Epidemiology of tuberculosis in Sardinia: a comparison of two data-gathering methods.

SETTING: A complete surveillance system for tuberculosis should be able to guarantee constant updating of incidence and provide useful data on a variety of problems related to tuberculosis such as drug resistance, co-infection with the human immunodeficiency virus (HIV), the geographic origin of patients, and mycobacterial species. OBJECTIVE AND DESIGN: To assess the completeness of the surveillance system currently operating in Sardinia, cases seen by all medical centres between 1987 and 1995 were compared with those notified to Sardinian Public Health Services for the same period. RESULTS: Each year, on average 39% of cases seen in Sardinia are notified; 646 (40%) of the 1591 patients notified during the study period were never seen by regional medical centres. An analysis of the results shows that from 1992 the decline recorded in incidence rates in previous years ceased: 1992 (26/100,000), 1993 (25/100,000), 1994 (28/100,000), and 1995 (24/100,000). CONCLUSIONS: The current surveillance system in Sardinia is inadequate for performing an accurate epidemiological survey of the disease. Epidemiological analysis based solely on notification can provide neither reliable incidence rates nor useful information concerning many aspects of tuberculosis.

Microbiological diagnosis of tuberculosis: a comparison of old and new methods.

The aim of our study was to evaluate the diagnostic value of various methods widely used in microbiological diagnosis of tuberculosis: direct smear examination for acid-fast bacilli, cultural identification in Lowestein-Jensen (L-J) medium, the radiometric BACTEC 460 system, and Polymerase Chain Reaction (PCR). Three hundred and ninety-three clinical samples of sputum (375), gastric aspirate (3), pleural fluid (12) and urine (3) were taken from 125 patients hospitalized at our Institute for suspected pulmonary tuberculosis, between January 1995 and June 1997. On completion of diagnosis, 35 were found to be affected by active tuberculosis (30 pulmonary, 4 pleural and 1 urinary) and 90 by other non-tubercular diseases (pneumonia, lung cancer, non-tubercular pleural effusion, etc.). In our study, direct smear examination for acid-smear bacilli gave diagnostic value results of 88% and positive predictive value of 91.67%. Cultural identification in L-J and BACTEC 460 TB radiometric system media resulted in diagnostic values of 96.80% and 94.40%, respectively, and positive predictive values of 100% for both of them. Finally, One-Tube Nested-PCR, a variant which uses specific primers for the IS6110 insertion sequence specific for Mycobacterium tuberculosis, gave us 88.80% (91.43% sensitivity and 87.78% specificity) diagnostic value results, and 74.42% (11 false-positives) positive predictive value. On the basis of our results, we can affirm that PCR is a good method for microbiological diagnosis of tuberculosis, given its high sensitivity and specificity and unparalleled rapidity. However, the high number of false-positives that we found suggests that results obtained should be confirmed with BACTEC, which considerably reduces the time required for identification, and makes it possible to carry out an antibiotic assay rapidly.