Comparison of Health Service Utilization and Determinants between Insured Women and Uninsured Women in the Sidama Region, Southern Ethiopia: A Multilevel Analysis

Research Article

Austin J Public Health Epidemiol. 2025; 12(1): 1174.

Comparison of Health Service Utilization and Determinants between Insured Women and Uninsured Women in the Sidama Region, Southern Ethiopia: A Multilevel Analysis

Debessa KC¹*, Negeri KG¹, Dangiso MH²

1School of Public Health, College of Medicine and Health Sciences, Hawassa University, Hawassa, Ethiopia

2Ethiopia Public Health Institute, Addis Ababa, Ethiopia

*Corresponding author: Kare Chawicha Debessa, School of Public Health, College of Medicine and Health Sciences, Hawassa University, Hawassa, Ethiopia Tel: +251900505922; Email: kare.debessa@gmail.com

Received: February 10, 2025; Accepted: February 28, 2025; Published: March 03, 2025

Abstract

Background: In Ethiopia, the community-based health insurance (CBHI) initiative was established to address financial risk protection and enhance healthcare accessibility, particularly for vulnerable populations like women.

Aim: This study compared healthcare utilization and determinants between insured women and uninsured women in the Central Zone of Sidama Region, Ethiopia.

Methods: From January 19th to February 20th, 2024, a community-based comparative cross-sectional design was employed to collect data from 1,280 women (640 CBHI members and 640 non-members) utilizing the multistage sampling technique. Data collection involved structured face-to-face interviews conducted using the KOBO Tool, focusing on healthcare utilization and related factors.

Findings: The analysis result revealed that CBHI membership was associated with a 77% higher frequency of health facility visits compared to non-members (APR = 1.77, 95% CI: 1.52 to 2.06). Key factors associated with health facility visits among CBHI members were age (APR = 1.01, 95% CI: 1.01- 1.02), woman’s educational level (APR = 1.3, 95% CI: 1.09-1.54), rural area of residence (APR = 0.73, 95% CI: 0.58-0.93), increased satisfaction with health services (APR = 1.04, 95% CI: 1.03-1.04), longer waiting times (APR = 0.68, 95% CI: 0.55-0.84), higher community-level literacy (APR = 1.75, 95% CI: 1.45- 2.12), and lower community-level poverty (APR = 1.33, 95% CI: 1.08-1.64) were associated with frequent healthcare visits.

Whereas for non-CBHI members women age (APR = 1.02, 95% CI: 1.01– 1.03), rural area (APR = 0.63, 95% CI: 0.46–0.87), and lower community-level poverty (APR = 2.36, 95% CI: 1.64–3.39) and waiting times (APR = 0.39, 95% CI: 0.30–0.49) were associated with frequency health facility visits.

Conclusions: The study’s findings provide valuable insights. Addressing challenges related to rural areas, waiting times, educational levels, communitylevel literacy, and poverty could enhance healthcare access and utilization, ultimately leading to improved health outcomes, especially among women.

Keywords: Community-based health Insurance; Healthcare; Utilization; Determinants; Negative binomial; Multilevel; Women; Ethiopia

Background

Access to healthcare is a basic human right important for human well-being. This right is embedded in various multilateral treaties, including the Universal Declaration of Human Rights and the International Covenant on Economic, Social, and Cultural Rights [1,2]. The right to healthcare includes essential components like availability, accessibility, acceptability, and quality, all vital for ensuring universal health coverage [2]. Availing quality health services should be safe, effective, people-centered, timely, equitable, integrated, and efficient [3].

Despite international efforts to improve healthcare access, many developing nations face barriers to achieving the goal of universal health coverage. The barriers include insufficient health infrastructure, shortage of trained workforce, and financial constraints [4]. Inadequate healthcare infrastructure includes a lack of well-equipped healthcare facilities and essential medicines, which hinders access to basic healthcare services. Additionally, a shortage of trained workforce amplified particularly in rural and underserved areas, further, this could exacerbate the problem [5].

Additionally, literature has identified several determinants of healthcare service utilization. These include socio-demographic factors such as gender, area of residence, marital status, literacy level, occupational status, family size, presence of children under five and elders in the household [6]. Likewise, economic factors, such as the average monthly income of households, also play a significant role in healthcare utilization [7].

Besides, healthcare access-related factors, such as the nearest health institution, time taken to reach health institution, presence of road for transportation, and availability of ambulance services are also important predictors of healthcare utilization [8]. Moreover, health perception and healthcare need-related factors, such as the presence of chronic illness in the household and attitude towards the CBHI scheme, have been identified as significant determinants that could affect healthcare utilization in a community [9].

On the other hand, financial constraints can also play a significant barrier to healthcare access and utilization. In many low-income countries, out-of-pocket payments often lead to financial burdens, affecting the need for sustainable healthcare financing mechanisms [10]. High costs of healthcare services, including consultation fees, diagnostic tests, and drugs can inhibit people from seeking timely health attention, leading to delayed diagnosis and care for health conditions [11].

The Sustainable Development Goals, particularly Goal 3, aimed to ensure healthy lives and promote well-being for all, targeting universal health coverage by 2030 [12]. As nations strive to expand healthcare coverage, they face structural challenges that impede progress. Community-based solutions, such as community-based health insurance (CBHI), have emerged as an alternative option to improve healthcare accessibility and affordability among the poorer segment of the population [13]. Hence, the CBHI scheme aimed to pool and generate financial resources to share risks among community members, consequently reducing the financial burden of health costs and improving access to essential health services in targeted communities [14].

Studies have revealed that CBHI membership can improve health service utilization. For instance, a study conducted in Ethiopia depicted that households enrolled in CBHI increased their health service utilization by 6.9 percentage points compared to non-CBHI members [15]. Another study done in the East Wallaga Zone of Oromia region, Ethiopia, showed that 60.5% of insured households used health services in the previous six months, compared to 45.9% of non-insured households [16].

On the other hand, successful implementation of CBHIs requires tackling challenges such as low enrollment rates, improving healthcare delivery, and ensuring its long-term sustainability [17]. Factors like lack of awareness about the benefits of CBHI, loss of trust in the scheme's management, and unaffordability of premiums can lead to low enrollment rates [14]. Therefore, ensuring the long-term viability of the CBHI program requires sustainable funding sources, an efficient management system, members' trust, and continuous monitoring and evaluation of the program to overcome emerging challenges [17].

For example, research in Ethiopia has identified the influence of socio-demographic and economic factors on CBHI enrollment and healthcare utilization, emphasizing the need to address the CBHI scheme to better serve the needs of women in the Sidama region [18]. Women often face unique barriers in seeking healthcare services, such as limited decision-making power, financial dependence, and constraints, as well as cultural norms that prioritize men's healthcare needs [19]. Understanding these specific barriers faced by women in the Sidama region is vital to designing targeted measures so as to improve access to healthcare.

In addition, in the southern part of Ethiopia, a substantial knowledge gap exists in the current research on healthcare utilization among community-based health insurance (CBHI) members. This gap is particularly pronounced in the methodologies used, as most studies rely on conventional regression models that fail to capture variations across different analytical levels, such as individual versus community levels [20,21]. This limitation hinders our understanding. of how factors at various levels interact to influence healthcare utilization among CBHI members and restrict the identification of detailed relationships between variables.

Another critical knowledge gap concerns the perspectives of women within households. Existing studies have primarily focused on the household level, inadvertently marginalizing women's experiences and views related to gender dynamics in healthcare utilization among CBHI members [22]. This oversight has resulted in a lack of information about how gender-specific factors affect healthcare utilization patterns among women of the CBHI members [23].

These gaps underscored the need for future research that incorporates sophisticated methodologies and perspectives, especially those related to gender differences and their impact on healthcare access and utilization behaviors. Therefore, this study investigated healthcare utilization and determinants between insured & uninsured women in the Sidama region, Ethiopia.

By addressing the knowledge gap on health service utilization determinants among CBHI members and non-member women in the Sidama region, studies aimed to contribute to universal health coverage and the Sustainable Development Goals [24]. Further, the findings will guide the development of targeted interventions to enhance healthcare access and quality for women in the Sidama region, ultimately leading to improved health outcomes for women and their families. Policymakers and program managers can also use the findings to design effective interventions that address the unique needs and barriers faced by women in accessing healthcare services, thereby contributing to the overall well-being of the population in the Sidama region.

Methods

Study area

The study was conducted in the Central Zone of Sidama region, Ethiopia. The Central Sidama Zone consists of six districts and one town administration, with a total population of 956,967 (2016 EFY) [25]. The study sites, Dale Woreda and Yirgalem City administration are located approximately 45 km south of the regional capital, Hawassa, and 320 of Addis Ababa, the capital of Ethiopia [26].

The community-based health insurance (CBHI) program was originally introduced in Ethiopia in 2011 at 13 sites, with Yirgalem City being one of them [27]. Yirgalem Hospital is known for being the home in addition to the Arbaminch Hospital where an innovative healthcare financing program was initiated as well it was one the first modern health institutions established in the region [28]. Determining the factors influencing healthcare utilization among CBHI members is made possible by the decision to assess the program in this particular area, with a focus on women. As so, this study offers important perspectives and insightful lessons for future work.

Study design and period

A community-based comparative cross-sectional study design was employed between January 19th and February 20th, 2024.

Source and study population

The study population was women aged 18 years and older residing in Dale Woreda and Yirgalem city administration in Sidama region, Ethiopia, both CBHI members and non-members. For the exposed group, women 18 years and older who are enrolled in CBHI were selected from kebeles within Dale Woreda and Yirgalem city administration. Meanwhile, for the unexposed group, women 18 years and older who were not enrolled in CBHI were selected from different households but from the same kebeles of Dale Woreda and Yirgalem City administration.

Inclusion and exclusion criteria

Participants' eligibility criteria include women aged 18 years and older who were both enrolled in CBHI and not enrolled. In households with polygamous marriages or where the mother was deceased and the father resides with a daughter aged 18+, the interview respondent was decided by the husband (household head). Households that have begun contributing premiums to CBHI but are not yet able to utilize health services were also considered non-members.

Sample size determination

The sample size was obtained using OpenEpi, Version 3.05.07, by considering a study conducted on "Community-based health insurance service utilization and associated factors in Addis Ababa, Ethiopia" [29]. The sample size calculation for this study was based on the following assumptions of the two population proportions; these were a 95% confidence interval, 80% power, a 1:1 ratio of exposed to unexposed, an outcome proportion of 67.35% among the unexposed group, and 55.58% among the exposed group, a design effect of 2, and a 10% non-response rate. Considering these assumptions, the maximum required sample size obtained for this particular study was 1,280 study participants (women), consisting of 640 CBHI members and 640 non-members.

Sampling procedure

A multi-stage sampling technique was applied to obtain the desired sample size. The estimated sample was proportionally allocated to Dale Woreda and Yirgalem city administration. Then, a total of 14 kebeles were drawn from Dale woreda (one urban and eight rural kebeles) and Yirgalem city administration (two urban and three rural) using a simple random technique. The total number of non-member households from each kebele was either obtained from the respective kebele or recorded in a separate sheet as non-members to CBHI in case of list absence. There was a total of 8,646 non-CBHI member households in the 14 kebeles at the time of data collection.

During the second stage, from the total number of 7,472 CBHI member households, 640 CBHI members were randomly selected from the CBHI registry. Similarly, 640 non-member households were randomly selected from each kebele until these number was obtained from the list. Data collectors contacted study participants (women) for the final interview.

When a woman was absent at home during data collection, data collectors made a maximum of three visits to the household before dropping the woman from the interview. When selected households lacked study participants drawn by simple random technique, women from the next household were included. If the selected household has an eligible participant, the interview continued.

Study variables

The outcome variable of interest is the frequency of health services utilization among women who were either CBHI members or nonmembers. Health services utilization was measured as the number of outpatient and inpatient healthcare visits made in the previous twelve months to a health facility. This was reported by the study participants at the time of the interview. Outpatient visits include trips to a health post, health center, clinics, or hospital for health care, while inpatient visits refer to overnight health center or hospital stays for healthcare and medical treatment [30].

The independent individuals and community-level variables were age, marital status, family size, religion, area of residence, household head, level of education, membership status, wealth index, distance to healthcare facility, decision-making at household, waiting times at healthcare facility, satisfaction level, community level women autonomy, community level women literacy, and community level women poverty.

Data source

The data sources for this study were women aged 18 years and above, households of CBHI members, and non-members at the time of data collection from the study areas. The details of the study variables measurement are provided in Supplementary File 1.

Data collection tool and procedure

The data collection tool was adapted from previous studies conducted elsewhere [16,20,31-41] and the questionnaire was provided as Supplementray File 2. In addition, before the data collection, the study team conducted a pilot test. From the pilot test, we identified and rectified some coding errors, labeling errors, and enhanced certain questions and responses.

Additionally, we incorporated the revisions, and feedback from supervisors, data collectors, and study participants to improve the quality of the study tools. Twenty-eight data collectors, who held bachelor's degrees, were trained and conducted face-to-face interviews using the ODK mobile application. Furthermore, the data collection process was supervised by five experienced supervisors, all of whom held master's degrees in public health. The collected data were then exported to Stata version 17 for further analysis.

To ensure data quality the study team (principal investigator, supervisors, and data collectors) implemented several quality control measures. These measures were training and pre-testing of data collectors, re-interviews, and daily data checks to identify and correct any errors, such as issues with labeling, incomplete answers, or formatting of some questions. Consequently, these steps helped to maintain and improve the quality of this study work.

Statistical methods

Before conducting data analysis, variable recoding, computations, and categorizations were performed. Consequently, for categorical variables, summary measures were expressed as absolute frequencies and percentages. In contrast, for continuous variables, the mean with standard deviation (SD) was used as a descriptive measure [42].

The wealth index was computed using principal component analysis (PCA) as a combined indicator of living standards. It was based on 42 questions related to ownership of selected household assets like house ownership, construction materials, number of rooms, agricultural land size, presence of livestock, cooking fuel types, and possession of improved sanitation and water facilities [43].

To estimate the adjusted prevalence ratios with 95% confidence intervals (CIs) for the associations with the outcome of interest, a multilevel negative binomial model was used. This model was preferred over multilevel poison regression as there was overdispersion observed with the data where variance was greater than the mean [44].

Before conducting a multilevel analysis, the need was examined by employing a random intercept model of a multilevel-negative binomial model. This model also generated the Intraclass Correlation Coefficient (ICC) by ensuring the necessity of a multilevel model. As the ICC value exceeded 5%, the multilevel analysis model became essential [45].

Additionally, in the multivariate analysis model, variables with p-values < 0.25 from bivariable analysis were included, along with variables supported by literature, to account for potential confounding [46]. Moreover, effect modification was evaluated by sequentially introducing interaction terms. Similarly, multicollinearity among independent variables was checked using multiple linear regression with a variance inflation factor threshold of < 5 [47].

To account for the hierarchical nature of the data and reduce potential standard error underestimation using the ordinary models, based on the preliminary analysis, a multilevel model was recommended [48]. Furthermore, to evaluate the fitness of a multilevel model, four models were assessed and evaluated.

These were; Model 0 (empty model), Model 1 (with only individual-level variables), Model 2 (with only community-level variables), and Model 3 (with both individual- and communitylevel variables). The summary of the models fitted in this study was described below (Tables 5 & 6). Finally, a statistically significant association was determined using adjusted prevalence ratios (APRs) with 95% CI with P < 0.05 between independent and dependent variables.

Results

The data in Table 1 describes the demographic and socioeconomic characteristics of the study respondents. It showed that 775 (60.5%) resided rurally, while 505 (39.5%) lived in urban areas. Additionally, 1143 (89.3%) identified as Protestant, with 137 (10.7%) following other religious categories. Ethnically, 1221 (95.4%) were Sidama, and 59 (4.6%) belonged to other ethnic categories. At the individual level, 726 (56.7%) had attended formal education, while 554 (45.2%) did not. Furthermore, 710 (54.8%) were autonomous women, and 579 (45.2%) were not (Table 1).