Funck‐Brentano 2005

Methods	Non‐randomised, 3‐arm controlled trial Setting: Paediatric university‐tertiary hospital (Paris, France) Adherence measures: Self‐reported problems with compliance: 5‐point perceived compliance area on the Perceived Treatment Inventory (PTI), where 1=low problems with compliance and 5=high problems with compliance Participants were followed for two years after study initiation.
Participants	Adolescents Mean age: 14.2 years (range 12.0‐17.4 years) Female: 63% Receiving ART: 90% Adherence considered not optimum by the medical team: 11%
Interventions	Adolescents met for a 90‐minute session once every 6 weeks for 26 months. The format was open‐ended. Participants were invited to determine their own focus themes for each meeting and encouraged to discuss their feelings spontaneously. The group was led by two therapists, a man and a woman, both trained in psychodynamic and family therapy.
Outcomes	Compliance area on PTI Intervention (group 1) Baseline: 1.0 (1‐2) Change at T2: 0 (‐1 to 0) Control, refused to participate (group 2) Baseline: 1.0 (1‐2) Change at T2: 0 (‐1 to 0) Control, not invited to participate (group 3) Baseline: 1.0 (1‐4) Change at T2: 0 (‐2 to 4) Viral load Intervention (group 1) Percent of participants with viral load ≤200 copies/mL at baseline: 30% Percent at T2: 80% (P=0.063) Control (group 2) Percent at baseline: 33% Percent at T2: 56% (not statistically significant) Control (group 3) Percent at baseline: 50% Percent at T2: 50% CD4 cell count: No significant change was observed in any group (data not shown)
Notes	The main objective of the study was to measure the effect of the intervention on various psychosocial measures, of which compliance to therapy was a small part. Inclusion criteria included: (1) HIV infection either by mother‐to‐child transmission or perinatal transfusion; (2) name of the virus and transmission route known by the patient; (3) evidence of self motivation to participate in a peer support group; (4) no previous participation in a therapy group and (5) adolescent's and guardian's agreement to participate in the study. Of 48 HIV‐infected adolescents being followed in the clinic during the study period, 38 met the inclusion criteria, and 23 were invited to participate. Of those, 10 self‐selected to participate in the intervention (group 1) and 13 declined (group 2). Reasons for declining were not systematically discussed, but the main reason was denial of any problem or needs for group therapy. The remaining 15 were not invited to participate because they lived to far from the clinic to participate in the support groups. They were nevertheless followed up and were included as a comparison group (group 3). Only adolescents who participated at both time points (baseline and 2 years follow‐up) were considered in the analyses; eight adolescents were excluded for this reason. 10% of subjects were not on ART. Details of the regimens were not reported. During the 26 months of the intervention, the number of adolescents fluctuated from one session to the next. Recently referred adolescents fulfilling the selection criteria joined the group after its initiation, and adolescents were allowed to withdraw at any time on the condition that they explain why.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	High risk	The groups self‐selected and were not randomised.
Blinding (performance bias and detection bias) All outcomes	High risk	Neither participants nor investigators were blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Although baseline data were collected for a total of 38 participants, follow‐up data were not available for eight (3 in Group 2 and 5 in Group 3), so these were excluded from the analysis. As reported in Table 3, which included compliance, Group 1 baseline data were available for eight of 10 participants and follow‐up data were available for seven; Group 2 baseline data were available for 10 of 10 participants and follow‐up data were available for nine; Group 3 baseline and follow‐up data were available for all 10. As reported in Table 4, viral load data were available at both time points for 10 (Group 1), nine (Group 2) and 10 (Group 3) participants. No reason is given for the missing data, although the Methods reports that psychological questionnaires missing more than 10% of the items were excluded from analysis.
Selective reporting (reporting bias)	Unclear risk	Patient's adherence as subjectively perceived by the paediatrician was listed as a measure in the Methods section but was only reported at baseline.
Other bias	High risk	Selection bias was a clear problem, as participants self‐selected to participate in the intervention or not.