The impact of a point-of-care visual clinical decision support tool on admissions for cellulitis in the University of Maryland medical system

Abstract

Introduction: Cellulitis is commonly diagnosed in emergency departments (EDs), yet roughly one third of ED patients admitted for presumed cellulitis have another, usually benign, condition instead (eg, stasis dermatitis). This suggests there is an opportunity to reduce health care resource use through improved diagnosis at the point of care. This study seeks to test whether a clinical decision support (CDS) tool interoperable with the electronic medical record (EMR) can reduce inappropriate hospital admissions and drive more appropriate and accurate care.

Methods: This study was a trial of an EMR-interoperable, image-based CDS tool for evaluation of ED patients with suspected cellulitis. At the point of assigning a provisional diagnosis of cellulitis in the EMR, the clinician was randomly prompted to use the CDS. Based on the patient features entered into the CDS by the clinician, the CDS provided the clinician a list of likely diagnoses. The following were recorded: patient demographics, disposition and final diagnosis of patients, and whether antibiotics were prescribed. Logistic regression methods were used to determine the impact of CDS engagement on our primary outcome of admission for cellulitis, adjusted for patient factors. Antibiotic use was a secondary end point.

Results: From September 2019 to February 2020 (or 7 months), the CDS tool was deployed in the EMR at 4 major hospitals in the University of Maryland Medical System. There were 1269 encounters for cellulitis during the study period. The engagement with the CDS was low (24.1%, 95/394), but engagement was associated with an absolute reduction in admissions (7.1%, p = 0.03). After adjusting for age greater than 65 years, female sex, non-White race, and private insurance, CDS engagement was associated with a significant reduction of admissions (adjusted OR = 0.62, 95% confidence interval (CI): 0.40-0.97, p = 0.04) and antibiotic use (Adjusted OR = 0.63, 95% CI: 0.40-0.99, p = 0.04).

Conclusions: CDS engagement was associated with decreased admissions for cellulitis and decreased antibiotic use in this study, despite low levels of CDS engagement. Further research should examine the impact of CDS engagement in other practice environments and measure longer-term outcomes in patients discharged from the ED.

Keywords: clinical decision support; dermatology; emergency medicine; health care use; infectious disease.

FIGURE 1

The Clinical Decision Support (CDS) prompt as implemented in EPIC. The link to the software is shown (green box and arrow).

FIGURE 2

Consolidated Standards of Reporting Trials (CONSORT) diagram for study.