Tuberculosis isolation: comparison of written procedures and actual practices in three California hospitals

To evaluate implementation of healthcare worker exposure control measures for tuberculosis (TB)-patient isolation, as specified by Centers for Disease Control and Prevention (CDC) guidelines and the hospital's TB-control policy.

Prospective multihospital study comparing CDC guidelines and hospital policy for TB-patient isolation to once-weekly observations of TB-patient isolation practices over 14 consecutive weeks at each hospital.

Three urban hospitals (two county, one private community) in counties in California with a high incidence rate of TB.

Work practices for TB-patient isolation were observed and ventilation performance of isolation rooms was assessed while patient rooms were in use for TB isolation.

Of 170 TB-patient rooms observed, 119 (70%) involved a patient in a designated TB isolation room, the room was under negative pressure, the door was closed, and a "respiratory precautions" sign was on the door; 32 patient-room units (19%) were not under negative pressure or not designated as negative-pressure rooms. Of 151 patient-room units mechanically capable of negative pressure at a prior point in time, 16 (11%) were not under negative pressure at the time of use. Of 67 patient-room units equipped with continuous monitoring devices, 8 (12%) involved devices that did not accurately reflect the direction of airflow. Of the 62 healthcare workers observed using a respirator for TB, 40 (65%) did not don the respirator properly.

Implementing CDC guidelines for TB-patient isolation was feasible but imperfect in the three hospitals. Day-to-day work practices deviated from hospital policy. Prospectively quantifying the implementation of a hospital TB isolation policy while the room is in use may lead to improved estimates of risk and may help to identify and thereby prevent avoidable healthcare worker exposures to Mycobacterium tuberculosis aerosol. Auditing practices and verifying equipment performance is likely to identify unexpected problems in implementation of the TB control program.