Duration of effectiveness of tuberculosis preventive treatment across various tuberculosis burden settings: a systematic review and individual-participant meta-analysis.
Lauren R Linde, Jeffrey I Campbell, C Robert Horsburgh et al.
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and individual-participant meta-analysis pooled data from 84 128 tuberculosis-infected close contacts in 44 cohort studies across 24 countries to assess how long tuberculosis preventive treatment (TPT) was associated with protection across settings of differing tuberculosis burden. TPT was most protective during the first year, with effects declining over follow-up in all settings. Protection persisted longer in medium- and low-burden settings than in high-burden settings: at 8 years, the adjusted HR was 0·15 (95% CI 0·07-0·33) in medium-burden settings, while in low-burden settings it was 0·65 (0·43-1·00) at 13 years.
Key findings
- At 1 year, adjusted HRs for tuberculosis with TPT were 0·57 (95% CI 0·31-1·07) in high-, 0·14 (0·09-0·21) in medium-, and 0·21 (0·17-0·26) in low-burden settings.
- Protection persisted to 8 years in medium-burden settings (adjusted HR 0·15 [95% CI 0·07-0·33]) and to 13 years in low-burden settings (0·65 [0·43-1·00]).
- In medium- and low-burden settings, TPT completers had greater initial protection than non-completers (adjusted HR at 1 year 0·59 [95% CI 0·36-0·95]); groups were not statistically different at 2 years.
Why it matters
The duration of observed TPT protection differed by background tuberculosis burden, with protection waning relatively quickly in high-burden settings but persisting longer in medium- and low-burden settings. These findings underscore that estimates of preventive benefit over time may not generalise uniformly across epidemiological settings and provide context for interpreting long-term tuberculosis risk among exposed people with evidence of infection.
What to keep in mind
The analysis pooled cohort studies, so treatment was not randomly assigned in the included data; the findings therefore describe associations and may be affected by differences between treated and untreated participants.