Faecal immunochemical tests versus colonoscopy for post-polypectomy surveillance: an accuracy, acceptability and economic study

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 09/22/192. The contractual start date was in January 2011. The draft report began editorial review in August 2017 and was accepted for publication in March 2018. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Wendy Atkin and Amanda J Cross report grants from the National Institute for Health Research (NIHR) Health Technology Assessment (HTA) programme, grants from Cancer Research UK (Population Research Committee – Programme Award C8171/A16894) and non-financial support from Eiken Chemical Co. Ltd (Tokyo, Japan) (MAST is UK distributor) during the conduct of the study. Stephen Morris is a member of the NIHR Health Services and Delivery Research funding board. Sheena Pearson, Carolyn Piggott and Julia Snowball all report grants from the NIHR HTA programme during the conduct of the study.

Disclaimer

This is a summary of independent research funded by the NIHR HTA programme and the Bobby Moore Fund for Cancer Research UK. The views expressed are those of the authors and not necessarily those of the NHS, NIHR, the Department of Health and Social Care or Cancer Research UK. Infractructure support for this work was provided by the NIHR Imperial Biomedical Research Centre.

Copyright statement

© Queen’s Printer and Controller of HMSO 2019. This work was produced by Atkin et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2019 Queen’s Printer and Controller of HMSO

Colorectal cancer screening

Screening enables the early detection of CRC and endoscopic removal of adenomas and is highly effective at reducing CRC mortality rates. 9–15 In England, the NHS Bowel Cancer Screening Programme (BCSP) offers CRC screening to men and women aged between 60 and 74 years. Every 2 years, patients are invited to complete a stool test, which is currently the guaiac faecal occult blood test (gFOBT). Among the first million people tested by the BCSP, 2% (21,106/1,079,293) had an abnormal gFOBT result using a three-test-kit algorithm, of whom 83% (17,518/21,106) had a follow-up colonoscopy or other investigative examination. 16 Adoption of an additional screening modality, flexible sigmoidoscopy at age 55 years, began in 2013. 17

Adenoma surveillance

Adenomas detected through screening, surveillance or among patients presenting symptomatically, are typically removed by polypectomy during colonoscopy, flexible sigmoidoscopy or surgery. Following adenoma removal, a proportion of people remain at increased risk of developing advanced adenomas (AAs) (i.e. sized ≥ 10 mm, tubulovillous or villous histology or high-grade dysplasia) or CRC, and are recommended to undergo a surveillance colonoscopy at a later date. 18–20 The risk of developing AAs or CRC following polypectomy is dependent on the characteristics of the baseline colonoscopy, such as completeness and quality of bowel preparation, and on the characteristics of adenomas found at baseline, including number of adenomas, size and pathology. 18–20 Based on the characteristics of adenomas found at baseline, the UK adenoma surveillance guideline (UK-ASG) defines three risk groups with consequent recommendations for surveillance colonoscopy following polypectomy (Figure 1). 21

FIGURE 1.

The UK-ASG. 21–23 a, Other considerations: age, comorbidity, family history, accuracy and completeness of examination. Reproduced from surveillance guidelines after removal of colorectal adenomatous polyps, Atkins and Saunders, vol. 51, pp. v6–9, 2002,21 with permission from BMJ Publishing Group Ltd.

Reproduced from Surveillance guidelines after removal of colorectal adenomatous polyps, Atkins and Saunders, 51, pp. v6–9, 2002,21 with permission from BMJ Publishing Group Ltd.

Colonoscopy in post-polypectomy surveillance

Colonoscopy is the most sensitive examination for CRC and adenomas. However, there are a number of problems with its use as a surveillance method.

Colonoscopy is an expensive procedure because of the requirement for skilled endoscopists and the use of sedation. The increased detection of adenomas requiring colonoscopy surveillance from the BCSP is reaching the point at which demand is overwhelming the available endoscopy workforce. 29 Currently, post-polypectomy surveillance accounts for approximately 20% of colonoscopies in the UK, and this figure will inevitably rise as more people enter surveillance. 30

There are disadvantages to colonoscopy for the patient undergoing the examination. Bowel preparation is typically unpleasant and the colonoscopy itself can be uncomfortable. 31,32 There is a small risk of serious complications (e.g. severe bleeding and colon perforation) and that risk increases with age. 33 Attendance for colonoscopy after a positive gFOBT is poor (around 80%), possibly because of procedural anxiety, in addition to other patient-reported barriers such as fear of finding cancer and anticipated pain. 16,34–36

Colonoscopy is not 100% sensitive and sometimes lesions are missed. Studies suggest that > 50% of post-colonoscopy CRCs result from missed lesions. 37 Delayed diagnosis of CRC occurring as a result of the surveillance interval could be detrimental to patient outcomes. 38

The yield of CRC and AAs at each surveillance colonoscopy is low. Even in higher-risk groups, the yield is < 20%, meaning that > 80% of colonoscopies will either be negative or detect only small adenomas of low malignant potential. 18,19,39,40 Negative colonoscopies provide no therapeutic benefit other than reassurance while contributing to cost, discomfort and risk.

For these reasons, a different, more cost-effective, method of surveillance following adenoma removal is urgently required. The faecal immunochemical test (FIT) for occult blood in stool may be an effective alternative.

Faecal immunochemical tests and guaiac faecal occult blood tests

Many screening programmes use gFOBT. In randomised controlled trials (RCTs), CRC mortality rates were reduced by 15%, on average, in people offered the gFOBT, and by 25% in those accepting the gFOBT. 13,41,42 However, an alternative stool test, FIT, which is due to replace gFOBT in the BCSP, has many advantages over gFOBT. 43,44

The FIT uses antibodies raised against the globin component of haemoglobin, whereas gFOBT is based on a chemical reaction involving haem. Unlike gFOBT, FIT is not subject to interference from dietary haemoglobin in red meat or peroxidases in vegetables. 45 Reduced dietary interference with FIT enables its increased analytical sensitivity to be exploited without diminution of analytical specificity. Furthermore, FIT is less susceptible than gFOBT to false positives from upper gastrointestinal (GI) tract bleeding as the globin component of haemoglobin is degraded by gastric proteases in the upper GI tract.

Quantitative FIT enables the selection of a preferred faecal haemoglobin threshold level for positivity, allowing the adjustment of clinical sensitivity and specificity so that the test is clinically acceptable while not overwhelming endoscopy resources. 46–48 At a low haemoglobin threshold, FIT can detect lower levels of bleeding than gFOBT and, therefore, has a higher sensitivity for CRC and AAs. 49–51 Quantitative FIT provides the opportunity to incorporate haemoglobin concentration into a multivariable CRC risk score, rather than the binary risk (positive or negative) afforded by gFOBT. 52

Test uptake is reported to be higher with FIT than gFOBT. 53,54 Individuals may find FIT more acceptable than gFOBT because the kit is typically simpler to use, less messy and usually requires only one stool sample (unlike the widely used three-sample, six-window gFOBT). 53

Given the advantages of FIT over gFOBT, many national screening programmes have started to use FIT to screen average-risk populations. 42 A systematic review examined the performance of FIT in the screening context and found that at a threshold of 20 µg haemoglobin per gram of faeces (hereafter referred to as µg/g), sensitivity for CRC approximated 90%. 55 Sensitivity for AAs is reported to be lower, typically < 50%. 56

Faecal immunochemical tests in post-polypectomy surveillance

Given the high sensitivity of low-threshold FIT for CRC in the screening setting, FIT could be useful for post-polypectomy surveillance. However, few published studies have looked at the use of FIT for this purpose and those few that have included patients undergoing surveillance for reasons other than adenomas (e.g. family or personal history of CRC). 46,57–62

A number of studies have examined the performance of interval FIT in addition to colonoscopy surveillance. 58,60,61 In the first of these studies,58 1641 individuals enrolled in colonoscopy surveillance (538 of whom were undergoing surveillance because of a personal history of neoplasia) were invited to complete a qualitative FIT [Inform, Enterix Pty Ltd, Sydney, NSW, Australia; Inform is known as InSure (Enterix Inc., Edison, NJ) in the USA] in the interval prior to colonoscopy. Among the 792 who responded and completed a FIT, 57 (7.2%) tested positive. Of the 57 who tested positive, six (10.5%) were diagnosed with CRC and eight (14.0%) had a significant adenoma. The study authors suggested that using FIT in surveillance could speed detection of interval CRCs. However, the study did not report the results of subsequent colonoscopy in those testing FIT negative and, therefore, sensitivity and specificity of FIT could not be calculated.

A second study, by Cole et al. ,60 is, to our knowledge, the only published study besides ours to have examined the performance of annual FIT in surveillance. A total of 1736 individuals enrolled in colonoscopy surveillance (984 of whom had prior adenomas) were invited to complete annual qualitative FIT (InSure) in the interval prior to surveillance colonoscopy. Of these 1736 individuals, 1071 (61.7%) completed at least one FIT (the median number completed was two) in the interval prior to surveillance colonoscopy, of whom 379 (35.4%) tested positive at least once. Colonoscopy was performed either following positive FIT or, in those testing FIT negative, after the designated surveillance interval. Sensitivity of repeated FIT was 85.7% (12/14) for CRC and 62.5% (60/96) for AAs. In cases in which CRC and AAs were diagnosed following a positive test, it is estimated that FIT brought forward colonoscopy by a median of 25 months and 24 months, respectively.

Most recently, Cole et al. 61 examined the performance of two-sample interval quantitative FIT (OC-Sensor, Eiken Chemical Co. Ltd, Tokyo, Japan; 20 µg/g threshold) in addition to colonoscopy surveillance. In total, 804 colonoscopies were performed early bacause of a positive FIT. As a result, nine (1.1%) patients were diagnosed with CRC and 162 (20.1%) patients were diagnosed with AAs. The results of subsequent surveillance colonoscopy in those patients testing negative were not reported and, therefore, sensitivity could not be calculated. However, the findings indicate that FITs could bring forward detection of significant neoplasia if used in combination with colonoscopy surveillance.

In addition to these studies, there have been a number of others that have tested the performance of one round of FIT directly prior to surveillance colonoscopy. Robinson et al. 57 invited 919 individuals (420 of whom had prior adenomas) to complete a qualitative FIT (HemeSelect, SmithKline Diagnostics, San Jose, CA, USA) before surveillance colonoscopy. In the 808 who complied, sensitivity for CRC was 70.0% (7/10) and sensitivity for large adenomas (i.e. sized ≥ 10 mm) was 44.4% (16/36). Hazazi et al. 59 invited 1469 individuals to complete three FITs (OC-Micro, Eiken Chemical Co. Ltd) prior to colonoscopy. Of 1469 invited, 1000 completed a FIT (337 of whom had prior polyps). Sensitivity of the first of three FITs at 10 µg/g was 100.0% (8/8) for CRC and 44.4% (32/72) for advanced colorectal neoplasia (ACN) (CRC and/or AAs). Using the highest haemoglobin concentration of all three FITs, sensitivity for ACN increased to 65.3% (47/72).

Although the study by Hazazi et al. 59 indicated that low-threshold quantitative FITs had high accuracy for ACN, a more recent study by Terhaar sive Droste et al. 46 of quantitative FIT (OC-Sensor) reported lower sensitivity. At a threshold of 10 µg/g, sensitivity was 80.0% (4/5) for CRC and 30.2% (32/106) for ACN. Both studies, however, had limited sample sizes and did not test the performance of multiple rounds of FIT.

These studies provide some evidence to suggest that FIT could be a useful tool for surveillance following adenoma removal. However, further research is needed on the accuracy, acceptability and cost-effectiveness of repeated FIT for detecting CRC and AAs in a large cohort of individuals. For this reason we developed the FIT for Follow-Up study.

Study aim

The aim of the FIT for Follow-Up study was to determine whether or not annual FIT is a feasible, safe, acceptable and cost-saving alternative to colonoscopy surveillance for the detection of CRC and AAs following adenoma removal. We proposed to offer annual FIT to people deemed at intermediate risk because of adenomas (see Figure 1) detected following a positive gFOBT completed as part of the BCSP in England. As is standard UK practice,21 these individuals were scheduled to have a surveillance colonoscopy at 3 years. In the BCSP, intermediate-risk patients account for approximately 40% of all patients diagnosed with adenomas, and approximately 5000 intermediate-risk patients a year are identified. 16 We hypothesised that, by using FIT as an alternative to colonoscopy surveillance, the number of colonoscopies could be greatly reduced, with minimal loss of sensitivity.

Primary objective

To determine the 3-year programme sensitivity of annual FITs compared with colonoscopy surveillance at 3 years, for the detection of CRC or AAs in patients categorised as at intermediate risk because of adenomas detected at colonoscopy following a positive gFOBT completed as part of the BCSP.

Secondary objectives

• To estimate the diagnostic accuracy of FIT at first, second and third tests and over two or three tests at various thresholds.

• To examine the acceptability of FIT, compared with colonoscopy, as a method of surveillance for people at increased risk of CRC.

• To calculate the incremental costs and cost-effectiveness of FIT versus colonoscopy surveillance.

Literature search

We searched MEDLINE and EMBASE, using text words and medical subject heading (MeSH)/EMTREE terms, to identify studies using FIT in a surveillance setting published on or before 6 April 2017. Reference lists of relevant articles were also reviewed to identify further pertinent studies.

Research governance and ethics arrangements

The FIT for Follow-Up study was prospectively registered in the International Standard Randomised Controlled Trial Number (ISRCTN) registry (ISRCTN18040196). Imperial College London was the nominated sponsor of the study. A trial steering committee provided independent oversight and advice.

In line with Imperial College London research governance procedures, relevant approvals were obtained before the study commenced, and appropriate regulations and guidelines were followed. The BCSP Research Committee granted support for the study on 7 October 2009. The London – City and East Research Ethics Committee approved the study on 17 May 2011 (reference number 11/LO/0326). Further approval was granted for substantial amendments to (1) invite a subset of participants to an end-of-study interview, (2) add a ‘health and lifestyle’ questionnaire to round 3 of testing and increase the FIT positivity threshold from 20 µg/g to 40 µg/g and (3) include a reminder letter to be sent to participants who had not yet attended their routine 3-year surveillance colonoscopy. The Ethics and Confidentiality Committee (ECC) of the National Information Governance Board (now known as the Health Research Authority Confidentiality Advisory Group) granted approval [reference number ECC 3-04(p)/2011] to access patient identifiers and contact details on 10 June 2011, allowing the identification of eligible participants and invitation by the BCSP.

Participant recruitment

Individuals who had been identified as being at intermediate risk (one or two adenomas of ≥ 10 mm in size, or three or four adenomas of < 10 mm in size, no CRC) within the previous 12 months were identified from those who had attended colonoscopy after positive gFOBT in the English BCSP.

Individuals meeting the eligibility criteria were identified by NHS Digital (formerly known as the Health and Social Care Information Centre) using the BCSP’s database, the Bowel Cancer Screening System (BCSS). NHS Digital sent encrypted information on potentially eligible individuals to the BCSP Southern Hub. In the BCSP, there are five regional hubs (Southern Hub, Eastern Hub, Midlands and North West Hub, London Hub and North East Hub) that receive and analyse gFOBT kits. 16 However, for the purpose of this study, analysis and distribution of kits were centralised; kits were sent and analysed from one hub, the BCSP Southern Hub, regardless of the invitee’s location. The BCSP Southern Hub sent an invitation letter, a detailed participant information sheet, a consent form, a short baseline questionnaire and a FIT kit to eligible individuals between January 2012 and December 2013 (see Report Supplementary Material 1 for study documentation). Individuals who returned a completed consent form and an analysable FIT constituted the study cohort.

Eligibility criteria

Criteria for inclusion were:

• aged 60–72 years at recruitment

• diagnosed < 1 year previously with intermediate-risk adenomas at colonoscopy following positive gFOBT in the BCSP

• scheduled for surveillance colonoscopy 3 years after initial colonoscopy, in line with BCSP guidelines for post-polypectomy surveillance of intermediate-risk patients.

Although individuals aged 72–74 years are invited to gFOBT screening in the BCSP, this group was excluded from this study. Given that they would be ≥ 75 years by the end of the study, they would not be eligible for a surveillance colonoscopy according to BCSP guidelines.

After study initiation, it was found that some potentially eligible individuals had received multiple baseline colonoscopies. These individuals were excluded from the study to prevent patient overinvestigation from a further colonoscopy at years 1 or 2 because of a positive FIT result.

Faecal immunochemical test kit processing

The FIT kit comprised a step-by-step instruction wallet, the FIT sampling device (OC-Sensor sampling device, Eiken Chemical Co. Ltd), a plastic zip-lock bag with absorbent material in case of leakage and a foil-lined pre-paid envelope in which to return the completed kit. The distribution of FIT kits and analysis of returned tests was conducted by the BCSP Southern Hub through the patient management system (PMS) (see Data processing and information governance).

Laboratory analysis of the FIT was carried out at the BCSP Southern Hub, using the OC-Sensor DIANA analyser (Eiken Chemical Co. Ltd) in accordance with the manufacturer’s instructions. Samples were refrigerated on receipt and analysis took place within 7 days. Results were uploaded onto the PMS as comma-separated values (CSV) files. Any samples that were > 10 days old on receipt were not analysed. If the sample was spoilt in any other way [damaged packaging, technical fail, non-technical fail (e.g. insufficient sample to analyse)], then a replacement kit was sent to the participant. The coefficient of variation for FIT was 4.5% at a concentration of 20 µg/g and 3.3% at a concentration of 90 µg/g. 64

In the pilot study, the threshold for a FIT positive result was set at 20 µg/g. However, as the percentage of participants testing positive in this pilot was higher than expected, the threshold was subsequently increased to 40 µg/g. This change of threshold was made to ensure that the range of predicted positivity rates would not lead to unsustainable numbers of colonoscopies that would swamp screening centres. The threshold was not changed retrospectively for participants screened within the pilot study (see Pilot study).

Study design

The study was divided into three annual FIT rounds at 1, 2 and 3 years after baseline colonoscopy (i.e. colonoscopy at which intermediate-risk adenomas were detected), as shown in the study flow diagram (Figure 2).

FIGURE 2.

The FIT for Follow-Up study flow diagram. PEQ, participant experience questionnaire; R1, round 1; R2, round 2; R3, round 3 (intervention rounds).

Participants who tested FIT positive at round 1 or 2 were offered an early surveillance colonic examination (typically colonoscopy) and were not invited to further FIT rounds. For participants who accepted the offer, the early colonic examination was the reference standard. If the participant declined the offer in round 1 or 2, they were still invited to an end-of-study colonic examination at 3 years after baseline colonoscopy, which served as the reference standard.

Participants who tested FIT negative in rounds 1 and 2 were eligible for further FIT rounds. All participants offered a round 3 FIT were invited to the routine 3-year surveillance colonic examination, which again served as the reference standard. Those who attended a colonic examination following positive FIT, because of symptoms or at the end of the study, returned to the appropriate surveillance pathway according to the outcomes of the examination (see Figure 1).

Questionnaires and interviews

Participant experience questionnaires were sent by post (1) with the consent form in round 1 (baseline questionnaire), (2) with FIT-negative results letters in rounds 1 and 2 (round 1 negative questionnaire, round 2 negative questionnaire), (3) after attending/declining an early surveillance colonoscopy after a FIT positive result in rounds 1 and 2 (round 1 positive questionnaire, round 2 positive questionnaire) and (4) after the participant had attended their year 3 end-of-study colonic examination (round 3 end-of-study questionnaire) (see Report Supplementary Material 2 for study questionnaires). To avoid unnecessary distress, participants diagnosed with cancer were not sent questionnaires.

End-of-study phone interviews were conducted with a subsample of participants in order to better understand attitudes towards the use of FIT in surveillance (see Chapter 4).

Colonic examinations

The default colonic examination was colonoscopy. However, as is current practice in the BCSP, computed tomographic (CT) colonography or flexible sigmoidoscopy were alternative options when colonoscopy was not deemed appropriate. In the BCSP, only a minority of patients (typically < 3%) are deemed unfit for colonoscopy. 16

If participants tested FIT positive in round 1 or 2, the participant’s affiliated BCSP screening centre was informed that the participant had consented to take part in the study, had tested FIT positive and should be offered an early colonic examination.

Before commencement of the study, all screening centre directors and nominated specialist screening practitioners at the 64 English screening centres were encouraged to support the study, and to contact FIT-positive participants to discuss arranging an earlier colonic examination in a timely manner to reduce participant anxiety. Collaborators at NHS Digital modified the BCSS to allow specialist screening practitioners to amend a participant’s surveillance due date, bring forward the date of the colonic examination and indicate that this was as a result of them taking part in the study and testing FIT positive. The Imperial College London study team contacted the appropriate screening centre on a case-by-case basis for all those who tested FIT positive, retrieved the relevant colonic examination and histology reports for those who attended and updated the PMS to show that the participant had declined an earlier examination.

All participants who did not attend (DNA) an early colonic examination as a result of a positive FIT were still invited for their year 3 surveillance examination. To manage the timely collection of the year 3 surveillance examination reports for participants who tested FIT negative, monthly data downloads from the BCSS with information on participants’ scheduled and attended colonic examinations were sent by NHS Digital to the Imperial College London study team. The team were responsible for contacting the screening centres to determine whether colonic examinations had been arranged or declined and, once colonic examinations had occurred, to retrieve copies of the relevant colonic examination and histology reports. Colonic examination and histology reports were sent directly to data clerks at the Imperial College London central study office by encrypted NHS.net e-mail or secure fax, and were immediately logged on receipt.

Data clerks entered data according to a standard operating procedure (see Report Supplementary Material 3). All data entered were audited by a second data clerk.

Colonic examination data entered included:

• date of examination

• type of examination

• pre-medications used (if any)

• name of the endoscopist

• bowel preparation used

• segment of the bowel reached

• whether or not the examination was completed

• the final results of the examination

• details of polyps discovered in the bowel

• whether or not a polyp was removed and successfully retrieved

• polyp pathology (if known)

• evidence of CRC or AA.

Data processing and information governance

The study was co-ordinated through a purpose-built PMS. The PMS was hosted in an ISO 27001 high-security data centre by ioko365 (Piksel Ltd, York, UK). The Clinical Informatics Research Unit at the University of Southampton processed the data, and was responsible for the development, maintenance and security of the PMS and had policies in place for information governance (IG) that were compliant with the Data Protection Act and the European Clinical Trials Directive.

Every aspect of the study was conducted through the PMS, creating a complete record of each individual’s passage through the study. The PMS was used by the following organisations that collaborated on this study:

• The BCSP Southern Hub (based at Royal Surrey County Hospital NHS Trust).

  • Nominated members of the BCSP Southern Hub had access to the PMS in order to upload information about potentially eligible individuals and for patient management purposes.

  • The IG toolkit organisation code was RA2.

• Cancer Screening and Prevention Research Group (CSPRG), Imperial College London.

  • Nominated members of the CSPRG could receive paper records from study participants and NHS sites involved with participants’ medical care. All records were stored in secure locked cabinets.

  • Nominated members of the CSPRG received pseudonymised electronic BCSS data, which were stored in a managed and secure area on the server.

  • Nominated members of the CSPRG had access to the PMS in order to carry out trial management duties, record the information received and extract information on consented participants for analysis.

  • The CSPRG held a level 3 IG certificate that was subsequently superseded by the IG toolkit version 14 (CSPRG organisational code: 8HL46-FOM-CSPRG).

  • The PMS was hosted and developed outside Imperial College London but complied with Imperial College London’s policy on data handling and data storage. ioko365 (Piksel Ltd, York, UK), the company that hosted the PMS, held an IG toolkit (IG Toolkit Organisation Code 8GX09). The Head of the CSPRG was the data controller.

• Department of Behavioural Science and Health, University College London.

  • Selected members had limited access to the PMS in order to record the questionnaires. They could not access participant-identifiable information. Access to the system was from the CSPRG office with a guest user account valid for a limited time period.

Statistical methods

Patient and public involvement

Patient and public representatives were involved throughout this study by means of workshops, individual consultations (e.g. to co-participate in designing patient materials and questionnaires) and as members of the independent steering committee. We consulted a range of representatives, some of whom had no history of colorectal investigations, some of whom had tested negative in the BCSP and had no experience of colonoscopy, some of whom had tested positive in the BCSP and had experienced an investigative colonoscopy, and some of whom were in an active adenoma surveillance programme.

While planning the study, we took advice from a panel of patient experts on contacting individuals prior to consent. The panel felt that, given the importance of the research, this was acceptable, as long as the details of individuals who had not given explicit consent to take part in the study were not shared outside the BCSP Southern Hub. Personnel at the BCSP Southern Hub were the only members of the study team who could view unconsented individuals’ personal details, in order to produce the initial invitation letters.

Representatives were consulted in the pre-pilot phase (see Preparation for the study), during which the PMS and study materials (e.g. flyer, letter, consent form, baseline questionnaire, FIT kit instructions, participant information sheet) were being developed. Representatives’ feedback strongly influenced the type of study materials that were used and their layout and content.

Service users were also interviewed to assess the likely acceptability of FIT as a CRC surveillance tool. A discussion group was convened in November 2011 to discuss the practicalities of using FIT, the implications of the results and the possible risks associated with FIT. As a result of users’ feedback about the psychological implications of receiving a false-positive test result, patient materials were modified to clearly demonstrate the possibility of testing false positive (see Chapter 4 and Bowyer et al. 67 for full details and results).

Users were also involved in assessment of the pilot study (see Pilot study), the aim of which was to test the processes for invitation, analysis and transmission of results, and storage of information on the PMS. A number of modifications to the PMS and to the study materials were made as a result of the pilot study.

Health psychology assessment

A secondary objective of the FIT for Follow-Up study was to examine the acceptability of FIT compared with colonoscopy as a surveillance mechanism for people at increased risk of developing CRC. Participant experience questionnaires after each round of FIT allowed us to assess various aspects of the study, including the FIT kit instructions, how easy participants found it to use the FIT kit, how they felt when receiving the results, how any subsequent colonoscopy affected them, and their screening preferences (see Chapter 4).

We assessed the psychological consequences of annual surveillance over time using an itinerary of questions, including a short version of the Spielberger State–Trait Anxiety Inventory (STAI)68 and more specific measures of CRC-related worry. 69 We also investigated the emotional impact of FIT outcomes (e.g. false-negative and false-positive results) through telephone interviews. Participants suspected of having cancer were not sent questionnaires to avoid causing unnecessary distress. Chapter 4 fully documents the methods and analyses pertaining to the health psychology assessment.

Economic evaluation

We assessed the cost and cost-effectiveness of annual FIT with colonoscopy only for a positive result versus 3-yearly colonoscopy surveillance. Further details are in Chapter 5.

Preparation for the study

In July 2011, we conducted a pre-pilot study to review study materials and to assess PMS functionality. Volunteers were asked to review and comment on the flyer introducing the study, invitation letter, participant information sheet, consent form, baseline questionnaire, FIT kit instruction wallet and foil-lined sample return envelope.

Feedback on the study materials was received from 27 age-appropriate volunteers (60–71 years), six other volunteers (< 60 years) and 13 members of the research teams.

It was decided that a flyer, designed to introduce potential participants to the study, should not be used in the main study, as feedback indicated that it did not contain enough information about the study and would be likely to confuse potential participants. Other feedback warranted changes to the invitation letter and the participant information sheet.

A qualitative study with discussion groups was conducted (November 2011) to gain an understanding of public attitudes towards FIT, identify potential issues for the main study and facilitate the design of study materials. Details of this study are further described in Chapter 4, Study 1: patient attitudes towards the faecal immunochemical test as an alternative to colonoscopy surveillance of groups at increased risk of colorectal cancer – a qualitative discussion group study.

Pilot study

The aim of the pilot study was to test the processes for invitation, analysis, transmission of results and storage of information on the PMS. It also enabled us to examine uptake, requirements for reminders, attendance for colonic examination of those with a positive FIT and the frequency and nature of calls to the freephone helpline. Following the pilot study, a number of modifications to the PMS and to the information and study materials were made. A major change was to the haemoglobin threshold set to define a positive FIT result. Initially, the threshold was set at 20 µg/g, as this is commonly used in screening programmes and has been used frequently by other studies. 70–76 It was subsequently changed to 40 µg/g as the proportion positive at a threshold of 20 µg/g was 6.5%, which was higher than we expected given the literature. 77 We were concerned that this level of positivity had the potential to cause an excess of colonoscopies, with consequences on patient risk and endoscopist workload.

Invitation to participate in the study

In total, 9851 individuals were identified as potentially eligible for the study. Of these, 296 were excluded after eligibility assessment for the following reasons: 187 individuals had received more than one colonoscopy prior to invitation and were therefore excluded to prevent patient overinvestigation, and 109 individuals were excluded by the BCSP for clinical reasons, or because of informed dissent, death or emigration. A further 1547 were not invited as the recruitment target of 8000 participants had been met. The remaining 8008 individuals were invited to participate in the study between 30 January 2012 and 30 December 2013 (Figure 3). The mean age of these invited individuals was 65.7 years (standard deviation 3.4 years; 49.3% were aged ≤ 65 years and 50.7% were aged > 65 years), and 65.3% were men (Table 1).

FIGURE 3.

Participant flow diagram from invitation through to end-of-study colonic examination. Invited: 9851 individuals were identified as potentially eligible for the study and were ordered by surveillance due date; 8008 individuals were invited to participate. A total of 1843 of the potentially eligible individuals were not invited to participate in the study; 296 were not invited as they were excluded after the eligibility assessment and a further 1547 individuals were not invited as the recruitment target of 8000 had already been met. Not recruited: 2055 were because of lack of consent, one consented but had not returned a FIT kit and four consented but returned a FIT kit that could not be analysed. DNA: includes participants who (using data collected on 6 October 2016) were offered, but DNA, a diagnostic examination after a positive kit, or DNA a study-end colonic examination after being offered such.

Participation in surveillance by round

Uptake and test positivity

Attendance for colonic examination and yield of colorectal cancer, advanced adenomas and advanced colorectal neoplasia

Performance of the faecal immunochemical test at different faecal haemoglobin thresholds

Accuracy of the faecal immunochemical test at different faecal haemoglobin thresholds

We estimated the sensitivity, specificity, PPV and NPV of FIT at different faecal haemoglobin thresholds among the 5199 participants who underwent a colonic examination (see Tables 4–6).

Introduction

In this chapter, we present studies evaluating perceptions and preferences for FIT surveillance and summarise the experience of participants in the main FIT for Follow-Up study. Colonoscopy surveillance is currently considered the optimal test for individuals at intermediate or high risk of CRC. 21 However, since the introduction of the BCSP, there have been concerns about the implications for endoscopy departments and the risk associated with repeat colonoscopies at regular intervals. These concerns are further compounded by the modest diagnostic yield of surveillance colonoscopy. 19,39 Another issue with surveillance colonoscopy is that one in five individuals at intermediate or high risk of CRC do not attend the surveillance examinations, which is comparable to the uptake of colonoscopy following an abnormal gFOBT in the BCSP. 16,35,36 Factors associated with non-attendance at colonoscopy include anxiety about having to prepare for the test with a laxative, anxiety about the procedure itself (such as the risk of bowel perforation) and expectations of pain and embarrassment. 34

FIT might be a viable alternative as the primary surveillance test. FIT is administered at home, is non-invasive and requires only a single stool sample, thereby significantly reducing the unpleasantness associated with other home-based stool tests. 67 As such, FIT is aligned with people’s preferences for CRC tests that are non-invasive, avoid pain as much as possible and do not require bowel preparation, while offering high levels of sensitivity and specificity. 78 One study found that average-risk participants who were concerned about procedural discomfort were more likely to choose FIT over colonoscopy as a screening modality. 79 In these patients, FIT completion rates were higher than colonoscopy attendance, a finding supported in a recent RCT. 80 However, there is a lack of research into the perceptions of and preferences for different forms of post-polypectomy surveillance among higher-risk individuals.

Here, we present the findings of the health psychology work stream nested within the FIT for Follow-Up study. Its aim was to better understand what people think about FIT in the context of surveillance, to monitor the actual participant-reported experience associated with an annual FIT and begin to investigate what preferences people at intermediate risk have for their adenoma surveillance.

Study 1: patient attitudes towards the faecal immunochemical test as an alternative to colonoscopy surveillance of groups at increased risk of colorectal cancer – a qualitative discussion group study

The following is an executive summary of a study into public attitudes towards FIT. Its main aim was to gain an early understanding of potential issues for the main study and to facilitate the design of participant materials. A full summary has been published by Bowyer et al. 67

Study 2: acceptability of annual faecal immunochemical tests for post-polypectomy surveillance – findings from the main study

General discussion

Participant-reported satisfaction with the FIT for Follow-Up study was very high. This was probably in part bolstered by the participants’ previous experience with gFOBT. Completing the FIT kit was unanimously perceived as easier and more convenient, especially as it required only one stool sample. However, many participants remained sceptical that a single sample stool kit would detect CRC reliably and preferred to have the test in addition to, rather than as a replacement for, colonoscopy.

Although the act of completing the test kit and waiting for the results was associated with some anxiety, this did not seem to have lasting effects as CRC-related worry or general anxiety was reported to be very low, even for those who had received positive FIT results.

Overall, the main message from these studies seems to be that participants valued colonoscopy for its ability to visually inspect the bowel and were concerned that a single stool sample test might miss intermittently bleeding polyps and cancers. From a participant perspective, therefore, the FIT was appreciated as offering a quick and easy way to gain additional reassurance rather than as a stand-alone test.

There are limitations to bear in mind when interpreting these results. Despite the high level of recruitment, our findings might not represent the entire surveillance population. Importantly, our end-of-study preference data are restricted to participants who had undergone an end-of-study colonic examination, excluding participants who withdrew during the study, failed to attend their end-of-study colonic examination or did not return their final questionnaire. As a result, our findings might over-represent people who are concerned about CRC and have positive attitudes towards colonoscopy.

Across all of our studies we were unable to record the ethnic mix of participants. If FIT for surveillance were to be implemented, it would be vital to ensure active participation from patients from ethnically diverse backgrounds. Future work should also investigate the views of patients who have not previously been exposed to colonoscopy surveillance. In a recent hypothetical survey of 491 screening-naive individuals, the majority (61%) stated a preference for a surveillance test resembling a home-based FIT, whereas only 31% reported a preference for colonoscopy. Increased frequency of testing was a commonly cited reason for the preference for FIT. Unfortunately, the response rate of the survey was only 16%, limiting its generalisability. Future work should investigate whether or not the preference for FIT in screening-naive individuals translates to greater adherence in the surveillance setting. 89

Future research should try to identify the extent to which FIT might enable the small but substantial proportion of people not attending surveillance colonoscopy to have adenoma surveillance. This study also revealed that the key concerns about annual FIT surveillance are the lack of visual inspection and the reliance on a single stool sample, which might miss intermittently bleeding polyps and cancers. These concerns are based on complex beliefs, which need to be explored further.

Another important question for future research will be to determine what precise additional benefit people expect to gain from annual FIT. In this study, it was interesting to observe the response from people who had CRC-related abnormalities detected following a positive FIT result in round 1 or 2. Here, our end-of-study interviews revealed ‘surprise’ or even ‘shock’, often combined with a realisation that abnormalities detected so soon after the baseline colonoscopy would probably have been missed at that examination. The extent to which people value FIT for its ability to detect disease that has been missed previously should be investigated further.

To conclude, these investigations have shown that, although there is enthusiasm for the use of FIT in CRC surveillance, it is often restricted to the use of FIT as an additional rather than a stand-alone test. Future research is needed to better understand the concerns of participants with regard to FIT missing polyps and cancers.

Introduction

We investigated the incremental cost and cost-effectiveness of annual FIT versus colonoscopy at 3 years post baseline colonoscopy for post-polypectomy surveillance of intermediate-risk patients.

As defined in the protocol, under the assumption that FIT was as accurate as colonoscopy surveillance, the aim of the economic evaluation was to undertake a cost-minimisation analysis of annual FIT surveillance versus colonoscopy surveillance at 3 years. In the event that the marginal effectiveness of FIT surveillance was significantly less than that of colonoscopy surveillance, the analysis would instead comprise a cost and cost-effectiveness analysis, balancing test costs against test outcomes.

Method

Generating a control group

The FIT for Follow-Up study was a single-arm trial, which meant that for the economic analysis we had to generate a control arm (a pseudo-control group) to consider the cost and outcomes of people who would have had routine 3-yearly colonoscopy surveillance in the absence of FIT.

In order to create this pseudo-control group, we assumed that all DNA participants for colonic examinations scheduled because of positive FIT would also be DNA participants for routine colonoscopy surveillance, and that those participants lost to follow-up during the study would also be lost to routine colonoscopy surveillance. As a starting point, we defined the size of the intervention and control groups as the 5946 participants who completed round 1 of the FIT regimen. To estimate the number of participants receiving colonoscopy surveillance at 3 years in the absence of FIT, we used the sum of all participants involved in the trial who had a colonic examination (4455 participants who had the routine 3-year colonic examination, plus 744 participants who had an examination after having a FIT positive result, which resulted in 5199 participants in total).

The appropriateness of the size of the pseudo-control arm receiving colonoscopy surveillance at 3 years can be confirmed by subtracting the participants who DNA their colonic examination and those who were lost to follow-up from the original 5946 participants who completed round 1 of the FIT regimen (5946 minus 584 DNA participants for the 3-year colonic examination, minus 51 DNA participants for early colonic examination after a FIT positive result, minus 112 participants who were lost to follow-up during the course of the 3-year study period, which resulted in 5199 participants in total).

Figure 4 shows the actual number of study participants completing each round of FIT and the estimated number undergoing colonoscopy surveillance at 3 years in the absence of FIT. Figure 5 shows the resulting costs, number of colonoscopies, and AAs and CRCs detected in each group.

FIGURE 4.

Participants included in the economic analysis. FIT round completed data were drawn from the FIT for Follow-Up study Consolidated Standards of Reporting Trials (CONSORT) diagram. The pseudo-control arm estimates the number of surveillance colonoscopies that would have taken place in the absence of the FIT for Follow-Up study.

FIGURE 5.

Estimated costs and outcomes associated with annual FIT surveillance and 3-yearly colonoscopy surveillance.

Results

Discussion

Our economic analysis to evaluate the cost-effectiveness of annual FIT surveillance versus 3-yearly colonoscopy surveillance showed that FIT surveillance was cheaper than colonoscopy surveillance, but also less effective at detecting AAs and CRCs. Sensitivity analysis suggested that the cost of DNAs did have an effect on the analysis, as did the cost of colonoscopies.

The total cost associated with FIT surveillance was sensitive to the faecal haemoglobin threshold selected to define FIT positivity. This is not surprising, as more people would be classed as positive and referred for colonic examination with lower thresholds. In all cases, annual FIT surveillance was less costly than 3-yearly colonoscopy surveillance; however, at the lowest threshold (10 µg/g), there is a chance that FIT surveillance was not less costly than colonoscopy surveillance. The incremental cost-effectiveness per AA and CRC not detected increased significantly (i.e. reflecting better value for money) at lower thresholds. This is because, despite higher diagnostic costs associated with higher numbers of colonic examinations at lower thresholds, FIT surveillance would miss fewer participants with AAs and CRC at these lower thresholds. This is despite the increase in diagnostic costs associated with the higher numbers of colonic examinations that would be performed. The findings suggest that using FIT as an alternative to colonoscopy for surveillance could result in significant financial savings, although this would mean missing a number of AAs and CRCs.

Main findings

In our study, the sensitivity of FIT was greater for CRC than for AAs and depended strongly on the faecal haemoglobin threshold applied and the number of completed FIT rounds. The sensitivity of the first completed FIT at a threshold of 40 µg/g was only half that estimated at a threshold of 10 µg/g (30.8% vs. 61.5% for CRC and 17.6% vs. 33.2% for AAs). Sensitivity was even higher at 10 µg/g in those who completed all FITs offered (91.7% and 59.3% for CRC and AAs, respectively, in cumulative test analysis).

Although using a low threshold and multiple rounds of FITs reduced the number of CRCs and AAs missed, this came at a cost of reduced specificity and an increase in cumulative positivity and requirement for colonoscopy. However, even when the lowest threshold of 10 µg/g was applied, 71% of participants did not test positive at any round. Therefore, using three rounds of annual FIT as an alternative to the first surveillance colonoscopy at year 3 would considerably reduce the number of colonoscopies performed.

The FIT positivity declined by round of testing, reducing from approximately 6% at the first round to 4% at the third round with a threshold of 40 µg/g, and from 14% to 9% with a threshold of 10 µg/g. The decrease in positivity is not surprising given that participants who would have tested positive at a given threshold would have been offered an early colonoscopy and, therefore, would not have been invited to subsequent FIT rounds.

Sensitivity also declined by round of testing; for example, at a threshold of 40 µg/g, sensitivity for ACN was estimated to be 17.9% for the first completed FIT, 11.7% for the second FIT and 12.1% for the third FIT. At 10 µg/g, sensitivity for ACN was 34.3% for the first FIT, 22.5% for the second FIT and 19.6% for the third FIT. A possible explanation for this is that participants with an AA or CRC that caused high faecal haemoglobin levels were likely to test positive at an early round.

There were differences in the performance of FIT by age and sex. FIT positivity was higher in older and male participants. This reflects the prevalence of ACN, which, in both this study and more generally, increases with age and is higher in men. 104,105 However, sensitivity for AAs and CRC was also generally higher, and specificity lower, in older participants and men. The reasons for these differences in FIT positivity and sensitivity between older and younger patients, as well as between men and women, are not yet fully understood.

Looking at the stage distribution of the CRCs detected during the study, 11 were pT stage three or four. Considering that these were detected within 3 years following colonoscopy, it is likely that they were present but missed during the baseline examination. Although advanced-stage CRCs are more likely to be detected by FIT than early-stage CRCs,75 our reported sensitivities for FIT are not likely to be significantly inflated compared with what would be achieved in clinical practice; colonoscopy is not perfectly sensitive and a small proportion of CRCs, including those at advanced stages, are routinely missed. 106–108 A surveillance programme employing annual FIT would therefore help to speed detection of missed CRCs compared with a programme adopting surveillance colonoscopy at 3 years. Indeed, in our study, FIT detected 8 of the 11 pT3/4 CRCs at 40 µg/g and 10 out of the 11 at 10 µg/g, with seven detectable in round 1 or 2 at either threshold.

Comparison with other studies of faecal immunochemical tests in surveillance

The FIT for Follow-Up study is, to our knowledge, the largest study to date to have examined the performance of FIT in post-polypectomy surveillance. In our study, 5199 participants completed at least one FIT and attended colonoscopy in the study period, during which time 26 CRCs and 443 AAs were diagnosed.

Estimated sensitivity for CRC of a single FIT at the lowest studied threshold (10 µg/g) was lower in our study (61.5%) than in previous studies of both qualitative and quantitative FIT in a surveillance setting.

Robinson et al. 57 reported the sensitivity for CRC of a qualitative FIT (Hemeselect, SmithKline Diagnostics, San Jose, CA, USA) before surveillance colonoscopy to be 70.0% (7/10). Terhaar sive Droste et al. 46 reported the sensitivity for CRC of a single quantitative FIT (OC-Sensor) to be 80.0% (4/5) at 10 µg/g. Hazazi et al. 59 found the sensitivity of three consecutive quantitative FITs (OC-Micro at 10 µg/g) for CRC before surveillance colonoscopy to be 100.0% (8/8). Low numbers of CRCs in each study limited the precision of these estimates. One potential reason for lower sensitivity in our study might be that the first FIT was completed 2 years prior to, rather than immediately before, the year 3 planned surveillance colonoscopy. Faecal haemoglobin concentration may have increased over time as neoplasia progressed.

The sensitivity for AAs or ACN of a single FIT at 10 µg/g was not markedly different in our study (33.2% for AAs and 34.3% for ACN) to figures reported in previous studies of FIT in surveillance. Robinson et al. 57 noted a sensitivity of 44.4% (16/36) of a qualitative FIT for large adenomas (i.e. sized ≥ 10 mm). Hazazi et al. 59 found the sensitivity for ACN of three consecutive quantitative FITs at 10 µg/g to be 44.4% (32/72). Terhaar sive Droste et al. 46 reported the sensitivity for AAs of a single FIT at 10 µg/g to be 27.7% (28/101).

The study conducted by Lane et al. 60 is the only other study to have examined the sensitivity of annual FIT in a surveillance setting. Annual qualitative FIT (median of two rounds) identified 12 out of 14 patients with CRC (sensitivity of 85.7%) and 60 of 96 patients with AAs (sensitivity of 64.5%). Comparable, but slightly lower, sensitivity was observed in our study with three annual FIT rounds and a threshold of 10 µg/g (programme analysis: 84.6% for CRC and 57.3% for AAs).

Comparison with other studies of faecal immunochemical tests in screening

More extensive research has been conducted into the use of FIT for CRC screening. A systematic review commissioned by the US Preventive Services Task Force,56 published in 2016, reported the sensitivity of a single FIT for CRC, from studies using a FDA-approved qualitative or quantitative FIT and colonoscopy as reference standard, to be in the range 73–88% and specificity to be in the range 90–98%. Sensitivity of a single FIT for AAs was reported to be in the range 22–40% and specificity to be in the range 91–97%. An earlier systematic review with meta-analysis of FIT for CRC screening calculated a pooled sensitivity and specificity for CRC of FIT at 20 µg/g of 86% (95% CI 75% to 92%) and 91% (95% CI 89% to 93%), respectively. 55

The sensitivity of FIT for CRC reported by these systematic reviews is higher than was observed in our study. One potential explanation for this may be differences in the size, site and stage of CRCs detected through surveillance relative to screening. A systematic review with meta-analysis by Hirai et al. 109 found lower sensitivity of FIT for proximal than distal CRC. Furthermore, faecal haemoglobin levels appear to be higher in patients with larger AAs and in those with a greater number of AAs. 110

Health psychology assessment

Participants generally reported high levels of satisfaction with the FIT for Follow-Up study. Although completing the test was associated with some increased anxiety, this did not seem to have a long-lasting effect. Participants also reported that FIT was easier and more convenient than the three stool sample gFOBT (hema-screen™; Immunostics, Eatontown, NJ, USA) used at the time in the BCSP.

There was some scepticism about whether or not FIT alone would detect cancer reliably, particularly with only one stool sample. Although participants were generally in favour of the use of FIT in surveillance, most preferred annual FIT in addition to colonoscopy every 3 years (57.9%, 2478/4279), with a lower proportion (31.5%, 1347/4279) preferring annual FIT alone. Many participants valued annual FIT in addition to colonoscopy for the more frequent reassurance this provided them.

Health economic assessment

In incremental cost and cost-effectiveness analyses, annual FIT surveillance was cheaper than 3-yearly colonoscopy surveillance. At a threshold of 40 µg/g, the cost of FIT surveillance for the 5946 study participants was estimated to be £485,236, compared with £2,633,382 for colonoscopy surveillance. However, compared with colonoscopy surveillance, FIT surveillance resulted in eight missed CRCs (8/26) and 291 missed AAs (291/443).

Reducing the FIT threshold increased the cost of FIT surveillance and reduced the number of colorectal lesions missed. With a threshold of 10 µg/g, the cost of FIT surveillance was estimated to be £956,602, with 4 out of 26 CRCs missed and 189 out of 443 AAs missed. Incremental cost-effectiveness per AA and per CRC not detected increased (i.e. improved) with lower thresholds. These findings demonstrate that low-threshold annual FIT could be a cheaper alternative strategy for surveillance of intermediate-risk patients following the removal of adenomas. However, further economic analysis would be beneficial to understand the cost implications of AAs and CRCs missed by FIT, and the impact on expected QALYs lost.

Strengths and limitations

Our study had a number of strengths, including a relatively large sample size, a well-defined population comprised solely of individuals undergoing surveillance for intermediate-risk adenomas, and results from multiple rounds, rather than a single round, of FIT.

The study also had limitations. One possible limitation is that we assumed that the population with ACN was static, such that any ACN detected was present and would remain present and unchanged in the absence of colonic examination at years 1, 2 and 3. This is a reasonable assumption as it is widely thought that annual polyp progression rates are low. A systematic review of the natural history of small polyps (i.e. sized < 10 mm) found that, in the few studies that have been conducted, the majority of untreated polyps did not progress to become an AA over 2–3 years. 111 Furthermore, in a study of 226 patients with large untreated polyps (i.e. sized ≥ 10 mm), the risk of CRC at 5 years post-baseline assessment was only 2.5%. 112

Under the assumption of a static ACN population and with a constant FIT sensitivity, we would expect the number of neoplasms detected to decrease with each round. The detection of eight CRC cases at round 1, eight CRC cases at round 2 and two CRC cases at round 3 is consistent with a static population assumption for CRC. However, the detection of 78 AAs at the first round and 37 at each of the second and third rounds is less consistent with a static population of AAs, and suggests that even within the study period of 3 years, there is some movement in and out of this population. However, if anything, the static population assumption is likely to lead to conservative results with respect to the effect of FIT at the various thresholds, as it takes no account of the improved outcome associated with lesions detected earlier.

A further limitation of the study is that we assumed that the colonic surveillance examination had perfect sensitivity for AAs and CRCs. Most participants attending colonic examination received a colonoscopy(98.0%, 5097/5199). Although considered the ‘gold-standard’ colonic examination, colonoscopy occasionally misses polyps and, on rare occasions, CRCs. 113,114 A small proportion of participants attending colonic examination did not receive colonoscopy but did receive CT colonography. A randomised trial of CT colonography and colonoscopy in 1610 symptomatic patients found little difference in the detection rate of CRC and large polyps between the examinations. 115 In a study of 2531 asymptomatic participants undergoing CT colonography followed by colonoscopy, sensitivity of CT colonography for adenomas or CRCs ≥ 10 mm in size was 90% relative to colonoscopy. 116 Flexible sigmoidoscopy has lower sensitivity than CT colonography or colonoscopy as the proximal colon is not examined. However, in our study, only two participants received flexible sigmoidoscopy without additional colonoscopy or CT colonography (0.04%, 2/5199).

The generalisability of our findings was limited to an extent by the inclusion criteria, which stipulated that all participants had to return a first round FIT to be included. Uptake of FIT at rounds 2 and 3 is likely to be higher in our study than if participants who did not respond to the first FIT invitation had also been reinvited. Participation was slightly higher in older invitees (75.6% among invitees aged > 65 years and 72.9% among invitees aged ≤ 65 years) and in men (74.6% of male invitees participated and 73.6% of female invitees participated). Given that we stratified our findings by age and sex, the negative consequences of this are limited.

The generalisability of the findings from the health psychology assessment are limited in a similar way, as questionnaires and interviews were administered only to individuals participating in the main study. The attitudes of these individuals towards FIT and surveillance are likely to have been more favourable than those of individuals who did not agree to participate.

A further potential issue of generalisability is that we evaluated just one FIT brand (OC-Sensor). Evidence from a recent study of nine different quantitative FITs does, however, suggest that this is not a significant limitation. 117 When directly compared in a single screening cohort, different FIT brands achieved near-equal sensitivities when thresholds were set to yield defined rates of test positivity.

As noted, the economic evaluation was limited as it was based on a single-arm trial and did not consider the cost implications of missed AAs and CRCs and impact on QALYs.

In addition, although we hoped to model the potential of FIT screening to replace colonoscopy surveillance for groups at intermediate risk of developing CRC, including those with a personal or family history of CRC, we unfortunately were not able to realise this aim in the study period. A cost-effectiveness analysis of different CRC screening and surveillance strategies has recently been conducted, using data from the Dutch general population. 118 The analysis used the Adenoma and Serrated pathway to Colorectal Cancer (ASCCA) model, populated with Dutch data for colorectal lesion prevalence rates and CRC incidence and mortality rates, to compare no screening or surveillance, FIT screening alone and FIT screening plus surveillance colonoscopy. It was revealed that both FIT screening alone and FIT screening with surveillance colonoscopy were more effective and cost saving than no screening or surveillance. FIT screening with surveillance colonoscopy was not, however, cost-effective compared with FIT screening alone (using the Dutch cost-effectiveness threshold).

Implications for clinical practice

Our findings suggest that a programme of low-threshold annual FIT could perform a useful role in the surveillance of intermediate-risk patients. However, in a recent study examining heterogeneity of CRC incidence in intermediate-risk patients, we identified a lower-risk subgroup, for whom the value of colonoscopy surveillance was unclear, and a higher-risk subgroup for whom colonoscopy surveillance was clearly beneficial. 102 The utility of FIT surveillance may differ dependent on risk subgroup.

Currently, most patients deemed at intermediate risk following polypectomy cease surveillance after two negative surveillance colonoscopies. 22 We have examined the performance of FIT relative to the initial surveillance colonoscopy but we do not have sufficient data on the performance of FIT in the longer term. FIT might also perform a useful role in patients diagnosed with high-risk adenomas instead of the first 12-month surveillance examination or after the first surveillance colonoscopy.

Recommendations for research

1. Further research on FIT-based surveillance is needed to examine the implications of missed ACN, considering effects on QALYs.

2. The identification of subgroups among intermediate-risk patients102 requires further research into the performance and value of FIT by subgroup.

3. The efficacy, safety and cost-effectiveness of FIT in longer-term surveillance, as well as in high-risk patients, should be evaluated in RCTs.

4. Future studies should aim to determine whether molecular stool testing in addition to FIT could improve sensitivity further. This is important as the diagnostic accuracy of FIT is to some extent limited given that, as it is a test for occult blood in stool, non-bleeding lesions will not be detected. Although van Lanschot et al. 119 are currently evaluating FIT and a stool DNA test performed separately in a surveillance setting, the sensitivity of these tests in combination requires investigation.

5. Finally, effort should be placed on building a natural history model of CRC, similar to the ASCCA model,118 which incorporates data from the British population and from intermediate-risk patients. This would enable evaluation of the potential to replace 3-yearly colonoscopy surveillance with annual FIT surveillance.

Conclusion

This study has demonstrated the potential value of FIT for the surveillance of patients deemed to be at intermediate risk following polypectomy. Annual rounds of FIT, using a low faecal haemoglobin threshold and with colonoscopy in positive cases, achieved high cumulative sensitivity for CRC, was well accepted by patients, and would be cost saving compared with 3-yearly colonoscopy surveillance. However, with higher thresholds, FIT surveillance could miss 15–30% of CRCs and 40–70% of AAs, and most participants preferred annual FIT in combination with 3-yearly colonoscopy. Further research is needed to define a clear role for FIT in post-polypectomy surveillance. A full RCT with cost–utility analysis that addresses the cost and health implications of AAs and CRCs that were not detected is recommended.

Trial Steering Committee

Dr Andrew Veitch (chairperson), Ms Helen Watson and Ms Lynn Faulds-Wood (patient representatives), Professor Allan Hackshaw, Professor Sue Moss, Professor Marco Novelli, Professor Matt Rutter and Professor Christopher Todd.

Patient and Public Involvement

Throughout the study, our patient and public involvement representatives sat on the Trial Steering Committee and provided relevant insight into the views of patients with regard to the study, and wider issues relating to bowel cancer diagnosis, screening and surveillance.

Trial office staff, Cancer Screening and Prevention Research Group

Ms Amy Brenner, Ms Elizabeth Coles, Dr Paula Kirby, Ms Jessica Martin, Miss Emma Robbins, Mrs Urvi Shah, Ms Irene Simmonds, Mr Iain Stenson and Dr Laura Turner.

Collaborators

Contributions of authors

Wendy Atkin was the principal investigator for the study.

Wendy Atkin, Amanda J Cross, Kate Wooldrage, Aaron Prendergast and Stephen W Duffy were responsible for data analysis and interpretation.

Wendy Atkin, Ines Kralj-Hans, Stephen W Duffy, Stephen Morris, Christian von Wagner and Stephen Halloran were responsible for obtaining funding and research approvals and study design.

Eilidh MacRae, Bhavita Patel, Kevin Pack and Rosemary Howe were responsible for trial management, including collecting, entering and cleaning the data.

Wendy Atkin, Amanda J Cross, Ines Kralj-Hans, Eilidh MacRae, Carolyn Piggott, Sheena Pearson, Kate Wooldrage, Jeremy Brown, Fiona Lucas, Aaron Prendergast, Natalie Marchevsky, Bhavita Patel, Kevin Pack, Rosemary Howe, Hanna Skrobanski, Robert Kerrison, Nicholas Swart, Julia Snowball, Stephen W Duffy, Stephen Morris, Christian von Wagner and Stephen Halloran reviewed and edited the report.

Jeremy Brown and Fiona Lucas drafted the final report.

Hanna Skrobanski, Robert Kerrison and Christian von Wagner were responsible for the psychological analyses.

Nicholas Swart and Stephen Morris were responsible for the economic analysis.

Publications

Bowyer HL, Vart G, Kralj-Hans I, Atkin W, Halloran SP, Seaman H, et al. Patient attitudes towards faecal immunochemical testing for haemoglobin as an alternative to colonoscopic surveillance of groups at increased risk of colorectal cancer. J Med Screen 2013;20:149–56.

Bonello B, Ghanouni A, Bowyer HL, MacRae E, Atkin W, Halloran SP, et al. Using a hypothetical scenario to assess public preferences for colorectal surveillance following screening-detected, intermediate-risk adenomas: annual home-based stool test vs. triennial colonoscopy. BMC Gastroenterol 2016;16:113.

Cross AJ, Wooldrage K, Robbins EC, Kralj-Hans I, MacRae E, Piggott C, et al. Faecal immunochemical tests (FIT) versus colonoscopy for surveillance after screening and polypectomy: a diagnostic accuracy and cost-effectiveness study [published online ahead of print December 11 2018]. Gut 2018.

Data-sharing statement

Data sharing requests should be directed to the corresponding author. Access to available anonymised, aggregated data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.