GENERATION OF REFERENCE VALUES FOR CARDIAC ENZYMES FROM HOSPITAL ADMISSION LABORATORY DATA

An approach is described for using patient databases of a hospital inf ormation system as a source of reference values for cardiac enzymes. O f a total of 2029 emergency admission patients with serial cardiac enz yme data, 538 patients were considered ''healthy'' (having no damage i n myocardium) because their discharge diagnoses suggested neither myoc ardial damage nor any other condition that could lead to elevated enzy me activities, and because their serially collected cardiac enzyme act ivities remained stable. Enzyme activities of creatine kinase (EC 2.7. 3.2), creatine kinase isoenzyme MB, lactate dehydrogenase (EC 1.1.1.28 ), and lactate dehydrogenase isoenzyme 1 of these patients at admissio n to hospital were considered as suitable health related reference val ues. The upper (97.5%) reference limits of activities, measured at 37 degrees C according to Scandinavian recommendations, were as follows ( age dependent limits given at 25 and at 75 years of age, U/l): creatin e kinase men 268, 192; creatine kinase women 200 (no age effect); crea tine kinase-MB 16, 24; lactate dehydrogenase 497, 603; lactate dehydro genase isoenzyme 1 103, 140. For comparison, reference values were als o produced conventionally from a group of 246 healthy subjects. Observ ed effects of age on enzyme activities were quite similar to those in the selected patient group. Calculated reference limits for isoenzymes creatine kinase-MB and lactate dehydrogenase isoenzyme 1 were also si milar but reference limits for less cardiospecific total enzyme activi ties, creatine kinase and lactate dehydrogenase, were more variable be tween these two groups. Observed differences in total enzyme activity levels may reflect different preanalytical conditions of emergency adm ission patients and conventional reference subjects, in which case ref erence values produced from selected ''healthy'' emergency admission p atients would be more suitable for clinical use. We conclude that refe rence values can be produced from patient databases with relatively li ttle effort and at low cost provided that the database information is clinically adequate and reliable.