Postabortion care (PAC) is an essential component of emergency obstetric care (EmOC) and is necessary to prevent unsafe abortion-related maternal mortality, but we know little regarding the preparedness of facilities to provide PAC services, the distribution of these services and disparities in their accessibility in low-resource settings. To address this knowledge gap, this study aims to describe PAC service availability, evaluate PAC readiness and measure inequities in access to PAC services in seven states of Nigeria and nationally in Côte d’Ivoire. We used survey data from reproductive-age women and the health facilities that serve the areas where they live. We linked facility readiness information, including PAC-specific signal functions, to female data using geospatial information. Findings revealed less than half of facilities provide basic PAC services in Nigeria (48.4%) but greater PAC availability in Côte d’Ivoire (70.5%). Only 33.5% and 36.9% of facilities with the capacity to provide basic PAC and only 23.9% and 37.5% of facilities with the capacity to provide comprehensive PAC had all the corresponding signal functions in Nigeria and Côte d’Ivoire, respectively. With regard to access, while ∼8 out of 10 women of reproductive age in Nigeria (81.3%) and Côte d’Ivoire (79.9%) lived within 10 km of a facility providing any PAC services, significantly lower levels of the population lived <10 km from a facility with all basic or comprehensive PAC signal functions, and we observed significant inequities in access for poor, rural and less educated women. Addressing facilities' service readiness will improve the quality of PAC provided and ensure postabortion complications can be treated in a timely and effective manner, while expanding the availability of services to additional primary-level facilities would increase access – both of which could help to reduce avoidable abortion-related maternal morbidity and mortality and associated inequities.
Data for this study come from Performance Monitoring and Accountability 2020 (PMA2020), Nigeria and Côte d’Ivoire. (PMA2020) uses mobile-assisted technology to implement low-cost, rapid turnaround national/regional family planning monitoring surveys annually (Performance Monitoring for Action (PMA) 2021; Zimmerman et al., 2017). In each country, a cadre of sentinel resident interviewers collect data at the household, individual and facility levels. The data we used for this study are cross-sectional and include surveys of service delivery points (SDPs) that serve a nationally representative population of reproductive-age women (15–49 years). In-country ethical review boards and the Johns Hopkins Bloomberg School of Public Health provided ethical approval for this study. The sampling for the female survey in both countries employed a multi-stage cluster design with probability proportional to size (PPS) sampling of enumeration areas (EAs) within urban/rural geography strata. In Nigeria, PMA2020 was originally implemented in two states, Lagos and Kaduna. As demand for nationally representative female data increased, PMA2020 investigators selected an additional five states using PPS. Overall, two states were selected from the North West where 25% of the country’s population lives and one state from each of five remaining six geopolitical zones. EAs were selected from urban/rural-state strata. In both Nigeria and Cote d’Ivoire, EAs are geographic units comprised of ∼200 households and are defined by the central statistical or census office of the country. A sample of 35 households from each EA (40 for EAs in Lagos state, Nigeria) was randomly selected, and all women of reproductive age (15–49 years) from the selected households were invited to participate. Women provided verbal informed consent prior to participating in both countries. In Nigeria and Côte d’Ivoire, the household response rates were 97.5% and 97.6%, respectively, and the female response rates were both 98.1%. At the same time, a sample of SDPs was created that included private and public SDPs serving the selected EAs. The sample was selected from a list of public sector facilities serving the geography, which we obtained from the local health authorities, and a list of all private SDPs within each EA, which interviewers created through mapping and listing. Up to three private SDPs were randomly selected per EA as well as up to three public facilities assigned to serve those EAs and representing each level of care (primary, secondary and tertiary). On average, each EA is served by less than one private SDP, and two to three public SDPs are designated as primary, secondary or tertiary levels of care for that area. The final SDP sample (excluding pharmacies and chemists) involved facilities that could potentially provide at least basic PAC. The SDP response rates in Nigeria and Côte d’Ivoire were 96.6% and 97.0%, respectively. We linked the SDP sample to the female data using geospatial data. For each woman, we used the central point of her EA as her Global Positioning System (GPS) point as we did not have ethical review board approval to use individual women’s GPS data in Nigeria. As such, when we present findings on access, we are presenting the percent of women who live in an EA in which the midpoint is within 10 km of a given facility. For simplicity, in the ‘Results’ section of this paper, we refer to these findings as ‘the percent of women living within 10 km of a given facility’. For each SDP, we used the GPS point taken at the time of the interview. Neither the EA GPS points nor the facility GPS points were displaced. We then linked each woman to each SDP sampled using Euclidean distance and determined the distance to the closest sampled facility that provided any PAC, provided PAC and had all basic PAC signal functions, and provided PAC and had all comprehensive PAC signal functions. In total, 11 106 and 2738 women completed the survey in Nigeria and Côte d’Ivoire, respectively. One urban EA in Nigeria had inaccurate GPS data; thus, we excluded the 24 women from this EA, resulting in an analytic sample of 11 082 women. The 24 excluded women were slightly older, more educated and more likely to have average wealth than women nationally. All women in Côte d’Ivoire had GPS data. To distinguish facility type, we divided facilities by level (referral vs primary) and sector (public vs private). In Nigeria, these categories corresponded to public referral (teaching hospitals, state hospitals and higher-level maternity centres), public primary (all lower-level facilities), private referral (tertiary hospitals and secondary hospitals) and private primary. Nigeria has a three-tier health system in the public sector; thus, the public referral category encompasses tertiary (or specialist) facilities managed by the federal government and secondary facilities managed by state governments; all public primary facilities are managed by local governments. In Côte d’Ivoire, we grouped facilities into public referral (teaching, regional and general hospitals), public primary (all lower-level facilities) and private primary (there were no referral-level private facilities in our Côte d’Ivoire sample). These categories were determined in conjunction with in-country partners based on the local healthcare system. Public and private referral facilities in Nigeria and public referral facilities in Côte d’Ivoire should have the capacity to provide comprehensive PAC; all primary facilities should have the capacity to provide basic PAC. Some primary facilities may potentially have the capacity to provide comprehensive PAC if a trained provider is present, but since this is rare, we assumed primary-level facilities would not be expected to provide this service. The one exception to excluding primary facilities from analyses of comprehensive PAC was in the context of assessing individual signal functions. We present the estimates of individual comprehensive PAC signal functions at primary facilities to show the extent to which these services are available even at some lower-level facilities, offering a more complete assessment of PAC readiness. The SDP survey covered structural features of the facility, provider information, family planning service availability, stockouts and patient caseloads. Specific to this study, we included an additional module on abortion and postabortion services. We assessed facility abortion service readiness by measuring signal functions necessary to provide basic and comprehensive PAC services (Table 1). We created an index that combined PAC signal function information for each level of care (basic and comprehensive). The index is additive, providing a more nuanced measure of basic and comprehensive PAC readiness than a simple ‘all or nothing’ measure. We refer to this index as a readiness score, which we converted into a percentage that ranged from 0 to 100, representing the percent of signal functions a given facility has. PAC caseloads were reported by the person at the facility most knowledgeable about PAC and abortion service delivery. These respondents provided separate estimates of the number of inpatient and outpatient PAC clients treated in the last completed month and the average month. We averaged these two numbers to account for potential seasonality and multiplied by 12 to get annual PAC caseload estimates for each facility. Basic and comprehensive PAC signal functions criteria Using the public SDP data and a sampling frame of public facilities provided by the government, we constructed public facility weights for each country that are the inverse of the probability of selection of each facility type (within each state in Nigeria) multiplied by the response rate for that facility type/state stratum. In Côte d’Ivoire, we had a sampling frame for private facilities as well, which we used to construct weights in a similar fashion. Since there was not a private facility sampling frame for Nigeria, we multiplied the household sample EA probabilities of selection and the response rate of that facility type within that EA and took the inverse to construct the private facility weights. With the weighted data, we sought to produce service readiness estimates that reflect the facility type distribution (among seven states in Nigeria and nationally in Côte d’Ivoire). We conducted a number of analyses to assess PAC availability, readiness and accessibility in Nigeria and Côte d’Ivoire. To determine PAC availability, we used SDP respondent reports of whether the facility provided the service. We evaluated PAC readiness in a number of ways, including the percentage of all facilities with each basic and comprehensive signal function by facility type; the percentage of facilities with ‘all’ basic and comprehensive signal functions by facility characteristics (i.e. type, sector, as well as state in Nigeria) and the basic and comprehensive readiness score among PAC providing facilities by facility characteristics. We then estimated the percentage of PAC patients receiving care in facilities meeting basic and comprehensive readiness criteria, overall and by facility characteristics. Lastly, we determined accessibility by estimating the proportion of reproductive-age women living within a 10-km radius of a facility providing any PAC services, a facility with all basic PAC readiness criteria, as well as a facility with all comprehensive PAC readiness criteria, and explored potential sociodemographic characteristics associated with a lack of PAC accessibility within 10 km through bivariate and multivariate logistic regression. We applied survey weights that account for the complex sampling design and clustering—including the probability of selection and response rate—to the female data for the final analysis. We conducted all analyses in Stata version 15.1 (Statacorp, 2017).
N/A