Quantitative faecal immunochemical testing to guide colorectal cancer pathway referral in primary care

Populations included in the evidence base

3.2 Most of the evidence was in populations that did not exactly match the population defined in the scope. Some populations only included people with 'high' or 'low' risk symptoms (see section 2.3), and some populations were unclear. The external assessment group (EAG) explained that sensitivity analyses indicated that these differences in study populations did not have a detectable effect on the estimates of diagnostic accuracy. But there was a large amount of variability between studies. Because some tests did not have evidence in the scope population, the EAG chose to include studies from a broader population. The committee concluded that the estimates of diagnostic accuracy based on this broad population were likely representative of the accuracy in the scope population.

Diagnostic accuracy of different tests

3.3 The quantity and quality of the evidence base varied between tests. The committee noted that most studies used either HM‑JACKarc (16 studies) or OC‑Sensor (17 studies). For FOB Gold, 3 studies were identified, but these were small studies and the estimates of accuracy were uncertain. The committee acknowledged that FOB Gold was recommended in NICE's diagnostics guidance on quantitative FIT to guide referral for colorectal cancer in primary care (DG30). During development of DG30 the committee concluded that, although there was less data for FOB Gold than for HM‑JACKarc or OC‑Sensor, it was likely to perform similarly in practice. However, in this assessment the committee observed that the evidence base for HM‑JACKarc and OC‑Sensor was now larger and the estimates of diagnostic accuracy were more certain than during the development of DG30. But the FOB Gold evidence base remained limited. The committee also noted recent evidence that different tests produce different results from the same samples, so equivalence between brands could not be assumed. Only 1 study was identified for each of IDK hemoglobin ELISA, IDK hemoglobin/haptoglobin complex ELISA, NS‑Prime and QuikRead go. No studies were found for IDK TurbiFIT. So, the committee recommended that HM-JACKarc and OC-Sensor could be used for FIT. It recommended further research on the clinical effectiveness (including diagnostic accuracy) of FOB Gold, IDK TurbiFIT, IDK hemoglobin ELISA, IDK hemoglobin/haptoglobin complex ELISA, NS‑Prime and QuikRead go.

Factors that could affect the performance of FIT

3.4 There was not enough evidence to make any recommendations on changes to how FIT should be used for people with different characteristics that could affect the performance of the test. During scoping, clinical experts suggested that factors such as age, sex, ethnicity, anaemia, or medications or conditions that change the risk of gastrointestinal bleeding, could influence the threshold that should be used to guide referral, or affect the diagnostic accuracy of the test. Some people may also have difficulty providing samples because of cognitive or physical disability. The EAG found limited evidence in these subgroups, and no conclusive evidence to determine whether FIT should be used differently in these groups. The EAG and committee members also noted that ethnicity and disability are generally poorly recorded in studies of FIT. A committee member suggested further research could be recommended for some subgroups. But, the committee noted that evidence is already developing in this area, with algorithms such as COLOFIT that incorporate multiple factors alongside a FIT result. This should address some of these uncertainties and allow these factors to be considered alongside a FIT result.

Uptake of faecal testing

3.5 The committee reviewed evidence showing differences in the rate of return of FIT between sociodemographic groups based on age, sex, ethnicity and socioeconomic status. The EAG highlighted publications that proposed strategies to help encourage test return in these groups, such as following up after a sample is not returned, providing information in multiple languages, or providing counselling and education services. But, it was not clear which methods would be the most effective, and different methods may be more appropriate for different groups. Therefore, the committee recommended social research to determine the best way to improve return of FIT, especially from groups in which engagement is less likely.

3.6 A patient expert suggested that healthcare professional involvement is important to drive engagement with testing. GP experts noted that the ability of primary care healthcare professionals to provide support is limited by workload. They noted that support would be hardest to implement in the most underserved areas where engagement with testing is likely to be lower. Guidance or educational resources to help improve test uptake would be helpful to minimise geographical differences in care.

Dual FIT

3.7 Dual FIT was considered as a testing strategy. This uses 2 separate faecal samples, usually collected within 2 weeks of each other, and a positive result from either sample would indicate a referral to secondary care. The committee considered evidence from the EAG's clinical-effectiveness review that found that dual FIT generally improved sensitivity but decreased specificity compared with single FIT at the same threshold. Some committee members suggested that a dual FIT strategy could reduce the risk of missing people with cancer and reported positive experiences with dual FIT in their local practice. Clinical experts noted that FIT results can vary between bowel movements because bleeding can be intermittent. However, the committee recalled that certain groups are less likely to return a sample. It was concerned that asking for 2 samples could particularly affect these groups (see section 3.5). This could increase inequality in access to healthcare. Experts also highlighted that there could be additional implementation issues if twice as many sample kits were needed, such as increased reliance on mail services or GP capacity.

Inflammatory bowel disease

3.8 Several conditions other than colorectal cancer can cause gastrointestinal symptoms and blood in faeces, including inflammatory bowel disease (IBD). IBD (Crohn's disease or ulcerative colitis) is also usually diagnosed in secondary care through investigations such as colonoscopy. The EAG's clinical review found that the estimates of the diagnostic accuracy of FIT for IBD were more uncertain than those for colorectal cancer, and the sensitivity was generally lower. However, clinical experts did not think that introducing FIT would have a substantial effect on people who have IBD, because GPs are likely to order a calprotectin test at the same time as FIT, which is a more accurate test for IBD (see NICE's guidance on faecal calprotectin diagnostic tests for inflammatory diseases of the bowel). Because the focus of this assessment was using FIT to guide referral pathways for colorectal cancer, other methods of detecting IBD were not considered.

Cost effectiveness of FIT

3.9 The committee agreed that using FIT for all people with signs or symptoms suggestive of colorectal cancer (excluding people with rectal mass, see recommendation 1.1) was likely to be cost effective compared with using FIT as outlined in NICE guidance (see section 2.3). The economic model estimated that all testing strategies using HM‑JACKarc or OC‑Sensor were cost effective. This was because costs were saved by reducing the overall number of colonoscopies, but there was also a very small loss of health resulting from people who had false negatives from their FIT test. The EAG stated that the QALY loss was equivalent to less than 1 day of full health for all people in the cohort. The committee noted that the model predicted that reducing the number of colonoscopy referrals would likely reduce secondary care waiting times for most people. However, the average time to diagnosis was increased overall because some people with false negative FIT results would have very long waiting times.

Assumptions in the economic model

3.10 The committee agreed that the overall conclusions of the economic model were reasonable. However, there was uncertainty in specific cost-effectiveness estimates because many inputs were based on clinical expert opinion when evidence was not available. Some committee members thought that the times to diagnosis used in the base case were pessimistic. But a scenario analysis that used shorter times to diagnosis resulted in a more favourable cost-effectiveness estimate for FIT than in the base case. Primary care experts thought that the number of additional GP appointments for people in primary care was too low, but not so low that the overall conclusion of cost effectiveness would be changed. The proportion of people who would be referred to secondary care despite a negative FIT result was based on clinical experts' experience with existing guidance (see sections 2.3 and 2.5), which recommends a threshold of 10 micrograms of haemoglobin per gram of faeces. Clinical experts thought that using a higher threshold would reduce physician confidence in the test. As a result, the proportion of people being referred without a positive FIT result would be higher than modelled. Therefore, the cost-effectiveness results at higher thresholds were more uncertain.

Testing strategies

3.11 The committee decided that testing a single faecal sample and using a single threshold to inform referral decisions was the best strategy. It noted that the economic model predicted that using dual FIT would be slightly less cost effective than single FIT, but would also reduce the QALY loss from false negatives. However, it recalled that dual FIT could disadvantage groups that are less likely to return samples, and introduce additional implementation issues (see section 3.5 and section 3.7). The committee concluded that drawbacks of dual FIT were likely to outweigh the potential benefits of increased sensitivity. The committee noted that using 2 thresholds appeared slightly less cost effective than using 1 threshold. Clinical experts also advised that using 2 thresholds would complicate referral decisions and make it harder to understand what the results mean in practice, which may reduce cost effectiveness more than predicted by the model.

Choice of threshold

3.12 The committee concluded that a threshold of 10 micrograms of haemoglobin per gram of faeces should be used to guide referral decisions. It acknowledged that the economic model suggested a threshold of 100 micrograms of haemoglobin per gram of faeces would be most cost effective. However, the committee recalled that the cost-effectiveness estimates at higher thresholds were more uncertain (see section 3.10). Thresholds below 10 micrograms of haemoglobin per gram of faeces were not considered. This was because they were less cost effective and approached the limits of quantitation for many of the tests, which may reduce the reliability of results (see section 2.8).

3.13 Economic experts highlighted that cost-effectiveness estimates improved the most between lower thresholds. They suggested that moving to a threshold of 20 micrograms of haemoglobin per gram of faeces could produce a gain in cost effectiveness without losing physician confidence. Clinical experts disagreed that physicians would accept a higher threshold because of the risk of false negative results, but conceded that there was no evidence on how the choice of threshold affects decision making. So, the committee recommended further research on the clinical impact of using different thresholds to guide referral to understand whether referrals would decrease by a similar proportion as predicted by the model.

Implementation of safety netting

3.14 The committee discussed safety netting for people with negative FIT results and ongoing symptoms. The committee stated that clear guidance will be needed to ensure that safety netting is implemented consistently and effectively, and noted that NICE's guideline on suspected cancer (NG12) includes guidance on this. Clinical experts highlighted that the exact form of safety netting available is likely to differ across the UK. The implementation of safety netting used in the model for people with negative FIT results or who did not return a test was based on clinical advice. Options included:

3.15 Clinical experts emphasised that having a positive FIT result should not be an absolute requirement for referral to secondary care. This is because it is possible to have a false negative result and some people may not be able to complete a test, either because of physical or cognitive disability or because of barriers to test uptake. So, the option to refer should always be available should GPs think it is needed.