Overall Health Care Cost During the Year Following Diagnosis of Colorectal Cancer Stratified by History of Colorectal Evaluative Procedures

Abstract Background The cost-effectiveness of colorectal screening has been modeled; however, the cost of health care following the diagnosis of colorectal cancer has not been described stratified by history of colorectal evaluative procedures. Methods We identified persons with first diagnosis of colorectal cancer between 2015 and 2017 from the Ontario Cancer Registry, and categorized them by history of colorectal evaluative procedures during Period 1 (the 10 years before the 6-month prediagnostic interval) with or without procedures during Period 2 (the 6 month prediagnostic interval), versus only during Period 2, versus none. We extracted overall health care cost 1 year following diagnosis from population-wide administrative databases. Results Among cases diagnosed at 52 to 74 years, overall health care cost among those with no colorectal evaluative procedures on or before the date of diagnosis is $71,039.65 (SD $51,825.18), compared to $48,406.15 (SD $38,843.64) among those who received colorectal evaluative procedures during Period 1, with or without procedures during Period 2. Among the population aged 20 to 74 years at diagnosis, cases with ≥1 screening colonoscopies for hereditary CRC syndrome, the mean overall initial cost was between $32,300.32 (SD) and $33,084.67 (SD $39,905.77), and those with ≥1 screening colonoscopies because of a first-degree relative with CRC, was between $36,344.71 (SD $35,539.85) and $45,456.41 (SD $49,818.59). Conclusions Overall health care cost is lower among cases who received colorectal evaluative procedures during Period 1, with or without procedures during Period 2, and among those with screening colonoscopy for hereditary CRC syndromes or affected first-degree relatives.


Introduction
Opportunistic colorectal screening in Ontario began during the 1990s (1)(2)(3). Ontario inaugurated its population-based colorectal screening program ColonCancerCheck (CCC) in 2008, recommending to 50 to 74-year-old persons biennial screening with guaiac fecal occult blood testing (gFOBT) except for those with affected first-degree relatives, to whom screening colonoscopy was recommended (4). Frequent colonoscopy is recommended beginning at a young age for those with hereditary syndromes (5,6).
The cost-effectiveness of colorectal screening in reducing mortality from colorectal cancer (CRC) in Ontario has been demonstrated by modeling (7,8), but variation in health care cost following diagnosis of CRC has not been reported stratified by history of colorectal screening. The aim of this study is to estimate mean overall health care cost per case borne by the government of Ontario, during the year following diagnosis of CRC, in the screening age-eligible population aged 52 to 74 years at diagnosis between 2015 and 2017, stratified by history of colorectal screening or evaluative procedures, and in the population diagnosed at age 20 to 74 years, by history of screening colonoscopy. The usefulness of this information would be primarily as inputs in modeling cost-effectiveness of interventions, including but not limited to, colorectal screening, and other policy questions focused on CRC control.

Identification of CRC Cohort in the Screening-Eligible Age Range 52 to 74 Years
To estimate the mean overall cost of care for CRC, we identified persons aged between 52 and 74 years at first diagnosis of CRC between 2015 and 2017 from the Ontario Cancer Registry (OCR). We excluded cases with previous diagnosis of CRC, any other invasive malignancy within 5 years before, or 1 year following, diagnosis of CRC, but did not exclude cases with multiple synchronous CRCs on the same diagnosis date or metachronous CRC diagnosed within 12 months from the index diagnosis date. We excluded cases with missing cancer stage. We identified deaths from the Registered Persons Database. We categorized cases broadly as colon versus rectosigmoid+ rectal, because of potential misclassification in registry and administrative data. Date of last follow-up was March 31, 2019.
We extracted billing claims for colonoscopy from the Ontario Health Insurance Plan (OHIP) database during 10.5 years before diagnosis, and records gFOBT 10.5 years before diagnosis, from the OHIP database, and after April 1, 2008, from the ColonCancerCheck (CCC) database of gFOBT reports. The indication for colonoscopy is not recorded. The indication for gFOBT is unknown in the OHIP database; gFOBT records in the CCC database should represent screening of asymptomatic individuals, however, CCC does not verify the asymptomatic status of those completing gFOBT.
The records of gFOBT and/or colonoscopy during Period 1 (the 10 years before the 6 month prediagnostic interval) would represent either diagnostic investigation, periodic colorectal screening or periodic postpolypectomy surveillance. We also identified gFOBT and/or colonoscopy during (Period 2) (the 6 month prediagnostic interval). The importance of the distinction between Period 1 and Period 2 derives from several issues including firstly, the inability to distinguish screening from diagnostic investigation in either period, secondly, the negative predictive value of colorectal evaluative procedures not leading to diagnosis of CRC during Period 1, which have effectively screened the patient, regardless of the intention or indication for the procedure, and thirdly, the prevalent nature of cancers detected without a prior history of colorectal evaluation during Period 1 and the detection bias associated with cases of cancer detected by the case's first ever screening test during Period 2. It is possible, but not verifiable, that few persons with colonoscopy in both Periods had a postcolonoscopy CRC, as well as a few persons with colonoscopy in Period 1 but no colonoscopy in Period 2.
To categorize comorbidity, we used the Ambulatory Care Group program (9)  We identified first diagnoses of CRC in the 20 to 74 age range from the OCR, excluding those with CRC at any time before 2015, those with missing stage, those with diagnosis of inflammatory bowel disease or with total colectomy before diagnosis of CRC, those whose residence was not in Ontario at the time of diagnosis and those with less than 36 months continuous eligibility for OHIP.
We identified billing claims for colonoscopy Period 1 (10 years before the 6 months before diagnosis), and during Period 2. We categorized them according to feecode in a hierarchical fashion in the following sequence: firstly, Z494 screening colonoscopy for persons with hereditary CRC syndrome (although this code may also be applied to persons with inflammatory bowel disease, we have already excluded such persons), secondly, Z499 screening colonoscopy for persons with affected first-degree relatives but not hereditary CRC syndromes, and thirdly, all other colonoscopies. The mean number of colonoscopies during Period 1 was computed, with the percent of cases with Stage 1 CRC.
Identification of Health Care Costs Borne by the Government of Ontario beginning on the date of diagnosis, using algorithms originally developed by Wodchis et al. (10) and subsequently enhanced (11)(12)(13)(14)(15)(16). Costs include inpatient, outpatient and emergency department hospital costs, visits and treatments at ambulatory cancer and dialysis centres, intravenous chemotherapy, radiation therapy, reimbursement of physicians and surgeons, medical laboratory services, oral medications including oral chemotherapy, in-patient rehabilitation, institutional complex continuing and long-term care, home care services and assistive device procurement. The data sources do not allow accurate separation of costs related to the initial care for CRC from costs incurred because of concurrent conditions. Costs were presented in 2017 Canadian dollars; on average during 2017, $ 1 United States currency = $1.2986 Canadian currency (17).

Approximating the Mean Overall Health Care Cost During the Year Following Diagnosis of CRC Among Cases Aged 52 to 74 Years
We computed the mean overall health care cost across 360 days beginning on the date of diagnosis of CRC, to approximate the cost incurred during the first year following diagnosis. We re-computed the mean overall health care cost across the first 360 days, excluding cases with ≤690 days of follow-up (to approximate the mean overall cost during the entire year following diagnosis of CRC without contribution from cost incurred during the final year of life).
We stratified the mean overall health care cost across 360 days, beginning on the date of diagnosis, by the history of colorectal evaluative procedures firstly, during Period 1, plus or minus during Period 2, secondly, first ever procedures only during Period 2, and thirdly, no procedures on or before the date of diagnosis The mean overall health care cost was also stratified by age group at diagnosis (52 to 64 years versus 65 to 74 years), sex, anatomic site (colon versus rectosigmoid+rectum) and cancer stage.
Approximating the mean overall health care cost during the year following diagnosis of CRC among cases first diagnosed at age 20 to 74 years.
The method of computation was identical to that used with the 52 to 74-year-old cohort, however, stratification was by colonoscopy, using the hierarchy of firstly, screening colonoscopy

Cohort of Persons With CRC Aged 52 to 74 Years
We identified 12,550 cases of CRC aged 52 to 74 years between 2015 and 2017 from the OCR. We excluded 570 cases because of previous diagnosis of CRC, 752 cases because of any other invasive malignancy within 5 years before diagnosis, 344 cases with any other invasive malignancy 1 year following the diagnosis of CRC, 903 cases lacking cancer stage, and 4 cases because the recorded diagnosis date fell after the recorded date of death, leaving a cohort of 9977 cases (comprising 6310 persons with colorectal evaluative procedures during Period 1 with or without procedures during Period 2, 2545 persons with procedures only during Period 2, and 1122 persons with no procedures at any time on or before the date of diagnosis. The cohort is described by history and timing of first gFOBT and/or colonoscopy in Table 1, and by history and timing of first gFOBT and/or colonoscopy stratified by age, and sex, for each stage of colon cancer and each stage of rectosigmoid+rectal cancer in Table 2. Among cases with colorectal evaluative procedures during Period 1, 3424/6310 (54.3%) underwent colonoscopy but no gFOBT during Period 2 and 1543/6310 (24.5%) underwent gFOBT plus colonoscopy, while 1343/6310 (21.3%) did not have colonoscopy during the 6-month prediagnostic interval. Cases with colorectal evaluative procedures during Period 1 were older and were more likely to have one or more major adult ADG diagnoses compared to those who did not. The percent of cases diagnosed with stage 1 CRC is highest among those who began to receive colorectal evaluative procedures during Period 1, conversely the per cent diagnosed with stage 4 CRC is highest among those with no record of gFOBT or colonoscopy on or before the date of diagnosis (  The number of cases aged 65 to 74 years in this colonoscopystratified cohort is lower than in the cohort aged 52 to 74 years because of the additional exclusion criteria, however, the estimates of mean overall health care cost are similar for the two cohorts (Tables 4 and 5).

Discussion
Overall health care cost during the year following diagnosis of CRC diagnosed in the screening age-eligible population aged 52 to 74 years, is lower among cases who received colorectal evaluative procedures during Period 1 (the 10 years before the 6-month prediagnostic interval), generally with further evaluation during Period 2 (the 6-month prediagnostic window), reflecting the likelihood that some of those cases had participated in periodic screening. Although we cannot distinguish asymptomatic screening from diagnostic investigation as the intention of any one record of gFOBT or colonoscopy, it is likely that many had been participating in periodic asymptomatic screening by gFOBT or colonoscopy, or periodic colonoscopic surveillance after prior polypectomy, in addition to those who were undergoing diagnostic investigation of symptoms on one or more occasions. In effect, these cases have been screened, regardless of the intention and indication for those procedures. The steady increments in mean overall health care cost during the year following diagnosis of CRC among strata simultaneously defined by cancer stage and anatomic site of primary CRC indicate that initial care year costs should be reported by this simultaneous stratification and not by simple stratification by stage or by anatomic site of primary CRC. The higher overall health care cost among those without colorectal evaluative procedures during Period 1 is not due to higher burden of comorbidity: The percent of cases with one or more major adult ADGs is highest among those with colorectal evaluative procedures during Period 1, who have the lowest cost during the year following the diagnosis of CRC. This supports the hypothesis that periodic screening, surveillance, or diagnostic investigation during Period 1 results in a higher percent of cases diagnosed in stage 1 and lower overall health care costs during the year following diagnosis.
Overall health care cost during the year following diagnosis of CRC between the ages of 20 to 74 years is lower among those who have had screening colonoscopies because of hereditary CRC syndromes or affected first-degree relatives. However, the higher CRC stage and higher overall mean cost for those with screening colonoscopy because of affected first-degree relatives in Period 1 who did not have colonoscopy in Period 2 illustrates that some postcolonoscopy CRCs will be associated with higher costs. The feecode for screening colonoscopy because of affected first-degree relatives was introduced at the end of 2011, so the maximum interval between the first family history colonoscopy and the diagnosis of CRC would have been 3 to 6 years, depending on the year of diagnosis.
There have been publications of health care costs in Ontario during the year following diagnosis of CRC among cohorts diagnosed before the 2015 to 2017 observation period.  (16). However, we are unaware of any prior publications from Ontario or any other jurisdiction stratifying health care costs after diagnosis of CRC by the presence or absence of a history of colorectal evaluative procedures during Period 1, or stratified by any other factor related to exposure to colorectal screening.
There are limitations to estimation of costs from administrative data, compared to prospective collection. There is the possibility of miscoding of diagnoses and procedures, which may occur more frequently in records completed by practitioners (e.g., OHIP physician billing claims database), compared to professional health records technologists (e.g., CIHI DAD and SDS databases), which in turn would distort the cost estimates. These estimates cannot be used to compute total government expenditures on the care of all CRC in Ontario by simple multiplication of the mean overall cost by the number of cases, due to the restrictive cohort definitions and exclusion criteria, which were required in order to meet the aims of this work. The estimates reflect the health care funding provided by the government of Ontario, and are not necessarily reflective of cost in other jurisdictions or in other currencies.

Conclusions
Mean overall health care cost during the year following diagnosis of CRC is lower among cases who received colorectal evaluative procedures during Period 1 (10 years before the 6 month prediagnostic interval), generally with further evaluation during Period 2 (the 6 month prediagnostic interval), and among those with screening colonoscopy for hereditary CRC syndromes or affected first-degree relatives.