Impact of referral bias on evaluation of cataract surgery.

Patients from a defined population (Olmsted County, Minnesota) who underwent cataract extraction during 1980 were compared with referral patients operated on at the Mayo Clinic to determine whether selective referral of patients introduces bias into studies conducted at academic centers. Rates of cataract extraction in Olmsted County increased continuously with increasing age and, in the older groups, were greater among females than males. The overall age- and sex-adjusted annual rate, 120 per 100,000 population, was considerably lower than that for 1980 in the United States, 209 per 100,000 population. A significantly greater proportion of referral patients had coexistent diabetes mellitus (P less than .05) and underwent other ophthalmic surgical procedures in combination with cataract extraction (P less than .001). Previous ophthalmic disorders were also more frequent among referral patients. Because selection for referral can distort data, this possible source of bias must be considered when interpreting data.