Comparative Effectiveness Research in Cancer

“The dearth of outcomes research on terminal brain cancer is pronounced…Their collaboration, novel in the neuro-oncology sphere, unites clinical expertise and comparative effectiveness research, conducted with innovative statistical methodology, to improve quality of life for elderly patients with glioblastoma.” Excerpt from article about Dr. Dominici’s research Big Data Benefits for Terminal Cancer Patients by Orna Feldman

Treatment strategies in cancer research are becoming increasingly complex and the need to compare these strategies on patient outcomes (e.g. survival and re-hospitalization) could not be greater. It is often not feasible to compare these complex treatments in randomized clinical trials (RCT) because the subject population isn’t necessarily an adequate representation of the afflicted population. For instance, although a clinical trial may prove that chemotherapy will extend the life of a 42-year-old with relatively good health by three months, this information does not help inform the treatment approach for a 75-year-old man with seizures and confusion. Identifying these gaps and developing strategies to address them will strengthen our ability to care for the vast population of those suffering from cancer.

Recently, large administrative databases, such as Medicare part A, SEER-Medicare have been used for Comparative Effectiveness Research (CER.) The use of these data in CER is attractive: they include large populations (e.g. Part A Medicare includes 95% of the entire population>65) and they allow us to compare complex treatment strategies. However, these large databases have many challenges that must be overcome to make reliable inference. First, these are observational data and therefore methods for confounding adjustment must be developed. Second, there is often a misclassification in the treatment assignment, that is, the incorrect diagnostic or procedural code is used. Third, in many cases more than one administrative database is available to make the comparison of interest, but these databases include different sets of measured potential confounders (e.g. SEER-Medicare, and Part A Medicare), yet undoubtedly careful integration of these databases will improve estimation. In summary, the imperative to develop new methods to make rigorous inference from the analysis of these large administrative databases could not be greater.

In close collaboration with researchers and clinicians at the Dana Farber Cancer Institute and at the Harvard Medical School we develop Bayesian methods in causal inference for Comparative Effectiveness Research (CER) that overcome the challenges described above. We also apply methods to administrative data to address key CER questions in the treatment of Glioblastoma and in other cancers.

Administrative data provide a wonderful opportunity to conduct Comparative Effectiveness Research (CER), but without rigorous methods, results can be untrustworthy. Our newly developed methods, which will be routinely tested by using RCT data, will overcome many of the most important challenges for CER analyses of administrative data. Their dissemination of methods to clinical investigators and their application to administrative will substantially elevate the quality of CER comparisons across many domains of cancer research.

Dr. Dominici is a member of the DF/HCC: