(especially those with large numbers of tuberculosis patients) that have low nosocomial tuberculosis rates and health care worker PPD conversion rates by following the CDC guidelines provide further assurance of the effectiveness of the guidelines.
An extremely detailed search for possible cases of nosocomial transmission of tuberculosis was done at Cook County Hospital in Chicago, where French and colleagues performed DNA fingerprinting on one isolate from every patient with tuberculosis for 1 year, from April 1995 through March 1996 (27). A comprehensive record review of patients whose isolates were in a fingerprint cluster was done to determine chance of cross-transmission. Overall, 91/168 (54 percent) isolates were in 15 clusters. There were six clusters of 2 isolates, 7 clusters of 3 to 8 isolates each, one cluster of 16 isolates, and one cluster of 29 isolates. The risk factors for clustering were birth in United States, male sex, African-American ethnicity, alcohol or illicit drug abuse, and homelessness. On multivariate analysis, only male sex and birth in the United States were associated with clustering.
For 13 of 15 clusters (46 patients), no instances were identified where two patients were inpatients or outpatients at same time. For the two largest clusters, 148 instances of two patients being on hospital grounds at same time were found. For 144/148 instances, cross-transmission was thought to be unlikely because of different sensitivity patterns (32 instances) or lack of geographic overlap of patients (112 instances). Of four remaining instances, the site of possible cross-transmission was ED (3 instances), and the HIV clinic (1 instance). In one case the possible source patient had only extrapulmonary tuberculosis, so nosocomial transmission was thought to be unlikely. In another case, only 5 weeks elapsed from the time of exposure to the diagnosis of fibrotic pulmonary disease in an immunocompetent patient. Cross-transmission in this case was thought to be implausible.
Of two remaining instances, the same source patient was involved. In one instance the source patient (patient A, HIV positive, CD4 count of 423 cells per milliliter) had a CXR consistent with miliary tuberculosis. The patient was masked and placed in isolation within 1.25 hours of admission. Patient B (also HIV positive) was brought to the ED by ambulance after patient A had been placed in isolation. Patient B had a history of a positive PPD test result and so was masked and placed in an isolation room 50 yards from patient A. Eight months later patient B developed pulmonary tuberculosis. Given the prompt masking and isolation of both patients and a history of a positive PPD, nosocomial transmission to patient B was thought to be unlikely (albeit possible, since droplet nuclei can stay suspended for some time).
The last possible patient exposure occurred in the HIV clinic. Patient C (also HIV positive) was in a clinic concurrent with patient A 5 weeks