Inaccuracy and expense of the leukocyte count in making urgent clinical decisions.

The care of 172 patients with abnormal white blood count (WBC) results was examined by chart review and physician interview. Neither the total leukocyte count, the neutrophil count, nor the band count reliably distinguished between bacterial, nonbacterial, and noninfectious disease. The positive predictive value for bacterial disease of a WBC greater than or equal to 12,500/mm3 was only 26%, with either a neutrophil count greater than or equal to 10,000/mm3 or a band count greater than or equal to 500/mm3, the positive predictive value was only 33%. Of WBCs, 21% were ordered by mistake when another (or no) test was desired. Only 18 abnormal WBCs of 172 (10.4%) (of the total 860 obtained) had any impact on the patient's clinical management, and only ten caused impact judged to be neutral or beneficial. Based on a WBC and differential costs of +23.50, obtaining an abnormal result that had a clearly beneficial impact on patient care would cost +10,105 in unneeded tests. The WBC does not in itself reliably predict severity or cause of disease in acutely ill adults, and its widespread use as a screening test in this setting probably should be abandoned.