Investigators from the SWOG Cancer Research Network have developed and validated a risk prediction model for identifying which patients with advanced cancer who are enrolled to clinical trials are at highest risk for unplanned emergency room (ER) visits and hospital stays.
Determining which patients are at significantly higher risk could inform interventions to reduce the need for such visits, improving care quality and reducing costs.
The work will be delivered as an oral presentation by Dawn L. Hershman, MD, MS, at the 2024 ASCO Quality Care Symposium, which will take place September 27 – 28 in San Francisco.
Dr. Hershman is the American Cancer Society Professor of Medicine and Epidemiology at Columbia University Irving Medical Center, deputy director of the Herbert Irving Comprehensive Cancer Center, and group co-chair-elect of SWOG Cancer Research Network, a clinical trials group funded by the National Cancer Institute (NCI), part of the National Institutes of Health (NIH).
“If we can identify patients easily who have the highest risk of acute care use,” Hershman said, “we can target interventions that have been proven to be beneficial, such as active symptom monitoring, and we can better study new strategies to mitigate this risk.”
Hershman’s team linked Medicare claims data to data from six SWOG advanced cancer clinical trials to identify hospital stays or ER visits by enrolled patients. They found that more than two-thirds (67.5 percent) of the 1,397 patients whose data they analyzed had made at least one such visit within one year of their trial enrollment.
The researchers split the patient data into a training data set (60 percent) to use in developing a model and a test data set (40 percent) for subsequently validating that model.
To build the risk prediction model, they considered 23 baseline factors (factors present when patients first enrolled), including sociodemographic, geographic, clinical, and treatment factors. Other factors examined included whether the patient had any of a selected set of additional illnesses or health conditions, known as comorbidities, factors not typically recorded in trial databases but available to the study team through SWOG’s unique linkage to Medicare claims.
From the training data set, the researchers derived a final risk model that incorporated four individual risk factors: the patient’s performance status (a measure of how well the patient is able to perform daily activities) and whether the patient had coronary artery disease, hypertension, or liver disease. The model also adjusted for the patient’s cancer type – specifically, whether or not the patient had prostate cancer.
Among patients in the training set, those with two or more of the four risk factors had more than three times the risk of acute care use compared to patients who had zero or one risk factors.
Researchers used the test data set to validate their model’s performance. Results were similar to those seen with the training data set, confirming the validity of the risk prediction model.
When all patients were considered, those in the highest risk quartile – patients with three or four risk factors – had more than four times the risk (odds ratio = 4.23) of a hospital stay or ER visit compared to patients in the lowest risk quartile (patients with zero risk factors).
Clinical trial eligibility criteria have often excluded patients with certain comorbid conditions. In recent years, there has been a concerted effort to remove some of these criteria, opening trial enrollment to patients with more comorbidities, which may mean more patients at increased risk for acute care use.
“Trials have become more inclusive of patients with some comorbid conditions due to the work of ASCO and other organizations,” said senior author on the work Joseph M. Unger, PhD, associate professor at Fred Hutch Cancer Center and a SWOG biostatistician and health services researcher.
“This could also have the salutary effect of reducing disparities in access to trials for patients of different sociodemographic backgrounds, who may differ in their prevalence of comorbid conditions. However, our work also highlights how investigators and trialists should anticipate a higher risk of acute care use.”
This work was funded by NIH/NCI/NCORP grant UG1CA189974 and The Hope Foundation for Cancer Research.
In addition to Hershman and Unger, the author team includes Cathee Till and Michael L. LeBlanc, both of SWOG Statistics and Data Management Center and Fred Hutch Cancer Center, and Scott D. Ramsey, of Fred Hutch Cancer Center.
SWOG Cancer Research Network is part of the National Cancer Institute's National Clinical Trials Network and the NCI Community Oncology Research Program and is part of the oldest and largest publicly funded cancer research network in the nation. SWOG has 20,000 members in 45 states and eight other countries who design and conduct clinical trials to improve the lives of people with cancer. SWOG trials have directly led to the approval of 14 cancer drugs, changed more than 100 standards of cancer care, and saved more than 3 million years of human life. Learn more at swog.org, and follow us on Twitter/X at @SWOG.
Reference:
“Development and validation of a risk prediction model for acute care use among patients with advanced cancer on clinical trials” (abstract 277)
Oral Abstract Session A, 9/27/2024 1:00 PM-2:30 PM PT
https://meetings.asco.org/meetings/2024-asco-quality-care-symposium/321/program-guide/scheduled-sessions
END
