Decreased Colorectal Cancer Incidence and Incidence-Based Mortality in the Screening-Age Population of Ontario

Abstract Background and Aims We aimed to evaluate trends in Ontario, Canada, 2002 to 2016, in uptake of colorectal evaluative procedures, colorectal cancer (CRC) incidence and incidence-based mortality in the colorectal screening-age population. Methods We defined the screening age-eligible population as persons 51 to 74 years of age with ≥1 year eligibility for the Ontario Health Insurance Plan, excluding those with a diagnosis of CRC in the Ontario Cancer Registry (OCR) prior to age 50 or January 1, 2002. We computed annual up-to-date status with colorectal evaluative procedures from billing claims, and CRC incidence from the OCR. In order to compute incidence-based CRC mortality, we included persons with a first diagnosis of CRC between the ages of 51 and 74, diagnosed between January 1, 1992 and December 31, 2001, still alive and <75 years of age on January 1, 2002, based on cause of death from the OCR. Overall, age-stratified and sex-stratified trends were evaluated by Cochran–Armitage trend tests. Results Persons up to date with colorectal evaluative procedures increased from 628,214/2,782,061 (22.6%) in 2002 to 2,584,570/4,179,789 (62.2%) in 2016. CRC incidence fell from 129.3/100,000 in 2002 to 94.54/100,000 in 2016, and incidence-based CRC mortality fell from 40.8/100,000 to 24.1/100,000. Decreasing trends in overall and stratified incidence and mortality were all significant, except among persons 51 to 54 years old. Conclusions There was continued increase in persons up-to-date with colorectal evaluative procedures, and significant decrease in CRC incidence and incidence-based CRC mortality from 2002 through 2016.


Background
Statistically significant declines in age-standardized colorectal cancer (CRC) incidence stratified by site and sex have been documented in Canada since 1983 for all combinations of colorectal subsites and sex, with the exception of rightsided colon cancer among males, for whom a statistically significant annual percent change increase in age-standardized Subsequent to a meta-analysis of trials of colorectal screening using fecal occult blood testing (FOBT) demonstrating significant reduction in CRC mortality in 1998 (2), two Canadian guidelines recommending biennial FOBT were issued in 2001 (3) and 2002 (4). At the time, there was evidence of substantial use of colorectal evaluative procedures among screening-age persons in Ontario without prior diagnoses of bowel diseases (5)(6)(7). In 2008, Ontario established ColonCancerCheck (CCC, a population-based colorectal screening program for persons at average risk for CRC aged 50 to 74 years recommending biennial FOBT, and screening by colonoscopy for those with a first-degree relative affected by CRC) (8).
In this paper, we aim to evaluate trends in colorectal evaluative procedures, CRC incidence and incidence-based mortality in the screening-age population of Ontario, 2002 to 2016.

METHODS
This work was approved by the Research Ethics Board of Sunnybrook Health Sciences Centre, Toronto, Ontario (REB 396-2017).

Identification of the Cohort of Colorectal Screening-Eligible Persons
We identified persons 51 to 74 years of age with ≥1 year eligibility for the Ontario Health Insurance Plan (OHIP) during 2002 to 2016 based on OHIP's Registered Persons Database (RPDB), excluding those with a diagnosis of CRC in the Ontario Cancer Registry (OCR) age <= 50 years or diagnosis prior to 2002. Records of FOBT were identified from the OHIP billing claims file and CCC's FOBT database. Flexible sigmoidoscopy (FS), CT colonography (CTC, data available since 2011 only) and colonoscopy were identified from the OHIP physician billing claims file. The intended indications are unavailable in the data. CRC was identified from the OCR. All persons were followed until December 31, 2016, their 75th birthday, last date of OHIP eligibility, first diagnosis of CRC or date of death, whichever came first.

Computation of Up-to-Date Status With Colorectal Evaluative Procedures
For each year, 2002 to 2016, we computed the up-to-date status of each person with colorectal evaluative procedures. Being up-to-date with FOBT was given priority over any other colorectal evaluative procedure performed during the 24 months following the date of the FOBT. If a person had a record of FOBT ≤24 months prior to the end of the year, but no record of colonoscopy within the prior 10 years or any record of CTC or FS within the prior 5 years, the person was classified as up-todate with FOBT. If not up-to-date with FOBT, but with a record of FS ≤60 months prior to the end of the year, with no colonoscopy within the prior 10 years or CTC within the prior 5 years, the person was classified as up-to-date with FS. If not up-to-date with either FOBT or FS but with a record of CTC ≤60 months prior to the end of the year, with no colonoscopy within the prior 10 years, the person was classified as up-to-date with CTC. If not up-to-date with FOBT, FS or CTC but with a record of colonoscopy ≤10 years prior to the end of the year, the person was classified as up-to-date with colonoscopy. Otherwise, the person was classified as not up-to-date with any colorectal evaluative procedure. For each calendar year, the percent up-to-date (overall and stratified by age and by testing modality as above), and the percent not up-to-date with any test, was computed and plotted. We also computed the percent up-to-date with FOBT among a subpopulation of those who were not up-to-date with colonoscopy. For each year, the rate of CRC per screening-eligible 100,000 persons was computed, with 95% confidence intervals (CIs), overall, and stratified by age, sex, and by ICD10 code C18 (colon) versus ICD10 codes C19 plus C20 (rectosigmoid plus rectum). We computed Cochran-Armitage trend tests on incidence from 2002 to 2016, overall, stratified by sex and stratified by age groups.

Computation of Incidence-Based CRC Mortality
To compute incidence-based CRC mortality (9,10) among persons aged 51 to 74 during 2002 to 2016, we identified the underlying population at risk for CRC death in the screening age-eligible age range by including persons diagnosed with CRC in the colorectal screening age range prior to January 1, 2002, still alive on that date, and still in the age range 51 to 74 years, followed until December 31, 2016, their 75th birthday, last date of OHIP eligibility or date of death, whichever came first. Cause of death was identified from the OCR.
For each year, incidence-based CRC mortality was computed with 95% CIs, overall, stratified by age and sex. We computed Cochran-Armitage trend tests on incidence-based mortality from 2002 to 2016, overall, stratified by age and stratified by sex. All analyses were performed using SAS version 9.3.

RESULTS
The Colorectal Screening-Eligible Persons, 2002 to 2016 (51.2% female) in 2016. Over 90% had been eligible for OHIP ≥10 years on the date of initial inclusion in the study population. Table 1   Several factors underlie the lack of increase in the percent up-to-date with FOBT. Although there is high-quality evidence that mailing FOBT kits to screening eligibles increases participation in screening (11), this was not a component of CCC.
In addition, primary care practitioners in Ontario believe that FOBT is inferior to colonoscopy as a screening test, and therefore often do not recommend it to their patients (12). In 2016, the percent up-to-date with colorectal evaluative procedures 8 years after the introduction of CCC was similar to the level of self-reported colorectal screening status in the 2015 United States National Health Interview Survey, which found that recent colorectal screening was reported by 63.4% of ageeligible females (compared to 65% observed to be up-to-date in Ontario) and 61.9% of age-eligible males (compared to 60% observed to be up-to-date in Ontario) (13). While increases in colorectal screening participation in North America are encouraging, further improvements in the prevention and early detection of CRC are required.
An international collaborative group of cancer screening researchers has examined barriers to effective screening, especially factors within screening programs as well as external health system factors, using the Barriers to Effective Screening Tool (BEST) (14,15). BEST was applied among cancer screening organizations, screening researchers and health policymakers, as well as by systematic literature review (16). Many of the barriers are outside the direct control of cancer screening programs, for example, incomplete and/or inaccurate lists of eligibles and their addresses (15), difficulties in access to screening (17), insufficient coordination (14,15,17) and expenditures on opportunistic screening (14,16). Some expenditures on opportunistic screening may add little to the effectiveness of colorectal screening in Ontario; however, if those resources were to be reallocated to programmatic screening, the levels of participation and effectiveness of screening might be enhanced. We have previously shown that 33.7% of patients in the colorectal screening-eligible age range who underwent a complete negative outpatient colonoscopy, without prior, new or subsequent diagnoses of CRC or inflammatory bowel disease, went on to have a repeat colonoscopy between 0.5 and 5.5 years later (18). During that follow-up window, only 0.5% received a new diagnosis of CRC or inflammatory bowel disease, or underwent bowel resection for any reason in the interval (18).
We have found statistically significant decreases in CRC incidence between 2002 and 2016, overall, and stratified by anatomic site, sex and age, continuing previously observed downward secular trends in CRC incidence in Canada (1983 to 2007) (1) and the United States (19,20). These decreases began earlier than widespread screening and colonoscopy utilization, and so should not be ascribed entirely to the impact of widespread colorectal screening on CRC prevention. Nevertheless, as in the United States (19), much of the decline in CRC incidence in Ontario is likely due to the increasing prevalence of colonoscopy for screening and other indications. We did not find a decrease in CRC incidence among 51-to 54-year-old persons. The incidence of rectosigmoid plus rectal carcinoma among this age group was higher in 2016 than in 2002. Colonoscopic polypectomy performed during the first few years of screening age-eligibility would not be expected to decrease incidence of CRC among those aged 51 to 54.
We have found statistically significant decreases in incidence-based CRC mortality 2002 to 2016, similar to long-term declines in CRC mortality documented in the Canada (1), United States (21) and Europe (22), which occurred in advance of large increases in the uptake of colorectal evaluative procedures and screening and in advance of the introduction of organized colorectal screening programs. The effectiveness of colonoscopic polypectomy in preventing CRC likely results in decreased CRC mortality more than a decade after the procedure (2,3). The decline in CRC incidence associated with the increasing prevalence of colonoscopic polypectomy provided in Ontario since the decade of the 1990s (7) is the most likely explanation for much of the decline in incidence-based CRC mortality across the entire time period of observation to 2016 (23). It is possible that some of the decline in incidence-based CRC mortality could be attributable the prior secular trends in CRC incidence and mortality (1), and also to better access to appropriate surgery and chemotherapy for CRC, both of which have been given a high priority by Cancer Care Ontario.
The strengths of this study include the availability of all records of all colorectal evaluative procedures, CRC diagnoses and deaths from CRC, which are of good quality (24,25). The entire geographically defined target population itself is a major strength, whose number increased from 2,782,061 to 4,179,789 persons between 2002 and 2016, including marginalized subpopulations and those who experience barriers to care. Over 90% of screening-eligible persons had residence in Ontario for >10 years, so the possibility of misclassification of any person's history of colorectal evaluative procedures or CRC is low. Limitations include the absence of information on the indication for colorectal evaluative procedures throughout the study period, and the inability to examine trends in FOBT positivity before and after the introduction of CCC. It is unknown if the fecal immunochemical test would have been associated with a higher percentage of persons up-to-date with stool testing. We emphasize that the decreases in incidence and incidencebased mortality are confined to the typical screening ageeligible population, that we did not observe decreases in the 51-to 54-year-old age group, and that our findings must not be extrapolated to those younger or older than the typical screening age-eligible population.

Conclusion
The percent of screening age-eligible permanent residents of Ontario up-to-date with colonoscopy continued to increase annually during 2002 to 2016, following the previously observed increased utilization in 1992 to 2001, which was likely the most significant factor in the decline of CRC incidence and incidence-based mortality in Ontario from 2002 to 2016, and likely more important than other factors such as long-term gradual declines in incidence and mortality, or the impact on mortality of improved treatment.

Funding
This work was supported by a grant from Cancer Care Ontario to LFP. This study was also supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). The funder had no role in the design of the study or collection, analysis and interpretation of the data or in writing the manuscript. The opinions, results, views and conclusions reported in this paper are those of the authors and do not necessarily reflect those of the ICES, the MOHLTC or Cancer Care Ontario, and no endorsement by these bodies is intended or should be inferred.