Guide to Surgical Oncologic Research: A Comparison of Four National Databases
Robyn N. Rubenstein, MD, Jonas A. Nelson, MD, MPH, Kathryn Haglich, BS, MS, Shen Yin, PhD, Carrie S. Stern, MD, Evan Matros, MD, MMSc., MPH
Memorial Sloan Kettering Cancer Center, New York, New York
Background: Various national databases assist physicians in understanding demographic and clinicopathologic patterns and surgical practices by providing large samples and numerous variables. with the benefit of large sample sizes. These databases differ in focus of pathology, specificity, size, types of populations, etc. The aim of this study is to compare four commonly used, large national datasets from the surgical and oncologic perspective.
Methods: NSQIP, NIS, SEER, and NCDB were evaluated from the lens of a surgical oncologist, comparing database characteristics. An examination was performed to assess the accessibility of the database, required software, size of the cohorts, distribution of patient populations, available demographics, oncologic and pathologic information offered, detail of surgical procedures, diagnoses, outcomes, complications, and limitations to the datasets.
Results: The time periods covered by databases are as follows: NSQIP 2005-2019, NIS 1988-2018, SEER 1975-2018, and NCDB 2004-2017. In 2017, NSQIP reported 1,028,713 cases from 708 hospitals and NIS reported 7,159,694 inpatients from 4,584 hospitals. SEER represents 35% of the U.S. population with 17 location-based registries which offers a national population-based sample; NCDB overall has more than 34 million records from over 1500 Commission-accredited cancer programs. The race distributions for NSQIP, NIS, and SEER differ - Caucasian percentages, for example, are 70.3%, 65.1%, 31.9%, respectively. NCDB is available in site-specific cohorts only so race distribution is not indicative of the entire cohort. NSQIP offers the most comprehensive collection of pre-operative comorbidities, post-operative complications and outcomes, and detail regarding both inpatient and outpatient surgical procedures. NIS offers detailed hospital/facility information and tracks inpatient procedures only, which is limited by the recent change from ICD9 to ICD10 procedure codes. Both SEER and NCDB are restricted by a single procedure code, limiting the operative detail that the other two offer. As cancer-focused databases, SEER and NCDB offer the most oncologic, pathologic, diagnostic, and treatment-related detail.
Conclusion: Understanding the qualities, benefits, and limitations to various national databases improves researcher experience and productivity. Recognizing the characteristics that each database offers allows researchers to efficiently use these national databases, leading to more effective investigative outcomes.
Back to 2021 Abstracts