Jawahar 2013.

Methods	Study design: randomized, open‐label, parallel‐group, 3‐armed, active comparator, equivalence trial Study period: started May 2004 for an anticipated duration of 5 years; terminated early (between February and October 2006) due to high recurrence rates in the shorter treatment arms Recruitment sites: Chennai and Madurai Country where the trial was undertaken: India Length of follow‐up: 24 months after treatment completion
Participants	No. of participants randomized: 429 (of 1200 anticipated) Intervention groups: 259 (gatifloxacin regimen 141; moxifloxacin regimen 118) Control group: 170 Age: < 40 years 72% (gatifloxacin 66%; moxifloxacin 77%; control 73%) Gender: male 74% (gatifloxacin 76%; moxifloxacin 72%; control 72%) Inclusion criteria: Adult patients 18 years or older with newly diagnosed pulmonary tuberculosis with at least 2 positive sputum cultures Resident within a designated study area and permitted home visits Exclusion criteria: Those with previous treatment for tuberculosis exceeding 30 days, weighing < 30 kg, pregnant or lactating women Those with concomitant diabetes mellitus, severe systemic hypertension, epilepsy, serious forms of extrapulmonary tuberculosis, or HIV infection Proportion with HIV seropositivity: nil (excluded) Proportion with cavitation: not reported Baseline drug resistance: isoniazid 7% overall (gatifloxacin 4%; moxifloxacin 1.2%; control 12%); rifampicin 0.2% overall (moxifloxacin 1%); ofloxacin 1.7% overall (gatifloxacin 2%; control 3%); isoniazid and ethambutol 0.4% overall (1% in each intervention arm); isoniazid and ofloxacin 0.2% overall (control 1%)
Interventions	Interventions: 4‐month ATT regimens Gatifloxacin regimen (gatifloxacin replacing ethambutol): N = 141; 136 eligible, 131 completed (93% of those randomized; 96% of those eligible) 2 months of gatifloxacin, isoniazid, rifampicin, and pyrazinamide thrice weekly, followed by 2 months of gatifloxacin, isoniazid, and rifampicin thrice weekly Moxifloxacin regimen (moxifloxacin replacing ethambutol): N = 118; 115 eligible, 113 completed (96% of those randomized; 98% of those eligible) 2 months of moxifloxacin, isoniazid, rifampicin, and pyrazinamide thrice weekly, followed by 2 months of moxifloxacin, isoniazid, and rifampicin thrice weekly Control: 6‐month ATT regimen: N = 170; 165 eligible, 159 completed (94% of those randomized; 96% of those eligible) 2 months of ethambutol, isoniazid, rifampicin, and pyrazinamide thrice weekly, followed by 4 months of isoniazid and rifampicin thrice weekly Dosage: gatifloxacin or moxifloxacin 400 mg, rifampicin 450 or 600 mg, depending on body weight (< 60 kg or ≥ 60 kg), pyrazinamide 1500 mg, and isoniazid 600 mg
Outcomes	Outcomes reported and used in this review: Recurrence of tuberculosis among those with a favourable response at the end of treatment Death from any cause Rates of treatment discontinuation Sputum smear/culture positivity at 8 weeks Proportion with bacteriologically or clinically defined failure Serious adverse events Other adverse events Development of secondary drug resistance to anti‐tuberculosis drugs, identified by drug susceptibility testing
Notes	Funding: Tuberculosis Research Centre of the Indian Council of Medical Research Follow‐up method: a physician examined the patient every month and recorded adherence to treatment, any adverse drug reactions, and the clinical response. Sputum specimens were examined every month by microscopy and culture: 2 (2 overnight and 1 spot) during the treatment phase, and 2 (1 overnight and 1 spot) during the follow‐up phase Treatment supervision: directly observed, thrice‐weekly treatment in all arms Trial registration ID: CTRI/2012/10/003060; retrospectively registered (trial commenced May 2004; trial registered 15/10/2012, after termination) Comment: the data safety monitoring board recommended termination of both intervention arms in 2006 due to high tuberculosis recurrence rates in the 2 arms compared to the standard 6‐month regimen
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote from report: "Restricted random allocation sequences were generated by a biostatistician using random number tables, separately for the two strata and sealed envelopes were used to assign regimens" Quote from trial registration document: "Stratified block randomization"
Allocation concealment (selection bias)	High risk	Quote from report: "Patients were enrolled by the physicians, and when ready for allocation, the biostatistician drew the regimen from sealed envelopes. Allocation was stratified on sputum smear grading and extent of lesions in chest x ray" Quote from report: "The study design envisaged enrolling 400 patients in each arm in a 1∶1:1 ratio. However, due to the non‐availability of one of the test drugs (M), patients were enrolled initially in a 1∶1 ratio in the G and control regimen arms commencing in May 2004. Subsequently, when M became available (May 2005), patients were enrolled to the G, M, and control regimen arms in a 1∶2:1 ratio to compensate for the delay in recruiting to the moxifloxacin arm at the onset" Quotes from correspondence with study authors: "When the first patient on Moxifloxacin was allocated, there were 110 patients randomised to the Gatifloxacin regimen and 110 to the Control regimen. The last patient was allocated to the Gatifloxacin regimen on 3 February 2006" Comment: alteration of recruitment ratios raises serious concerns that allocation concealment was compromised. Even though biostatisticians implemented allocation after clinicians confirmed eligibility, by the time the first patient was recruited to the moxifloxacin regimen in May 2005, 110 allocated to the gatifloxacin regimen (80% of 136 eligible among those finally recruited), and 110 allocated to the control regimen (67% of the 165 eligible) had already been recruited. This would have alerted investigators that most of those to be recruited over the following year would be allocated to the moxifloxacin regimen. In addition, premature termination of the trial, combined with the alteration in allocation ratios, appears to have led to imbalance in the numbers recruited to the gatifloxacin (141), moxifloxacin (118), and control (170) regimens that is not explained, given that block randomization was used. There were also baseline imbalances in proportions resistant at baseline to any of the anti‐tuberculosis drugs tested (6%, 3%, and 16% respectively)
Blinding of participants and personnel (performance bias) All outcomes	Low risk	This was an open‐label trial; given the likelihood that allocation concealment was compromised, treating personnel may have had knowledge of allocation. However, it is unlikely that this led to performance bias
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote from report: "Sputum specimens were given identification laboratory numbers, and bacteriological investigations were carried out by technicians who were blinded to the clinical status of the patient and the regimen. ECG was done every month" Comment: although bacteriologic outcomes were done blind to treatment allocation, clinical efficacy and safety outcomes were undertaken by study personnel. But results at the time of termination favouring the standard and not experimental interventions suggest that detection bias was unlikely
Incomplete outcome data (attrition bias): At the end of ATT (Treatment failure, positive sputum culture, treatment discontinuation, adverse events)	Low risk	In the combined fluoroquinolone arms, 228 of 259 (88%) completed treatment compared to 152 of 170 (89%) in the control arms. Study results did not differ between per‐protocol and modified intention‐to‐treat analyses (that excluded only late‐screening failures). The modified‐ITT analysis included 97%, 98%, and 97% of those in the gatifloxacin, moxifloxacin, and standard ATT arms, respectively. The results of both analyses were consistent
Incomplete outcome data (attrition bias): At the end of follow‐up (Relapse, deaths)	Low risk	In the combined fluoroquinolone arms, 230 of 259 (89%) were assessed for tuberculosis recurrence compared to 154 of 170 (91%) in the control arm. Per‐protocol and modified intention‐to‐treat analyses did not significantly alter the results. Early termination led to recruitment of only a third of the estimated 1200 participants required to prove equivalence, but although this reduces the power of the trial to detect equivalence, lack of differential attrition, with similar reasons for exclusion, is unlikely to affect the reported relative effect estimates
Selective reporting (reporting bias)	Low risk	Although the trial was retrospectively registered, stated outcomes in trial registry documents and in the online study protocol were available in the trial report and do not suggest selective reporting
Other bias	Low risk	The trial was terminated early at the recommendation of the Data Safety Monitoring Board after an interim analysis showed high recurrence rates in the fluoroquinolone arms compared to the control arm. Because this was a planned interim analysis, it is unlikely to have introduced bias, other than that discussed under allocation concealment