NCT04606849

Adaptation and Pilot Implementation of ePNa Clinical Decision Support for Utah Urgent Care Clinics

Adaptation and Pilot Implementation of a Validated, Electronic Real-Time Clinical Decision Support Tool for Care of Pneumonia Patients in 10 Utah Urgent Care Centers

• Status: Recruiting
• Phase: N/A
• Study type: Interventional
• Enrollment: 4,000 (estimated)
• Sponsor: Intermountain Health Care, Inc. · Academic / Other
• Sex: All
• Age: 12 Years
• Healthy volunteers: Accepted

Summary

We plan to adapt an innovative, validated emergency department (ED) CDS tool based on consensus guidelines for pneumonia care (ePNa) to function in urgent care clinics (Instacares at Intermountain) and combine it seamlessly with Stanford's CheXED artificial intelligence model using an interoperable platform currently under development by Care Transformation Information Services at Intermountain. We will then deploy it to one of two groups of Instacares (randomly selected) using the CFIR framework for Implementation Science best practice.

Detailed description

Clinicians' ability to accurately diagnose pneumonia and then choose the most appropriate treatment options is enhanced by well-designed clinical decision support (CDS). Pneumonia CDS has historically been focused on inpatient settings, but ambulatory care settings with high pneumonia patient volumes also might benefit. The investigators propose to adapt an innovative, validated emergency department (ED) CDS tool based on consensus guidelines for pneumonia care (ePNa) and deploy it to urgent care centers (UCC) using the CFIR framework. Electronic tools such as ePNa may become even more useful within UCCs as the COVID-19 pandemic evolves, since recommendations can be readily updated as better methods of diagnosis and effective treatment develop. ePNa within the ED has already been adapted to recommend SARS-coV-2 testing for patients with pneumonia and signs and symptoms characteristic of viral pneumonia. The proposal supports four aims: 1. Adapt ePNa for UCC and after in silico testing, pilot it among "super user" clinicians during UCC shifts and assess its usability. ePNa needs adaptation for more limited patient data available in UCCs, calibration of severity measures for lower observed mortality, and a chest imaging prompt in patients with pneumonia signs and symptoms. ePNa for UCC will incorporate Stanford University's artificial intelligence CheXED model to provide electronic classification of chest images in \<10 seconds for elements of pneumonia diagnosis and treatment (radiographic pneumonia, single vs multiple lobes, and pleural effusion). 2. Using the CFIR framework, our prior ED implementation experience, a focus group of UCC clinicians, semi-structured interviews, and direct observations of workflow including ePNa guided transitions of care between clinicians, the investigators will identify barriers and facilitators to adaptation and implementation of ePNa to UCCs. 3. Test the implementation strategy by deploying ePNa at one of two randomly chosen Intermountain Healthcare UCC clusters each with about 800 annual pneumonia patients - the other a usual care control. 4. Co-primary outcomes are a) accuracy of pneumonia diagnosis defined by compatible chief complaint plus ≥ 1 pneumonia sign/symptom and radiographic confirmation will be ≥10% higher in the ePNa cluster, and b) the percent of UCC pneumonia patients transferred to an emergency department for further evaluation will decrease by ≥ 3% in the ePNa cluster replaced by more direct hospital admissions or discharge home. Safety measures will be unplanned subsequent 7-day ED visits/hospitalizations and 30-day mortality. Based on this rigorous pilot study, the investigators anticipate a subsequent multi-system cluster-randomized trial. Our work incorporates the Five Rights of CDS to ensure that the strengths of this technology are optimized in the clinical environment. The investigators will leverage experience in innovative pneumonia research, pioneering CDS, and implementation science available at Intermountain to successfully complete this proposal.

Conditions

• Pneumonia

Interventions

Type	Name	Description
OTHER	Physician Survey	Our questionnaire includes questions on respondent demographics and Likert-style questions about respondent experiences with ePNa. We will validate our modified questionnaire by calculating component loadings and Cronbach Alphas (i.e., internal consistency) of Likert questions loading onto the same components.
DEVICE	ePNa-CheXED	ePNa-CheXED will incorporate Stanford University's artificial intelligence CheXED model to provide electronic classification of chest images in \<1 second for elements of pneumonia diagnosis and treatment (radiographic pneumonia, single vs multiple lobes, and pleural effusion).

Timeline

• Start date: 2020-11-12
• Primary completion: 2024-09-30
• Completion: 2024-09-30
• First posted: 2020-10-28
• Last updated: 2024-08-26

Locations

12 sites across 1 country: United States