Study settings, period and design

The study was conducted among selected public hospitals and health centers of East Shawa Zone, Oromia Region. East Shawa Zone is one of the largest and most populous areas in the Oromia Region, having an estimated total population of 1,513,063 based on the 2007 population and housing census of Ethiopia [23, 24]. Among a population 764,097 were females, 334,841 were women in reproductive age groups (15–49 years), 52,503 were pregnant and 5250 pregnant women expected to be eligible for abortion services.

Under the Zonal Health Department there are 69 midwives, 16 clinical nurses and six gynecologists trained on CAC. There are 652 Health Extension Professionals working on awareness creation and providing primary health care services at the community level. Administratively, the zone is divided into one town and ten districts. There are five government hospitals and 59 health centers.

An institution-based cross-sectional study design was used to assess service availability, facility readiness, client satisfaction and associated factors. The study was conducted from March 1 to July 31, 2020.

Study participants

The level of service availability and facility readiness for providing CAC was investigated at the facility level. Hence, all 64 public hospitals and health centers in East Showa Zone were considered to be the source population while those randomly selected from among them were considered the study population. For assessment of client satisfaction and associated factors, all women who received CAC from public health facilities of East Shawa Zone were the source population, while those receiving CAC services at the selected facilities were considered the study population. Women who were unable to communicate due to medical and related problems and women referred to other health facilities were excluded.

Sample size determination

To incorporate an adequate sample of health facilities and women in the study, sample sizes were determined independently for each specific objective. Furthermore the required statistical assumptions were considered while adequate sample size was determined for each specific objective. For assessments of service availability and facility readiness, 50% of public health facilities in East Showa Zone were included. Accordingly, all five hospitals and 25 health centers were selected. For assessment of satisfaction with services, sample size was calculated using a single population proportion formula. Accordingly, since there was no multi-center study in similar study settings, the percentage of women satisfied with abortion services provided at public health facilities was considered to be 50%, and the desired level of confidence in estimating the level of satisfaction was 95%. Hence, the corresponding standardized value for 95% confidence was Z = 1.96, and the maximum tolerable margin of error in estimating the proportion was d = 0.05. Accordingly, the estimated sample size was 384 women. Since the sampling design was two-stage sampling, in order to compensate for the complexities of the design effect, the estimated sample size was multiplied by 1.5; the final sample was computed as 384*1.5 = 576. Furthermore, the study was designed to use multilevel modeling to undertake the assessment at facility and individual levels simultaneously. With these considerations, sample size was determined based on the recommendation of Kreft (1996) rule of thumb (25). In the type of study where group and subjects in the group are simultaneously investigated, the rule suggested that 30 women be taken per group if the overall number of groups is 30. Accordingly, 30 health facilities and 30 women from each health facility were sampled. Thus, 900 women were sampled to determine the level of satisfaction with CAC services and to identify individual- and facility-level factors associated with the level of satisfaction.

Sampling procedure

Two-stage sampling was used to select the sample of women who received abortion services at public health facilities of East Showa Zone. Health facilities were selected from all 11 districts in East Showa Zone using a stratified sampling technique. First, the public health facilities were stratified into hospitals and health centers. Then, all five hospitals in the zone (Adama Hospital Medical College, Bishoftu Referral Hospital, Olenchiti Hospital and Batu and Modjo hospitals) were included in the current study. There are 59 health centers in East Showa Zone. First, lists of all health centres providing abortion service in the 11 districts of the zone were prepared, and a sample of 25 health centres was selected proportionally from all districts, using simple random sampling technique. Since the attendance to service delivery units is assumed to be random, 30 women from each selected health center were incorporated into the study.

Variables of the study and operational definitionsDependent variable

Service availability, facility readiness and client satisfaction were considered as outcome variables in the current study and they were operationalized per the description given below.

Service availability Service availability was assessed through availability and functionality of the six basic SAC signal functions at the health center level and the additional four comprehensive SAC signal functions at the hospital level. The service was assumed to be available where at least 75% of the components were present and functional at the facility.

Facility readiness General service readiness among facilities was assessed at both health center and hospital levels by gauging the presence of six basic amenities and nine standard precautions for infection prevention. The six basic amenities are; availability of electric power during normal working hours, availability of improved water source within 500 m of health facility, availability of communication equipment, either phone or short-wave radio and room with auditory and visual privacy for patient consultations, accessibility of adequate sanitation facilities for clients and computer with e-mail or Internet service. The nine standard precautions for infection prevention are; availability of safe and protected final disposal of sharps, safe and protected final infectious wastes, appropriate storage of sharps waste, appropriate storage of infectious waste, disinfectant, single-use disposable or auto-disable syringes, soap and running water or alcohol-based hand rub, latex gloves, and availability of guidelines for standard precautions. Specific service readiness among facilities was assessed by the availability of staff trained in two major areas of the service (comprehensive abortion care and family planning) at the health center level and two additional major areas of the service(surgery and Anesthesia) at the hospital level. Specific service readiness was also assessed by the availability of five pieces of equipment at the health center level and one additional piece of equipment at the hospital level, the availability of four diagnostics at the health center level and one additional diagnostic facility at the hospital level, the availability of six materials for MVA at both health centers and hospitals, and the availability of 12 medicines at health centers and six additional medicines at hospitals. The facility was deemed ready to provide service where at least 75% of the components were present and functional.

Client satisfaction was assessed by 26 service components and measured using five Likert scale values based on women’s responses. The scales were labeled and coded as: 1. highly dissatisfied, 2. dissatisfied, 3. Neutral, 4. satisfied and 5. highly satisfied. Then, the satisfaction scores on 26 service components for each woman were added to create an overall score. Based on the overall scores, the level of client satisfaction with the service was operationalized into five scaled ordinal variables using the measure of relative standing. First the overall scores were arranged in ascending order and then their relative standing was calculated using percentile. Accordingly, women whose overall satisfaction scores were above 80th percentile were grouped as “very high” level of satisfaction. Similarly, women whose overall scores were between 61 and 80th percentile were grouped as “high”, between 41 and 60th percentile were grouped as “average”, between 21 and 40th percentile were grouped as “low” and below the 21th percentile were grouped as “very low” level of satisfaction.

Based on the data 20% of households earn below 1500birr on average per month and classified under low income group. Similarly, 25% of households earn above 2000birr on average per month and classified under high income group. The remaining 50% of households earn between 1500 and 2000 birr per month and classified as middle income group.

Independent variables

Socio-demographic and economic factors: place of residence, maternal age, maternal education, partner’s education, occupation, average monthly income, marital status, distance from health facility, access to transport to health facility.

Education status: for both women and partners, the educational status was classified into groups labeled as; no education, primary, secondary and higher education. Women or partners who didn’t attend any formal education were classified under the group labeled with “no education”, those who attended formal education from grade one to eight were grouped under “primary”, from grade nine to twelve were grouped under “secondary” and those who attended above grade twelve were classified under “higher education”.

Income: It was used to measure the household average monthly income in Ethiopian Birr from the respondents report. Based on the data, households average monthly income ranges from 0 to 20,000 Birr per month. Considering the contextual variability of income across different settings and time; we used measure of relative standing (quartile and percentile) for classification. Based on the data 25% of households earn below 1500birr on average per month and classified under low income group. Similarly, 25% of households earn above 2000birr on average per month and classified under high income group. The remaining 50% of households earn between 1500 and 2000 birr per month and classified as middle income group.

Obstetric factors: parity, gravidity, history of abortion, number of previous abortions, pregnancy intention, number of living children, family size, history of obstetric complications, gestational age, diagnosis type, type of procedure done, procedure outcome,

Diagnostic type of abortions: Diagnostic type can be either induced or spontaneous abortion. Spontaneous abortion can be completed without intervention or it may require interventions by surgical or medical means.

Types of procedures: the procedures by which the abortion is performed could be either medical or manual vacuum aspiration. The procedure was classified as medical abortion when the medications like mifepristone or misoprostol are used for either induced or spontaneous abortion based on the facilities protocol. However we classified the procedures as Manual Vacuum Aspiration (MVA) when Ipas MVA plus charged aspirator with easy grip cannula are used for surgical abortion of first trimester pregnancy.

Procedure outcome: the outcome of abortion procedures could be “with complication” or “without complication”. The outcome was considered as without complication when the process of abortion is completed without any complication otherwise considered as with complication when the process of abortion is followed by certain complications like, sepsis, anemia, uterine perforation, organ injuries.

Provider-related factors: age, provider’s profession, level of education, total years of experience, duration of experience in CAC service, provider’s sex, monthly income, marital status, family size, provider’s living conditions.

Provider’s monthly income: It was used to measure the provider’s average monthly income in Ethiopian Birr. Considering the contextual variability of income across different settings and time; we used measure of relative standing (median) for classification. Based on the data 50% of providers earn below 5342birr on average per month and classified under low income group. Similarly, 50% households earn above 5342 birr on average per month and classified under high income group.

Data collection tools and procedure

Data were collected using an interviewer-administered questionnaire and observational checklist. These tools were adapted from the safe abortion care model. Women were interviewed through a semi-structured questionnaire to assess their satisfaction with CAC services and associated factors. Observational checklists were used to assess service availability and facility readiness.

Data quality assurance

Data regarding service availability and facility readiness were collected by health professionals, particularly midwives who were at least first degree holders and trained in CAC services. The data for assessment of service satisfaction were collected by high school teachers in order to avoid the introduction of the potential bias of health care provider conducting interviews potential. The interviewer-administered questionnaires were translated from English to the locally spoken language. The internal validity of the tools, specifically the observational checklist and assessment tools for client satisfaction, was tested using Kappa statistics. To assess the practicability of data collection tools in the study settings, a pre-test was conducted before embarking on the main study. To ensure data quality, data collectors and supervisors were trained on use of the data collection tools and procedures as well as study purposes. The data collection activities were regularly supervised to check completeness and consistency. Corrections were made and feedback provided as needed.

Data processing and analysis

Data were coded and entered into a computer, then processed and analyzed using Stata-13. Before analysis, data processing tasks such as data cleaning, counting, categorizing and computing were performed. Then, descriptive analysis was performed to characterize facilities and to explore the level of service availability and facility readiness to provide CAC services. The women’s characteristics across all variables were explored using descriptive statistics. The level of client satisfaction was estimated using a 95% CI. In order to identify and measure the effects of both facility’s and woman’s characteristics on client satisfaction, a multilevel approach using a two-level mixed effects ordinal logistic regression model was used. In using this model, the effects of explanatory variables on the level of client satisfaction were analyzed simultaneously at two levels, those of the woman and those of the facilities. Four models were fitted to estimate both the independent and combined effects of individual and facility factors and random effect of between facilities variations.

Model-I:

was a null model with no covariates;

Model-II:

included only individual-level factors;

Model-III:

included only facility-level factors; and

Model-IV:

was a combined model that included both individual and facility factors.

The null model (Model-I) represents random intercept and was fitted without any predictor so as to examine the random effect of variability between facilities. The random effect was described by the intra-class correlation coefficient (ICC), which was calculated using between and within group variance. The existence of a nonzero ICC in the model was considered to select multilevel model analysis technique over single-level regression. Proportional change in variance (PCV) was also calculated for successive models to see the contribution of variables at individual and facility levels to explaining the level of client satisfaction in reference to the null model. Finally, the net effects of both individual- and facility-level factors on client satisfaction were estimated by the combined model (Model-IV). In this model both the net fixed and random effects are revealed simultaneously. The effects of individual- and facility-level factors were estimated using AORs by controlling for the effects of all remaining variables in the combined model. The magnitude of association was estimated AORs with 95% CI.