Cancer incidence estimation from mortality data: a validation study within a population-based cancer registry

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Population-based cancer registries are crucial for calculating cancer incidence.
• Methods exist to estimate cancer incidence in areas lacking registries, but their validity requires further analysis.

Purpose of the Study:

• To assess the validity of the incidence-to-mortality ratio (IMR) method for estimating cancer incidence.
• To compare estimated cancer cases with observed cases and identify the best estimation scenario.

Main Methods:

• Utilized 15-year cancer mortality data to estimate yearly cancer cases (2004-2013) for six cancer sites.
• Employed generalized linear mixed models within a Bayesian framework and Markov chain Monte Carlo methods.
• Applied the IMR method under five different assumptions and compared estimations to observed incidence data.

Main Results:

• Relative differences between observed and predicted cancer cases were under 10% for most sites.
• A constant IMR trend assumption yielded the best goodness-of-fit for several cancers in men and women.
• Linear IMR trend assumption was optimal for lung, ovarian, and prostate cancers; overall mean absolute percentage error was 6% (men) and 4% (women).

Conclusions:

• The IMR method is a valid tool for estimating cancer incidence in areas without registries.
• The proposed goodness-of-fit indicator aids in statistically selecting the optimal IMR assumption for accurate cancer incidence estimation.