and 2016 Burundi Demographic and Health Surveys (BDHS) and analyzed through the recently updated Health Equity Assessment Toolkit (HEAT) of the World Health Organization. CS delivery was disaggregated by four equity stratifiers, namely education, wealth, residence and sub-national region. For each equity stratifier, relative and absolute summary measures were calculated. We built a 95% uncertainty interval around the point estimate to determine statistical significance. Main findings: Disparity in CS was present in both survey years and increased over time. The disparity systematically favored wealthy women (SII = 10.53, 95% UI; 8.97, 12.10), women who were more educated (PAR = 8.89, 95% UI; 8.51, 9.26), women living in urban areas (D = 12.32, 95% UI; 9.00, 15.63) and some regions such as Bujumbura (PAR = 11.27, 95% UI; 10.52, 12.02). Conclusions: Burundi had not recorded any progress in ensuring equity regarding CS coverage between 2010 and 2016. It is important to launch interventions that promote justified use of CS among all subpopulations and discourage overuse among high income, more educated women and urban dwellers.
Home to over 11 million people, Burundi is the third most densely populated country in sub-Saharan Africa (SAA) with an estimated 463 inhabitants per km2 [23], and an increasing population that is expected to double by 2040 [18]. Plagued by political uncertainty and violence, the country is poor and severely fragile in terms of its security, economy, society, politics and environment [24]. Although the under-five child mortality rate in Burundi has gradually decreased from 156.4 deaths per 1000 live births in 2000 to 58.5 deaths per 1000 in 2018 [25], it is still over twice the target of less than 25 per 1000 set by the UN 2030 Agenda [26]. The maternal mortality ratio per 100,000 live births has also dropped in Burundi over the past two decades from 1010 deaths per 100,000 live births in 2000 to 538 deaths per 100,000 live births in 2017 and still above the UN 2030 Agenda target of less than 70 deaths per 100,000 by 2030 [26, 27]. Although under-five child and maternal mortality rates have improved over the past two decades in Burundi, they continue to be higher than target indicators [28]. Furthermore, the 2015 political crisis hampered service delivery, particularly affecting maternal and child health services during this time [28]. Significant disparities in coverage and utilization due to socioeconomic status (i.e. financial barriers) and access (i.e. rurality, transport) to maternal and child health services continue in Burundi [28]. The 2010 and 2016 Burundi Demographic and Health Surveys (BDHS) were used in this study. DHS uses a stratified two-stage cluster design where the first stage includes Enumeration Areas (EA) selected through a Probability Proportional to Size approach where large more representative EAs have a higher chance of being in the sample than the small EAs. In the second stage, a random sample of households are drawn from the selected EAs. The household surveys collected data on maternal reproductive and child health in Burundi representative at the national, residence and regional level [29]. Even though the BDHS was carried out in 1987, 2010 and 2016, the 1987 BDHS is not available in the WHO HEAT software; therefore, we confined our analysis to the 2010 and 2016 rounds. Detailed information regarding the BDHS study design are published elsewhere for 2010 [30] and 2016 [31]. A total sample of 24,520 women participated in the 2010 BDHS, of whom 16,778 had a birth in the last 5 years, and 375 births were through CS while remaining 7323 did not have a CS [30]. In 2016, a total sample of 45,419 women were surveyed, of whom 32, 312 had birthed a child and answered the question whether the child was born by caesarean section; 786 had a caesarean and 12,321 did not have a caesarian [31]. CS is defined as the percentage of births delivered by caesarean section among all live births in the 5 years prior to the surveys for the last birth and the next-to-last birth. The question asked: “Was (name) delivered by caesarean, that is, did they cut your belly open to take the baby out?”; the answer options provided were “yes” or “no” [30, 31]. CS was the outcome variable of interest and inequality was measured according to the four equity stratifiers: economic status, education, place of residence, and subnational region. Economic status was approximated through wealth index and is classified into five quintiles: poorest, poor, middle, rich and richest. Wealth index is a standard socioeconomic variable in the DHS and we used the variable as it is. Using a statistical data reduction technique, Principal Component Analysis, wealth index is computed based on household assets and possessions following methods introduced by the Rutstein SO and Johnson K [5]. Educational status is categorized as no education, primary, secondary or higher; place of residence as urban versus rural, and the sub-national region included five and 18 regions in 2010 and 2016 surveys, respectively. Education and wealth have a natural ordering and are known as ordered equity stratifiers whereas place of residence and regions are non-ordered equity stratifiers. The type of summary measures to be calculated are partly determined by whether or not the equity stratifiers are ordered or not [21]. Using the recently updated WHO’s HEAT software (2019 update) [22], we analyzed the socioeconomic and area-based CS disparities. The inequality analysis was completed by, first, disaggregating CS according to the above-mentioned dimensions of inequality and, subsequently interpreting the findings derived from summary measures. Then, we calculated absolute inequality summary measures that included Difference (D), Population Attributable Risk (PAR), and Slope Index of Inequality (SII). These measures were calculated for the four equity stratifiers; for the wealth and education dimensions of inequality, we calculated all of the three inequality measures, and for the region and place of residence, we only calculated the D and PAR. The detailed methods regarding calculation and interpretation of the measures used in the study have been detailed elsewhere [21]. Since CS is a favorable indicator (i.e. a lifesaving measure), positive values of a measure show disproportionate use of the service among the advantaged sub-groups, while negative values indicate that disadvantaged groups are using the service most. The higher their absolute value, the greater the inequality. Zero value for the measures show absence of inequality. D is a simple measure that shows absolute difference between two categories. The other two (PAR, SII), on the other hand, are weighted complex measures of inequality and take into account sizes of all the subpopulations used in the calculation, thereby producing more robust estimates that could represent the entire subpopulation [20, 21]. SII was computed for education and wealth equity stratifiers as it requires an ordered variable. Each point estimate is accompanied by an Uncertainty Interval (95%UI) to identify CS disparities that are statistically significant and to determine whether or not the inequality changed with time. For all inequality measures, the lower and upper bounds of the UI must not contain zero for CS inequality to exist. We assessed the trend of inequality for each summary measure by referring to the UIs for the different survey years; if the UIs did not overlap, inequality existed. The findings were presented per the recommendation of the Strengthening Reporting of Observational studies in Epidemiology (STROBE) reporting guidelines [32]. The demographic health surveys are available publicly and ethics approvals were completed by institutions that commissioned, funded, and managed the surveys. DHS surveys are approved by Inner City Fund (ICF) International and an in-country Institutional Review Board (IRB) to ensure protocols are in compliance with the U.S. Department of Health and Human Services regulations for the protection of human subjects.
N/A
