Background: Duplicate as-needed (referred to hereafter as “PRN”) orders for common indications such as pain, nausea, insomnia, and constipation are frequent in hospitalized patients. Without explicit instructions for circumstances or order of administration, therapeutic duplication can cause confusion for nurses and violates both Join Commission and the Centers for Medicare & Medicaid Services (CMS) guidelines.
Purpose: To reduce the incidence of duplicate PRN medications ordered without explicit administration instructions using passive clinical decision support, performance feedback, and education.
Description: We evaluated the duplicate PRN medication order rate at an 804-bed academic medical center with approximately 30,000 annual hospitalizations. An interprofessional team comprised of physicians, pharmacists, and nurses, some with expertise in the Epic electronic health record (EHR), convened to address this system weakness. We focused on three indications, pain, constipation, and nausea/vomiting, as being responsible for the majority of therapeutic duplication in PRN orders. Up until this point, our healthcare system used a resource intensive, monthly process to manually audit patient charts. Following a fish bone analysis (Figure 1), several quality improvement initiatives were gradually implemented from 2018 to 2019. These initiatives included 1) developing and refining an automated data extraction workflow from the Epic clinical data warehouse to reliably measure therapeutic duplication, 2) providing regular feedback to clinicians, 3) creating a “Best Practices” guide for PRN medication orders for prescribers, pharmacists, and nurses, 4) updating existing pain and constipation medication order sets to include less ambiguous PRN indications, and 5) consolidating separate “nausea,” “vomiting,” and “nausea or vomiting” indications into a single “nausea/vomiting” option. We also provided focused education to hospital medicine physicians and pharmacists to promote awareness and ensure buy-in from the involved staff members. We leveraged the EHR data warehouse to track therapeutic duplication error rate and provide individualized feedback for prescribers and pharmacists. We analyzed performance using the raw, observed duplicate PRN order rate as well as a generalized linear autoregressive moving average (GLARMA) negative binomial model to assess the intervention’s efficacy (Figure 2). When accounting for the data’s temporal autocorrelation using GLARMA, our intervention led to a 32.7% (95% CI: 22.2 – 41.8, p < 0.001) reduction for inappropriate PRN therapeutic duplication.
Conclusions: We implemented a multi-component program to reduce inappropriate duplicate PRN medication orders using a multidisciplinary team and leveraging continuous data extraction and individualized performance feedback. Though the intervention was aimed at three specific, common PRN indications, we achieved a sustained reduction in the overall inappropriate PRN order rate consistent with our internal quality metrics and recommended Joint Commission and CMS guidelines. Disambiguating PRN medications can minimize nursing staff confusion and may mitigate adverse drug effects for hospitalized patients. Future directions include expanding this workflow to the other four University of California health systems for benchmarking and alignment of improvement efforts.