Treatment of latent tuberculosis infection: A network meta-analysis
In many countries with a low incidence of tuberculosis (TB), many new cases emerge as a result of reactivation of latent TB infection (LTBI), which is often acquired in high-incidence areas or from recent exposure in occasional outbreaks. Therefore, such countries have had a renewed interest in LTBI screening and treatment, generally for groups at particularly high risk for reactivation, such as contacts of patients with pulmonary TB, persons who are immunocompromised, and migrants from high-incidence areas.
Although efficient and safe, LTBI treatment regimens are lengthy. It is thus essential to offer the least toxic and shortest possible effective regimen to ensure high completion rates. Globally, it is most common to use 6 to 9 months of isoniazid (INH) monotherapy; in the United States, 9 months is recommended. Three months of INH plus rifampicin (RMP) and 3 to 4 months of RMP alone may be equally as efficacious as INH regimens. Twelve weeks of INH plus rifapentine (RPT) has been shown to be noninferior to 9 months of INH alone and is now included in Centers for Disease Control and Prevention guidelines.
To date, all reviews of LTBI treatment have used conventional meta-analyses. By allowing only direct comparisons between regimens, such analyses were severely limited in the inferences they could make about relative efficacy and toxicity. Bayesian hierarchical models use a network approach that also enables the indirect comparison of regimens and thus produces inferences of relative efficacy that would not otherwise be possible. We therefore undertook a systematic review using such an analytic approach to provide an up-to-date summary of the randomized, controlled trials (RCTs) that have evaluated LTBI treatment and an informative comparison of the relative efficacies and adverse event (AE) profiles of different regimens.
Source: Annals of Internal Medicine
