The systematic use of directly observed therapy (DOT) for tuberculosis (TB) is widely recommended by TB authorities, but some have questioned its effectiveness and universal advisability. We review reports describing the outcome of DOT in cohorts treated in the human immunodeficiency virus (HIV) era (publication period, 1990-2000). The 34 studies represent a wide variety of geographic, cultural, and medical settings and include both routine program data and clinical trials. Most reported acceptably low rates (<5%) of treatment failure and posttreatment relapse (mean ± SD, 2.4 ± 2.0% [n = 30] and 3.6 ± 2.3% [n = 23], respectively). Overall effectiveness was substantially lower in many cohorts because of nonadherence: rates of treatment completion and cure were unsatisfactory (<85%) in 9 and 14 cohorts, respectively. The methods of outcome assessment varied, and surveillance for posttreatment relapse was often absent (11 studies) or incomplete, so that long-term cure rates remain uncertain in most cohorts. Regarding possible mechanisms of failure/relapse, initial drug resistance was not involved in the majority of instances but contributed substantially at some sites. HIV coinfection had little demonstrable impact on outcome (leaving aside non-TB-related deaths during treatment) but likely accounts for occasional early TB-related deaths and late recurrences. To better understand why poor outcomes occur despite DOT, we need systematic data on actual adherence, which was usually not reported, and on disease site and severity, drug bioavailability, and mycobacterial strain typing (to exclude reinfection). Well-implemented DOT is superior to unsupervised pill-taking, although its independent contribution to recent improvements in TB control cannot be quantified. Poor adherence remains the chief obstacle to success, signaling the need to focus on behavioral and logistic aspects of program performance.
ASJC Scopus subject areas