Recognition of information in renal biopsies of patients with lupus nephritis

To evaluate the informational value of renal biopsy in nephritis of systemic lupus erythematosus, we selected the records of 30 patients who had a renal biopsy done and also had a known clinical outcome. Detailed case histories were prepared, and three distinct randomly chosen cases were given to 197 academic rheumatologists. The rheumatologists estimated the probability of future clinical events (worsened serum creatinine, worsened urine protein, renal death, and aggressive therapy) at 3 and 12 months after the biopsy. Biopsy results were given in detail, and probability estimates were made of the same clinical events using the additional information. The accuracy of each probability estimate was measured using a scoring function that depends on the estimates and the actual outcomes. Knowledge of the renal biopsy failed to improve predictive accuracy scores of estimates of future serum creatinine levels, urine protein levels, and renal death at 3 and 12 months (p less than 0.0001), and for estimates of the probability of the use of aggressive therapy at 12 months (p less than 0.007). The renal biopsy information improved only the accuracy of predictions concerning the use of aggressive therapy at 3 months (p less than or equal to 0.0003). Knowledge of the renal biopsy results failed to add important prognostic information about the future course of treated lupus nephritis to information already obtained from history, physical examination, and laboratory tests.