Cancer occurrence in the elderly: agreement between three major data sources

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Accurate cancer data is crucial for epidemiological research and public health surveillance.
• Multiple data sources, including hospitalization records, cancer registries, and death certificates, are used to ascertain cancer cases.
• Understanding the agreement between these sources is essential for data validation and interpretation.

Purpose of the Study:

• To evaluate the agreement on cancer occurrence and cancer site between Medicare Part A hospitalization data, the Massachusetts Cancer Registry, and death certificates.
• To assess the reliability of these data sources for cancer surveillance.

Main Methods:

• A cohort of 905 subjects from the East Boston Senior Health Project, who died between 1986 and 1990, was utilized.
• Data linkage was performed between Medicare Part A, the Massachusetts Cancer Registry, and death certificates for the study cohort.
• Agreement was quantified using kappa statistics.

Main Results:

• Substantial agreement was observed between hospitalization data and death certificates (kappa = 0.70).
• Moderate agreement was found between hospitalization data and cancer registry data (kappa = 0.59), and between cancer registry and death certificate data (kappa = 0.50).
• Agreement levels remained consistent across various demographic and clinical subgroups and were higher for colorectal and respiratory cancers compared to breast and prostate cancers.

Conclusions:

• Medicare hospitalization data, cancer registries, and death certificates demonstrate reasonable agreement for cancer occurrence ascertainment.
• These findings provide valuable insights into the strengths and limitations of each data source for cancer epidemiology.
• Epidemiologists can leverage this information to optimize study design and data interpretation in cancer research.