University of Louisville: Building an evidence-based algorithm to guide autoimmune neurology patient care
UofL Health – UofL Hospital physicians and laboratory scientists faced challenges in the ordering of autoimmune and paraneoplastic panels: duplicate orders that add unnecessary cost and omitted test orders that could provide the key to the right diagnosis for patients.
The Department of Pathology and Laboratory Medicine at UofL Hospital had an established policy to review any order for a test that cost over a certain amount. Autoimmune and paraneoplastic panels, which frequently fall into the higher-cost category, were often caught by this policy. This meant the Neurology Department was fielding frequent phone calls to ensure that these critical tests being ordered were truly needed.
To help streamline this process, Anita Fletcher, M.D., who was then a resident physician specializing in neurology at the University of Louisville School of Medicine (now a clinical fellow in the Division of Neuroimmunology and Neurovirology, NINDS, NIH), initiated a project to develop an algorithm that standardized ordering practices for autoimmune and paraneoplastic panels. The project was advised by UofL Health – UofL Hospital – Multiple Sclerosis Clinic neurologists David Robertson, M.D., Neuroimmunology Director at University of Louisville, and Michael Sweeney, M.D., an autoimmune neurologist at University of Louisville.
Anita Fletcher, M.D.
David Robertson, M.D.
Michael Sweeney, M.D.
It was especially critical to ensure that the right test was being ordered for this type of testing, as an accurate diagnosis is essential to guide patient care. Based on recent studies, a considerable portion of patients with encephalopathy and/or epilepsy of unknown etiology may have an autoimmune or paraneoplastic cause. For patients whose condition is due to an underlying malignancy, an accurate diagnosis will allow them to get the right care team and plan. When the cause is an autoimmune disorder, there are equally important decisions to be made for chronic care and treatment.
By the numbers
sensitivity and 71% specificity
in cost savings
true positive rate increase
Initially, Dr. Fletcher worked with Dr. Cierra Sharp, then a clinical chemistry fellow from the Department of Pathology and Laboratory Medicine, to analyze all orders for autoimmune and paraneoplastic panels in 2018. They gathered data on the type of panel used, body fluid collected, and type of antibodies recorded. Then, Dr. Fletcher used diagnostic criteria from published articles to develop an algorithm to help physicians decide when to order a specific panel. Her sources included the 2016 autoimmune encephalitis criteria1, the Antibody Prevalence in Epilepsy and Encephalopathy (APE2) score2, and clinical description and autoantibody correlation used primarily for movement disorder characterization3.
The APE2 score, developed by Mayo Clinic’s Divyanshu Dubey, M.B.B.S., predicts how likely a patient with epilepsy and encephalopathy is to be positive for certain antibodies that indicate an autoimmune cause. The score a patient receives is based on multiple factors—taken altogether, these factors indicate, with a high level of certainty, whether or not patient’s epilepsy or encephalopathy might have an autoimmune cause.
After Dr. Fletcher developed the algorithm using the APE2 score and other criteria, she contacted Dr. Dubey, who reviewed the algorithm and provided suggestions. “Because not every physician is aware of these disorders, it’s good to have scoring systems and algorithms like this,” said Dr. Dubey. “This allows testing to be applied in a more consistent manner.”
After establishing the criteria to determine when an autoimmune or paraneoplastic panel should be ordered, Dr. Fletcher wanted to create consistency in the last step of the process—which reference lab to order the panels from. This was critical, because in order to accurately analyze predictive values, there had to be panel-to-panel consistency.
Dr. Fletcher’s preference was to send to a lab with a two-tier model, which meant that after a patient sample was tested, the remaining sample would be used for other training and research purposes. This led Dr. Fletcher to Mayo Clinic Laboratories. At Mayo, the Neuroimmunology Laboratory, Autoimmune Neurology Clinic, and research labs are fully integrated, which means that samples from the lab are also used in staff competency training programs, as well as the test development process, supporting the discovery of novel antibodies. This approach applies to relationships with other medical institutions as well. “Testing services are just one part of what we do. Education and research are equally important,” said Dr. Dubey. “Collaboration helps transform and transition the practice of neuroimmunology, when this sort of research ends up becoming clinical practice.”
“This is a remarkable study which evolved from the need for enhanced collaboration between our hospital systems and the bedside delivery of health care. The results combine the rapidly evolving field of neuroimmunology with real world, everyday application.”David Robertson, Neuroimmunology Director at University of Louisville and UofL Health – UofL Hospital – Multiple Sclerosis Clinic neurologist
In addition, the ability to quickly be put in contact with an expert at Mayo when questions arose was a key factor in Dr. Fletcher’s decision. With experts in Mayo Clinic’s Neuroimmunology Laboratory available for questions 24/7, “it was clear if I needed to get ahold of someone I could,” said Dr. Fletcher. “I have been so incredibly pleased outside of this project, when I had a difficult case and just wanted to talk it through. It is a service that you can’t put money on.”
When the new algorithm was applied to the 2018 data, the APE2 portion of the algorithm showed 100% sensitivity and 71% specificity for neural specific autoantibodies. The results of the project were presented at 2020 American Academy of Neurology meeting and were published in Neurology supplement.
The project data and algorithm was presented to the Diagnostic Stewardship Committee at University of Louisville Hospital and implemented as policy. “We have since been able to streamline our diagnostic workup in these difficult cases and reduce overall costs without sacrificing testing sensitivity. As new antibody tests become available and new testing techniques are developed and validated, this process requires ongoing evaluation,” said Dr. Sweeney.
An assessment was also conducted in 2019, six months after implementing the new algorithm. It showed a true positive rate increase of nearly 33%, a reduction in the number of panels ordered, and a cost savings of $54,900 based on pre-algorithm cost projection.
Dr. Fletcher notes: “An algorithm for physicians to use is an evidence-based guide and should come with references that are strong and peer reviewed. But there is never a case where an algorithm completely takes away a physician’s judgment.”
“The most important thing we do, especially in a rapidly changing field like autoimmune neurology, is to improve patient care from diagnosis to treatment, to ultimately improve quality of life. This project was a step toward that goal, and it became hospital policy. I was honored to be part of it.”Anita Fletcher, M.D.
Fletcher A, Sharp C, Sweeney M. Integration of an Evidence-Based Algorithm for Inpatient Autoimmune and Paraneoplastic Neurologic Syndrome Autoantibody Panel Ordering Practice to Improve Diagnostic Evaluation Quality. Presentation presented at: American Academy of Neurology Annual Meeting 2020. Accessed January 15, 2021. https://cslide-us.ctimeetingtech.com/aan2020/attendee/eposter/poster/3603?r=pt%7E27
Graus F, Titulaer M, Balu R, et al. A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurol. 2016 Apr;15(4):391-404.
Dubey D, Pittock S, McKeon A. Antibody Prevalence in Epilepsy and Encephalopathy score: Increased specificity and applicability. Epilepsia. 2019 Feb;60(2):367-369.
Damato V, Balint B, Kienzler A, Irani S. Mov Disord. The clinical features, underlying immunology, and treatment of autoantibody-mediated movement disorders 2018 Sep;33(9):1376-1389.