Colorectal cancer in patients with single versus double positive faecal immunochemical test results: A retrospective cohort study

BACKGROUND Screening for colorectal cancer (CRC) using the faecal immunochemical test (FIT) is widely advocated. Few studies have compared the rate of detecting colonoscopic pathologies in single compared to double FIT-positive follow-up colonoscopy-compliant individuals in a two-sample national FIT screening program. OBJECTIVE To compare CRC incidence in double FIT-positive versus single FIT-positive individuals using a retrospective cohort of patients from a tertiary hospital in Singapore. DESIGN Retrospective cohort study. SETTING Data was extracted from one regional acute hospital in Singapore. PARTICIPANTS 1,539 FIT-positive individuals from the national FIT screening program who were referred to the hospital from 1st January 2017 to 31st September 2019 for follow-up consultation and diagnostic colonoscopy. MEASUREMENTS The exposure of interest was a positive result on both FIT kits. The main outcome was a follow-up diagnostic colonoscopy finding of CRC. The secondary outcome was a diagnostic colonoscopy finding of a colorectal polyp. RESULTS Incidence density of CRC was 1.53 (95% CI = 0.61, 3.15) and 17.88 (95% CI = 11.67, 26.19) per 100,000 person-months, in the single and double FIT-positive group, respectively. This resulted in an incidence rate ratio of 11.71 (95% CI = 5.25, 29.08). Colorectal polyp detection was significantly higher (p < 0.01) in the double (107 of 157 participants; 68.2%) compared to the single (310 of 585 participants; 53.0%) FIT-positive group. LIMITATIONS The key limitation of this study was the relatively small cohort derived from a single regional hospital, as this had the effect of limiting the number of incident cases, resulting in comparatively imprecise CIs. CONCLUSIONS Double FIT-positive individuals are significantly more likely to have a colonoscopy finding of incident CRC or premalignant polyp than single FIT-positive individuals. Clinicians and policymakers should consider updating their CRC screening protocols accordingly.


SETTING
Data was extracted from one regional acute hospital in Singapore.

RESULTS
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 14, 2020. per 100,000 person-months, in the single and double FIT-positive group, respectively. This resulted in an incidence rate ratio of 11.71 (95% CI = 5.25, 29.08). Colorectal polyp detection was significantly higher (p < 0.01) in the double (107 of 157 participants; 68.2%) compared to the single (310 of 585 participants; 53.0%) FIT-positive group.

LIMITATIONS
The key limitation of this study was the relatively small cohort derived from a single regional hospital, as this had the effect of limiting the number of incident cases, resulting in comparatively imprecise CIs.

CONCLUSIONS
Double FIT-positive individuals are significantly more likely to have a colonoscopy finding of incident CRC or premalignant polyp than single FIT-positive individuals. Clinicians and policymakers should consider updating their CRC screening protocols accordingly.

FUNDING SOURCE:
This study was supported by the Singapore Population Health Improvement Centre (SPHERiC) [NMRC/CG/C026/2017_NUHS]. The funders had no role in the study design, execution, analyses, interpretation of the data, or decision to submit results. We confirm that all authors had full access to all of the data (including statistical reports and tables) in the study and can take responsibility for the integrity and accuracy of the data analysis.
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Singapore has adopted two-sample FIT in its national CRC screening program as this has been shown to increase sensitivity by potentially detecting neoplasms that may have been missed in one-sample FIT (3,4). The national CRC screening program advises age-eligible Singapore residents to complete a free two-sample FIT kit screening every year (3). Should either sample be positive, the individual would be contacted by a dedicated hospital coordinator to arrange for consultation and colonoscopy with a colorectal surgeon or gastroenterologist (5).
Individuals with two positive FIT samples ("double FIT-positive") -which would have been sampled over two days -should intuitively be more likely to have colonic pathologies compared to individuals with only one positive FIT sample ("single FIT-positive"). However, few studies have examined these outcomes (4,6). Thus, this study aimed to compare CRC incidence in double FIT-positive versus single FIT-positive individuals using a retrospective cohort of patients from a tertiary hospital in Singapore.

Ethical approval
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 14, 2020.

Patient and public involvement statement
This research was done without patient involvement. Patients were not invited to comment on the study design and were not consulted to develop patient relevant outcomes or interpret the results. Patients were not invited to contribute to the writing or editing of this document for readability or accuracy.

Study design and participants
The study population (N = 1,641) was a retrospective cohort comprising all individuals who

Primary exposure and outcomes of interest
The exposure of interest was defined as a positive result on both FIT kits. The main outcome (incident CRC cases) was defined as a follow-up diagnostic colonoscopy finding of a CRC, regardless of staging. The secondary outcome was defined as a follow-up diagnostic . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 14, 2020. . https://doi.org/10.1101/2020.05.11.20097881 doi: medRxiv preprint colonoscopy finding of a colorectal polyp. Colonoscopy reports were verified by using the hospital's electronic medical records system

Other participant factors
Apart from the main exposure and outcome of interest, our data included participants' ethnicity (defined as Chinese, Malay, Indian, or Others), age, date of two-sample FIT kit collection, date of referral to NUH, and date of follow-up medical consultation and colonoscopy.

Statistical analysis
For demographics, frequencies and proportions were used to report ethnicity and compliance to follow-up colonoscopy. Median and interquartile range were used to summarise participant age. For colonoscopy-compliant participants, median and interquartile range was used to report follow-up time (in days) between date of referral to the hospital and date of colonoscopy. Incident CRC cases were summarised using frequencies and proportions.
To examine possible effects of confounding, chi-square tests were used to compare ethnicity and colonoscopy compliance, and Mann-Whitney U tests for age and follow-up time, between single and double FIT-positive groups.
A chi-square test was used to compare incident CRC cases and polyp detection between the groups. Incidence density (ID) of CRC cases was derived using person-months at risk, calculated from month of birth to the month of colonoscopy. Incidence rate ratio (IRR) was used to compare relative risk of CRC between the single FIT-positive and double FITpositive groups. Confidence intervals (95% CIs) were reported for both ID and IRR.

RESULTS
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 14, 2020. . https://doi.org/10.1101/2020.05.11.20097881 doi: medRxiv preprint The median age of the sample was 65 years (IQR = 58 -71 years). Overall compliance to follow-up colonoscopy was 50.4%. The distribution of sample demographics, colonoscopy compliance, and outcomes of interest between single and double FIT-positive patients can be found in Table 1.  is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 14, 2020. . https://doi.org/10.1101/2020.05.11.20097881 doi: medRxiv preprint At the national level, countries that are still advocating CRC screening using only one-sample FIT should reconsider their approach. Double FIT-positive is likely to have increased specificity, resulting in fewer false positive findings during follow-up (4,6). Recent findings have also highlighted the significantly higher positive predictive value of double FIT-positive compared to single FIT-positive in detecting CRC (4). Although compliance rates of individuals performing two-sample FIT over two separate days are likely to be slightly lower than one-sample FIT, this can be managed through public education and targeted interventions (6). Countries concerned about costs to the health system should consider performing a cost-effectiveness analysis, which will likely favour a two-sample FIT program (6).
More work is also required to reinforce the importance of prompt follow-up after a positive FIT result. As observed in our study and the literature, colonoscopy compliance rate after a positive FIT is approximately 50% (7). Using simple arithmetic, another 33 CRC cases could have been diagnosed and managed promptly should there have been perfect compliance.
Prior literature has suggested a combination of patient and provider factors influencing follow-up compliance (7). As these have been known to vary between populations, public health professionals and policymakers must understand population-specific barriers and facilitators in order to develop tailored, cost-effective interventions.
The key strengths of this study were (I) the representativeness of the sample to the averagerisk study population recommended to our hospital via the national screening programme, with a loss to follow up under 10%, and (II) the ability to verify outcomes of interest via the hospital's clinical database. The key limitation of this study was the relatively small cohort derived from a single regional hospital, as this had the effect of limiting the number of incident cases, resulting in comparatively imprecise CIs. Our findings should nonetheless be representative of national demographics as FIT-positive hospital referrals are performed . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 14, 2020. . https://doi.org/10.1101/2020.05.11.20097881 doi: medRxiv preprint geographically, and population sociodemographic characteristics should be similarly distributed across Singapore (8). Although there is certainly the need to validate our findings in larger two-sample FIT cohorts, the significant difference in the rates of CRC detected should prompt clinicians and policymakers to consider examining this issue with greater urgency.
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 14, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 14, 2020. . https://doi.org/10.1101/2020.05.11.20097881 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 14, 2020. . https://doi.org/10.1101/2020.05.11.20097881 doi: medRxiv preprint