Data Source
The data was from the 2010 and 2017-2018 Multiple Indicator Cluster Survey (MICS) in the DRC. MICS, which has been conducted in 116 countries in the world, is one of the largest international household survey programs developed by UNICEF. It used a set of survey tools to generate statistically sound and internationally comparable data. To date, there have been six international rounds of the MICS. The DRC have taken part in MICS1 (1995), MICS2 (2001), MICS4 (2010) and MICS6 (2017-2018). Among these four rounds, MICS4 and MICS6 had similar questionnaire structures. They were both conducted by the National Institute of Statistics with support from UNICEF. Following the same protocol, multi-stage stratified sampling processes have been applied in the DRC, proportional to population size in each province. Overall, the surveys included 11,393 (MICS4) and 20,792 (MICS6) successfully interviewed households. Detailed sampling details can be found in the MICS4 and MICS6 summary paper[19, 20].
Participants
Of 12,851 and 21,756 women aged 15-49 in the MICS4 and the MICS6, we included women giving live births within the two years preceding the survey to assess the coverages of ANC and skilled birth attendance in their last delivery. We excluded participants with missing values in age, education attainment, marital status, household head’s sex and education attainment, household wealth index group, residential region, province, self-reported ANC visits, and skilled birth attendance. After applying the exclusion criteria, the total sample consisted of 13,313 women, with 4,759 from the MICS4 and 8,554 from the MICS6.
Exposures
The present study aims to explore the trends in coverages of ANC and skilled birth attendance among provinces in the DRC. In 2015, the original 11 provinces in the DRC were divided into the new 26 provinces. To make the province districts comparable between the MICS4 (2010) and the MICS6 (2017-2018), we rearranged the province categories in the MICS6 (2017-2018). The detailed correspondence can be found in Additional file Table S1 [see Additional file 1]. Totally, 11 historical province categories were used in the analysis: Katanga, Kasai Oriental, Kasai Occidental, Kinshasa, Bas Congo, Bandundu, Equateur, Province Orientale, Maniema, Nord Kivu, and Sud Kivu.
Additionally, we aimed to study the differentiated trends of ANC coverage and skilled birth attendance among different wealth index groups. In the MICS data, the household wealth index was categorized into five quantiles (poorest, poorer, middle, wealthier, and wealthiest) based on an asset-based wealth index[21]. In the present study, household index groups were used to present the SES of participants.
Outcomes
To ensure comparability with previous studies, we adopted the definitions from WHO[22]. Antenatal care coverage was defined as the percentage of women who utilized antenatal care provided by skilled health personnel for reasons related to pregnancy at least once during pregnancy among all women who gave birth to a live child in a given period. Qualified health personnel includes physicians, nurses, and midwives. For skilled birth attendance coverage, the definition was the proportion of births attended by skilled health personnel among all births. Information about whether the participants having at least one antenatal care provided by skilled birth attendants or having skilled attendance at delivery was retrieved from the MICS database.
Covariates
Based on the literature review[23–27], we chose women’s age, education attainment, marital status, household heads’ sex, household heads’ education attainment, and residence region as covariates. Women were classified into four age groups:15-19, 20-29, 30-39, and 40-49. Women and their household heads’ education attainment were re-categorized into three groups: below the primary school, primary school, and secondary or higher school. Women’s marital status information was also retrieved from the original datasets and re-categorized into two groups: currently single and married or living with a partner. For the residence region, it is worth noting that Kinshasa, the capital of the DRC, refers both to a city and a province. Thus, there is no rural region in Kinshasa.
Statistical analysis
Descriptive statistics were calculated to show the characteristics of survey participants. The overall coverages of ANC and skilled birth attendance were calculated using sampling weights provided in the MICS database. The weighted coverages were then cross-validated with the survey findings in the UNICEF reports[19, 20]. A logistic regression-based adjusted prevalence method was used to calculate the adjusted coverages of ANC and skilled birth attendance. Details for the method were described elsewhere[28, 29]. Briefly, the method is based on logistic regression. Adjusted values were calculated using the floating method. By attributing variance to the reference group, the floating method makes it possible for comparisons of risks between any two groups[30]. The coverages of ANC and skilled birth attendance for participants from different provinces in 2010 and 2018 were calculated, adjusted for women’s age, education attainment, marital status, household heads’ sex and education attainment, residence region, and household wealth status. Maps were drawn to visualize the trends in coverages of adjusted ANC and adjusted skilled birth attendance from 2010 to 2018. To further compare the trends in conflicted but unassisted regions (Kasai region), conflicted and already assisted regions (Kivu region), and relatively non-conflicted regions (other provinces), subgroup analyses were applied. In each region group, adjusted ANC coverage and adjusted skilled birth attendance coverage in different wealth index groups were calculated separately. Percentages with 95% confidence intervals were reported for the adjusted coverage. Data were processed by R 4.0.0 (R Core Team, 2020). Adjusted coverages were calculated by SAS 9.4 statistical software (SAS Institute, Cary NC).