Although 1.3 million people in sub-Saharan Africa were receiving antiretroviral therapy (ART) in December 2006 , this represented only 28% coverage. Expanding access to the next 3 million needing treatment requires substantial resources for drugs, medical personnel and infrastructure. Intermittent ART has the potential to reduce long-term toxicity and costs of ART. However, it may increase the risk of HIV-related events, and its impact on long-term adherence and emergence of resistance is unclear.
‘Structured’ treatment interruptions (STI) can be guided by CD4 cell count or of fixed duration: evaluations in chronic infection have demonstrated mixed results [2,3]. The large SMART trial recently terminated early and reported inferiority of STI guided by CD4 cell count with thresholds for ART interruption and restart of 350 and 250 cells/μl, respectively . The CD4 cell count-driven STI arm of the TRIVACAN trial in Côte d'Ivoire with the same CD4 cell count thresholds also terminated early . Other trials with similar designs but higher CD4 cell count thresholds have recently reported that their CD4 cell count-driven STI strategies were not inferior to continuous therapy (CT) [6,7], but these trials were much smaller than SMART. A prerequisite for STI based on CD4 cell count is regular access to reliable testing: high variability of CD4 cell counts will likely limit the feasibility of CD4 cell count-driven STI strategies, particularly in resource-limited settings where access to laboratory monitoring is also often limited [8,9]. In such settings, fixed-duration STI may be more feasible, being less dependent on access to CD4 cell estimation.
This report describes a randomized trial comparing fixed-length STI and CT in adults on first-line ART in Uganda and Zimbabwe that was nested within the ongoing DART trial .
Trial design and participants
STI was compared with CT in an open randomized trial (nested within DART ) in two centres in Uganda (the MRC/UVRI Uganda Research Unit on AIDS, Entebbe, and the Joint Clinical Research Centre, Kampala, with a satellite at the Infectious Diseases Institute, Mulago) and one in Zimbabwe (University of Zimbabwe, Harare). At each, the study was managed by a local trials centre, connected to but not located within the clinical centre. At DART enrolment, all participants received triple drug ART (zidovudine/lamivudine, coformulated as Combivir, plus abacavir, tenofovir DF or nevirapine) and were randomized to clinical monitoring only (CMO) or laboratory plus clinical monitoring (LCM). Haematology, biochemistry (renal and liver function) and CD4 cell counts were performed 12 weekly in all participants: all results were returned to clinicians for the LCM group, whereas haematology/biochemistry results were only returned for the CMO group if grade 4 toxicity occurred or the test was clinically indicated, and CD4 cell count results were not returned. Laboratory tests could be requested for clinical indications at any time. No real-time viral load testing was performed. At enrolment, each participant gave informed consent for randomization to (i) the two monitoring strategies and (ii) STI or CT, if eligible. The protocol included a nonrandomized pilot study of one or two 12-week STI with CD4 cell count estimation every 4 weeks for 100 participants to inform the final STI trial design. DART received ethics committee approval in Uganda, Zimbabwe and the UK (ISCRTN 13968779).
Participants with CD4 cell count ≥ 300 cells/μl at 48 or 72 weeks after ART initiation were randomized to CT or STI with repeated 12 week periods on or off therapy. The main randomization precluded reporting CD4 cell counts for CMO participants, so at 48 weeks (or 72 weeks if not already randomized) CD4 cell count results were communicated as ≥ 300 or < 300 cells/μl to determine eligibility. Other eligibility criteria were: taking ART at the randomization visit (i.e., not stopped ART for toxicity/personal reasons); not pregnant or breastfeeding; no WHO stage 3 or 4 events in the prior 12 weeks; and no intercurrent illnesses. Eligible participants were randomized at the next 4-weekly visit following the qualifying CD4 cell count, which would be 52 or 76 weeks after ART initiation. The STI pilot study (monitored in real-time by two members of the Trial Steering Committee) informed the final design by raising the CD4 cell count threshold from ≥ 250 to ≥ 300 cells/μl and extending the period of initial CT from 28 to 52/76 weeks. Intensive CD4 cell count monitoring (4-weekly) was also performed for a year in the first 100 participants randomized to STI/CT.
Randomization to STI/CT was stratified by centre, weeks of CT (52/76) and allocation to CMO/LCM. The computer-generated sequentially numbered randomization list (variable block sizes) was pre-prepared by the trial statistician and securely incorporated within the trial database, allowing the trial manager at each centre to access the next number but not the whole list. Randomization was undertaken by the clinician phoning the local trial centre.
Interventions, management and follow-up
Participants randomized to STI stopped ART immediately. Twelve weeks later, the same ART was restarted for 12 weeks; the off/on cycles continued until trial termination. Those taking nevirapine had a staggered stop (nucleoside reverse transcriptase inhibitors continued for 7 days after stopping nevirapine ), and nevirapine was dose-escalated at each restart. Co-trimoxazole prophylaxis was used at the discretion of the physician and participant but, if used, was continued during treatment interruptions. All DART participants attended the study clinic every 4 weeks when a nurse administered a standard symptom checklist and adherence questionnaire and dispensed a repeat prescription (28 days) for participants taking ART. The participant was seen by a doctor every 12 weeks or if there was cause for concern. Participants were also asked to return for extra visits if unwell.
Treatment interruptions were timed so that scheduled 12-weekly laboratory tests fell 8 weeks into the STI and 8 weeks after restarting ART, allowing clinicians to shorten or defer interuptions based on clinical judgement or test results. Criteria for early ART restart were HIV-related clinical events including new/recurrent WHO stage 3/4 events, or CD4 cell count < 50 cells/μl (LCM only). Criteria for deferring the next STI were having received < 12 weeks of continuous ART (e.g., participant did not attend scheduled visit for previous ART restart); CD4 cell count < 200 cells/μl 4 weeks before the scheduled STI start (LCM only); and pregnant (currently or within the last 24 weeks) or breastfeeding. Participants with WHO stage 3/4 events or CD4 cell count < 50 cells/μl (LCM only) at any time after STI randomization had no further treatment interruption, nor did participants with clinical/immunological failure on first-line ART.
The primary endpoints were (i) progression to a new (never reported before or during DART prior to STI/CT randomization) WHO stage 4 event  or death, and (ii) serious adverse events, defined as events not related to HIV only, and fatal, life threatening, causing unplanned or prolonging hospitalization, causing permanent or significant disability, or other important medical conditions. There was no requirement to report laboratory or clinical grade 4 adverse events as serious adverse events unless they met these criteria. Secondary endpoints were progression to a new/recurrent WHO stage 4 event or death, CD4 cell count and grade 3/4 adverse events (graded accorded to minor modification of the AIDS Clinical Trials Group criteria ). All primary endpoints were reviewed against protocol-specified criteria for presumptive/definitive diagnosis by an endpoint review committee, blinded to randomized allocation, who also reviewed cause of death. WHO stage 3 events, malaria diagnoses and non-serious adverse events were not reviewed.
Data Safety and Monitoring Committee
The DART Data Safety and Monitoring Committee met every 9–12 months to review data on safety, adherence to randomized strategies and efficacy, and took into account findings from other relevant studies. A flat stopping boundary (the Haybittle–Peto criterion; P < 0.001) was the statistical guide for considering recommending stopping or modifying the trial at interim analysis. The Data Safety and Monitoring Committee reviewed the STI/CT randomization in July 2005 (data to May 2005) based on 146 person-years at risk (no deaths). Following the second review in March 2006 following the termination of SMART  (data to January 2006; total 603 person-years; two deaths), the Data Safety and Monitoring Committee recommended termination of the STI/CT randomisation, and the Trial Steering Committee decided to terminate the STI/CT randomization on 15 March 2006. All STI participants changed to CT.
The intention was to randomize all eligible participants to STI/CT: the original estimated sample size of 600 provided 80% power to demonstrate non-inferiority of STI to CT, defined as the upper limit of the 95% confidence interval (CI) for the hazard ratio (HR, for STI:CT) being no greater than 1.6 (a relative increase of 60%, equivalent to an annual rate of progression of 16% in the STI compared with 10% in the CT arm) with probability (power) 0.80 if STI and CT were truly equivalent, based on 2 year recruitment, 3 year follow-up, loss to follow-up 3.3% per year, and type I error probability (α) 0.05 (two sided).
Time-to-event outcomes were compared across randomized groups using Kaplan–Meier plots, log-rank test and proportional hazards models censoring at the earlier of 15 March 2006 or last follow-up. Categorical variables were compared using exact tests, continuous variables using t-tests and the Wilcoxon rank-sum test. All comparisons between arms were as randomized (intent-to-treat). Baseline values were those nearest to but before and within 120 days of STI/CT randomization (> 99% within 42 days). CD4 cell counts and laboratory parameters were compared across randomized groups over time using generalized estimating equations (independent correlation structure) and the closest measurement to each scheduled visit within equally spaced windows. All P values were two sided. To confirm that no major imbalances affected the results, all analyses were repeated with and without stratification for weeks of ART before STI randomization, centre and randomization to CMO/LCM.
Between 26 July 2004 and 13 March 2006, 817 individuals were randomized to CT (405) or STI (412). One participant (STI) randomized in error with a CD4 cell count < 300 cells/μl was recalled and ART restarted. Three more with CD4 cell count ≥ 300 cells/μl but clinical conditions contraindicating STI were also randomized to STI in error and never interrupted ART (one awaiting surgery that required CD4 cell count > 350 cells/μl, one with Kaposi's sarcoma diagnosed immediately after randomization, and one poorly controlled diabetic). After these four exclusions, 813 participants (STI, 408; CT, 405) were included in analyses (Fig. 1).
Baseline characteristics were broadly similar between the two groups (Table 1). Differences in the numbers recruited into the STI/CT randomization from each centre reflected different total numbers and pre-ART characteristics of DART participants (e.g., on average, higher pre-ART CD4 cell count in Entebbe). Similar proportions of eligible participants were not randomized in the four centres (2–7%).
Median follow-up to 15 March 2006 was 51 weeks in both groups [interquartile range (IQR), 37–64; range, 0–85], with only 21 participants (3%: STI, 12; CT, 9) having less than 12 weeks of follow-up (all but three were randomized within the 12 weeks before trial termination). Nine participants died (STI, 5; CT, 4), and four were lost to follow-up (STI, 3; CT, 1) (Fig. 1). Overall 99% of 386 person-years in CT were spent taking triple-drug ART, compared with 50% of 388 person-years in STI. In the STI arm, this followed the STI schedule, the proportion of time spent on triple drug ART being 4%, 98%, 15%, 98%, 17%, and 99% from 0–12 (off ART), 12–24 (on ART), 24–36 (off), 36–48 (on), 48–60 (off), and 60–72 (on) weeks from randomization, respectively. Of the 408 STI participants, four (1%) never stopped ART (Fig. 1) and 85 (21%), 149 (37%), 139 (34%) and 31 (8%) started one, two, three or four STI cycles, respectively. ART was restarted early by 38 (9%) participants during an STI for WHO stage 3/4 events, low CD4 cell count or intercurrent illness. One participant randomized to CT underwent three STI (wrong allocation at randomization, Fig. 1). Of 174 STI in which nevirapine was stopped, only 10 (6%) did not stagger this with nucleoside reverse transcriptase inhibitor use (eight clinical error, two returned late already off ART). Overall somewhat less time was spent taking co-trimoxazole prophylaxis in CT (42%) than STI (50%), although this varied across centres (2–75% in CT versus 16–85% in STI).
Nine (1%) participants died (STI, 5; CT, 4) (Table 2, Fig. 2a). Three deaths were considered definitely/probably HIV related: two in the STI group (meningitis, strongyloides diarrhoea) and one in the CT group (recurrent cryptococcal meningitis). One death was definitely/probably drug related: in the CT group (acute renal failure judged to be related to tenofovir DF). One death (CT group) was uncertain whether HIV related but was not drug related (cervical carcinoma). The remainder of deaths were unlikely to be related to HIV or drug therapy: one in the STI group was a road traffic accident and three were unknown: two in the STI group and one in the CT group (all died at home).
The incidence of first new WHO stage 4 events or death was increased more than twofold in the STI arm: from nine (2.4/100 person-years) in CT to 24 (6.4/100 person-years) in STI (HR, 2.73; 95% CI, 1.27–5.88; P = 0.007; Table 2 and Fig. 2a). Oesophageal candidiasis was the most frequent first new WHO stage 4 event (STI, 13; CT, 3) followed by extrapulmonary tuberculosis (STI, 4; CT, 1); other new WHO stage 4 events were Pneumocystis jiroveci pneumonia (CT), cryptococcosis (STI) and chronic mucocutaneous herpes simplex (STI). Nine patients died without a confirmed new WHO stage 4 event (STI, 5; CT, 4). There was a trend towards an increased rate of WHO stage 4 events other than oesophageal candidiasis or death (HR, 2.05; 95% CI, 0.77–5.47; P = 0.14).
Similar results were obtained if recurrences of previous events were included in the endpoint (Table 2). Including new and recurrent WHO stage 3 events increased the difference between randomized arms, driven predominantly by differences in oral candidiasis.
In exploratory analyses by subgroups (centre; sex; year, CD4 cell count or age at ART initiation; LCM/CMO; weeks of continuous ART; CD4 cell count or age at STI/CT randomization), there was no evidence of significant heterogeneity in the relative difference between STI and CT, although power was low. There was no evidence of a subgroup favouring the STI arm. There was an excess of new WHO stage 4 events or death in the STI arm in each 12 week period following randomization (heterogeneity P = 0.94), although absolute event rates were higher in the first 12 weeks after randomization [12.0 for STI versus 4.4 for CT; HR, 2.81 (95% CI, 0.89–8.82)] than subsequently [4.6 versus 1.7; HR, 2.64 (95% CI, 0.94–7.40)], possibly reflecting early events in STI participants most at risk.
Hospitalization was required for 46 (11%) STI participants after STI/CT randomization compared with 30 (7%) in the CT arm [exact P = 0.07; median duration 6 days (IQR, 3–14) and 4 days (IQR, 3–6), respectively; rank sum P = 0.16]. On self-reported questionnaire, 172 (43%) in the STI arm and 117 (29%) in the CT arm reported being too sick to work at one or more visits (P < 0.001), and 251 (62%) and 186 (46%), respectively, had visited a healthcare centre (P < 0.001). Although participants made similar numbers of visits to DART clinics in the two arms [STI, median 13 (IQR, 9–16); CT, median 12 (IQR, 9–16); rank-sum P = 0.30), more STI participants saw the doctor (median, 9 doctor visits; IQR, 6–14) than did CT participants (median, 6 (IQR, 4–9) (P < 0.0001; referral reasons not recorded).
CD4 cell count changes
Interrupting ART resulted in rapid and significant declines in CD4 cell count: the mean decrease compared with baseline (STI/CT randomization) was 194, 174 and 165 cells/μl 8 weeks after stopping ART for the first, second and third STI, respectively (Fig. 2b). Eight weeks after restarting ART following the first, second and third STI, there was still a net decrease in CD4 cell count in the STI arm compared with baseline (mean decrease of 70, 81 and 63 cells/μl, respectively). In contrast, after an initial small decline, CD4 cell count increased slightly in the CT arm, likely owing to regression to the mean as eligibility was based on a threshold of ≥ 300 cells/μl (mean increase, 25 cells/μl at 56 weeks after randomization) (global P < 0.001). Around half the STI participants had CD4 cell counts < 200 cells/μl during treatment interruption and approximately 10% had < 100 cells/μl.
There was no difference between the STI and CT arms in time to first serious adverse event (P = 0.78), first new grade 4 adverse event (P = 0.47) or first new grade 3/4 adverse event (P = 0.90; Table 2). The most common grade 3/4 adverse event was neutropenia (Table 3) but only 21 occurrences were grade 4 (STI, 12; CT, 9). However, more ART changes for toxicity occurred in the CT than the STI arm (P = 0.02).
Statistically significant differences between STI and CT arms were observed for changes in haemoglobin, platelets, aspartate aminotransferase and weight after randomization (all global P < 0.0001; Table 4), although mean changes were small and most values remained in the normal range. Haemoglobin was lower in the STI arm during the 12 weeks back on ART (90% participants received Combivir-containing regimens: the rest stavudine/lamivudine). In contrast, platelets and aspartate aminotransferase declined substantially off ART in the STI arm, similarly to CD4 cell counts. There was a mean weight loss of 1.5 kg at the end of the first STI (week 12) compared with gain of 0.1 kg in CT arm (t-test, P < 0.0001): but from the end of the second period back on ART for the STI group (week 48) the difference between the groups had attenuated, with a small mean weight loss in both groups (Table 4). There were smaller differences between groups in neutrophils (global P = 0.03), and no difference in alanine aminotransferase (P = 0.98), creatinine (P = 0.49) or estimated glomerular filtration rate (P = 0.85).
The rationale for investigating ART interruptions during chronic HIV infection is similar in resource-rich and resource-limited settings: if safe, such a strategy could reduce long-term toxicity, costs and treatment fatigue. However, the population of patients treated, disease manifestations and healthcare infrastructure vary greatly between these settings, so approaches evaluated in resource-rich settings cannot necessarily be generalized to all settings. Given the limited access to CD4 cell count monitoring, an STI approach guided by CD4 cell counts will not be feasible in many resource-limited settings. DART, therefore, investigated fixed-duration STI and is the only STI trial to be undertaken in patients also randomized to CMO versus LCM. Review of unblinded data by the Data Safety and Monitoring Committee as well as additional frequent observation of CD4 cell count data in the STI arm by two members of the Trial Steering Committee were instituted as safety measures.
DART is also the largest fixed-period STI trial to be undertaken to date. Other fixed-period STI trials have all had slightly different designs. Two with short cycles off/on ART (7–28 days) were terminated early because of increased resistance [14,15]. The WINDOW trial  randomized 403 patients with HIV-1 RNA < 50 copies/μl and CD4 cell count > 450 cells/μl between CT and a strategy of cycles of 2 months on or off therapy, and found no difference in the proportion with CD4 cell count < 300 cell/μl at 96 weeks. The fixed period STI TRIVACAN trial is the only other study in a resource-limited setting: although the CD4 cell count-driven STI arm was terminated early , the fixed-period STI and CT arms are continuing. Results are awaited with interest, as, compared with DART, participants had higher pre-ART CD4 cell counts and spent less time off therapy in the STI arm (2 months off; 4 months on ART). A number of studies from resource-rich countries have suggested that CD4 cell count-guided interruptions are shorter in those with lower CD4 cell count nadir [17,18], and, therefore, shorter fixed-duration interruptions may be more feasible in patients with low pre-ART CD4 cell count, as in our study (approximately 130 cells/μl), provided that CD4 cell count gains on ART are substantial.
As observed by others [4–7,19], CD4 cell counts in the STI arm fell sharply after stopping ART, although there was some attenuation during later interruptions, possibly because of selection bias if those patients with largest CD4 cell count falls did not interrupt therapy repeatedly. CD4 cell counts rose substantially following ART reinitiation, although we cannot estimate maximum CD4 cell count loss and regain as measurements were taken 8 weeks into each 12 week off/on ART cycle. While our results clearly indicate increased risk of the STI strategy compared with CT in terms of disease progression, it should be noted that over 90% of patients in the STI arm had no WHO stage 3/4 clinical events despite low CD4 cell counts during the interruptions; further, nearly 80% had two or more STI cycles and over 40% had three or four cycles. Although approximately 10% had CD4 cell count < 100 cells/μl during the interruptions, relatively few of these developed WHO stage 3/4 events, and most with clinical events did not experience CD4 cell count decreases to these levels. Whether these patients would have eventually developed clinical events, or whether substantial but very transient CD4 cell count decreases could have been maintained over the longer-term without clinical deterioration remains unknown. In addition, although initial weight loss occurred in the STI arm, differences from the CT arm attenuated over time, a pattern also seen with other laboratory parameters, suggesting that some patients in the STI group were able to tolerate repeated STI cycles. The WHO stage 4 events in the STI arm were mainly oesophageal candidiasis and extrapulmonary tuberculosis; mortality was very similar in both arms. The same WHO stage 4 events, as well as bacterial infections, were the commonest events in the CD4 cell count-guided STI arm of the West African TRIVACAN trial .
Compliance with the randomized strategy was high so that ART exposure was almost 100% in the CT arm compared with 49% in the STI arm, implying significant drug cost savings if such a strategy were successful. Further, there were significantly fewer toxicity-related ART changes in the STI arm. However, these need to be balanced against the increased resource utilization for visits to healthcare personnel and hospital admissions that were reported in the STI arm. Despite staggered stop over 7 days for the NRTI in those receiving nevirapine, the potential for emergence of resistance cannot be excluded, particularly over repeated STI cycles: viral load and genotypes have not been assayed in real-time, although retrospective testing of stored samples would be possible. Increased viral load during treatment interruptions also carries a potential risk for increased HIV transmission, not evaluated in our study.
The largest STI trial to date, the CD4 cell count-driven SMART trial, was stopped early because of a highly statistically significant 2.6-fold relative increase in HIV disease progression in the STI arm , although absolute differences in event rates were small (3.3 versus 1.3 per 100 person-years). As in DART, subgroup analysis could not identify any group that might benefit from this strategy. However, an unexpected finding in SMART was the significant increase in renal, hepatic and cardiac events in the STI arm, postulated to be related to immune activation following viral load rebound. Although we did not observe an excess in such events in DART, these were not specifically sought and other clinical events may have predominated given the lower average CD4 cell counts.
In conclusion, a fixed 12 week off/12 week on ART STI strategy for symptomatic HIV-infected adults initiating ART with low CD4 cell counts in a resource-limited setting, as undertaken in DART, resulted in a greater than twofold risk of disease progression compared with CT and cannot be recommended in this population. Additional data on adherence and patient-reported acceptability, and predictors of response to STI in DART, along with information from completed and ongoing trials in different populations of HIV-infected adults and children may determine whether there are particular populations who could potentially benefit from periods of time off-treatment . In the meantime, all patients in the STI arm in DART have moved to CT and follow-up of the main DART trial, evaluating whether ART can be given safely and successfully in resource-limited settings in the absence of CD4 cell count monitoring and laboratory monitoring for toxicity (unless prompted for clinical reasons), is continuing.
We thank all the participants and staff from all the centres participating in the DART trial.
We particularly acknowledge the contribution of Professor Anne McLaren (Chair of Data Safety Monitoring Committee) who died in July 2007.
Sponsorship: DART is funded by the UK Medical Research Council, the UK Department for International Development (DFID), and the Rockefeller Foundation. GlaxoSmithKline, Gilead and Boehringer-Ingelheim donated first-line drugs for DART.
Analysis and Writing Committee: J. Hakim, D.M. Gibb, A.S. Walker, P. Munderi, C. Kityo, A. Reid, A. Babiker, F. Ssali, A. Kambugu, D. Bray, E. Katabira, P. Muygenyi, H. Grosskurth, C. Gilks, J.H. Darbyshire.
DART Trial Team: A. Latif, J. Hakim, V. Robertson, A. Reid, R. Bulaya-Tembo, C. Chimbetete, A. Jamu, S. Makota, T. Mupudzi, G. Musoro, N. Ngorima, M.J. Pascoe, F. Taziwa, C. Warambwa, E. Chidziva, E. Chigwedere, M. Phiri, L. Chakonza, G. Tinago, H. Chirairo, S. Chitsungo, F. Mapinge, A. Mawora, C. Muvirimi, R. Shoniwa, G.E. Zengeza, J. Chimanzi, J. Machingura, C. Maweni, S. Mutsai, R. Warara, M. Matongo, T. Bafana, N. Mdege, S. Mudzingwa, M. Jangano, I. Machingura, K. Moyo, C. Tsiga, L. Vere, S. Chitsungo, S. Chitongo, K. Manyevere, E. Chivige (University of Zimbabwe, Harare, Zimbabwe); P. Mugyenyi, C. Kityo, D. Tumukunde, F. Ssali, D. Atwine, G. Mulindwa, G. Kabuye, R. Byaruhanga, T. Bakeimyaga-Grace, H. Katabira, E. Nimwesiga, G. Barungi, S. Atwiine, F. Ahimbisibwe, S. Tugume, T. Otim, J. Takubwa, M. Mulindwa, S. Murungi, J. Tukamushaba, D. Muebesa, H. Kyomugisha, J. Kagina, L. Namale (Joint Clinical Research Centre, Kampala, Uganda); H. Grosskurth, P. Munderi, K. Wangati, G. Kabuye, A. Ruberantwari, B. Amuron, D. Nsibambi; R. Kasirye, E. Zalwango, M. Nakazibwe, B. Kikaire, G. Nassuna, R. Massa, M. Aber, M. Namyalo, A. Zalwango, L. Generous, P. Khauka, N. Rutikarayo, W. Nakahima, A. Mugisha, W. Omwony, J. Nakiyingi, S. Muyingo, P. Hughes (MRC Research Unit on AIDS/Uganda Virus Research Institute, Entebbe, Uganda); E. Katabira, J. Oyugi, A. Ronald, A. Kambungu, J. Martin, R. Nalumenya, R. Nairubi, E. Bulume, M. Teopista, C. Twijukye, F. Sematala, H. Byakwaga (Academic Alliance, Mulago Hospital, Uganda); A. Coutinho, B. Etukoit [AIDS Support Organization (TASO), Uganda]; C. Gilks, K. Boocock, C. Puddephatt, D. Winogron (Imperial College); J. Darbyshire, D.M. Gibb, A. Burke, D. Bray, A. Babiker, A.S. Walker, H. Wilkes, M. Rauchenberger, S. Sheehan, L. Peto, K. Taylor (MRC Clinical Trials Unit).
Trial Steering Committee: I. Weller (chair), A. Babiker (trial statistician), S. Bahendeka, M. Bassett, A. Chogo Wapakhabulo, J. Darbyshire, B. Gazzard, C. Gilks, H. Grosskurth, J. Hakim, A. Latif, O. Mugurungi, P. Mugyenyi: Observers S. Jones, E. Loeliger, P. Naidoo, M. Palmer, J. Rooney, J.-M. Steens.
Data and Safety Monitoring Committee: A. McLaren (chair), C. Hill, J. Matenga, A. Pozniak, D. Serwadda
Endpoint Review Committee: T. Peto (chair), A. Palfreeman, M. Borok, E. Katabira, F. Ssali, P. Munderi, A. Reid, D.M. Gibb, H. Bwakaga, D. Winogron.
© 2008 Lippincott Williams & Wilkins, Inc.