Response

We thank Molenaar et al. for their extension of our analysis and Murphy et al. for their comments on our recent article in the Journal, which reported long-term increases in colorectal cancer (CRC) incidence in age groups younger than age 55 years coinciding with rapid declines in older age groups. We agree with Murphy et al. on the importance of conveying absolute risk in communicating information to physicians and the public and acknowledged the substantially lower incidence of CRC in younger age groups in the text of the article and in Figures 1 and 2. Despite relatively low risk, however, approximately 27 000 cases of CRC were diagnosed in patients age 20 to 54 years in 2013, more than double the total number of cervical cancers diagnosed (1). Moreover, the focus of our study was not differences in the risk of CRC by age, which are well documented, but differences in how these risks are changing. Temporal trends are conventionally described in relative terms when comparing different populations because of differences in scale. Relative differences, like the annual percent change we report, are comparable across age groups because they account for differences in the absolute risk.

While a CRC diagnosis before age 50 years remains uncommon, the sustained increase in incidence, especially in the youngest age groups, deserves attention for several reasons. First, these cohorts will carry elevated risk with them as they age, forecasting an increase in the disease burden decades into the future. Notably, new data from the National Cancer Institute indicate that after two decades of rapid declines, the most recent joinpoint trend for CRC in all ages combined is stable (2). Second, trends in young age groups are most indicative of progress against cancer because they can only reflect relatively recent exposures (3). The unknown cause of this trend makes it that much more concerning. Third, largely due to lack of awareness, 28% of CRC patients younger than age 50 years were diagnosed with metastatic disease in 2013, compared with 20% of patients age 50 years or older (4). As Molenaar et al. in their correspondence and others have reported, the most rapid increase in CRC incidence rates in young age groups is for distant-stage tumors (5). This, as well as the consistency of Molenaar et al.’s results with our own, despite different data sources and risk measurements, suggest a true increase in disease occurrence among young adults. Finally, with the age-adjusted percentage of CRC patients younger than age 55 years almost doubling in the past two decades, the burden of this disease is shifting to those with young families who are in their most productive years of life. We do not recommend screening everyone at average risk before age 50 years, but we do believe that this new evidence, along with recent findings from the Cancer Intervention and Surveillance Modeling Network that beginning screening for those at average risk at age 45 years instead of age 50 years provides a more favorable balance of life-years gained and screening burden (6), should be considered when the age to initiate screening is reevaluated.