OBJECTIVES: To derive outcome-based critical result thresholds in the adult patient population. METHODS: We extracted deidentified laboratory results and outcomes (death or discharged) of patients 18 years and older from the Medical Information Mart for Intensive Care database. The lower and upper critical result thresholds were obtained from the nearest minimum and maximum laboratory values, which corresponded to predicted probability of death at 90%. RESULTS: The critical value thresholds were sodium (<123, >153 mmol/L), potassium (<2.2, >6.6 mmol/L), bicarbonate (<15, >49 mmol/L), chloride (<82, >121 mmol/L), urea (>20 mmol/L), creatinine (>1,052 μmol/L), glucose (<1.5, >23.8 mmol/L), total calcium (<1.62, >2.95 mmol/L), magnesium (<0.37, >1.48 mmol/L), phosphate (<0.19, >2.52 mmol/L), pH (<7.22, >7.57), lactate (>5.0 mmol/L), hemoglobin (<4.6 g/dL), WBCs (>32 × 103/μL), prothrombin time (>90 seconds), and international normalized ratio (>10). CONCLUSIONS: The indirect approach described in this study is a pragmatic way to obtain threshold values that are clinically and operationally meaningful.
Am j clin pathol
177 - 184
Critical reporting, Critical results, Critical value, Panic value, Postanalytical, Adolescent, Adult, Aged, Aged, 80 and over, Clinical Laboratory Techniques, Critical Care, Databases, Factual, Female, Humans, Male, Middle Aged, Young Adult