This study employed an analytical cross-sectional design and was conducted in the urban setting of Brazzaville, Republic of the Congo. The study was conducted in January 2024, with data collection carried out from January 10 to January 28, 2024. The target population consisted of households within the city limits of Brazzaville containing at least one adult respondent (aged ≥18 years) deemed capable of providing reliable and informed responses to the structured questionnaire. Households without an eligible respondent present at the time of the visit were not considered for inclusion.

The sample size N of households was calculated using the following formula:

$N=e\left[\frac{Z^2.p\left(1-p\right)}{d^{{}^2}}\right]$

where, Z = 1.96 (corresponding to a 95% confidence level), p is the expected proportion set at 50%, d is the margin of error set at 5%, and e = 1.875 is the design effect used to account for the sampling method. Using this formula, a total sample size of 720 households was obtained, i.e., a sample of 180 households per health district will be included, based on 20 households per counting zone (CZ) and 6 CZs per district.

A three-stage stratified sampling design was employed to generate a representative sample of the population residing in Brazzaville. At the first stage, four health districts, Bacongo, Djiri, Ouenzé, and Talangaï, were randomly selected from the ten administrative health districts constituting the city, using simple random sampling without replacement. At the second stage, within each selected district, six census zones (CZs) were selected using probability proportional to size (PPS) sampling. The selection probability for each CZ was proportional to the number of households recorded in the Fifth General Population and Housing Census (RGPH-5, 2023). This method ensured adequate representation of more densely populated zones and minimized spatial clustering bias. At the third stage, systematic random sampling was applied to select households within each selected CZ. The sampling interval was calculated by dividing the total number of households in the CZ by the target number of households to be surveyed (30 per CZ). The first household was randomly selected within the initial interval range, and subsequent households were selected using a fixed sampling interval. In each selected household, all adults present at the time of the survey visit were considered eligible. Adults who were present at the time of the survey but declined to participate were excluded from the study. Similarly, individuals who provided incomplete or internally inconsistent responses were not included in the final analysis. Non-respondents were immediately replaced by another member of the household concerned. If it was not possible to substitute in the surveyed household, the next household was used and so on. A total of 734 individuals were interviewed across 720 households surveyed.

Data collection was conducted between 10 and 28 January 2024 in four selected health districts of Brazzaville: Bacongo, Djiri, Ouenzé, and Talangaï. A structured questionnaire was administered face-to-face to a responsible adult present in each selected household. The questionnaire was pre-tested on a pilot sample of 30 households to assess the clarity of the questions, item validity, and contextual relevance. Necessary modifications were made following the pre-test. Specifically, some questions were reworded to improve clarity, redundant items were removed, and certain culturally sensitive or ambiguous terms were revised to ensure better comprehension. Additionally, response options for several closed-ended questions were expanded to capture a broader range of participant experiences. The final questionnaire comprised sections on socio-demographic, economic, cultural, and perceptual characteristics, as well as on the use of street medicines. Interviews were carried out by trained fieldworkers under the supervision of experienced supervisors. Prior to the survey, all fieldworkers received specific training covering the objectives of the study, interview techniques, ethical considerations, and data confidentiality. Data were collected using paper-based forms and subsequently entered into a secure database using R software (version 4.4.1). A double data entry process was employed to minimise errors.

Outcome variable

The outcome variable, use of street medicines (SM), was measured as a binary variable, distinguishing between individuals who reported using such products (users) and those who did not (non-users). This variable was assessed through a specific question addressed to each adult present in the household at the time of the survey: “Have you used any medicines purchased from informal vendors or in the street during the past 12 months?” Responses were coded as “1” for users and “0” for non-users. This operational definition allowed for a clear individual-level classification based on recent use of street medicines.

Predictor variables

Among the socio-demographic variables, age was categorised into three groups: under 30 years, 30–49 years, and 50 years or older, the latter representing a population potentially more exposed to chronic illnesses and therefore more likely to use medications. Sex was dichotomised as male and female, aiming to explore potential gender-related differences in health-seeking behaviours and access to care.

Monthly income was categorised into three groups: below the minimum wage, equal to the minimum wage, and above the minimum wage, in order to highlight the potential impact of economic constraints on health-related behaviours. Educational level ranged from no formal schooling to higher education, allowing for the assessment of how educational capital may influence knowledge, perceptions, and the use of pharmaceutical products.

Marital status was classified as single, married, divorced/widowed, or in a consensual union. This variable may reflect levels of social or economic stability, which can in turn affect health behaviours.

Religion was categorised as Christianity, local/traditional religion, or no religion, with the aim of exploring the influence of spiritual beliefs and representations on health practices.

Socio-professional category distinguished between professionals, traders, students/pupils, manual labourers, and unemployed individuals, in order to capture occupational dynamics and their impact on healthcare access.

The number of dependents was classified as none, 1 to 5, and more than 5, reflecting the household’s economic burden, which may influence therapeutic choices. Place of residence (Bacongo, Djiri, Ouenzé, Talangaï) was included to account for territorial disparities, both in terms of formal healthcare provision and the density of the informal medicine market.

Nationality (Congolese or foreign) was considered in order to assess potential discrimination or inequalities in access to healthcare linked to migration status.

Two perceptual variables were considered:

Data were entered and analysed using R software, version 4.4.1. The first step consisted of a descriptive analysis to summarise the socio-demographic, economic, and perceptual characteristics of the participants. Qualitative variables were presented as frequencies and percentages. In the second step, a bivariate analysis was conducted to identify variables potentially associated with the use of street medicines (SM). Comparisons between “users” and “non-users” were performed using the chi-square test (or Fisher’s exact test when expected cell counts were low). A p-value < 0.05 was considered statistically significant. Variables showing a significant association in the bivariate analysis (p < 0.20) were included in a multivariable binary logistic regression model to identify factors independently associated with the use of SM. Results were expressed as odds ratios (ORs) with their corresponding 95% confidence intervals (95% CI). Statistical significance was set at p < 0.05. The performance of the final model was assessed using the pseudo R² index, the likelihood ratio (chi-square) test, and the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). Model validity was checked by examining residuals and ensuring the absence of multicollinearity between explanatory variables.

The study adhered to ethical standards for research involving humans. Informed consent was obtained after explaining the study purpose, voluntary participation, and confidentiality assurances. Data were managed solely by the principal investigator to ensure privacy. All required approvals were secured from local health authorities and academic committees.

Table 1 presents the socio-demographic and economic characteristics of participants according to their status of street medicine (SM) use. Among the 734 respondents, 88.4% (n = 649) were users. Users were predominantly aged 50 years or older (60.4%, p < 0.001) and male (65.5%), although the latter was not statistically significant (p = 0.113). They more frequently lived in cohabitation (34.4%, p < 0.001) and mainly had a secondary education level (52.4%, p < 0.001). Most users identified as Christian (64.4%) or adhered to local religions (22.8%, p < 0.001), and were often engaged in commercial activities (43.1%, p < 0.001). The majority had between one and five dependents (66.1%, p < 0.001) and reported a monthly income below the minimum wage (41.8%, p = 0.027). Geographically, they mainly resided in Ouenzé (31.0%, p < 0.001) and were primarily of Congolese nationality (87.8%, p = 0.017).

Table 2 highlights several factors significantly associated with the use of street

medicines (SM). Compared to individuals earning above the minimum wage (SMIG), those with an income equal to the SMIG have a higher risk of SM use (OR = 3.643; p = 0.024). Certain occupational categories are also more exposed, notably traders (OR = 3.35; p = 0.026) and students/pupils (OR = 4.034; p = 0.011). Residence in Ouenzé (OR = 10.682; p < 0.001) is strongly associated with SM use, as it has a secondary level of education (OR = 4.569; p < 0.001) compared to higher education. A positive perception of the quality of SM, particularly when judged “better” than formal medicines (OR = 7.81; p = 0.001), significantly increases the likelihood of use. Finally, a lack of awareness about the dangers of SM is also a strongly associated factor (OR ranging from 4.158 to 4.914; p ≤ 0.001).