Menu
Menu
Menu
Menu
Background: Giving birth in health facilities with skilled birth attendants (SBAs) is one of the key efforts promoted to reduce preventable maternal deaths in sub-Saharan Africa. However, research has revealed large socioeconomic status (SES) disparities in facility-based childbirth. We seek to extend the literature on the factors underlying these SES disparities. Drawing on the Disparities in Skilled Birth Attendance (DiSBA) framework, we examined the contribution of three proximal factors—perceived need, accessibility, and quality of care—that influence the use of SBAs. Methods: We used data from a survey conducted in Migori County, Kenya in 2016, among women aged 15–49 years who gave birth nine weeks before the survey (N = 1020). The primary outcome is facility-based childbirth. The primary predictors are wealth, measured in quintiles calculated from a wealth index based on principal component analysis of household assets, and highest education level attained. Proposed mediating variables include maternal perceptions of need, accessibility (physical and financial), and quality of care (antenatal services received and experience of care). Logistic regression with mediation analysis was used to investigate the mediating effects. Results: Overall, 85% of women in the sample gave birth in a health facility. Women in the highest wealth quintile were more likely to give birth in a facility than women in the lowest quintile, controlling for demographic factors (adjusted odds ratio [aOR]: 2.97, 95% CI: 1.69–5.22). College-educated women were five times more likely than women with no formal education or primary education to give birth in a health facility (aOR: 4.96; 95% CI: 1.43–17.3). Women who gave birth in health facilities had higher perceived accessibility and quality of care than those who gave birth at home. The five mediators were estimated to account for between 15% and 48% of the differences in facility births between women in the lowest and higher wealth quintiles. Conclusion: Our results confirm SES disparities in facility-based childbirth, with the proximal factors accounting for some of these differences. These proximal factors – particularly perceived accessibility and quality of care – warrant attention due to their relationship with facility-birth overall, and their impact on inequities in this care.
The data for this analysis are from a 2016 cross-sectional survey conducted in Migori County, Kenya which is described in detail elsewhere [6, 22, 23]. To summarize, data were collected from women aged 15–49 years who delivered in the nine weeks before survey administration. Migori is a predominantly rural county in western Kenya with 8 sub-counties and a population of about one million people [6]. The county has one referral hospital, seven sub-county hospitals, 18 health centers, several dispensaries, and a few faith-based and private health facilities. The estimated total fertility rate for the county is 5.2 children per woman [6]. A multistage sampling approach explained in detail elsewhere [23] was used to select women from each of the 8 sub-counties, with a target of interviewing 200 women from each sub-county. First, Migori County was divided into 8 strata (the 8 sub-counties), within each stratum, 10 community health units were randomly selected. In the Kenyan health service delivery structure, a community health unit is a geographic area set to include approximately 5000 people [23, 24]. From the health unit, women who gave birth within 9 weeks were identified with the help of community health volunteers that were assigned to that community health unit. The study interviews were conducted by trained study field staff in English, KiSwahili, and DhLuo. A total of 1,052 women were interviewed. For this analysis, we used data from 1,020 women with complete information on all the variables of interest. All participants provided informed consent after receiving information about the research from the study team. Participants under the age of 18 years were considered emancipated minors with the ability to give informed consent for themselves because they had recently given birth and were included in the study to represent the population of emancipated minors. Ethical approval for the study was provided by the University of California San Francisco and Kenya Medical Research Institute IRBs. The measures used for the analysis were informed by the DiSBA framework. Health Facility Delivery. Participants were asked: “Did you deliver in a home or health facility?” From the responses, Home (0) or Health facility (1), we created a binary outcome variable: facility-based delivery. Household wealth was measured in quintiles calculated from a wealth index based on the principal component analysis of variables on household assets [25]. The wealth variable was coded as a categorical variable: Poorest (0), Poorer (1), Middle (2), Richer (3), and Richest (4). Education. To determine the highest level of education attained, participants were asked: “What is the highest grade or class that you completed at school?” Response options included the following: No school (0), Attended primary but did not finish (1), Primary (2), Post primary or vocational (3), Secondary (4), College (middle level; 5), and University or above (6). We recoded the education variable by combining smaller categories: No school/primary (0), Post-primary/Vocational/Secondary (1), and College or above (2). We measured the latent variables of perceived need, accessibility, and quality of care using additive indices. Lower scores indicate lower or less positive perceptions, and higher scores denote higher and more positive perceptions. The perceived need for maternal health services variable (Cronbach’s alpha = 0.6) is based on five survey questions, each with a 5-point Likert scale response [17]: Strongly disagree (0), Disagree (1), Neither agree nor disagree (2), Agree (3), and strongly agree (4). The survey statements below attempt to capture the perceived need for maternal health services. The responses to the first three statements were reverse coded, and the responses to all the questions were summed to create a score ranging from 0 to 20. The perceived financial access to health services during childbirth variable (Cronbach’s alpha = 0.7) is based on two survey questions: To both questions, women responded: Very easy (0), Easy (1), Difficult (2), and Very difficult (3). We reverse coded both questions and added the responses to create a score ranging from 0 to 6 (2*3). The perceived physical access to health services during childbirth variable (Cronbach’s alpha = 0.7) is based on two survey questions [17]: Very easy (0), Easy (1), Difficult (2), and Very difficult (3). We reverse coded both questions so that higher perceived physical access produced higher scores. The lowest possible score was 0, and the highest possible score was 6 (2*3). The perceived provision of care variable (Cronbach’s alpha = 0.4) is based on service provision measures during antenatal care, presented as nine survey questions (see Appendix I) asking whether the participants received various services during antenatal care. The responses to the nine questions varied. Five questions were binary: No (0); Yes (1). One question had a three-level response: No (0); Yes, once (1); Yes, more than once (2). Three questions had four-level responses: No (0); Yes, a few times (1); Yes, most times (2); Yes, all of the time (3). The lowest possible score was 0, and the highest possible score was 16 (5*1 + 1*2 + 3*3). The perceived experience of care variable (Cronbach’s alpha = 0.8) is based on 17 experience-of-care survey questions during antenatal visits (see Appendix II in Supplementary file 1). Most women (n = 1019) attended at least one antenatal care appointment during their pregnancy regardless of their childbirth location. The experience-of-care questions capture the provider-patient communication and feeling of dignity and respect. Six questions had binary responses: No (0); Yes (1). Eleven questions had four-level categorical responses: No (0); Yes, a few times (1); Yes, most times (2); Yes, all of the time (3). The lowest possible score was 0, and the highest possible score was 39 (6*1 + 11*3). Age, marital status, parity (the number of prior births), prior facility-birth, literacy, partner occupation, and paid employment were specified as control variables; they are hypothesized to confound the estimated effect of SES on the delivery outcome. Our analysis is informed by the DiSBA conceptual framework and relies on the causal relationships represented in the directed acyclic graph in Fig. 1 [17]. We present descriptive statistics to depict the distributional characteristics of key variables. The characteristics of women who gave birth at a health facility and those who did not were compared using chi-squared tests and t-tests for categorical and continuous variables, respectively. To address possible violations of distributional assumptions we verified results using Fisher’s Exact and Wilcoxon tests. Directed acyclic graph describing the association between SES predictors and delivery facility with control variables to illustrate perception-mediating pathways We first examined the association between wealth (4-level categorical) and facility-based delivery (binary) using logistic regression. The models were constructed by first adding the primary predictor (model 1), followed by the covariates represented as possible confounders in Fig. 1 (model 2), and finally, the mediating predictors (model 3). We followed the same analytic approach in examining mediation of the association between education (3-level categorical) and facility-based delivery. Standard methods for mediation analysis based on linear models for continuous outcomes correctly estimate the indirect or mediated effect as the difference between the coefficients in models including and excluding the mediating predictor [26]. This approach does not reliably estimate indirect effects for logistic regression related to the non-collapsibility property of odds ratios [27]. Our analysis is based on the Karlson-Holm-Breen (KHB) rescaling method to account for this limitation [27, 28]. The binary logistic model without the mediators (reduced model) was rescaled so that the coefficients of the independent key variables, household wealth and education (cn), in the reduced model were comparable to the coefficients of wealth and education in the full model (cn’) containing the mediating variables. The indirect effect or mediated effect was calculated as cn-cn’ for each coefficient, and the total mediated effect percentage was [(cn-cn’)/cn] * 100. We performed diagnostic tests to ensure the logistic regression models were well-specified (using a goodness-of-fit test) and checked for collinearity between included predictors [29]. We also estimated Bayesian information criterion (BIC) values to compare the overall fit between nested models. A sensitivity analysis was carried out to assess the change in target odds ratio estimates resulting from exclusion of SES covariates. Stata (version 17.0) was used for all analyses [30].
