What are the key risk factors for drug-related problems in Emergency Departments? A multinational modified Delphi study

Technical Summary: Key Risk Factors for Drug-Related Problems in Emergency Departments

Problem Statement
Drug-related problems (DRP) represent a significant public health challenge, contributing to substantial morbidity and mortality, with up to 16.6% of hospital admissions linked to adverse events and approximately 10% specifically drug-related. In Emergency Departments (EDs), the early identification of patients at high risk for DRPs is critical for targeted intervention. However, existing predictive models often rely on low-level evidence or utilize data points that are not readily available in real-time clinical practice, limiting their utility. Furthermore, the lack of 24-hour pharmacist coverage in most EDs necessitates efficient prioritization tools to allocate limited pharmaceutical resources effectively. Currently, clinicians rely on variable, experience-dependent processes to identify patients requiring medication review, highlighting a gap for structured, consensus-based screening tools.

Methodology
This study employed a multinational, multidisciplinary modified Delphi method, specifically the RAND/UCLA Appropriateness Method, to establish expert consensus. The panel consisted of 51 healthcare professionals, including hospital pharmacists, physicians, nurses, and social workers, primarily from Spain, with additional representation from Australia and the United States.

The study proceeded through the following stages:

1. Open Consultation: An initial list of 33 DRP risk factors, derived from literature and expert experience, was presented to participants. During this phase, the panel added 22 new factors, modified 1, and excluded 2, resulting in a final list of 53 factors categorized into clinical situation, patient characteristics, and chronic treatment characteristics.
2. Delphi Rounds: Two iterative rounds of structured evaluation were conducted. Participants rated each factor on three dimensions using a 10-point Likert scale or binary response:
   • Relevance/Scientific Evidence (1 = very low, 10 = very high).
   • Real-time availability of data (Yes/No).
   • Ease of data retrieval in clinical practice (1 = very difficult, 10 = very easy).
3. Consensus Analysis: Consensus was determined using the RAND/UCLA criteria, comparing the Interpercentile Range (IPR) against the IPR adjusted for symmetry (IPRAS). Factors were classified as "Appropriate" (median >6), "Uncertain" (median 4–6), or "Inappropriate" (median <4). Disagreement was defined as an IPR greater than the IPRAS.

Key Results
Following the open consultation and two Delphi rounds (with participation rates of 84% in Round 1 and 78% in Round 2), the study yielded the following findings regarding the 53 evaluated factors:

• Classification: Only one factor, "employment status," was classified as inappropriate (median score of 3). Seven factors were classified as uncertain, including requests for non-surgical intervention, mode of transport, gender, private insurance, primary care visit timing, specialist consultation timing, and body mass index.
• Disagreement: Significant disagreement was observed regarding 21 factors, indicating a lack of consensus on their predictive value or data availability.
• High-Relevance Factors: The factors identified as most relevant (appropriate) included high-risk prehospital drug administration, requests for pharmacist intervention, and chronic drug use identified by STOPP (Screening Tool of Older Persons' Prescriptions) criteria. Other robust candidates included the number of chronic drugs, renal impairment, and polypharmacy with multimorbidity.
• Data Availability: The study highlighted variability in the perceived real-time availability and ease of retrieving data for specific factors, which is crucial for the integration of screening tools into electronic health records.

Significance and Claims
The authors position this study as a foundational step rather than a definitive validation. The primary contribution is the generation of a consensus-derived set of candidate risk factors that can inform the development of screening tools to prioritize pharmaceutical care in EDs.

The paper claims that:

• Tool Development: These consensus factors may support the creation of screening tools to help emergency department pharmacists prioritize patients for medication reviews.
• Implementation Feasibility: By assessing real-time data availability and ease of retrieval, the study addresses the practical barriers to integrating prioritization tools into routine ED workflows and electronic health records.
• Resource Allocation: The findings aim to facilitate the earlier identification of DRPs and more efficient allocation of limited pharmacy resources.

The authors explicitly state that these findings should be regarded as a basis for model development rather than a validated prediction rule. They emphasize the need for prospective multicenter validation to evaluate the predictive performance, clinical utility, and actual impact on medication safety and resource allocation. The study acknowledges limitations, including potential fatigue from the expanded factor list, ambiguity in interpreting data availability, and the reliance on expert opinion rather than empirical prospective data.