The discussion notes that improvements in case follow-up led to a decreased length of stay, so that fewer patients were in the hospital at any given time. Overall, there was significant improvement in compliance with CDC guidelines, although glitches (e.g., open doors and windows) persisted.

Unfortunately, as noted previously, survey results may give an incomplete picture. Sutton and colleagues (55) reported the results of a questionnaire and direct observation at three California hospitals (two county, one private-community) in an area where tuberculosis is highly endemic. This was done over 1 year (1994–1995, [exact dates not given]).

All of the hospitals had written tuberculosis plans consistent with CDC guidelines, but none of the hospitals performed routine assessment of their tuberculosis control practices. There were 13–17 tuberculosis isolation rooms available, including at least 1 each in the ICU and ED, at each hospital. Negative pressure was documented in 18/25 (72 percent) tested rooms, and 19/22 (86 percent) tested rooms had ≥6 ACH (6/16 [38 percent] had ≥12 ACH, even though they were not new rooms). However, only 1/27 (4 percent) rooms tested met recommended airflow pattern, and 20/24 (83 percent) had poor-fair air mixing (<10 seconds for puff of smoke to disperse, equivalent to ≥2 breaths for health care workers). The latter measurements are rarely reported in other studies.

One hospital provided HEPA masks with a fit testing program, one used dust-mist (DM) masks without a fit testing program, and one used DM masks but had HEPA masks with fit testing available (but the paper notes that in practice these were not used).

Practices noted on direct observations included lack of regular checks of negative pressure; windows that could be opened (with potential changes