Trends in the Incidence of Young-Onset Colorectal Cancer With a Focus on Years Approaching Screening Age: A Population-Based Longitudinal Study

Abstract

Background

With recent evidence for the increasing risk of young-onset colorectal cancer (yCRC), we had the objective to evaluate the incidence of yCRC in 1-year age increments, particularly focusing around the screening age of 50 years.

Methods

We conducted a longitudinal study using linked administrative health databases in British Columbia, Canada, including a provincial cancer registry, inpatient and outpatient visits, and vital statistics from January 1, 1986, to December 31, 2016. We calculated incidence rates per 100 000 at every age from 20 to 60 years and estimated annual percent change in incidence (APCi) of yCRC using joinpoint regression analysis.

Results

We identified 3614 individuals with yCRC (49.9% women). The incidence of CRC steadily increased from 20 to 60 years, with a marked increase from 49 to 50 years (incidence rate ratio = 1.19, 95% confidence interval [CI] = 1.04 to 1.34). Furthermore, there was a trend of increased incidence of yCRC among women (APCi = 0.79%, 95% CI = 0.22% to 1.36%) and men (APCi = 2.17%, 95% CI = 1.59% to 2.76%). Analyses stratified by age yielded APCis of 2.49% (95% CI = 1.36% to 3.63%) and 0.12% (95% CI = −0.54% to 0.79%) for women aged 30-39 years and 40-49 years, respectively, and 2.97% (95% CI = 1.65% to 4.31%) and 1.86% (95% CI = 1.19% to 2.53%) for men.

Conclusions

Our findings indicate a steady increase over 1-year age increments in the risk of yCRC during the years approaching and beyond screening age. These findings highlight the need to raise awareness as well as continue discussions regarding considerations of lowering the screening age.

Colorectal cancer (CRC) is the third most commonly diagnosed cancer among both men and women in Canada and is traditionally considered a disease of older adults, with marked onset after the age of 50 years (1,2). However, research over the past decade indicated an increase in young-onset colorectal cancer (yCRC), that is, CRC diagnosed in individuals younger than 50 years of age. A systematic review and meta-analysis published in 2020 that included 39 articles on trends of yCRC incidence reported a pooled annual percent change in incidence (APCi) for yCRC of 1.33% (95% confidence interval [CI] = 0.97% to 1.68%) internationally and 1.59% (95% CI = 1.24% to 1.93%) when pooling was restricted to studies conducted in North America (3). The rising trend of yCRC incidence may be attributed to an increase in rectal cancer. A US population-based study by Bailey et al. (4) showed a statistically significantly increased APCi for yCRC over the period of 1975-2010, with the most drastic changes observed for cases of rectosigmoid and rectal cancer in the 20- to 34-year age group, with APCis of 2.66%, 3.05%, and 4.03% for distant, regional, and localized cancers, respectively.

Canadian guidelines introduced in 2001 recommend that screening for CRC begins for individuals 50 years or older (5). However, recent evidence on the increasing risk of yCRC has called for discussions to assess whether the guideline-recommended initial screening age for CRC should be lowered (6). In 2018, the American Cancer Society made a qualified recommendation of lowering the recommended age for average-risk adults to initiate screening from 50 to 45 years (7). Recently in 2020, Abualkhair et al. (8) used the US Surveillance, Epidemiology, and End Results registries from 2000 to 2015 to analyze CRC incidence over 1-year age increments. The study authors observed the most pronounced incidence increase (46.1%) in the 49- to 50-year age transition with an incidence rate ratio (IRR) of 1.46 (95% CI = 1.42 to 1.51) (8). These findings suggest that an excess of yCRC cases are going undetected and becoming apparent only when screening is implemented at 50 years of age (8). To evaluate whether such trends are also happening in Canada, a nation where average-risk screening begins at 50 years of age but has universal medical and hospital care, we used a population-level cancer registry from British Columbia to assess the incidence of CRC in 1-year age increments.

Methods

Data Source and Study Population

We established a population-based CRC cohort for epidemiologic and health services research by linking administrative databases held in British Columbia (BC), Canada (9-15). Specifically, Population Data BC contains longitudinal and deidentified individual-level health services data for the population of BC (approximately 4.86 million in 2016) (16) since April 1985, including information on outpatient visits from the Medical Service Plan database (12), inpatient visits from the Discharge Abstract Database (15), and vital statistics (11). PharmaNet captures information on all community-dispensed prescriptions from 1996 onward (10). Finally, the BC Cancer Registry captures all new cancers diagnosed in BC residents since 1985, including information on diagnosis (eg, date, tumor group, sites, stage) and treatment (eg, dates, modality) (9). Data are linked at the individual level across databases using provincial health numbers, which are replaced by data stewards with random depersonalized identifiers to maintain patient anonymity.

We established a source population from the period of January 1, 1986, to December 31, 2018, which included individuals with CRC identified in the BC Cancer Registry using the following International Classification of Diseases for Oncology, 3rd Edition codes: C18.2-C18.9 (colon); C19.9 (rectosigmoid); and C20, C21.8 (rectum). The source population consisted of 54 971 individuals diagnosed with CRC. We classified age of CRC diagnosis using the age of diagnosis variable from the BC Cancer Registry, and cases of yCRC were defined as individuals who received their diagnosis at younger than 50 years of age (see Supplementary Figure 1 [available online] for data sources and source population). For individuals included in the source population, the end of follow-up corresponded with the last recorded health care use visit in Medical Service Plan, Discharge Abstract Database, or PharmaNet databases.

Statistical Analysis

For our primary analyses, we calculated the incidence rate (IR) of CRC per 100 000 population from January 1, 1986, to December 31, 2016, for each 1-year age interval from 20 to 60 years using the 2016 population distribution of BC as the standard to obtain age-adjusted estimates. We truncated the study period to account for 2-year lag in data reporting and submitting cycles. Descriptive statistics (eg, age, sex) were determined for all incident CRC cases diagnosed between the ages of 20 and 60 years. We concentrated on incident cases of CRC among individuals receiving a diagnosis between 20 and 60 years of age because this permitted us to evaluate cumulative incidence trends encompassing the transition from 49 to 50 years of age, that is, the transition into the age for average-risk screening for Canadians. In addition, we calculated the percentage rate increase in incidence as well as IRR for each 1-year age increment (eg, IRR for 49-50 years). To reflect the introduction of CRC screening guidelines by the Canadian Task Force of Preventative Care in 2001, we additionally conducted stratified analyses by the calendar year period of 2001-2016. Analyses were also stratified according to sex and cancer site. We completed analyses using SAS statistical software v. 9.4 (SAS Institute, Cary, NC).

As a secondary analysis, we evaluated temporal trends in the incidence of yCRC, focusing on 20 to 49 years of age, over the study period of 1986 through 2016. We also calculated descriptive statistics for incident yCRC cases receiving a diagnosis between the ages of 20 and 49 years given our focus on individuals excluded from average-risk screening protocols. Exponentiated regression coefficients minus 1 estimate the APCi of yCRC for a given segment, with potentially several segments over the study period. We conducted this overall as well as according to sex, cancer site, and the following 10-year age groups: 30-39 years and 40-49 years. The 20- to 29-year age group was not examined because of too few events. We completed analyses using the Joinpoint Regression Program (v. 4.7.0.0).

Study Conduct

This study was approved by the University of British Columbia (H17-03530). All inferences, opinions, and conclusions drawn in this manuscript are those of the authors and do not reflect the opinions or policies of the data steward(s).

Results

Study Sample

A total of 12 770 incident cases (44.8% women; mean [SD] age = 51.6 [7.1] years) of CRC were diagnosed in individuals between the ages of 20 and 60 years over the period of 1986-2016. Of these, 3614 cases (49.9% women) were of incident yCRC (20-49 years), where the mean age at diagnosis was 42.3 years (SD = 5.8 years) (Table 1).

Table 1.

Characteristics of individuals with yCRC diagnosed between the ages of 20 and 49 years (n = 3614)^a

Characteristic	No. (%)
Age, mean (SD)	42.3 (5.8)
Women	1804 (49.9)
Cancer site
Colon	1816 (50.3)
Rectum	1440 (39.9)
Rectosigmoid	358 (9.9)
Neighborhood income quintile
Quintile 1	678 (18.8)
Quintile 2	696 (19.3)
Quintile 3	855 (23.7)
Quintile 4	710 (19.7)
Quintile 5	675 (18.7)
Residence
Urban	3161 (87.5)
Rural	453 (12.5)

^ayCRC = young-onset colorectal cancer.

Annual Incidence Rates of CRC (20-60 Years)

Findings from our primary analyses evaluating the IR of CRC per 100 000 population for each 1-year age interval from 20 to 60 years are illustrated in Figure 1. Overall, CRC IRs presented according to 1-year age increments and spanning our 30-year study period showed a steady increase in incidence during the 49- to 50-year age transition (see Table 2 for IRs and IRRs of surrounding age transitions). The 49- to 50-year age transition is specifically characterized by a CRC rate increase of 18.5%, increasing slightly from an IR of 26.2 (95% CI = 23.9 to 28.7) per 100 000 population at 49 years of age to an IR of 31.1 (95% CI = 28.5 to 33.8) per 100 000 population at 50 years of age, with a corresponding IRR of 1.19 (95% CI = 1.04 to 1.34). This analysis was repeated for the period of 2001-2016 to reflect when guidelines for CRC screening were introduced by the Canadian Task Force on Preventive Health Care (5), and findings similarly suggest an IR increase of 19.7% during the transition from 49 to 50 years of age (27.5 [95% CI = 24.5 to 30.7] per 100 000 population to 32.9 [95% CI = 29.6 to 36.4] per 100 000 population) (Table 2; Supplementary Figure 2, available online). This increase from 49 to 50 years of age corresponds to an IRR of 1.20 (95% CI = 1.02 to 1.40), which is similar to the IRR for the consecutive age transition from 50 to 51 years of age (IRR = 1.17, 95% CI = 1.02 to 1.35).

Figure 1.

Incidence rates of colorectal cancer (per 100 000 population) in 1-year age increments (1986-2016). Incidence rates are shown for A) overall population, B) women, and C) men.

Table 2.

Incidence rates and IRRs describing the incremental 1-year age transitions surrounding the initial age (50 years) for average-risk screening of CRC^a

Age transition	Study period
	1985 to 2016		2001 to 2016
	Incidence rate (95% CI) per 100 000	IRR (95% CI)	Incidence rate (95% CI) per 100 000	IRR (95% CI)
51-52 y	36.7 (33.9 to 39.7) to 42.3 (39.3 to 45.6)	1.15 (1.03 to 1.29)	38.5 (34.9 to 42.4) to 43.1 (39.3 to 47.2)	1.12 (0.98 to 1.28)
50-51 y	31.1 (28.5 to 33.8) to 36.7 (33.9 to 39.7)	1.18 (1.05 to 1.33)	32.9 (29.6 to 36.4) to 38.5 (34.9 to 42.4)	1.17 (1.02 to 1.35)
49-50 y	26.2 (23.9 to 28.7) to 31.1 (28.5 to 33.8)	1.19 (1.04 to 1.34)	27.5 (24.5 to 30.7) to 32.9 (29.6 to 36.4)	1.20 (1.02 to 1.40)
48-49 y	25.1 (22.8 to 27.5) to 26.2 (23.9 to 28.7)	1.05 (0.92 to 1.20)	25.1 (22.2 to 28.2) to 27.5 (24.5 to 30.7)	1.09 (0.93 to 1.29)
47-48 y	20.7 (18.7 to 22.9) to 25.1 (22.8 to 27.5)	1.21 (1.05 to 1.39)	21.5 (18.9 to 24.4) to 25.1 (22.2 to 28.2)	1.17 (0.98 to 1.39)

^aCI = confidence interval; CRC = colorectal cancer; IRR = incidence rate ratio.

IRs according to cancer site over the period of 1986-2016 are presented in Figure 2. The corresponding IRRs for the 49- to 50-year age transition are as follows: colon, 1.08 (95% CI = 0.90 to 1.30); rectosigmoid, 1.17 (95% CI = 0.80 to 1.73); and rectum, 1.33 (95% CI = 1.09 to 1.62). A marked increase in the incidence of rectosigmoid cancer was observed over the 54- to 55-year age transition with an IRR of 1.63 (95% CI = 1.21 to 2.20). The 1-year IRs of CRC according to cancer site for the contemporary cohort (2001-2016) are shown in Supplementary Figure 3 (available online).

Figure 2.

Incidence rates of colorectal cancer (per 100 000 population) in 1-year age increments according to cancer site (1986-2016). Incidence rates are shown according the following sites: A) colon, B) rectosigmoid, and C) rectum.

Trends in the Incidence of yCRC (<50 Years)

Trends in the annual incidence of yCRC during the study period are reported in Table 3. Results show an increase in the overall rate of yCRC from 1986 to 2016 by an APCi of 1.48% (95% CI = 1.02% to 1.93%). Stratified analyses suggest the increasing rate of yCRC is largely driven by rectal cancer (APCi = 2.35%, 95% CI = 1.58% to 3.14%) and by yCRC among men, as evidenced by an APCi of 2.17% (95% CI = 1.59% to 2.76%) compared with an APCi of 0.79% (95% CI = 0.22% to 1.36%) for women. When evaluating the rate of yCRC incidence according to 10-year age groupings, the largest APCi is observed in the 30- to 39-year age group, with an APCi of 2.82% (95% CI = 1.87% to 3.78%) for women and men combined. Stratification by age and sex yielded APCis of 2.49% (95% CI = 1.36% to 3.63%) for women and 2.97% (95% CI = 1.65% to 4.31%) for men in the 30- to 39-year age group. For the 40- to 49-year age group, the APCi was 0.12% (95% CI = −0.54% to 0.79%) for women and 1.86% (95% CI = 1.19% to 2.53%) for men. The 20- to 29-year age group was omitted in the stratified analysis because this age group had several years with zero yCRC cases.

Table 3.

Average APCi of yCRC from 1986 to 2016^a

Overall and subgroups	APCi, % (95% CI)
	Overall	Women	Men
All yCRC (20-49 y)	1.48 (1.02 to 1.93)	0.79 (0.22 to 1.36)	2.17 (1.59 to 2.76)
10-year age grouping
30-39 y	2.82 (1.87 to 3.78)	2.49 (1.36 to 3.63)	2.97 (1.65 to 4.31)
40-49 y	1.02 (0.54 to 1.50)	0.12 (−0.54 to 0.79)	1.86 (1.19 to 2.53)
Cancer site
Colon	1.33 (0.77 to 1.90)	0.83 (-0.05 to 1.72)	1.80 (1.07 to 2.55)
Rectosigmoid	−1.38 (−2.68 to −0.06)	−1.83 (−3.36 to −0.27)	−1.34 (−3.19 to 0.55)
Rectum	2.35 (1.58 to 3.14)	1.45 (0.60 to 2.31)	3.21 (2.12 to 4.31)

^aAPCi = annual percent change in incidence; CI = confidence interval; yCRC = young-onset colorectal cancer.

Discussion

Our population-based evaluation using administrative health data from 1986 to 2016 that included 12 770 individuals diagnosed with CRC between the ages of 20 and 60 years showed a steady increase in the incidence of CRC, particularly over the 49- to 50-year age transition, when incidence was presented in 1-year age increments. These findings persisted when assessment was restricted to the more contemporary time period of 2001 to 2016 with the intent of encompassing when CRC screening guidelines were introduced in Canada. Specifically, the 49- to 50-year age transition was characterized by an IRR of 1.20 (95% CI = 1.02 to 1.40) compared with an IRR of 1.19 (95% CI = 1.04 to 1.34) when evaluating data for all study years (1986-2016). When considered collectively with results of our secondary analysis that shows an increasing trend of yCRC incidence from 1986 to 2016 (APCi = 1.48%, 95% CI = 1.02% to 1.93%), our study highlights the changing epidemiology of CRC among younger adults and the need for widespread awareness among patients and health care providers.

It is important to contextualize our findings with those reported by Abualkhair et al. (8). These authors used population-based US data from 2000 to 2015 to evaluate the incidence of CRC in 1-year age increments among individuals 30-60 years of age with the rationale that a drastic increase of CRC cases as individuals shift from 49 to 50 years of age, when average-risk screening begins, would suggest an abundance of preclinical undetected yCRC cases and thereby support a lower age for guideline-based screening. Abualkhair et al. (8) found that this age transition captured the largest increase of CRC incidence, characterized by a 46.1% increase and an IRR of 1.46 (95% CI = 1.42 to 1.51). In contrast, our replication of this analysis in a Canadian cohort displayed a steady increase of CRC incidence through the same age transition and represented a 19.7% increase (IRR = 1.20, 95% CI = 1.02 to 1.40) in the contemporary cohort. This discrepancy between our study and that of Abualkhair et al. (8) could be a result of differing health care systems as well as the timing of guideline implementation. First, fewer CRC cases clustered at 50 years of age in our study may be explained by Canadians experiencing fewer barriers to early CRC symptom workup as a result of publicly funded health care, leading to CRC case detection being more distributed over each year of age. In addition, although Canadian guidelines were published in 2001 (5), the BC Colon Screening Program, which screens asymptomatic individuals between the ages of 50 and 74 every 2 years, was not piloted until 2009 and was fully implemented in 2013 (17,18). The recentness of the BC screening program is reflected in results from the nationally administered Canadian Community Health Survey in 2012 that indicate only 49.6% of eligible BC residents received up-to-date CRC screening (19). Given that our study captured the early implementation phase of CRC screening, we would expect to observe an increase in CRC incidence for all ages older than 50 years rather than an isolated increase. Indeed, the constant increase in CRC incidence from 49 to 50 years (IRR = 1.20, 95% CI = 1.02 to 1.40) and from 50 to 51 years (IRR = 1.17, 95% CI = 1.02 to 1.35) lends support to this explanation.

Differences in CRC screening modalities used in Canada compared with the United States may also explain the steady increase. In BC, for the majority of the study period, the fecal occult blood test was the first screening test performed, and only if this test was positive would patients be referred for colonoscopy. In contrast, in the United States there is a larger use of colonoscopy, which is considered a gold standard when evaluating assay sensitivity. Although the newer fecal immunochemical test (FIT) has reduced some of the disparity in sensitivity (20), these practice differences may explain the larger increase in CRC incidence reported by Abualkhair et al. (8). The influence of screening practices also extends to our understanding of the large increase in rectosigmoid cancer incidence noted at age 55 years and periodic increases for rectal cancer occurring every 2 years. The 5-year interval increase may be explained by individuals who had a low-risk adenoma detected at age 50 years and subsequently had a follow-up colonoscopy at age 55 years as per current screening guidelines in BC. Periodic increases of rectal cancer at 2-year age intervals could reflect the average-risk screening pathway that includes FIT testing every 2 years.

Our study also reinforces the notion that incidence of yCRC is increasing. In 2019, Brenner et al. (21) assessed CRC incidence in Canada according to sex and age (<50 years, ≥50 years) from 1969 to 2015 and found that yCRC incidence increased for both women (APCi = 4.45%) and men (APCi = 3.47%) beginning in 2010 and 2006, respectively. An evaluation of yCRC incidence trends worldwide by Siegel et al. (22) further emphasizes this phenomenon, with incidence particularly increased in high-income countries and in the form of rectal cancer among Canadians, a finding corroborated in our study. Our analysis illustrates that yCRC incidence is specifically accelerating for both women and men between 30 and 39 years of age (APCi = 2.49% for women and 2.97% for men), and yCRC incidence remains increased for men between 40 and 49 years of age (APCi = 1.86%). A concentrated increase in the incidence of yCRC among 30- to 39-year olds has also been reported in a Canada-wide cohort study by Patel et al. (23) (1996-2010 APCi = 2.4%, overall; 2.3%, women; 2.5%, men) as well as a 2017 US study by Siegel et al. (24) that reported steeper increases in rectal cancers (1980-2013 APCi = 3.2%) over colon cancers (1988-2013 APCi = 1.0%) for the 30- to 39-year age group. Low case counts prohibited our analysis of yCRC incidence in individuals younger than 30 years of age; however, North American studies have also noted an increasing trend in the risk of yCRC for this age group, with APCis ranging from 2.4% to 6.3% and 3.2% to 7.0%, for colon and rectal cancers, respectively (23,24). The increasing incidence of yCRC calls for greater recognition of symptoms by both patients and physicians and provides support for discussion and consideration of lowering the recommended age for initiating screening, especially given that patients with yCRC often present with more advanced-stage disease (25).

The strengths and limitations of our study must also be addressed. Our CRC study sample that spans a 30-year period was drawn from a population-based cohort, created by linking data from Population Data BC and the BC Cancer Registry, which capture data on approximately 95% of all cancer cases in the province. The BC Cancer Registry is reviewed annually for quality, completeness, and accuracy by the North American Association of Central Cancer Registries (26). Importantly, data in the BC Cancer Registry are collected for the purpose of generating cancer statistics and conducting surveillance on the burden of cancer in BC (26). However, administrative health data are not without their limitations. The BC Cancer Registry lacks sufficient data on CRC disease stage, which were collected beginning in 2010 and are not acquired using a systematic approach because information on stage relies on a variety of data sources such as death certificates and pathology reports. Access to disease stage would have permitted a more comprehensive assessment of CRC incidence among young adults. However, the absence of these data does not influence the results because our study did not involve any analyses on impacts of stage on patient outcomes.

Altogether, in evaluating trends in the incidence of yCRC at the population level, we observed a steady increase in the annual rate of yCRC over the age span from 20 to 60 years, which was characterized by a marked increase around screening age from 49 to 50 years. Combined with the observed rising incidence of yCRC in the population of BC, with a notable increase for individuals between 30 and 39 years of age, our study demonstrates the need to increase education and awareness on the changing epidemiology of CRC, particularly the increased risk among younger adults. Consideration on whether to lower the age for average-risk screening in BC to ensure that guidelines recognize the growing risk of yCRC and minimize the occurrence of delayed diagnoses necessitates ongoing discussion and evaluation as the screening program becomes well established.

Funding

This research was funded by a Project Grant from the Canadian Institutes of Health Research, “Examining the epidemiology, treatment, and outcomes in young-onset colorectal cancer” (funding reference number: PJT-159467).

Notes

Role of the funder: The funder had no role in study design, data collection and analysis, interpretation of the data, decision to publish, or preparation of the manuscript.

Disclosures: The study authors have no conflicts of interest to declare.

Author contributions: MDV contributed to conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, supervision, visualization, data interpretation, and writing original draft. AH contributed to conceptualization, formal analysis, investigation, methodology, project administration, visualization, data interpretation, and writing original draft. ECS contributed to data curation, formal analysis, investigation, methodology, software, validation, and visualization. JML, SG, CJB, MJR, and AF contributed to conceptualization, methodology, and data interpretation. MDV, JML, SG, and CJB contributed to funding acquisition. All study authors reviewed and edited the manuscript.

Data Availability

Data used for this study are protected by a strict data sharing agreement between the researchers and data stewards.