1. Effects of patient tracing on estimates of lost to follow-up, mortality and retention in antiretroviral therapy programs in low-middle income countries: a systematic review James H. McMahon 1,2, Julian H. Elliott 1,3,4, Steven Y. Hong 2, Michael R. Jordan 2 1 Infectious Diseases Unit, Alfred Hospital, Melbourne, Australia; 2 Department of Public Health and Community Medicine, Tufts University School of Medicine, Boston, Massachusetts, USA; 3 Department of Medicine, Monash University, and 4 Burnet Institute, Melbourne, Australia 1

2. Background Frequently reported outcomes for populations receiving ART include the number of patients: –Alive and on ART –Died –Transferring care from one facility to another (‘transfer out’) –Stopping ART (physician directed or patient initiated) but remaining in care –Lost to follow-up (LTFU) 2

3. Background - Definitions LTFU - generic term for patients initiating ART with unknown treatment outcomes –Unreported deaths –Unknown transfer of care without documentation –Disengagement from care Retention on ART: patients alive and receiving ART 1 –Retained on ART = 1 – LTFU - died - stopped ART Retention at the original site: individuals retained on ART and excludes transfers out 1 –Retained at the original site = 1 – LTFU – died – stopped ART – transfer out 3 1 Fox TMIH 2010, Rosen PLoS Med 2007

4. Background Patient Tracing - Potential benefits: –Improved classification of unknown outcomes –Linking patients disengaged from care back into the health system Methods of tracing: –Telephone tracing –Physical tracing Prior reviews 1 provide summary estimates of LTFU, mortality and retention but have not incorporated the potential for patient tracing to affect these outcomes Or combination of both 1 Fox TMIH 2010, Rosen PLoS Med 2007, Gupta PLoS One 2011, Lawn AIDS 2008 4

5. Objective Compare summary estimates of LTFU, mortality and retention in low- and middle- income countries (LMICs) 12 months after ART initiation in cohorts of patients with and without physical tracing 5

6. Methods Systematic review for studies in LMIC programmatic settings –MEDLINE (2003-2011) –HIV conferences (CROI and IAS 2009-2011) MeSH and search terms for LTFU and retention Included studies: reported proportion LTFU 12-months after ART initiation Excluded studies: majority children, patients received mono- or dual-therapy, not performed in LMICs, clinical trials (non-programmatic setting) 6

7. Methods Tracing activities determined from studies or contacting study authors –Classified as tracing study if physical tracing available for majority of patients Summary estimates –Medians (IQR) if estimates non-normally distributed or; –Weighted means (± SD) if normally distributed Weighting of proportions was by the inverse of its variance [1/(p x [1-p]/n); where p is proportion and n is sample size] –Compared by Student’s t-test if normally distributed, or Wilcoxon rank sum test if non-normal 7

8. Identified studies 261 papers Identified studies 616 conference abstracts Excluded after reviewing titles and abstracts 149 papers Excluded after reviewing titles 334 conference abstracts Full text review 112 papers Included in the review 32 papers Full text review 282 conference abstracts Included in the review 7 conference abstracts 32 papers and 7 conference abstracts included in the review Search strategy and study selection 8 Excluded after reviewing full text 80 papers Excluded after reviewing full text 275 conference abstracts

9. Comparison of summary estimates with and without physical tracing Outcome of interest With tracingWithout tracing P value # Cohorts (n) Starting ART (n) Range of estimates (%) Summary estimate* (%) Cohorts (n) Starting ART (n) Range of estimates (%) Summary estimate* (%) LTFU 25627910.3 - 15.07.6 ± 1.1291248750.8 - 34.815.1 ± 1.7< 0.001 Mortality 25627914.2 – 29.7 10.5 (7.0 – 12.7) 251136931.1 - 15.3 6.6 (4.3 – 9.6) 0.006 Stopped ART 13439750.5 – 5.82.8 ± 0.27108410.8 – 8.53.2 ± 0.80.5 Transfer out 569451.0 – 14.02.7 ± 1.9761951.2 – 14.53.9 ± 1.30.6 Retention on ART 256279158.4 – 88.5 80.0 (76.5 – 84.5) 2511369358.5 – 91.0 75.8 (70.0 – 81.2) 0.04 Retention at original site 256279147.5 – 88.5 80.0 (76.0 – 84.0) 2511369358.5 – 90.6 72.9 (68.5 – 79.8) 0.02 * Values represent median (Q1–Q3), or weighted mean ± SE (estimates weighted by the inverse of their variance) # Comparing summary estimates for the 2 groups of studies (tracing and non-tracing) by Wilcoxon rank-sum test for medians or student’s t test for weighted means Notes: LTFU, lost to follow up; ART, antiretroviral therapy 9

10. Discussion  LTFU and  mortality with physical tracing –Uncertain by how much the  LTFU was a result of re-engagement into care versus re-classification of unknown outcomes However, in addition to  LTFU and  mortality, we report  in retention at the original site –Suggests tracing may  re-engagement in care Retention at the original site definition accounts for re- classification of lost patients as died or transferred out 10

11. Discussion  re-engagement would lead to beneficial effects of ART 1 –survival, fewer opportunistic infections, limiting treatment interruptions (minimizing emergence of HIV drug resistance),  in community HIV viral load Cost-effectiveness (CE) of tracing not known –Prior CE analyses on reducing LTFU have not considered tracing 2 11 1 Pallella NEJM 1998, Parienti CID 2004, Oyugi AIDS 2007, Das PLoS One 2010, Montaner JAIDS 2010, Andrews JID 2012. 2 Losina PLoS Med 2009

12. Discussion Difference in summary estimates emphasizes the importance of knowing whether physical tracing occurs within an ART program or clinic when interpreting LTFU, mortality or retention data 1 12 1 2006 WHO IMAI guidelines, 2010 WHO HIVDR Early Warning Indicators, 2009 UNGASS indicators, 2009 PEPFAR indicators

13. Limitations ART clinics with physical tracing may have  resources resulting in improved outcomes –Review of randomized controlled trials (RCTs) with tracing interventions may provide more accurate assessments of the impact of tracing on LTFU, mortality and retention –RCTs not found  Needed to quantify benefits and CE Transfer out data available in a minority of studies –Estimates of retention at the original site could differ if complete transfer out data available –Emphasises the importance of understanding transfer out to accurately interpret estimates of retention 13

14. Conclusions Physical tracing leads to: –  unknown outcomes –Suggests improved re-engagement in care Critical need for studies to assess tracing interventions for: –Ability to improve re-engagement of patients on ART –Optimal methods of tracing –Cost effectiveness Programs providing ART in LMICs should consider physically tracing patients who have unknown outcomes as an intervention to improve individual outcomes and programmatic evaluation of populations receiving ART 14

15. Acknowledgements Financial support –National Health and Medical Research Council Postgraduate Scholarship - J.H.M –National Institutes of Health 5K23AI074423-04 - M.R.J., 1K23AI097010-01A1 - S.Y.H. In addition to study authors –Tufts Medical Center / Tufts University Christine Wanke –Alfred Hospital / Monash University Sharon Lewin –World Health Organization, HIV Department Silvia Bertagnolio 15