Analytic precision in clinical chemistry and medical decisions

A study was undertaken to answer the question: how precise do measurements in clinical chemistry have to be to contribute positively to patient treatment. Sixty-three senior specialists in internal medicine, working in university and teaching hospitals, were interviewed on a number of subjects related to medical decision-making. Hospital laboratories tested samples of quality-control sera to obtain information about intra- and interlaboratory variability. Data for normal range values were reported by laboratories. The tests evaluated in this study were: sodium, potassium, chloride, calcium, inorganic phosphorus, urea, creatinine, glucose, cholesterol, total protein, lactate dehydrogenase, alkaline phosphatase and hemoglobin. Three criteria were applied to test whether analytic precision met present medical needs: (1) the difference between normal range limits and action levels (a value prompting the clinician to action, e.g., repeated or additional testing, in the outpatient situation; (2) a medically significant change in a patient's condition, assuming an initial value in the near-abnormal range; (3) clinicians' feeling of satisfaction or dissatisfaction with laboratory performance. Except for calcium, median analytic precisions of laboratories participating in the study were adequate to record a change in the condition of a patient with a value in the near-abnormal range (second criterion). In order to distinguish the individual with a disease from the normal population, more tests need improvement: calcium, creatinine, glucose, cholesterol, total protein, alkaline phosphatase, lactate dehydrogenase and hemoglobin (first criterion). This finding largely agrees with physicians' satisfaction or dissatisfaciton (third criterion).