Tropical Medicine and International Health
doi:10.1111/tmi.12387
volume 19 no 12 pp 1411–1419 december 2014
Impact of systematic HIV testing on case finding and retention in care at a primary care clinic in South Africa Kate Clouse1, Colleen F. Hanrahan1, Jean Bassett2, Matthew P. Fox3,4,5, Ian Sanne3 and Annelies Van Rie1 1 Department of Epidemiology, University of North Carolina Gillings School of Global Public Health, Chapel Hill, NC, USA 2 Witkoppen Health and Welfare Centre, Johannesburg, South Africa 3 Health Economics and Epidemiology Research Office, Department of Medicine, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa 4 Department of Epidemiology, Boston University, Boston, MA, USA 5 Center for Global Health & Development, Boston University, Boston, MA, USA
Abstract
objective Systematic, opt-out HIV counselling and testing (HCT) may diagnose individuals at lower levels of immunodeficiency but may impact loss to follow-up (LTFU) if healthier people are less motivated to engage and remain in HIV care. We explored LTFU and patient clinical outcomes under two different HIV testing strategies. methods We compared patient characteristics and retention in care between adults newly diagnosed with HIV by either voluntary counselling and testing (VCT) plus targeted provider-initiated counselling and testing (PITC) or systematic HCT at a primary care clinic in Johannesburg, South Africa. results One thousand one hundred and forty-four adults were newly diagnosed by VCT/PITC and 1124 by systematic HCT. Two-thirds of diagnoses were in women. Median CD4 count at HIV diagnosis (251 vs. 264 cells/ll, P = 0.19) and proportion of individuals eligible for antiretroviral therapy (ART) (67.2% vs. 66.7%, P = 0.80) did not differ by HCT strategy. Within 1 year of HIV diagnosis, half were LTFU: 50.5% under VCT/PITC and 49.6% under systematic HCT (P = 0.64). The overall hazard of LTFU was not affected by testing policy (aHR 0.98, 95%CI: 0.87–1.10). Independent of HCT strategy, males, younger adults and those ineligible for ART were at higher risk of LTFU. conclusions Implementation of systematic HCT did not increase baseline CD4 count. Overall retention in the first year after HIV diagnosis was low (37.9%), especially among those ineligible for ART, but did not differ by testing strategy. Expansion of HIV testing should coincide with effective strategies to increase retention in care, especially among those not yet eligible for ART at initial diagnosis. keywords HIV/AIDS, retention in care, loss to follow-up, voluntary counselling and testing, provider-initiated HIV counselling and testing, HIV counselling and testing
Introduction South Africa has more people living with HIV than any other country in the world (Joint United Nations Programme on HIV/AIDS (UNAIDS) 2010). In a generalised HIV epidemic WHO recommends that providerinitiated HIV counselling and testing (PITC) should be the standard of care and performed irrespective of symptoms or reason for clinic visit (WHO/UNAIDS 2007). South African HIV testing guidelines recommend PITC in all settings (Republic of South Africa Department of Health 2010), but PITC continues to mainly target specific clinic populations, such as those
© 2014 John Wiley & Sons Ltd
attending antenatal, tuberculosis (TB) or sexually transmitted infection (STI) services (Pope et al. 2008; Kharsany et al. 2010; Leon et al. 2010b). Benefits of PITC include increased uptake of testing (Bassett & Walensky 2010; Leon et al. 2010b; Dalal et al. 2011; Kennedy et al. 2013; Roura et al. 2013) and earlier HIV diagnosis prior to advanced immunosuppression (Bassett & Walensky 2010; Republic of South Africa Department of Health 2010). Better integration of HIV testing with routine clinic services can also contribute to normalising HIV care and reduce HIV exceptionalism and stigma (Leon et al. 2010a; Dalal et al. 2011). 1411
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
A well-documented challenge of providing effective HIV services in sub-Saharan Africa is retention in care. Retention has been shown to be especially poor in the period between HIV diagnosis and antiretroviral therapy (ART) initiation (Bassett et al. 2010; Lessells et al. 2011; Rosen & Fox 2011; Mugglin et al. 2012; Clouse et al. 2013), and somewhat better but still problematic in the period after ART initiation (Cornell et al. 2010; Fox & Rosen 2010; Kranzer et al. 2012). One concern related to expanding testing is that those who are identified may not be adequately linked to care or motivated to engage in care, particularly those who feel healthy or are ineligible for ART, thus exacerbating retention challenges (Roura et al. 2013). To understand the impact of systematic HIV counselling and testing (HCT) on retention in care and patient clinical outcomes, we compared data from a period when voluntary testing (VCT) plus targeted PITC was implemented at a primary care clinic to a period when a strategy of clinicwide, systematic HCT was implemented at the same clinic.
Methods Study setting Witkoppen Health and Wellness Centre (WHWC) is a high-volume primary healthcare clinic in Johannesburg, South Africa, which primarily serves a population from neighbouring formal and informal settlements characterised by widespread poverty, frequent mobility and high HIV and TB burden. Prior to 2011, WHWC offered VCT, plus targeted PITC within antenatal, TB and STI services, or at any time at the request of a care provider. Individual pre- and post-test counselling accompanied HIV testing. In January 2011, the clinic’s HIV testing strategy was changed to systematic HCT for all adult clinic clients. At the start of each visit, prior to any clinic activity, current HIV status was assessed by three specially-assigned clinic staff. All adult patients with a negative result ≥3 months prior to the current visit and all clients new to the clinic were sent first to group pre-test HIV counselling. This was followed by a one-on-one interaction with an HIV counsellor during which the individual was asked to accept or decline rapid on-site HIV testing. Individual post-test counselling followed testing. HIV testing for both strategies occurred in the same testing room at the clinic, using the same staff component and same HIV test method (rapid, point-of-care tests using dried blood spots). For individuals newly diagnosed with HIV at WHWC, blood is drawn for CD4 count and sent to an outside laboratory, and results are shared and discussed at a followup visit at WHWC about 1–4 weeks later. For those with 1412
CD4 counts ≤350 cells/ll, ART initiation occurs 1– 4 weeks after that within the same clinic. After ART initiation, visits are scheduled monthly until the individual achieves viral suppression, and then are scheduled quarterly. For individuals ineligible for ART based on CD4 results, follow-up visits to monitor CD4 values are scheduled about every 6 months. The policy of initiation at ≤350 cells/ll was adopted in mid-2011 throughout South Africa. However, WHWC adopted the policy earlier, in early 2010, and during data collection, all patients in the earlier cohort were individually assessed for ART eligibility, per clinician decision, to ensure correct classification. Study design We conducted an observational cohort study of adults (age ≥18) newly diagnosed with HIV infection through the VCT plus targeted PITC strategy (January–June 2010) or through systematic HCT (February–July 2012). File reviews were conducted at least 15 months after each patient tested HIV-positive. The primary exposure of interest is the HIV testing strategy: VCT plus targeted PITC or systematic HCT. The outcomes of interest were the case finding yield (number of new HIV diagnoses during the 6 month periods), characteristics of individuals newly diagnosed with HIV infection, ART initiation and retention in care 12 months after HIV diagnosis. Loss to follow-up (LTFU) was defined as not returning to the clinic ≥3 months after the last scheduled visit. Patients who requested transfer to another clinic were identified as ‘transferred out’, considered a separate outcome category and not counted as LTFU. Statistical analysis Patient characteristics at the time of testing HIV-positive are presented using counts and proportions for categorical variables, and medians and interquartile ranges for continuous variables. Retention outcomes 1 year after HIV testing are reported as proportions and 95% confidence intervals (95%CI), with P-values determined by chi-square testing. We created crude Kaplan–Meier curves of time to LTFU by study period and compared the distributions using the log-rank test of equality for survival functions. We identified effect measure modification of the association between testing strategy and LTFU by pregnancy status and ART eligibility status. To account for this, we stratify results by a six-level variable combining gender, pregnancy and ART eligibility. We used Cox proportional hazards models methods, adjusting for
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
covariates selected as potential confounders based on a review of the literature and causal diagrams, to estimate adjusted hazard ratios (aHR) and 95%CI. Ethical consideration This study was approved by the institutional review boards of the University of North Carolina at Chapel Hill and the University of the Witwatersrand.
Results Description of individuals recently diagnosed with HIV In total, 2266 individuals were newly diagnosed with HIV during the two 6-month study periods: 1144 (50.5%) individuals were newly diagnosed by VCT with targeted PITC and 1122 (49.5%) by systematic HCT. For both 6-month periods, over 90% of files of individuals newly diagnosed with HIV could be located and reviewed for inclusion in the analysis. The median age for both groups was 32 years, and the distributions by age category were similar (Table 1). Almost half (43.8%) had immigrated to South Africa from neighbouring countries and unemployment was high (46.7%). Independent of the HIV testing strategy, most new HIV diagnoses were made in women (66.9% in VCT/PITC group and 65.3% in HCT group). The proportion of pregnant women was higher among those diagnosed through VCT/PITC (39.2%) than those diagnosed through systematic HCT (26.3%) (P < 0.0001). The median CD4 count was 256 cells/ll, and similar between groups (251 for VCT/ PITC and 264 for systematic HCT, P = 0.19). Independent of HIV testing strategy, about two-thirds of individuals newly diagnosed with HIV were eligible for ART based on a CD4 count ≤350 cells/ll: 67.2% for VCT/PITC and 66.7% for systematic HCT strategy, P = 0.80, Figure 1). Overall, 61.2% of those eligible initiated ART, 58.7% of those diagnosed by VCT/PITC and 63.7% among those diagnosed by systematic HCT (P = 0.05, Figure 1). Retention in care during the first year following HIV diagnosis Only 37.9% (95%CI: 35.9–39.9%) of individuals newly diagnosed with HIV were retained in care at WHWC during the 12-month period following HIV diagnosis, with similar proportions independent of HIV testing strategy: 36.5% (95%CI: 33.7–39.3%) for VCT/PITC
© 2014 John Wiley & Sons Ltd
Table 1 Characteristics of 2266 individuals newly diagnosed with HIV at a primary care clinic in Johannesburg, South Africa, overall and by HIV testing strategy
Patient characteristic Sex, n (%) Female Male Age at HIV testing, median (IQR) Age at HIV testing, n (%) 18–29 years 30–39 years 40 years and older First CD4 value (cells/ll), median (IQR) First CD4 value, n (%) 350 cells/ll Missing Nationality, n (%) Born in South Africa Born outside of South Africa Missing Employment status, n (%) Employed Not employed Missing Initiated ART*, n (%) No Yes Pregnant at HIV diagnosis, n (%)† No Yes Missing
Total
VCT/targeted PITC (n = 1144, 50.5%)
Systematic HCT (n = 1122, 49.5%)
1498 (66.1) 768 (33.9) 32 (27–39)
765 (66.9) 379 (33.1) 32 (27–38)
733 (65.3) 389 (34.7) 32 (28–39)
864 (38.1) 898 (39.6) 504 (22.2)
452 (39.5) 451 (39.4) 241 (21.1)
412 (36.7) 447 (39.8) 263 (23.4)
256 (128–403)
251 (120–397)
264 (131–410)
209 626 639 707 85
106 333 318 349 38
103 293 321 358 47
(9.2) (27.6) (28.2) (31.2) (3.8)
(9.3) (29.1) (27.8) (30.5) (3.3)
(9.2) (26.1) (28.6) (31.9) (4.2)
1240 (54.7)
625 (54.6)
615 (54.8)
992 (43.8)
485 (42.4)
507 (45.2)
34 (1.5)
34 (3.0)
1168 (51.5) 1059 (46.7) 39 (1.7)
541 (47.3) 573 (50.1) 30 (2.6)
627 (55.9) 486 (43.3) 9 (0.8)
567 (38.8) 894 (61.2)
307 (41.3) 437 (58.7)
260 (36.3) 457 (63.7)
991 (66.2) 493 (32.9) 14 (0.9)
465 (60.8) 300 (39.2) 0
526 (53.1) 193 (26.3) 14 (1.9)
0
ART, antiretroviral therapy; HCT, HIV counselling and testing; IQR, interquartile range; PITC, provider-initiated HIV counselling and testing; VCT, voluntary HIV counselling and testing. *Among patients eligible for ART per CD4 value ≤350 cells/ll (n = 1461). †Among female participants (n = 1498).
1413
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
(a) 100%
VCT/targeted PITC period
Systematic HCT period
90% P = 0.80
P = 0.05
80% 70%
67.2%
66.7% 58.7%
60%
63.7%
P = 0.02 47.8%
50%
41.7%
40% 30% 20% 10% 0% Eligible for ART
Initiated ART within 12 months Retained in care at WHWC12 of HIV diagnosis months after HIV diagnosis
(b) 100%
VCT/targeted PITC period
Systematic HCT period
90% 80% 70% 60% 50% 40%
P = 0.80 32.8%
33.3%
30%
P = 0.53 29.2%
27.1%
20% 10% 0% Ineligible for ART
Retained in care at WHWC 12 months after HIV diagnosis
and 39.4% (95%CI: 36.6–42.3%) for systematic HCT. Independent of testing strategy, few individuals were transferred to another healthcare facility (8.8%, 95%CI: 7.7–10.0%) and very few were known to have died (3.2%, 95%CI: 2.6–4.0%). About half were lost to follow-up within the first year of HIV diagnosis: 50.5% (95%CI: 47.6–53.4%) for VCT/PITC and 49.6% (95% CI: 46.6–52.3%) for systematic HCT. Kaplan–Meier estimates of time to LTFU were also similar by HCT strategy (log-rank test P = 0.74, Figure 2). When assessing factors associated with LTFU (Table 2), we found no difference in the overall hazard by HIV testing strategy (aHR 0.98, 95%CI: 0.87–1.10). Independent of gender or pregnancy status, those who were ineligible for ART were 18–40% more likely to be LTFU compared to their ART-eligible counterparts. Older individuals (30+ years) were 31–40% less likely to be LTFU compared to those age 18–29 years. Those born outside of South Africa were at slightly higher risk of 1414
Figure 1 Retention in care of 2181 newly diagnosed individuals by voluntary counselling and testing (VCT) or targeted provider-initiated HIV counselling and testing (light grey), or systematic HIV counselling and testing (HCT) (dark grey). (a) Among individuals eligible for antiretroviral treatment (ART) at time of HIV diagnosis (CD4 count ≤350 cells/ll) and (b) among individuals not yet eligible for ART at the time of HIV diagnosis (CD4 count >350 cells/ll). WHWC, Witkoppen Health and Welfare Centre. Eighty-five (3.8%) patients missing a CD4 count at HIV diagnosis were excluded.
LTFU (1.17, 95%CI: 1.04–1.32) than those born in South Africa. There was no relationship between employment status and LTFU. Examining factors associated with LTFU separately for each HCT strategy, risk factors were similar with a few exceptions. Among non-pregnant women diagnosed under systematic HCT, those ineligible for ART were 48% more likely to be LTFU than those eligible for ART (aHR 1.48, 95%CI: 1.15–1.91), whereas ART eligibility did not determine the hazard of LTFU among non-pregnant women diagnosed by VCT/PITC (aHR 0.88, 95% CI: 0.65–1.19). The strongest predictor of LTFU was among men ineligible for ART at HIV diagnosis who were diagnosed by systematic HCT (aHR 2.02, 95% CI: 1.50–2.74); a much stronger association than that observed among ART-ineligible men diagnosed by VCT/ PITC (aHR 1.19, 95% CI: 0.84–1.69). Overall, individuals ineligible for ART at the time of HIV diagnosis had poorer retention than those eligible
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
0.75 0.50 0.25
VCT/targeted PITC
Log-rank P = 0.74
3
6
9
12
Months since HIV diagnosis Number at risk VCT/targeted PITC Systematic HCT
for ART (28.2% vs. 44.7%, P < 0.0001). Among individuals eligible for ART, those diagnosed by systematic HCT tended to have better retention in care than those diagnosed by VCT/PITC (47.8% vs. 41.7%, P = 0.02 – Figure 1; log-rank test P = 0.05 – Figure 3). In contrast, the proportion of individuals retained in care 12 months after diagnosis among those ineligible for ART was similar between HCT strategies, (27.1% vs. 29.2%, P = 0.53 – Figure 1) but dropout occurred sooner after HIV diagnosis among those diagnosed by systematic HCT (log-rank test P = 0.03 – Figure 3). Discussion Systematic HCT at primary care facilities is recommended by the WHO but rarely implemented. In contrast to our expectation, we did not observe an increase in HIV case finding yield, found no difference in demographic and clinical characteristics of individuals newly diagnosed with HIV, and observed similar overall rates of retention in care during the first 12 months following HIV diagnosis when moving from VCT with targeted PITC to systematic HCT for all adult clients at a primary care clinic in Johannesburg, South Africa. The number of clinic clients newly diagnosed with HIV during a 6-month period was almost exactly the same for both HIV testing strategies (1144 vs. 1122), a surprising finding considering the explicit expansion of HIV testing. Others have shown mixed results regarding case finding after transitioning from VCT to PITC, reporting both increases (Bassett et al. 2007) and decreases (Topp et al.
© 2014 John Wiley & Sons Ltd
Systematic HCT
0.00
Figure 2 Kaplan–Meier estimates of time to loss to follow-up (LTFU) by HIV testing strategy among 2154 individuals newly diagnosed with HIV at a primary care clinic in Johannesburg, South Africa. VCT, voluntary HIV counselling and testing; PITC, provider-initiated HIV counselling and testing; HCT, HIV counselling and testing. LTFU is defined as not returning at least 3 months after the last scheduled visit; thus, person-time begins accumulating at 3 months. This analysis excludes 112 patients who died or transferred prior to 3 months followup (N = 2154).
Kaplan-Meier estimate
1.00
K. Clouse et al. Impact of systematic HIV testing
1092 1062
644 611
502 514
418 442
2011). The reason for this surprising finding is unclear. Because this was not a randomised controlled trial, other factors could have contributed to our observation. For example, the national South African HIV testing campaign launched in April 2010 could have decreased the proportion of individuals with unknown HIV status between 2010 and 2012. The characteristics of those newly diagnosed with HIV also did not substantially change between strategies. Independent of testing strategy, women comprised nearly two-thirds of those testing HIV-positive. The proportion of those newly diagnosed that were pregnant women decreased from 39.2% in the VCT and targeted HCT group to 26.3% in the systematic HCT group (P < 0.0001). This may be because pregnant women were one of the key groups targeted for HIV testing prior to systematic HCT. Surprisingly, median CD4 barely increased – from 251 to 264 cells/ll – when expanding HIV testing. This is in contrast to a study in Zambia that found a decrease in median CD4 value after implementing PITC, suggesting that PITC was providing a ‘safety net’ for individuals who did not seek VCT (Topp et al. 2012). Community-based HIV testing studies in South Africa reported higher median first CD4 values (>400 cells/ll) (Auvert et al. 2004; Kranzer et al. 2013). Taken together, these data suggest that, independent of HIV testing strategy, clinic-based HIV testing identifies individuals at more advanced HIV disease stage compared to community-based testing. Increasing the number of people who know their HIV status will only reduce the burden of HIV if those 1415
1416
40 years and older Nationality Born in South Africa Born outside of South Africa
Year of HIV testing Pre-intervention (2010) Post-intervention (2012) Gender/pregnancy and ART-eligibility status† Female, not pregnant at HIV testing, ART-eligible Female, not pregnant at HIV testing, ART-ineligible Female, pregnant at HIV testing, ART-eligible Female, pregnant at HIV testing, ART-ineligible Male, ART-eligible Male, ART-ineligible Age at HIV testing 18–29 years 30–39 years 1.08 (0.88, 1.33) 1.55 (1.29, 1.87) 1.15 (0.98, 1.36) 1.54 (1.23, 1.92) 1 0.69 (0.60, 0.78) 0.60 (0.51, 0.71)
124/244 (50.8)
167/229 (72.9)
261/527 (49.5)
526/827 (63.6) 407/852 (47.8)
552/961 (57.4)
572/1159 (49.4)
201/475 (42.3)
1.26 (1.12, 1.41)
1
1.27 (1.05, 1.53)
161/291 (55.3)
102/161 (63.4)
1
0.98 (0.87, 1.10)
556/1062 (52.4)
239/612 (39.1)
1
Crude HR (95% CI)
578/1092 (52.9)
n (%)
Overall
1.17 (1.04, 1.32)
1
1 0.69 (0.60, 0.79) 0.60 (0.50, 0.72)
1.20 (1.01, 1.41) 1.57 (1.25, 1.97)
1.30 (1.06, 1.58)
0.90 (0.73, 1.12)
1.18 (0.97, 1.43)
1
0.98 (0.87, 1.10)
1
Adjusted HR (95% CI)
290/473 (61.3)
278/585 (47.5)
92/227 (40.5)
285/438 (65.1) 201/427 (47.1)
43/74 (58.1)
140/268 (52.2)
105/140 (75.0)
77/148 (52.0)
63/138 (45.7)
1.44 (1.22, 1.70)
1
1 0.64 (0.54, 0.77) 0.56 (0.44, 0.71)
1.26 (1.00, 1.58) 1.20 (0.86, 1.69)
1.65 (1.29, 2.12)
1.11 (0.85, 1.47)
0.95 (0.71, 1.27)
1
–
–
118/290 (40.7)
–
Crude HR (95% CI)
–
n (%)
VCT/targeted PITC
1.35 (1.14, 1.60)
1
1 0.64 (0.53, 0.79) 0.56 (0.43, 0.73)
1.33 (1.05, 1.69) 1.19 (0.84, 1.69)
1.41 (1.09, 1.84)
0.90 (0.68, 1.21)
0.88 (0.65, 1.19)
1
–
–
Adjusted HR (95% CI)
262/488 (53.7)
294/574 (51.2)
109/248 (44.0)
241/389 (62.0) 206/425 (48.5)
59/87 (67.8)
121/259 (46.7)
62/89 (69.7)
47/96 (49.0)
98/153 (64.1)
121/322 (37.6)
–
–
n (%)
Systematic HCT
Table 2 Multivariate analysis of factors associated with patient LTFU* 1 year after HIV at a primary care clinic in South Africa
1.10 (0.93, 1.30)
1
1 0.73 (0.61, 0.88) 0.64 (0.51, 0.81)
1.05 (0.83, 1.33) 1.91 (1.42, 2.57)
1.40 (1.05, 1.88)
1.02 (0.74, 1.41)
1.60 (1.25, 2.05)
1
–
–
Crude HR (95% CI)
1.04 (0.88, 1.24)
1
1 0.74 (0.61, 0.91) 0.64 (0.49, 0.82)
1.09 (0.86, 1.38) 2.03 (1.50, 2.74)
1.16 (0.86, 1.57)
0.87 (0.62, 1.21)
1.48 (1.15, 1.91)
1
–
–
Adjusted HR (95% CI)
Tropical Medicine and International Health volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
1 1.06 (0.89, 1.27) 1 1.12 (0.95, 1.32) 263/516 (51.0) 302/547 (55.2) 1 1.07 (0.95, 1.21) 1 1.12 (1.00, 1.26) 564/1112 (50.7) 552/1004 (55.0) Employment status Employed Unemployed
1.00 0.75 0.50
VCT/targeted PITC, eligible for ART VCT/targeted PITC, ineligible for ART
0.25
Kaplan-Meier estimate
Systematic HCT, eligible for ART
Systematic HCT, ineligible for ART
0.00
ART, antiretroviral therapy; CI, confidence interval; HR, hazard ratio; HCT, HIV counselling and testing; LTFU, loss to follow-up; PITC, provider-initiated HIV counselling and testing; VCT, voluntary HIV counselling and testing. n (%) show proportion lost to follow-up in each category. Models are adjusted for all variables listed. *LTFU is defined as not returning ≥3 months after the last scheduled visit; this analysis excludes 112 patients who died or transferred prior to 3 months follow-up (N = 2154). †ART eligibility is defined as CD4 count ≤350 cells/ll and is based on the first CD4 value after HIV diagnosis.
1 1.09 (0.92, 1.30) 1 1.12 (0.95, 1.33)
© 2014 John Wiley & Sons Ltd
301/596 (50.5) 250/457 (54.7)
Adjusted HR (95% CI) Crude HR (95% CI) n (%) Crude HR (95% CI) n (%) n (%)
Table 2 (Continued)
Overall
Crude HR (95% CI)
Adjusted HR (95% CI)
VCT/targeted PITC
Adjusted HR (95% CI)
Systematic HCT
K. Clouse et al. Impact of systematic HIV testing
3 0
6 9 3 6 Months since HIV diagnosis
12 9
Figure 3 Kaplan–Meier estimates of time to loss to follow-up (LTFU) among 2154 individuals newly diagnosed with HIV, by HIV testing strategy and eligibility for antiretroviral therapy (ART) at HIV diagnosis. ART eligibility is defined as the first CD4 value after HIV diagnosis ≤350 cells/ll. VCT, voluntary HIV counselling and testing; PITC, provider-initiated HIV counselling and testing; HCT, HIV counselling and testing. LTFU is identified as not returning at least 3 months after the last scheduled visit; thus, person-time begins accumulating at 3 months. This analysis excludes 112 patients who died or transferred prior to 3 months follow-up (N = 2154).
diagnosed are engaged and retained in care. Independent of HIV testing strategy, overall LTFU was high (50.0%) 1 year after HIV diagnosis, and especially high among those ineligible for ART at diagnosis (59.9%). These findings correspond to earlier studies in the sub-Saharan African region observing LTFU ranging between 13% and 75% (Cornell et al. 2010; Rosen & Fox 2011; Mugglin et al. 2012) and with especially high LTFU among individuals ineligible for ART (Larson et al. 2010; Lessells et al. 2011; Rosen & Fox 2011; Macpherson et al. 2012; Mugglin et al. 2012). One positive finding is that LTFU did not increase after implementation of routine testing. This is in contrast to a study in Zambia which found significantly higher 6-month LTFU after transitioning from VCT to PITC (Topp et al. 2012). When examining risk factors for LTFU, we again found that these were independent of HCT strategy. Men and younger adults were at increased risk of LTFU, a finding confirmed by other retention studies throughout Southern Africa (Fox & Rosen 2010; Kranzer et al. 2010; Boyles et al. 2011). A notable difference between HCT strategies was the reduced LTFU among those diagnosed by systematic HCT and eligible for ART. This may be due to the ‘normalisation’ of HIV testing under the systematic approach, which could reduce the stigma that 1417
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
is still associated with HIV care in many settings. Another difference observed was dropout occurred faster among those ineligible for ART when diagnosed through systematic HCT. This may be because individuals diagnosed by VCT or targeted PITC are more motivated to stay in care even without direct benefits (ART initiation) as they came forward themselves for HIV testing or required care for other reasons (such as TB, antenatal and postnatal care). Independent of the reason, high LTFU in this population highlights the importance of tailored counselling and care programs for clients not yet eligible for ART so that they remain engaged in care until they are eligible to initiate treatment. High dropout prior to ART initiation may also provide support for earlier initiation of ART through expanded eligibility criteria or immediate treatment for all new cases (‘test-and-treat’). Despite the large sample size, limited missing data and the ability to assess the implementation of systematic HIV testing under real-world conditions at a busy clinic outside of the confines of a clinical trial, our results should be interpreted within the context of the study limitations. Because of this observational ‘real-world’ before/after design, differences between groups cannot conclusively be assigned solely to the differences in HCT strategies and may be influenced by other changes occurring between 2010 and 2012. We are reporting results from one clinic only, which can limit generalisability. However, our retention findings and participant characteristics are similar to those from other settings in southern Africa, suggesting that our study population is likely representative for other urban settings in sub-Saharan Africa. Patient-level data in South Africa are not linked across facilities; we therefore cannot know if clients who drop out of care at WHWC are receiving care elsewhere, a limitation common to most retention studies in the African region. We limited our analysis to patient characteristics and retention in care of those newly diagnosed with HIV. We are unable to comment on the proportion testing HIV-positive during the two periods because the denominator of total individuals tested is unavailable. File reviews relying on paper clinic records may be prone to selection bias and/or outcome misclassification due to missing records; in our sample, after multiple attempts to locate missing files, a similar proportion of files was considered lost in each arm, representing
doi:10.1111/tmi.12387
volume 19 no 12 pp 1411–1419 december 2014
Impact of systematic HIV testing on case finding and retention in care at a primary care clinic in South Africa Kate Clouse1, Colleen F. Hanrahan1, Jean Bassett2, Matthew P. Fox3,4,5, Ian Sanne3 and Annelies Van Rie1 1 Department of Epidemiology, University of North Carolina Gillings School of Global Public Health, Chapel Hill, NC, USA 2 Witkoppen Health and Welfare Centre, Johannesburg, South Africa 3 Health Economics and Epidemiology Research Office, Department of Medicine, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa 4 Department of Epidemiology, Boston University, Boston, MA, USA 5 Center for Global Health & Development, Boston University, Boston, MA, USA
Abstract
objective Systematic, opt-out HIV counselling and testing (HCT) may diagnose individuals at lower levels of immunodeficiency but may impact loss to follow-up (LTFU) if healthier people are less motivated to engage and remain in HIV care. We explored LTFU and patient clinical outcomes under two different HIV testing strategies. methods We compared patient characteristics and retention in care between adults newly diagnosed with HIV by either voluntary counselling and testing (VCT) plus targeted provider-initiated counselling and testing (PITC) or systematic HCT at a primary care clinic in Johannesburg, South Africa. results One thousand one hundred and forty-four adults were newly diagnosed by VCT/PITC and 1124 by systematic HCT. Two-thirds of diagnoses were in women. Median CD4 count at HIV diagnosis (251 vs. 264 cells/ll, P = 0.19) and proportion of individuals eligible for antiretroviral therapy (ART) (67.2% vs. 66.7%, P = 0.80) did not differ by HCT strategy. Within 1 year of HIV diagnosis, half were LTFU: 50.5% under VCT/PITC and 49.6% under systematic HCT (P = 0.64). The overall hazard of LTFU was not affected by testing policy (aHR 0.98, 95%CI: 0.87–1.10). Independent of HCT strategy, males, younger adults and those ineligible for ART were at higher risk of LTFU. conclusions Implementation of systematic HCT did not increase baseline CD4 count. Overall retention in the first year after HIV diagnosis was low (37.9%), especially among those ineligible for ART, but did not differ by testing strategy. Expansion of HIV testing should coincide with effective strategies to increase retention in care, especially among those not yet eligible for ART at initial diagnosis. keywords HIV/AIDS, retention in care, loss to follow-up, voluntary counselling and testing, provider-initiated HIV counselling and testing, HIV counselling and testing
Introduction South Africa has more people living with HIV than any other country in the world (Joint United Nations Programme on HIV/AIDS (UNAIDS) 2010). In a generalised HIV epidemic WHO recommends that providerinitiated HIV counselling and testing (PITC) should be the standard of care and performed irrespective of symptoms or reason for clinic visit (WHO/UNAIDS 2007). South African HIV testing guidelines recommend PITC in all settings (Republic of South Africa Department of Health 2010), but PITC continues to mainly target specific clinic populations, such as those
© 2014 John Wiley & Sons Ltd
attending antenatal, tuberculosis (TB) or sexually transmitted infection (STI) services (Pope et al. 2008; Kharsany et al. 2010; Leon et al. 2010b). Benefits of PITC include increased uptake of testing (Bassett & Walensky 2010; Leon et al. 2010b; Dalal et al. 2011; Kennedy et al. 2013; Roura et al. 2013) and earlier HIV diagnosis prior to advanced immunosuppression (Bassett & Walensky 2010; Republic of South Africa Department of Health 2010). Better integration of HIV testing with routine clinic services can also contribute to normalising HIV care and reduce HIV exceptionalism and stigma (Leon et al. 2010a; Dalal et al. 2011). 1411
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
A well-documented challenge of providing effective HIV services in sub-Saharan Africa is retention in care. Retention has been shown to be especially poor in the period between HIV diagnosis and antiretroviral therapy (ART) initiation (Bassett et al. 2010; Lessells et al. 2011; Rosen & Fox 2011; Mugglin et al. 2012; Clouse et al. 2013), and somewhat better but still problematic in the period after ART initiation (Cornell et al. 2010; Fox & Rosen 2010; Kranzer et al. 2012). One concern related to expanding testing is that those who are identified may not be adequately linked to care or motivated to engage in care, particularly those who feel healthy or are ineligible for ART, thus exacerbating retention challenges (Roura et al. 2013). To understand the impact of systematic HIV counselling and testing (HCT) on retention in care and patient clinical outcomes, we compared data from a period when voluntary testing (VCT) plus targeted PITC was implemented at a primary care clinic to a period when a strategy of clinicwide, systematic HCT was implemented at the same clinic.
Methods Study setting Witkoppen Health and Wellness Centre (WHWC) is a high-volume primary healthcare clinic in Johannesburg, South Africa, which primarily serves a population from neighbouring formal and informal settlements characterised by widespread poverty, frequent mobility and high HIV and TB burden. Prior to 2011, WHWC offered VCT, plus targeted PITC within antenatal, TB and STI services, or at any time at the request of a care provider. Individual pre- and post-test counselling accompanied HIV testing. In January 2011, the clinic’s HIV testing strategy was changed to systematic HCT for all adult clinic clients. At the start of each visit, prior to any clinic activity, current HIV status was assessed by three specially-assigned clinic staff. All adult patients with a negative result ≥3 months prior to the current visit and all clients new to the clinic were sent first to group pre-test HIV counselling. This was followed by a one-on-one interaction with an HIV counsellor during which the individual was asked to accept or decline rapid on-site HIV testing. Individual post-test counselling followed testing. HIV testing for both strategies occurred in the same testing room at the clinic, using the same staff component and same HIV test method (rapid, point-of-care tests using dried blood spots). For individuals newly diagnosed with HIV at WHWC, blood is drawn for CD4 count and sent to an outside laboratory, and results are shared and discussed at a followup visit at WHWC about 1–4 weeks later. For those with 1412
CD4 counts ≤350 cells/ll, ART initiation occurs 1– 4 weeks after that within the same clinic. After ART initiation, visits are scheduled monthly until the individual achieves viral suppression, and then are scheduled quarterly. For individuals ineligible for ART based on CD4 results, follow-up visits to monitor CD4 values are scheduled about every 6 months. The policy of initiation at ≤350 cells/ll was adopted in mid-2011 throughout South Africa. However, WHWC adopted the policy earlier, in early 2010, and during data collection, all patients in the earlier cohort were individually assessed for ART eligibility, per clinician decision, to ensure correct classification. Study design We conducted an observational cohort study of adults (age ≥18) newly diagnosed with HIV infection through the VCT plus targeted PITC strategy (January–June 2010) or through systematic HCT (February–July 2012). File reviews were conducted at least 15 months after each patient tested HIV-positive. The primary exposure of interest is the HIV testing strategy: VCT plus targeted PITC or systematic HCT. The outcomes of interest were the case finding yield (number of new HIV diagnoses during the 6 month periods), characteristics of individuals newly diagnosed with HIV infection, ART initiation and retention in care 12 months after HIV diagnosis. Loss to follow-up (LTFU) was defined as not returning to the clinic ≥3 months after the last scheduled visit. Patients who requested transfer to another clinic were identified as ‘transferred out’, considered a separate outcome category and not counted as LTFU. Statistical analysis Patient characteristics at the time of testing HIV-positive are presented using counts and proportions for categorical variables, and medians and interquartile ranges for continuous variables. Retention outcomes 1 year after HIV testing are reported as proportions and 95% confidence intervals (95%CI), with P-values determined by chi-square testing. We created crude Kaplan–Meier curves of time to LTFU by study period and compared the distributions using the log-rank test of equality for survival functions. We identified effect measure modification of the association between testing strategy and LTFU by pregnancy status and ART eligibility status. To account for this, we stratify results by a six-level variable combining gender, pregnancy and ART eligibility. We used Cox proportional hazards models methods, adjusting for
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
covariates selected as potential confounders based on a review of the literature and causal diagrams, to estimate adjusted hazard ratios (aHR) and 95%CI. Ethical consideration This study was approved by the institutional review boards of the University of North Carolina at Chapel Hill and the University of the Witwatersrand.
Results Description of individuals recently diagnosed with HIV In total, 2266 individuals were newly diagnosed with HIV during the two 6-month study periods: 1144 (50.5%) individuals were newly diagnosed by VCT with targeted PITC and 1122 (49.5%) by systematic HCT. For both 6-month periods, over 90% of files of individuals newly diagnosed with HIV could be located and reviewed for inclusion in the analysis. The median age for both groups was 32 years, and the distributions by age category were similar (Table 1). Almost half (43.8%) had immigrated to South Africa from neighbouring countries and unemployment was high (46.7%). Independent of the HIV testing strategy, most new HIV diagnoses were made in women (66.9% in VCT/PITC group and 65.3% in HCT group). The proportion of pregnant women was higher among those diagnosed through VCT/PITC (39.2%) than those diagnosed through systematic HCT (26.3%) (P < 0.0001). The median CD4 count was 256 cells/ll, and similar between groups (251 for VCT/ PITC and 264 for systematic HCT, P = 0.19). Independent of HIV testing strategy, about two-thirds of individuals newly diagnosed with HIV were eligible for ART based on a CD4 count ≤350 cells/ll: 67.2% for VCT/PITC and 66.7% for systematic HCT strategy, P = 0.80, Figure 1). Overall, 61.2% of those eligible initiated ART, 58.7% of those diagnosed by VCT/PITC and 63.7% among those diagnosed by systematic HCT (P = 0.05, Figure 1). Retention in care during the first year following HIV diagnosis Only 37.9% (95%CI: 35.9–39.9%) of individuals newly diagnosed with HIV were retained in care at WHWC during the 12-month period following HIV diagnosis, with similar proportions independent of HIV testing strategy: 36.5% (95%CI: 33.7–39.3%) for VCT/PITC
© 2014 John Wiley & Sons Ltd
Table 1 Characteristics of 2266 individuals newly diagnosed with HIV at a primary care clinic in Johannesburg, South Africa, overall and by HIV testing strategy
Patient characteristic Sex, n (%) Female Male Age at HIV testing, median (IQR) Age at HIV testing, n (%) 18–29 years 30–39 years 40 years and older First CD4 value (cells/ll), median (IQR) First CD4 value, n (%) 350 cells/ll Missing Nationality, n (%) Born in South Africa Born outside of South Africa Missing Employment status, n (%) Employed Not employed Missing Initiated ART*, n (%) No Yes Pregnant at HIV diagnosis, n (%)† No Yes Missing
Total
VCT/targeted PITC (n = 1144, 50.5%)
Systematic HCT (n = 1122, 49.5%)
1498 (66.1) 768 (33.9) 32 (27–39)
765 (66.9) 379 (33.1) 32 (27–38)
733 (65.3) 389 (34.7) 32 (28–39)
864 (38.1) 898 (39.6) 504 (22.2)
452 (39.5) 451 (39.4) 241 (21.1)
412 (36.7) 447 (39.8) 263 (23.4)
256 (128–403)
251 (120–397)
264 (131–410)
209 626 639 707 85
106 333 318 349 38
103 293 321 358 47
(9.2) (27.6) (28.2) (31.2) (3.8)
(9.3) (29.1) (27.8) (30.5) (3.3)
(9.2) (26.1) (28.6) (31.9) (4.2)
1240 (54.7)
625 (54.6)
615 (54.8)
992 (43.8)
485 (42.4)
507 (45.2)
34 (1.5)
34 (3.0)
1168 (51.5) 1059 (46.7) 39 (1.7)
541 (47.3) 573 (50.1) 30 (2.6)
627 (55.9) 486 (43.3) 9 (0.8)
567 (38.8) 894 (61.2)
307 (41.3) 437 (58.7)
260 (36.3) 457 (63.7)
991 (66.2) 493 (32.9) 14 (0.9)
465 (60.8) 300 (39.2) 0
526 (53.1) 193 (26.3) 14 (1.9)
0
ART, antiretroviral therapy; HCT, HIV counselling and testing; IQR, interquartile range; PITC, provider-initiated HIV counselling and testing; VCT, voluntary HIV counselling and testing. *Among patients eligible for ART per CD4 value ≤350 cells/ll (n = 1461). †Among female participants (n = 1498).
1413
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
(a) 100%
VCT/targeted PITC period
Systematic HCT period
90% P = 0.80
P = 0.05
80% 70%
67.2%
66.7% 58.7%
60%
63.7%
P = 0.02 47.8%
50%
41.7%
40% 30% 20% 10% 0% Eligible for ART
Initiated ART within 12 months Retained in care at WHWC12 of HIV diagnosis months after HIV diagnosis
(b) 100%
VCT/targeted PITC period
Systematic HCT period
90% 80% 70% 60% 50% 40%
P = 0.80 32.8%
33.3%
30%
P = 0.53 29.2%
27.1%
20% 10% 0% Ineligible for ART
Retained in care at WHWC 12 months after HIV diagnosis
and 39.4% (95%CI: 36.6–42.3%) for systematic HCT. Independent of testing strategy, few individuals were transferred to another healthcare facility (8.8%, 95%CI: 7.7–10.0%) and very few were known to have died (3.2%, 95%CI: 2.6–4.0%). About half were lost to follow-up within the first year of HIV diagnosis: 50.5% (95%CI: 47.6–53.4%) for VCT/PITC and 49.6% (95% CI: 46.6–52.3%) for systematic HCT. Kaplan–Meier estimates of time to LTFU were also similar by HCT strategy (log-rank test P = 0.74, Figure 2). When assessing factors associated with LTFU (Table 2), we found no difference in the overall hazard by HIV testing strategy (aHR 0.98, 95%CI: 0.87–1.10). Independent of gender or pregnancy status, those who were ineligible for ART were 18–40% more likely to be LTFU compared to their ART-eligible counterparts. Older individuals (30+ years) were 31–40% less likely to be LTFU compared to those age 18–29 years. Those born outside of South Africa were at slightly higher risk of 1414
Figure 1 Retention in care of 2181 newly diagnosed individuals by voluntary counselling and testing (VCT) or targeted provider-initiated HIV counselling and testing (light grey), or systematic HIV counselling and testing (HCT) (dark grey). (a) Among individuals eligible for antiretroviral treatment (ART) at time of HIV diagnosis (CD4 count ≤350 cells/ll) and (b) among individuals not yet eligible for ART at the time of HIV diagnosis (CD4 count >350 cells/ll). WHWC, Witkoppen Health and Welfare Centre. Eighty-five (3.8%) patients missing a CD4 count at HIV diagnosis were excluded.
LTFU (1.17, 95%CI: 1.04–1.32) than those born in South Africa. There was no relationship between employment status and LTFU. Examining factors associated with LTFU separately for each HCT strategy, risk factors were similar with a few exceptions. Among non-pregnant women diagnosed under systematic HCT, those ineligible for ART were 48% more likely to be LTFU than those eligible for ART (aHR 1.48, 95%CI: 1.15–1.91), whereas ART eligibility did not determine the hazard of LTFU among non-pregnant women diagnosed by VCT/PITC (aHR 0.88, 95% CI: 0.65–1.19). The strongest predictor of LTFU was among men ineligible for ART at HIV diagnosis who were diagnosed by systematic HCT (aHR 2.02, 95% CI: 1.50–2.74); a much stronger association than that observed among ART-ineligible men diagnosed by VCT/ PITC (aHR 1.19, 95% CI: 0.84–1.69). Overall, individuals ineligible for ART at the time of HIV diagnosis had poorer retention than those eligible
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
0.75 0.50 0.25
VCT/targeted PITC
Log-rank P = 0.74
3
6
9
12
Months since HIV diagnosis Number at risk VCT/targeted PITC Systematic HCT
for ART (28.2% vs. 44.7%, P < 0.0001). Among individuals eligible for ART, those diagnosed by systematic HCT tended to have better retention in care than those diagnosed by VCT/PITC (47.8% vs. 41.7%, P = 0.02 – Figure 1; log-rank test P = 0.05 – Figure 3). In contrast, the proportion of individuals retained in care 12 months after diagnosis among those ineligible for ART was similar between HCT strategies, (27.1% vs. 29.2%, P = 0.53 – Figure 1) but dropout occurred sooner after HIV diagnosis among those diagnosed by systematic HCT (log-rank test P = 0.03 – Figure 3). Discussion Systematic HCT at primary care facilities is recommended by the WHO but rarely implemented. In contrast to our expectation, we did not observe an increase in HIV case finding yield, found no difference in demographic and clinical characteristics of individuals newly diagnosed with HIV, and observed similar overall rates of retention in care during the first 12 months following HIV diagnosis when moving from VCT with targeted PITC to systematic HCT for all adult clients at a primary care clinic in Johannesburg, South Africa. The number of clinic clients newly diagnosed with HIV during a 6-month period was almost exactly the same for both HIV testing strategies (1144 vs. 1122), a surprising finding considering the explicit expansion of HIV testing. Others have shown mixed results regarding case finding after transitioning from VCT to PITC, reporting both increases (Bassett et al. 2007) and decreases (Topp et al.
© 2014 John Wiley & Sons Ltd
Systematic HCT
0.00
Figure 2 Kaplan–Meier estimates of time to loss to follow-up (LTFU) by HIV testing strategy among 2154 individuals newly diagnosed with HIV at a primary care clinic in Johannesburg, South Africa. VCT, voluntary HIV counselling and testing; PITC, provider-initiated HIV counselling and testing; HCT, HIV counselling and testing. LTFU is defined as not returning at least 3 months after the last scheduled visit; thus, person-time begins accumulating at 3 months. This analysis excludes 112 patients who died or transferred prior to 3 months followup (N = 2154).
Kaplan-Meier estimate
1.00
K. Clouse et al. Impact of systematic HIV testing
1092 1062
644 611
502 514
418 442
2011). The reason for this surprising finding is unclear. Because this was not a randomised controlled trial, other factors could have contributed to our observation. For example, the national South African HIV testing campaign launched in April 2010 could have decreased the proportion of individuals with unknown HIV status between 2010 and 2012. The characteristics of those newly diagnosed with HIV also did not substantially change between strategies. Independent of testing strategy, women comprised nearly two-thirds of those testing HIV-positive. The proportion of those newly diagnosed that were pregnant women decreased from 39.2% in the VCT and targeted HCT group to 26.3% in the systematic HCT group (P < 0.0001). This may be because pregnant women were one of the key groups targeted for HIV testing prior to systematic HCT. Surprisingly, median CD4 barely increased – from 251 to 264 cells/ll – when expanding HIV testing. This is in contrast to a study in Zambia that found a decrease in median CD4 value after implementing PITC, suggesting that PITC was providing a ‘safety net’ for individuals who did not seek VCT (Topp et al. 2012). Community-based HIV testing studies in South Africa reported higher median first CD4 values (>400 cells/ll) (Auvert et al. 2004; Kranzer et al. 2013). Taken together, these data suggest that, independent of HIV testing strategy, clinic-based HIV testing identifies individuals at more advanced HIV disease stage compared to community-based testing. Increasing the number of people who know their HIV status will only reduce the burden of HIV if those 1415
1416
40 years and older Nationality Born in South Africa Born outside of South Africa
Year of HIV testing Pre-intervention (2010) Post-intervention (2012) Gender/pregnancy and ART-eligibility status† Female, not pregnant at HIV testing, ART-eligible Female, not pregnant at HIV testing, ART-ineligible Female, pregnant at HIV testing, ART-eligible Female, pregnant at HIV testing, ART-ineligible Male, ART-eligible Male, ART-ineligible Age at HIV testing 18–29 years 30–39 years 1.08 (0.88, 1.33) 1.55 (1.29, 1.87) 1.15 (0.98, 1.36) 1.54 (1.23, 1.92) 1 0.69 (0.60, 0.78) 0.60 (0.51, 0.71)
124/244 (50.8)
167/229 (72.9)
261/527 (49.5)
526/827 (63.6) 407/852 (47.8)
552/961 (57.4)
572/1159 (49.4)
201/475 (42.3)
1.26 (1.12, 1.41)
1
1.27 (1.05, 1.53)
161/291 (55.3)
102/161 (63.4)
1
0.98 (0.87, 1.10)
556/1062 (52.4)
239/612 (39.1)
1
Crude HR (95% CI)
578/1092 (52.9)
n (%)
Overall
1.17 (1.04, 1.32)
1
1 0.69 (0.60, 0.79) 0.60 (0.50, 0.72)
1.20 (1.01, 1.41) 1.57 (1.25, 1.97)
1.30 (1.06, 1.58)
0.90 (0.73, 1.12)
1.18 (0.97, 1.43)
1
0.98 (0.87, 1.10)
1
Adjusted HR (95% CI)
290/473 (61.3)
278/585 (47.5)
92/227 (40.5)
285/438 (65.1) 201/427 (47.1)
43/74 (58.1)
140/268 (52.2)
105/140 (75.0)
77/148 (52.0)
63/138 (45.7)
1.44 (1.22, 1.70)
1
1 0.64 (0.54, 0.77) 0.56 (0.44, 0.71)
1.26 (1.00, 1.58) 1.20 (0.86, 1.69)
1.65 (1.29, 2.12)
1.11 (0.85, 1.47)
0.95 (0.71, 1.27)
1
–
–
118/290 (40.7)
–
Crude HR (95% CI)
–
n (%)
VCT/targeted PITC
1.35 (1.14, 1.60)
1
1 0.64 (0.53, 0.79) 0.56 (0.43, 0.73)
1.33 (1.05, 1.69) 1.19 (0.84, 1.69)
1.41 (1.09, 1.84)
0.90 (0.68, 1.21)
0.88 (0.65, 1.19)
1
–
–
Adjusted HR (95% CI)
262/488 (53.7)
294/574 (51.2)
109/248 (44.0)
241/389 (62.0) 206/425 (48.5)
59/87 (67.8)
121/259 (46.7)
62/89 (69.7)
47/96 (49.0)
98/153 (64.1)
121/322 (37.6)
–
–
n (%)
Systematic HCT
Table 2 Multivariate analysis of factors associated with patient LTFU* 1 year after HIV at a primary care clinic in South Africa
1.10 (0.93, 1.30)
1
1 0.73 (0.61, 0.88) 0.64 (0.51, 0.81)
1.05 (0.83, 1.33) 1.91 (1.42, 2.57)
1.40 (1.05, 1.88)
1.02 (0.74, 1.41)
1.60 (1.25, 2.05)
1
–
–
Crude HR (95% CI)
1.04 (0.88, 1.24)
1
1 0.74 (0.61, 0.91) 0.64 (0.49, 0.82)
1.09 (0.86, 1.38) 2.03 (1.50, 2.74)
1.16 (0.86, 1.57)
0.87 (0.62, 1.21)
1.48 (1.15, 1.91)
1
–
–
Adjusted HR (95% CI)
Tropical Medicine and International Health volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
1 1.06 (0.89, 1.27) 1 1.12 (0.95, 1.32) 263/516 (51.0) 302/547 (55.2) 1 1.07 (0.95, 1.21) 1 1.12 (1.00, 1.26) 564/1112 (50.7) 552/1004 (55.0) Employment status Employed Unemployed
1.00 0.75 0.50
VCT/targeted PITC, eligible for ART VCT/targeted PITC, ineligible for ART
0.25
Kaplan-Meier estimate
Systematic HCT, eligible for ART
Systematic HCT, ineligible for ART
0.00
ART, antiretroviral therapy; CI, confidence interval; HR, hazard ratio; HCT, HIV counselling and testing; LTFU, loss to follow-up; PITC, provider-initiated HIV counselling and testing; VCT, voluntary HIV counselling and testing. n (%) show proportion lost to follow-up in each category. Models are adjusted for all variables listed. *LTFU is defined as not returning ≥3 months after the last scheduled visit; this analysis excludes 112 patients who died or transferred prior to 3 months follow-up (N = 2154). †ART eligibility is defined as CD4 count ≤350 cells/ll and is based on the first CD4 value after HIV diagnosis.
1 1.09 (0.92, 1.30) 1 1.12 (0.95, 1.33)
© 2014 John Wiley & Sons Ltd
301/596 (50.5) 250/457 (54.7)
Adjusted HR (95% CI) Crude HR (95% CI) n (%) Crude HR (95% CI) n (%) n (%)
Table 2 (Continued)
Overall
Crude HR (95% CI)
Adjusted HR (95% CI)
VCT/targeted PITC
Adjusted HR (95% CI)
Systematic HCT
K. Clouse et al. Impact of systematic HIV testing
3 0
6 9 3 6 Months since HIV diagnosis
12 9
Figure 3 Kaplan–Meier estimates of time to loss to follow-up (LTFU) among 2154 individuals newly diagnosed with HIV, by HIV testing strategy and eligibility for antiretroviral therapy (ART) at HIV diagnosis. ART eligibility is defined as the first CD4 value after HIV diagnosis ≤350 cells/ll. VCT, voluntary HIV counselling and testing; PITC, provider-initiated HIV counselling and testing; HCT, HIV counselling and testing. LTFU is identified as not returning at least 3 months after the last scheduled visit; thus, person-time begins accumulating at 3 months. This analysis excludes 112 patients who died or transferred prior to 3 months follow-up (N = 2154).
diagnosed are engaged and retained in care. Independent of HIV testing strategy, overall LTFU was high (50.0%) 1 year after HIV diagnosis, and especially high among those ineligible for ART at diagnosis (59.9%). These findings correspond to earlier studies in the sub-Saharan African region observing LTFU ranging between 13% and 75% (Cornell et al. 2010; Rosen & Fox 2011; Mugglin et al. 2012) and with especially high LTFU among individuals ineligible for ART (Larson et al. 2010; Lessells et al. 2011; Rosen & Fox 2011; Macpherson et al. 2012; Mugglin et al. 2012). One positive finding is that LTFU did not increase after implementation of routine testing. This is in contrast to a study in Zambia which found significantly higher 6-month LTFU after transitioning from VCT to PITC (Topp et al. 2012). When examining risk factors for LTFU, we again found that these were independent of HCT strategy. Men and younger adults were at increased risk of LTFU, a finding confirmed by other retention studies throughout Southern Africa (Fox & Rosen 2010; Kranzer et al. 2010; Boyles et al. 2011). A notable difference between HCT strategies was the reduced LTFU among those diagnosed by systematic HCT and eligible for ART. This may be due to the ‘normalisation’ of HIV testing under the systematic approach, which could reduce the stigma that 1417
Tropical Medicine and International Health
volume 19 no 12 pp 1411–1419 december 2014
K. Clouse et al. Impact of systematic HIV testing
is still associated with HIV care in many settings. Another difference observed was dropout occurred faster among those ineligible for ART when diagnosed through systematic HCT. This may be because individuals diagnosed by VCT or targeted PITC are more motivated to stay in care even without direct benefits (ART initiation) as they came forward themselves for HIV testing or required care for other reasons (such as TB, antenatal and postnatal care). Independent of the reason, high LTFU in this population highlights the importance of tailored counselling and care programs for clients not yet eligible for ART so that they remain engaged in care until they are eligible to initiate treatment. High dropout prior to ART initiation may also provide support for earlier initiation of ART through expanded eligibility criteria or immediate treatment for all new cases (‘test-and-treat’). Despite the large sample size, limited missing data and the ability to assess the implementation of systematic HIV testing under real-world conditions at a busy clinic outside of the confines of a clinical trial, our results should be interpreted within the context of the study limitations. Because of this observational ‘real-world’ before/after design, differences between groups cannot conclusively be assigned solely to the differences in HCT strategies and may be influenced by other changes occurring between 2010 and 2012. We are reporting results from one clinic only, which can limit generalisability. However, our retention findings and participant characteristics are similar to those from other settings in southern Africa, suggesting that our study population is likely representative for other urban settings in sub-Saharan Africa. Patient-level data in South Africa are not linked across facilities; we therefore cannot know if clients who drop out of care at WHWC are receiving care elsewhere, a limitation common to most retention studies in the African region. We limited our analysis to patient characteristics and retention in care of those newly diagnosed with HIV. We are unable to comment on the proportion testing HIV-positive during the two periods because the denominator of total individuals tested is unavailable. File reviews relying on paper clinic records may be prone to selection bias and/or outcome misclassification due to missing records; in our sample, after multiple attempts to locate missing files, a similar proportion of files was considered lost in each arm, representing
