Effects of referral bias on surgical outcomes: a population-based study of surgical patients 90 years of age or older.

During the 11-year period 1975 through 1985, 1,063 surgical procedures were performed on 795 Mayo Clinic patients 90 years of age or older. Preoperative conditions, surgical setting, and perioperative morbidity and mortality were analyzed in a comparison of the local residents of Olmsted County, Minnesota (N = 224), with patients from outside the county but within 250 miles (N = 456) or referrals from a distance of 250 miles or further (N = 115). In comparison with non-Olmsted County patients, Olmsted County patients were generally older, had more preoperative chronic diseases, and underwent more emergency operations. Patients who had traveled 250 miles or more to the Mayo Clinic were more likely to be men and referred for cancer-related surgical procedures. The risks of major morbidity and mortality within 48 hours postoperatively were increased in patients with more preoperative chronic diseases and those undergoing emergency procedures, characteristics most common in Olmsted County patients. Because of differences in these factors between groups, perioperative risks averaged over our entire patient series underestimated risks that would be expected from a population-based cohort such as Olmsted County residents. In general, data from tertiary medical centers probably do not accurately reflect overall practice or outcomes in community settings.