Secondary data analysis offers ‘high-impact’ approach to improving patient care

MIAMI — Secondary data analysis enables researchers to provide high-impact data; however, implementation of this approach can be complex, Maxine Sun, PhD, MPH, said at the International Kidney Cancer Symposium.

“Essentially it is using data that has been collected, but not for you. Conceptually, it means that someone else’s secondary data becomes your primary data,” Sun, a research scientist at Dana-Farber Cancer Institute, Brigham and Women’s Hospital, and Harvard Medical School, said during the presentation.

Secondary data analysis offers a variety of benefits for researchers compared with collecting primary data including good sample sizes; providing a variety of information, including on costs and value in care; reduced logistical and financial challenges; and accurate depictions of “real-world” problems.

However, a variety of factors should also be taken into consideration prior to initiation of secondary data analysis.

For example, researchers should understand why and how the data were collected, Sun said.

Types of secondary data that can be analyzed include:

administrative data, primarily hospital discharge data reported to a government agency;
claims-based data, designed to describe the billable interactions between insured patients and the health care delivery system;
electronic health records, which are obtained at the point of care at a medical facility, hospital, clinic or practice;
health surveys, designed to provide an accurate evaluation of the population health;
patient or disease registries, defined as clinical information systems that track a narrow range of key data for certain chronic conditions; and
clinical trial data, comprised of free digital library-laboratory information that provides one location for the research community to broadly share.

Once a database is chosen, Sun noted the importance of using a statistical analysis plan.

“By having a statistical analysis plan a priority, it will minimize the amount of errors and surprises you will encounter; it will solidify the study design; and most important, it will diminish the temptation to do data mining,” Sun said.

A statistical analysis plan should include primary and secondary aims, hypotheses, data source, inclusion/exclusion criteria, primary variable of interest, covariates, handling of missing data, statistical methods and power analysis.

“Using secondary data analysis to answer a clinical and healthy policy question is very complex, and can be very tricky,” Sun said. “But, with the appropriate and rigorous approaches, there is really an opportunity to provide high-impact research to improve the care of patients with kidney cancer.” by Kristie L. Kahl

Reference:

Sun M. Utility, pitfalls and perils of using public databases for RCC research. Presented at: International Kidney Cancer Symposium; Nov. 3-4, 2017; Miami.

Disclosure: Sun reports no relevant financial disclosures.