This was a multi-country study to assess the preparedness of public health facilities to deliver PAC services in Burkina Faso, Kenya and Nigeria. The three countries offer both similar and dissimilar contexts for investigating the quality of PAC. For instance, abortion is largely restricted across the three countries, and they all report high incidences of unsafe abortion [8,10,11]. These settings offer worthy contexts to examine the preparedness of their health facilities to provide PAC services.
Study design and population
A cross-sectional survey was conducted among a representative sample of primary, secondary and tertiary health facilities in the aforementioned countries. Health system across the three countries is organized according to hierarchical levels. Health facility levels are generally categorized as primary, secondary and tertiary-levels. Primary health facilities are the first point of contact for the majority of community members’ health needs, and include community facilities, dispensaries and clinics. In Kenya, primary-level facilities handle the Kenya Essential Package for Health (KEPH), which encompass activities related to health promotion, preventive care, and curative services. Secondary facilities are mainly sub-regional and regional and serve as referral facilities for the primary-level facilities. They undertake curative and rehabilitative care and address a limited extent of preventive care and health promotion. Tertiary facilities are mainly national referral and teaching hospitals. All health facilities capable of conducting normal deliveries were included in our sample frame. Data was collected in facilities over a 30-day period between November 2018 and February 2019.
Sampling and recruitment
A two stage stratified sampling procedure was used in each country, that is, a) the highest sub-national administrative units (i.e. counties in Kenya, states in Nigeria and regions in Burkina Faso), and b) the levels of health facilities. The sub-national levels represented by “counties”, “states” and “regions”, denote the geopolitical zone below national and above district levels. At the first stage, in each country, a random sample of six regions, counties or states was drawn, and excluding the administrative unit hosting the national capital regions – i.e. Centre in Burkina, Nairobi in Kenya, and Abuja – Federal Capital Territory (FCT) in Nigeria. Thereafter, the capital regions were added to the regions purposely to make seven regions/counties/states in each country.
The selected administrative units included, Burkina Faso (seven regions from the 13: Boucle du Mouhoun, Cascades, Centre, Centre-Ouest, Centre-sud, Haut-Bassins, and Nord); Kenya (seven counties from 47: Garissa, Kajiado, Kiambu, Laikipia, Mandera, Migori, and Nairobi); Nigeria (seven states from 36: namely Anambra, Bauchi, Cross-River, Edo, Federal Capital Territory, Kano and Kogi.
At the second stage, the researchers obtained from government records updated master lists of all public health facilities in the different sub-national units. Burkina Faso and Nigeria’s list were updated up to July 2018 while Kenya was updated in February 2018. A requisite sample of facilities in each country was determined using a formula for known populations and known proportion estimates by: ∆=z√ ((p (1-p))/n).
In all cases, the known estimate p represented the proportion of facilities capable of providing PAC contraceptive counseling, which was the lowest measure for quality of PAC in Kenya (19.4%) and Nigeria (16%) [22,37]. Because we did not find any recent estimate in Burkina Faso, we used the 50% proxy in order to generate the maximum sample size possible. These calculations yielded the number of facilities required for each country, and upon accounting for a response rate of approximately 93%, the estimated sample size of facilities was determined as follows: 414 in Burkina Faso, 259 in Kenya, and 223 in Nigeria.
The total sample size of health facilities was allocated to each of the seven sub-national units in each country depending on the population of eligible facilities in a specific region/county or state. Therefore, health facilities were randomly selected within each sub-national level with all tertiary health facilities included and a sample of primary and secondary health facilities. Eligible facilities were those that could provide normal delivery services, were publicly owned (government owned) and operational at the time of survey. As such, we excluded some specialized facilities including mental and spinal hospitals as well as military and prison hospitals known not to offer services to the public. Our focus on public health facilities is because government investments in health services primarily go to these facilities. During the survey, some facilities were dropped and replaced with similar facilities within the same locality, due to insecurity and travel inaccessibility. In addition, sampled facilities that declined to participate in the study were replaced with similar facilities from the sampling frame, which had been identified a priori.
Trained field workers visited each eligible facility and administered the signal functions questionnaire which had been adopted from previous versions [38,39]. The questionnaire was further refined to the contexts following extensive discussions with experienced obstetricians and gynecologists in each country. The questionnaires captured details on availability of key equipment, supplies and commodities, staffing and staff training, facility operation hours and ability to perform various sexual and reproductive health services. Uniform tools were used across all countries. However, some aspects were adapted to fit in national standards (e.g. facilities categorizations). We asked the providers whether they were currently providing the listed services (Figure 1). Whenever the provider indicated that a particular service was currently unavailable, the next sets of questions probed for the reasons why the service is not available. In response, the providers listed all possible reasons why the service was unavailable at that time. The tools were pre-tested to enhance conceptual clarity and logical flow. At large referral hospitals, respondents were the head of the obstetrics and gynecology department, or a key obstetrician gynecologist working in the facility. However, at lower level facilities, a nurse, a midwife or another health worker who was knowledgeable on PAC services provided in the facility was interviewed. The quantitative data were collected using tablets and hosted on the SurveyCTO platform. Completed and verified data were uploaded unto the APHRC cloud server for safe storage. Spot-checks were performed on 5% of the sample by the lead for each country.
Using the Ministry of Health master list of health facilities in Burkina Faso, Kenya and Nigeria, and the sampling frame of public facilities, we constructed facility-levels weights accounting for the sample design and adjusting for stratification by regions/counties/states, and facility non-response, as well as applying a finite population correction. The statistical analysis was conducted in Stata version 15.0 . We therefore use weighted data to describe the capacity of health facilities to deliver PAC services. We drew from the Owolabi et al (2019) approach and constructed composite or aggregate indicators of signal functions to provide basic and comprehensive PAC using a signal functions approach . By calculating the availability of specific health interventions that are key to PAC—i.e., the signal functions—we measure the capacity for, and quality of, PAC from a health systems perspective. We do this by summating or combining sets of indicators that constitute the two delineated levels of care - basic and comprehensive PAC, that roughly correspond to care that should be provided at both the primary level and at the referral level hospitals respectively (Box 1). A key departure from the Owolabi approach is that under basic PAC indicators, we excluded the ability to communicate with referral facilities. This was mainly because this variable was not captured in our data collection tool and was proxied with having an established referral pathway between different health facilities. We also explored another level of analysis, again adopted from the Owolabi paper, which included developing case scenarios by excluding some PAC signal functions to have a less restrictive criterion at various stages. At first, we analyzed all PAC signal functions for each facility levels. Secondly, we excluded the availability of staff capable of conducting normal deliveries, thirdly, we excluded - staff with delivery capabilities; having a referral capacity; availability of short and long-acting, or permanent family planning methods. At the fourth stage, we examined PAC capability by excluding the ability of a facility to conduct referrals (through having a vehicle fueled). “Capacity” or “preparedness” was conceptualized as the ability of health facilities to deliver services based on signal function indicators . Proportions of facilities capable of delivering basic and comprehensive PAC were generated.