Universal “Test and Treat” for HIV Had Little Effect on Outcomes,but Missed Clinic Visits Threaten Success of Botswana’s National Antiretroviral Treatment Program

Abstract

In 2016, Botswana changed the policy to institute universal “test and treat” (UTT) in people with human immunodeficiency virus or HIV (PWH). It is unclear whether these policy changes have yielded any clinical benefits or harms. We conducted a retrospective cohort analysis of PWH aged ≥18 years to compare patient outcomes in individuals who received antiretroviral treatment (ART) under (1) the introduction of UTT and (2) rapid versus delayed ART start, at two clinics in Gaborone, Botswana, between 2014 and 2020. Multivariate logistic regression and propensity score models were used to control for potential confounding and selection bias. Of the 2008 participants who had a complete plasma HIV RNA at 1 year following ART initiation, 59 (2.9%) experienced virologic failure, and 665 (33.1%) were lost to follow-up (LTFU). Higher LTFU was recorded in UTT than in delayed ART period (43% vs 31%, p < 0.001); the same trend was upheld on further examination treating all LTFU as treatment failure (47% vs 37%, p < 0.001). In adjusted models, neither the UTT policy: odds ratio (OR) 1.91 [95% confidence interval (CI): 0.90, 3.56] nor rapid ART start: OR 1.31 (95% CI: 0.75, 2.34) was associated with viral failure. UTT was not associated with LTFU: OR 1.04, 95% CI (0.75, 1.45). Missed clinic visits were associated with viral failure regardless of policy period: OR 1.17 (95% CI: 1.03, 1.31) or rapid start: OR 1.20 (95% CI: 1.07,1.35). Neither UTT policy nor rapid ART start was associated with any one of the composite unfavorable outcomes. However, missing clinic visits was an independent risk factor for unfavorable outcomes.

Introduction

The human immunodeficiency virus (HIV) is one of the world’s most serious challenges affecting over 39 million people worldwide, of whom 1.3 million were new infections and caused 630,000 deaths in 2022.¹ While there has been scale-up of treatments and prevention strategies, challenges related to poor linkage to care, suboptimal medication adherence, treatment failure, missed clinic appointments, and loss to care remain.^2
–4 Traditionally, in low- and middle-income countries’ health settings, patients diagnosed with HIV were required to undergo pre-treatment evaluation steps including screening for baseline laboratory tests such as CD4 cell counts to assess immune status, if eligible, then extensive counseling for medication adherence was required before antiretroviral treatment (ART) could be started.

Before a change in HIV guidelines in 2016 in Botswana, individuals diagnosed with HIV were not eligible for treatment until a World Health Organization’s (WHO) stage 3 or 4 opportunistic disease was diagnosed or a CD4 threshold of 350 cells/mL or less was met.⁵ Individuals were required to choose an “adherence partner,” someone in their social network intended to enhance personal accountability during treatment. The pair underwent counseling sessions largely related to adherence. While these health care visits and tests were widely considered essential to enhance social support, adherence, and reduce the risk of treatment failure,^6,7 they utilized more resources, but findings for their clinical benefit were inconclusive.^8
–10

In 2016, following consistent evidence from randomized controlled trials (RCTs) showing that initiating ART rapidly following HIV diagnosis reduced the risk of developing serious AIDS-related morbidities and mortality by over 50%,^11
–13 Botswana adopted a national universal “test and treat” (UTT) policy for ART, removing both CD4 cell count criteria for eligibility and prolonged counseling before prescribing ART.^14,15

While these policy changes, particularly removing pre-initiation screenings and extensive pre-ART counseling, likely reduced the burden these tasks created on the health system, which included cost and time, this reduction in workload likely only resulted in decreased work burden for non-prescriber staff. These policy changes, however, did result in a sudden increase in the number of individuals eligible for ART, which may have led to additional work burden for prescribers. It is unknown if this policy change and, therefore, workflow had beneficial or detrimental effects on patient outcomes and other factors affecting treatment success during this policy change.

In patients already receiving ART, missing clinic visits has been identified as an important risk factor for viral failure^16,17 and attrition from care.¹⁸ This outcome is known to be associated with patient and health care system-related factors, including inadequate patient involvement, nonflexible clinic visit schedules, poor provider–patient relationships, and long patient waiting times.¹⁹ It is unclear whether the transition to UTT would potentially impact clinic attendance, given the simultaneous changes of streamlining ART initiation workflow and a significant increase in individuals initiating ART.

After eliminating the CD4 criteria for prescribing ART, the rapid ART start policy was initially piloted at only a limited number of clinics. This allowed for the opportunity to test the impact of both the expanded pool of patients initiating ART overall and the removal of the counseling sessions for a subset of individuals prescribed ART nearly immediately upon diagnosis (“rapid ART starters”).

Our study aimed to compare treatment outcomes between individuals who were treated during the UTT policy period to those who were treated during the CD4 restricted period and those who underwent a rapid ART start to those who underwent a delayed ART start within Botswana’s national ART program. We also aimed to assess the associations between missed clinic visits and treatment outcomes. We postulated that individuals treated during the “test and treat” policy period and rapid ART starters would have better treatment outcomes. We also hypothesized that missed clinic visits would result in higher rates of both viral failure and loss from care, regardless of the policy period or model of care (i.e., rapid ART start or delayed) received.

Methods

This observational study was conducted at two clinics in Botswana’s capital city, Gaborone. Tebelopele Clinic (site 1), located in the center of the capital city, Gaborone, is a nongovernmental organization clinic run by Botswana’s largest nongovernmental provider of voluntary HIV counseling and testing (VCT), providing about 32% of HIV testing in Botswana.^20

–23,27 In 2018, Tebelopele began offering HIV prevention and treatment services. Lesirane Clinic (site 0), located in a peri-urban setting just outside the capital, is a large government-run clinic that offers outpatient HIV services. Both these outpatient facilities had on-site laboratory and pharmacy services.

Data were extracted from the medical charts of all adults aged ≥18 years who tested positive for HIV and started ART at these clinics between 2014 and 2020. Patients who tested positive but had no follow-up information were not included in this analysis. Data collected included demographics and clinical and laboratory data such as timing of HIV diagnosis and ART initiation, baseline CD4 counts, plasma HIV RNA, opportunistic infections at baseline, and the number of missed clinic appointments. When laboratory data were not available in the patient chart, the national electronic medical records were used to collect laboratory data. All data were collected, uploaded into, and managed in Research Electronic Data Capture (REDCap).^24,25

The study was reviewed and approved by the Health Research Committee of the Ministry of Health in Botswana and the Institutional Review Board of the University of Pennsylvania, USA. Consistent with 45 CFR 46.116, a consent waiver was applied for and obtained for this study.

Analyses

In primary analysis, we compared proportions of participants who were lost to follow-up, experienced virologic failure, or experienced both between participants who started during the UTT policy versus the CD4 restricted criteria period. Chi-squared tests and t-tests were used to compare categorical and continuous variables, respectively. In secondary analyses, we compared these same outcomes between those who were rapid starters versus delayed ART starters. Multivariable logistic regression models were used to control for the effect of potential confounding variables.

Sample size estimations

Assuming the viral failure rate during the CD4 restricted ART period to be 10% and defining a 5% difference during the “test and treat” period to be clinically significant, we estimated that we needed a sample size of 1438 participants to have 95% statistical power with a type I error rate of 5%.

Exposure variables

In the primary analysis, our exposure variable of interest was the ART policy period, a binary variable with 0 indicating that the individual initiated ART during the CD4 restricted period (before June 3, 2016) or 1 during the UTT period (after June 3, 2016).

In the secondary analyses, our exposure variable of interest was rapid start. Consistent with WHO guidelines,²⁶ we considered individuals who were prescribed ART within 7 days of their HIV diagnosis to be rapid ART starters, and anyone who was prescribed ART later than 7 days to be a delayed ART starter. Missed clinic visits were defined as any scheduled clinic visit that the participant did not keep. It was captured as both a count of “no-shows” and as a binary variable indicating whether a patient has a history of missing a visit recorded in the patient’s medical chart or not at 1 year since ART initiation.

Study end-points

Our main outcomes of interest were (1) virologic failure defined as never achieving HIV RNA <400 copies/mL and (2) loss to follow-up 1 year after ART initiation. A patient was declared lost to follow-up if they had not presented for a scheduled clinic visit for ≥90 days without any documented reason and despite follow-up attempts by site personnel. A third outcome of interest (unfavorable outcomes) was a composite outcome consisting of patients who either experienced virologic failure, loss to follow-up or both.

Confounding

Potential confounding variables in all analyses included age, biological sex, baseline CD4 cell count, site of ART care, number of missed clinic visits (counts), history of missed clinic visits (binary), time to ART start (weeks), opportunistic infections at baseline, and key population status [which included men who have sex with men (MSM), sexual partners of MSM, sex workers, and sexual partners of sex workers]. In all three analyses, we individually tested each of the variables for associations with both exposure variables (ART period and rapid ART) and outcome variables (viral failure, loss to follow-up, and unfavorable outcomes) using univariate logistic regression analysis. We then adjusted for all variables individually associated with both the exposure variables and outcome variables at p ≤ 0.20 in the model. We then further subjected this model to a stepwise regression to remove highly collinear variables and retained all other variables in the final logistic regression models as confounders. We adjusted for age and sex in all final models for epidemiological reasons, even if either they were not associated in univariate analyses or were removed in the stepwise regression analyses.

Sensitivity analysis

Additionally, we were concerned with the potential effect of selection bias given differential enrolments between policy period groups at the two study sites. We therefore carried out a sensitivity analysis using propensity score matching to control for both confounding and potential selection bias. The propensity score was created by assessing the relation between each of the baseline characteristics, with both “outcomes” being receipt of ART during test and treat and CD4 restricted period and rapid ART versus delayed ART start. We matched on the propensity scores using the 1:2 matching with replacement method because it provided superior covariate balance after matching. We then used a conditional logistic regression model to estimate the average treatment effects on the treated on outcomes of interest (viral failure, loss to follow up, and the composite outcome).

Results

We enrolled a total of 2008 participants, who had complete data on the two outcomes of viral failure and loss to follow up in this analysis. Table 1 shows that there was a slight predominance of women, and the failure rate was less than 3% despite one-third of the cohort meeting the definition of loss to follow-up. Nearly 60% of all participants had undergone rapid ART start, although most of these rapid starts happened before the adoption of the universal test and treat period. Most participants were on first line regimen consisting of a backbone of two nucleoside(tide) reverse transcriptase inhibitors (NRTIs) and an integrase strand transfer inhibitor (INSTI; 94%), a non-nucleoside(tide) reverse transcriptase inhibitor (NNRTI; 5.8%) and fewer were on second line regimens consisting of 2NRTIs and a protease inhibitor(PI) <1%. There was a predominance of INSTI-based first-line regimen among CD4 restricted period participants (99.2% vs 68.6%), while NNRTI-based regimens were more common among UTT initiators. Only 17% of included individuals started ART after the UTT policy change.

Table 1.

The Distribution of Participants Characteristics by ART Policy Period

	Universal test and treat period	CD4 restricted period	p
n	337	1671
Age [mean (SD)]	35.15 (9.92)	35.91 (9.46)	0.181
Sex: Male (%)	124 (36.8)	704 (42.1)	0.079
Site: 1	336 (99.7)	929 (55.6)	<0.001
ART regimens (%):			<0.001
2NRTI + INSTI	232 (68.6)	1658 (99.2)
2NRTI + NNRTI	104 (30.9)	13 (0.8)
2NRTI + PI	1 (0.3)	0 (0.0)
Baseline CD4 count [mean (SD)]	260 (227)	396 (270)	<0.001
Baseline HIV RNA in log₁₀ [mean (SD)]	2.67 (0.32)	2.78 (0.61)	0.002
Missed clinic visits [mean (SD)]	3.27 (1.93)	2.46 (1.63)	<0.001
History of missed clinic visits (%)	299 (88.7)	1295 (77.8)	<0.001
Key population (%)	0 (0.0)	223 (13.3)	N/A
ART start: Rapid (%)	14 (4.2)	1170 (70.0)	<0.001
Opportunistic infection: Yes (%)	35 (10.4)	126 (7.7)	0.130

ART, antiretroviral treatment; INSTI, integrase strand transfer inhibitor; NNRTI, non-nucleoside(tide) reverse transcriptase inhibitor; NRTI, nucleoside(tide) reverse transcriptase inhibitors; SD, standard deviation.

The distribution of participants characteristics and outcomes of viral failure, loss to follow-up, and a composite unfavorable outcome are also shown in Table 2. The odds of viral failure for participants who started during the UTT versus those starting in the CD4 restricted ART period were not significantly different in either unadjusted or adjusted models, odds ratio (OR) 1.45 [95% confidence interval (CI): 0.72, 2.64] or adjusted OR (aOR) 1.91 (95% CI: 0.90: 3.96), respectively, as shown in Table 3.

Table 2.

Showing Associations Between Baseline Characteristics and Outcomes

	Viral failure			Loss to follow-up			Unfavorable outcome
Characteristics	Suppressed	Failed	p	Retained	Lost	p	Favorable	Unfavorable	p
n	1949	59		1343	665		1222	786
Age in years [mean (SD)]	36 (9.6)	34 (8.2)	0.243	36 (9.5)	35 (9.7)	0.088	36 (9.4)	35 (9.7)	0.026
Sex: (%)			0.395			0.371			0.897
Female	1149 (59)	31 (52.5)		799 (59)	381 (57)		720 (59)	460 (58)
Male	800 (41.0)	28 (47.5)		544 (41)	284 (43)		502 (41)	326 (42)
Site (%):			1.000			<0.001			<0.001
1	721 (37%)	22 (37.3)		580	163		560 (46)	183 (33)
2	1228 (63.0)	37 (62.7)		763 (57)	502 (76)		662 (54)	603 (77)
ART regimens			0.708			0.001			0.003
2NRTI + INSTI	1833	57		1283	607		1167	723
2NRTI+ NNRTI	115	2		59	58		54	63
2NRTI + PI	1	0		1	0		1	0
Baseline CD4 count in cells/µL [mean (SD)]	377 (267)	294 (212)	0.017	376 (262)	373 (274)	0.781	381 (267)	365 (265)	0.167
Missed clinic visits in counts [mean (SD)]	2.37 (1.59)	3.24 (2.31)	<0.001	2.08 (1.22)	3.03 (2.09)	<0.001	2.05 (1.19)	2.93 (2.01)	<0.001
Baseline HIV RNA in log₁₀ [mean (SD)]	2.74 (0.55)	3.23 (1.01)	<0.001	2.73 (0.53)	2.80 (0.64)	0.012	2.60 (0.03)	3.00 (0.86)	<0.001
Key population (%)	217 (11.1)	6 (10.2)	0.980	147 (11)	76 (11)	0.812	145 (12)	78 (10)	0.201
Rapid ART starters (%)	1151 (59.1)	33 (55.9)	0.729	809 (60)	375 (56)	0.109	758 (62)	426 (54)	0.001
Opportunistic infection (%)	1557 (8.1)	6 (10.2)	0.731	189 (14.1)	140 (21.2)	<0.001	172 (14.1)	157 (20.1)	0.001
Universal test and treat period (%)	316 (16.3)	13 (22.0)	0.321	96 (7)	67 (10)	0.030	86 (7)	77 (10)	0.034
History of missed clinic visits (%)	1539 (79.4)	52 (88.1)	0.138	941 (70.3)	650 (98.5)	<0.001	846 (69.5)	745 (95.5)	<0.001

Table 3.

Showing the Adjusted and Unadjusted Associations Between Universal Test and Treat and Viral Failure

	Unadjusted analysis	Adjusted analysis
Characteristics	OR (95% CI)	OR (95% CI)
Universal “test and treat” period	1.45 (0.72, 2.64)	1.91 (0.90, 3.96)
Age (years)	0.98 (0.96, 1.01)	0.98 (0.95, 1.01)
Sex: Male	1.30 (0.77, 2.18)	1.26 (0.72, 2.19)
Missed clinic visits	1.23 (1.10, 1.37)	1.17 (1.04, 1.31)
Site: 1	1.00 (0.59, 1.73)	0.56 (0.30, 1.06)
History of missed clinic visits	1.93 (0.93,4.69)	1.46 (0.67, 3.67)
ART regimen: modified first-line	2.61 (0.62,7.48)
Baseline viral load	2.04 (1.58,2.58)	2.13 (1.61, 2.76)
Rapid ART start	0.90 (0.52,1.48)
Baseline CD4 cell count	1.00 (1.00, 1.00)

ART, antiretroviral treatment; CI, confidence interval; OR, odds ratio.

As shown in Table 4, participants who started ART during the UTT period versus during the CD4 restricted period had significantly higher odds of loss to follow-up in unadjusted analyses, an association that was no longer significant after adjustment, OR 1.64 (95% CI: 1.28, 2.08) and aOR 1.04 (95% CI: 0.75: 1.45), respectively.

Table 4.

Associations Between Universal Test and Treat and Loss to Follow-Up

	Unadjusted analysis	Adjusted analysis
Characteristics	OR (95% CI)	OR (95% CI)
Universal “test and treat” period	1.64 (1.28, 2.08)	1.04 (0.75, 1.45)
Age (years)	0.99 (0.98, 1.00)	0.99 (0.98, 1.00)
Sex: Male	1.07 (0.90, 1.31)	1.05 (0.84, 1.30)
Missed clinic visits	1.47 (1.16, 1.30)	1.29 (1.21, 1.38)
Site: 1	2.33 (1.90, 2.88)	1.68 (1.33, 2.14)
History of missed clinic visits	27.4 (15.3, 55.4)	19.1 (10.6, 38.7)
ART regimen: modified first-line	0.91 (0.45,1.72)
Baseline viral load	1.20 (1.03, 1.40)	1.07 (0.90,1.28)
Rapid ART start	0.86 (0.71, 1.04)	1.17 (0.91,1.50)
Baseline CD4 cell count	1.22 (1.04, 1.42)

ART, antiretroviral treatment; CI, confidence interval; OR, odds ratio.

In Table 5, it shows that the UTT period was also significantly associated with the composite unfavorable outcomes in unadjusted OR 1.53 (95% CI: 1.21, 1.94) but not in adjusted analysis; aOR 0.97 (95% CI: 0.72, 1.30). Adjusting for a number of missed clinic visits and history nullified the association between “test and treat” period and unfavorable outcomes, OR 1.07 (95% CI: 0.83,1.39), while both variables remained independently associated with unfavorable outcomes.

Table 5.

Association Between Universal Test and Treat and Unfavorable Outcome

	Unadjusted analysis	Adjusted analysis
Characteristics	OR (95% CI)	OR (95% CI)
Universal “test and treat” period	1.53 (1.21, 1.94)	0.97 (0.72, 1.30)
Age (years)	0.99 (0.98, 1.00)	0.99 (0.98, 1.00)
Sex: Male	0.96 (0.78, 1.18)	0.96 (0.76, 1.20)
Missed clinic visits (counts)	1.46 (1.36,1.55)	1.30 (1.22, 1.39)
Site: 1	2.33 (1.90,2.88)
History of missed clinic visits	27.4 (15.3,55.4)	7.12 (5.00, 10.4)
Rapid ART start	0.90 (0.52,1.48)	0.87 (0.69, 1.09)
ART regimen: modified first-line	0.86 (0.71,1.04)
Baseline viral load	2.04 (1.58,2.58)
Baseline CD4 cell count	1.22 (1.04,1.42)

ART, antiretroviral treatment; CI, confidence interval; OR, odds ratio.

Rapid ART and Outcomes

The odds of viral failure for those who started on a rapid basis were not different from those of participants who started on a delayed ART basis in either unadjusted OR 0.87 (95% CI: 0.52, 1.48) or adjusted aOR 1.27 (95% CI: 0.73, 2.26) analysis. Rapid ART start was not associated with loss to follow-up in unadjusted OR 0.86 (95% CI: 0.71, 1.04) or adjusted analysis, aOR 1.10 (95% CI: 0.88, 1.39). While rapid ART start was associated with a composite outcome of unfavorable outcomes in unadjusted analysis, OR 0.74 (95% CI: 0.62, 0.89), it was not associated in adjusted analysis: aOR 0.83 (95% CI: 0.66, 1.03).

Missed clinic visits

Regardless of policy period or ART start strategy, the counts or history of missed clinic visits were associated with higher odds of viral failure. For instance, in the model assessing the association between policy period and viral failure, the unadjusted OR and aOR of viral failure for one additional missed clinic visit were OR 1.20 (95% CI: 1.08, 1.30) and aOR 1.15 (95% CI: 1.03, 1.27), respectively. In the model assessing policy period and loss to follow-up, the OR of loss to follow-up for both missed appointment counts and history of missed visits were: aOR 1.25 (95% CI: 1.18, 1.31) and aOR 28.9 (95% CI: 17.3, 52.4), respectively. In the policy period versus unfavorable outcomes, the OR of unfavorable outcomes for an additional missed appointment and history of missed appointment was aOR 1.30 (95% CI: 1.22, 1.39) and aOR 7.12 (95% CI: 5.00, 10.4), respectively.

Sensitivity analyses

While different in magnitude, all results except for the association between the “test and treat” period and loss to follow-up, in the sensitivity analyses using inverse propensity score matching were consistent with conclusions from the logistic regression models, as shown in Table 6. In this analysis, in contrast with the main analysis, the “test and treat” period was significantly associated with high odds of loss to follow-up, OR 1.82 (95% CI: 1.24, 2.65). The estimated sensitivity/dispersion parameter Gamma (Γ) value for this analysis was an OR of 1.10 with a lower bound of 1.05.

Table 6.

Results of Sensitivity Analyses Using Propensity Score Matching

		Matched analysis
Model exposure	Outcome	OR (95% CI)
Rapid ART	Viral failure	0.76 (0.44, 1.32)
	Loss to follow-up	0.99 (0.82, 1.19)
Missed clinic visits	Viral failure	1.28 (1.03, 1.59)
Universal test & treat period	Viral failure	1.69 (0.83, 3.47)
Universal test & treat period	Loss to follow-up	1.82 (1.24, 2.65)
Missed clinic visits	Loss to follow-up	1.47 (1.33, 1.63)

ART, antiretroviral treatment.

Discussion

Our study found not only high viral suppression but also high loss to follow-up rates among those who had documented outcomes, consistent with earlier studies in Botswana.^15,27,28 The observed predominance of INSTIs-based regimen among CD4 restricted starters and NNRTI among UTT likely reflects the sequence in which the transition from PIs and NNRTI-based regimens to INSTIs-based first-line regimens was rolled out in this setting, that is, participants on Protease-Inhibitors (PI)-based regimens and the most treatment-experienced patients on NNRTIs were initially prioritized for treatment switch to more convenient and safer INSTIs-based regimens.

We found no difference in viral failure rates by either policy period or rapid ART start. These findings indicate that despite the documented benefit of UTT in clinical trials and other settings,^12,13,29 its adoption in 2016 added little benefit to what was already being done to achieve very high viral suppression rates in Botswana. This is likely due to the progress Botswana had made toward achieving the 95-95-95 goals before the policy change.^21,27,30 While most participants were enrolled during the CD4 restricted period than in the “test and treat” period, these results are unlikely to have resulted from bias due to this differential enrolment as evidenced by the results of the propensity score-matched analysis that adjusted for potential different selection, based on covariates we were able to measure. These findings are inconsistent with studies from other resource-limited settings, such as China, where rapid ART start is associated with reduced viral failure rates.³¹

The “test and treat” policy period was not significantly associated with higher odds of loss to follow-up. These findings are consistent with studies that found no association with loss to follow-up^15,30 or viral suppression rates,^14,29,32 but they are inconsistent with those that found that “test and treat” worsens attrition rates,^3,33,34 or is protective from loss from care.³¹ While this finding is reassuring, in the sensitivity analysis using propensity score, the “test and treat” policy period was significantly associated with higher odds of loss from care. If the propensity score results are valid, these findings raise concerns about the worsening patient attrition from care following test and treat policy adoption. However, this concern should be tempered by our finding that an unadjusted confounder with an association (OR) of 1.10 or greater would abrogate the effect.

We also found no difference in the odds of unfavorable outcomes between participants who started ART in the “test and treat” policy period versus in the CD4 restricted criteria period. The association between “test and treat” policy period and unfavorable outcomes was attenuated by adjusting for the history of missed clinic visits and was completely nullified by the number of missed clinic visits, while both measures of missed clinic visits remained independent risk factors for unfavorable outcomes. This suggests that the heightened risk of unfavorable outcomes in the “test and treat” period may be explained by missed clinic visits.

Regardless of ART initiation strategy or period, the number of and/or history of missed clinic visits remained significant risk factors for viral failure, loss to follow-up, and unfavorable treatment outcomes. Missing a clinic visit should be considered by clinicians as a potential early indicator that a patient is at risk for unfavorable treatment outcomes consistent with prior work in the United States.³⁵ Strategies aimed at the reduction of missed clinic visits in the first 6 months of ART may be an important intervention area of focus to improve outcomes in the Botswana National ART program. The latest findings of higher acceptability rates of long-acting injectable (LAI) ARTs both in sub-Saharan Africa³⁶ and other health settings³⁷ are compatible with these findings and should be tested in Botswana to mitigate missed clinical visits.

Our findings should be interpreted considering several limitations. First, referral data from testing sites to ART initiation clinics were often incomplete during the CD4 restricted ART period, meaning that some participants lost between testing positive and ART initiation would have been excluded from the analysis. By doing so, we may have inherently excluded individuals destined to have virologic failure in the CD4 restricted group had they started ART. This would bias the association between the rapid start group and virologic failure toward the null. Thus, while we have a null finding of the effects of rapid start ART on most outcomes, due to the decrease in time and effort for patients and non-prescriber adherence counsellors overall, our findings favor the rapid ART start policy. Given higher acceptability rates of LAI in sub-Saharan Africa³⁶ and other health settings,³⁷it may be beneficial to introduce them in Botswana to alleviate the burden of clinic visits and enhance adherence. However, these findings may not be generalizable to other sub-Saharan countries, where, unlike in Botswana, viral suppression rates are relatively lower.^38
–40

Due to the elimination of referral for ART start and many pre-initiation visits before ART start, the “test and treat” policy period may have initiated more participants who would have been destined to be lost from care had they started in the CD4 restricted period. This bias would have inflated our effect estimate away from the null, favoring the CD4 restricted period. Our analytic approach eliminated this bias only to the extent that it is correlated with variables included in the propensity score calculation.

We did not have long enough follow-up to determine if the “test and treat” or rapid start ART prevented other adverse clinical outcomes such as opportunistic infections and death. However, given the strong association between these and viral suppression, retention in care, and mortality,^31,41 we think the “test and treat” approach is favorable.

The high rates of loss to follow-up and missed clinic visits reported in our data are worrisome and pose a serious threat to the Botswana National ART program. While other barriers to ART care in this setting have been described,⁴ more research needs to be done in these areas to explore and elucidate the intricate contextual ways through which missed clinic visits occur and affect patient-level outcomes.

These findings support the continued UTT policy that eliminates most pre-ART assessments since it does not appear to compromise patient-level treatment outcomes. Clinicians should consider a missed clinic visit an early indicator of an unfavorable outcome. An evaluation of the reasons for missed clinic visits will be important to determine effective ways to develop interventions aimed at mitigating loss from care. Continued monitoring of the program will be important as it is scaled up.

Footnotes

Acknowledgments

The study acknowledges the immense contributions of all staff at the Ministry of Health, Botswana, especially those at the collaborating research sites and research assistants, whose meticulous attention to detail made this study possible: (1) Ms. Oaitse Nteba (Research Assistant); (2) Mr. Katlego Gammu (Research Assistant); (3) Mr. Malense Keagile Malense (Nurse at Lesirane Clinic); (4) Ms. Onalenna Tshipayagae (Nurse & Matron at Lesirane Clinic).

Authors’ Contributions

L.M. compiled and submitted applications for ethics approval, supervised Research Assistants, designed data collection instruments on Redcap, supervised data collection, cleaned the data, analyzed the data, and drafted the article. R.G. supervised L.M., reviewed the analytic approach, drafted the article, and coordinated all other study-related activities. B.L. reviewed the analysis, contributed to the article draft writing, and continued the qualitative aim of the study. A.M. supervised the extraction of laboratory data from the Integrated Patient Information Management Systems.

Author Disclosure Statement

The authors have no conflicts of interest. R.G. serves on a DMSB for Pfizer Inc. for a drug unrelated to HIV treatment.

Funding Information

This study was funded by the PENN Center for AIDS Research (CFAR-PI Collman). CFAR GRANT NUMBER: 2-P30-AI- 045008-16.

References

HIV and AIDS Epidemic Global Statistics. HIV.gov. Available from: https://www.hiv.gov/hiv-basics/overview/data-and-trends/global-statistics [Last accessed: October 20, 2023 ].

Mantshonyane

, Roy

, Levy

, et al. Participants switching to second-line antiretroviral therapy with susceptible virus display inferior adherence and worse outcomes: An observational analysis. AIDS Patient Care STDS, 2021; 35(12):467–473; doi: 10.1089/apc.2021.0115

Onoya

, Hendrickson

, Sineke

, et al. Attrition in HIV care following HIV diagnosis: A comparison of the pre-UTT and UTT eras in South Africa. J Int AIDS Soc, 2021; 24(2):e25652; doi: 10.1002/jia2.25652

Kebaabetswe

, Manyake

, Kadima

, et al. Barriers and facilitators to linkage to care and ART initiation in the setting of high ART coverage in Botswana. AIDS Care, 2020; 32(6):722–728; doi: 10.1080/09540121.2019.1640843

MEDBOX. Handbook of the Botswana 2016 Integrated HIV clinical care guidelines. Available from: https://www.medbox.org/document/handbook-of-the-botswana-2016-integrated-hiv-clinical-care-guidelines#GO [Last accessed: October 23, 2023 ].

Knight

, Schatz

. Social support for improved ART adherence and retention in care among older people living with HIV in Urban South Africa: A complex balance between disclosure and stigma. Int J Environ Res Public Health, 2022; 19(18):11473; doi: 10.3390/ijerph191811473

Byrd

, Hou

, Bush

, et al.; Patient-centered HIV Care Model Team. Adherence and viral suppression among participants of the patient-centered Human Immunodeficiency Virus (HIV) care model project: A collaboration between community-based pharmacists and HIV clinical providers. Clin Infect Dis, 2020; 70(5):789–797; doi: 10.1093/cid/ciz276

Bogart

, Mosepele

, Phaladze

, et al. A social network analysis of HIV treatment partners and patient viral suppression in Botswana. J Acquir Immune Defic Syndr, 2018; 78(2):183–192; doi: 10.1097/QAI.0000000000001661

Baker

, Gorbach

, de Melo

, et al. The effect of partnership presence and support on HIV viral suppression among serodiscordant partnered and single heterosexual HIV-Positive individuals in Brazil. AIDS Behav, 2021; 25(6):1946–1953; doi: 10.1007/s10461-020-03124-5

10.

Nakalega

, Mukiza

, Debem

, et al. Linkage to intensive adherence counselling among HIV-positive persons on ART with detectable viral load in Gomba district, rural Uganda. AIDS Res Ther, 2021; 18(1):15; doi: 10.1186/s12981-021-00349-9

11.

Danel

, Moh

, Gabillard

, et al. A trial of early antiretrovirals and isoniazid preventive therapy in Africa. New England J Medicine, 2015; 373(9):808–822; doi: 10.1056/NEJMoa1507198

12.

Lundgren

, Babiker

, Gordin

, et al. Initiation of antiretroviral therapy in early asymptomatic HIV infection. New England J Medicine, 2015; 373(9):795–807; doi: 10.1056/NEJMoa1506816

13.

Cohen

, Chen

, McCauley

, et al.; HPTN 052 Study Team. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med, 2011; 365(6):493–505; doi: 10.1056/NEJMoa1105243

14.

Langwenya

, Phillips

, Brittain

, et al. Same-day antiretroviral therapy (ART) initiation in pregnancy is not associated with viral suppression or engagement in care: A cohort study. J Int AIDS Soc, 2018; 21(6):e25133; doi: 10.1002/jia2.25133

15.

Lebelonyane

, Bachanas

, Block

, et al. Rapid antiretroviral therapy initiation in the Botswana Combination Prevention Project: A quasi-experimental before and after study. Lancet HIV, 2020; 7(8):e545–e553; doi: 10.1016/S2352-3018(20)30187-9

16.

Shumetie

, Moges

, Teshome

, et al. Determinants of virological failure among HIV-infected children on first-line antiretroviral therapy in West Gojjam Zone, Amhara Region, Ethiopia. HIV AIDS (Auckl), 2021; 13:1035–1044; doi: 10.2147/HIV.S334067

17.

Zinski

, Westfall

, Gardner

, et al. The contribution of missed clinic visits to disparities in HIV viral load outcomes. Am J Public Health, 2015; 105(10):2068–2075; doi: 10.2105/AJPH.2015.302695

18.

Masoud

. Previous Missed Appointments Predict Likelihood of Trend in Patients With HIV. Infectious Disease Advisor. August 15, 2018. Available from: https://www.infectiousdiseaseadvisor.com/home/topics/hiv-aids/previous-missed-appointments-predict-likelihood-of-trend-in-patients-with-hiv/ [Last accessed: January 8, 2024 ].

19.

Lowane

, Lebese

. Missing appointments by patients on antiretroviral therapy: Professional nurses’ perspective. Curationis, 2022; 45(1):e1–e7; doi: 10.4102/curationis.v45i1.2213

20.

Botswana Integrated HIV Clinical Care Guidelines. Children & AIDS. Available from: https://www.childrenandaids.org/Botswana_Integrated-HIV-Clinical-Care-Guidelines_2016 [Last accessed: October 18, 2023 ].

21.

Botswana HIV/AIDS Impact Survey Report. UNICEF Botswana. Available from: https://www.unicef.org/botswana/documents/botswana-hivaids-impact-survey-report-0 [Last accessed: September 6, 2023 ].

22.

Pappas-DeLuca

, Kraft

, Galavotti

, et al. Entertainment-education radio serial drama and outcomes related to HIV testing in Botswana. AIDS Educ Prev, 2008; 20(6):486–503; doi: 10.1521/aeap.2008.20.6.486

23.

Creek

, Alwano

, Molosiwa

, et al. Botswana’s Tebelopele voluntary HIV counseling and testing network: Use and client risk factors for HIV infection, 2000–2004. J Acquir Immune Defic Syndr, 2006; 43(2):210–218; doi: 10.1097/01.qai.0000230525.71717.5d

24.

Harris

, Taylor

, Thielke

, et al. Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform, 2009; 42(2):377–381; doi: 10.1016/j.jbi.2008.08.010

25.

Harris

, Taylor

, Minor

, et al.; REDCap Consortium. The REDCap consortium: Building an international community of software platform partners. J Biomed Inform, 2019; 95:103208; doi: 10.1016/j.jbi.2019.103208

26.

WHO. Consolidated Guidelines on HIV Prevention, Testing, Treatment, Service Delivery and Monitoring: Recommendations for a Public Health Approach. WHO; 2021.

27.

Gaolathe

, Wirth

, Holme

, et al.; Botswana Combination Prevention Project study team. Botswana’s progress toward achieving the 2020 UNAIDS 90-90-90 antiretroviral therapy and virological suppression goals: A population-based survey. Lancet HIV, 2016; 3(5):e221–e230; doi: 10.1016/S2352-3018(16)00037-0

28.

Bachanas

, Alwano

, Lebelonyane

, et al. Finding, treating and retaining persons with HIV in a high HIV prevalence and high treatment coverage country: Results from the Botswana Combination Prevention Project. PLoS One, 2021; 16(4):e0250211; doi: 10.1371/journal.pone.0250211

29.

Danel

, Moh

, Gabillard

, et al.; TEMPRANO ANRS 12136 Study Group. A trial of early antiretrovirals and isoniazid preventive therapy in Africa. N Engl J Med, 2015; 373(9):808–822; doi: 10.1056/NEJMoa1507198

30.

Essex

, Makhema

, Lockman

. Reaching 90-90-90 in Botswana. Curr Opin HIV AIDS, 2019; 14(6):442–448; doi: 10.1097/COH.0000000000000580

31.

Xia

, Li

, Wu

, et al. Rapid initiation of antiretroviral therapy under the treat-all policy reduces loss to follow-up and virological failure in routine human immunodeficiency virus care settings in China: A retrospective cohort study (2016–2022). AIDS Patient Care STDS, 2024; 38(4):168–176; doi: 10.1089/apc.2024.0045

32.

Glass

, Günthard

, Calmy

, et al. The role of human immunodeficiency virus (HIV) asymptomatic status when starting antiretroviral therapy on adherence and treatment outcomes and implications for test and treat: The Swiss HIV Cohort Study. Clin Infect Dis, 2021; 72(8):1413–1421; doi: 10.1093/cid/ciaa239

33.

Joseph Davey

, Kehoe

, Serrao

, et al. Same-day antiretroviral therapy is associated with increased loss to follow-up in South African public health facilities: A prospective cohort study of patients diagnosed with HIV. J Int AIDS Soc, 2020; 23(6):e25529; doi: 10.1002/jia2.25529

34.

Ahmed

, Demissie

, Worku

, et al. Effectiveness of same-day antiretroviral therapy initiation in retention outcomes among people living with human immunodeficiency virus in Ethiopia: Empirical evidence. BMC Public Health, 2020; 20(1):1802; doi: 10.1186/s12889-020-09887-9

35.

Bonacci

, Frasca

, Swift

, et al. Antiretroviral refill adherence correlates with, but poorly predicts retention in HIV Care. AIDS Behav, 2016; 20(5):1060–1067; doi: 10.1007/s10461-015-1205-9

36.

Kennedy

, Zhao

, Vo

, et al. High acceptability and perceived feasibility of long-acting injectable antiretroviral treatment among people living with HIV who are viremic and health workers in Uganda. AIDS Patient Care STDS, 2023; 37(6):316–322; doi: 10.1089/apc.2023.0017

37.

Moreno

, Izquierdo

, Alejos

, et al.; CoRIS Cohort. Acceptability of long-acting injectable antiretroviral treatment for HIV management: Perspectives of patients and physicians in Spain. AIDS Patient Care STDS, 2024; 38(7):305–314; doi: 10.1089/apc.2024.0093

38.

Lecher

, Fonjungo

, Ellenberger

, et al. HIV viral load monitoring among patients receiving antiretroviral therapy—Eight Sub-Saharan Africa Countries, 2013–2018. MMWR Morb Mortal Wkly Rep, 2021; 70(21):775–778; doi: 10.1558/5/mmwr.mm7021a2

39.

Fonjungo

, Lecher

, Zeh

, et al. Progress in scale up of HIV viral load testing in select sub-Saharan African countries 2016–2018. PLoS One, 2023; 18(3):e0282652; doi: 10.1371/journal.pone.0282652

40.

Gandla

, Nakka

, Khan

, et al. Biological and social predictors of HIV-1 RNA viral suppression in ART treated PWLH in Sub-Saharan Africa. Trop Med Infect Dis, 2025; 10(1):24; doi: 10.3390/tropicalmed10010024

41.

Zhao

, Wu

, McGoogan

, et al. Immediate antiretroviral therapy decreases mortality among patients with high CD4 counts in China: A nationwide, retrospective cohort study. Clin Infect Dis, 2018; 66(5):727–734; doi: 10.1093/cid/cix878