Rich sources of data are available in hospitals, but are often housed in different systems. These include electronic health records (EHRs), physician scheduling software, and existing tools that can be used to electronically contact clinicians. Developing a streamlined way to integrate and connect disparate data has enormous potential for clinical care, operations, and research.
Purpose:
To develop and pilot-test a novel software platform linking clinical data from the EHR with an automated, personalized physician alert and survey system.
Description:
Background and Development:
Our 600-bed hospital uses various software platforms to facilitate clinical operations, including an Epic™-based EHR and the AMiON physician scheduling system. Our hospitalist group developed a software program, called “Murmur”, that could draw information from the EHR, AMiON, and a voluntarily-provided internal database of cellular phone numbers. We programmed Murmur to deliver text messages to provider phones (with their permission) that included a personalized link to a secure, HIPAA-compliant REDCap™ survey, formatted for a smartphone screen. An automatic timer on a desktop computer ran Murmur every day. The clinician end-users did not need to download or learn any new software.
Use case testing:
We tested Murmur in two ways: first, to conduct daily surveys of hospitalists on service, and second, to automatically alert physicians of a clinical event based on data obtained from the EHR. (Figure)
Case 1: We used Murmur to ask physicians on service to identify patients for whom a delay in care will lead to an unnecessary day in the hospital. Murmur identified the hospitalists working each day and texted each a personalized survey link with questions about potential delays. During a one-month trial, Murmur collected 74 responses about avoidable hospital days.
Case 2: Murmur performed a daily query of the EHR to identify patients readmitted to the hospital within five days of a prior admission, and texted both the discharge and readmitting attendings. This text included a survey link with questions about preventability. A time gap between the texts allowed Murmur to record a “sign out” from the discharge attending and forward it to the readmitting attending. During a two-week trial, Murmur notified attendings about 14 readmitted patients.
Conclusions:
We created Murmur, a low-cost, easy-to-use, automated software platform that leveraged existing data systems to personalize data collection and generate alerts to practicing clinicians. This software, with small modifications, could be implemented at any institution that has access to the underlying building blocks (e.g. Epic EHR, REDCap, etc.).