HIV antiretroviral therapy (ART) access is being rapidly and systematically scaled up throughout sub-Saharan Africa according to World Health Organization guidelines.1,2 However, persistent challenges remain in ensuring that patients are retained and engaged as they progress through the intended “cascade of HIV care” consisting of diagnosis, linkage to care, initiation of ART, and achievement of a long-term undetectable viral load (VL).3,4
One potential way to assess the effectiveness of ART delivery is to estimate metrics of “population VL,” a term that represents the aggregate distribution of HIV RNA levels within a population and includes both persons with known diagnoses of HIV and persons who are as-yet undiagnosed.5 Such metrics are an objective measure of success in achieving a full cascade of HIV care. Population VL metrics include the proportion of the population with an undetectable VL as well as the median and geometric mean VL of the population. We previously reported a method for assessing population VL in resource-limited settings using a fingerprick-based blood plasma collection method deployed at a community-wide health campaign and showed that a substantial fraction of a community can be recruited and tested for HIV in a short 1-week period of time5,6 and that population VL metrics can thereby be calculated.
In the Mbarara District of southwestern Uganda, an agrarian region with an HIV prevalence of 8%, ART access has recently expanded through several new activities. First, the aforementioned community health campaign (CHC) we conducted in 2011 recruited residents of a parish of 6300 persons for HIV testing and actively linked HIV-positive persons to care, many of whom were eligible for ART initiation.6 Concomitantly, in early 2011, the Uganda Ministry of Health changed its ART initiation guidelines to a threshold CD4+ cell count <350 instead of CD4+ <250.7 Finally, our group initiated a clinical research study in this region offering ART to persons with CD4 ≥350 who were otherwise ineligible for government-sponsored ART.8
In this environment of multimodal ART scale-up activities, we sought to measure population VL metrics in 2012, one year our first measurement was made in 2011. We sought to determine if population VL was lower during this period of ART expansion, and whether the proportion of the population with an undetectable VL was higher in 2012 compared with 2011.
2011 and 2012 Community Health Screening and Treatment Campaigns
Fingerprick blood was collected for HIV-1 RNA (VL) measurement during 5-day community-wide multidisease screening and treatment health campaigns (CHC) in Kakyerere Parish, Mbarara District, southwestern Uganda (May 16–21, 2011,6 and April 30 to May 2, 2012 [unpublished data]). Both health campaigns received advance ethics and regulatory approval from the institutional review boards of Makerere University School of Medicine, Kampala, Uganda, and the University of California, San Francisco and from the Uganda National Council for Science and Technology. Patients provided informed verbal consent for participation before each health campaign.
Kakyerere Parish is a geographic area whose 2011 population was estimated as 6300 persons.9 As previously described,6 the 2011 health campaign offered HIV testing as follows: after adults completed an anonymous questionnaire ascertaining previous knowledge of HIV diagnosis, point-of-care fingerprick-based HIV testing was performed (Determine, Inverness), followed by confirmatory testing if indicated (STAT-PAK; Chembio Diagnostic Systems; and UniGold RecombiGen; Trinity) according to Ugandan testing guidelines.10 Participants who tested HIV positive were provided with a referral appointment to the health center serving Kakyerere Parish (Bwizibwera Health Center) for intake into HIV care.
In 2012, the health campaign was repeated with similar procedures but was preceded by a baseline census of the parish for more accurate estimation of its population and to better estimate the uptake of the campaign relative to the parish population. Participants underwent digital fingerprint measurement during census enumeration (U Are U 4500 Reader; Digital Persona, Redwood City, CA), and their attendance at the 2012 CHC was recorded by matching a digital fingerprint obtained on arrival at the campaign to those obtained during the baseline census. Participants diagnosed with HIV in 2012 were offered linkage-to-care appointments as in 2011. In addition, in 2012, participants were asked if they had attended the 2011 health campaign.
HIV-1 Plasma RNA and CD4+ T-Cell Count Measurements
Persons testing positive for HIV in 2012 had a plasma HIV-1 RNA level measured through a fingerprick-derived blood sample that was collected at the CHC, processed in a central laboratory and tested using a modified commercial assay (RealTime HIV-1 Viral Load; Abbott, Chicago, IL) reported previously by our group.5 This fingerprick-adapted assay has a lower limit of detection of 486 copies HIV-1 RNA per milliliter due to reduced input volume and incorporation of a dilution factor of 12.14-fold into the final result. Fingerprick VL results correlate well with phlebotomy-derived results, and this modified technique is compatible with the demands of high-throughput health campaigns where phlebotomy is infeasible (due to the time and expertise needed, the requirement for sample storage capability, and biosafety concerns). Plasma HIV RNA levels were categorized as follows: (1) undetectable, (2) detectable with VL <486 to 10,000 copies/mL, (3) VL 10,001–100,000 copies/mL, or (4) VL >100,000 copies/mL. HIV-1 RNA copy numbers below 486 copies/mL were determined from valid raw Ct values by extrapolation against the assay standard curve and multiplied by the appropriate dilution factor to account for the reduced sample input volume. CD4+ T-cell counts were measured using the point-of-care PIMA platform (Alere), yielding results in 20 minutes that correlate well with phlebotomy-derived samples.11
ART Scale-up in Kakyerere Parish, 2011–2012
In early 2011, the Uganda Ministry of Health revised its CD4+ T-cell count threshold for ART initiation from 250 to 350 cells per microliter in concert with World Health Organization guidelines.2,7 This new national recommendation was operationalized at the health center serving Kakyerere Parish (Bwizibwera Health Center) with assistance from a PEPFAR-supported implementing partner (Mulago-Mbarara Joint AIDS Program).
Additionally, in September 2011, our research group initiated a single-arm clinical trial studying ART administration to individuals who had CD4+ counts of ≥350 cells per microliter and were thus above the national guideline threshold. This study (EARLI: early antiretroviral therapy in resource limited settings in patients with high CD4+ cell counts, NCT: 01479634) is based at Bwizibwera Health Center, the clinic serving the community where the 2011 and 2012 health campaigns were conducted. The EARLI study began enrolling high CD4+ count individuals within a 25-km radius in September 2011 and reached a target enrollment of 200 individuals. Therefore, residents of Kakyerere Parish with CD4+ ≥350 were eligible for screening and potentially for ART initiation, further expanding ART access locally. The EARLI Study has received ethical/regulatory approval from Makerere University School of Medicine, University of California, San Francisco, Uganda National Council for Science and Technology, and the Uganda National Drug Authority.
Estimation of Population RNA Metrics
Within each of the 2 CHCs (2011 and 2012), we assessed the distribution of VL values among HIV-positive individuals by computing 3 sets of metrics: (1) the proportion of participants with an undetectable VL [with 95% binomial confidence intervals (CIs)] and the proportion with VL values in 3 higher strata [<486 (detectable) to 10,000 copies/mL; 10,001–100,000 copies/mL; and >100,000 copies/mL]; (2) the median VL with interquartile range (IQR); and (3) the mean log (VL), that is, the geometric mean VL, with 95% CIs. Population VL metrics for 2011 were reported previously.5
We compared the 2011 and 2012 proportions of participants with undetectable VL through χ2 tests, median VLs through rank-sum tests, and geometric mean VLs through 2-sample t tests. Analyses were performed in Stata 12/SE (Stata Corp, College Station, TX).
To examine the potential impact of a change in the sex distribution of sampled adults between 2011 and 2012 on population VL results, we compared the VL distribution (as described above) within the male and female populations of the 2011 and 2012 CHCs.
Finally, to examine whether changes in population VL metrics were related to whether individuals had been connected to HIV care for a longer time (and thus more likely to have been offered ART if eligible), we compared the VL distribution of 2012 health campaign participants who reported attending the health campaign for the first time in 2012 (“2012 first-time testers”) versus persons who also participated in 2011 (“2012 repeat testers”).
CHC Participants in 2011 and 2012
In 2011, a total of 4343 individuals from Kakyerere Parish participated in the health campaign, including 2323 adults aged ≥18 years, of whom 2282 adults underwent HIV testing (Table 1, reproduced from Chamie et al.6). This represented a 72% community participation rate based on Ugandan Bureau of Statistics data. In 2012, 4879 individuals (2687 adults) participated, of whom 4282 were Parish residents (2204 adult residents), achieving an estimated 63% community participation based on the population enumerated (n = 6844) during the census done before the campaign.12 Of the 2204 adult parish residents participating in the 2012 health campaign, 2197 (99.7%) underwent HIV testing.
In both 2011 and 2012, CHC participants were young, with a median age of 18–19 years, and predominantly women (34% men in 2011 and 46% men in 2012; Table 1). Overall, in 2011, 179 of 2282 adults (7.8%) were HIV positive, whereas in 2012, the prevalence was 223 of 2197 (10%). CD4+ T-cell counts were measured in 184 of 223 HIV-positive adults; the median CD4+ count was 423 cells per microliter (IQR: 307–611 cells/μL). The fraction of participants self-reporting their HIV-positive status as a new diagnosis was 46% in 2011 and 47% in 2012. CD4+ counts were similar between persons self-reporting their HIV status as a new diagnosis (n = 85; median CD4+ count 445 cells/μL; IQR: 306–617) compared with a previously known diagnosis (median CD4+ count 424 cells/μL; IQR: 313–609; P = 0.81). VL was successfully determined in 92%–93% of HIV-positive participants in both years (Table 1, reproduced from Jain et al.5 and Chamie et al.6).
Population-Level HIV RNA Metrics
As shown in Table 2, 37% (95% CI: 30 to 45) of adult participants in 2011 had an undetectable VL5 whereas in 2012, this was substantially higher at 53% undetectable (95% CI: 46 to 60; P = 0.02). Examining the distribution of detectable VL values, we found that individuals with VL >100,000 copies/mL comprised 13% (95% CI: 8 to 18) of the sample in 2011, but only 3% of the sample in 2012 (95% CI: 1.4 to 6.8; P = 0.0007). The median VL was 2185 copies/mL (IQR: <486–33,045 copies/mL) in 2011 and was lower in 2012: <486 (IQR: <486–5931 copies/mL; P = 0.0001). Finally the mean log (VL)—that is, the geometric mean VL—was 3.62 log copies/mL in 2011 (95% CI: 3.46 to 3.78 log) and substantially lower in 2012: 3.18 log copies/mL (95% CI: 3.06 to 3.29 log; P < 0.0001; Table 1 and Jain et al.5). Among persons with both VL and CD4+ count measured, median CD4+ T-cell counts did not differ substantially between persons with undetectable VL [n = 97; 419 cells/μL (IQR: 303–617)] compared with persons with detectable viremia [n = 79; 452 cells/μL (IQR: 323–593); P = 0.46].
To examine whether the lower population VL metrics we observed in 2012 might be because of changes predominantly within either men or women alone, we examined the VL distribution within sex for 2011 and 2012 (Table 3). From 2011 to 2012, virological suppression among female participants increased from 40% to 54% (+14%; P = 0.022) and increased to a similar degree among male participants, rising from 37% to 52% (+15%; P = 0.097). Persons with VL >100,000 copies/mL declined from 12% to 4% among women (P = 0.018) and declined from 27% to 2.5% among men (P < 0.001). Geometric mean VL also declined among women from 3.65 log in 2011 (95% CI: 3.47 to 3.82 log) to 3.18 log in 2012 (95% CI: 3.03 to 3.32 log; P = 0.001) and among men [from 3.55 log in 2011 (95% CI: 3.17 to 3.93 log) to 3.18 log in 2012 (95% CI: 2.99 to 3.36 log; P = 0.047); Table 3 and Jain et al.5].
Finally, we examined whether a higher fraction of HIV-infected persons who self-identified as having tested at both the 2011 and 2012 campaigns (termed “2012 repeat testers”) had an undetectable VL in 2012 compared with the group who self-identified as testing in 2012 for the first time (termed “2012 first-time testers”). This was motivated by the idea that repeat testers may have had more time to become eligible for ART, initiate therapy, and achieve lower VLs before their measurement at the 2012 campaign. Of the 208 total adult HIV-positive participants in the 2012 campaign, repeat participation was ascertained in 197. Overall, 115 individuals identified as first-time testers, 82 as repeat testers, 5 reported an unknown status, and in 6 participants this question was not answered. As shown in Table 4, virological suppression among 2012 first-time testers (n = 115/197) was 46% (95% CI: 37 to 56) and among 2012 repeat testers (n = 82/197) was higher at 66% (95% CI: 55 to 76; P = 0.006). Median VL in both groups was <486 copies/mL. Geometric mean VL was lower among 2012 repeat testers compared with 2012 first-time testers (3.05 log vs. 3.25 log; P = 0.083).
We report here a substantial reduction in population VL metrics in a rural southwest Ugandan geographic region from 2011 to 2012, finding that the proportion of residents with an undetectable VL was higher and that the median and geometric mean VL were both lower. To our knowledge, this is the first report from a resource-limited region in sub-Saharan Africa detailing population VL metrics over time during a period of expanding ART scale-up activities.
Our findings suggest a rising level of success in diagnosing and treating patients and of retention through the cascade of care that culminates in ART-mediated virological suppression. The fact that 53% of HIV-positive Ugandan participants in the 2012 health campaign had an undetectable VL means this metric is approaching a level seen in certain resource-rich settings in the United States such as Washington, DC, where 57% of the population was virologically suppressed in 2008.13 However, despite improvements, there remains a substantial proportion (47%) of the HIV-positive study population with measurable viremia, likely representing individuals with as-yet undiagnosed HIV, persons in care but still ineligible for ART by Ugandan guidelines (i.e., persons with CD4 >350), and persons on ART experiencing virological failure. Although we offered ART access to individuals with CD4+ counts >350 cells per microliter living in this geographic unit, the EARLI Study was only in operation for 6 months before the 2012 health campaign; the impact of this expanded ART access may have been too early to ascertain and may be more evident in future assessments.
Our results come at a time of expanding interest within resource-limited settings in the potential impact of expanding ART access on HIV transmission and incident HIV. The largest and most comprehensive study on this question is a report by Tanser et al14 who showed, within a large longitudinal population-based cohort in rural Kwa-Zulu Natal, South Africa, that regions with higher fractions of residents receiving ART had markedly lower HIV incidence rates over time than regions with lower rates of ART uptake. One presumed mechanism for this encouraging and important finding is that individuals' levels of HIV RNA were decreased by expanding ART coverage leading to less transmission. Although this report did not enumerate individuals' VL characteristics over time, South Africa is one of the few locations where VL testing is a part of routine care, so future evidence may emerge strengthening the link between population VL metrics and ART effectiveness.
Our results lend strength to the theoretical mechanistic link between ART, population VL metrics, and HIV transmission—a link that will be tested in upcoming large-scale cluster randomized trials that will compare strategies of universal ART at all CD4+ cell counts (“test and treat”) versus standard CD4-guided therapy.15,16 Many of the upcoming trials will measure individual participant VLs during multiyear observation and will track incident HIV as a primary outcome, offering the potential to causally link ART scale-up, population VL metrics, and HIV transmission.
Interpretation of our findings should be considered in light of our study approach. Because of the anonymous registration of the 2011 health campaign, individual participants could not be linked across the 2 health campaigns. As such, we could not precisely determine the degree to which individual VLs decreased from 2011 to 2012. However, our findings are unlikely to be primarily driven by different individuals attending the 2011 versus 2012 health campaigns since they reached 64%–72% of the Parish population each year. A related issue is that our community sampling was unlikely to be totally random, introducing the possibility of bias if there were systematic differences between persons attending versus not attending the health campaigns. Missing individuals, for example, could have introduced both upward and downward bias in either or both years.17 Nonattendees may have had higher rates of undiagnosed HIV and may have raised VL metrics. Conversely, HIV-positive persons on ART may not have attended if they felt that they would not benefit from the community campaign services. Had they participated, population VL reductions may have been larger. We did assess gender participation and found that both men and women showed similar declines in population VL metrics despite different levels of participation.
In summary, by using a fingerprick-based VL ascertainment performed within the context of high-throughput community-wide health campaigns, we demonstrate in this study that population VL metrics were substantially lower in 2012 than 2011 in a rural Ugandan community undergoing rapid ART scale-up activities. This included a greater fraction of individuals with an undetectable VL and lower VL levels in viremic individuals. Our results suggest that periodic estimations of population VL may provide useful time-updated information about the effectiveness of ongoing ART delivery, particularly during the current era of expanding ART eligibility and delivery in sub-Saharan Africa.
The authors are grateful to the residents of Kakyerere Parish, Mbarara District, Uganda, for their participation in the 2011 and 2012 CHCs that underlie this research study and to the patients participating in the EARLI Study. The authors thank the staff of the Bwizibwera Level-IV Health Center and the Makerere University Joint AIDS Program for their expertise and the Infectious Disease Research Collaboration, Kampala, Uganda, for administrative, logistic, and scientific support. Finally, the authors thank the UCSF-GIVI CFAR Virology Core for diagnostic testing and assay development, and A.M. for feedback and editing.
1. World Health Organization. Consolidated Guidelines on the Use of Antiretroviral Drugs for Treating and Preventing HIV Infection: Recommendations for a Public Health Approach. Geneva, Switzerland: World Health Organization; 2013.
2. World Health Organization. Antiretroviral Therapy for HIV Infection in Adults and Adolescents: Recommendations for a Public Health Approach: 2010 Revision. Geneva, Switzerland: World Health Organization; 2010.
3. Rosen S, Fox MP. Retention in HIV care between testing and treatment in sub-Saharan Africa: a systematic review. PLoS Med. 2011;8:e1001056.
4. Kranzer K, Zeinecker J, Ginsberg P, et al.. Linkage to HIV care and antiretroviral therapy in Cape Town, South Africa. PLoS One. 2010;5:e13801.
5. Jain V, Liegler T, Kabami J, et al.. Assessment of population-based HIV RNA levels in a rural east African setting using a fingerprick-based blood collection method. Clin Infect Dis. 2013;56:598–605.
6. Chamie G, Kwarisiima D, Clark TD, et al.. Leveraging rapid community-based HIV testing campaigns for non-communicable diseases in rural Uganda. PLoS One. 2012;7:e43400.
7. Uganda Ministry of Health. Uganda Integrated National Guidelines on Antiretroviral Therapy, Prevention of Mother to Child Transmission of HIV, and Infant & Young Child Feeding. Kampala, Uganda: Uganda Ministry of Health; 2012.
8. Jain V, Kamya MR, Havlir DV. The EARLI study: early HIV therapy in patients with high CD4 cell counts. 2011. Available at: http://clinicaltrials.gov/show/NCT01479634
. Accessed June 1, 2013.
9. Chamie G, Kwarisiima D, Clark TD, et al.. Uptake of community-based HIV testing during a multi-disease health campaign in rural Uganda. Submitted. 2013.
10. Ugandan Bureau of Statistics. Population Projections 2008–2012. Kampala, Uganda: Ugandan Bureau of Statistics; 2008.
11. Uganda Ministry of Health. Uganda National Policy Guidelines for HIV Voluntary Counseling and Testing. Kampala, Uganda: Uganda Ministry of Health; 2005.
12. Mtapuri-Zinyowera S, Chideme M, Mangwanya D, et al.. Evaluation of the PIMA point-of-care CD4 analyzer in VCT clinics in Zimbabwe. J Acquir Immune Defic Syndr. 2010;55:1–7.
13. Castel AD, Befus M, Willis S, et al.. Use of the community viral load as a population-based biomarker of HIV burden. AIDS. 2012;26:345–353.
14. Tanser F, Barnighausen T, Grapsa E, et al.. High coverage of ART associated with decline in risk of HIV acquisition in rural KwaZulu-Natal, South Africa. Science. 2013;339:966–971.
15. Granich R, Williams B, Montaner J. Fifteen million people on antiretroviral treatment by 2015: treatment as prevention. Curr Opin HIV AIDS. 2013;8:41–49.
16. Granich R, Gupta S, Suthar AB, et al.. Antiretroviral therapy in prevention of HIV and TB: update on current research efforts. Curr HIV Res. 2011;9:446–469.
17. Miller WC, Powers KA, Smith MK, et al.. Community viral load as a measure for assessment of HIV treatment as prevention. Lancet Infect Dis. 2013;13:459–464.