THE USE OF CLINICAL CHARACTERISTICS TO PREDICT THE RESULTS OF TEMPORAL ARTERY BIOPSY AMONG PATIENTS WITH SUSPECTED GIANT-CELL ARTERITIS

Objective. To develop a mathematical model which predicts temporal art ery biopsy results. Methods. We collected clinical and laboratory data as well as biopsy results among a consecutive cohort of all individua ls who underwent temporal artery biopsy at Mayo Medical Center between January 1, 1988 and December 31, 1991. All biopsies were independentl y reviewed by one pathologist. Logistic regression was used to identif y a set of variables which best predicted the biopsy results. This mod el was then used to identify patients who were highly likely (greater than or equal to 95% predictive value) to have either a negative or a positive biopsy. A receiver operating characteristic (ROC) curve was g enerated using the best fit model. Results. Of the 525 people in the s tudy, there were 187 men and 338 women. The logistic regression model and the ROC curve generated from this model were of modest value in pr edicting biopsy results from prebiopsy clinical characteristics. Howev er, this model identified 60 (11%) individuals who had a greater than or equal to 95% probability of having a negative biopsy. None of these individuals had any symptoms of claudication, only 5 of 60 (8%) had t emporal artery abnormalities on examination, 45 (75%) had synovitis (s uggesting an alternate diagnosis), and their median erythrocyte sedime ntation rate was only 31 mm/h (Westergren). Conclusions. In individual s with these findings, we recommend a careful search for other diagnos es before temporal artery biopsy.