Maloney and colleagues [1] claim that implementing tuberculosis control measures, including prompt isolation and treatment, rapid diagnosis, negative-pressure ventilation, and use of molded surgical masks reduced transmission to patients and staff. Given the presented data, the authors are half-right: Patient-to-patient transmission was reduced, probably because patients suspected of having tuberculosis were quickly placed in isolation. Even the most ardent proponents of the above approach would be hard-pressed to ascribe the interruption of patient-to-patient spread of infection to the use of molded surgical masks by workers.

The authors fail, however, to show a decrease in transmission to staff. As they point out, the overall purified protein derivative (PPD) conversion rate among workers in the preintervention and intervention periods was the same: 26 (3.1%) of 840 workers compared with 22 (3%) of 727 workers. Conversion rates by job category and ward location were then analyzed. However, each analysis is problematic.

The authors grouped workers by job into those with and those without patient contact and then compared relative risk for conversion between the two groups, before and during the interventions. The pertinent comparison, however, is the conversion rate in workers with patient contact before compared with during intervention. This analysis shows no difference: Conversions occurred in 22 (6.4%) of 342 workers with direct patient contact before intervention compared with 14 (4.7%) of 296 during intervention (P > 0.2).

Analyzing rates according to hospital ward assumes that the whereabouts of 560 employees can be established. However, many employees (such as physicians, respiratory therapists, and housekeepers) work on several wards daily, whereas others (such as nurses and clerks) have a "home base" but may work else-where during overtime or to help with staffing shortages. Given this large uncertainty and the small number of PPD conversions, calling the interventions "effective" is unjustified.

Thus, workers were not protected beyond the benefit derived from placing patients into isolation. The failure to show effectiveness is important and questions the wisdom of current guidelines [2]. I realize that the horse is already hopelessly out of the barn, but we need either more convincing data or else a major reconsideration of these guidelines beyond the "evolutionary process" of review promised in an article accompanying the paper by Maloney and colleagues [3].