Validity of Hematologic Data in Veterans Administration Hospital Laboratories: A Veterans Administration Cooperative Study

Abstract

Twenty laboratories in Veterans Administration Hospitals, similar in sophistication, were assessed for accuracy and precision in performing hemoglobin (Hb), hematocrit (Hct), erythrocyte count, mean corpuscular volume (MCV), and leukocyte count. Nineteen of the laboratories used the Coulter S counter. Ten laboratories were classified as “superior” by past performance in College of American Pathologists (CAP) surveys. The remainder were selected from other V. A. hospitals. All laboratories analyzed two “masked” and two “open” samples from two pools (normal and abnormal) on the same day, once weekly, for three consecutive weeks (12 samples/pool/lab). Results from 19 laboratories were evaluated for within-day, day-to-day, and interlaboratory variability, effects of masking, and previous performance in CAP surveys. Overall study means were in agreement with target values. An unexpected finding was that withinday variability was greater than day-to-day variability, which was greater than interlaboratory variability (variance components). Precisions for Hb, Hct, erythrocyte count, and MCV were greater than that for leukocyte count. Interlaboratory Standard deviations were less than or close to “medical needs.” The data indicated that for three laboratories, “open” samples resulted in a significant bias. It is concluded that V. A. hospital laboratories using similar instrumentation perform at the level of the “state of the art” and could serve as a common resource for cooperative studies requiring Coulter S data.