Objective. To analyze the reasons for the discrepancy between the preliminary and final clinical diagnoses at a tuberculosis (TB) hospital and to determine the ways of improving the accuracy of diagnosis.
Materials and methods. Sixty-nine medical records were studied in the patients treated at the Republican Clinical Tuberculosis Hospital in Donetsk in 2014–2018, in whom the diagnosis of tuberculosis was removed after examination in the TB department. The patients were divided into groups: 1) HIV-positive (n = 31) and 2) HIV-negative (n = 38). There was a male preponderance in both groups: 17 and 21, respectively; the mean age was 42.3 ± 1.8 and 57.8 ± 2.4 years and the median bed/days were 10 and 11. Bacteriological assays, chest and head spiral computed tomography (SCT), and diagnostic fibrobronchoscopy (FBS) were used.
Results. Exclusion of tuberculosis and clarification of its diagnosis were made in 44 (63.8%) patients within 2 weeks; tuberculosis test therapy was required in 7 (10.1%) patients. The HIV-infected patients were most frequently diagnosed with HIV/AIDS-associated nervous system lesions and pneumonia (including atypical ones); the HIV-negative patients had lung oncopathology (cancer, carcinomatosis). Acid-resistant bacteria detected prior to hospital, unconfirmed in a specialized dispensary, were present in 15 (39.5%) HIV-negative patients.
Conclusion. To improve the accuracy of tuberculosis diagnosis in the facilities of the general medical network, it is necessary to improve the quality of microscopic examination and to prevent the contamination of biological samples. If there are clinical and radiological signs of pneumonia in HIV-infected persons with a negative sputum microscopy, their treatment should be performed in the specialized departments of AIDS centers with repeated consultation of a TB specialist if necessary.

Keywords: tuberculosis dispensary, diagnostic discrepancy