Contact email

erittenberg@bwh.harvard.edu

Overview

This project within the inpatient setting of an academic medical center utilized an automated alert system for critical laboratory results to decrease unnecessary delays in appropriate treatment. 

Collaborators

Jonathan M. Teich, MD, PhD
Milenko J. Tanasijevic MD
Nell Ma’Luf
Eve Rittenberg, MA
Ashish Jha, MD
Julie Fiskio
James Winkelman, MD

Status/Stage of Development

Terminated

Funding Sources

This work was supported in part by research grant RO1-HS08927 from the Agency for Health Care Policy and Research.

Practice Setting

Academic Medical Center

National/Policy Context

• Errors and delays are common in medical practice. Increasingly, health information technology is being used to reduce errors by helping systems detect and inform clinicians about key clinical events. 

Local/Organizational Context

• When the project was implemented in 1996, approximately 3 million inpatient chemistry and hematology results per year were generated by Brigham and Women’s Hospital laboratories, of which ~1% were critical results. Prior to intervention, the policy was for critical results to be called by the laboratory technologists to the patient’s floor to a unit secretary or a nurse as soon as they were available, prior to when they appeared in the hospital’s record.  

Patient Population Served and Payor Information

• Patients served by this study were medical and surgical inpatients. 

Leadership

• Drs. Kuperman and Bates started this and worked in close conjunction with the lab team, especially Drs. Tanasijevic and Winkelman.  This is no longer active.

Project Research + Planning

• More than a year was spent developing the consensus on which “events” merited the creation of corresponding alerts.

Tools or Products Developed

• Alerting Rules: 12 alerting rules were developed to cover “events” including single lab values, lab values changing over time, and a drug-laboratory interaction based on a set of alerting conditions developed previously (a full list can be found in Table 1 of Kuperman et al).
  • Examples of alerting rules included: 
    • Hematocrit has fallen 10% or more since last result and is now less than 26%
    • Serum glucose is greater than or equal to 400 mg/dL  
    • Serum potassium less than 3.3 mEq/L and patient has an active order for digoxin

Training

• No training required. 

Tech Involved

• Desktop computer
• Electronic medical record
• Email
• Pagers
• Statistical software

Team Members Involved

• Administrative Assistant
• NPs
• Physicians
• Support Staff

Workflow Steps

• A continuously running event monitor algorithm determines whether any new patient data matches any of the critical result alerting criteria. 
• If an alert is generated, a notification system automatically pages the responding clinician for the patient, determined from an automated coverage list database
• The physician can log onto any computer workstation to review the alert on a screen which displays patient identifying information, the alert message, alert details, active medications relevant to the alert, and alert specific action suggestions such as orders for therapeutics or additional lab tests. The suggestions can be completed simply with one additional keystroke or completion of an order. 
• If there is no physician response within 15 minutes, a fail safe notification sequence is initiated. The border of the computer screens on the patient’s floor turns red indicating a patient on the floor has an automated result, indicating to nurses on the floor to review the alert. 
• If after 30 minutes the alert has not been reviewed a computer workstation in the telecommunications office begins to beep. A telephone operator then reviews the alert and calls the patient’s floor with details of the alert and records the name of the person to whom the details were relayed. 

Outcomes

• Time to Treatment: The main outcome measured was the time from when a critical result was available to the time an appropriate treatment was ordered. The intervention group had a 38% shorter time interval until appropriate treatment was ordered (median of 1 hour vs. 1.6 hours and mean of 4.1 vs. 4.6 hours, p = 0.003).
• Time until resolution of critical condition: The time until the alerting condition resolved was less in the intervention group (median of 8.4 hours vs. 8.9 hours, mean of 14.4 hours vs. 20.2 hours, p = 0.11). 
• Frequency of adverse events: There was no significant difference between groups for adverse events. 

Future Outcomes

•  Not applicable.

Benefits

• The automated system reduced the time for appropriate treatment after a critical result was generated. This supports the use of information technology to help alert physicians of important clinical events and improve patient safety. 

Intervention-Specific Challenges

• It is time consuming to determine which events are worthy of an automated page, automated email, or other special communication methods. It is also time consuming specifying new rules unique to an institution.
• Difficulty transferring electronic knowledge between institutions because of practice pattern variations and differing technical systems. 
• There is a risk of overburdening physicians if the number of rules for automated alerts expands
• The technical requirements for the project are non-trivial to implement if not already existing. However, in the time span since this project was initially published, improvement of existing technology infrastructure may make similar implementation today more straightforward.

Glossary

• Critical Result: Signs of major perturbations of key physiologic systems. Examples include a drop in hematocrit of greater than 10% and now less than 26%, serum glucose higher than 400, serum potassium greater than 6 mEq/L or more, etc. The 12 included in this study can be found in Table 1 of the study, cited below. 

Sources

• Kuperman, G. J., Teich, J. M., Tanasijevic, M. J., Ma’Luf, N., Rittenberg, E., Jha, A., … Bates, D. W. (1999). Improving response to critical laboratory results with automation: results of a randomized controlled trial. Journal of the American Medical Informatics Association : JAMIA, 6(6), 512–522. doi:10.1136/jamia.1999.0060512
• Tate KE, Gardner RM, Weaver LK. A computerized laboratory alerting system. MD Comput. 1990;7(5):296-301