provides an excellent review of the social costs of an outbreak and associated controls.

A significant fraction of the ongoing cost of a tuberculosis program may be in evaluating patients who do not have tuberculosis but who meet criteria to be evaluated. Scott and colleagues (68) evaluated the experience at the University of Iowa Hospital and Clinics. All patients with a positive sputum culture for tuberculosis between January 1, 1987, and September 24, 1992, were considered a case. Forty-four patients were identified, and charts were available for review for 43. Control patients were chosen randomly from patients who had had sputum submitted for AFB but who had negative cultures. Since bronchoscopy specimens were routinely sent for AFB smear and culture regardless of clinical suspicion of tuberculosis, patients who had specimens only from bronchoalveolar lavage were excluded. Of 92 potential controls for every case, 43 random controls chosen matched by location (inpatient/outpatient) and service.

Of the case patients, 39 (91 percent) were smear positive; 25 (58 percent) were positive on the first smear. Only one test for AFB was sent from 48 percent of the control subjects. Of 24 inpatients with pulmonary tuberculosis, only 10 (42 percent) were isolated upon admission. A total of 37/ 43 (86 percent) case patients had a CXR consistent with tuberculosis, as did 7/43 (16 percent) controls. If same rate held for all patients, ~670 patients would have had abnormal CXRs. The six other case patients had abnormal CXRs, but not “typical” for tuberculosis. From July 1, 1991, through June 30, 1992, there were 12 “exposure” workups for an AFB+ smear, with 363 contacts. Only 4 of the 12 had tuberculosis; the others had infection with non-tuberculous mycobacteria (NTM).

Scott and colleagues (68) calculated the cost of diagnosing a case of tuberculosis: $18.30 was spent for an AFB smear and culture. Control patients had an average of 2 sputum specimens sent, while case patients had an average of 3.2 specimens sent. With 92 control patients for every tuberculosis patient, this led to a cost of $3,426 per case of tuberculosis diagnosed. The authors also estimated that 15 minutes/person of nurse epidemiologist time was spent tracing and contacting health care workers exposed to a case of tuberculosis, with an additional $6.00 to $11.00 per employee for PPD testing.

The authors state that a policy of isolating everyone for whom an AFB smear was sent would be unreasonable, causing a 92-fold overuse of isolation rooms. However, this is within the range reported in the Veterans Affairs hospital study by Roy et al. (30). Although not discussed, if only the estimated 670 patients with “typical” CXRs were isolated, the overisolation ratio would be ~18:1, which does not seem unreasonable.