Skip to main content
PMC home page
HHS Author Manuscripts logoLink to HHS Author Manuscripts
. Author manuscript; available in PMC: 2021 Oct 1.
Published in final edited form as: Lancet HIV. 2020 Sep 1;7(10):e699–e710. doi: 10.1016/S2352-3018(20)30199-5

Community-wide HIV testing, linkage case management, and defaulter tracing in Bukoba, Tanzania: pre-intervention and post-intervention, population-based survey evaluation

Claire Steiner 1,*, Duncan MacKellar 1,*, Haddi Jatou Cham 1, Oscar Ernest Rwabiyago 1, Haruka Maruyama 1, Omari Msumi 1, Sherri Pals 1, Rachel Weber 1, Gerald Kundi 1, Johnita Byrd 1, Kokuhumbya Kazaura 1, Caitlin Madevu-Matson 1, Fernando Morales 1,†, Jessica Justman 1, Thomas Rutachunzibwa 1, Anath Rwebembera 1
PMCID: PMC7748378  NIHMSID: NIHMS1651582  PMID: 32888413

Summary

Background

Community randomised trials have had mixed success in implementing combination prevention strategies that diagnose 90% of people living with HIV, initiate and retain on antiretroviral therapy (ART) 90% of those diagnosed, and achieve viral load suppression in 90% of those on ART (90-90-90). The Bukoba Combination Prevention Evaluation (BCPE) aimed to achieve 90-90-90 in Bukoba Municipal Council, Tanzania, by scaling up new HIV testing, linkage, and retention interventions.

Method

We did population-based, cross-sectional surveys before and after our community-wide intervention in Bukoba—a mixed urban and rural council of approximately 150 000 residents located on the western shore of Lake Victoria in Tanzania. BCPE interventions were implemented in 11 government-supported health-care facilities throughout Bukoba from Oct 1, 2014, to March 31, 2017, when national ART-eligibility guidelines expanded from CD4 counts of less than 350 cells per μL (Oct 1, 2014–Dec 31, 2015) and 500 or less cells per μL (Jan 1, 2016–Sept 30, 2016) to any CD4 cell count (test and treat, Oct 1, 2016–March 31, 2017). We used pre-intervention (Nov 4, 2013–Jan 25, 2014) and post-intervention (June 21, 2017–Sept 20, 2017) population-based household surveys to assess population prevalence of undiagnosed HIV infection and ART coverage, and progress towards 90-90-90, among residents aged 18–49 years.

Findings

During the 2·5-year intervention, BCPE did 133 695 HIV tests, diagnosed and linked 3918 people living with HIV to HIV care at 11 Bukoba facilities, and returned to HIV care 604 patients who had stopped care. 4795 and 5067 residents aged 18–49 years participated in pre-intervention and post-intervention surveys. HIV prevalence before and after the intervention was similar: pre-intervention 8·9% (95% CI 7·5–10·4); post-intervention 8·4% (6·9–9·9). Prevalence of undiagnosed HIV infection decreased from 4·7% to 2·0% (prevalence ratio 0·42, 95% CI 0·31–0·57), and current ART use among all people living with HIV increased from 32·2% to 70·9% (2·20, 1·82–2·66) overall, 23·0% to 62·1% among men (2·70, 1·84–3·96), and 16·7% to 64·4% among people aged 18–29 years (3·87, 2·54–5·89). Of 436 and 435 people living with HIV aged 18–49 years who participated in pre-intervention and post-intervention surveys, previous HIV diagnosis increased from 47·4% (41·3–53·4) to 76·2% (71·8–80·6), ART use among diagnosed people living with HIV increased from 68·0% (60·9–75·2) to 93·1% (90·2–96·0), and viral load suppression of those on ART increased from 88·7% (83·6–93·8) to 91·3% (88·6–94·1).

Interpretation

BCPE findings suggest scaling up recommended HIV testing, linkage, and retention interventions can help reduce prevalence of undiagnosed HIV infection, increase ART use among all people living with HIV, and make substantial progress towards achieving 90-90-90 in a relatively short period. BCPE facility-based testing and linkage interventions are undergoing national scale up to help achieve 90-90-90 in Tanzania.

Funding

US Presidents’ Emergency Plan for AIDS Relief.

Introduction

To achieve HIV epidemic control, UNAIDS set targets for countries to diagnose 90% of people living with HIV, initiate and retain on antiretroviral therapy (ART) 90% of those diagnosed, and achieve viral load suppression for 90% of those on ART (90-90-90) by 2020.1 To help optimise the clinical benefit of ART and potential for epidemic control, WHO recommends that ART should be initiated for all people living with HIV without contraindications within 7 days of diagnosis.2,3

In sub-Saharan Africa, Tanzania has an estimated 1·4 million adults living with HIV, of whom only 61% had been diagnosed, 57% were using ART, and 52% were virally suppressed in 2017.4 ART coverage among HIV-infected men (47%) and people aged 15–24 years (46%) in Tanzania is particularly low, attributed primarily to lower access to or uptake of HIV testing services compared with women and older people.4–6 Thus, to substantially reduce the estimated 81 000 annual HIV infections in Tanzania, age and sex disparities must be reduced to help achieve 81% (90% × 90%) or greater ART coverage for all people living with HIV.1,4

Several experimental combination prevention trials in sub-Saharan Africa that aimed to achieve 81% or greater ART coverage among all people living with HIV have had mixed success.7–11 Three of the four studies relied primarily on community-based testing, referral, and universal ART initiation without comprehensive linkage or defaulter tracing services.7–10 No combination prevention studies in Tanzania or elsewhere have evaluated whether implementation of facility-based and community-based testing, combined with comprehensive linkage and defaulter tracing services can achieve 81% or greater ART coverage in a community under non-experimental conditions.

Implemented during a period of expanding national ART policies, the Bukoba Combination Prevention Evaluation (BCPE) aimed to achieve 90-90-90 among all residents living with HIV aged 18–49 years in Bukoba Municipal Council (Bukoba) through a community-wide HIV testing, linkage, and defaulter tracing combination prevention intervention. Here, we describe pre-intervention and post-intervention differences in population prevalence of undiagnosed HIV infection and ART coverage among all people living with HIV, diagnostic-associated and ART-associated sociodemographic disparities, particularly among men and young adults, and progress towards 90-90-90.

Methods

Study design and participants

We did population-based, cross-sectional surveys in Bukoba before and after our community-wide intervention to assess changes in prevalence of undiagnosed HIV infection and progress towards 90-90-90. As the capital of Kagera region on the western shore of Lake Victoria, Bukoba is a mixed urban-rural council with approximately 150 000 residents (figure 1). Excluding military and police facilities, all 11 government-supported health-care facilities in Bukoba participated in BCPE, including eight public facilities (one regional referral hospital, two health centres, five dispensaries), and three faith-based health centres (figure 1).

Figure 1: Bukoba Municipal Council wards and participating health-care facilities.

Figure 1:

Open in a new tab

Estimated population in 2017 approximately 150 000. Councils are designated by the Tanzania National Bureau of Statistics as districts, municipalities, or towns.

HIV testing, linkage, and defaulter-tracing interventions were implemented throughout Bukoba over 2·5 years (Oct 1, 2014–March 31, 2017) when national guidelines for ART eligibility expanded from CD4 counts of less than 350 cells per μL (Oct 1, 2014–Dec 31, 2015) and 500 or less cells per μL (Jan 1, 2016–Sept 30, 2016) to any CD4 cell count (test and treat, Oct 1, 2016–March 31, 2017). Beginning Dec 1, 2015, eligible patients ready to begin treatment were initiated on ART at their first clinic visit (same-day ART). Beginning Nov 1, 2016, same-day ART was also offered to all people living with HIV at the point of diagnosis during community-based testing events.

Institutional review boards of the Government of Tanzania, National Institute for Medical Research, and Columbia University approved BCPE testing, linkage, and defaulter tracing interventions, and all survey methods and data collection activities. The study was also reviewed and approved in accordance with US Centers for Disease Control and Prevention (CDC) human participants research protection procedures. Research staff obtained written informed consent from pre-intervention and post-intervention survey participants. All identified people living with HIV who enrolled and remained in HIV care were provided ART when they became eligible, according to national guidelines.

Procedures

Methods, outcomes, and costs of BCPE HIV testing and linkage interventions have been previously described in detail.12,13 HIV testing was offered to all eligible patients attending outpatient department clinics in the 11 participating facilities. Testing was also offered at all occupied homes at least once and at 79 male-frequented venues throughout Bukoba.12 Peer-delivered, linkage case management was routinely offered to all diagnosed people living with HIV who were referred to participating facilities.13 As part of linkage case management, HIV-positive peer counsellors provided a comprehensive package of services recommended by the CDC and WHO for up to 90 days for all consenting people living with HIV.3,13,14 Defaulter tracing was implemented July 1, 2016–May 31, 2017, for patients who had defaulted from care since Oct 1, 2014. Electronic medical records were queried quarterly to identify patients who had not received care within 90 days. Lay counsellors did defaulter tracing for up to 30 days via phone and home visits. Patients who were contacted and wished to return to care received treatment navigation, expedited care, and same-day ART services. Patients who were not contacted or refused services once were re-traced in a subsequent quarter to ensure all reasonable efforts were made to help patients restart care.

We did identical pre-intervention and post-intervention single-stage cluster sample surveys from Nov 4, 2013, to Jan 25, 2014, and from June 21, to Sept 20, 2017, respectively. Survey sample sizes were estimated based on Fisher’s exact test for expected differences between two independent sample proportions. In Bukoba census enumeration areas randomly selected in proportion to ward population, all consenting household members aged 18–49 years were eligible to participate in a computer-assisted personal interview and test for HIV.

Staff did interviews in Kiswhali or English on sociodemographic characteristics, previous HIV testing and diagnosis, and use of ART. Nurse counsellors did HIV testing in accordance with national guidelines after the interview. Blood specimens from HIV-positive participants were sent to the national laboratory in Dar es Salaam, Tanzania, for viral quantification using the COBAS TaqMan HIV-1 test (Roche Molecular Systems, Branchburg, NJ, USA).

Outcomes

For the intervention, staff recorded testing, linkage, and defaulter tracing outcomes on standard forms. For linkage clients, enrolment in HIV care and ART initiation were measured during the 90-day case management period.13 Staff compiled outcomes monthly, by sex and age group. For the evaluation, classification of previous HIV diagnosis and current ART use among HIV-positive survey participants was based on: interview self-report of these conditions, medical record confirmation, or having achieved viral load suppression (<1000 HIV RNA copies per mL). Undiagnosed HIV infection among HIV-positive participants was defined as not having received a previous diagnosis.

Statistical analysis

We compared characteristics of study participants in pre-intervention and post-intervention surveys using Rao-Scott χ2 tests, incorporating survey weights and adjusting for clustering within enumeration areas. Population prevalence of undiagnosed HIV infection, previous HIV diagnosis, and ART use among people living with HIV before and after the intervention used Taylor series linearisation variance estimation to account for clustering within enumeration areas. Prevalence ratios (PRs) and adjusted prevalence ratios (APRs) were estimated using generalised estimating equations models with a log link and empirical sandwich standard errors to account for clustering within enumeration areas. For both surveys, we divided participants and census data into cells defined by four age groups, sex, and three geographical areas, and computed weights by dividing the percentage of census participants in each cell by the percentage of survey participants. We incorporated weights into all statistical tests and models to address under-coverage of the sample and survey nonresponse related to age, sex, and geographical area. SAS version 9.3 was used for all analyses; pre-intervention and post-intervention population prevalence of HIV infection, undiagnosed HIV infection, and current ART use was estimated with SURVEYFREQ, and intrasurvey and intersurvey PRs and APRs were estimated with GENMOD.

Role of the funding source

Authors from CDC contributed to the study design, selection of settings investigated, strategies assessed, conduct of the analyses, and interpretation of results. ICF supported the salary for JB and had no role in study design, data collection and analysis, or manuscript development. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Results

Between Oct 1, 2014, and March 31, 2017, 133 695 HIV tests were done, 56 304 (42%) among men and 43 247 (32%) among young adults aged 15–24 years. Of 4732 people who had tested HIV-positive and were not in care (4143 newly diagnosed), 1822 (1583 newly diagnosed) were men, and 993 (881 newly diagnosed) were young adults aged 15–24 years. Of 4206 linkage clients who participated during CD4 counts of less than 350 cells per μL (n=2233), 500 or less cells per μL (n=1221), and test-and-treat (n=752) periods, 2018 (90%), 1168 (96%), and 732 (97%) enrolled in HIV care, of whom 1057 (52%), 815 (70%), and 649 (89%) initiated ART within 90 days of diagnosis, respectively. Excluding patients who upon tracing were found to have moved (n=504), transferred HIV care (n=300), or died (n=172), of 1650 patients who defaulted from care, 820 (50%) were contacted and received defaulter tracing services, of whom 604 (74%) returned to care and 573 (70%) initiated or re-initiated ART; the remaining 830 patients were lost to follow-up.

In pre-intervention and post-intervention surveys, 4795 (73%) of 6532 and 5067 (74%) of 6844 enumerated household residents aged 15–49 years were interviewed and tested for HIV, respectively. In both surveys, proportionally fewer men than women participated due to challenges contacting men for enrolment (figure 2). Proportionally more post-intervention than pre-intervention survey participants reported at least some secondary-level education, being employed, owning a mobile phone or television, difficulty meeting household food needs, never using alcohol, engaging in unprotected sexual intercourse in the last 6 months, and testing for HIV in the past 2 years (table 1).

Figure 2: Pre-intervention and post-intervention household survey participation cascades, 2014–17.

Figure 2:

Open in a new tab

Table 1:

Pre-intervention and post-intervention household survey participant characteristics

Pre-intervention survey (November, 2013-January, 2014) Post-intervention survey (June, 2017-September, 2017) p value*
n Unweighted % Weighted† % (95% CI) n Unweighted % Weighted† % (95% CI)
Total 4795 100% 100% 5067 100% 100% ··
Sex
 Male 1834 38.25% 49.04% (47.02–51.07) 2091 41.27% 49.04% (47.27–50.81) ··
 Female 2961 61.75% 50.96% (48.93–52.98) 2976 58.73% 50.96% (49.19–52.73) ··
Age group (years)
 18–29 2749 57.33% 54.39% (52.07–56.71) 2832 55.89% 54.39% (51.94–56.84) ··
 30–39 1385 28.88% 30.40% (28.51–32.30) 1461 28.83% 30.41% (28.51–32.30) ··
 40–49 661 13.79% 15.21% (13.79–16.62) 774 15.28% 15.20% (13.72–16.69) ··
Residence
 Urban 3791 79.06% 75.58% (61.66–89.50) 3724 73.50% 75.58% (60.83–90.33) ··
 Rural 1004 20.94% 24.42% (10.50–38.34) 1343 26.50% 24.42% (9.67–39.17) ··
Duration of current home residence (years) ·· ·· ·· ·· ·· ·· 0.33
 <1 1559 32.51% 31.34% (28.75–33.94) 1519 29.98% 29.66% (27.08–32.23) ··
 1–2 1059 22.09% 22.14% (19.96–24.33) 1226 24.20% 24.43% (22.93–25.93) ··
 >2 2177 45.40% 46.52% (42.74–50.29) 2322 45.83% 45.91% (42.86–48.97) ··
Highest level of education completed‡ ·· ·· ·· ·· ·· ·· 0.0092
 None or some primary 556 11.60% 11.92% (9.11–14.73) 532 10.50% 10.77% (9.10–12.46) ··
 Completed primary 2596 54.17% 54.54% (51.66–57.42) 2546 50.27% 49.55% (46.50–52.59) ··
 At least some secondary 1640 34.22% 33.54% (30.05–37.03) 1987 39.23% 39.68% (35.60–43.75) ··
Employed ·· ·· ·· ·· ·· ·· 0.021
 No 983 20.50% 17.45% (15.05–19.84) 767 15.14% 14.08% (12.16–15.99) ··
 Yes 3812 79.50% 82.55% (80.16–84.95) 4300 84.86% 85.92% (84.01–87.84) ··
Ownership of mobile phone or television ·· ·· ·· ·· ·· ·· 0.0002
 No 364 7.59% 8.04% (5.49–10.59) 209 4.12% 4.42% (3.03–5.81) ··
 Yes 4431 92.41% 91.96% (89.41–94.51) 4858 95.88% 95.58% (94.19–96.97) ··
Trouble satisfying household food needs ·· ·· ·· ·· ·· ·· <0.0001
 Never 2707 56.64% 56.01% (50.49–61.52) 2285 45.34% 45.93% (40.30–51.57) ··
 Seldom 1822 38.13% 38.69% (33.35–44.02) 2118 42.02% 41.20% (36.25–45.15) ··
 Sometimes, often, or always 250 5.23% 5.31% (4.12–6.49) 637 12.64% 12.86% (9.56–16.17) ··
Frequency of alcohol use§ ·· ·· ·· ·· ·· ·· 0.0005
 Never 2971 62.03% 58.97% (55.03–62.90) 3376 66.65% 65.14% (62.04–68.24) ··
 Monthly 985 20.56% 20.78% (18.80–22.77) 1044 20.61% 20.93% (18.55–23.29) ··
 Weekly or daily 834 17.41% 20.25% (17.25–23.26) 645 12.73% 13.93% (11.33–16.54) ··
Sexual behaviour in the past 6 months ·· ·· ·· ·· ·· ·· 0.040
 Protected intercourse 1122 23.40% 22.85% (21.07–24.63) 1067 21.06% 20.83% (19.40–22.27) ··
 Unprotected intercourse 3019 62.96% 63.76% (61.59–65.94) 3380 66.71% 67.25% (65.10–69.40) ··
 No sexual partners 654 13.64% 13.39% (12.16–14.61) 620 12.24% 11.92% (10.24–13.60) ··
Last tested for HIV in past 2 years|| ·· ·· ·· ·· ·· ·· <0.0001
 No 1957 40.82% 41.89% (39.51–44.27) 1091 21.53% 21.46% (19.85–23.07) ··
 Yes 2837 59.18% 58.11% (55.73–60.49) 3976 78.47% 78.54% (76.93–80.15) ··
HIV infected ·· ·· ·· ·· ·· ·· 0.48
 No 4359 90.91% 91.07% (89.63–92.51) 4632 91.42% 91.63% (90.14–93.11) ··
 Yes 436 9.09% 8.93% (7.49–10.37) 435 8.58% 8.37% (6.89–9.86) ··
Open in a new tab
*

Rao-Scott χ2 test for differences in pre-intervention and post-intervention prevalence estimates; not reported for variables used for census-based weights.

†

All estimates were census weighted by sex, age group, and residence, and adjusted for clustering within census enumeration areas.

‡

Pre-intervention, 3 participants did not respond; post-intervention, 2 did not respond.

§

Pre-intervention, 5 participants did not respond; post-intervention, 2 did not respond.

Unprotected indicates condom use for <100% of sexual-intercourse acts; protected indicates condom use for 100% of sexual-intercourse acts.

||

Pre-intervention, 1 participant did not respond.

Overall prevalence of HIV infection was similar between pre-intervention and post-intervention surveys (table 1). In both surveys, prevalence of HIV infection was higher among women than men, and residents aged 40–49 and 30–39 years than residents aged 18–29 years (figure 3). In 2017, post-intervention prevalence of previously diagnosed HIV infection across each demographic subgroup was greater than undiagnosed infection in each subgroup (figure 3).

Figure 3: Pre-intervention (A) and post-intervention (B) prevalence of HIV infection in Bukoba Municipal Council among residents aged 18–49 years, by sex and age group (years).

Figure 3:

Open in a new tab

Residents with HIV infection who had received a previous diagnosis of HIV infection (red) and residents who had not been diagnosed previously and were newly HIV diagnosed at the time of the survey (blue).

By 2017, prevalence of undiagnosed HIV infection decreased by approximately half overall, and at least by half among most subgroups including men and women, urban and rural residents, employed and unemployed, and those who engaged in unprotected sexual intercourse (table 2). Prevalence of undiagnosed infection decreased by approximately two-thirds among residents aged 18–29 years and those who had not completed a primary education.

Table 2:

Pre-intervention and post-intervention population prevalence of undiagnosed HIV infection among residents aged 18–49 years, and intrasurvey and intersurvey undiagnosed infection PRs

Pre-intervention survey (November, 2013-January, 2014) Post-intervention survey (June, 2017-September, 2017) Intersurvey PR (95% CI)
Total (n) Prevalence of undiagnosed HIV infection* Total (n) Prevalence of undiagnosed HIV infection*
n % (95% CI) PR (95% CI) APR† (95% CI) n % (95% CI) PR (95% CI) APR† (95% CI)
Total 4795 222 4.70% (3.76–5.64) ·· ·· 5067 96 1.99% (1.37–2.61) ·· ·· 0.42 (0.31–0.57)
Sex
 Male 1834 72 4.28% (3.15–5.40) 1 (ref) 1 (ref) 2091 36 1.90% (1.13–2.67) 1 (ref) 1 (ref) 0.44 (0.29–0.69)
 Female 2961 150 5.10% (4.04–6.17) 1.19 (0.94–1.51) 1.20 (0.95–1.52) 2976 60 2.08% (1.45–2.71) 1.10 (0.79–1.52) 1.14 (0.82–1.58) 0.41 (0.31–0.54)
Age group (years)
 18–29 2749 110 3.95% (3.01–4.88) 1 (ref) 1 (ref) 2832 39 1.35% (0.93–1.76) 1 (ref) 1 (ref) 0.34 (0.24–0.48)
 30–39 1385 87 6.56% (4.92–8.21) 1.66 (1.36–2.04) 1.67 (1.37–2.05) 1461 37 2.67% (1.48–3.87) 1.99 (1.36–2.91) 2.00 (1.37–2.92) 0.41 (0.26–0.64)
 40–49 661 25 3.66% (2.13–5.20) 0.93 (0.58–1.48) 0.94 (0.59–1.50) 774 20 2.94% (1.52–4.36) 2.18 (1.28–3.73) 2.20 (1.29–3.76) 0.80 (0.44–1.46)
Residence
 Urban 3791 170 4.54% (3.41–5.66) 1 (ref) ·· 3724 70 1.93% (1.20–2.65) 1 (ref) ·· 0.42 (0.30–0.59)
 Rural 1004 52 5.20% (3.64–6.77) 1.15 (0.79–1.67) ·· 1343 26 2.19% (1.02–3.37) 1.14 (0.61–2.14) ·· 0.42 (0.23–0.76)
Duration of current home residence (years)
 ≥2 2177 81 3.90% (2.81–4.99) 1 (ref) 1 (ref) 2322 40 1.89% (1.22–2.57) 1 (ref) 1 (ref) 0.49 (0.33–0.71)
 1–2 1059 56 5.32% (3.73–6.90) 1.36 (0.98–1.91) 1.48 (1.05–2.10) 1226 19 1.45% (0.79–2.11) 0.76 (0.46–1.27) 0.95 (0.57–1.57) 0.27 (0.16–0.45)
 <1 1559 85 5.45% (3.99–6.91) 1.40 (1.01–1.94) 1.65 (1.18–2.30) 1519 37 2.59% (1.55–3.63) 1.37 (0.95–1.98) 1.84 (1.25–2.69) 0.48 (0.31–0.72)
Highest level of education completed‡
 At least some secondary 1640 41 2.63% (1.75–3.51) 1 (ref) 1 (ref) 1987 16 0.81% (0.41–1.21) 1 (ref) 1 (ref) 0.31 (0.18–0.53)
 Completed primary 2596 130 4.92% (3.81–6.03) 1.87 (1.27–2.76) 1.73 (1.16–2.57) 2546 63 2.68% (1.83–3.53) 3.33 (2.12–5.21) 2.88 (1.77–4.70) 0.55 (0.38–0.77)
 None or some primary 556 51 9.53% (6.96–12.10) 3.62 (2.45–5.36) 3.42 (2.29–5.10) 532 17 3.19% (1.53–4.85) 3.95 (2.23–7.02) 3.52 (1.99–6.22) 0.33 (0.20–0.57)
Employed in past 12 months
 No 983 30 3.02% (1.75–4.29) 1 (ref) 1 (ref) 767 10 1.22% (0.49–1.95) 1 (ref) 1 (ref) 0.40 (0.20–0.80)
 Yes 3812 192 5.05% (4.04–6.06) 1.67 (1.11–2.52) 1.65 (1.06–2.56) 4300 86 2.12% (1.42–2.82) 1.74 (0.92–3.29) 1.40 (0.73–2.68) 0.42 (0.30–0.58)
Ownership of television or mobile phone
 Yes 4431 186 4.23% (3.47–4.99) 1 (ref) 1 (ref) 4858 78 1.66% (1.14–2.19) 1 (ref) 1 (ref) 0.39 (0.29–0.53)
 No 364 36 10.08% (6.28–13.88) 2.38 (1.70–3.34) 2.41 (1.72–3.39) 209 18 9.05% (5.54–12.57) 5.44 (3.52–8.40) 5.05 (3.16–8.07) 0.90 (0.53–1.51)
Trouble satisfying food needs in past 12 months§
 Never 2707 97 3.68% (2.79–4.57) 1 (ref) 1 (ref) 2285 27 1.24% (0.80–1.69) 1 (ref) 1 (ref) 0.34 (0.22–0.52)
 Seldom 1822 102 5.65% (4.36–6.95) 1.54 (1.22–1.93) 1.48 (1.17–1.88) 2118 45 2.25% (1.26–3.24) 1.81 (1.01–3.25) 1.70 (0.94–3.08) 0.40 (0.27–0.59)
 Sometimes, often, or always 250 22 8.46% (5.42–11.50) 2.30 (1.63–3.23) 2.25 (1.60–3.16) 637 24 3.91% (2.13–5.68) 3.14 (1.74–5.65) 2.72 (1.44–5.11) 0.46 (0.27–0.80)
Frequency of alcohol use i n past 6 months
 Never 2971 105 3.54% (2.70–4.37) 1 (ref) 1 (ref) 3376 47 1.36% (0.94–1.78) 1 (ref) 1 (ref) 0.38 (0.27–0.54)
 Less than monthly or monthly 985 49 4.85% (3.11–6.58) 1.37 (0.98–1.92) 1.31 (0.93–1.86) 1044 23 2.39% (0.95–3.83) 1.76 (0.95–3.25) 1.62 (0.86–3.07) 0.49 (0.26–0.95)
 Weekly or daily 834 68 7.95% (6.13–9.78) 2.25 (1.69–2.99) 2.31 (1.66–3.23) 645 26 4.35% (2.14–6.55) 3.20 (1.93–5.28) 2.98 (1.71–5.17) 0.55 (0.34–0.89)
Sexual behaviour in past 6 months||
 Protected intercourse 1122 32 3.01% (1.93–4.08) 1 (ref) 1 (ref) 1067 20 1.94% (0.68–3.20) 1 (ref) 1 (ref) 0.65 (0.35–1.20)
 Unprotected intercourse 3019 153 5.12% (3.99–6.25) 1.70 (1.22–2.37) 1.52 (1.07–2.17) 3380 65 2.01% (1.32–2.69) 1.03 (0.53–2.01) ·· 0.39 (0.27–0.56)
 No sexual partners 654 37 5.60% (3.48–7.72) 1.86 (1.15–3.01) 1.83 (1.14–2.93) 620 11 1.99% (0.75–3.23) 1.02 (0.48–2.20) ·· 0.36 (0.18–0.72)
Open in a new tab

PR=prevalence ratio. APR=adjusted prevalence ratio.

*

Weighted by sex, age group, and geographical area of the Bukoba Municipal Council census, and adjusted for clustering within census enumeration areas. All PRs and APRs with 95% CIs that exclude 1·0 are statistically significant (p<0·05).

†

Variables noted with ·· were not included in the multivariate model.

‡

Pre-intervention, 3 participants did not respond; post-intervention, 2 did not respond.

§

Pre-intervention, 16 participants did not respond.

Pre-intervention, 5 participants did not respond; post-intervention, 2 did not respond.

||

Unprotected indicates condom use for <100% of sexual-intercourse acts; protected indicates condom use for 100% of sexual-intercourse acts.

In 2017, post-intervention prevalence of undiagnosed HIV infection was similar among men and women, and urban and rural residents (table 2). Compared with reference subgroups, prevalence of undiagnosed HIV infection was approximately two times higher among residents aged 30–39 and 40–49 years, approximately three times higher among residents with only a primary education, having difficulty meeting household food needs, or using alcohol at least weekly; and approximately five times higher among residents without a mobile phone or television (table 2).

By 2017, current ART use increased two times overall and among women; nearly three times among men; three times among people living with HIV who had a secondary-level education, or reported unprotected intercourse in the past 6 months; and nearly four times among young adults aged 18–29 years (table 3). By 2017, disparities in ART use were substantially reduced between men and women, and people living with HIV aged 30–39 and 40–49 years compared with those aged 18–29 years (table 3).

Table 3:

Pre-intervention and post-intervention population prevalence of ART use among people living with HIV aged 18–49 years, and intrasurvey and intersurvey ART use PRs

Pre-intervention HIV survey (November, 2013-January, 2014) Post-intervention HIV survey (June, 2017-September, 2017) Intersurvey PR (95% CI)
People living with HIV (n) Prevalence of current ART use* People living with HIV (n) Prevalence of current ART use*
n % (95% CI) PR (95% CI) APR† (95% CI) n % (95% CI) PR (95% CI) APR† (95% CI)
Total 436 147 32.23% (26.36–38.10) ·· ·· 435 315 70.94% (65.52–76.37) ·· ·· 2.20 (1.82–2.66)
Sex
 Male 113 25 23.01% (14.80–31.22) 1 (ref) 1 (ref) 112 69 62.06% (53.55–70.57) 1 (ref) 1 (ref) 2.70 (1.84–3.96)
 Female 323 122 37.68% (31.15–44.20) 1.64 (1.15–2.32) 1.52 (1.14–2.04) 323 246 75.60% (70.51–80.69) 1.22 (1.07–1.38) 1.15 (1.04–1.27) 2.01 (1.69–2.38)
Age group (years)
 18–29 163 30 16.65% (9.63–23.67) 1 (ref) 1 (ref) 136 89 64.39% (56.31–72.46) 1 (ref) 1 (ref) 3.87 (2.54–5.89)
 30–39 180 60 29.81% (23.26–36.35) 1.79 (1.17–2.73) 1.95 (1.30–2.93) 172 124 70.56% (62.77–78.35) 1.10 (0.95–1.26) 1.18 (1.05–1.34) 2.37 (1.85–3.03)
 40–49 93 57 60.38% (47.85–73.91) 3.63 (2.44–5.38) 3.24 (2.20–4.79) 127 102 78.19% (68.85–87.53) 1.21 (1.01–1.46) 1.26 (1.08–1.47) 1.29 (1.04–1.62)
Residence
 Urban 344 123 34.00% (27.66–40.34) 1 (ref) ·· 313 227 71.11% (64.12–78.10) 1 (ref) ·· 2.09 (1.72–2.55)
 Rural 92 24 26.85% (14.33–39.38) 0.79 (0.49–1.28) ·· 122 88 70.49% (64.00–76.99) 0.99 (0.87–1.13) ·· 2.63 (1.66–4.16)
Duration of current home residence (years)
 <1 137 32 22.46% (16.51–28.40) 1 (ref) 1 (ref) 142 95 65.24% (56.50–73.99) 1 (ref) 1 (ref) 2.91 (2.17–3.89)
 1–2 99 29 28.82% (17.27–40.36) 1.28 (0.88–1.87) 1.10 (0.71–1.70) 89 63 70.90% (60.99–80.80) 1.09 (0.91–1.29) 1.01 (0.86–1.18) 2.46 (1.63–3.70)
 >2 200 86 40.19% (31.56–48.82) 1.79 (1.31–2.45) 1.38 (1.03–1.84) 204 157 74.92% (67.98–81.87) 1.15 (1.01–1.31) 1.03 (0.93–1.14) 1.86 (1.49–2.33)
Highest level of education completed
 At least some secondary 65 17 23.88% (13.45–34.31) 1 (ref) 1 (ref) 76 56 72.03% (61.10–82.95) 1 (ref) ·· 3.02 (1.96–4.64)
 Completed primary 279 103 35.52% (27.48–43.56) 1.49 (0.95–2.33) 1.15 (0.85–1.56) 276 198 69.82% (63.83–75.82) 0.97 (0.84–1.12) ·· 1.97 (1.56–2.47)
 None or some primar 92 27 28.22% (19.48–36.96) 1.18 (0.69–2.01) 1.07 (0.66–1.73) 83 61 73.59% (63.76–83.41) 1.02 (0.83–1.26) ·· 2.61 (1.86–3.65)
Employed in past 12 months
 No 56 21 38.18% (22.97–53.39) 1 (ref) ·· 56 43 75.80% (65.14–86.46) 1 (ref) ·· 1.99 (1.32–2.98)
 Yes 380 126 31.48% (25.56–37.40) 0.82 (0.55–1.23) ·· 379 272 70.31% (64.66–75.97) 0.93 (0.80–1.07) ·· 2.23 (1.83–2.72)
Ownership of mobile phone or television
 No 68 19 26.62% (16.92–36.31) 1 (ref) 1 (ref) 44 24 54.94% (39.61–70.26) 1 (ref) 1 (ref) 2.06 (1.31–3.25)
 Yes 368 128 33.33% (27.07–39.60) 1.25 (0.87–1.80) 1.06 (0.81–1.39) 391 291 73.06% (67.68–78.44) 1.33 (1.01–1.76) 1.22 (0.94–1.58) 2.19 (1.81–2.66)
Trouble satisfying household food needs‡
 Never 193 68 33.02% (25.02–41.02) 1 (ref) ·· 120 85 69.04% (61.37–76.71) 1 (ref) ·· 2.09 (1.62–2.71)
 Seldom 196 63 31.38% (23.90–38.86) 0.95 (0.71–1.27) ·· 206 148 70.68% (63.39–77.98) 1.02 (0.88–1.20) ·· 2.25 (1.73–2.93)
 Sometimes, often, or always 46 16 33.14% (19.35–46.92) 1.00 (0.66–1.54) ·· 108 81 73.32% (63.18–83.47) 1.06 (0.90–1.25) ·· 2.21 (1.47–3.32)
Frequency of alcohol use in past 6 months§
 Weekly or daily 93 14 14.95% (6.54–23.35) 1 (ref) 1 (ref) 67 39 54.98% (41.93–68.03) 1 (ref) 1 (ref) 3.68 (1.96–6.90)
 Less than monthly or monthly 106 34 32.21% (22.44–41.98) 2.16 (1.20–3.89) 1.77 (0.99–3.17) 114 84 72.86% (62.81–82.92) 1.33 (1.03–1.71) 1.16 (0.93–1.45) 2.26 (1.65–3.11)
 Never 237 99 40.36% (32.74–47.97) 2.70 (1.58–4.63) 2.36 (1.45–3.85) 253 191 74.90% (69.35–80.45) 1.36 (1.08–1.72) 1.21 (1.00–1.45) 1.86 (1.53–2.26)
Sexual behaviour in the past 6 months
 Unprotected intercourse 234 50 19.99% (14.54–25.44) 1 (ref) ·· 192 117 60.19% (52.68–67.70) 1 (ref) 1 (ref) 3.01 (2.27–4.00)
 Protected intercourse 106 51 47.27% (36.08–58.46) 2.37 (1.71–3.27) ·· 151 122 80.05% (73.16–86.94) 1.33 (1.15–1.54) 1.25 (1.09–1.43) 1.69 (1.33–2.15)
 No sexual partners 96 46 46.50% (34.23–58.77) 2.33 (1.72–3.15) ·· 92 76 80.54% (70.41–90.67) 1.34 (1.15–1.55) 1.20 (1.05–1.38) 1.73 (1.29–2.33)
Open in a new tab

ART=antiretroviral therapy. PR=prevalence ratio. APR=adjusted prevalence ratio.

*

Weighted by sex, age group, and geographical area of the Bukoba Municipal Council census, and adjusted for clustering within census enumeration areas. All PRs and APRs with 95% CI that exclude 1·0 are statistically significant (p<0·05). Current ART use included people living with HIV who either: reported ART use as part of the standard survey interview, were confirmed by medical record to be on ART, or had achieved viral load suppression (HIV-1 RNA concentration <1000 per μL on a viral load assay). ART use included viral load suppression because of low sensitivity (77%) of self-reported ART use15 and that only 1% of people living with HIV achieve durable viral load suppression in the absence of ART adherence.16

†

Variables noted with ·· were not included in the multivariate model.

‡

Pre-intervention, 1 participant did not respond; post-intervention, 1 did not respond

§

Post-intervention, 1 participant did not respond.

Unprotected indicates condom use for <100% of sexual-intercourse acts; protected indicates condom use for 100% of sexual-intercourse acts.

By 2017, diagnostic coverage among people living with HIV increased from 47·4% (95% CI 41·3–53·4) to 76·2% (71·8–80·6); ART use among diagnosed people living with HIV increased from 68·0% (60·9–75·2) to 93·1% (90·2–96·0), and viral load suppression among people living with HIV on ART increased from 88·7% (83·6–93·8) to 91·3% (88·6–94·1), with similar gains in most sex and age groups (figure 4). In 2017, diagnostic coverage among people living with HIV was lowest among men and residents aged 18–29 years, and highest among women and residents aged 40–49 years (figure 4).

Figure 4: Conditional population prevalence of previous HIV diagnosis, current ART use, and viral load suppression among people living with HIV aged 18–49 years, by sex (A) and age group (B), pre-intervention and post-intervention household surveys.

Figure 4:

Open in a new tab

Error bars show 95% CIs. Previous HIV diagnosis and current ART use included people living with HIV who either: reported these conditions as part of the standard survey interview, were confirmed by medical record to have these conditions, or had achieved viral load suppression (HIV-1 RNA concentration <1000 per μL on a viral load assay). Conditions included viral load suppression because of low sensitivity (77%) of self-reported ART use15 and that only 1% of people living with HIV achieve durable viral load suppression in the absence of ART adherence.16 Pre-intervention (October, 2013–January, 2014) and post-intervention (June, 2017–September, 2017) prevalence estimates and 95% CI for previous HIV diagnosis, ART use, and viral load suppression among people living with HIV were weighted by sex, age group, and geographical area of the Bukoba Municipal Council census and adjusted for clustering within census enumeration areas. ART=antiretroviral therapy.

Discussion

After implementation of a comprehensive combination prevention intervention during a 2·5 year period of expanding ART policy, ART coverage among people living with HIV in Bukoba increased two times overall, and nearly three times among men and four times among young adults—populations that consistently have low ART coverage.5,6 In 2017, more than 90% of diagnosed people living with HIV in Bukoba were using ART and more than 90% of those on ART were virally suppressed, meeting the last two goals of 90-90-90 that are necessary to reduce HIV transmission.1 Notably, BCPE was not an experimental test-and-treat community trial. ART was provided in accordance with national treatment guidelines, and only in the final 6 months of the intervention were all HIV-positive people eligible to receive ART. Our findings suggest that comprehensive facility-based and community-based HIV testing, combined with recommended linkage, defaulter tracing, and test-and-treat services can help reduce gender and age-group disparities and increase HIV diagnostic, ART, and viral load suppression coverage in urban and rural communities in a relatively short period.12,13,17

Although BCPE facility-based and community-based testing programmes newly diagnosed more than 4000 people living with HIV and helped to substantially increase HIV diagnostic coverage, the 90% coverage target was not met, contributing to higher prevalence of undiagnosed infection and lower ART coverage among people living with HIV in several sociodemographic groups. Prevalence of undiagnosed infection remained high among residents who had never completed primary education, frequently used alcohol, or were likely impoverished (suggested by not owning a mobile phone or difficulty meeting family food needs).

Although diagnostic coverage among people living with HIV in 2017 was substantially higher among women than men, post-intervention prevalence of undiagnosed HIV infection between women and men was similar. Similarly, although diagnostic coverage in 2017 was higher among people living with HIV aged 30–49 than 18–29 years, residents aged 30–49 years had twice the prevalence of undiagnosed infection than residents aged 18–29 years. Applying our estimated prevalence of undiagnosed infection to the census of Bukoba residents aged 18–49 years suggests that 660 (34 724 × 0·0190) and 750 (36 078 × 0·0208) men and women living with HIV, and 520 (38 510 × 0·0135) and 891 (32 292 × 0·0276) people living with HIV aged 18–29 and 30–49 years, respectively, remained unaware of their HIV infection and needed diagnosis in 2017.18

Although considerable emphasis has been placed on increasing diagnostic coverage among younger people and men (due to lower diagnostic coverage among people living with HIV in these subgroups), as noted in our study and elsewhere, total undiagnosed infections might remain higher in older people and women due to higher prevalence of HIV infection.6 In Bukoba, our findings suggest testing programmes should equally target men and women and similar if not greater efforts should be made to reach and test high-risk people aged 30–49 years. Because prevalence of undiagnosed HIV infection remained substantially higher in lower socioeconomic and alcohol-using groups, our findings underscore the importance of other testing strategies needed to reach those who might not frequent health-care facilities to achieve more than 90% diagnostic coverage.

Over the 2·5 year intervention, we did 27 407 home-based and 17 475 venue-based tests (34% of all tests) to reach people living with HIV who might not regularly access health-care services.12 Lower socioeconomic neighbourhoods were not specifically targeted for home-based testing and the 79 community venues were mostly high-volume market, street locations, or community events.12 Our findings suggest that community-based testing strategies targeting impoverished neighbourhoods, the homeless, and bars and clubs might be important to achieve diagnostic coverage in low socioeconomic and alcohol-using groups.19 Also, while we diagnosed new HIV infections in 25% of 691 sexual partners and family members of linkage clients who were HIV tested, we had limited success in reaching most sexual partners.13 Safe and appropriate index client testing services might also be effective in reducing undiagnosed infections among residents of all ages, including those aged 30 years or older.20–23

Of the four community-randomised, combination prevention test-and-treat trials, three had similarly low ART coverage as BCPE at baseline and offer useful comparisons.7–11 Of these, only the SEARCH trial8 in Kenya and Uganda reached 90-90-90, achieving 89% ART coverage among all stable-resident people living with HIV.8 In KwaZulu-Natal, South Africa, TasP trial9 made modest progress towards 90-90-90, achieving 48% ART coverage among all people living with HIV aged 16 years or older.9 Over an intervention of 2 or more years, all identified people living with HIV in these experimental test-and-treat trials were informed they would be initiated on ART regardless of CD4 cell count.7–10 In 16 small rural communities of approximately 5000 residents each, SEARCH relied primarily on home-based and community-based testing campaigns to achieve 90-90-90, a strategy that might not be as effective in larger urban communities.8 Notably, only 48–65% of people living with HIV in these trials linked to care within 6 months of diagnosis and 63–73% did so within 12 months.7–9 Linkage services in these trials, at least at first, relied primarily on point-of-diagnosis counselling and referral, with home-visit follow-up for clients who failed to enrol in care.7–9 By contrast, during the 6-month test-and-treat period, BCPE’s linkage case management intervention, combined with community-based and facility-based same-day ART services, enrolled in HIV care 97% of all clients within 3 months of diagnosis.13 Our peer-delivered, linkage case management intervention provided a comprehensive package of CDC and WHO recommended linkage services to consenting clients from the point of diagnosis for up to 90 days.13 HIV-positive, peer counsellors served as linkage champions for their clients, sharing their own personal experiences of living with HIV to help clients overcome the shock of diagnosis, escorted and facilitated their enrolment in care, ensured they understood how to navigate care, and helped mitigate real or perceived barriers to care (eg, stigmatisation, perceived wellness, denial, concerns about partner response or loss, etc). These proactive, peer-delivered services might have been particularly helpful to younger people and men who are more likely to delay enrolment in care.5,6,13,19 Our linkage-to-care findings are similar to those reported by a peer-delivered linkage case management programme in eSwatini, and is consistent with other studies in sub-Saharan Africa that report the benefits of integrated testing and linkage support for early ART initiation.6,19,24–27 Not a major component of the test-and-treat experimental trials, BCPE defaulter tracing services also contributed to improving ART coverage in Bukoba by initiating or reinitiating ART on over 500 patients who had stopped care. As experienced throughout sub-Saharan Africa, many pre-ART patients in Bukoba defaulted from HIV care, and defaulter tracing services were implemented after the first ART-eligibility expansion in 2016 as an effort to help former patients receive ART.28,29 Because of low retention in pre-ART care, more patients might have been retained in care and higher ART coverage achieved in Bukoba if both test-and-treat and defaulter tracing services were provided from the beginning of our intervention.28,29 Many patients were traced 1–2 years after defaulting from pre-ART care and were lost to follow-up. Findings from our defaulter tracing programme underscores both the value of test and treat and same-day ART to avert early drop-out, and need for more immediate initiation of defaulter tracing services to prevent patients from becoming lost to follow-up.29

Our findings are subject to at least five limitations. First, because our research design did not include control communities, BCPE intervention effects on population-level coverage of previous HIV diagnosis and ART could not be estimated. Although BCPE provided a large majority of testing services during the intervention, some residents were diagnosed and linked to ART as hospital inpatients, and pre-natal and post-natal care, tuberculosis, and voluntary medical male circumcision outpatients. Nonetheless, we believe a large intervention effect is likely. As reported elsewhere, BCPE testing and linkage interventions diagnosed and enrolled in HIV care 3488 people living with HIV aged 15–49 years during the intervention period, representing nearly 100% of the estimated 3493 people living with HIV residents aged 18–49 years who were not in HIV care in 2014.12,13,17 Second, contributing to uncertainty in estimating BCPE intervention effects, residence of clients who received BCPE testing and linkage services was not collected and is unknown. Some BCPE participants who were diagnosed and linked to HIV care in Bukoba facilities might not have been residents of Bukoba, which is home to the regional referral hospital and two health centres known to provide medical services to some residents of other districts. Third, although prevalence estimates were weighted to the census population, residual bias might reduce the validity of estimates for men who were underrepresented in both surveys. Fourth, although we used a combination of self-report, medical record review, and viral load suppression to define previous HIV diagnosis and ART use, these outcomes might remain underestimated given the low sensitivity of self-report among people living with HIV and likelihood that medical record reviews did not capture all survey participants who were currently in HIV care.15 Finally, the scalability of our combined interventions might be limited to well resourced HIV prevention and treatment programmes. However, to help enable adoption, all our interventions were developed to be reasonably low cost, relying primarily on peer lay counsellors to deliver services. For example, estimated peer-delivered costs per HIV test (US$, 2017) for BCPE facility-based, home-based, and venue-based strategies were $3·06, $4·81, and $5·45, respectively, similar with other reported HIV-test costs.12 Estimated per client cost for peer-delivered linkage case management was $18·00 for a facility-based model.13

Interventions modelled on BCPE when implemented under a test-and-treat programme are promising strategies that might help countries achieve 90-90-90 goals. In 2017, BCPE facility-based testing and linkage case management interventions were approved by the Ministry of Health, Community Development, Gender, Elderly and Children as new service delivery models. In 2019, these interventions were implemented by four HIV prevention organisations and scaled up in all regions of Tanzania.30 The United States President’s Emergency Plan for AIDS Relief recommends optimised testing in facilities and peer-delivered linkage case management as possible solutions for country programmes striving to achieve 90-90-90.31

Research in context.

Evidence before this study

On Jan 15, 2019, we searched PubMed for studies that assessed the impact of universal HIV testing and antiretroviral therapy (ART; test and treat) on HIV incidence in sub-Saharan Africa. We used search terms “HIV” AND “incidence” AND “test and treat” AND “community randomized” AND “Africa” in English. We found four community randomised, test and treat trials conducted in Kenya and Uganda (SEARCH), South Africa (TasP), Zambia and South Africa (PopART), and Botswana (BCCP). All trials were conducted for approximately 3 years and included comprehensive repeat community-based HIV testing, referral and linkage services, and ART for all people living with HIV in intervention communities. To evaluate incidence, SEARCH studied 95 083 residents from 32 rural communities in Kenya and Uganda; TasP studied 14 223 residents in 22 communities in rural KwaZulu-Natal, South Africa; PopART studied 48 301 residents of 12 periurban communities in Zambia and nine communities in Western Cape Province, South Africa; and BCCP studied 8974 residents of 30 rural or periurban communities in Botswana. By the end of the study in intervention communities, more than 90% of all people with HIV had been diagnosed, 58–97% of those diagnosed were on ART, and 46–88% of all people with HIV had achieved viral load suppression. Compared with study control arms, HIV incidence was significantly lower by approximately 30% in two of five intervention arms of the four trials. Mixed results from the universal test and treat trials have been attributed to multiple factors including increased ART access in control communities and delay (or failure) in reaching ART coverage targets in intervention communities. Notably, only 48–65% and 63–73% of people with HIV linked to care within 6 and 12 months of diagnosis, respectively, in three of the four trials (TasP, PopART, and SEARCH), underscoring the need for new prevention interventions to optimise test and treat.

Added value of this study

Implemented during a period of expanding national ART policies, the Bukoba Combination Prevention Evaluation (BCPE) aimed to achieve more than 80% ART coverage of all residents with HIV aged 18–49 years in Bukoba Municipal Council, a mixed urban and rural council in Tanzania with approximately 150 000 residents. BCPE is the first study in Tanzania to assess the scale-up of a comprehensive package of services that included facility and community-based HIV testing, peer-delivered linkage case management, and defaulter tracing. We demonstrate that our combination prevention intervention helped to substantially reduce prevalence of undiagnosed HIV infection, and as national treatment guidelines expanded access to ART, helped to substantially increase ART coverage and reduce considerable sex and age-group ART disparities.

Implications of all the available evidence

BCPE testing, linkage, and defaulter tracing interventions are promising strategies that can help strengthen test and treat to achieve epidemic control. In 2017, BCPE facility-based testing and linkage case-management interventions were approved by the Tanzania Ministry of Health, Community Development, Gender, Elderly and Children as new service delivery models. In 2019, these interventions were implemented by four HIV prevention organisations and scaled up in all regions of Tanzania.

Acknowledgments

In memory of our dear colleague and friend Dr Fernando Morales who sadly passed away before publication of this Article. He was a loving father, inspiring leader, and dedicated mentor who humbly dedicated his career trying to improve the health of those less fortunate throughout sub-Saharan Africa. In memory of Dr Subilaga Kaganda, a public health practitioner determined to address the HIV/AIDS epidemic in Tanzania, who sadly passed away before the project was completed. We are thankful for the support of the Tanzania’s Ministry of Health, Community Development, Gender, Elderly and Children, National AIDS Control Program, Tanzania Commission for AIDS, Bukoba Municipal Council and Kagera region leadership, and staff at the 11 health-care facilities. We are also grateful to the residents of Bukoba Municipal Council for embracing the study. We acknowledge the invaluable contributions of the lay counsellors, health-care workers, and data clerks without whom these achievements would not be possible. This research was supported by the United States’ President’s Emergency Plan for AIDS Relief through the US Centers for Disease Control and Prevention under the terms of cooperative agreement 5UGPS001998-05. The findings and conclusions in this Article are those of the authors and do not necessarily represent the official position of the funding agencies. Editorial note: the Lancet Group takes a neutral position with respect to territorial claims in published maps and institutional affiliations.

Footnotes

Declaration of interests

We declare no competing interests.

Data sharing

Study protocol, statistical analysis plan, and analytical code will be made available upon request to the corresponding author.

References

RESOURCES