Study objective: To determine a threshold dose for parenteral lorazepam when screening for propylene glycol toxicity with the osmol gap, and to characterize which osmol gap values are more predictive of toxic propylene glycol concentrations and resultant clinical toxicity.
Design: Prospective, two-phase observational study.
Setting: Thirty-two bed, multidisciplinary intensive care unit.
Patients: Thirty-five adult patients receiving any dose of parenteral lorazepam (phase 1), and 14 patients receiving lorazepam in doses of 1 mg/kg/day or higher (phase 2).
Measurements and main results: Serum osmolality was measured every other day during lorazepam therapy, and the osmol gap (measured osmolality minus calculated osmolarity) was determined. A serum propylene glycol concentration was obtained when the osmol gap first exceeded 10. In phase 1, 35 patients were monitored for 186 patient-days. Ten patients (29%) developed an osmol gap greater than 10; only one (10%) of these patients had propylene glycol concentrations greater than 18 mg/dl. In phase 2, 14 patients received lorazepam at a median dose of 631 mg (interquartile range [IQR] 437-972 mg) over a median of 5.5 days (IQR 4-8.75 days). Nine patients (64%) had propylene glycol concentrations greater than 18 mg/dl; six (67%) of these nine developed transient acute kidney injury, metabolic acidosis, or both. The correlation between osmol gap and propylene glycol concentration was 0.44 (p=0.006). An osmol gap of 10 or greater had a likelihood ratio of 4.4 to predict a propylene glycol concentration greater than 18 mg/dl. An osmol gap of 12 or greater had a likelihood ratio of 2.7 to predict the development of possible propylene glycol clinical toxicity.
Conclusion: Screening for propylene glycol toxicity with the osmol gap may be helpful for patients receiving intravenous lorazepam in doses of 1 mg/kg/day or higher. An osmol gap of 10 or greater was predictive of elevated propylene glycol concentrations, and values of 12 or greater were predictive of clinical changes suggestive of propylene glycol toxicity.