Thread by @MariaGlymour

Dr Julia Adler-Milstein re turning digital fumes into fresh air (ie useful evidence for clinical care & system design) at @UCSF_Epibiostat seminar. Super cool new Center for Clinical Informatics and Improvement Research. medicine.ucsf.edu/cliir @CliirUcsf 1/n

Digital transformation: requires constant evolution & improving tools we use, which is achieved by observing how users interact w/ & use tools. In health care, we're still in early stages of this work-need to move into an era where we adopt continuous test/refine cycle.

@johnwitte: does better user experience=spending more $? JAM: maybe for Amazon, but not us.
Digital transformation enables new discovery of *what* care should be delivered AND *how* care should be delivered. Provide tools to support clinical decisions & care team design.

We spend a lot of time thinking about how to change patient behavior, but not much time thinking about how changing clinician behavior may improve care. Amazon is changing site every 11.6 secs in constant little RCTs to see what works (!!). Health care systems?

Tons of data elements are already captured -very granular data re what EHR data is viewed by whom, for how long, in what order, how data are documented, what care decisions are made, what outcomes patients experience. Twitter thread below redacted b/c papers are under review.

Study 1: Relationship b/w EHR use and MD burnout (=cynicism + exhaustion). Data=EHR use measures +burnout survey. Are there types of EHR activities that are contributing to burnout?

Exposures:
1.After hour minutes in EHR?
2.Minutes active on EHR on unscheduled days
3.Message volume from EHR
Did these predict cynicism or exhaustion?? We have to wait for the paper....

Study 2. Relationship between EHR use & MD gender. Estimate mixed effects model clustered by specialty.
Were there differences in minutes spent writing notes after hours or on unscheduled days for women vs men?
(results forthcoming)

Proposal from Mark Pletcher-why not evaluate time per RVU by specialty?

Examples of more data: event log or audit log. a running log of actions taken by a user within the EHR. Timestamped, workstation location, user. Most actions tied to a patient MRN and encounter ID.

Starting to work w/ clinical notes. To give a sense of data size: started with 1300 ED visits & subsequent hospitalizations generated 34.5K unique notes w/ 1.1 million actions by 7,700 users.

Study 3: Nat experiment on interoperability to allow clinicians to see outside records data. On a specific date, EHR began presenting an integrated list of encounters including from UCSF & from other EPIC using institutions. Did providers use outside records more after change?

Check out their center: medicine.ucsf.edu/cliir
@CliirUcsf and JAM: profiles.ucsf.edu/julia.adler-mi…