Estimation of age-standardized net survival, even when age-specific data are sparse. (August 2020)

• Record Type: Journal Article
• Authors: Rutherford, Mark J.; Dickman, Paul W.; Coviello, Enzo; Lambert, Paul C.
• Is Part Of: Cancer epidemiology. Volume 67(2020:Aug.)
• Publication Date: 2020-08
• Subjects: Age-standardization -- Net survival -- Period analysis -- Population-based data; Cancer -- Epidemiology -- Periodicals; Cancer -- Prevention -- Periodicals; Cancer -- Diagnosis -- Periodicals; Carcinogenesis -- Periodicals; 616.994005
• DOI: 10.1016/j.canep.2020.101745
• Languages: English
• ISSNs: 1877-7821

Abstract

Highlights: Age standardization is vital for comparability of cancer survival metrics. Traditional approaches to standardization can suffer issues in the face of sparse data. Age standardization using pre-weighting coupled with the Pohar Perme estimator offers a solution in these scenarios.Age standardization with a pre-weighting approach provides consistent estimates for non-sparse data. A slight adaptation is needed for calculating the pre-weights when considering estimates used in period analysis. Abstract: Background: Age-standardization is vital in international comparison studies of cancer patient survival, but standard approaches can fail to produce estimates in the case of sparsity. Methods: The purpose of this paper is to demonstrate that using a standardization pre-weighting approach is a viable alternative approach for external age-standardization in population-based cancer data and performs well in cases of sparsity. We further de;1;scribe how the pre-weighting approach to age-standardization can be coupled with the Pohar Perme estimator in both a cohort and period analysis setting. For period analysis, we compare approaches for defining the internal age distribution. We use SEER public use data to illustrate our approach and estimate survival for Connecticut and by race to create a scenario with sufficient sparsity. Results: The pre-weighting approach gives comparable estimates to traditional age-standardization in cases with sufficient data and produces estimates throughout follow-up in cases of sparsity when a traditional approach would fail. Conclusion: International comparison studies and other national population-based survival studies that need to age-standardize estimates for comparability purposes should adopt the Pohar Perme estimator with pre-weighting. This approach avoids issues of non-estimation in the case of sparsity and will allow more consistent comparisons across the produced estimates.