287 - Identifying Inpatient Pediatric Prescribing Errors Associated with Prescriber and Contextual Factors

Publication Number: 287.5451

Daniel Tawfik, Stanford University School of Medicine, Palo Alto, CA, United States; Liem M.. Nguyen, Stanford University School of Medicine, Palo Alto, CA, United States; Dane Jacobson, Stanford University, reno, NV, United States; Kimberley Kirk, Stanford University School of Medicine, Palo Alto, CA, United States; Jochen Profit, Stanford University School of Medicine, Palo Alto, CA, United States; Christian Rose, Stanford University School of Medicine, Palo Alto, CA, United States; A Jay Holmgren, University of California, San Francisco, School of Medicine, San Francisco, CA, United States; Thomas Kannampallil, Washington University in St. Louis School of Medicine, St Louis, MO, United States; Stefanie Sebok-Syer, Stanford University School of Medicine, Palo Alto, CA, United States

Background: Up to 20% of hospitalized children experience a preventable adverse drug event, yet only 1% of prescribing errors are reported. Automated error detection tools (e.g., the wrong-patient retract-and-reorder tool) show promise for reliably identifying errors, but are limited in the specific error types they detect.

Objective: Using electronic health record (EHR) data and associated metadata about prescribing activities (e.g., order voiding and modifications), this study aimed to identify patterns consistent with pediatric prescribing errors, and quantify their associations with prescriber and contextual factors.

Design/Methods: This retrospective, observational study included inpatient pediatric prescribers who placed prescriptions from 2020 through 2022. We excluded prescribers who placed fewer than 50 prescriptions and patients admitted < 48 hours to reduce small sample bias. We identified 5 indicators of intercepted prescribing errors reflecting wrong-patient, wrong-dose, wrong-route (parenteral vs. enteral vs. inhaled vs. other), and wrong-medication (within the same pharmaceutical class) self-intercepted errors, as well as pharmacist-intercepted wrong-dose errors.

Results: We quantified 1911 prescribers, 1,033,642 prescriptions, and 16,244 (1.6%) suspected errors (Figure 1), including 219 (2% of dosing errors) with suspected transposition of weight-based vs. absolute dosing and 1537 (11% of dosing errors) with >10-fold dosing error. Intercepted errors were independently associated with patient, prescriber, and time contextual factors (Table 1). For patient context this included prescribed dose below patient’s prior prescribed range for the medication (OR 3.86, 95%CI 3.66-4.08) and prescription for a medication that is new to the patient (OR 1.24, 95%CI 1.20-1.28) For provider context this included prior prescriptions of the medication by the prescriber (OR 0.92 per 100 orders, 95%CI 0.90-0.93) and years of experience (OR 0.94 per year, 95%CI 0.92-0.95). For time context this included evening orders (OR 1.09, 95%CI 1.03-1.16).

Conclusion(s): Intercepted prescribing errors are common within inpatient pediatrics, with higher incidence than manual error reports and congruent with previously published estimates of algorithmically-detected errors in other settings. The overall detected error incidence is 14-fold that of wrong-patient errors alone and over half the incidence of these error types found by manual audit (89 out of 2955, 3%), with several identified patient and provider context associations, highlighting the potential for expanding error detection capabilities to promote patient safety.

Figure 1
Action sequences for 5 methods of suspected prescribing error detection, along with incidence for each among 1,033,642 inpatient pediatric prescriptions from 2020 through 2022.

Table 1
Multivariable logistic regression results for intercepted prescribing errors, among 1,033,205 inpatient pediatric prescriptions. Model also adjusted for month and year.