Determination of Lung Cancer Incidence in the Elderly using Medicare Claims Data

Abstract

The Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute provides data for making national estimates of lung cancer incidence and for monitoring secular trends. The authors compared the number of cases of lung cancer and the incidence rates among elderly residents of the five states included in the SEER program in 1986-1987 with the number of incident cases identified and the rates calculated using hospitalization and enrollment data on elderly Medicare beneficiaries maintained by the Health Care Financing Administration (HCFA) for the same years. The SEER program state registries identified 5.9% more cases than did HCFA (p < 0.01). However, the overall rates were similar (274.2/100,000 population for SEER and 264.7/100,000 population for HCFA), as were the majority of the rates for the different demographic subgroups examined. Age-adjusted lung cancer incidence rates for 1986 through 1990 among elderly Medicare beneficiaries residing outside of all nine SEER areas were 8-13 percent higher than the rates calculated for SEER-area residents. This observation is supported by the existence of similar differences in the age-adjusted lung cancer mortality rates for 1979 through 1988 in the same populations. Because the SEER areas may not be representative of the entire nation for lung cancer incidence and HCFA data cover the entire country, the authors recommend using HCFA information to complement the SEER data system.

Administrative Data for Public Health Surveillance and Planning

“…Despite the observation that exact values may not match between administrative data and data from other sources, relative differences tend to be highly consistent (e.g. 18,48,67).…”

Use of Medicare claims data to estimate national trends in stroke incidence, 1985-1991.

“…Our approach resembles that used by Whittle and coworkers 7 and McBean and coworkers 8 for evaluating the accuracy of Medicare hospitalization data for estimating cancer incidence. In addition, we used other data sources for evaluating the validity of the derived incidence trends.…”

Comorbid illnesses and health care utilization among Medicare beneficiaries with lung cancer

“…We included patients if there was one MEDPAR code for cancer of the lung or bronchus or two carrier codes for cancer of the lung or bronchus more than 30 days apart. Several previous studies have shown that lung cancer [16,17] and incident cancer [18] can be identified reliably using Medicare data with specific criteria. We excluded non-US residents and beneficiaries for whom there was any termination of coverage other than for hospice or death.…”