A guide to large data sets for population-based cancer research: Strengths, limitations, and pitfalls

With the proliferation of cancer research based on large databases, misalignment of research questions and data set capabilities is inevitable. Nationally maintained databases are appealing to cancer researchers because of the ease of access to large amounts of patient data available for analysis and risk estimation. Data sets that are commonly used in cancer research include the National Cancer Database, the SEER (Surveillance, Epidemiology, and End Results) program of the National Cancer Institute, the SEER-Medicare database, the American College of Surgeons National Surgical Quality Improvement Program, and the Healthcare Cost and Utilization Project databases, among others. Each data set has pros and cons with respect to variable availability and the ability to analyze cancer-specific outcomes. It is critical for researchers to understand the strengths and limitations of each database. Changing variable definitions, the length of postoperative data collection, and the availability of patient-reported outcomes or social determinants of health data are examples of factors that researchers must consider when selecting a data set for research purposes. For the current review, the authors summarized the advantages and disadvantages of various national data sets for cohort studies in cancer populations.