3 min read
Schipper reports personal fees from Innovative Analytics. Rini reports research grants and/or personal fees from Alkermes, Aravive, AstraZeneca, Aveo Pharmaceuticals, Bristol Myers Squibb, Genentech, GlaxoSmithKline, Merck and Pfizer outside the submitted work. Garrett-Mayer reports no relevant financial disclosures.
A new web-based application can help estimate the impact of treatment delays for individual patients with cancer and optimize allocation of resources during the COVID-19 pandemic, according to results of a study published in JAMA Oncology.
The application, called OncCOVID, provides survival estimates for patients who receive immediate vs. delayed cancer treatment. The model incorporates several individual risk factors for COVID-19 and its potential complications.
The COVID-19 pandemic has led to delays in cancer treatment for patients throughout the country, particularly in areas where cases of the virus have spiked.
The decision analytical modeling study, which included data of more than 5 million patients with cancer, showed the OncCOVID model revealed associations between the impact of treatment delays and individual patient factors not currently captured by common triage systems, including age, comorbidities, treatment received and local community estimates of COVID-19 risk.
“The OncCOVID model was developed to provide an opportunity for cancer care professionals to triage patient care based on individual risk measures that indicate the potential risks (or benefits) of delaying treatment depending on the individual patient’s risk [for] developing COVID-19 illness,” Elizabeth Garrett-Mayer, PhD, division director of biostatistics and research data governance in Center for Research and Analytics at ASCO, and Brian I. Rini, MD, professor of medicine at Vanderbilt University and HemOnc Today Editorial Board Member, wrote in an editorial accompanying the study.
Garrett-Mayer and Rini commended the researchers for developing a more personalized approach to decision-making, but added that the OncCOVID model’s estimates should be interpreted with some caution and its risk parameters updated as new data emerge.
HemOnc Today spoke with Matthew J. Schipper, PhD, research professor in biostatistics and research associate professor in radiation oncology at University of Michigan, about the development of the app and how it could be used in clinical practice.
Question: How did you come up with the idea for this app?
Answer: This came about from a prostate cancer research group that we have. We do a lot of predictive modeling-type projects, which we have been implementing via web-based apps. During one of our group meetings, the topic of COVID-19 came up and its differential impact on patients with cancer and in different areas of the country, so we started discussing how to put this together.
Q: How can this application be used to help health care centers and patients?
A: There are multiple potential uses. If a hospital or cancer center has limited resources and there is more demand than availability, you can use this app to prioritize patients who need the most help. This is similar to the three-tier system that many hospitals used, but it is more individualized because it takes into account the patient’s age and their comorbidities. It also can be used by individual clinicians or patients with cancer. If you are a patient worried about optimal timing of treatment, you could see your predicted risk for getting or dying of COVID-19 if you delay treatment for 2 months. You could also see how a delay in chemotherapy treatment could impact your risk if you become infected with COVID-19. If you are a clinician, you could make personalized decisions for patients with cancer and decide what is best for them in terms of the optimal timing of their treatment.
Q: How has it come together, and how is it working so far?
A: This was the biggest team project in which I have ever participated. Many statisticians and clinicians worked on all different aspects of it. One of my students, Holly E. Hartman, MS, did an incredible amount of work and built the web application and did the SIR (susceptible-infected-recovered) models predicting risk for COVID-19 infection over time. It was then linked with Johns Hopkins University’s COVID tracker, so the user could enter the county and see the real-time data. Daniel Spratt, MD, and Robert Dess, MD, spent a lot of time providing input and reviewing the models and app. A group from Pennsylvania State University conducted an analysis using SEER and National Cancer Database data that estimated the impact of treatment delays on COVID-19 mortality. Analyses of case mortality rates based on age and the number of comorbidities also were included. Overall, the statistical analysis was extensive.
For more information:
Matthew J. Schipper, PhD, can be reached at email@example.com.