in response to nosocomial transmission of tuberculosis to patients and/or health care workers. Implementation of these measures then led to decreases in nosocomial cases of tuberculosis infection or disease.

Wenger and coworkers (13) reported the experience of Jackson Memorial Hospital, Miami, Florida, following an outbreak of MDR tuberculosis from 1988 to 1990 on an HIV ward (4, 14). Control measures were implemented over time, starting in March 1990. The measures implemented included the following:

March

• Stricter enforcement of isolation policy to include isolation of any HIV-positive patient with an abnormal chest radiograph (CXR)

• Change in criteria for stopping isolation from discontinuation after 7 days on therapy to discontinuation only after three negative smears for acid-fast bacilli (AFB) (or after reduction in AFB on three smears plus a clinical response)

• Enforcement of policy to keep tuberculosis patients in their rooms unless medically necessary and having patients wear a surgical mask when out of their rooms

• Sputum induction done only in isolation rooms

• Initial therapy for tuberculosis with four drugs

April (through April 1991)

• The 6 tuberculosis isolation rooms (of 23) without negative pressure were repaired, and the ventilation in the other rooms was made more consistent

June

• Aerosolized pentamidine administered only in isolation room

September

• Change from cup-type surgical mask to submicron mask for health care workers

A review of admissions of HIV-positive patients with MDR tuberculosis was performed, covering three time periods: initial period (January 1990 to May 1990), early follow-up (June 1990 to February 1991), and late follow-up (March 1991 to June 1992). There was a decrease in MDR tuberculosis patient-days over the three periods (222/100 real days initially, then 119/100, and finally 16/100). Fifteen patients with MDR tuberculosis were admitted during the initial period: 12 (80 percent) had been exposed while on the HIV ward. Eleven patients were admitted during the follow-up periods, only five of whom had been exposed on the ward, all during the initial period. No known patient exposures occurred during the follow-up periods.