Electronic Health Record (EHR)-based asynchronous messaging is a form of secure, email-style message exchange used widely by healthcare teams to communicate time-sensitive patient information without requiring sender and receiver to engage simultaneously. Asynchronous alerts are messages whose primary goal is to prompt timely, relevant healthcare team action aimed at improving patient outcomes. Unlike alerts that interrupt users mid-task, asynchronous alerts are delivered to a secure electronic """"""""In-Basket"""""""" where they wait to be opened. While numerous studies describe interruptive alerts, little is known about the use, impact and risk of asynchronous alerts. This is an important gap. Often asynchronous alerts serve as the only notification to healthcare team members of time-sensitive issues;unopened alerts represent a potentially dangerous communication breakdown. Through previous AHRQ funding (R18 HS017203) we implemented an alert intervention notifying primary care teams of medication safety concerns and the need for close office follow-up post hospitalization. Automated asynchronous alerts were sent following 1282 hospital discharges of elderly patients but we found no evidence of an impact on patient outcomes. Preliminary review showed that 40% of staff-directed alerts and 19% of primary care provider alerts remained unopened at the end of day 1 post- delivery. A variety of sociotechnical factors likely influenced alert opening and subsequent actions. We compiled a rich collection of data with detailed information on alerts, the discharges triggering those alerts, contextual factors (e.g. number of unopened messages in the recipient's """"""""In-Basket"""""""" at the time of alert delivery), patient and provider characteristics, and clinical outcome data. We enriched this data with electronic tracking data (""""""""digital crumbs"""""""") recording the date and time of the creation and review of EHR data elements. We propose a secondary data analysis to identify contextual factors, types of alerts and characteristics of providers, patients and hospital discharges that relate to (a) timely opening of alerts following receipt and (b) timely first responsive action after alerts are receive and opened. We will also conduct a sequential pattern analysis describing the relevant actions taken by healthcare providers immediately following alert opening. Finally, we will assess the relationship between timely first responsive action following opening of alerts and key patient outcomes (office visits, rehospitalization, ER and urgicare visits, and adverse drug events). Our study is timely and innovative in its proposed use of digital crumbs, a largely untapped data source that will allow us to identify points of communication breakdown on the alert-response pathway. Epic Systems Corporation, the EHR system from which we have gathered our data, stores medical information for close to 127 million patients or nearly 40% of the U.S. population. What our research team learns by using digital crumbs to track alert opening and response will be broadly applicable to healthcare teams and researchers across the nation, and will inform design and implementation of future asynchronous alert systems.

Public Health Relevance

Electronic Health Record (EHR)-based asynchronous alerts are used widely to deliver time-sensitive information and prompt healthcare team action. Alerts are intended to improve patient outcomes but little data exists on factors influencing alert effectiveness or on the potential risks incurred when alerts remain unopened. Building on our prior work implementing an asynchronous alert intervention and making innovative use of electronic tracking data ('digital crumbs') which capture date and time of EHR element review, we propose a secondary database analysis aimed at identifying factors associated with timely alert opening, timely response and patient outcomes.

National Institute of Health (NIH)
Agency for Healthcare Research and Quality (AHRQ)
Exploratory/Developmental Grants (R21)
Project #
Application #
Study Section
Health Care Technology and Decision Science (HTDS)
Program Officer
Bernstein, Steve
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of Massachusetts Medical School Worcester
Internal Medicine/Medicine
Schools of Medicine
United States
Zip Code
Sadasivam, Rajani Shankar; Cutrona, Sarah L; Kinney, Rebecca L et al. (2016) Collective-Intelligence Recommender Systems: Advancing Computer Tailoring for Health Behavior Change Into the 21st Century. J Med Internet Res 18:e42