Sarna 2008.

Methods	Randomized controlled trial
Participants	The study location was outpatient HIV treatment clinics in Mombasa, Kenya 116 participants were randomized to the intervention group and 118 participants were randomized to the control group The inclusion criteria were ART‐naive adults (> 18 years) residing in Mombasa who were eligible for ART (CD4 cell count < 200 cells/mm3 or WHO clinical stage 3 or 4)
Interventions	Intervention: MODIFIED DIRECTLY OBSERVED THERAPY (M‐DOT) Patients in the intervention arm received M‐DOT services for the first 24 weeks of ART, in addition to standard care. Although consideration was given to provision of home‐based M‐DOT, formative research revealed that patients were concerned this would undermine their privacy and confidentiality. They also believed that clinic visits would provide beneficial access to healthcare workers, so M‐DOT services were provided during twice‐weekly clinic visits. To enhance convenience, participants could select 1 of 6 health centers for their M‐DOT visits. At each visit, participants met with M‐DOT observers (nurses) who observed ingestion of the patient's ART dose, inquired about difficulties encountered, and provided individualized adherence support. Used medication containers were also collected, pill counts recorded, and medication dispensed for the subsequent 3 or 4 days. During the monthly clinic visits, efforts were made to ensure that participants in the M‐DOT and control group received equal contact time with study staff to minimize the likelihood that any differences observed would be due to nonspecific effects of increased attention given to the intervention group. CHW traced M‐DOT participants who missed visits and brought medications to participants who were unable to visit the facility. After cessation of M‐DOT at week 24, patients had no further contact with M‐DOT observation centers. Like the control group, they collected monthly medication refills at recruitment clinics and received standard adherence support during weeks 25 to 72. No compensation was given for study participation, though travel costs were reimbursed for 21 M‐DOT participants with financial constraints (USD 0.65 per visit) Control: STANDARD CARE As standard care, all participants attended 3 one‐on‐one adherence counseling sessions before initiating ART. In these, trained nurse counselors emphasized the importance of adherence; educated patients on the treatment regimen, dosing instructions, side effects, and dietary considerations; tailored the regimen to daily activities; and identified social issues like living conditions and family support. After ART initiation, patients visited treatment centers every 4 weeks for follow‐up. At the first 2 follow‐up visits, general adherence counseling was provided, as was discussion of specific emerging problems with side effects or medication intake; thereafter, counseling was tailored to individual needs. All patients were encouraged to bring a family member or friend to clinic visits and counseling sessions
Outcomes	The measures of adherence were a self reported adherence and clinic‐based pill counts. Self reported adherence (4‐day recall) was assessed 7.0 times (8 weekly from 1 to 48 weeks plus at week 72) and was dichotomized into those having missed or not having missed doses in the past 4 days. Pill counts were made at each M‐DOT visit between weeks 4 and 24 when used medication containers were returned (twice weekly). These counts were aggregated to produce a 4‐week measure. From weeks 25 to 48 and for participants in the control arm (weeks 4 to 48), pill counts were measured every 4 weeks and at week 72 (a total of 13 data points for both groups). Pill counts, measured as a percentage and dichotomized to < 95% and > 95% adherence, were calculated as: number of pills taken/number of pills expected to have taken *100 The patient outcome was depression, assessed at baseline and weeks 24, 48, and 72 using a culturally adapted 21‐item Beck's Depression Inventory (BDI), translated into Swahili. Depression was categorized as none (0 to 9), mild (10 to 18), moderate (19 to 29), and severe (30 to 63) as per BDI guidelines. Body weight was collected monthly and body mass index (BMI) is presented. CD4 cell counts were determined at baseline and weeks 24, 48, and 72 using PARTEC and FACS counters. CD4 measures were extracted from medical records by nurses and the study co‐ordinator. For analysis, cell counts at baseline and week 24 were dichotomized above and below the median. Plasma viral load was measured at 48 and 72 weeks. Viral load was dichotomized to those with, and those without, viral suppression
Notes	―
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A computer‐generated random number sequence was used. "Computer‐generated random number assignment was used, allocating an equal number of participants to treatment and control groups. Allocation concealment was maintained with sequentially numbered, opaque sealed envelopes. Before ART initiation, participants were randomly assigned to study groups, in blocks of 40. Only laboratory personnel were blinded to study group allocation. Standardized data collection tools, staff training, and regular supervision by the study coordinator aimed to ensure that study activities were uniform across sites" (pg 612)
Allocation concealment (selection bias)	Low risk	"Computer‐generated random number assignment was used, allocating an equal number of participants to treatment and control groups. Allocation concealment was maintained with sequentially numbered, opaque sealed envelopes." (pg 612)
Selective reporting (reporting bias)	Unclear risk	No protocol available; although it appears that everything was reported it is difficult to determine this without a protocol
Other bias	Unclear risk	"As the study began when ART programs were first introduced in Kenya, treatment was available for a limited number of patients. With this constraint, the study lacks power to detect smaller but nevertheless clinically important differences in outcomes. Differences in the way pill counts were measured for the 2 groups during the first 24 weeks could have introduced bias in these measures." (pg 613)
Blinding of outcome assessment (detection bias) Adherence measure	Unclear risk	(PRIMARY) PILL COUNT ‐ "Only laboratory personnel were blinded to study group allocation" (pg 612). However, there is insufficient information about who did the pill count
Blinding of outcome assessment (detection bias) Patient outcome	Unclear risk	(PRIMARY) VIRAL LOAD AND CD4 COUNTS ‐ CD4 cell counts were determined at baseline and weeks 24, 48, and 72 using PARTEC and FACS counters. CD4 measures were extracted from medical records by nurses and the study co‐ordinator. No mention of blinding
Blinding of participants (performance bias) Adherence measure	High risk	(PRIMARY) PILL COUNT ‐ Patients not blinded and pill count may be altered by patients. "Some came back with fewer pills than expected due to misplaced pills, repeated ingestion if pills had been vomited, or "pill dumping" (as occasionally admitted to counsellors and CHW)."
Blinding of participants (performance bias) Patient outcome	Low risk	(PRIMARY) VIRAL LOAD AND CD4 COUNTS ‐ This is an objective measure of outcome
Blinding of personnel (performance bias) Adherence measure	Unclear risk	(PRIMARY) PILL COUNT ‐ No information on blinding given. There is insufficient information to permit judgment of 'Low risk' or 'High risk'
Blinding of personnel (performance bias) Patient outcome	Unclear risk	(PRIMARY) VIRAL LOAD AND CD4 COUNTS ‐ No information on blinding given. There is insufficient information to permit judgment of 'Low risk' or 'High risk'
Incomplete outcome data (attrition bias) Adherence measure	Unclear risk	(PRIMARY) PILL COUNT ‐ Reasons for discontinuation are unclear; unclear how this may have affected results
Incomplete outcome data (attrition bias) Patient outcome	Low risk	(PRIMARY) VIRAL LOAD AND CD4 COUNTS ‐ Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups; from M‐DOT group, a total of 89/116 lasted in the intervention; 7 discontinued the intervention, 15 died and 2 were lost to follow‐up. From the standard care, a total of 94/118 lasted until the end; 10 discontinued the intervention, 12 died, and 4 were lost to follow‐up(Figure 1)