tuberculosis control measures (Blumberg, 1999). A multivariate analysis found no association between tuberculin skin test conversions and patient contact (frequent contact versus no contact). The analysis did, however, show an association between skin conversions and bacille Calmette-Guérin (BCG) vaccination, lower salary levels, and shorter time of employment. This suggests that community exposure was likely important for lower-salary workers, who probably come from parts of Atlanta with high rates of active tuberculosis.
Jackson Memorial Hospital (Miami) From 1988 to 1990, Jackson Memorial Hospital in Miami experienced an outbreak of multidrug-resistant tuberculosis related to patient-to-patient transmission on an HIV ward (Beck-Sague et al., 1992; Fischl et al., 1992; Wenger et al., 1995). After reviewing their infection control policies and work practices, hospital managers implemented a series of more stringent tuberculosis control measures (Wenger et al., 1995). The first control measures, which were implemented in March 1990, included a four-drug initial treatment regimen and more rigorous isolation policies on the ward (i.e., stricter isolation criteria for HIV infected patients, stricter criteria for discontinuing isolation; stricter enforcement of policies that infectious patients stay in their rooms unless medically necessary and wear surgical mask when out of their rooms, and restriction of sputum induction procedures to isolation rooms). In April 1990, the hospital repaired improperly functioning isolation rooms and improved the ventilation in other rooms. In June 1990, the hospital instituted a policy that aerosolized pentamidine would be administered only in isolation rooms. In the following months the hospital switched respiratory protections for health care workers from a surgical mask to a submicron mask (September 1990), established and staffed a separate unit for tuberculosis control (October 1990), added laboratory staff to improve turnaround times for specimen results (December 1990), required isolation for all patients with multidrug-resistant tuberculosis (February 1991), and switched to dust-mist respirators (April 1992). NIOSH checked the ventilation in the isolation rooms, and hospital staff checked negative pressure daily. Implementation of other practices (e.g., keeping doors to isolation rooms closed, and wearing of respirators) was checked by observation.
The effects of these changes were monitored for three time periods: January through May 1990 (which overlaps the first interventions), June 1990 through February 1991 (which overlaps most of the remaining interventions), and March 1991 through June 1992. The investigators found that all patients with multidrug-resistant tuberculosis who were admitted during the first monitoring period had been exposed to other such patients while on the HIV ward. In contrast, none of the patients with multidrug-resistant disease admitted during the subsequent monitoring