You are here

Article Text

Article menu

Download PDFPDF

Colon
Original article
Population-based assessment of the outcomes in patients with postcolonoscopy colorectal cancers
  1. Anand Govindarajan1,2,
  2. Linda Rabeneck2,3,4,5,
  3. Lingsong Yun2,
  4. Jill Tinmouth2,3,4,5,
  5. Lawrence F Paszat2,5,
  6. Nancy N Baxter2,5,6
  1. 1Department of Surgery, Mount Sinai Hospital, University of Toronto, Toronto, Ontario, Canada
  2. 2Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada
  3. 3Cancer Care Ontario, Toronto, Ontario, Canada
  4. 4Department of Medicine, University of Toronto, Toronto, Ontario, Canada
  5. 5Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada
  6. 6Department of Surgery, Keenan Research Centre, Li Ka Shing Knowledge Institute, St. Michael's Hospital, University of Toronto, Toronto, Ontario, Canada
  1. Correspondence to Dr Anand Govindarajan, Department of Surgery, Mount Sinai Hospital, University of Toronto, 600 University Avenue, Suite 1220, Toronto, Ontario, Canada M5G1X5; agovindarajan{at}mtsinai.on.ca

Abstract

Objective The potential for cancers to not be detected on colonoscopy is increasingly recognised, but little is known about patient outcomes. The objective of this study was to assess the outcomes of patients diagnosed with postcolonoscopy colorectal cancers (PCCRCs).

Design We conducted a population-based retrospective cohort study, including all patients diagnosed with colorectal cancer (CRC) in Ontario, Canada from 2003 to 2009. Patients were categorised into three groups: DETECTED (diagnosed within 6 months of first colonoscopy), PCCRC (diagnosed 6–36 months after first colonoscopy) or NOSCOPE (no colonoscopy within 36 months of diagnosis). Univariate and multivariable analyses were conducted to study overall survival, surgical treatment, emergency presentation and surgical complications.

Results Overall, 45 104 patients were included, with 2804 being classified as having a PCCRC. Compared with the DETECTED group, PCCRC was associated with a significantly higher likelihood of stage IV disease (17.2% vs 12.9%), worse overall survival (5 year OS: 60.8% vs 68.3%, p<0.0001; adjusted HR: 1.25, 95% CI 1.17 to 1.32, p<0.0001), a higher likelihood of emergency presentation (OR: 2.86, 95% CI 2.56 to 3.13, p<0.001) and lower likelihood of surgical resection (OR: 0.61, 95% CI 0.55 to 0.67, p<0.001). However, patients with PCCRC had significantly better outcomes than those in the NOSCOPE group (stage IV: 37.1%, 5 year OS: 38.9%)

Conclusions Compared with CRC detected by colonoscopy, PCCRCs are associated with a higher risk of emergent presentation, a lower likelihood of surgical resection and most notably, significantly worse oncological outcomes. However, they have better outcomes than patients with no recent colonoscopy.

  • COLORECTAL CARCINOMA
  • COLONOSCOPY

https://doi.org/10.1136/gutjnl-2014-308578

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Significance of this study

    What is already known on this subject?

    • Development of postcolonoscopy colorectal cancers (PCCRCs) is increasingly recognised.

    • Postcolonoscopy cancers may be a result of cancers that were missed or incompletely excised on initial colonoscopy, or developed de novo since the initial colonoscopy.

    What are the new findings?

    • PCCRCs present at a more advanced stage than colonoscopy-detected cancers and have inferior overall survival.

    • Patients with PCCRCs are more likely to present non-electively with their cancer and are less likely to undergo surgical resection.

    • Patients with PCCRCs have better outcomes than patients with no recent colonoscopy prior to diagnosis.

    How might it impact on clinical practice in the foreseeable future?

    • Increased focus on access, uptake and quality of colonoscopy is crucial to improving colorectal cancer outcomes at a population level.

    Introduction

    Colorectal cancer (CRC) is the second leading cause of cancer-related mortality in the USA and Canada.1 Colonoscopy is the gold standard for the diagnosis of CRC, and endoscopic screening is associated with a reduction in CRC incidence and mortality by the detection of premalignant lesions and cancers at an earlier stage.2 ,3 There is, however, increasing recognition of the potential for cancers developing post colonoscopy.4–8 These cancers may have newly developed since the colonoscopy, but may also have been ‘missed’ on colonoscopy. To date, little is known about the natural history and outcomes of patients with postcolonoscopy CRC (PCCRC).

    The present study was conducted to compare, in a population-based manner, patients with PCCRC, patients with colonoscopy-detected cancers and patients diagnosed with CRC with no recent prior colonoscopy. The primary objective was to determine if overall survival was compromised in patients with PCCRC. Secondary objectives included assessment of stage at diagnosis, the likelihood of patients with PCCRC having an emergent presentation related to their cancer, assessment of the type of treatment received by these patients and short-term outcomes related to treatment.

    Methods

    Study design

    The study was a retrospective population-based cohort study conducted using information from health databases. The study population included all individuals diagnosed with CRC from 1 January 2003 to 31 December 2009 in Ontario, Canada.

    Data sources

    We used four data sources:

    1. Canadian Institute for Health Information (CIHI) discharge abstract database: This database contains information on all patients discharged from hospital or same day surgery units in Ontario. Data elements include patient demographics, diagnostic and procedure codes (International Classification of Disease 10th revision, ICD-10) and discharge status.

    2. Ontario Health Insurance Plan (OHIP): This database contains information about all claims for physicians’ services provided to Ontario residents. As Ontario is a single-payer system and all residents are covered, this database includes virtually all healthcare services provided by physicians to Ontario patients (including endoscopy).

    3. Ontario Cancer Registry (OCR): This database has recorded all incident cancer cases in Ontario since 1964 and is estimated to be 95% complete.9 Complete stage information was available from 2007 to 2009.

    4. Registered Persons Database (RPDB): This database contains vital statistics on all residents of Ontario including demographics and vital status.

    All data sources were deterministically linked using encrypted unique patient identifiers that allow for longitudinal tracking of patients across all data sources.

    Study population

    From the OCR database, we identified all persons 20 years of age and older, who were newly diagnosed with CRC from 1 January 2003 to 31 December 2009 (ICD-9-CM codes 153.x excluding patients with appendix cancers, 154.0, 154.1). Patients with a history of IBD (ICD9 codes: 556.x, 555.x; ICD10 codes: K50.0, 50.1, 50.8, 50.9, 51.9) were excluded to minimise bias from diseases that may have a different progression to carcinoma than sporadic cases. Pathological diagnosis of CRC could be based on tissue obtained at the time of surgery, endoscopy or via percutaneous biopsy (in patients who did not undergo surgery or endoscopy), although the data sources do not allow these different methods to be distinguished.

    Using the OHIP database, we determined whether patients had undergone a colonoscopy (defined by OHIP code Z555) in the 36 months preceding diagnosis. The ‘index colonoscopy’ was defined as the first colonoscopy performed in the 36 months prior to the date of cancer diagnosis. Using similar methods and time windows from previously published research,7 ,8 ,10 patients were categorised into three groups based on the interval from the index colonoscopy to date of cancer diagnosis:

    1. Postcolonoscopy colorectal cancer (PCCRC): Patients whose index colonoscopy occurred 6–36 months prior to the date of cancer diagnosis. Since the index colonoscopy was defined as the first colonoscopy in the 36 month time window, patients who initially had a colonoscopy 6–36 months prior to diagnosis in which a cancer was not detected, and a repeat colonoscopy within 6 months of diagnosis at which the cancer was diagnosed would be categorised in the PCCRC group.

    2. DETECTED: Patients whose index colonoscopy preceded the date of cancer diagnosis by 6 months or less, with no other prior colonoscopy within the previous 36 months. This time frame is based on the assumption that virtually all patients who had a cancer detected by colonoscopy would have a definitive diagnosis registered in the OCR within 6 months.

    3. No recent colonoscopy (NOSCOPE): Patients who have not had a colonoscopy within the 36 months preceding cancer diagnosis.

    As cancer stage was only available from 2007 to 2009, we defined a subgroup for analysis of patients whose cancer diagnosis date was between 1 January 2007 and 31 December 2009.

    Definition of covariates

    Patient-related variables including age, sex, geographical location, income quintile and comorbidities were recorded. Comorbidities were identified using ICD-9 and ICD-10 diagnosis codes using a 5-year lookback window from the diagnosis date, and were aggregated into the Charlson-Deyo comorbidity score.11 Geographical location was categorised as urban versus rural based on census data, with rurality being defined as a community size of at most 10 000 people, a standard definition used in population-based research in Ontario. Site of the cancer within the colorectum was recorded and categorised as right colon, transverse colon, left colon (splenic flexure, descending colon and sigmoid colon) or rectum based on ICD-9-CM cancer diagnosis code recorded in the OCR. Missing or unspecified site within the colon was treated as a separate category.

    Definition of outcomes

    The primary outcome was overall survival, determined from the RPDB, with the last date of potential follow-up being 31 December 2011. Secondary outcomes were stage at diagnosis, incidence of non-elective presentation, proportion of patients treated with surgical resection, and postoperative morbidity and mortality. American Joint Committee on Cancer (AJCC) stage at diagnosis was only assessed for the subgroup of patients diagnosed from 2007 to 2009, and was directly obtained from the OCR database. Non-elective presentation was identified through the CIHI database using an identifier that classifies all hospital admissions as elective or urgent/emergent. Patients were classified as having a non-elective presentation if they had an urgent/emergent admission for CRC in the 6 months prior to the date of diagnosis and/or if their index admission included concomitant ICD-10 diagnosis codes for obstruction (K56.6), perforation (K63.1) or peritonitis (K65.0, K65.9). Surgical treatment was categorised as ‘no surgical treatment’, ‘stoma only’, ‘resection and stoma’ or ‘resection without stoma’ based on ICD-10 Canadian Classification of Health Interventions (CCI) procedure codes in the CIHI database. Patients who underwent a resection and underwent stoma formation on separate occasions were categorised as ‘resection and stoma’. Surgical treatment up to 6 months post diagnosis was considered related to the diagnosis of cancer. Postoperative mortality was defined in the standard manner as death within 30 days or inhospital during the same admission as the surgical procedure and was identified from RPBD and CIHI.

    Analyses

    Descriptive statistics for the study population were calculated for the entire population and stratified by patient category (PCCRC, DETECTED, NOSCOPE). Continuous variables were analysed using the t test and categorical variables using the χ2 test. For the primary outcome (overall survival), Kaplan-Meier analysis was used to perform univariate survival analysis, and Cox proportional hazards were used to perform multivariable analyses. For the 2007–2009 subgroup, survival analysis was first performed using the same variables as the full cohort and then repeated with the addition of the stage variable. The analyses on this subgroup were then repeated after stratifying the patients by stage (stages 0, I, II vs stage III vs stage IV). Patients with missing stage information were excluded from this subanalysis. Patients were censored if they had not experienced an event at the time of last possible follow-up in the RPDB. Univariate and multivariate logistic regressions were used to examine the association between patient groups (PCCRC, DETECTED, NOSCOPE) and non-elective presentation, receipt of surgical resection and postoperative morbidity. Covariates included in multivariate models included age (continuous variable), sex, rurality, Charlson-Deyo score (categorised as 0, 1, 2, 3, 4+) and cancer site (right, transverse, left, unspecified colon, rectum). All statistical tests were two-sided, with p values <0.05 considered statistically significant. All analyses were conducted using SAS V.9.3 (Cary, North Carolina, USA) at the Institute for Clinical Evaluative Sciences, which is a Section 45(1) prescribed entity in Ontario's Personal Health Information Protection Act.

    Results

    Overall, 45 104 patients were diagnosed with CRC during the study period, of which 2804 (6.2%) were in the PCCRC group, 27 671 (61.3%) were in the DETECTED group and 14 629 (32.4%) were in the NOSCOPE group. Baseline characteristics of the study population are shown in table 1. The median time from index colonoscopy to diagnosis of PCCRC was 20 months. In the PCCRC group, a polypectomy was performed in 40.1% of patients at the time of the index colonoscopy. Colon cancers were more common in the PCCRC and NOSCOPE groups than in the DETECTED group (p<0.001 for both pairwise comparisons). Compared with patients in the DETECTED group, patients in the PCCRC group were more likely to be of older age (p<0.001), female sex (p<0.001), rural residence (p=0.023) and higher Charlson comorbidity score (p<0.001). Compared with the DETECTED and PCCRC groups, the NOSCOPE group had a lower proportion of rural patients (p<0.001 for both comparisons) and patients in the highest income quintile (p<0.001 for both comparisons). The median follow-up among all patients (including non-survivors) was 3 years.

    View this table:
    Table 1

    Baseline characteristics of the entire cohort and categorised by DETECTED, PCCRC or NOSCOPE

    There was a statistically significant difference in the stage at diagnosis between the DETECTED, PCCRC and NOSCOPE groups (table 2, p<0.001). Compared with the DETECTED group, the PCCRC group had a similar proportion of patients diagnosed with an early (stage 0–II) CRC (53.6% vs 56.0%, p=0.22). Conversely, patients in the PCCRC group had a 33% relative increase in the risk of being diagnosed with stage IV disease compared with DETECTED patients (17.2% vs 12.9%, absolute difference: 4.3%, p<0.001). Patients in the NOSCOPE group had the highest proportion of patients with stage IV disease at diagnosis (37.1%, p<0.001 vs DETECTED and PCCRC).

    View this table:
    Table 2

    Stage at diagnosis of study cohort (2007–2009 subgroup)

    Overall survival

    A total of 19 810 patients died in the study period, 8982 (32.5%) in the DETECTED group, 1185 in the PCCRC group (42.3%) and 9643 in the NOSCOPE group (65.9%). Figure 1 shows actuarial survival curves of the three groups. Five-year overall survival was significantly different among the three groups (DETECTED: 68.3%; PCCRC: 60.8%; NOSCOPE: 38.9%, p<0.001). After adjusting for covariates, patients in the PCCRC group had a significantly worse survival than patients in the DETECTED group (adjusted HR: 1.25, 95% CI 1.17 to 1.32, p<0.001, table 3) and patients in the NOSCOPE group had a significantly worse survival than the PCCRC and DETECTED groups (adjusted HR: 2.57 relative to DETECTED, 95% CI 2.49 to 2.65, p<0.001).

    View this table:
    Table 3

    Multivariable survival analysis (Cox regression) examining overall survival

    Figure 1
    Figure 1

    Kaplan-Meier survival curve comparing the overall survival of the DETECTED, postcolonoscopy colorectal cancer (PCCRC) and NOSCOPE groups.

    In the 2007–2009 subgroup, the adjusted HR (in the model without stage) for the PCCRC group was 1.33 (95% CI 1.20 to 1.47) and for the NOSCOPE group was 3.03 (95% CI 2.88 to 3.19). When stage was added to the model, the adjusted HR for PCCRC was 1.14 (95% CI 1.02 to 1.27, p=0.024) and for the NOSCOPE group was 1.93 (95% CI 1.83 to 2.04, p<0.0001). Subgroup analyses were performed stratifying the groups by early stage (stages 0, I, II), intermediate stage (stage III) and advanced stage (stage IV). The adjusted HR for PCCRC in each strata was similar in magnitude to the non-stratified analysis (early stage: HR 1.15, 95% CI 0.93 to 1.42, p=0.19; intermediate stage: HR 1.11, 95% CI 0.90 to 1.37, p=0.33; late stage: HR 1.13, 95% CI 0.95 to 1.34, p=0.15). The adjusted HR for the NOSCOPE group in early stage patients was 2.35 (95% CI 2.10 to 2.62, p<0.0001), in intermediate stage patients was 1.90 (95% CI 1.71 to 2.12, p<0.0001) and in late stage patients was 1.77 (95% CI 1.63 to 1.91, p<0.0001).

    Non-elective presentation

    During the study period, 13 076 patients had a non-elective presentation related to CRC within the 6 months preceding diagnosis. Patients in the PCCRC group were significantly more likely to have an emergency presentation compared with patients in the DETECTED group (p<0.001), and patients in the NOSCOPE group had the highest incidence of non-elective presentations (DETECTED: 9.7%; PCCRC: 27.0%, NOSCOPE: 65.4%, p<0.001 for DETECTED vs NOSCOPE, p=0.007 for PCCRC vs NOSCOPE).

    Surgical treatment

    In the DETECTED group, 84.1% of patients underwent surgical resection, whereas the PCCRC group (76.1%, absolute difference: 8.0%) and NOSCOPE groups (63.2%, absolute difference: 20.9%) had significantly lower rates of surgical resection (p<0.001 for DETECTED vs PCCRC and DETECTED vs NOSCOPE). There was no between-group difference in the number of patients treated with a stoma or intestinal bypass without resection. In the DETECTED group, 13.9% of patients received no surgical treatment compared with 21.9% in the PCCRC group and 31.0% in the NOSCOPE group. After adjusting for covariates, patients in the PCCRC group were significantly less likely to receive a surgical resection than patients in the DETECTED group (adjusted OR: 0.65, 95% CI 0.59 to 0.72, p<0.001), and patients in the NOSCOPE group were the least likely to receive a surgical resection (adjusted OR: 0.36, 95% CI 0.34 to 0.38, p<0.001). The overall postoperative mortality rate was 3.5% (DETECTED: 2.2%, PCCRC: 3.1%, NOSCOPE: 6.7%, p<0.001). In multivariable analysis, there was no statistically significant difference between the DETECTED and PCCRC groups (adjusted OR: 1.01, 95% CI 0.77 to 1.31, p=0.97), but the NOSCOPE group had a significantly higher perioperative mortality (OR: 1.93, 95% CI 1.71 to 2.19, p<0.0001).

    Discussion

    Although colonoscopy is the gold standard for the diagnosis of CRC, there is increasing recognition of the development of PCCRCs. The incidence of PCCRC in the literature is variable, due in part to differences in the method of calculation.12 The underlying aetiology of PCCRCs may be related to cancers that were ‘missed’ at the index colonoscopy, incompletely resected dysplastic polyps, or tumours that are biologically more rapidly progressive and have developed de novo shortly after colonoscopy. Some studies have suggested that there may be some inherent biological differences between PCCRCs and screen-detected cancers.13–16 A recent study by le Clercq et al17 identified all PCCRCs at three large hospitals in the Netherlands and adjudicated their probable aetiology on a case-by-case basis using predefined criteria. They suggested that the minority (13%) were de novo cancers, while the remainder were missed cancers (78%) or incompletely resected dysplastic polyps (9%). A similar study by Robertson et al18 suggested that approximately 24% of PCCRCs were likely de novo cancers. Although there is ample literature on factors contributing to missed cancers, there is little known to date about the outcomes of these patients.

    In this study, we found that patients with PCCRC were three times more likely to present to the hospital emergently, and had significantly worse overall survival than patients with a DETECTED cancer with a 25% increase in their hazard of death and an absolute survival difference of 7.5% at 5 years. Correspondingly, patients with PCCRC had an absolute increase of 4.3% in the diagnosis of stage IV disease compared with DETECTED patients, representing a 33% relative increase, and were less likely to undergo surgical resection of their cancer. The more advanced stage at diagnosis was found to explain a portion of the decreased overall survival associated with PCCRC. Nonetheless, even after adjusting for stage, PCCRC was still associated with a worse overall survival. In stage-specific subgroup analyses, similar adjusted HRs were obtained for each stage as the non-stratified analysis, although they did not reach statistical significance within individual strata, likely due to the smaller sample sizes. This could suggest that some of the survival difference between PCCRC and DETECTED cancers may be attributable to differences in biological behaviour as well.

    In this study, we were unable to determine the indication for the colonoscopies, a possible confounding factor. Theoretically, if the DETECTED group had a higher frequency of screening colonoscopies than the index colonoscopies in the PCCRC group, this could be responsible for some of the differences seen in stage and survival. However, there are no data in the literature to suggest that the DETECTED group should have more colonoscopies performed for asymptomatic indications (ie, screening or surveillance) than the index colonoscopy in the PCCRC group. Indeed, the reverse is likely true as asymptomatic cancers are more likely to be smaller and thus more likely to be missed on initial colonoscopy and to present as a PCCRC. Corley et al19 found that screening or surveillance colonoscopies were more prevalent in the interval cancer group than the rest of the colonoscopy cohort. Additionally, differences in oncological outcomes were seen with and without correction for differences in stage, which is the principal benefit of screening. Therefore, an imbalance in screening colonoscopies between groups is unlikely to fully explain the study findings.

    Few studies in the literature have examined the clinical outcomes associated with PCCRCs. A population-based study in Denmark compared patients with interval cancers (index colonoscopy between 1 year and 5 years before CRC diagnosis) with patients with remote colonoscopies (≥10 years ago) or no colonoscopies prior to CRC diagnosis.20 No survival differences were noted, implying that interval cancers may not be more aggressive than sporadic CRCs. However, CRCs that were diagnosed by recent colonoscopies (ie, non-interval cancers) were not studied and therefore, the results are not directly comparable with the present one. A population-based study conducted in Manitoba, Canada, found no survival difference between detected and interval cancers.6 In contrast, a study of the UK National Health Service Bowel Cancer Screening Programme that compared screen-detected cancers (based on positive faecal occult blood testing) against non-screen-detected cancers including interval cancers, did find a significant survival difference between the two groups.21 Although a statistical comparison between interval cancers and screen-detected cancers is not specifically reported, the respective survival curves suggest that they may be associated with a worse survival. The sample size of interval cancers in that study was relatively small and the definition of interval cancers included CRCs diagnosed after any negative screening test (including faecal occult blood testing only without colonoscopy) and before the next round of screening. In the present study, we included only patients that had undergone an index colonoscopy within a strict time window (within 6–36 months of index colonoscopy) in the definition of PCCRC, which reduces the heterogeneity in this group by reducing the inclusion of sporadic cancers in this group. Our study is also the largest to date examining the outcomes of PCCRCs, and its results are readily generalisable owing to its population-based design.

    Although PCCRCs had worse outcomes than DETECTED patients in our study, they had markedly better outcomes than patients who had not had a recent colonoscopy at all, with improved surgical resection rates, decreased emergency presentations, lower stage at diagnosis and improved overall survival compared with this group. There are several possible reasons underlying this finding. A portion of the improved outcomes is explained by the higher incidence of late-stage cancers and emergency presentations in the group with no colonoscopy compared with the PCCRC group. This suggests that PCCRCs that developed as a result of being ‘missed’ were likely to be small and at an early stage in their carcinogenesis trajectory (including missed or incompletely excised precursor lesions) at the time of their colonoscopy. Interval growth of these missed cancers likely resulted in only a small degree of progression and upstaging, in keeping with the relatively slow sojourn of sporadic CRC. We also found that despite adjusting for stage and patient factors including comorbidities, patients with no recent colonoscopy had worse survival than PCCRC and DETECTED cancers. This may be a result of a ‘healthy user’ bias, wherein patients who choose to undergo colonoscopy for either screening or symptom evaluation are also more likely to have better health outcomes in general.22

    The present study has certain limitations. Although the time frame used in the study to define PCCRC has been used previously by our group and others in the literature, we recognise that there is no universally accepted definition of PCCRC, and this can lead to variable PCCRC rates.12 In this study, we did not aim to determine the incidence of PCCRCs, but rather to assess their outcomes. It is possible that PCCRCs may be detected outside of the defined time window. If a PCCRC was diagnosed more than 36 months after index colonoscopy, it would be excluded from our definition; however, the 36 months time frame chosen for this study is based on the mean sojourn time for CRC23–25 and extending the time frame beyond 36 months would increase the contamination of the PCCRC group with new sporadic cancers. Misclassification of PCCRC as DETECTED may occur in a situation when two colonoscopies are both performed within 6 months of diagnosis, with the cancer not detected by the first one but detected by the second one. However, any bias from PCCRCs being misclassified as DETECTED would only serve to narrow any potential differences between groups; therefore the significant survival difference we have demonstrated may be even larger. The retrospective design of the study and the use of health databases allows for the potential for bias from unmeasured variables. One such variable for which there is a lack of information is, as discussed earlier, the exact indication for colonoscopy. We were also not able to assess factors associated with the quality of the index colonoscopy such as the adequacy of bowel preparation and the withdrawal time. However, the purpose of this study was not to determine the effectiveness of colonoscopic screening per se, or the reasons underlying why a colonoscopy did not detect a cancer. Rather, this study was designed to determine the implications for patients of a cancer not detected by a previous colonoscopy, regardless of the indication for the colonoscopy or the reason the cancer was not diagnosed. We also did not have information on the prevalence of hereditary or familial CRC syndromes in the different groups, though we were able to exclude patients with IBD. However, syndromes known to accelerate the adenoma-carcinoma sequence such as Lynch syndrome make up less than 5% of all CRCs26 and therefore, are unlikely to fully explain the results seen.

    In summary, we found that PCCRCs were associated with significantly worse patient outcomes than cancers detected at colonoscopy, with a higher likelihood of emergency presentation, a lower likelihood of surgical resection, higher proportion of late-stage disease and most notably a significantly worse overall survival. However, patients with PCCRC do have markedly better outcomes than patients with no recent colonoscopy. These findings reinforce the critical importance of studying and improving the uptake, access and quality of colonoscopy in detecting CRC.

    References

    1. ↵
      1. Siegel R,
      2. Naishadham D,
      3. Jemal A
      . Cancer statistics, 2013. CA Cancer J Clin 2013;63:11–30. doi:10.3322/caac.21166
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Atkin WS,
      2. Edwards R,
      3. Kralj-Hans I, et al
      . Once-only flexible sigmoidoscopy screening in prevention of colorectal cancer: a multicentre randomised controlled trial. Lancet 2010;375:1624–33. doi:10.1016/S0140-6736(10)60551-X
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Schoen RE,
      2. Pinsky PF,
      3. Weissfeld JL, et al
      . Colorectal-cancer incidence and mortality with screening flexible sigmoidoscopy. N Engl J Med 2012;366: 2345–57. doi:10.1056/NEJMoa1114635
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Cooper GS,
      2. Xu F,
      3. Barnholtz Sloan JS, et al
      . Prevalence and predictors of interval colorectal cancers in medicare beneficiaries. Cancer 2012;118: 3044–52. doi:10.1002/cncr.26602
      OpenUrlCrossRefPubMed
    5. ↵
      1. Rex DK,
      2. Rahmani EY,
      3. Haseman JH, et al
      . Relative sensitivity of colonoscopy and barium enema for detection of colorectal cancer in clinical practice. Gastroenterology 1997;112:17–23. doi:10.1016/S0016-5085(97)70213-0
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Singh H,
      2. Nugent Z,
      3. Demers AA, et al
      . Rate and predictors of early/missed colorectal cancers after colonoscopy in Manitoba: a population-based study. Am J Gastroenterol 2010;105:2588–96. doi:10.1038/ajg.2010.390
      OpenUrlCrossRefPubMed
    7. ↵
      1. Bressler B,
      2. Paszat LF,
      3. Chen Z, et al
      . Rates of new or missed colorectal cancers after colonoscopy and their risk factors: a population-based analysis. Gastroenterology 2007;132:96–102. doi:10.1053/j.gastro.2006.10.027
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Baxter NN,
      2. Sutradhar R,
      3. Forbes SS, et al
      . Analysis of administrative data finds endoscopist quality measures associated with postcolonoscopy colorectal cancer. Gastroenterology 2011;140:65–72. doi:10.1053/j.gastro.2010.09.006
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Robles SC,
      2. Marrett LD,
      3. Clarke EA, et al
      . An application of capture-recapture methods to the estimation of completeness of cancer registration. J Clin Epidemiol 1988;41:495–501. doi:10.1016/0895-4356(88)90052-2
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Baxter NN,
      2. Goldwasser MA,
      3. Paszat LF, et al
      . Association of colonoscopy and death from colorectal cancer. Ann Intern Med 2009;150:1–8. doi:10.7326/0003-4819-150-1-200901060-00306
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Deyo RA,
      2. Cherkin DC,
      3. Ciol MA
      . Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–19. doi:10.1016/0895-4356(92)90133-8
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Morris EJA,
      2. Rutter MD,
      3. Finan PJ, et al
      . Post-colonoscopy colorectal cancer (PCCRC) rates vary considerably depending on the method used to calculate them: a retrospective observational population-based study of PCCRC in the English National Health Service. Gut 2015;64:124856. doi:10.1136/gutjnl-2014-308362
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Shaukat A,
      2. Arain M,
      3. Anway R, et al
      . Is KRAS mutation associated with interval colorectal cancers? Dig Dis Sci 2012;57:913–7. doi:10.1007/s10620-011-1974-6
      OpenUrlCrossRefPubMed
    14. ↵
      1. Arain MA,
      2. Sawhney M,
      3. Sheikh S, et al
      . CIMP status of interval colon cancers: another piece to the puzzle. Am J Gastroenterol 2010;105:1189–95. doi:10.1038/ajg.2009.699
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Sawhney MS,
      2. Farrar WD,
      3. Gudiseva S, et al
      . Microsatellite instability in interval colon cancers. Gastroenterology 2006;131:1700–5. doi:10.1053/j.gastro.2006.10.022
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Nishihara R,
      2. Wu K,
      3. Lochhead P, et al
      . Long-term colorectal-cancer incidence and mortality after lower endoscopy. N Engl J Med 2013;369:1095–105. doi:10.1056/NEJMoa1301969
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. le Clercq CMC,
      2. Bouwens MWE,
      3. Rondagh EJA, et al
      . Postcolonoscopy colorectal cancers are preventable: a population-based study. Gut 2014;63:957–63. doi:10.1136/gutjnl-2013-304880
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Robertson DJ,
      2. Lieberman DA,
      3. Winawer SJ, et al
      . Colorectal cancers soon after colonoscopy: a pooled multicohort analysis. Gut 2014;63:949–56. doi:10.1136/gutjnl-2012-303796
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Corley DA,
      2. Jensen CD,
      3. Marks AR, et al
      . Adenoma detection rate and risk of colorectal cancer and death. N Engl J Med 2014;370:1298–306. doi:10.1056/NEJMoa1309086
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Erichsen R,
      2. Baron JA,
      3. Stoffel EM, et al
      . Characteristics and survival of interval and sporadic colorectal cancer patients: a nationwide population-based cohort study. Am J Gastroenterol 2013;108:1332–40. doi:10.1038/ajg.2013.175
      OpenUrlCrossRefPubMed
    21. ↵
      1. Gill MD,
      2. Bramble MG,
      3. Rees CJ, et al
      . Comparison of screen-detected and interval colorectal cancers in the Bowel Cancer Screening Programme. Br J Cancer 2012;107:417–21. doi:10.1038/bjc.2012.305
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Shrank WH,
      2. Patrick AR,
      3. Brookhart MA
      . Healthy user and related biases in observational studies of preventive interventions: a primer for physicians. J Gen Intern Med 2011;26:546–50. doi:10.1007/s11606-010-1609-1
      OpenUrlCrossRefPubMed
    23. ↵
      1. Chen TH,
      2. Yen MF,
      3. Lai MS, et al
      . Evaluation of a selective screening for colorectal carcinoma: the Taiwan Multicenter Cancer Screening (TAMCAS) project. Cancer 1999;86:1116–28. doi:10.1002/(SICI)1097-0142(19991001)86:7<1116::AID-CNCR4>3.0.CO;2-D
      OpenUrlCrossRefPubMed
    24. ↵
      1. Jouve JL,
      2. Remontet L,
      3. Dancourt V, et al
      . Estimation of screening test (Hemoccult) sensitivity in colorectal cancer mass screening. Br J Cancer 2001;84:1477–81. doi:10.1054/bjoc.2001.1752
      OpenUrlCrossRefPubMed
    25. ↵
      1. Brenner H,
      2. Altenhofen L,
      3. Katalinic A, et al
      . Sojourn time of preclinical colorectal cancer by sex and age: estimates from the German national screening colonoscopy database. Am J Epidemiol 2011;174:1140–6. doi:10.1093/aje/kwr188
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Burt RW
      . Colon cancer screening. Gastroenterology 2000;119:837–53. doi:10.1053/gast.2000.16508
      OpenUrlCrossRefPubMedWeb of Science

    Footnotes

    • Some data from this paper was presented in abstract form at the Gastrointestinal Symposium of the American Society for Clinical Oncology (ASCO-GI) in January 2014.

    • Correction notice This article has been corrected since it was published Online First. The provenance and peer review statement has been corrected.

    • Contributors Study concept and design, analysis and interpretation of data, critical revision of manuscript for important intellectual content—AG, LR, JT, LFP, NNB. Drafting of the manuscript—AG, LR, NNB. Statistical analysis—LY.

    • Funding JT holds a New Investigator Award from the Canadian Institutes of Health Research. This study was supported by the Institute for Clinical Evaluative Sciences, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care. The opinions, results and conclusions reported in this paper are those of the authors and are independent from the funding sources. No endorsement by the Institute for Clinical Evaluative Sciences or the Ontario Ministry of Health and Long-Term Care is intended or should be inferred.

    • Competing interests None.

    • Ethics approval Research Ethics Board at Sunnybrook Health Sciences Centre.

    • Provenance and peer review Not commissioned; externally peer reviewed.