Introduction

Access to medicines, especially antibiotics, is critical in the Universal Health Coverage (UHC) plan1. It is an essential element in the effort to achieve the Sustainable Development Goal (SDG)2, with SDG 3.8 aiming to ensure safe, effective, high-quality, and affordable medicines for all3. In this sense, access to working and quality-assured antibiotics is relevant for treating bacterial infections4. Nonetheless, inappropriate antibiotic access practices can impend their efficacy and lead to bacterial antimicrobial resistance, which has emerged as one of the biggest threats to human survival, contributing to up to 7.7 million estimated deaths worldwide in 20195,6. The highest death rate was evident for sub-Saharan Africa (SSA), with 230 death records per 100,000 population6. Bacterial infection, possibly involving resistant strains, is a leading cause of death in Ethiopia7,8. Furthermore, inappropriate antibiotic access practices can lead to other adverse consequences, such as side effects, toxicities, and risks of antibiotic resistance9.

Antibiotics are prescription-only medications that require prescriptions before access10. Despite this restriction, non-prescribed antibiotic access practices are widespread in resource-limited countries10,11,12,13. Access to antibiotics without prescription can surpass 93% in some nations11,13, where prescription-only dispensing restrictions are not strictly implemented14. People in these countries also face further challenges in obtaining antibiotics from official sources due to the scarcity of public facilities in rural and remote areas and a lack of sustainable financing and drug distribution systems15,16. In this sense, unlicensed drug providers are among the principal sources of non-prescribed access to antibiotics, accounting for more than half of all supplied medicines17,18,19. However, unrestricted access to antibiotics without prescription through unofficial routes is problematic since it may result in low-quality and ineffective drugs, also known as sub-standard and falsified medicines9, increasing the risk of resistance. Unofficial drug sellers may not limit obtaining sub-optimal antibiotic doses20, which are insufficient to treat bacterial infections and can trigger repeat antibiotic access21, perhaps resulting in an increased risk of resistance22. It requires a contextual knowledge of the extent and predictors of inappropriate antibiotic access practices at the community level in line with policy frameworks and healthcare systems across nations23.

In 2023, the World Health Organization (WHO) set a research agenda that included determining the levels and contextual predictors of inappropriate access to antibiotics at the community level24. Ethiopia also planned to strengthen the evidence through surveillance and study25. In this regard, previous studies conducted in Ethiopia26,27,28,29,30,31,32,33,34 focused on the antibiotic dispensing practices of the supply side only. Nonetheless, assessing the practices of antibiotic access on the demand (user) side is equally relevant. Previous research primarily focused on the adult population. However, the types and burdens of illnesses that drive antibiotic usage are likely to differ among population groups of all ages, with children under five years at the highest risk of antimicrobial resistance35. They did not also address potential supply sources for non-prescribed antibiotic access. The antibiotic access points and behaviors of urban and rural populations may also differ, and perhaps farmers are common antibiotic users for animal farms36,37. Hence, there is a shortage of contextual evidence on antibiotic access practices among all population ages at the community level in eastern Ethiopia. The community-based study on access practices helps understand the complex interplay of antibiotic behaviors between users and providers, addressing the supply and demand issues. It also helps to look into other actors in the antibiotic access transaction, such as unlicensed drug sellers. These assessments can help determine a tailored antibiotic stewardship approach that aids in preserving these medications. Therefore, this study aimed to assess the appropriateness of antibiotic access practices at the community level in eastern Ethiopia.

Methods

Study design and setting

We applied a community-based cross-sectional study design using household surveys. It draws from data collected for a research project on antibiotic access and usage practices at the community level using a combination of qualitative and quantitative studies. With this mixed design, we used pragmatism, a viewpoint that focuses on the practical implications of what works to answer the research question38.

The study was conducted in selected sub-districts of Haramaya and Harar town in eastern Ethiopia. Haramaya district is located in East Hararghe, Oromia region, while Harar town is in the Harari region. These study sites are part of the Haramaya University's urban and rural health research centers. Specifically, we chose these two sites because they had readily available demographics and relevant sampling frames. The catchment area of the Haramaya research site includes 12 rural sub-districts39, with approximately 20,580 active households enrolled in 2022. The Harar research site also has 12 urban sub-districts40, with around 12,876 active households registered in 2022.

Study participants

Participants were adult household representatives, preferably mothers. In our setting, mothers are the usual caregivers for sick household members. They are more likely to qualify as household representatives in survey studies because they meet most WHO criteria, including the highest knowledge of household members' health, healthcare costs, and healthcare utilization41. An adult member who had just received antibiotics also supported the household representative in answering antibiotic-specific questions, while caregivers provided support on behalf of the child user. We considered the support of adult household members who have recently used antibiotics to improve the accuracy of antibiotic-specific responses by household representatives. Because the study assesses antibiotic access practices over the last six months, a household representative and anyone else who had been away for six months before data collection were excluded. For this study, therefore, a total sample of 1263 was determined using a single population proportion formula, with the assumption the margin of error does not exceed one-tenth of the sample value of the proportion42, following Cochran's sampling theory43. However, since the study was part of a larger research project with additional questions comparing two proportions44, the final sample size was 2310 households. We allocated 900 households to the Harar site and 1410 households to the Haramaya site. This allocation was proportional to the number of active households available at each site.

We used a stratified two-stage cluster sampling procedure. First, 77 clusters were selected randomly. Second, we carried out systematic random samplings of 30 households for each chosen cluster. This sampling strategy involved a random choice of the initial direction (as indicated by tossing a bottle) from each cluster's center to list active households and select the first sample using a lottery method. Subsequent household sampling from each cluster occurred at regular intervals, using an assigned 'k' value based on the number of active households in each cluster. With this, we considered the systematic random sampling approach as a typical random sampling method as long as the starting point was random for each cluster and the interval for sampling households was determined before the actual data collection occurred45.

Data collection methods

The data were collected from July to September 2023 using a semi-structured questionnaire. The questionnaire was developed by considering and customizing variables of interest from the WHO's standardized household survey manual on the access and use of medicines41, the national wealth, demographic, and health survey questionnaire46, and reviews of related studies. The questionnaire was pilot-tested and adjusted for contexts. We then entered the final form into the KoBo toolbox (https://www.kobotoolbox.org/) and deployed it to mobile applications to collect data. We also developed a show card based on a preliminary antibiotic inventory assessment for the most commonly accessed antibiotics in the settings, which aimed to facilitate the medication access recall.

The data collectors were experienced pharmacists. They were well trained and prepared to consistently respond to all clarification requests from study participants about the question items before data collection. The training also enabled them to precisely follow all instructions in the protocol and field guide, which entailed up to three times repeat visits to closed houses.

The study measured antibiotic access practices, including access sources and prescription status, within just six months before the data collection time. Though shorter recall periods (e.g., 30 days or less) are advised for measuring medication behaviors47, we chose six months to assess antibiotic access practices because of time and resource constraints. We confirmed the actual access to antibiotics by visualizing the drug or its packaging48. However, household members who claimed to have taken an antibiotic but could not recall its name or discarded its package were recorded as non-users because we could not determine whether the accessed medicine was an antibiotic. We preferred the most recent access time in cases where household members had received multiple antibiotic courses during the last six months. Accordingly, fulfillment of a directive recommendation for antibiotic access, with prescription access from an official supplier, was considered appropriate49,50. Our approach to verifying non-prescribed access was the inquiry of where they had received their most recent antibiotic medicines. This evaluation also included whether they obtained antibiotics on their own.

Data analysis

Completed data were downloaded from the KoBo toolbox platform in Excel form and imported to Stata 14.0 for further cleaning and analysis. Descriptive analyses were used to summarize sociodemographic and household data, together with other pertinent variables. The outcome of this analysis was inappropriate antibiotic access practice and fitted with a modified Poisson regression with robust standard error51. We also used multinomial regression to identify specific factors associated with each inappropriate antibiotic access practice category (i.e., non-prescribed, unlicensed, and informal sources). We considered these inappropriate antibiotic practice categories because antibiotics can be obtained without a prescription from pharmacies in our setting, with additional options including unlicensed providers (e.g., primary clinics) and informal sellers (e.g., kiosks), which may provide antibiotics in remote areas. We then performed a Mediation analysis using Generalized Structural Equation Modeling (GSEM) to look at the direct and indirect relationships between the identified factors and each of the inappropriate antibiotic access practices while also adjusting for multilevel variances52. Factors of significant association with each outcome domain were chosen to build the GSEM. We began the analysis by hypothesizing that the usual drug supplier attended (UDSA) would mediate the appropriateness of antibiotic access practice53. A community's preferred drug supplier can be formal or informal, based on the available providers in the area. In this study, antibiotic access was considered appropriate when accessed through official routes and prescriptions from licensed practitioners. Furthermore, receiving sub-optimal antibiotic doses (i.e., lower doses or shorter course duration than required) can trigger repeat access21, with personal, community, and institutional factors known to influence drug needs, including antibiotics54. The UDSA was classified as inappropriate (for an unlicensed drug seller), potentially appropriate (for a pharmacy licensed to dispense over-the-counter drugs directly or a primary clinic licensed to prescribe emergency medicines only), and appropriate (for a private or public facility authorized to prescribe and dispense drugs)55,56 (Fig. 1).

Figure 1
figure 1

Hypothesized model of relationships between factors and antibiotic access practice appropriateness among study participants, eastern Ethiopia, 2023.

Full size image

The final GSEM model was selected based on the statistical significance of path coefficients, theoretical meaningfulness of the relationships, and minimum information criterion. The path coefficients indicated the direction in which each predictor or mediating variable influenced the outcome variable. Statistically significant effects were considered at a 95% confidence (95% CI) level, excluding zero.

Ethics declarations

The research was approved (with a reference number of IHRERC/007/2023) by the Institutional Health Research Ethics Review Committee of the Haramaya University at which the authors work. All methods of the study were performed in accordance with relevant guidelines and regulations. Before the data collection began, we also obtained informed consent from study participants.

Results

Participant characteristics

Of the 2310 houses sampled, 2256 (97.6%) household representatives responded. The average age of participants was 35.8 years. The majority of the participants were rural residents (61.2%), females (83.8%), married (79.9%), Muslim (75.9%), and Oromo (75.3%). Nearly half of the participants had no formal education (45.5%) and were housewives (46.7%). Most households had community-based health insurance (CBHI) (75.2%) (Table 1).

Table 1 Sociodemographic and household characteristics among study participants from selected sub-districts of Harar and Haramaya, eastern Ethiopia, 2023.
Full size table

Appropriateness of antibiotic access practices

Of 1245 individuals who received antibiotics, the majority were females (61.0%), under-five children (30.3%) and young adults (29.8%). Though official drug suppliers (i.e., hospitals, health centers, pharmacies, or drugstores) provided 69.0% of recent antibiotic access, primary clinics (22.2%) and kiosks (8.0%) also played significant roles. Antibiotics were obtained inappropriately in 45.6% of access practices, with 15.1% not prescribed, 21.8% unlicensed providers, and 8.7% informal sellers (Table 2 and Supplementary Table S1).

Table 2 Antibiotic access practices among study participants from selected sub-districts of Harar and Haramaya, eastern Ethiopia, 2023.
Full size table

Predictors of inappropriate antibiotic access practices

Participants who attended authorized or registered drug suppliers were 40% less likely to access antibiotics inappropriately than those who did not (adjusted prevalence ratio (APR): 0.60; 95% CI: 0.45–0.78). Participants who were not enrolled in CBHI had 1.37 times higher practice of accessing antibiotics inappropriately than those who were enrolled (APR: 1.37; 95% CI: 1.20–1.57). Farmers had 37% more increased practice of inappropriate antibiotic access than non-farmers (APR: 1.37; 95% CI: 1.001–1.88). Participants who were concerned about their distance to the healthcare site were 1.27 times more likely to obtain antibiotics inappropriately than those who were not (APR: 1.27; 95% CI: 1.10–1.45). Participants who used sub-optimal antibiotic doses in the past six months were 1.30 times more likely to access them improperly than those who did not (APR: 1.30; 95% CI: 1.11–1.50). Urban participants with sub-optimal antibiotic dose use habits were 1.59 times more likely to access antibiotics inappropriately than rural participants with sub-optimal antibiotic dose use practices (APR: 1.59; 95% CI: 1.22–2.07) (Table 3).

Table 3 Predictors of inappropriate antibiotic access practices among study participants from selected sub-districts of Harar and Haramaya, eastern Ethiopia, 2023.
Full size table

In GSEM, the final model included nine exogenous variables (gender, residence, CBHI, distance, cost of medicine, drug availability, trust in drug provider, healthcare inconvenience, and number of antibiotic courses received); two mediator variables (UDSA and sub-optimal antibiotic dose use practices), and three endogenous (outcome) variables (non-prescribed access, access from unlicensed providers, and access from informal sellers). The multilevel variance component considered was village (Fig. 2).

Figure 2
figure 2

Relationships between factors and inappropriate antibiotic access categories among participants from selected sub-districts of Harar and Haramaya, eastern Ethiopia, 2023. Note: ina3: inappropriate antibiotic access categories (1: non-prescribed access; 2: access from unlicensed providers; 3: access from informal sources); few_dose: sub-optimal antibiotic dose use practices (yes, no); UDSA: usual drug supplier type attended; CHBI: community-based health insurance; conven: healthcare convenience; avail: drug availability; resid: residence; and course: number of antibiotic courses received.

Full size image

As shown in Table 4, urban residence had positive direct (β = 2.45; 95% CI: 1.85–3.05), indirect (β = 0.76; 95% CI: − 0.02–1.55), and total (β = 3.21; 95% CI: 2.20–4.23) effects on the practice of non-prescribed antibiotic access. Similarly, concern about drug availability had positive direct (β = 0.96; 95% CI: 0.38–1.54), indirect (β = 1.36; 95% CI: − 0.01–2.74), and total (β = 2.32; 95% CI: 0.83–3.82) effects on the practice of non-prescribed antibiotic access. Healthcare inconvenience had positive direct (β = 0.49; 95% CI: 0.04–0.93), indirect (β = 0.79; 95% CI: 0.26–1.32), and total (β = 1.27; 95% CI: 0.58–1.97) effects on the practice of non-prescribed antibiotic access. Distance to the healthcare site also had positive direct (β = 0.58; 95% CI: 0.08–1.09), indirect (β = 0.18; 95% CI: − 0.06–0.42), and total (β = 0.76; 95% CI: 0.20–1.32) effects on the practice of non-prescribed antibiotic access. Furthermore, other factors with direct positive effects on non-prescribed antibiotic access included sub-optimal antibiotic dose use practices (β = 1.78; 95% CI: 1.35–2.21), non-enrollment to CBHI (β = 0.75; 95% CI: 0.35–1.16), and trust in drug suppliers (β = 0.62; 95% CI: 0.20–1.04).

Table 4 Direct, indirect and total effects of factors and mediators on the inappropriate antibiotic access domains among study participants from selected sub-districts of Harar and Haramaya, eastern Ethiopia, 2023.
Full size table

Rural residence had a positive direct effect (β = 2.14; 95% CI: 1.62–2.66) on the practice of accessing antibiotics from unlicensed providers. Distance to the healthcare site also had positive direct (β = 0.50; 95% CI: 0.17–0.83), indirect (β = 0.27; 95% CI: − 0.004–0.55), and total (β = 0.77; 95% CI: 0.34–1.21) effects on the practice of antibiotic access from unlicensed providers. Nonetheless, concerns about the cost of medicine had negative direct (β =  − 0.65; 95% CI: − 0.99 to − 0.32), indirect (β =  − 0.45; 95% CI: − 0. 0.75 to − 0.15), and total (β =  − 1.10; 95% CI: − 1.55 to − 0.66) effects on the practice of antibiotic access from unlicensed providers. Similarly, appropriate usual drug supplier type attendance had direct negative effects on the practice of accessing antibiotics from unlicensed providers (β =  − 0.94; 95% CI: − 1.37 to − 0.52) and informal sources (β =  − 2.53; 95% CI: − 4.06 to − 1.00). Other factors, such as repeat courses with sub-optimal dose uses than needed (β = 1.20; 95% CI: 0.83–1.58) and female gender (β = 1.15; 95% CI: 0.23–2.07) had direct positive effects on the practice of receiving antibiotics from informal sources (Table 4).

Discussion

We found that nearly half of the study participants' antibiotic access practices in the past six months were inappropriate, with non-prescribed access and purchases from unlicensed providers or sellers being the main domains assessed. Independent predictors of the non-prescribed access were urban residence, lack of health insurance, drug availability concern, sub-optimal antibiotic dose use habits, and healthcare inconvenience. Antibiotic access from unlicensed providers/sellers was predicted favorably by rural residence, distance, female gender, and repeat courses with sub-optimal dose uses, while cost and appropriate drug supplier type attendance were negatively associated.

Unauthorized antibiotic access was evident in 45.6% of the antibiotics received in the last six months. These were obtained without prescription or through unlicensed sales. Antibiotics are prescription-only medicines that must be accessed via prescription from a licensed practitioner. Antibiotics dispensed by pharmacists without prescriptions are inappropriate, as are sales by unlicensed personnel and premises57. Many community members in developing nations access antibiotics without prescriptions58,59. Health system gaps in drug affordability and availability13, along with health insurance limits and wait times60, can trigger this type of inappropriate antibiotic access. However, there is no guarantee of correct indication for all antibiotics obtained this way61,62. It is more likely to result in misuse63 and clinical risks such as superinfection and drug resistance62. With this, our study findings and the literature reveal that unlicensed drug sellers are the prominent sources of healthcare and antibiotics in rural areas64,65. The quality of antibiotics obtained from informal drug sellers, such as kiosks and street vendors, is frequently compromised with evidence that medicines contain less than 90% of the Pharmacopoeial limits for active pharmaceutical ingredients66,67. There are significant risks associated with sub-standard and falsified antibiotics68,69 that can result in inadequate infection treatment and increase the risk of antibiotic resistance. These informal sellers are also known as the principal suppliers of substandard antibiotics in resource-constrained settings, accounting for up to 30% of the market share70. Obtaining antibiotics in this manner may result in access to repackaged medications that lack the original packaging label and information71, exposing clients to access expired antibiotics with risks of toxicity and sub-optimal therapy. Furthermore, due to a lack of evidence of quality, substandard antibiotics are more likely to make their way into the hands of informal sellers9, resulting in sub-optimal dose use, resistance, treatment failure, and adverse effects72.

Antibiotic access from pharmacies without a prescription was more common in urban areas73,74, while unlicensed providers supplied antibiotics more commonly in rural areas75. Rural areas had the lowest density of registered antibiotic providers, such as pharmacies, hospitals, and health centers23. In this setting, primary clinics serve as the key source of antibiotics in rural areas23,76. However, primary clinics are licensed to give emergency medicines only, and this does not cover the provision of antibiotics56. Such transactions are unlicensed because they are not authorized supplies23. Though informal drug sellers can be found in any health system77, the quality of drugs they provide cannot be assured78. These supplies have an added risk of poor drug storage due to covert operations and the difficulties of meeting requirements by non-professionals. These unlicensed sellers are also not authorized to procure medicines, which may pose questions about the source of antibiotics they dispense79.

The practice of antibiotic access without prescription was more common among participants who were not enrolled in the CBHI scheme than those enrolled. In Ethiopia, the CBHI plan is a paid membership for households and enterprises to help cover the healthcare needs of low-income community members80,81. Household members enrolled in the plan often obtain drugs from public health facilities via prescription81,82. However, distance and health system skepticism, including concerns such as drug unavailability and inconvenience with the healthcare process, can challenge visits to public health facilities for access to prescribed antibiotics83. Accessing official healthcare may be problematic due to location and institutional disadvantages, including poor road conditions84 and long wait times85, which may encourage the public to access antibiotics from local markets and stores without prescriptions. Furthermore, a cost concern may push clients to consider antibiotic access from unlicensed sellers to minimize the extra costs incurred58. Community members often seek healthcare options in local markets and private premises to obtain antibiotics, as these access points may be more convenient than public health facilities13.

Individuals who obtained multiple antibiotic courses in the last six months did so mostly from informal sources, and they were more likely to use them in lower doses than needed. Sub-optimal dosing appeared to result in repeated antibiotic courses being accessed to treat unresolved infections86, as a partial dosage regimen is inadequate to treat the target condition. Additionally, prescribed use of antibiotics for minor illnesses may increase the chance of subsequent non-prescribed access for similar symptoms87. Though the pattern is uneven and varies by context88, the habit of using lower antibiotic doses than needed was more prevalent in urban than rural residents. In line with this, women are more likely to obtain antibiotics without a prescription for themselves and their children12,89, which was primarily from informal sources. Our qualitative findings also reveal that females frequently take one or two doses of antibiotic from a nearby kiosk to treat their gynecological discomfort, indicating that women are misinformed about how to treat their symptoms and the risks associated with this practice, putting them at future risk of drug-resistant infections.

Despite a large sample size and urban–rural comparison, this study has some limitations. First, the study was carried out in specific rural and urban areas of eastern Ethiopia. It may not be representative of the antibiotic access practices throughout the country. Second, we may not have recorded all antibiotics received by household members in the last six months due to poor recall with self-reports, failure to keep medicine packs, or a lack of public awareness that the pills taken were antibiotics. Third, our stratified two-stage sampling strategy may have resulted in an unequal distribution of local healthcare options due to a smaller proportion of formal versus informal drug suppliers in some areas. Although we collected information regarding the sources of antibiotics from participants, they may have provided socially desirable responses, especially in the case of access from informal sources. In this regard, we could have underestimated the prevalence of inappropriate antibiotic access practices. Despite these limitations, the findings of this study add to the body of knowledge on the contextual predictors of inappropriate antibiotic access practices by domains, enabling a targeted intervention to preserve these drugs from resistance. The study also sheds light on the magnitude of inappropriate antibiotic practices, such as non-prescribed access and access from unlicensed providers and informal sources.

Conclusions

We found that nearly half of antibiotic access practices at the community level in eastern Ethiopia were inappropriate and differed significantly between urban and rural areas. Antibiotics were widely available without prescription in urban areas due to non-restrictive antibiotic dispensing practices by pharmacies, while access from unlicensed drug sellers was more prevalent in rural areas. Sub-optimal antibiotic dosages were routinely consumed by women in rural areas, primarily from informal sources. Hence, the contextual predictors we identified can help design context-specific interventions that address the provider and user side domains, promoting antibiotic access from official suppliers through prescription recommendations and contributing to worldwide efforts to contain antibiotic resistance. In this regard, designing an efficient regulatory intervention is mandatory to continually oversee the practices of pharmacies/drugstores, primary clinics, and kiosks in rural areas. An educational intervention is also required for urban people on the risks of using non-prescribed and sub-optimal antibiotic dosages from pharmacies. Furthermore, educating women about the hazards of using lower antibiotic doses from unlicensed sellers may help lessen their misconceptions about where to seek medical care.