Fail-safe: Automating Critical-results Notification
The radiology department at Brigham and Women’s Hospital (BWH), Boston, Massachusetts, developed a policy for communicating critical and discrepant results after the Joint Commission made communications among caregivers a national priority for health-care providers. When the goal was expanded in 2007, the department took the next step and used IT to automate the process and to embed it into radiologist and referring-physician workflow. 
Nonetheless, the journey begins with a policy based on strict definitions of critical and discrepant findings, according to Luciano Prevedello, MD, a neuroradiology and informatics fellow in BWH’s radiology department. He described the solution in “Methods for Effective Communication and Reporting,” presented on June 4, 2010, at the annual Society for Imaging Informatics in Medicine meeting in Minneapolis, Minnesota. Noting that it is common, in the reading room, to hear radiologists question whether a particular result is critical, he says, “If you don’t have strict definitions of what a critical or discrepant finding is, then it is going to be a problem for people to use the policy and be compliant.”
BWH must have gotten it right. Since February 2006, the department’s compliance has moved from 30% to almost 95% in 2010. This is how they did it.Setting a PolicyIn 2004, the Joint Commission released a list of National Patient Safety Goals to be used in accrediting health-care providers, and goal number 2 was of particular relevance to radiology: Improve effectiveness of communication among caregivers. With each passing year, the Joint Commission added further requirements to this goal, until by 2007, it included four specifications for hospitals:
• for verbal or phone orders, as well as for reporting critical test results by phone, verify the complete order (or the test results) by having the person who receives the order or result record the information and read it to the speaker/caller;
• create a standardized list, applicable throughout the organization, of abbreviations, acronyms, symbols, and dose designations that are not to be used;
• measure and assess timelines for the reporting of all critical test results and for the receipt of those results by the responsible licensed caregiver; as needed, take action to improve those timelines; and
• implement a standardized approach to handing off messages or other communications, including in that approach a chance for both parties to ask and respond to questions.
BWH began by creating a policy for all critical and discrepant results that included mandating use of the language found in the templateapproved by the Structured Reporting Subcommittee of the RSNA in all reports that have critical or discrepant results.
“In our system, a critical result is defined as a new or unexpected finding that could result in mortality or significant morbidity if appropriate diagnostic or therapeutic follow-up steps are not undertaken,” Prevedello explains. “Discrepant findings are defined as an interpretation that is significantly different from a preliminary interpretation, when the preliminary interpretation has been accessible to the patient-care team and the difference in the interpretation may alter the patient’s diagnostic work-up or management.”Stratifying Critical ResultsCritical alerts were divided into three categories (with appropriate responses).
Red: findings that are potentially immediately life threatening; these require immediate interruptive communication, such as face-to-face communication or phone contact.
Orange: findings that could result in mortality or significant morbidity if not appropriately treated urgently; these require face-to-face communication and phone contact.
Yellow: findings that could result in mortality or significant morbidity if not appropriately treated, but that are not immediately life threatening or urgent; face-to-face contact, phone contact, or other verifiable methods of communication are required. Some hospitals do not include these results as critical, but BWH chose to set the bar high.
The time frames for notifying the appropriate personnel (and documenting that notification) were established as less than 60 minutes for red alerts, within three hours for orange alerts, and within three days for yellow alerts.
In compliance with the recommendation of the RSNA subcommittee, the policy requires documentation to be included in the radiology report. It must contain the following information: the name of the communicator, the name of recipient of that notification, the date and time that the result was reported, and a simple statement such as, “Critical findings were communicated by [radiologist’s name] to [surgeon’s name] at 5 pm, Wednesday, December 15, 2005.”
After it first implemented the policy, in February 2006, BWH monitored compliance with the policy using periodic manual review of one day’s worth of reports generated in the department. Section heads were required to review reports manually and to discuss any lapses with individual faculty members. Automating the ProcessAdherence increased dramatically, but there were still several opportunities to improve the process. The manual review required of section heads was labor intensive, and the standard communication methods—paging and phone calls—were inefficient because they can be interruptive, especially when findings do not pose an immediate safety risk.
“For example, the surgeon is operating and is paged and notified that there is a pulmonary nodule on a patient,” Prevedello posits. “You are interrupting at a time when it may be impossible to act appropriately on that finding, and the surgeon may later forget, so there are problems with communicating synchronously or interruptively.”
Standard email also has gaps, including documentation in the medical record and verification of receipt, Prevedello adds. “If you do not close the communications loop, we do not consider that compliant communication,” he says.
Before BWH automated the process, it identified a list of technology requirements that would need to be met. First, radiologists would generate alerts; therefore, the process would have to be embedded in the radiologists’ workflow. Second, responsible providers had to be able to consume those alerts within their workflow, and both radiologists and referrers would need access to alert audit trails. Third, those alerts would need to be patient centered, not exam centered. Fourth (and most critical), the application had to be integrated into existing IT systems.Radiologist WorkflowWith those requirements in mind, BWH developed Alert and Notification of Critical Results (ANCR), funded by a multicenter grant from The Risk Management Foundation of the Harvard Medical Institutions through CRICO (Cambridge, Massachusetts), a malpractice-insurance company. The system was developed at the BWH Center for Evidence-based Imaging, under the guidance of Ramin Khorasani, MD. Beth Israel Deaconess Medical Center, Boston, and the University of Chicago also collaborated.
To avoid an extra sign-on step, ANCR was integrated into the physician-authentication database and also with the PACS (so that all of the information that passed through the alert-notification system would be integrated, in context, with the PACS information). The email and paging systems also needed to be connected to give physicians the same ability to find each other that they had prior to automation.
When the radiologist generates an alert, all of the stratified notification methods are displayed along, with their descriptions, and the radiologist selects the color. For red and orange alerts that need to be communicated right away, the radiologist next selects the provider to whom the results will be communicated. A brief description of the finding is displayed on the screen with patient information, all in context from PACS. ANCR then sends a notification to the provider via pager. The physician is notified, he or she calls the radiologist or a face-to-face conversation takes place, and the radiologist closes the loop. All this is documented within ANCR.
Yellow alerts, which have a three-day turnaround, work differently. ANCR sets up that notification to be sent by email. Once the notification is submitted via email, it is added to the radiologist’s list of all pending alerts that he or she has generated until the loop is closed. Referrer FriendlyReferring-provider workflow is also integrated with the authentication database and with the email and paging systems, enabling the referrer to review the images, review the radiology report, and order follow-up studies without logging into other systems.
For red and orange alerts, referrers receive a page with the radiology-result notification, the patient information, and a request to call the radiologist, prompting immediate closure of the communications loop.
For yellow alerts, referrers receive an email with a link embedded that goes to the ANCR application. Once logged in, referrers can see the images and the full radiology report and can order follow-up studies. They can also click a button that enables them to see the contact information for that radiologist, in the event that they want to discuss the results. Finally, they click a button that says, “I acknowledge these findings, I understand the results,” and the radiologist is notified that the communications loop is closed.
ANCR also brings the institution significant quality-assurance benefits in that clinical administrators can track pending alerts (as well as alert-notification compliance) by individual physicians. “You can press a button and find out what portion of critical results are red, yellow, or orange; you can see the turnaround time for communication, notification, and acknowledgement; you can understand compliance by single physician; and you can pretty much drill down into this information in a matter of seconds,” Prevedello says. “Suboptimal communication handoffs are important sources of errors and can negatively affect safety and quality, which is why communication of critical results is included in the National Patient Safety Goals,” Prevedello concludes.Cheryl Proval is editorial director of RadInformatics.com and vice president, publishing, of imagingBiz, Tustin, California.
Nonetheless, the journey begins with a policy based on strict definitions of critical and discrepant findings, according to Luciano Prevedello, MD, a neuroradiology and informatics fellow in BWH’s radiology department. He described the solution in “Methods for Effective Communication and Reporting,” presented on June 4, 2010, at the annual Society for Imaging Informatics in Medicine meeting in Minneapolis, Minnesota. Noting that it is common, in the reading room, to hear radiologists question whether a particular result is critical, he says, “If you don’t have strict definitions of what a critical or discrepant finding is, then it is going to be a problem for people to use the policy and be compliant.”
BWH must have gotten it right. Since February 2006, the department’s compliance has moved from 30% to almost 95% in 2010. This is how they did it.Setting a PolicyIn 2004, the Joint Commission released a list of National Patient Safety Goals to be used in accrediting health-care providers, and goal number 2 was of particular relevance to radiology: Improve effectiveness of communication among caregivers. With each passing year, the Joint Commission added further requirements to this goal, until by 2007, it included four specifications for hospitals:
• for verbal or phone orders, as well as for reporting critical test results by phone, verify the complete order (or the test results) by having the person who receives the order or result record the information and read it to the speaker/caller;
• create a standardized list, applicable throughout the organization, of abbreviations, acronyms, symbols, and dose designations that are not to be used;
• measure and assess timelines for the reporting of all critical test results and for the receipt of those results by the responsible licensed caregiver; as needed, take action to improve those timelines; and
• implement a standardized approach to handing off messages or other communications, including in that approach a chance for both parties to ask and respond to questions.
BWH began by creating a policy for all critical and discrepant results that included mandating use of the language found in the templateapproved by the Structured Reporting Subcommittee of the RSNA in all reports that have critical or discrepant results.
“In our system, a critical result is defined as a new or unexpected finding that could result in mortality or significant morbidity if appropriate diagnostic or therapeutic follow-up steps are not undertaken,” Prevedello explains. “Discrepant findings are defined as an interpretation that is significantly different from a preliminary interpretation, when the preliminary interpretation has been accessible to the patient-care team and the difference in the interpretation may alter the patient’s diagnostic work-up or management.”Stratifying Critical ResultsCritical alerts were divided into three categories (with appropriate responses).
Red: findings that are potentially immediately life threatening; these require immediate interruptive communication, such as face-to-face communication or phone contact.
Orange: findings that could result in mortality or significant morbidity if not appropriately treated urgently; these require face-to-face communication and phone contact.
Yellow: findings that could result in mortality or significant morbidity if not appropriately treated, but that are not immediately life threatening or urgent; face-to-face contact, phone contact, or other verifiable methods of communication are required. Some hospitals do not include these results as critical, but BWH chose to set the bar high.
The time frames for notifying the appropriate personnel (and documenting that notification) were established as less than 60 minutes for red alerts, within three hours for orange alerts, and within three days for yellow alerts.
In compliance with the recommendation of the RSNA subcommittee, the policy requires documentation to be included in the radiology report. It must contain the following information: the name of the communicator, the name of recipient of that notification, the date and time that the result was reported, and a simple statement such as, “Critical findings were communicated by [radiologist’s name] to [surgeon’s name] at 5 pm, Wednesday, December 15, 2005.”
After it first implemented the policy, in February 2006, BWH monitored compliance with the policy using periodic manual review of one day’s worth of reports generated in the department. Section heads were required to review reports manually and to discuss any lapses with individual faculty members. Automating the ProcessAdherence increased dramatically, but there were still several opportunities to improve the process. The manual review required of section heads was labor intensive, and the standard communication methods—paging and phone calls—were inefficient because they can be interruptive, especially when findings do not pose an immediate safety risk.
“For example, the surgeon is operating and is paged and notified that there is a pulmonary nodule on a patient,” Prevedello posits. “You are interrupting at a time when it may be impossible to act appropriately on that finding, and the surgeon may later forget, so there are problems with communicating synchronously or interruptively.”
Standard email also has gaps, including documentation in the medical record and verification of receipt, Prevedello adds. “If you do not close the communications loop, we do not consider that compliant communication,” he says.
Before BWH automated the process, it identified a list of technology requirements that would need to be met. First, radiologists would generate alerts; therefore, the process would have to be embedded in the radiologists’ workflow. Second, responsible providers had to be able to consume those alerts within their workflow, and both radiologists and referrers would need access to alert audit trails. Third, those alerts would need to be patient centered, not exam centered. Fourth (and most critical), the application had to be integrated into existing IT systems.Radiologist WorkflowWith those requirements in mind, BWH developed Alert and Notification of Critical Results (ANCR), funded by a multicenter grant from The Risk Management Foundation of the Harvard Medical Institutions through CRICO (Cambridge, Massachusetts), a malpractice-insurance company. The system was developed at the BWH Center for Evidence-based Imaging, under the guidance of Ramin Khorasani, MD. Beth Israel Deaconess Medical Center, Boston, and the University of Chicago also collaborated.
To avoid an extra sign-on step, ANCR was integrated into the physician-authentication database and also with the PACS (so that all of the information that passed through the alert-notification system would be integrated, in context, with the PACS information). The email and paging systems also needed to be connected to give physicians the same ability to find each other that they had prior to automation.
When the radiologist generates an alert, all of the stratified notification methods are displayed along, with their descriptions, and the radiologist selects the color. For red and orange alerts that need to be communicated right away, the radiologist next selects the provider to whom the results will be communicated. A brief description of the finding is displayed on the screen with patient information, all in context from PACS. ANCR then sends a notification to the provider via pager. The physician is notified, he or she calls the radiologist or a face-to-face conversation takes place, and the radiologist closes the loop. All this is documented within ANCR.
Yellow alerts, which have a three-day turnaround, work differently. ANCR sets up that notification to be sent by email. Once the notification is submitted via email, it is added to the radiologist’s list of all pending alerts that he or she has generated until the loop is closed. Referrer FriendlyReferring-provider workflow is also integrated with the authentication database and with the email and paging systems, enabling the referrer to review the images, review the radiology report, and order follow-up studies without logging into other systems.
For red and orange alerts, referrers receive a page with the radiology-result notification, the patient information, and a request to call the radiologist, prompting immediate closure of the communications loop.
For yellow alerts, referrers receive an email with a link embedded that goes to the ANCR application. Once logged in, referrers can see the images and the full radiology report and can order follow-up studies. They can also click a button that enables them to see the contact information for that radiologist, in the event that they want to discuss the results. Finally, they click a button that says, “I acknowledge these findings, I understand the results,” and the radiologist is notified that the communications loop is closed.
ANCR also brings the institution significant quality-assurance benefits in that clinical administrators can track pending alerts (as well as alert-notification compliance) by individual physicians. “You can press a button and find out what portion of critical results are red, yellow, or orange; you can see the turnaround time for communication, notification, and acknowledgement; you can understand compliance by single physician; and you can pretty much drill down into this information in a matter of seconds,” Prevedello says. “Suboptimal communication handoffs are important sources of errors and can negatively affect safety and quality, which is why communication of critical results is included in the National Patient Safety Goals,” Prevedello concludes.Cheryl Proval is editorial director of RadInformatics.com and vice president, publishing, of imagingBiz, Tustin, California.
Cheryl Proval,
Vice President, Executive Editor, Radiology Business
Cheryl began her career in journalism when Wite-Out was a relatively new technology. During the past 16 years, she has covered radiology and followed developments in healthcare policy. She holds a BA in History from the University of Delaware and likes nothing better than a good story, well told.