AUTHOR=Alsadhan Norah , Almaiman Alaa , Pujades-Rodriguez Mar , Brennan Cathy , Shuweihdi Farag , Alhurishi Sultana A. , West Robert M. TITLE=Statistical methods for measuring trends in colorectal cancer incidence in registries: A systematic review JOURNAL=Frontiers in Oncology VOLUME=12 YEAR=2022 URL=https://www.frontiersin.org/journals/oncology/articles/10.3389/fonc.2022.1049486 DOI=10.3389/fonc.2022.1049486 ISSN=2234-943X ABSTRACT=Background

Monitoring cancer trends in a population is essential for tracking the disease’s burden, allocating resources, and informing public health policies. This review describes variations in commonly employed methods to estimate colorectal cancer (CRC) incidence trends.

Methods

We performed a systematic literature search in four databases to identify population-based studies reporting CRC incidence trends, published between January 2010 and May 2020. We extracted and described data on methods to estimate trends and assess model validity, and the software used.

Results

This review included 145 articles based on studies conducted in five continents. The majority (93%) presented visual summaries of trends combined with absolute, relative, or annual change estimates. Fourteen (10%) articles exclusively calculated the relative change in incidence over a given time interval, presented as the percentage of change in rates. Joinpoint regression analysis was the most commonly used method for assessing incidence trends (n= 65, 45%), providing estimates of the annual percentage change (APC) in rates. Nineteen (13%) studies performed Poisson regression and 18 (12%) linear regression analysis. Age-period-cohort modeling- a type of generalized linear models- was conducted in 18 (12%) studies. Thirty-nine (37%) of the studies modeling incidence trends (n=104, 72%) indicated the method used to evaluate model fitness. The joinpoint program (52%) was the statistical software most commonly used.

Conclusion

This review identified variation in the calculation of CRC incidence trends and inadequate reporting of model fit statistics. Our findings highlight the need for increasing clarity and transparency in reporting methods to facilitate interpretation, reproduction, and comparison with findings from previous studies.