DERIVATION AND VALIDATION OF A PULMONARY TUBERCULOSIS PREDICTION MODEL

OBJECTIVE: To describe the derivation and validation of a pulmonary tu berculosis (TB) prediction model that would enable early discontinuati on of unnecessary respiratory isolation. DESIGN: Patients placed in is olation for suspected pulmonary TB were studied retrospectively (deriv ation cohort) and prospectively (validation cohort). Independent predi ctors of pulmonary TB in the derivation cohort (January 1992-March 199 4) were identified by retrospective analysis. Predictors in the model were assigned weights on the basis of the results of the multivariate analysis in order to quantitate the risk of TB in an individual patien t. The prospective validation consisted of application of the model to patients placed in isolation during the period April 1994 to June 199 5. The predictability of the model in the derivation and validation co horts was evaluated using receiver operating characteristics (ROC), cu rve analysis, and calculation of the area under the ROC curve (AUC). S ETTING: A university-affiliated, urban, public hospital with a large p opulation of prison inmates and patients with human immunodeficiency v irus infection.INTERVENTIONS: Prospective application of the predictio n model to patients placed in isolation during the validation period. RESULTS: Four factors were found to be independent predictors of pulmo nary TB among 296 isolation episodes in the derivation cohort: positiv e acid-fast sputum smear (odds ratio [OR], 5.8; 95% confidence interva l [CI95], 3.0-11.0; weight=3 points), localized chest radiograph findi ngs (OR, 2.5; CI95, 1.3-4.9; weight=2 points), residence in a correcti onal facility (OR 2.3; CI95, 1.2-4.4; weight=2 points), and history of weight loss (OR 1.8; CI95, 1.03.2; weight=1 point). Infection control practitioners applied the model prospectively to 220 isolation episod es. The mean (+/-SE) AUCs of the ROC curve for the derivation and vali dation cohorts were not significantly different (.86+/-.04 vs .86+/-.0 7; P=.90). There was a significant decline in the mean duration of iso lation from the onset of an automatic TB isolation policy in August 19 92 to the end of the study (P=.045 by analysis of variance). CONCLUSIO NS: A pulmonary TB prediction model was derived and validated prospect ively in a hospital with a moderately high prevalence of TB. The model quantitated the risk of TB in an individual patient and aided infecti on control practitioners and primary-care physicians in their decision s to discontinue isolation during the validation period. Utilization o f the model was responsible, in part, for a decrease in the mean durat ion of isolation during the study period. Although the model may not h ave general applicability due to the uniqueness of the patient populat ion studied, this study illustrates how prediction models can be devel oped and used effectively to deal with a clinical problem (Infect Cont rol Hosp Epidemiol 1997; 18:554-560).