Is it possible to characterize recurrent urinary stone formers who benefit from thiazide prophylaxis? Application of discrimination analysis.

Twenty-three patients with recurrent urinary stones have been treated for 12-54 months with hydrochlorthiazide and 25 with placebo. Discrimination analysis was performed in order to distinguish patients with new stone formation (non-responders) from those without stone recurrence (responders) during the treatment. The discriminant function classified 16 out of 18 responders as such and 5 non-responders as such by using a combination of the three initial variables urinary pH, 24-h urine volume and serum phosphate. Eighteen other analysed variables and patient factors did not contribute to this discrimination. By using "leaving-one-out-technique", the probability for erroneously classifying a patient increased from 9 to 17%. Applying the discriminant function on the group receiving placebo, 92% of these patients were classified as responders to thiazide. It is concluded that discrimination analysis may give valuable information in characterizing patients who will benefit from a special therapy.