Predictors of sustained human immunodeficiency virus viral-load suppression before and after the adoption of Treat All policy in Rwanda

Abstract

BACKGROUND

Sustained viral load (VL) suppression is an important indicator of successful treatment among people living with human immunodeficiency virus (HIV).

AIM

To assess trends of different VL outcomes before and after adoption of the Treat All policy among people living with HIV in Rwanda.

METHODS

Between 2014 and 2017, VL suppression [VL suppression (VLS) < 200 copies/mL] was measured among people living with HIV from 28 healthcare facilities in Rwanda. Participant VL was measured at 6 months, 18 months, and 30 months. The unit of analysis was visit-pair, with subjects across four visit-pair categories: (1) Sustained VL suppression (VL < 200 copies/mL at two consecutive visits); (2) Persistent viremia (VL ≥ 200 copies/mL at two consecutive visits); (3) Viral rebound (VL < 200 copies/mL at prior visit only); and (4) Newly suppressed (VL < 200 copies/mL at subsequent visit only). Poisson regression models with generalized estimating equations were used to estimate adjusted incidence risk ratio (aIRR) and 95% confidence intervals (CIs) for factors associated with sustained VLS. To handle missing data, multiple imputations was performed.

RESULTS

A total of 634 participants contributed 973 visit-pairs (295 single pairs and 339 double pairs). The median age was 37 years (interquartile range: 32-43 years). The incidence rates of sustained VLS, persistent viremia, viral rebound, and new suppression were 85.2%, 4.3%, 4.6%, and 5.7%, respectively. Young individuals aged 18-24 years had higher incidence of viral rebound compared to those 25 years or older (14.8% vs 4.3%; P = 0.011). Of the visit-pairs that had sustained VLS during the first two visits (49.8%; n = 485), 56.7% exhibited sustained VLS throughout follow-up. Compared to having no education, having at least primary education was associated with an increased likelihood of sustained VLS (aIRR = 1.09; 95%CI: 1.01-1.17). Those who presented with advanced HIV disease at baseline had a 12% reduced likelihood of sustained VLS (aIRR = 0.88; 95%CI: 0.79-0.99). Achieving sustained VLS did not differ before or after adoption of the Treat All policy. When the analysis was repeated on imputed datasets, similar results were found.

CONCLUSION

Although most people living with HIV have sustained VLS in Rwanda, individuals without formal education, those presenting with advanced HIV, and younger individuals were lagging on multiple outcomes. Interventions tailored to these individuals would improve treatment outcomes to achieve epidemic control.

Key Words: Sustained viral load suppression; Viremia; Viral rebound; Newly suppressed; Treat all policy; Rwanda

Core Tip: There is limited data on longitudinal, sustained viral load suppression, a critical component of human immunodeficiency virus (HIV) care among people living with HIV in Rwanda. This cohort study analyzed 634 participants with 973 visit-pairs between 2017 and 2020. Nearly 85% of the visit-pairs exhibited sustained viral load suppression. The incidence rate of sustained viral load suppression was significantly lower among those with no education and those who presented with advanced HIV disease. Young participants were more likely to present with viral load rebound. Among people living with HIV in Rwanda, being young, having no education, and presenting with advanced HIV disease were associated with poor HIV treatment outcomes.

INTRODUCTION

To control the human immunodeficiency virus (HIV) epidemic, United Nations Programme on HIV/Acquired Immunodeficiency Syndrome has proposed targets to be met by 2025, requiring 95% of people living with HIV to know their status, of those people living with HIV who know their status be linked to care and of those linked to care, 95% of them to be virally suppressed. Rwanda has made significant strides and achieved these benchmarks according to the United Nations Programme on HIV/Acquired Immunodeficiency Syndrome 2023 report[1].

Following the expansion of antiretroviral therapy (ART) and benefits of rapid ART initiation, the World Health Organization recommended rapid ART initiation (same-day or within 7 days of HIV diagnosis)[2]. Rapid ART initiation is associated with reduced time to viral load suppression (VLS) and reduced risk of HIV transmission[3-5]. Despite these benefits, rapid ART initiation, particularly same-day initiation, has been associated with loss to follow-up[6]. Recent data from Rwanda have also shown that same-day ART initiation is associated with a 39% increase in risk of loss to follow-up compared to initiation at 7 days or more[7].

Generally, people living with HIV who are diagnosed early on and take their medication as prescribed attain a comparable life expectancy to those who are not living with HIV[8,9]. To achieve this success, optimal adherence to medication and sustained VLS are paramount. Because the definition of sustained VLS varies from one study to another, there are challenges in summarizing the overall sustained or durable VLS. Jefferson et al[10] defined sustained VLS as having two consecutive suppressed viral load tests (≤ 200 copies/mL) that were at least 90 days apart from one another with no intervening unsuppressed viral load tests or not having any unsuppressed viral load tests after achieving durable VLS. Robertson et al[11] defined sustained VLS as all viral loads ≤ 200 copies/mL in months 13-36 of follow-up. Enns et al[12] defined sustained VLS as having a viral load of ≤ 200 copies/mL at least 12 months prior to the last VLS test, as well as suppressed status on any intervening viral load test results[12]. Menza et al[13], Huang et al[14], and Rosen et al[15] defined durable VLS as two consecutive viral loads < 200 copies/mL[13]. The prevalence of durable VLS reported by Jefferson et al[10], Robertson et al[11], Enns et al[12], Menza et al[13], Huang et al[14], and Rosen et al[15] were 36%, 36.8%, 60.6%,77.3, 87%, and 72.4%, respectively.

A literature review did not identify studies that assessed sustained VLS in Rwanda; however, several studies have documented factors associated with VLS (HIV-RNA < 1000 copies/mL) at specific time points, with the majority showing a VLS greater than 90%[16,17]. Even when a more stringent definition of VLS was used (HIV-RNA < 200 copies/mL), reported VLS in the country was high[18]. While these data are important, they do not provide a trend of sustained VLS, which is the ultimate goal of HIV treatment. A person may be suppressed at a prior clinic visit but not at a subsequent visit and vice versa. Due to the possibility of alternating viral load status, understanding the magnitude of, and factors associated with, sustained VLS in Rwanda is crucial. Since the adoption of the Treat All policy in Rwanda, there has been limited data on the trend of sustained VLS. Whether the adoption of the Treat All policy has impacted sustained VLS is unknown. In this analysis, we assess: (1) The incidence rates of sustained VLS, viremia, viral rebound, and new suppression among people living with HIV on treatment before and after the adoption of the Treat All policy in Rwanda; and (2) The factors associated with sustained VLS among people living with HIV in Rwanda.

MATERIALS AND METHODS

Study design, setting and population

A retrospective cohort study was performed by using data collected from people living with HIV from 28 primary healthcare facilities in Rwanda as previously described[19]. In brief, the cohort consisted of adults 18 years and older living with HIV who initiated ART from medium- (serving ≥ 200 patients) and high- (serving ≥ 350 patients) volume facilities and were followed for 30 months. A total of 976 individuals initiated treatment from these healthcare facilities, of whom 634 (65%) had two or more viral load results documented. Of the 634 individuals with two or more documented viral load results, 297 (47%) initiated treatment before the adoption of the Treat All policy (between January and April 2014) and 332 initiated treatments after the adoption of the Treat All policy (January and April 2017).

Inclusion and exclusion criteria

Adults who were 18 years of age and older living with HIV and attending one of the 28 identified healthcare facilities were eligible for the study. As the outcome of interest was sustained VLS, only those who had two or more viral load test results during the follow-up period were included in the study. Those who only had a single viral load result were excluded from the study.

Sample size estimation

This was a secondary data analysis using data that was primarily intended to assess the association between differentiated HIV care delivery models and low-level viremia in Rwanda as previously described[19]. All participants from this database were evaluated and those who met inclusion criteria for the current analysis were included without a sample size calculation.

Definition of variables

Viral loads were measured at 6 months, 18 months, and 30 months per the Ministry of Health recommendations in Rwanda. Individuals included in this analysis had a minimum of two and maximum of three viral load test results. Visit pairs using these viral load test results were constructed to define the outcome of interest and create a unit of analysis. The main outcome of interest was sustained VLS, defined as achieving a suppressed viral load (< 200 copies/mL) at two consecutive visits. A level of 200 copies/mL was chosen as the threshold per the current definition of VLS from the Rwanda national HIV treatment guidelines[20]. Secondary outcomes included: (1) Persistent viremia: Viral load ≥ 200 copies/mL at two consecutive visits; (2) Viral load rebound: Viral load suppressed at the first visit but unsuppressed at the subsequent visit; and (3) Newly suppressed: Unsuppressed viral load at the first visit but suppressed at the subsequent visit. The main predictor variable was timing of ART initiation categorized as initiating treatment before or after the adoption of the Treat All policy. Additional variables explored in this analysis included age (18-24, 25-34, 35-44, 45-54, ≥ 55 years), sex (male, female), the status of HIV disclosure (disclosed, did not disclose), and the extent of advanced HIV disease based on baseline CD4 count ≤ 200 cells/mm³ or WHO stage III/IV. Disclosure of HIV status was defined as declaring living with HIV to a partner, peer educator, friend, or family member. Facilities assessed self-reported adherence based on 30-day recall. To quantify the magnitude of adherence, the latter was computed as the proportion of pills taken out of the number prescribed within 30 days. Thereafter, the values were dichotomized as optimal if adherence was ≥ 90% or suboptimal if adherence was < 90%.

Viral load quantification

The Abbott RealTime HIV-1 Viral Load Assay was used for the quantitative measurement of HIV-1 viral load in plasma or dried blood spots.

Statistical analysis

Data was presented by frequencies and proportions for categorical variables. Median and interquartile range were reported for continuous variables. Furthermore, categorical participant characteristics were stratified by the number of viral load results available [one viral load result (excluded from the study) vs two or more viral load results (included in the study)] and compared using χ² tests. Comparison of continuous variables was performed using independent t-tests. The unit of analysis was a visit-pair. To assess the effect of timing of ART initiation on sustained VLS, Poisson regression models with generalized estimating equation were used to calculate incidence risk ratios (IRRs) and its associated 95% confidence intervals (CIs). First, bivariate Poisson regression models were fit. Then, all variables explored in the bivariate analysis were included in multivariable models. Because study participants were recruited from multiple healthcare facilities, they were clustered within these healthcare facilities. To account for this clustering, random effects models were used to account for clustering in both bivariate and multivariable analyses. A complete case analysis was initially performed. Then, to account for missing data in the covariates, five multiple imputations were conducted. Multiple imputations were made by assuming data was missing-at-random. Timing of ART initiation, age, gender, education, disclosure of HIV status, advanced HIV disease, and viral load test results were included in the imputation model as predictors. Poisson regression models were used to analyze each imputed dataset, and Rubin’s Rules were employed to generate final estimates[21]. Then, a side-by-side comparison of the results from complete case analysis was performed from the results of imputed data. All associations were presented as an adjusted IRR (aIRR) and 95%CIs. A subgroup analysis involving people who had only three viral load test results was done to assess trajectories of the incidence of sustained VLS. Statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, United States).

RESULTS

Overall

Between 2014 and 2017, a total of 981 people living with HIV who initiated treatment from 28 healthcare facilities in Rwanda were recruited. Of these, 634 (64.6%) had at least two viral load test results and were included in the final analysis. Table 1 compares characteristics of all participants included and excluded from the study. Age, gender, level of education, and baseline status of advanced HIV disease were comparable between those included and those excluded from the study. Compared to those included in the study, those excluded were less likely to self-report adherence ≥ 90% (65.2% vs 74.5%; P = 0.004) or initiate treatment after the adoption of the Treat All policy (37.6% vs 52.8%; P < 0.001) but were more likely to have disclosed their HIV status to others (64.8% vs 58.0%; P = 0.036). A total of 634 participants were included in the study and contributed 973 visit pairs (295 single-pair and 339 double-pair).

Table 1 Comparison of characteristics between included study participants and those excluded from the analysis based on the documented number of viral load results, n (%).

Characteristics	n = 981	Number of viral load results
		One and excluded (n = 348)	≥ two and included (n = 634)	P value
Age (year), mean ± SD	39.2 ± 10.2	38.6 ± 9.6	39.5 ± 10.4	0.173
Age, year
18-24	24 (2.5)	7 (2.0)	17 (2.7)	0.702
25-34	328 (33.6)	121 (34.8)	208 (32.8)	-
35-44	411 (42.1)	152 (43.7)	263 (41.5)	-
45-54	132 (13.5)	42 (12.1)	90 (14.2)	-
≥ 55	82 (8.4)	26 (7.5)	56 (8.8)	-
Gender
Female	649 (66.1)	241 (69.5)	408 (64.5)	0.114
Male	331 (33.7)	106 (30.5)	225 (35.5)	-
Missing	2 (0.2)	1	1	-
Education
No education	432 (44.0)	162 (47.8)	270 (43.8)	0.240
≥ primary	523 (53.3)	177 (52.2)	346 (56.2)	-
Missing	27 (2.7)	9	18	-
HIV status disclosed
No	387 (39.4)	122 (35.2)	265 (42.0)	0.036a
Yes	588 (60.2)	225 (64.8)	591 (58.0)	-
Missing	4 (0.4)	1	3	-
Adherence
< 90%	250 (25.5)	105 (34.8)	145 (25.5)	0.004a
≥ 90%	621 (63.2)	197 (65.2)	424 (74.5)	-
Missing	111 (11.3)	46	65	-
Advance HIV disease
< 200	161 (16.4)	60 (17.2)	101 (15.9)	0.274
≥ 200	660 (67.2)	216 (62.1)	444 (70.0)	-
Missing	161 (16.4)	72 (20.7)	89 (14.0)	-
Timing of ART initiation
Before treat all	514 (52.3)	217 (62.4)	297 (47.2)	< 0.001a
After treat all	463 (47.2)	131 (37.6)	332 (52.8)	-
Missing	5 (0.5)	0	5	-

^aP < 0.05, indicates statistically significant results.

HIV: Human immunodeficiency virus; ART: Antiretroviral therapy.

Prevalence of viral load outcome

The incidence rates of sustained VLS, persistent viremia, viral rebound, and new suppression were 85.2%, 4.3%, 4.6%, and 5.7%, respectively (Table 2). Those who were 45-54 years old, those who had at least a primary education, and those without advanced HIV disease exhibited more sustained VLS compared to those who were 24 and younger, those who had no primary education, and those who had advanced HIV disease, respectively. Compared to not having a formal education, completing at least primary education was associated with a higher incidence of persistent viremia (6.4% vs 2.7%; P = 0.005). Similarly, those who presented with advanced HIV disease had a higher incidence of persistent viremia compared to those who did not have advanced HIV disease (8.9% vs 2.7%; P < 0.001). Moreover, participants who were 24 years old or younger had a significantly higher incidence of viral rebound compared to other age groups (14.8% vs 4.3%; P = 0.011).

Table 2 Incidence of sustained viral load suppression, persistent viremia, viral rebound, and new suppression among people living with human immunodeficiency virus in Rwanda.

Characteristics	Sustained VLS		Persistent viremia		Viral rebound		Newly suppressed
	n/N	%	n/N	%	n/N	%	n/N	%
Overall	829/973	85.2	42/973	4.3	45/973	4.6	57/973	5.9
Age, year
18-24	21/27	77.8	1/27	3.7	4/27	14.8	1/27	3.7
25-34	271/314	86.3	10/314	3.2	15/314	4.8	18/314	5.7
35-44	332/405	82.2	22/332	5.4	21/405	5.2	30/405	7.4
45-54	130/137	94.9	3/137	2.2	3/137	2.2	1/137	0.7
≥ 55	75/90	83.3	6/90	6.7	2/90	2.2	7/90	7.8
Gender
Female	539/622	86.7	21/622	3.4	26/622	4.2	36/622	5.8
Male	288/349	82.5	21/349	6.0	19/349	5.4	21/349	6.0
Education
No education	341/423	80.6	27/423	6.4	24/423	5.7	31/423	7.3
≥ Primary	465/523	88.9	14/523	2.7	19/523	3.6	25/523	4.8
HIV status disclosed
No	348/406	85.7	13/406	3.2	19/406	4.7	26/406	6.4
Yes	475/561	84.7	29/561	5.2	26/561	4.6	31/561	5.5
Adherence
< 90%	175/217	80.6	17/217	7.8	14/217	6.5	11/217	5.1
≥ 90%	562/648	86.7	20/648	3.1	26/648	4.0	40/648	6.2
Advance HIV disease
Yes	122/157	77.7	14/157	8.9	12/157	7.6	9/157	5.7
No	585/671	87.2	18/671	2.7	28/671	4.2	40/671	6.0
Timing of ART initiation
Before treat all	389/451	86.3	24/451	5.3	17/451	3.8	21/451	4.7
After treat all	433/513	84.4	18/513	3.5	27/513	5.3	35/513	6.8

HIV: Human immunodeficiency virus; VLS: Viral load suppression; ART: Antiretroviral therapy.

Predictors of sustained viral load suppression: Modelling analysis

Table 3 presents bivariate and multivariable factors associated with sustained VLS using complete case analysis and imputed datasets. Although the effect sizes under complete case analysis were portrayed by wider 95%CIs due to missing data, the effect size from both complete case and imputed data were similar in magnitude and directionality. In the bivariate analysis, under complete case analysis, education (primary vs no education; IRR = 1.08; 95%CI: 1.01-1.17), adherence (≥ 90% vs < 90%; IRR = 1.09; 95%CI: 1.01-1.17) and advanced HIV disease (No vs Yes; IRR = 0.87; 95%CI: 0.78-0.97) were associated with sustained VLS. In the multivariable analysis, under complete case analysis, the association between education and advanced HIV disease with respect to sustained VLS persisted. When we used imputed datasets, bivariate analysis showed that participants with at least primary education had a 9% increase in the likelihood of attaining sustained VLS compared to having no education. Participants who presented with advanced HIV disease had an 11% reduced likelihood of achieving sustained VLS compared to those who did not present with advanced HIV disease. In the multivariable analysis, education (primary vs no education; aIRR = 1.09; 95%CI: 1.02-1.16) and advanced HIV disease (No vs Yes; aIRR = 0.89; 95%CI: 0.82-0.98) were associated with sustained VLS. Initiating treatment before and after adoption of the Treat All policy was not associated with sustained VLS. Furthermore, we did not find an association between sex (male vs female; aIRR = 0.95; 95%CI: 0.89-1.02) or disclosure of HIV status (aIRR = 1.00; 95%CI: 0.94-1.06) and sustained VLS.

Table 3 Predictors of sustained viral load suppression among people living with human immunodeficiency virus in Rwanda.

Characteristics	Complete case analysis				Imputed datasets
	Bivariate		Multivariable		Bivariate		Multivariable
	IRR (95%CI)	P value	aIRR (95%CI)	P value	IRR (95%CI)	P value	aIRR (95%CI)	P value
Age, year
18-24	Reference	-	Reference	-	Reference	-	Reference	-
25-34	1.10 (0.86-1.40)	0.455	0.99 (0.83-1.19)	0.944	1.10 (0.86-1.40)	0.455	1.09 (0.87-1.35)	0.484
35-44	1.03 (0.81-1.32)	0.792	0.94 (0.78-1.14)	0.540	1.03 (0.81-1.32)	0.792	1.03 (0.83-1.29)	0.776
45-54	1.20 (0.94-1.52)	0.148	1.12 (0.92-1.35)	0.251	1.20 (0.94-1.52)	0.148	1.22 (0.97-1.52)	0.086
≥ 55	1.05 (0.81-1.37)	0.691	0.98 (0.80-1.22)	0.879	1.05 (0.81-1.37)	0.691	1.09 (0.85-1.36)	0.492
Gender
Female	Reference	-	Reference	-	Reference	-	Reference	-
Male	0.95 (0.89-1.02)	0.139	0.96 (0.89-1.03)	0.230	0.95 (0.89-1.02)	0.139	0.95 (0.89-1.02)	0.136
Education
No education	Reference	-	Reference	-	Reference	-	Reference	-
≥ Primary	1.08 (1.01-1.16)	0.012a	1.09 (1.01-1.17)	0.031a	1.09 (1.02-1.16)	0.012a	1.09 (1.02-1.16)	0.012a
HIV status disclosed
No	Reference	-	Reference	-	Reference	-	Reference	-
Yes	0.99 (0.93-1.05)	0.733	1.00 (0.94-1.09)	0.840	0.99 (0.93-1.05)	0.771	1.00 (0.94-1.06)	0.952
Adherence
< 90%	Reference	-	Reference	-	Reference	-	Reference	-
≥ 90%	1.09 (1.00-1.20)	0.050a	1.05 (0.96-1.15)	0.309	1.06 (0.99-1.15)	0.111	1.04 (0.96-1.12)	0.349
Advance HIV disease
No	Reference	-	Reference	-	Reference	-	Reference	-
Yes	0.87 (0.78-0.97)	0.014a	0.88 (0.79-0.99)	0.034a	0.89 (0.81-0.98)	0.020a	0.89 (0.82-0.98)	0.019a
Timing of ART initiation
Before treat all	Reference	-	Reference	-	Reference	-	Reference	-
After treat all	0.98 (0.92-1.04)	0.543	0.97 (0.90-1.05)	0.466	0.97 (0.92-1.04)	0.432	0.97 (0.91-1.04)	0.385

^aP < 0.05, indicates statistically significant results.

HIV: Human immunodeficiency virus; ART: Antiretroviral therapy; IRR: Incidence risk ratio; aIRR: Adjusted incidence risk ratio; CI: Confidence interval.

Subgroup analysis

We performed subgroup analysis among 339 people who had three viral load test results to explore incidence rates of sustained VLS. Of these, 275 (81.2%) had sustained VLS throughout the follow-up period, and only eight (2.4%) remained unsuppressed throughout the follow-up period (Table 4).

Table 4 Gradual viral load outcomes among 339 people living with human immunodeficiency virus who had three viral load results available.

Number of people (%)	6 months	18 months	30 months
275 (81.2)	Suppressed	Suppressed	Suppressed
8 (2.4)	Suppressed	Suppressed	Unsuppressed
4 (1.2)	Suppressed	Unsuppressed	Suppressed
15 (4.4)	Suppressed	Unsuppressed	Suppressed
24 (7.1)	Unsuppressed	Suppressed	Suppressed
3 (0.9)	Unsuppressed	Unsuppressed	Suppressed
2 (0.6)	Unsuppressed	Suppressed	Unsuppressed
8 (2.4)	Unsuppressed	Unsuppressed	Unsuppressed

DISCUSSION

We assessed trends of viral load outcomes among people living with HIV using routinely collected data from 28 healthcare facilities in Rwanda. The incidence rates of sustained VLS, persistent viremia, viral rebound, and new VLS were 85.2%, 4.3%, 4.6%, and 5.7%, respectively. Our results showed a higher percentage of people with sustained VLS compared to a study performed in Uganda that used the same threshold for VLS[15]. Having primary school education or higher and having ≥ 90% adherence was associated with an increased likelihood of achieving sustained VLS and reduced risk of persistent viremia compared to having no education and having < 90% adherence. Presenting with advanced HIV was associated with a reduced likelihood of achieving sustained VLS and an increased risk of persistent viremia compared to not presenting with advanced HIV disease.

Despite increased access to ART, some patients still present with advanced HIV disease, and stigma remains a predominant factor in delayed care-seeking[22]. Advanced HIV disease is associated with increased treatment costs[23] and a higher risk of lethal opportunistic infections, such as cryptococcal meningitis and tuberculosis[24,25]. In this analysis, we further demonstrate that advanced HIV disease reduced the likelihood of attaining sustained VLS, a critical goal of HIV treatment for epidemic control. Greater focus should be given to addressing stigma associated with HIV to reduce the burden of patients presenting with advanced HIV disease. A qualitative study from Rwanda showed that stigma from family and community members towards people living with HIV was a barrier to engagement in care; however, psychosocial support provided coping mechanisms for stigma[26]. A focus group discussion from Rwanda among women living with HIV also identified support groups to address stigma[27]. Collectively, leveraging the in-country stigma reduction interventions along with other data-driven interventions applied elsewhere could reduce the stigma around HIV. In addition to stigma, lack of awareness of early symptoms of HIV could contribute to people presenting late for HIV diagnosis, which may further be related to the low levels of education documented among participants.

Nearly half of the participants in this study had no formal education, which impacted on their ability to achieve sustained VLS and increased their risk of persistent viremia. Having knowledge about HIV, VLS, drug adherence and the benefits associated with being adherent to medications, such as VLS, reduced risk of HIV transmission and opportunistic infections and improved quality of life may motivate treatment adherence[28]. Furthermore, information about treatment benefits may be challenging to individuals without at least a primary education to comprehend[29,30]. Associations between education attainment and poor VLS have been previously reported[31]; still, this analysis found a disproportionate achievement of sustained VLS between those with formal and informal education, even in the era of expanded ART. HIV programs should consider a form of differentiated service delivery that identifies those with informal education and provides structured information about HIV treatment and expectations. Previous studies have demonstrated that the issue of low literacy can be addressed via interventions that provide simple instructions, use visual aids with limited written content, and build skills aimed at taking medication as prescribed[32,33]. The use of video and other related media to assist in the breakdown of complex information has been shown to improve knowledge among people with low literacy levels[34]. Promoting adoption of these proven interventions, along with other supportive systems like family, is critical to improve viral load outcomes.

Participants who were 24 years of age or younger exhibited a disproportionately higher incidence of viral rebound compared to other age groups, demonstrating challenges with drug adherence facing young individuals. Our findings are comparable to previous studies that found that young adults are more likely to experience viral rebound[35,36]. Developmental challenges, along with stigma, alcoholism, and limited disclosures, are some of the potential barriers to optimal adherence among young people[37,38]. Tailored interventions specific to young people are paramount, as viral rebound increases the risk of HIV transmission[39] and is a precursor to treatment failure[40].

We reviewed the literature for studies that used a viral load threshold of 200 copies/mL for two or more consecutive viral load test results to define sustained VLS, similar to our study methods. The review revealed that while Rosen et al[15] showed higher preponderance of durable VLS among women compared to men (77.7% vs 63.1%), Crepaz et al[41] showed fewer women than men having durable VLS (54.8% vs 64.0%). In our analysis, we did not find an association between sex and sustained VLS. Our findings are consistent with other studies[11,42], which did not find statistically significant disparities in sustained/durable VLS between men and women. Although it is intuitive that that disclosure of HIV status is associated with improved adherence[43] and VLS[44,45], we did not identify a statistically significant association between disclosure and sustained VLS.

This study has a few limitations. First, just over one-third of study participants were excluded from this analysis because they had only one viral load test result. Although some characteristics of included vs excluded participants were comparable, there were differences that may be associated with viral load outcomes. For example, those excluded were more likely to have less than 90% adherence and less likely to disclose their HIV status. Their exclusion may have underestimated/overestimated the associations of these characteristics and viral load outcomes and could have introduced selection bias. Second, adherence assessments were based on self-reported information, which is subject to recall bias. Despite the recall bias associated with self-reported information, adherence measurements based on self-reported data have been used extensively, and the association with viral load outcomes has been previously demonstrated[46-48]. Third, nearly 45% of participants included in the study had only two viral load measurements contributing to a single visit-pair, which limited the exploration of long-term viral load outcomes for most patients. In addition, viral load was measured at 6 months, 18 months, and 30 months per the Rwandan HIV treatment guidelines. A more frequent viral load measurement, such as every 6 months, would be more informative to capture the fluctuation of viral load over time; however, the cost of measuring viral load limits more frequent testing. Because this study was not primarily designed to examine factors associated with sustained viral load suppression, other potential factors affecting the healthcare system, such as staffing, healthcare policies, transportation to the clinic, and communication between healthcare providers and the patient, were lacking. The main strength of this study was the use of routinely collected data in an ideal clinic environment in Africa. Our results are also generalizable to other patients attending primary healthcare facilities. Furthermore, the use of multiple viral load test results enabled us to assess trends of viral load outcomes over time, which is more informative than reporting viral load outcomes at a single time point.

CONCLUSION

Although most people living with HIV have sustained VLS in Rwanda, individuals without formal education, those presenting with advanced HIV, and younger individuals were lagging in multiple outcomes. Interventions tailored to these individuals would improve treatment outcomes to achieve epidemic control.