The facility-based retrospective follow-up study was conducted in ART clinics of public health facilities in Gebi Resu (Zone 3), Afar National Regional State, Ethiopia.

The zone has 11 health centers and 1 general hospital, but only 8 health centers and 1 hospital offer ART services. The study was conducted among PLWH receiving ART services at selected health facilities in the Gebi Resu zone from 11 September 2017 to 10 September 2022. All PLWH who initiated ART for the first time during this period were eligible for inclusion. Since at least two consecutive measurements are required to declare viral load suppression, individuals with less than two consecutive viral load measurements were excluded from the study.

The sample size was calculated using the formulas designed for survival analysis through STATA statistical software, the log-rank test, and the Cox proportional hazard model. The following assumptions were taken into consideration: confidence level = 95%, power = 90%, sample size allocation ratio = 1:1, the hazard ratio of patient’s baseline CD4 level of less than 200 compared to their counterpart 0.683, the hazard ratio of good ART adherence level 2.648, and 1.85 of hazard ratio among patients having opportunistic infection taken from the study conducted in Arba Minch (12) and West Gojam zone (14). Finally, a sample of 298 PLWH was determined after comparing the calculated sample sizes.

Based on the difference in the level of care, we stratified them into hospitals and health centers from a total of nine public health facilities providing ART service in the Gebi Resu zone of the Afar regional state. Only one hospital (Mohammed Akle Referral Hospital) in the zone provided ART service. In addition, three health centers were selected from the seven listed health centers by employing a simple random sampling method. Then, according to the inclusion criteria, eligible PLWH who started ART from 11 September 2017 to 10 September 2022, were identified from the registration book of each selected ART clinic. Finally, study units and selected PLWH receiving ART were randomly selected and allocated proportionally to each selected health facility.

Viral suppression refers to a viral load less than or equal to 1,000 copies/mL.

The adherence level is classified as good if average adherence is >95% or the patient misses ≤2 from 30 doses and < 3 from 60 doses, fair if the average adherence is 85–94% or the patient misses 3–5 from 30 doses and 3–9 from 60 doses, and poor if the average adherence is <85% or the patient misses ≥6 from 30 doses and > 9 from 60 doses.

Body Mass Index (BMI) is categorized as follows: underweight <18.5 kg/m2, normal weight 18.5–24.99 kg/m2, overweight 25–29.99 kg/m2, and obesity > = 30 kg/m2.

The data for the study were collected by four data collectors who were working on ART service provision, and the collection process was supervised by four trained clinical officers. Data were collected using a structured data abstraction checklist that included sociodemographic variables, clinical factors, and treatment-related variables. The data abstraction format was prepared in the English language. The data were collected through a review of patient medical records, electronic ART databases, and other related registration books (ART registration book, viral load registration book, and HIV-positive tracking registration book) at the selected facilities.

The study’s objectives, purposes, questionnaire, data collection procedures, roles in the evaluation, best practices for high-quality data, confidentiality, and supervision techniques were all covered in a 2-day training session for all data collectors (ART providers) and supervisors (clinical officers). Additionally, the data-gathering tool was pre-tested, and supervisors monitored and verified the data collection procedures. Before processing, the gathered data were stored in a secure location and verified every day for accuracy. After the data were collected, each questionnaire was coded separately.

Patients were not involved in this study.

The collected data were coded and entered into Epi-Info version 7.2. These data were then exported to STATA version 16 for processing and analysis. Descriptive statistics such as frequency distribution, percentage, median, and interquartile range were calculated to summarize the characteristics of the study participants. The results of the variables were displayed using frequency tables and graphs. Kaplan–Meier survival curves were used to compare the event times between two or more groups. Observed survival differences were assessed using log-rank tests, with a significance level set at 5%.

Cox proportional-hazard regression was used to identify the predictors of time to viral load suppression among PLWH on ART. The event of interest under this particular objective was viral load suppression after enrolling in ART. The proportional hazard assumption was checked statistically using the global goodness-of-fit test proposed by Schoenfeld. The proportional hazard assumption was fulfilled with a global test value of 0.5661. The assumption was also checked for each predictor with minimum and maximum p-values of 0.2350 and 0.9753, respectively. The goodness of fit was checked using the Cox-Snell residual test.

Predictors with p-value of less than 0.25 in the bivariate Cox regression analysis were selected as candidates for multiple regression analysis. All predictors were then included in the multiple regression model to identify those independently associated with the outcomes of interest, adjusting for potential confounding variables. Variables with a p-value of less than 0.05 were considered as significantly associated with the dependent variable. The strength of the association between dependent and independent variables was expressed as a hazard ratio with a 95% confidence interval.

A total of 812 PLWH received ART at public health facilities in the Gebi Resu zone during the data collection period. Of these, 298 participants who fulfilled the inclusion criteria were randomly selected (Figure 1). From a total of 298 PLWH on ART, 204 participants (68.5%) were male individuals, 136 (45.6%) of them were married, 107 participants (35.9%) had no formal education, 205 participants (68.8%) resided in urban areas, 151 participants (50.7%) were Orthodox Christians, 88 participants (29.4%) were private employees, 39 participants (13.1%) were government employees, 27 participants (9.1%) were merchants, and 11 participants (3.7%) were pastoralists. (Table 1).

Of the 298 study participants, 109 participants (36.6%) were on a baseline ART regimen of ABC/3TC/NVP or EFV, 254 participants (85.2%) demonstrated good medication adherence, 162 participants (54.4%) received cotrimoxazole prophylaxis therapy (CPT), 195 participants (65.4%) received isoniazid preventive therapy (IPT), and 216 participants (72.5%) disclosed their HIV status to family and friends (Table 2).

Of the 298,298 study participants, 208 participants (69.8%) were at WHO clinical stages 1 and 2, 98 participants (32.9%) had previously developed an opportunistic infection, 78 participants (26.2%) had a tuberculosis co-infection, and 213 participants (71.55%) had a baseline CD4 count of greater than or equal to 200 cells/mm.³ Additionally, 275 participants (92.3%) had a baseline viral load of less than 1,000 copies/ml, and 208 participants (69.8%) had a baseline body mass index (BMI) greater than 18.5 kg/m² (Table 3).

The incidence proportions of people who have achieved first viral load suppression were 83.89% (95% CI (79.25, 87.66)). The total observation time for the 298 patients was 2,644 person-months, and the viral load suppression incidence rate was 9.46 per 100 person-months. The median time to viral load suppression was 7.7 months with an interquartile range of 3.8 months (Q1 = 6.47 and Q3 = 10.27).

The Kaplan–Meier survival function graph presents the overall time to viral load suppression, with a median survival time of approximately 8 months, indicating that 50% of PLWH achieve viral load suppression approximately 8 months after initiating ART (Figure 2).

The Kaplan–Meier survival estimate and log-rank test were used to compare the probability of time to viral load suppression across different variable groups. Significant differences in time to viral load suppression were observed between the categories of WHO clinical stage, CPT use, baseline CD4 count, and drug adherence groups (Figures 3–6).

Variables with a p-value of <0.25 in the bivariate Cox regression analysis, such as sex, age, residence, BMI, WHO clinical stage, adherence, CPT, IPT, baseline CD4, and past opportunistic infection, were included in the multivariate Cox regression model (Table 4).

In the multivariate Cox regression model, WHO clinical stage, CPT, and drug adherence were significant predictors of viral load suppression among PLWH with a p-value of <0.05 within a 95% confidence interval.

Patients at WHO clinical stages 3 and 4 had lower hazards of achieving viral load suppression by 33% compared to patients at clinical stages 1 and 2 (AHR = 0.67, 95%CI (0.47, 0.96)). Patients who received CPT had 1.47 times higher probability of achieving viral load suppression (AHR = 1.47, 95%CI (1.12, 1.92)). Patients with poor drug adherence had a reduced hazard of viral load suppression by 60% compared to patients with good drug adherence (AHR = 0.40, 95%CI (0.18, 0.90)) (Table 5).