Kamolratanakul 1999 THA.

Methods	Generation of allocation sequence: central block random allocation scheme prepared for each of 15 trial sites; random‐number table used. Allocation concealment: none. Blinding: no blinding of assessors. Completeness of follow‐up: 100% (no losses).
Participants	Number: 837 randomized; 73% male. Included: new smear positive adults (aged 15+).
Interventions	Daily supervision: participants chose their supervisor from (a) health centre staff, (b) community members, or (c) family members; for (b) and (c) health workers visited homes twice monthly (first 2 months) or monthly for checking of treatment cards, pill counts, and urine tests. Self‐administration of treatment: 1 month drug supply given at diagnosis and after each follow‐up visit; no treatment supervision between visits. All participants received the same drug regimen: isoniazid‐rifampicin‐pyrazinamide‐ethambutol for 2 months and isoniazid‐rifampicin for 4 months.
Outcomes	Cure rate (primary outcome): completed 6 months antituberculous therapy, with 2 negative sputum exams, 1 at end of treatment. Treatment completion: completed 6 months antituberculous therapy but less than 2 sputum exams. Sputum conversion rate: negative sputum at end of third month. Percentage defaults. Percentage transfers. Caseholding rate.
Notes	Location: Thailand. Date: 1996 to 1997. Duration of DOT not stated. Informed consent not obtained as participants were not told that they were participating in a study. Choice of supervisor for DOT participants: 352 chose a family member; 34 chose a community member; and 24 chose health centre staff. One participant in daily supervision arm excluded due to protocol violation so not strictly intention‐to‐treat.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Generated using random number tables.
Allocation concealment (selection bias)	Unclear risk	Inadequate information.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Investigators not blinded though the patients were blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	There were no exclusions.
Selective reporting (reporting bias)	Unclear risk	All outcomes stated in the methodology are reported.
Other bias	Unclear risk	Not applicable.