Preventing adverse events in the pediatric intensive care unit: prospectively targeting factors that lead to intravenous potassium chloride order errors.

Objective: To prospectively identify and reduce proximal causes of error contributing to inappropriate intravenous potassium chloride orders and reduce adverse events subsequent to these ordering errors.

Design: Pre-post cohort study of the reduction in both proximal causes of error and number of postinfusion elevated serum potassium levels after intervention.

Setting: Sixteen-bed, tertiary care, urban, academic pediatric intensive care unit.

Patients: Children 0-18 yrs old receiving intravenous potassium chloride in the pediatric intensive care unit.

Interventions: A multidisciplinary team determined proximal causes of error that were likely contributors to the occurrence of the outcome measure, elevated potassium levels after intravenous potassium chloride. A mandatory drug request form was designed for physicians ordering intravenous potassium chloride. The drug request form was designed to reduce proximal causes of error and, as a result, elevated potassium levels after intravenous potassium chloride. Demographic and laboratory data on children receiving intravenous potassium chloride in the pediatric intensive care unit and details of the drug order were analyzed.

Measurements And Main Results: Data from 1,492 intravenous potassium chloride administration-events before implementation of the drug request form were collected. After the drug request form was mandated, 166 consecutively completed forms were collected and analyzed. The incidence of postinfusion elevations in serum potassium decreased from a rate of 7.7% (103 of 1,341) before the drug request form to 0% (0 of 150) after the drug request form (p < .001). Proximal causes of error were also reduced. The number of patients with a creatinine >/=2 mg/dL receiving intravenous potassium chloride decreased from 28.4% to 14.2% (p < .001). The number of intravenous potassium chloride infusions administered to patients where serum potassium value was >4.5 mmol/L decreased significantly (2.9% vs. 0.0%, p < .02). The incidence rates of both verbal orders and failure to write the order in a correct format were reduced to zero.

Conclusions: Simple, blame-free, system-wide interventions designed to reduce proximal causes of error can be an effective, proactive means of reducing the likelihood of medical morbidity.