Study objective: To validate the predictive abilities of a retrospectively developed set of clinical criteria for detecting clinically significant electrolyte abnormalities, using a different patient population.
Design: Cross-sectional study.
Setting: The emergency department of a busy public hospital.
Type of participants: Nine hundred eighty-two patients on whom the emergency physician ordered serum electrolytes.
Interventions: The predictive properties of ten clinical criteria were evaluated; these included poor oral intake, vomiting, chronic hypertension, taking a diuretic, recent seizure, muscle weakness, age of 65 years or more, alcoholism, abnormal mental status, and recent history of electrolyte abnormality.
Measurements and main results: Seven hundred thirty patients (74.3%) had one or more electrolytes outside of the laboratory normal range, but only 143 (14.6%) had clinically significant electrolyte abnormalities. The clinical criteria predicted 135 of the clinically significant electrolyte abnormalities (sensitivity, 94.4%). When the eight "false-negative" cases were reviewed, none of the electrolyte abnormalities affected patient outcome. Implementation of the criteria would have avoided unnecessary testing in 233 patients (23.7%).
Conclusion: Although no set of clinical criteria can eliminate the need for clinical judgment, use of a set of clinical criteria could substantially decrease electrolyte ordering without compromising patient care.