Table 1 Summary of included studies

Kamolratanakul et al. 1999 [13]

Thailand, mixed urban/rural

RCT (DOT versus SAT) - for DOT arm supervisor self-selected.

837 total in study. 415 randomised to DOT (provider type known for 410; 1 other did not receive DOT as allocated, 352 received DOT via family member), 422 randomised to SAT.

Treatment success:

CB DOT: 27 of 34 (79%)

Clinic DOT: 21 of 24 (88%)

58 included in meta-analysis.

Loss to follow-up:

CB DOT: 5 of 34 (15%)

Clinic DOT: 1 of 24 (4%)

Kironde and Meintjies 2002 [26]

South Africa, mixed urban/rural

Prospective cohort study

769 total in study 50 transferred away from area and not included. 598 new patients (93 of these received SAT) and 121 retreatment patients (not included).

Treatment success:

CB DOT: 164 of 228 (72%)

Clinic DOT: 189 of 277 (68%)

505 included in meta-analysis

Loss to follow-up:

18.7% reported for the study overall but not broken down according to provider type.

Lwilla et al. 2003 [20]

Tanzania, rural

Open cluster RCT

522 total in study and all included in meta-analysis.

Treatment success:

CB DOT: 117 of 221 (53%)

Clinic DOT: 148 of 301 (49%)

Loss to follow-up:

CB DOT: 88 of 221 (40%).

Clinic DOT: 74 of 301 (25%).

Miti et al. 2003 [27]

Zambia, urban

Non-randomised trial

168 total in study and all included in meta-analysis

Treatment success:

CB DOT: 44 of 72 (61%)

Clinic DOT: 47 of 96 (49%)

Loss to follow-up:

CB DOT: 6 of 72 (8%)

Clinic DOT: 22 of 96 (23%)

Niazi and Al-Delaimi 2003 [28]

Iraq, urban

Non-randomised trial (sequential allocation to one treatment arm or the other)

172 total in study and all included in meta-analysis

Treatment success:

CB DOT: 72 of 86 (84%)

Clinic DOT: 59 of 86 (69%)

Loss to follow-up:

CB DOT: 10 of 86 (12%)

Clinic DOT: 9 of 86 (10%)

Nirupa et al. 2005 [31]

India, rural

Retrospective cohort study

3019 total in study

Treatment success:

2661 (88%) could be contacted for the study. Treatment results for only new sputum positive TB patients, N = 1131. 28 patients received SAT. Outreach workers (N = 238) excluded as neither CB DOT nor clinic DOT.

CB DOT: 526 of 666 (79%)

Clinic DOT: 147 of 199 (74%)

865 included in meta-analysis

Loss to follow-up:

CB DOT 92 of 666 (14%)

Clinic DOT: 34 of 199 (17%)

Singh et al. 2004 [29]

India, urban

Retrospective cohort study

617 total in study and all included in meta-analysis

Treatment success:

CB DOT: 110 of 141 (78%)

Clinic DOT: 367 of 476 (77%)

Loss to follow-up:

CB DOT: 21 of 141 (15%)

Clinic DOT: 69 of 476 (14%)

Tripathy et al. 2013 [30]

India, urban

Retrospective cohort study

2099 total in study

Treatment success:

Treatment cards of 1864 (89%) available for evaluation. Patients supervised by physicians (N = 95) removed from CB DOT results.

CB DOT: 475 of 509 (93%)

Clinic DOT: 951 of 1260 (75%)

1769 included in meta-analysis

Loss to follow-up:

CB DOT: 13 of 509 (3%)

Clinic DOT: 88 of 1260 (7%)