JMIR Publications

ABSTRACT

Background:

Prior literature suggests that alert dismissal could be linked to habits and automaticity formed by physicians. The evidence for this perspective is largely observational. This study uses log data from an electronic medical records system to empirically validate this perspective.

Objective:

We seek to quantify the prevalence of alert dismissal due to habitual dismissal and the role of habitual dismissal in impacting subsequent alert dismissal.

Methods:

We conduct a retrospective analysis using 66,049 alerts log data of hospitalized patients in a hospital from March 2017 to December 2018. We examine a specific clinical support alert triggered in a hospital’s electronic medical record system and record the time taken by the physician to respond to the alert. We collect other clinical and environmental factors related to the alert as covariates for the analysis.

Results:

Habitual dismissal of alerts are characterized by very short processing time. 72.5% of alerts are dismissed under three seconds after presentation, and 13.2% of all alerts are dismissed under one second. Physicians who habitually dismiss alerts are more likely to dismiss and habitually dismiss subsequent alerts. A one percent increase in physician's total habitual dismissal rate (cumulative number of habitual dismissal by physician ÷ cumulative total alert exposure received by the physician) is associated with an increase in odds for habitually dismissing another subsequent alert by 6 to 21 times.

Conclusions:

We find empirical support that habitual dismissal is one of the key factors driving alert dismissal. We also find that habitual dismissal is self-reinforcing, suggesting possible challenges in disrupting or changing alert dismissal habits once they are formed. Habitual dismissal of alerts is pervasive across all physician experience levels, and the effect is self-reinforcing.