SCCM is performing maintenance on its websites. For the best browsing experience, please use Microsoft Edge or Safari. Those using Chrome or Firefox may experience access issues at this time.

Advancing Research for Effective Alerts to Avoid Alert Fatigue

visual bubble
visual bubble
visual bubble
visual bubble

Sandra L. Kane-Gill, PharmD, MSc, FCCM,

The idea of clinical decision support (CDS) may generate some mixed feelings in clinicians. CDS has been used in many ways, such as for medication error reduction and as an early warning disease tool.1,2 Overall, CDS has demonstrated modest improvements in patient outcomes.2,3 Quality improvement efforts often focus on fixing problems with the aid of technology, including CDS; unfortunately, no follow-up assessment of CDS effectiveness has been conducted. The “set it and forget it” mentality results in numerous alerts, which are often sensitive but lack specificity. Oversensitivity results in nonactionable alerts.

In the context of medication-related CDS, 80% of alerts are appropriately overridden; however, overriding alerts is still a cause of concern because it is associated with an increased risk of adverse drug events (ADEs).4,5 This immunity in clinicians’ response to alerts and concern for missing an actionable alert is referred to as alert fatigue (see Table 1 for definitions).6-11 It is important to note that alert fatigue extends beyond the number of alerts received to more complex metrics involving appropriate responses, inappropriate alerts, number of overridden alerts, desensitization, and resulting harm.9 More rigorous studies of these detailed metrics, extending beyond quantity, are needed since quantity reduction alone has not consistently resulted in better clinician responsiveness.12

Some institutions are now devoting their efforts to improving the performance of alert systems to contend with user dissatisfaction and potential burnout.13-16 The development of an effective alert requires multiple iterations using advanced underlying knowledge and programming. Also needed is an active quality improvement plan to meet the intended outcome of alert development. Examples of possible alert modifications are shown in
Table 2. The overarching problem is that institutions make independent efforts to improve alert performance without sharing results, which is evident by the lack of studies on effective approaches to managing alert fatigue in critically ill patients.9 More institutions committed to improving the performance characteristics of alerts and a united public format for sharing results are needed.

Another area for improvement in alerts is to move away from detection and toward prevention.17 This is occurring in the case of early warning alerts for sepsis and acute kidney injury.2 However, some medication-related alerts, such as those for antidotes, are focused on the detection of an ADE after the injury has occurred rather than at the point of a drug-related hazardous condition, when there is still an opportunity for intervention.18 A preventive approach would be better. In the example of hyperkalemia, a preventive approach would mean generating an alert when the patient is receiving a drug that can cause hyperkalemia and the potassium level is 5.6 mEq/L without the patient having any ECG changes, rather than generating a detection alert for the administration of sodium polystyrene sulfonate. While detection alerts have a role, they are unlikely to have a major impact on averting ADEs and improving patient outcomes.

This call to action for effective alert development and alert fatigue reduction comes with substantial effort and cost. Unfortunately, obtaining funding for such research can be challenging because CDS advancement alone does not bring with it high marks for novelty. The key may be incorporating CDS advancement ideas with other innovative technologies or novel therapies for improvement of patient outcomes. Still, federal funding can be sought through organizations such as the Agency for Healthcare Research and Quality, which announces programs for developing new CDS to disseminate and implement evidence-based research findings and health information technology to improve healthcare quality and outcomes. The U.S. National Library of Medicine has funded projects on machine learning that can be incorporated into the enhancement of better-performing predictive alerts.19,20

Effective alert development is an area of much-needed research to allow the enhancement of patient safety by using health information technology for better surveillance in resource-conscious environments without burdening the clinician. Some day, it would be amazing to say that clinicians are 100% responsive to alerts because they know with confidence that the alerts they receive are 100% actionable and, as a result, patient safety has been extraordinarily improved.

References
 

  1. Nuckols TK, Smith-Spangler C, Morton SC, et al. The effectiveness of computerized order entry at reducing preventable adverse drug events and medication errors in hospital settings: a systematic review and meta-analysis. Syst Rev. 2014 Jun 4;3:56.
  2. Guidi JL, Clark K, Upton MT, et al. Clinician perception of the effectiveness of an automated early warning and response system for sepsis in an academic medical center. Ann Am Thorac Soc. 2015 Oct;12(10):1514-1519.
  3. Varghese J, Kleine M, Gessner SI, Sandmann S, Dugas M. Effects of computerized decision support system implemenatations on patient outcomes in inpatient care: a systematic reivew. J Am Med Inform Assoc. 2018 May 1;25(5):593-602.
  4. Wong A, Amato MG, Seger DL, et al. Evaluation of medication-related clinical decision support alert overrides in the intensive care unit. J Crit Care. 2017 Jun;39:156-161.
  5. Wong A, Amato MG, Seger DL, et al. Prospective evaluation of medication-related clinical decision support over-rides in the intensive care unit. BMJ Qual Saf 2018 Feb 9. [Epub ahead of print].
  6. Office of the National Coordinator for Health Information Technology. Clinical decision support: more than just ‘alerts’ tipsheet. https://www.healthit.gov/sites/default/files/clinicaldecisionsupport_tipsheet.pdf Last updated 2014. Accessed May 2, 2018.
  7. Kane-Gill SL, Visweswaran S, Saul MI, Wong AK, Penrod LE, Handler SM. Computerized detection of adverse drug reactions in the medical intensive care unit. Int J Med Inform. 2011 Aug;80(8):570-578.
  1. Agency for Healthcare Research and Quality. Patient Safety Primer. Alert fatigue. 2017 Available from: https://psnet.ahrq.gov/primers/primer/28. Last updated June 2017. Accessed May 2, 2018
  2. Kane-Gill SL, O’Connor MF, Rothschild JM, et al. Technologic distractions (part 1): summary of approaches to manage alert quantity with intent to reduce alert fatigue and suggestions for alert metrics. Crit Care Med. 2017 Sep;45(9):1481-1488.
  3. Agency for Healthcare Research and Quality. Glossary. Clinical decision support system. 
  4. Beeler PE, Bates DW, Hug BL. Clinical decision support systems. Swiss Med Wkly. 2014 Dec 23;144:w14073.
  5. Dexheimer JW, Kirkendall ES, Kouri M, et al. The effects of medication alerts on prescriber response in a pediatric hospital. Appl Clin Inform. 2017 May 10;8(2):491-501.
  6. Gouveia WA. Alert fatigue: a lesson relearned. Am J Health Syst Pharm. 2010 Apr 15;67(8):603-604; author reply 604.
  7. Hysong SJ, Spitzmuller C, Espadas D, Sittig DF, Singh H. Electronic alerts and clinician turnover: the influence of user acceptance. Am J Manag Care. 2014 Nov;20(11 Spec No 17):SP520-SP530.
  8. Gregory ME, Russo E, Singh H. Electronic health record alert-related workload as a predictor of burnout in primary care providers. Appl Clin Inform. 2017 Jul 5;8(3):686-697.
  9. Downing NL, Bates DW, Longhurst CA. Physician burnout in the electronic health record era: Are we ignoring the real cause? Ann Intern Med. 2018. doi:10.7326/M18-0139.
  10. Kane-Gill SL, Achanta A, Kellum JA, Handler SM. Clinical decision support for drug related events: moving towards better prevention. World J Crit Care Med. 2016 Nov 4;5(4):204-211.
  11. Kane-Gill SL, Dasta JF, Schneider PJ, Cook CH. Monitoring abnormal laboratory values as antecedents to drug-induced injury. J Trauma. 2005 Dec;59(6):1457-1462.
  12. U.S. National Library of Medicine. Research awards, 2004–present. https://www.nlm.nih.gov/ep/AwardsResearch.html. Accessed May 2, 2018.
  13. Sutherland SM, Chawla LS, Kane-Gill SL, et al. Utilizing electronic health records to predict acute kidney injury risk and outcomes: workgroup statements from the 15th ADQI consensus conference. Can J Kidney Health Dis. 2016 Feb 26;3:11.