Methodological issues in the use of administrative claims data to study surveillance after cancer treatment

Administrative claims databases linked to tumor registry data provide an excellent opportunity for studying the follow-up care of cancer patients.

Methodological principles of study design are reviewed, using examples drawn from the recently published literature.

Most follow-up care is outpatient-based. Therefore, in studies using Medicare claims, the patient should be eligible for Medicare Part B and should not be enrolled in a Medicare HMO. In studies of surveillance testing, it may be appropriate to exclude subjects who are near death, in a hospice, or in whom a new or recurrent cancer develops. The definition of the period for measurements of surveillance testing requires consideration, as does periodicity in patterns of testing. Several analytic methods can be employed, from proportions undergoing testing to survival analysis methods. Measurement of 'surveillance tests' among control subjects (those without cancer) may be useful for comparison with cancer patients. To date, administrative claims data have been most useful in studies investigating the quality of, and disparities in, cancer care.

With appropriate attention to methodological issues, linked tumor registry and administrative databases can provide important insights into the quality of survivorship care for cancer.