The "iron screen": modification of standard laboratory practice with data analysis.

Multivariate analysis was applied to iron deficiency anemia to generate an efficient sequence of diagnostic laboratory tests. A three step diagnostic system--serum ferritin level and mean corpuscular volume as a screen in all patients, followed by serum iron level and total iron binding capacity in some patients, and by erythrocyte sedimentation rate in a few patients--was constructed using a previously validated data reduction system. When compared to bone marrow iron stores, this system was found to have 96 per cent accuracy. In one year of clinical trial the "iron screen" classified 396 of 416 patients in a hospital setting. This sequential strategy shows how clinical laboratory data can be utilized to render diagnoses of defined probability.