Regimens of less than six months for treating tuberculosis

Plain language summary pending

Authors' conclusions: 

Longer periods of treatment (at least up to 6 months) result in higher success rates in patients with active TB, but the differences are small. Under field conditions, where adherence to treatment is a big problem, and shorter regimens might improve adherence, these differences may not be evident. A comparison of <6 months vs. 6 months of treatment under programme conditions would be needed to determine this.

Read the full abstract...

WHO recommends 6 months of treatment in TB programmes.


The purpose of this review is to assess the effects of regimens lasting less than 6 months compared with longer regimens in the treatment of active TB.

Search strategy: 

Search strategy: MEDLINE 1955 to October 2005, Cochrane Infectious Diseases Specialized Register, existing reviews, and researchers in the field. Date of the most recent search: May 2004.

Selection criteria: 

Randomized trials comparing two or more TB drug regimens, in which at least one regimen was <6 months and it was compared with at least one regimen that lasted longer, in any patients with active TB.

Data collection and analysis: 

One reviewer extracted data and assessed trial quality.

Main results: 

Seven trials with a total of 9 comparisons of <6 months (range: 2-5 months) versus longer treatment were included. About 2200 patients were in the shorter regimens and about 1900 in the longer regimens (the same comparison groups were used for more than one shorter regimen, in two studies).

Relapse rates were consistently higher after shorter duration treatment regimens, regardless of the comparison made, though they were all relatively low. Results were significantly better in the longer groups in the meta-analyses of 2, 3, and 4 months of treatment vs longer treatment (Peto OR = 6.1 [95%CI 2.19,17.01], 3.67 [2.42,5.58], 3.64 [1.71,7.75] but not in the single trial of 5 vs. 7 months (Peto OR = 2.24 [0.90,5.59].

Relapse rates after longer (comparison) regimens ranged from 0-7% at one year (or more), and in the shorter treatment arms, they ranged from 1-9% in 8 trials, and18% relapsed in the one remaining.

There was little or no difference in the rates of adverse reactions or toxicity requiring a change of regimen or discontinuation of treatment. The "sterilizing efficacy" at the end of treatment varied little among treatments, providing no predictive value for relapse rates. Few or no deaths were reported in the individual trials, and in no case did enough deaths occur for a comparison of short vs. long regimens.