Second-line antiretroviral therapy success rates and predictors among people living with HIV in Rwanda, 2019–2023 Authors: Jean Claude Kwizera, Clarisse Marie Claudine Simbi, Gallican Nshogoza Rwibasira, Bibiane Uwamahoro, Amanuel Kidane Andegiorgish Published: 2026 ABSTRACT AbstractBackgroundDespite advances in HIV treatment, managing people living with HIV (PLHIV) on second-line antiretroviral therapy (ART) remains challenging. Globally, second-line ART achieves an average viral suppression rate (VLS) of 81.2%. In Rwanda, evidence is limited; the most recent national study in 2016 reported a 17% failure rate. This study assessed current VLS proportions and predictors of second-line ART success in Rwanda.MethodsA retrospective cohort study was conducted at 277 health facilities across Rwanda, including PLHIV who initiated second-line ART between January 2019 and December 2023. The primary outcome was VLS, defined as a viral load of ≤ 1000 copies/mL at the most recent measurement recorded at least six months after second-line initiation. Data were abstracted from patient files using a standardized tool, and multivariate logistic regression identified independent predictors of VLS.ResultsAnalysis included 726 individuals; overall VLS was 88.8% (645/726). Most participants were female (63.9%), aged 35–49 years (38.7%), on protease inhibitor-based regimens (78.7%), and on ART for 10 years or more (80.9%). Participants aged 15–24 years had lower odds of VLS compared with the 35–49 year reference group (aOR = 0.34; 95% CI: 0.13–0.92), and students had lower odds than participants (aOR = 0.18; 95% CI: 0.06–0.51). Residence in Kigali City (aOR = 2.47; 95% CI: 1.07–6.26), Northern (aOR = 3.57; 95% CI: 1.37–10.95), and Southern Province (aOR = 3.84; 95% CI: 1.58–10.53) was independently associated with higher odds of VLS compared with Eastern Province. PLHIV with poor or moderate ART adherence had 0.25 times the odds of VLS (aOR = 0.25; 95% CI: 0.14–0.44); alcohol or illicit drug use (aOR = 0.44; 95% CI: 0.22–0.94) and NCD comorbidity (aOR = 0.22; 95% CI: 0.06–0.80) were also independently associated with lower VLS.ConclusionRwanda’s second-line VLS of 88.8% exceeds the global average and improves on the 2016 estimate but remains below the UNAIDS 95% target. Targeted interventions addressing adherence, substance use, and regional disparities, and integrated NCD services are needed. Tailored youth-friendly care, peer support, and integrated mental health services should be prioritized to improve outcomes among younger populations.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12879-026-13851-0.Keywords: Second-line ART, HIV, Treatment outcomes, Rwanda, Predictors, Virological suppression, Comorbidities Background The HIV continuum of care is a framework that encompasses the sequential steps in HIV management, from initial diagnosis through to sustained viral suppression [1]. Central to this continuum is antiretroviral therapy (ART), which has transformed HIV care globally by substantially improving quality of life and life expectancy for people living with HIV (PLHIV) [2]. Over the past decades, ART has evolved considerably, progressing from less safe drugs with significant side effects to more advanced and safer antiretroviral medications. These modern drugs, available as combination antiretroviral therapy (cART), are associated with reduced side effects, especially in first-line ART [3]. However, some patients on first-line ART do not achieve the desired treatment outcomes and may experience treatment failure due to drug resistance, poor adherence, or clinical complications. In such cases, second-line ART becomes necessary to achieve viral load suppression (VLS) and prevent disease progression [4]. Viral load (VL) monitoring is the cornerstone in evaluating the effectiveness of ART and detecting treatment failure. WHO guidelines recommend conducting routine viral load testing at the sixth and twelfth month after ART initiation, and then every twelve months for virally suppressed clients [5]. More frequent testing is recommended for patients with detectable VL, typically every three months, until the patient achieves viral suppression. Regular VL monitoring helps to identify therapy failure promptly, indicating the need for ART line transition [6]. Optimal second-line ART effectiveness is essential for individual health and public health alike, as it prevents the development of additional drug resistance and opportunistic infections, and reduces the risk of HIV transmission [7]. However, managing second-line ART is more challenging than first-line therapy, given the complexity of the regimens involved and the risk of cross-resistance between antiretroviral drug classes. In Rwanda, patients failing second-line ART are eligible for third-line regimens; however, these are even more complex and costly, underscoring the importance of optimizing outcomes at the second-line stage to prevent further treatment failure. VLS rates among PLHIV on second-line ART vary across regions and settings. Globally, second-line ART failure remains a significant concern, with a systematic review reporting a failure rate of 18.8% in low- and middle-income countries [8]. In sub-Saharan Africa (SSA), incidence of second-line treatment failure has been reported between 18 and 23 per hundred person-years between 12 and 18 months of second-line initiation [9]. Within East Africa, failure rates range from 12.18% to 23% across studies from Uganda and Tanzania [10, 11]. In Rwanda specifically, a 2016 national study reported a 17% treatment failure rate among adults aged 15 years and older who transitioned to second-line ART between 2004 and 2016 [12]. Major recognized factors of virological failure among individuals on second-line ART, include low CD4 count or high VL at therapy line transition, suboptimal adherence, and drug-resistant mutations, ART regimen type, low body weight at the time of ART line transition, presence of opportunistic infections (OI), sex, younger age, and non-disclosure of HIV status [13–15]. Regionally, same predictors were identified, for example, in Uganda, poor adherence, low CD4 count at the switch, and high baseline VL, younger age, male sex, and the co-administration of tuberculosis treatment were key predictors of second-line ART failure [10]. Also in Rwanda, second-line therapy failure was linked to younger age groups (15–29 years), initiation of ART at late-stage disease (HIV clinical stage III or IV), reduced CD4 (≤ 500) at regimen transition, as well as being on an ATV-based therapy [12]. Drug-resistant mutations and adherence challenges significantly increase the risk of treatment failure, emphasizing the need to reinforce adherence strategies and ensure consistent clinical follow-up to achieve better therapeutic results [16]. Since the last national study on second-line ART outcomes in Rwanda in 2016, the HIV treatment landscape has evolved considerably. The PLHIV population has shifted toward older age groups, with the Rwanda 2024 HIV Annual Report indicating that approximately 35% of those on ART are now aged 50 years or older [17–19]. Treatment regimens have also advanced, with the introduction of integrase strand transfer inhibitors such as dolutegravir, now incorporated across all ART lines in Rwanda [19]. Concurrently, the growing burden of non-communicable disease comorbidities among PLHIV adds further complexity to second-line ART management, due to increased pill burden, potential drug-drug interactions, and competing clinical priorities. The COVID-19 pandemic further disrupted HIV service delivery, affecting clinic attendance, viral load monitoring, and ART adherence support. Together, these shifts in population characteristics, treatment regimens, comorbidity profiles, and pandemic-related disruptions create a substantially different clinical context, warranting an updated assessment of second-line ART outcomes in Rwanda. This study therefore aimed to assess VLS proportions and identify independent predictors of second-line ART success among PLHIV in Rwanda from 2019 to 2023. Methods Study design, setting and population A retrospective cohort was conducted to assess the outcomes and predictors of second-line ART among PLHIV in Rwanda. It included PLHIV who were transitioned on second-line ART between January 2019 and December 2023. The study was conducted in Rwanda, a landlocked country located in the Eastern Africa region, with an estimated 13.1 million population in 2022 [20]. By June 2024, there were 222,604 PLHIV on ART served from 591 health facilities [18]. Of them, 6,419 (2.9%) were on second-line ART. The study participants comprised PLHIV on second-line ART in Rwanda. All eligible PLHIV who initiated second-line ART between January 2019 and December 2023 were included. The inclusion criteria are: Documented failure of first-line ART based on virological criteria as per the Rwandan national guidelines and initiated second-line regimen between 2019 and 2023. Documented failure of first-line ART based on virological criteria as per the Rwandan national guidelines and initiated second-line regimen between 2019 and 2023. Availability of at least one VL measurement recorded at least six months after second-line ART initiation. Availability of at least one VL measurement recorded at least six months after second-line ART initiation. Exclusion criteria include: PLHIV without a VL measurement at least six months after second-line ART initiation. PLHIV without a VL measurement at least six months after second-line ART initiation. Sample size and data collection The minimum required sample size was estimated using the single proportion formula: where Z = 1.96 (Z-score for 95% confidence level), p = 0.17 (virological failure rate from a prior national Rwanda study [12]), and d = 0.05 (margin of error), yielding a minimum of 217 participants. To maximize statistical power, all PLHIV who initiated second-line ART between January 2019 and December 2023 considered in data collection and all those who met the inclusion criteria were included in the final analysis. To identify eligible participants, aggregate reports from the Rwanda Health Management Information System (RHMIS) were reviewed. This identified 283 health facilities across all 29 administrative districts of Rwanda that had initiated PLHIV on second-line ART during the study period, from which 821 PLHIV were recorded as potential candidates for inclusion. Data were collected through retrospective review of physical patient files at each participating facility, using a pre-tested structured data abstraction checklist to extract clinical, behavioral, and socio-demographic variables of interest. Each patient was identified using their tracnet number, a unique ART enrolment identifier assigned at HIV programme entry, to prevent duplicate records. Extracted data were entered into the Research Electronic Data Capture (REDCap) platform, from which a consolidated dataset was generated for analysis. Variables Dependent variable Second-Line ART Outcome: The outcome variable was viral load suppression status, defined as the most recent viral load measurement recorded at least six months after second-line ART initiation, in accordance with WHO guidelines threshold [5]. This was measured as a dichotomous variable with two categories: Viral load suppression: A viral load count of ≤ 1000 copies/mL at the most recent measurement taken at least six months after second-line ART initiation.Virological failure: A viral load count of > 1000 copies/mL at the most recent measurement taken at least six months after second-line ART initiation. Second-Line ART Outcome: The outcome variable was viral load suppression status, defined as the most recent viral load measurement recorded at least six months after second-line ART initiation, in accordance with WHO guidelines threshold [5]. This was measured as a dichotomous variable with two categories: Viral load suppression: A viral load count of ≤ 1000 copies/mL at the most recent measurement taken at least six months after second-line ART initiation.Virological failure: A viral load count of > 1000 copies/mL at the most recent measurement taken at least six months after second-line ART initiation. Viral load suppression: A viral load count of ≤ 1000 copies/mL at the most recent measurement taken at least six months after second-line ART initiation. Viral load suppression: A viral load count of ≤ 1000 copies/mL at the most recent measurement taken at least six months after second-line ART initiation. Virological failure: A viral load count of > 1000 copies/mL at the most recent measurement taken at least six months after second-line ART initiation. Virological failure: A viral load count of > 1000 copies/mL at the most recent measurement taken at least six months after second-line ART initiation. Independent variables The independent variables influencing second-line ART outcomes are categorized into socio-demographic, behavioral, and clinical factors. Socio-demographic factors: Sex: Male or Female.Age: Age at the time of data extraction, calculated from the date of birth as documented in the patient file.Employment status: classified as unemployed, employed, or student.Residence: Province of residence (Eastern, Kigali City, Southern, Northern, and Western).Marital status: Single, married/cohabiting, divorced/separated, or widowed. Socio-demographic factors: Sex: Male or Female.Age: Age at the time of data extraction, calculated from the date of birth as documented in the patient file.Employment status: classified as unemployed, employed, or student.Residence: Province of residence (Eastern, Kigali City, Southern, Northern, and Western).Marital status: Single, married/cohabiting, divorced/separated, or widowed. Sex: Male or Female. Sex: Male or Female. Age: Age at the time of data extraction, calculated from the date of birth as documented in the patient file. Age: Age at the time of data extraction, calculated from the date of birth as documented in the patient file. Employment status: classified as unemployed, employed, or student. Employment status: classified as unemployed, employed, or student. Residence: Province of residence (Eastern, Kigali City, Southern, Northern, and Western). Residence: Province of residence (Eastern, Kigali City, Southern, Northern, and Western). Marital status: Single, married/cohabiting, divorced/separated, or widowed. Marital status: Single, married/cohabiting, divorced/separated, or widowed. 2.Behavioral factors: Alcohol or illicit drug use: Current alcohol or any illicit drug use, as recorded in the patient file at the most recent clinic visit prior to data extraction.ART adherence: measured using the proportion of pills taken over the prescribed period as documented at the most recent visit prior to data extraction, categorized as good adherence (≥ 95%) or poor/moderate adherence (< 95%). Behavioral factors: Alcohol or illicit drug use: Current alcohol or any illicit drug use, as recorded in the patient file at the most recent clinic visit prior to data extraction.ART adherence: measured using the proportion of pills taken over the prescribed period as documented at the most recent visit prior to data extraction, categorized as good adherence (≥ 95%) or poor/moderate adherence (< 95%). Alcohol or illicit drug use: Current alcohol or any illicit drug use, as recorded in the patient file at the most recent clinic visit prior to data extraction. Alcohol or illicit drug use: Current alcohol or any illicit drug use, as recorded in the patient file at the most recent clinic visit prior to data extraction. ART adherence: measured using the proportion of pills taken over the prescribed period as documented at the most recent visit prior to data extraction, categorized as good adherence (≥ 95%) or poor/moderate adherence (< 95%). ART adherence: measured using the proportion of pills taken over the prescribed period as documented at the most recent visit prior to data extraction, categorized as good adherence (≥ 95%) or poor/moderate adherence (< 95%). 3.Clinical factors: Duration on ART: Total duration the patient has been on ART, including both first-line and second-line regimens.History of opportunistic infection: Whether the client had a documented diagnosis of, or treatment for, any opportunistic infection (including but not limited to tuberculosis, cryptococcal meningitis, Pneumocystis pneumonia, and toxoplasmosis) at any point during their ART history up to the time of data extraction. This was recorded as Yes/No.NCD comorbidity: The presence of at least one documented NCD diagnosis in the patient file, including but not limited to hypertension, diabetes mellitus, chronic kidney disease, and cardiovascular disease. This was recorded as Yes/No.First-line regimen: Type of first-line ART regimen used before transitioning to second-line ART. Regimens were classified by NRTI backbone: tenofovir (TDF)-based, zidovudine (AZT)-based, abacavir (ABC)-based, or stavudine (d4T)-based.Second-line regimen: The specific second-line ART regimen the patient was receiving at the time of data extraction, classified by the third agent as: atazanavir/ritonavir (ATV/r)-based, lopinavir/ritonavir (LPV/r)-based, or dolutegravir (DTG)-based.Viral load at ART line transition: Viral load count at the time of the transition from first-line to second-line ART.Period of the latest VL measurement: The calendar period in which the most recent viral load measurement was recorded after second-line ART initiation, categorized as: pre-COVID-19 period (before 2020), COVID-19 period (2020–2022), and post-COVID-19 period (2023 and beyond). Clinical factors: Duration on ART: Total duration the patient has been on ART, including both first-line and second-line regimens.History of opportunistic infection: Whether the client had a documented diagnosis of, or treatment for, any opportunistic infection (including but not limited to tuberculosis, cryptococcal meningitis, Pneumocystis pneumonia, and toxoplasmosis) at any point during their ART history up to the time of data extraction. This was recorded as Yes/No.NCD comorbidity: The presence of at least one documented NCD diagnosis in the patient file, including but not limited to hypertension, diabetes mellitus, chronic kidney disease, and cardiovascular disease. This was recorded as Yes/No.First-line regimen: Type of first-line ART regimen used before transitioning to second-line ART. Regimens were classified by NRTI backbone: tenofovir (TDF)-based, zidovudine (AZT)-based, abacavir (ABC)-based, or stavudine (d4T)-based.Second-line regimen: The specific second-line ART regimen the patient was receiving at the time of data extraction, classified by the third agent as: atazanavir/ritonavir (ATV/r)-based, lopinavir/ritonavir (LPV/r)-based, or dolutegravir (DTG)-based.Viral load at ART line transition: Viral load count at the time of the transition from first-line to second-line ART.Period of the latest VL measurement: The calendar period in which the most recent viral load measurement was recorded after second-line ART initiation, categorized as: pre-COVID-19 period (before 2020), COVID-19 period (2020–2022), and post-COVID-19 period (2023 and beyond). Duration on ART: Total duration the patient has been on ART, including both first-line and second-line regimens. Duration on ART: Total duration the patient has been on ART, including both first-line and second-line regimens. History of opportunistic infection: Whether the client had a documented diagnosis of, or treatment for, any opportunistic infection (including but not limited to tuberculosis, cryptococcal meningitis, Pneumocystis pneumonia, and toxoplasmosis) at any point during their ART history up to the time of data extraction. This was recorded as Yes/No. History of opportunistic infection: Whether the client had a documented diagnosis of, or treatment for, any opportunistic infection (including but not limited to tuberculosis, cryptococcal meningitis, Pneumocystis pneumonia, and toxoplasmosis) at any point during their ART history up to the time of data extraction. This was recorded as Yes/No. NCD comorbidity: The presence of at least one documented NCD diagnosis in the patient file, including but not limited to hypertension, diabetes mellitus, chronic kidney disease, and cardiovascular disease. This was recorded as Yes/No. NCD comorbidity: The presence of at least one documented NCD diagnosis in the patient file, including but not limited to hypertension, diabetes mellitus, chronic kidney disease, and cardiovascular disease. This was recorded as Yes/No. First-line regimen: Type of first-line ART regimen used before transitioning to second-line ART. Regimens were classified by NRTI backbone: tenofovir (TDF)-based, zidovudine (AZT)-based, abacavir (ABC)-based, or stavudine (d4T)-based. First-line regimen: Type of first-line ART regimen used before transitioning to second-line ART. Regimens were classified by NRTI backbone: tenofovir (TDF)-based, zidovudine (AZT)-based, abacavir (ABC)-based, or stavudine (d4T)-based. Second-line regimen: The specific second-line ART regimen the patient was receiving at the time of data extraction, classified by the third agent as: atazanavir/ritonavir (ATV/r)-based, lopinavir/ritonavir (LPV/r)-based, or dolutegravir (DTG)-based. Second-line regimen: The specific second-line ART regimen the patient was receiving at the time of data extraction, classified by the third agent as: atazanavir/ritonavir (ATV/r)-based, lopinavir/ritonavir (LPV/r)-based, or dolutegravir (DTG)-based. Viral load at ART line transition: Viral load count at the time of the transition from first-line to second-line ART. Viral load at ART line transition: Viral load count at the time of the transition from first-line to second-line ART. Period of the latest VL measurement: The calendar period in which the most recent viral load measurement was recorded after second-line ART initiation, categorized as: pre-COVID-19 period (before 2020), COVID-19 period (2020–2022), and post-COVID-19 period (2023 and beyond). Period of the latest VL measurement: The calendar period in which the most recent viral load measurement was recorded after second-line ART initiation, categorized as: pre-COVID-19 period (before 2020), COVID-19 period (2020–2022), and post-COVID-19 period (2023 and beyond). Data analysis procedures Data analysis was performed using R statistical software, version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria, 2023). Descriptive statistics were used to summarize the characteristics of the study population; categorical variables were presented as frequencies and percentages, while continuous variables were described using median and interquartile range (IQR). Bivariate analysis using the Chi-square test was conducted to assess the association between each independent variable and the outcome of interest, viral load suppression. Variables with a p-value of less than 0.20 in the bivariate analysis, along with variables of a priori importance (age and sex), were considered for entry into the multivariate logistic regression model. Prior to model fitting, multicollinearity among independent variables was assessed using variance inflation factors (VIF), with a cut-off value of 5, to ensure the reliability of model estimates. Variables meeting these criteria were then entered into the multivariate logistic regression model to identify independent predictors of viral load suppression, accounting for potential confounding. The model estimated crude and adjusted odds ratios (cOR and aOR) with 95% confidence intervals (CI), with statistical significance set at p < 0.05. Results are presented in three tables covering descriptive, bivariate, and multivariate analyses. Ethical approval This study adhered to the ethical principles outlined in the Declaration of Helsinki and complied with the guidelines established by the Rwanda National Ethics Committee. Ethical approval for this study was sought from the University of Rwanda Institutional Review Board (UR-IRB) and received ethical clearance No. CMHS/IRB/183/2025. The study involved a retrospective review of existing medical records and used de-identified data, ensuring that no personal identifiers were collected. Data were collected using electronic tools, and all records are stored in an encrypted database accessible only to authorized personnel. Results are presented in aggregate form, and no identifying information is included in reports or publications. Results After patient selection process (Fig. 1), the final analytical sample comprised 726 participants from 277 health facilities. Flow diagram of patient selection for the study of second-line ART outcomes in Rwanda, 2019–2023. *Number of patients initiated on second-line ART during the study period was initially identified from aggregate report from Rwanda Health Management Information System (RHMIS). ART: Antiretroviral Therapy; VL: Viral Load Participant characteristics The socio-demographic, behavioral, and clinical characteristics of participants are summarized in Table 1. Overall, 645 of 726 participants (88.8%) had achieved viral load suppression. Most participants were female (63.9%), with a median age of 41 years (IQR: 26–49); the 35–49 year age group predominated (38.7%). The majority had been on ART for 10 years or more (71.7%) and were receiving PI-based second-line regimens (78.7%), of whom 539 were on ATV/r-based and 32 on LPV/r-based regimens, and only 155 patients (21.3%) on DTG-based regimens. Good ART adherence was documented in 80.0% of participants at their latest clinic visit. Opportunistic infections were documented in 46 participants (6.4%), and at least one NCD comorbidity was recorded in 16 participants (2.2%). Table 1Socio-demographic, behavioral, and clinical characteristics of PLHIV on Second-Line ART in Rwanda, 2019–2023 (N = 726)CharacteristicFrequencyPercent (%) Age group Under 15395.4 15–2413318.3 25–349212.7 35–4928138.7 50+18124.9 Sex Female46463.9 Male26236.1 Marital status Single21831.1 Married/Cohabiting34449 Divorced/Separated628.8 Widower7811.1 Employment status Employed53275.4 Unemployed8812.5 Student8612.2 Province of residence Eastern20528.2 Kigali City12016.5 Northern9813.5 Southern13518.6 Western16823.1 ART adherence Good58180 Poor/Moderate14520 Alcohol or illicit drug use No62986.6 Yes9713.4 Initial (1st-line) regimen TDF-based30241.6 AZT-based17724.4 ABC-based13718.9 d4T-based8411.6 Other/Unknown263.6 Current (2nd-line) regimen DTG-based15521.3 ATV/r-based53974.2 LPV/r-based324.4 Duration on ART Less than 5 years60.8 5–9 years13218.3 10 years and above58580.9 NCD comorbidity No71097.8 Yes162.2 Opportunistic infections No67893.6 Yes466.4 Period of VL measurement a COVID period (2020–2022)111.5 Post-COVID (> 2022)71598.5 VL at ART line transition (copies/mL) 1,001–9,99912216.8 ≥ 10,00060483.2 VL Suppression status Viral load suppression64588.8 Virological failure8111.2aAll the latest VL measurements were done from 2020 and beyondART: Antiretroviral Therapy; DTG: Dolutegravir; ATV/r: Atazanavir/ritonavir; LPV/r: Lopinavir/ritonavir; TDF: Tenofovir; AZT: Zidovudine; ABC: Abacavir; d4T: Stavudine; NCD: Non-communicable disease; VL: Viral Load Socio-demographic, behavioral, and clinical characteristics of PLHIV on Second-Line ART in Rwanda, 2019–2023 (N = 726) aAll the latest VL measurements were done from 2020 and beyond ART: Antiretroviral Therapy; DTG: Dolutegravir; ATV/r: Atazanavir/ritonavir; LPV/r: Lopinavir/ritonavir; TDF: Tenofovir; AZT: Zidovudine; ABC: Abacavir; d4T: Stavudine; NCD: Non-communicable disease; VL: Viral Load Bivariate analysis of variables associated with viral load suppression In the bivariate analysis (Table 2), VLS was significantly associated with age group (p = 0.004), sex (p = 0.001), marital status (p = 0183), province of residence (p = 0.029), employment status (p < 0.001), ART adherence (p < 0.001), current alcohol or illicit drug use (p = 0.008), NCD comorbidity (p < 0.001), and period of VL measurement (p = 0.028). VLS was highest among participants aged 35–49 years (92.2%) and lowest among those aged 15–24 years (80.5%). Females had higher VLS than males (91.8% vs. 83.6%). The Southern Province had the highest VLS (94.1%) and the Eastern Province the lowest (83.4%), while students recorded the lowest VLS among employment status group (75.6%). Participants with good ART adherence had markedly higher VLS than those with poor or moderate adherence (92.9% vs. 72.4%), and those with current alcohol or illicit drug use had lower VLS than non-users (80.4% vs. 90.1%). Among those with NCD comorbidity, VLS was substantially lower compared with those without (56.2% vs. 89.6%), and VLS was considerably lower during the COVID period (2020–2022) than in the post-COVID period (63.6% vs. 89.2%); however, only 11 participants had their latest VL measurement within this period. Table 2Association between participant characteristics and viral load suppression statusCharacteristicViral Load Suppression (VL ≤ 1000 copies/mL)p-valueSuppressed n (%)Not Suppressed n (%) Age group 0.004 Under 1532 (82.1%)7 (17.9%) 15–24107 (80.5%)26 (19.5%) 25–3482 (89.1%)10 (10.9%) 35–49259 (92.2%)22 (7.8%) 50+165 (91.2%)16 (8.8%) Sex 0.001 Female426 (91.8%)38 (8.2%) Male219 (83.6%)43 (16.4%) Marital status 0.183 Single188 (86.2%)30 (13.8%) Married/Cohabiting309 (89.8%)35 (10.2%) Divorced/Separated56 (90.3%)6 (9.7%) Widower74 (94.9%)4 (5.1%) Employment status < 0.001 Employed485 (91.2%)47 (8.8%) Unemployed76 (86.4%)12 (13.6%) Student65 (75.6%)21 (24.4%) Province of residence 0.029 Eastern171 (83.4%)34 (16.6%) Kigali City107 (89.2%)13 (10.8%) Northern90 (91.8%)8 (8.2%) Southern127 (94.1%)8 (5.9%) Western150 (89.3%)18 (10.7%) ART Adherence < 0.001 Good540 (92.9%)41 (7.1%) Poor/Moderate105 (72.4%)40 (27.6%) Alcohol or illicit drug use 0.008 No567 (90.1%)62 (9.9%) Yes78 (80.4%)19 (19.6%) Initial (1st-line) regimen 0.754 TDF-based266 (88.1%)36 (11.9%) AZT-based156 (88.1%)21 (11.9%) ABC-based122 (89.1%)15 (10.9%) d4T-based76 (90.5%)8 (9.5%) Other/Unknown25 (96.2%)1 (3.8%) Current (2nd-line) regimen 0.662 DTG-based137 (88.4%)18 (11.6%) ATV/r-based478 (88.7%)61 (11.3%) LPV/r-based30 (93.8%)2 (6.2%) Duration on ART 0.222 Less than 5 years4 (66.7%)2 (33.3%) 5–9 years118 (89.4%)14 (10.6%) 10 years and above520 (88.9%)65 (11.1%) NCD comorbidity < 0.001 No636 (89.6%)74 (10.4%) Yes9 (56.2%)7 (43.8%) Opportunistic infections 0.255 No605 (89.2%)73 (10.8%) Yes38 (82.6%)8 (17.4%) Period of VL measurement 0.028 COVID period (2020–2022)7 (63.6%)4 (36.4%) Post-COVID (> 2022)638 (89.2%)77 (10.8%) VL at ART line transition (copies/mL) 0.779 1,001–9,999107 (87.7%)15 (12.3%) ≥ 10,000538 (89.1%)66 (10.9%)p-values in bold indicate statistically significant associations (p < 0.20)ART: Antiretroviral Therapy; NCD: Non-Communicable Disease; VL: Viral Load; DTG: Dolutegravir; ATV/r: Atazanavir/ritonavir; LPV/r: Lopinavir/ritonavir; TDF: Tenofovir; AZT: Zidovudine; ABC: Abacavir; d4T: Stavudine Association between participant characteristics and viral load suppression status p-values in bold indicate statistically significant associations (p < 0.20) ART: Antiretroviral Therapy; NCD: Non-Communicable Disease; VL: Viral Load; DTG: Dolutegravir; ATV/r: Atazanavir/ritonavir; LPV/r: Lopinavir/ritonavir; TDF: Tenofovir; AZT: Zidovudine; ABC: Abacavir; d4T: Stavudine Independent predictors of viral load suppression among patients on second-line ART After adjustment for potential confounders, seven factors were independently associated with VLS in the multivariate logistic regression model (Table 3). Compared with the 35–49 year group, participants aged 15–24 years had lower odds of achieving VLS (aOR = 0.34; 95% CI: 0.13–0.92; p = 0.034). Students had 0.18 times the odds of VLS compared with employed participants (aOR = 0.18; 95% CI: 0.06–0.51; p = 0.002). Compared with Eastern Province, participants from Kigali City (aOR = 2.47; 95% CI: 1.07–6.26; p = 0.043), Northern (aOR = 3.57; 95% CI: 1.37–10.95; p = 0.015), and Southern Province (aOR = 3.84; 95% CI: 1.58–10.53; p = 0.005) had higher odds of VLS. PLHIV with poor or moderate ART adherence had 0.25 times the odds of VLS compared with those with good adherence (aOR = 0.25; 95% CI: 0.14–0.44; p < 0.001), and those with alcohol or illicit drug use had lower odds compared with non-users (aOR = 0.44; 95% CI: 0.22–0.94; p = 0.029). PLHIV with NCD comorbidity had lower odds of VLS compared with those without NCD (aOR = 0.22; 95% CI: 0.06–0.80; p = 0.019). PLHIV whose latest viral load measurement was recorded during the COVID period (≤ 2022) had lower odds of achieving VLS compared with those measured in the post-COVID period (aOR = 0.21; 95% CI: 0.04–1.06; p = 0.049). Table 3Multivariate logistic regression analysis of predictors of viral load suppression among people living with HIV initiated on second-line ART from 2019 to 2023FactorsCrudeAdjustedcOR (95% CI)p-valueaOR (95% CI)p-value Age Group 35–49RefRef Under 150.39 (0.16–1.05)0.0451.14 (0.25–5.92)0.871 15–24 0.35 (0.19–0.64) < 0.001 0.34 (0.13–0.92) 0.034 25–340.70 (0.32–1.59)0.3680.55 (0.22–1.45)0.210 50+0.88 (0.45–1.74)0.7001.09 (0.49–2.48)0.838 Sex FemaleRefRef Male0.45 (0.28–0.72)< 0.0010.78 (0.43–1.40)0.400 Marital Status SingleRefRef Married/Cohabiting1.41 (0.83–2.37)0.1970.39 (0.14–1.02)0.062 Divorced/Separated1.49 (0.63–4.13)0.3990.36 (0.10–1.35)0.122 Widower2.95 (1.12–10.19)0.0491.14 (0.28–5.59)0.865 Employment status EmployedRefRef Unemployed0.61 (0.32–1.26)0.1580.51 (0.22–1.23)0.123 Student 0.30 (0.17–0.54) < 0.001 0.18 (0.06–0.51) 0.002 Province of residence EasternRefRef Kigali City 1.64 (0.84–3.35) 0.158 2.47 (1.07–6.26) 0.043 Northern 2.24 (1.04–5.38) 0.052 3.57 (1.37–10.95) 0.015 Southern 3.16 (1.48–7.54) 0.005 3.84 (1.58–10.53) 0.005 Western1.66 (0.91–3.11)0.1061.62 (0.81–3.33)0.180 ART adherence GoodRefRef Poor/Moderate 0.20 (0.12–0.32) < 0.001 0.25 (0.14–0.44) < 0.001 Alcohol or illicit drug use NoRefRef Yes 0.45 (0.26–0.81) 0.006 0.44 (0.22–0.94) 0.029 NCD comorbidity NoRefRef Yes 0.15 (0.05–0.43) < 0.001 0.22 (0.06–0.80) 0.019 Period of VL Measurement Post-COVID (> 2022)RefRef COVID period (≤ 2022) 0.21 (0.06–0.82) 0.015 0.21 (0.04–1.06) 0.049 Bold rows indicate statistically significant associations (adjusted p < 0.05)ART: Antiretroviral Therapy; CI: Confidence Interval; cOR: Crude Odds Ratio; aOR: Adjusted Odds Ratio; NCD: Non-Communicable Disease; Ref: Reference category Multivariate logistic regression analysis of predictors of viral load suppression among people living with HIV initiated on second-line ART from 2019 to 2023 Bold rows indicate statistically significant associations (adjusted p < 0.05) ART: Antiretroviral Therapy; CI: Confidence Interval; cOR: Crude Odds Ratio; aOR: Adjusted Odds Ratio; NCD: Non-Communicable Disease; Ref: Reference category Discussion This study assessed viral load suppression (VLS) and its independent predictors among 726 PLHIV on second-line antiretroviral therapy (ART) in Rwanda between 2019 and 2023, identifying an overall VLS proportion of 88.8%. Multivariate logistic regression identified seven factors independently associated with suppression: age group, employment status, province of residence, ART adherence, alcohol or illicit drug use, NCD comorbidity, and the calendar period of the most recent viral load (VL) measurement. These findings extend the evidence base on second-line ART outcomes in Rwanda and sub-Saharan Africa and provide actionable insights for national programme improvement. The observed VLS of 88.8% exceeds the global average of approximately 81.2% reported in a recent systematic review of ART outcomes in low- and middle-income countries [8] and falls within the 85–90% range documented for second-line ART in sub-Saharan Africa [9]. However, it remains below the UNAIDS 95% VLS target [21] and Rwanda’s national goal of achieving 95% suppression across all treatment lines and population groups [19], indicating that important programmatic gaps persist. Also this proportion represents apparent progress relative to the 17% failure rate (approximately 83% suppression) reported in the 2016 Rwandan national study [12], however these two studies differ in their patient enrolment periods (2004–2016 versus 2019–2023), the antiretroviral regimens available, and the approaches used to measure outcomes. Participants aged 15–24 years had significantly lower odds of achieving VLS compared with the 35–49 year reference group, consistent with findings from Uganda, Ethiopia, and other sub-Saharan African settings [10, 15, 22]. This pattern likely reflects the multifaceted adherence challenges faced by adolescents and young adults, including social stigma, fear of HIV status disclosure, peer pressure, and the psychosocial burden of managing a chronic condition during critical developmental years, all of which can disrupt treatment engagement [23]. Beyond age group, students had markedly lower odds of suppression compared with other participants. This finding can reflect structural barriers specific to this group, including clinic schedules that conflict with academic timetables, heightened concerns about privacy among peers, and limited awareness of the clinical consequences of treatment interruption. As perinatally infected adolescents continue to transition into adult HIV care, there is a need to strengthen youth-friendly differentiated service delivery, peer support networks, extended and flexible clinic hours, and targeted adherence counselling tailored to the social realities of young people living with HIV. Participants residing in Kigali City, Northern Province, and Southern Province had significantly higher odds of VLS compared with those in Eastern Province, while Western Province did not differ significantly. Regional disparities in HIV treatment outcomes are well documented across sub-Saharan Africa and are frequently attributed to differences in health system infrastructure, human resource capacity, community-based support mechanisms, and population mobility [24–28]. The lower suppression observed in Eastern Province may be partly explained by the mobility of populations living in cross-border areas with Uganda and Tanzania: a study on the HIV care continuum in East African border areas found that non-resident individuals were significantly less likely to achieve VLS, and Eastern Rwanda’s shared borders create conditions where consistent care continuity is particularly challenging to sustain [29]. To address these inequities, targeted interventions such as on-site mentorship, mobile adherence support teams, and continuous quality improvement (CQI) activities should be prioritized in underperforming regions. Poor or moderate ART adherence was among the strongest independent predictors of virological failure, with PLHIV with suboptimal adherence having substantially lower odds of VLS compared with those maintaining good adherence. This finding is consistent with evidence from Kenya and Ethiopia and reflects the well-established biological mechanism by which inadequate drug concentrations resulting from missed doses permit ongoing viral replication, promote the selection of drug-resistant mutations, and progressively narrow future treatment options [22, 30, 31]. In the context of second-line ART, where regimen choices are already more limited than at first line and progression to third-line therapy is clinically complex and costly, the consequences of adherence failure are particularly serious. Strengthening adherence support in Rwanda’s second-line programme through individualized counselling, peer mentorship, psychosocial support addressing barriers such as stigma and treatment fatigue, and systematic identification of patients at risk of non-adherence remains a central priority for sustained VLS. Current alcohol or illicit drug use was independently associated with lower odds of VLS, corroborating findings from South Africa and Uganda [32, 33]. Substance use can undermine ART adherence through multiple pathways, including disruption of daily pill-taking routines, impaired treatment-seeking behavior, co-occurring depression, and disengagement from HIV care services [32], and harmful alcohol use is associated with higher odds of missed ART doses [33], leading to suboptimal VLS. The measurement limitations of the present study should be noted; we did not quantify the amount or frequency of alcohol consumed, and did not document the specific illicit substances involved, precluding any dose-response analysis or substance-specific interpretation. However, these findings underscore the importance of strengthening mental health and substance use screening within routine HIV care. Integrating standardized alcohol and drug use assessments, coupled with systematic referral pathways to addiction counseling or rehabilitation services, should become standard practice to enhance ART outcomes, like VLS, and reduce the risk of treatment failure. Non-communicable disease comorbidity was independently associated with lower odds of VLS, a finding with particular relevance given that approximately 35% of PLHIV on ART in Rwanda are now aged 50 years or older [18]. This finding can be explained by concurrent NCD treatment increases overall pill burden, which may reduce adherence to ART. However, note that only 16 participants (2.2%) had a documented NCD diagnosis, a proportion substantially lower than reported prevalence of NCD among PLHIV in sub-Saharan Africa and similar settings [34–36]. This low proportion likely reflects underreporting in clinical records, rather than a true absence of NCD comorbidity, given that NCD services are typically delivered through vertical, condition-specific programs that operate independently of HIV care. Additionally, the small sample size reduces the precision of the estimates. Nonetheless, it reinforces the clinical rationale for integrated HIV-NCD care and routine comorbidity screening as a standard component of ART management in Rwanda. Another independent predictor of VLS in our study was the calendar period of the most recent VL measurement with participants whose latest VL was recorded during the COVID-19 period (2020–2022) having lower odds of suppression compared with those measured post-2022. This is consistent with documented disruptions to HIV service delivery across sub-Saharan Africa during the pandemic, including reduced clinic attendance, interrupted VL monitoring, ART supply chain disruptions, and diminished adherence counselling capacity [37–39]. While Rwanda maintained core HIV services throughout the pandemic, lockdown and social distancing measures likely curtailed adherence counselling and disrupted peer support networks, which may have contributed to suboptimal VLS during 2020–2022 even without major supply-side failures. This finding must nonetheless be interpreted with caution given that only 11 participants had their most recent VL recorded during the COVID period, most having accumulated a subsequent measurement post-2022, producing a highly imprecise estimate that limits the ability of this study to characterize the true magnitude of COVID-19’s impact on VLS in Rwanda. In this study, neither the type of first-line nor the second-line regimen was significantly associated with VLS. This contrasts with studies from Uganda and South Africa, where regimen type influenced treatment outcomes [10, 40]. For example, a retrospective study in Eastern Uganda found that clients who had initiated AZT-based regimens had lower odds of treatment failure compared to those initiated on TDF [10]. Similarly, in Johannesburg, South Africa, patients on second-line PI-based regimens had significantly lower odds of VLS compared to those on first-line DTG-based regimens [40]. In Rwanda, this null finding may reflect the strength and uniformity of national HIV treatment protocols, which minimize variation in outcomes attributable to regimen type, and the consistent clinical support available across facilities. It is important to emphasize, however, that the DTG-based subgroup comprised only 155 participants (21.3% of the cohort), limiting statistical power to detect clinically meaningful differences. The absence of a statistically significant finding should not be construed as evidence of therapeutic equivalence, and prospective studies with larger DTG-exposed cohorts are needed to adequately characterize outcomes on DTG-based second-line regimens in the Rwandan context. This study has the strengths of using nationally representative data from 277 health facilities across all five provinces and 29 administrative districts of Rwanda and including all eligible PLHIV who initiated second-line ART during the study period, which minimize selection bias and enhance the generalizability of findings to PLHIV on second-line regimens in Rwanda and comparable settings; however, it also has some limitations that should be considered when interpreting the findings and developing policy recommendations. As a retrospective cohort study, the design precludes causal inference; all findings represent statistical associations between patient characteristics and VLS status at the time of the most recent VL measurement. Reliance on clinical documentation introduces information bias across several key variables: ART adherence, alcohol or drug use, and NCD comorbidity may all be underreported in patient files, as clinicians may not consistently document behavioral factors or secondary diagnoses, and patients may underreport stigmatized behaviors. The notably low documented prevalences of NCD comorbidity (2.2%) and opportunistic infections (6.4%) likely reflect this limitation rather than true clinical absence. ART adherence was assessed using clinician-documented pill-taking proportion at the most recent visit, a measure known to overestimate true adherence relative to objective methods such as pharmacy refill records, which may introduce misclassification, and is subject to social desirability bias when patient self-report is used. The small subgroup sizes for the COVID period (n = 11) and NCD comorbidity (n = 16) limit the precision of estimates for these variables. Conclusion This nationally representative study found that 88.8% of PLHIV on second-line ART in Rwanda from 2019 to 2023 achieved viral load suppression, exceeding the global second-line average but remaining below the UNAIDS 95% target and Rwanda’s national goal. Major identified independent predictors of VLS were younger age (15–24 years), being a student, province of residence, suboptimal ART adherence, alcohol or illicit drug use, and NCD comorbidity. To achieve optimal VLS among PLHIV on second-line ART will require strengthened youth-friendly differentiated service delivery, integrated HIV-NCD care models, region-specific quality improvement in Eastern Province, and systematic screening for adherence barriers and substance use within Rwanda’s ART programme. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements We acknowledge the Rwanda Biomedical Centre (RBC) for authorizing data collection in health facilities, and the healthcare providers who facilitated data collection. We thank the director generals of the hospitals who permitted data collection. We also acknowledge the University of Rwanda and its lecturers and reviewers for their constructive comments and suggestions. Furthermore, we thank the U.S. Centers for Disease Control and Prevention (CDC) for supporting the data collection process. Abbreviations Abacavir Adjusted odds ratio Antiretroviral therapy Atazanavir Zidovudine Combination antiretroviral therapy Cluster of differentiation 4 (T-lymphocyte count) U. S. Centers for Disease Control and Prevention Confidence interval College of Medicine and Health Sciences Crude odds ratio Continuous quality improvement Stavudine Dolutegravir Field Epidemiology Training Program Human Immunodeficiency Virus Interquartile range Institutional Review Board National Institute of Statistics of Rwanda Opportunistic infection(s) President’s Emergency Plan for AIDS Relief Protease inhibitor People living with HIV Research Electronic Data Capture Rwanda Health Management Information System Sub-Saharan Africa Tenofovir disoproxil fumarate Joint United Nations Programme on HIV/AIDS University of Rwanda University of Rwanda Institutional Review Board Variance inflation factor Viral load Viral load suppression World Health Organization Author contributions JCK: Conceptualization, Methodology, Formal analysis, Writing original draft. SCMC: Writing review and editing. GNR: Writing review and editing. BU: Supervision, Writing review and editing. KAA: Supervision, Writing review and editing. All authors reviewed the manuscript and approved this manuscript. Funding This study was supported by the President’s Emergency Plan for AIDS Relief (PEPFAR) through the U.S. Centers for Disease Control and Prevention (CDC) under the terms of the Field Epidemiology Training Program (FETP). Funding was provided specifically to support data collection activities. The funders had no role in study design, data analysis, interpretation of findings, or the decision to publish. Data availability The main data supporting the findings of this study are contained within the manuscript. Any additional data are available from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate This study adhered to the ethical principles outlined in the Declaration of Helsinki. Ethical approval was obtained from the University of Rwanda Institutional Review Board (UR-IRB), clearance No. CMHS/IRB/183/2025. Data collection in health facilities was authorized by the Rwanda Biomedical Centre (RBC). The study involved retrospective review of de-identified existing medical records; therefore, individual patient consent was waived by the University of Rwanda Institutional Review Board (UR-IRB). Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The main data supporting the findings of this study are contained within the manuscript. Any additional data are available from the corresponding author upon reasonable request.