An Analysis of 34,218 Pediatric Outpatient Controlled Substance Prescriptions

BACKGROUND:Prescription errors are among the most common types of iatrogenic errors. Because of a previously reported 82% error rate in handwritten discharge narcotic prescriptions, we developed a computerized, web-based, controlled substance prescription writer that includes weight-based dosing logic and alerts to reduce the error rate to (virtually) zero. Over the past 7 years, >34,000 prescriptions have been created by hospital providers using this platform. We sought to determine the ongoing efficacy of the program in prescription error reduction and the patterns with which providers prescribe controlled substances for children and young adults (ages 0–21 years) at hospital discharge. METHODS:We examined a database of 34,218 controlled substance discharge prescriptions written by our institutional providers from January 1, 2007 to February 14, 2014, for demographic information, including age and weight, type of medication prescribed based on patient age, formulation of dispensed medication, and amount of drug to be dispensed at hospital discharge. In addition, we randomly regenerated 2% (700) of prescriptions based on stored data and analyzed them for errors using previously established error criteria. Weights that were manually entered into the prescription writer by the prescriber were compared with the patient’s weight in the hospital’s electronic medical record. RESULTS:Patients in the database averaged 9 ± 6.1 (range, 0–21) years of age and 36.7 ± 24.9 (1–195) kg. Regardless of age, the most commonly prescribed opioid was oxycodone (73%), which was prescribed as a single agent uncombined with acetaminophen. Codeine was prescribed to 7% of patients and always in a formulation containing acetaminophen. Liquid formulations were prescribed to 98% of children 12 years of age (the remaining 84% received tablet formulations). Regardless of opioid prescribed, the amount of liquid dispensed averaged 106 ± 125 (range, 2–3240) mL, and the number of tablets dispensed averaged 51 ± 51 (range, 1–1080). Of the subset of 700 regenerated prescriptions, all were legible (drug, amount dispensed, dose, patient demographics, and provider name) and used best prescribing practice (e.g., no trailing zero after a decimal point, leading zero for doses 15%. Of th