Participants
One thousand and nine patients with pulmonary tuberculosis who were admitted to the Government Tuberculosis Sanatorium, Bangalore, India were screened. The criteria for participation were: age between 20 and 55 years, sputum-positivity on direct microscopy,  no history of prior ATT and no evidence of comorbidities or extrapulmonary tuberculosis. Most patients were excluded as they were already being treated with ATT in varying combinations for varying durations. These patients were excluded as prior (often inadequate) treatment would have influ- enced their response to the ATT, with possible drug- resistance.After screening, 73 patients fulfilled the criteria.

Study design

Seventy-three patients were randomly assigned to two groups, yoga and breath awareness. The charac- teristics of the yoga group were as follows: average age 32.2  10.0 years; gender, 17 males; drug sensitivity, 13 sensitive, six resistant, and six contaminated. The characteristics of the breath awareness group were as follows: average age 37.3  9.6 years; gender, 13 males; drug sensitivity, 13 sensitive, seven resistant, and three contaminated. Forty-eight completed the trial. Out of 73 patients, 25 were excluded for the following reasons: (i) they left the sanatorium against medical advice before the 2 months of their prescribed stay was over, and (ii) they failed to stay continuously in the sanatorium for 2 months. The protocol was approved by the directorate of the tuberculosis control program, Department of Health and Family Welfare, Government of Karnataka, India and by the superintendent of the Government Tuber- culosis Sanatorium, Bangalore, Karnataka, India. Written informed consent was obtained from all par- ticipants. All assessments were made at commencement and at the end of 2 months stay in the sanatorium. Treatment compliance was the only measurement that was made every month for the 4 months after the patients were discharged from the sanatorium, during which time patients were asked to continue the intervention to which they had been allocated. However, they were not supervised.

Outcome measures
The outcome measures were symptom scores, body- weight, FVC, FEV /FVC%, sputum microscopy, sputum culture, and CXR. The method and frequency of the measures are detailed in Table 1.

Interventions
The ATT regimen used in the trial consisted of 2 months of an intensive phase of treatment with 0.75 mg of streptomycin (intramuscular injection administered daily), 300 mg of isoniazid (orally, daily), and 150 mg of thioacetazone (orally, daily). This was followed by a 10 month continuous phase of 300 mg isoniazid, and 150 mg thioacetazone (orally, daily).
The yoga group practised 60 min of yoga tech- niques. The 60 min consisted of simple breathing exercises (30 min), specific yoga voluntarily regulated breathing (pranayamas, 20 min), and relaxation in a supine posture (10 min). During simple breathing exercises the following points were noted: (i) inhala- tion and exhalation were in the ratio of 1 : 2, (ii) inhalation and exhalation were synchronized with separate body movements (e.g. stretching the arms wide during inhalation and bringing them together during exhalation), and (iii) awareness of breathing. This was followed by the specific yoga breathing prac- tices (pranayamas). These practices are traditionally believed to reduce body temperature and to be espe- cially useful for diseases of the respiratory system.
During pranayamas, as for breathing exercises, the same inhalation : exhalation ratio (1 : 2) was main- tained with awareness of breathing, along with other voluntary regulations of breathing.  The session ended with relaxation in shavasana (the corpse pos- ture), practised for 10 min.
The non-yoga group performed breath awareness (as a control treatment), so that they would also have intervention and interaction with the instructor. It was hoped that this would reduce any difference between the groups due to the psychological benefits that have been ascribed to additional care.  During the breath awareness session, there were no body movements or changes in the inhalation : exhalation ratio. Patients were asked to direct their awareness to their breathing during the 60-min session, keeping their eyes closed. Breath awareness was selected as it is one of the aspects of yoga that does not include the following: (i) body movements synchronized with breathing (called ‘breathing exercises’), (ii) changing the inhalation : exhalation ratio voluntarily, and (iii) relaxation while supine (called shavasana). Breath awareness need not be given with instructions to relax. The patients had no knowledge or experience of either of the interventions before they started practis- ing them for the study. Both groups had equal inter- action with the instructor. The instructor for both types of sessions (yoga and breath awareness) recorded the participation of the patients every day. The patients who completed the trial were without exception able to attend all the sessions (six sessions per week, for 8 weeks). Hence there were 48 sessions of each intervention, during the 2-month stay at the sanatorium.

Randomization
Eligible patients were randomized by alternate allo- cation on admission to the two groups (initially des- ignated as group 1 and group 2), by a medical officer authorized to admit the patients but who had no role in the trial. The two groups were then designated as the ‘yoga group’ and the ‘breath awareness group’ by the superintendent of the sanatorium who was blinded to the patient allocation, and whose other involvement in the trial was as one of the three read- ers of the coded CXR of the patients.

Masking
It is difficult to assess yoga practices in double blind trials because the intervention requires active partic- ipation of the subject and hence the identities of the interventions become known after allocation.  However, the assessments for symptoms, bodyweight, sputum microscopy, sputum culture, and CXR were made and scored by a blinded investigator. CXR were read independently by three radiologists. Two consistent readings were considered for analysis, using the rating scale detailed in Table 1. This was possible in all cases. The spirometry recordings were made by the instructor administering the interven- tions (yoga and breath awareness), under the super- vision of a doctor who did not otherwise participate in the trial.

Statistical methods
The t-test for unpaired data was used to compare the baseline (i.e. day 1) values for the yoga and the breath awareness groups with respect to: (i) symptom scores, (ii) bodyweight, (iii) FEV/ FVC%. The McNemar’s test for significance of change was used to compare the number of patients with different grades in the two groups at baseline with respect to sputum microscopy, culture and CXR.

The t-test for paired data was used to compare the day 1 and day 60 values for each group separately, with respect to: (i) symptom scores, (ii) bodyweight, (iii) FVC, (iv) FEV , and (v) FEV1/FVC%. Separate analyses were carried out on data for sputum micros- copy comparing day 1 with days 30, 45, and 60. For sputum culture and radiography, day 1 was compared to day 60 ( test was used to compare the number of patients in the two groups who (i) improved by changing from a higher to a lower grade, (ii) showed no change, or (iii) worsened by changing from a lower to a higher grade.