ASTROnews: Comparative effectiveness research for radiation oncology

A new report from the Institute of Medicine (IOM) sheds a harsh light on the National Cancer Institute’s clinical cancer trials process—a 40 percent incomplete rate and a process that touches barely a tiny fraction of the 1.5 million new cancer patients and the 11 million patients living with cancer. The IOM rightly criticizes the burdensome bureaucracy that stretches the development of a concept to the start of the trial over two and a half years. Yet, even completely overhauling the clinical trial process, without adding comparative effectiveness research (CER), we simply don’t know how cancer care actually should be done.

Even though our own Radiation Therapy Oncology Group (RTOG) has a very high level of successful completion, CER must also be given a mature role in growing a strategic approach to the evidence basis of our care. The power of modern computing and the Internet give us access to practice-based evidence/data and advanced analytic tools that can yield answers that will never come from randomized controlled trials. Organizing and funding this research, as well as focusing our academic and community practice-based leaders on the correct research questions, will take much coordination. I have learned much about this through my involvement with Quality Research in Radiation Oncology (QRRO), the successor to the Patterns of Care Study. QRRO seeks to develop meaningful electronic data retrieval and analysis for a new wave of CER for meaningful radiation oncology endpoints, all linked into the larger NCI Bioinformatics Grid.

Whereas there are not enough column inches to give a full review of CER here, the potential is limitless. As studies on acute stroke victims have demonstrated, amassing thousands of data elements per subject gives a strong ability not only to see comparative primary outcomes but also to do ever greater analyses of potential confounding factors. It is growing ever easier to safely and remotely access, deidentify and aggregate data from hospital and radiation oncology electronic medical records. Advanced image and RT data acquisition and analysis are already being perfected and will contribute as we increase our image guided care. CER’s work is done entirely on practice-based evidence and can perform prospective cohort studies. We can build a novel national multidisciplinary network and perform comparative effectiveness analyses of different approaches in the treatment of all cancers with an emphasis on the effectiveness of radiation therapy.        

By using CER’s advanced remote data extraction, mining and analytic tools to retrieve data with common and consistent definitions, CER can address important clinical questions that cannot be answered through randomized clinical trials. These studies can test hypotheses of comparative effectiveness in diseases where varied treatment approaches appear to have similar outcomes, where compliance with guideline recommendations may affect outcomes and where advanced technologies are rapidly entering national practice without adequate testing by traditional scientific methods.

Both randomized trials and the 35-year Patterns of Care Study/QRRO project have informed our modern practices. However, the importance of CER for radiation oncology should not be understated. The strength of hypothesis- driven and statistically validated data of the randomized trial can now meet the practice based evidence of PCS/QRRO and modern CER’s speed, flexibility,
advanced data mining, analysis and sharing. I predict these elements will form the basis of a more favorable future IOM review of the NCI research process.

Potential benefits of cer in emr era

• The ability to collect data from much larger and more representative groups of patients, including those who might not normally participate in randomized trials.
• The ability to simultaneously gather and manage thousands of potential data points on each participant, with the ability to analyze these data in multiple ways.
• The ability to generate new hypotheses based on scientific questions that arise even after data have been gathered and to use these hypotheses to mine collected data.
• The ability to initiate and sustain real-time or near real-time collection of data and analyses, including the integration of new technologies and medical treatments into study databases.
• The ability to synthesize and publish both raw data and the results of analyses much more rapidly than is possible in current studies.
• The ability to discard original hypotheses or approaches when proven faulty or inefficient and to discover and correct such challenges much more rapidly.
• The ability to leverage economies of scale in the generation of standardized consent forms, standardized statistical analysis approaches, rapid generation of best practice documents and optimal results guidelines, and generation and agreement on standard terminologies and interoperability protocols.

Dr. Devlin practices at Brigham and Women’s Hospital in Boston. He welcomes comments on his editorials, at communications@astro.org.