Setting: The four hospitals and a tuberculosis clinic in the province of Zeeland, The Netherlands.

Objective: To assess the usefulness of PPD antibody measurement in the diagnosis of tuberculosis in patients admitted to hospital.

Patients and methods: Sixty-one patients presenting with active tuberculosis, and 215 control patients were included in the study. Initial serum PPD antibody titres were determined with a macrophage uptake Fluorescent antibody test (MuFat) to construct a discrimination model between Tuberculosis (TB) and non-TB. We also retrospectively collected clinical parameters of the TB patients at presentation. Univariate and multivariate logistic regression are used to identify variables predicting high antibody titres.

Results: In TB patients, the presence of clinical symptoms ( ) and the presence of at least two concurrent non-lymph node disease localizations outside thorax and abdomen ( ) are necessary and sufficient to predict high titres.

The logistic model shows a significant contribution of the ²log (titre) to the discrimination between TB and non-TB patients. At a cut-off value of 128, a specificity, sensitivity, and positive predictive and negative predictive values of 97%, 39%, 80% and 85%, respectively, are calculated in the study cohort.

Conclusion: Our data suggest an application of the test at high cut-off values for timely diagnosis of difficult-to-diagnose TB patients. The results of this retrospective study will have to be confirmed in further prospective studies.