As described above in the three-hospital study by Sutton and colleagues, nearly two-thirds of workers observed using a respirator did not put it on properly. In another study at a single facility (the University of California at San Diego), researchers observed health care workers over a 14-week period (LoBue et al., 1999). They recorded 64 violations (during 541 observations) that included 36 failures to maintain isolation (e.g., leaving a door open) and 28 failures to use respirators properly. Medical students, residents, and fellows accounted for 17 percent of the study observations and 45 percent of the violations. (Eight of the 29 violations for this group were described as not clinically important; for example, isolation had been ordered discontinued, but a sign was still on the door.)
Asimos and colleagues (1999) reported results of a survey of emergency medicine residents conducted in 1998 in conjunction with the annual in-service examination of the American Board of Emergency Medicine. Nearly 90 percent of the residents responded to at least part of the survey. Half reported that they did not routinely wear a NIOSH-approved respirator during contact with patients at risk of having tuberculosis. Almost half reported that the reason for lack of compliance was a lack of easy availability of respirators, and about a third reported a lack of fit testing as the reason. Just under one-third reported they had not been offered fit testing, and 8 percent reported being offered but not going through fit testing.
The interplay between institutional and individual practices is also suggested by another observational study. During the investigation of 22 New York City hospitals described earlier, Stricof and colleagues found that the rate of use of approved respirators was higher when the respirators were placed outside isolation rooms rather than at nursing stations and when only approved respirators were available (i.e., surgical masks were not available) (Rachel Stricof, New York State Department of Health, personal communication, August 28, 2000).
A 1993 study of health care workers focused on knowledge rather than practice (Lai et al., 1996). Two hundred health care workers with patient contact were tested. Just under half reported some education on tuberculosis during the preceding 2 years. Nearly all (98 percent) knew that coughing or sneezing could spread tuberculosis, but more than a quarter (28 percent) thought that it could be transmitted by a handshake. The great majority (88 percent) knew that masks should be used in the rooms of patients with tuberculosis, but a third also thought that gowns were needed.
Ideally, the 1994 CDC guidelines would have been based on rigorous, prospective, controlled studies demonstrating the effectiveness of each