Alert placement in clinical workflow

From Clinfowiki
Jump to: navigation, search

Despite the potential of alerting systems to prevent potential errors, several studies have found that a large amount of the alerts are overridden by physicians.


A study on the prescriber override reasons revealed that the physicians thought [1]

  • the interactions were not clinically significant (21.6%)
  • the patient currently tolerated the drug (21.6%)
  • the patient was no longer taking 1 of the offending agents (8.0%)
  • the patient had previously tolerated the combination (12.3%)"

Studies suggest that one of the fundamental requirements for effective alerting is that eligible patients are accurately identified.[2] In most enterprise alerting systems, “an event monitor inspects the data stream in real time, and constantly “sniffs out” eligible patients. This task requires real time access to the relevant data stream, computer interpretable rules, and an inference engine”.[3]


A study by the Kaiser Permanente Clinical Information Systems Department, Portland, Oregon, was done to find a “viable method” of rendering quality alerts through off-line data analysis.


They chose to implement a low dose aspirin therapy alert as their first test alert. Data from the Kaiser Permanente Northwest (KPNW) region’s EMR systems were extracted nightly to a separate data warehouse. From the data warehouse, analysts identified a cohort of patients eligible for a clinical alert and their electronic records were marked with a “flag”. The patients in the cohort with documented aspirin allergies were excluded quarterly and patients who had an active aspirin order were determined weekly. Hence a newly documented aspirin allergy patient would remain in the cohort for up to 3 months while a new prescription would turn off the flag within 1 week. One hundred clinicians were randomly assigned either to a control group or to a group that received the alert when viewing the electronic medical record of eligible patients. 315 of the 580 patients (54.3%) seen by alerted clinicians were no longer eligible for the alert at the end of the one month study, compared to 128 of the 496 patients (25.8%) seen by control clinicians(p<.001).[3]

Results of off-line data analysis system

The results of this study suggests that off-line identification of an eligible patient cohort by flagging their electronic chart, and then presenting a predetermined alert to the clinician at the point of care can be an effective alerting strategy while conserving computing resources and speed for real time transaction processing. This would be particularly useful when there are multiple data sources as in a group practice Health Maintenance Organization where not all the data is accessible to an event monitor in real time. Accurate patient active medication and allergy lists, rapid update of the cohort database, and a mechanism to easily document exceptions will be required to create quality alerts. Though there were no differences in response by specialty or gender of the clinicians in this study, how long the effect would persist over time remains to be found.

Previous studies have indicated there may be decline in benefit over even short periods (< 6 months) when alerts remain active. [4] But off-line data analysis could prove to be an effective method of initiating a clinical alert and can be an effective alternative to an event monitor. Implementing an off-line data analysis system could save time and increase the speed of the system.


  1. A. J. Grizzle, M. H. Mahmood, Yu Ko, J. E. Murphy, E. P. Armstrong, G. H. Skrepnek, W. N. Jones, G. P. Schepers W. Paul Nichol, A. Houranieh, D. C. Dare, C. T. Hoey, and D. C. Malone. Reasons Provided by Prescribers When Overriding Drug-Drug Interaction Alerts. Am J Manag Care. 2007;13:573-580
  2. Krall MA, Sittig DF. Clinician’s assessments of outpatient electronic medical record alert and reminder usability and usefulness requirements. AMIA Proceedings 2002:400-404.
  3. Effectiveness of an Electronic Medical Record Clinical Quality Alert Prepared by Off-line Data Analysis. Michael A. Krall, Kati Traunweiser, William Towery
  4. Demakis JG, Beauchamp C, Cull WL, Denwood R, et al. Improving residents’ compli-ance with standards of ambulatory care. Results from the VA cooperative study on computerized reminders. JAMA 2000;284:1411-1416.

Neena Elizabeth Augustine BMI 512