5.1 The committee discussed the current standard of care for people who present to their GP with symptoms, but who are at a low risk of having colorectal cancer. It noted that recommendations had previously been made for using faecal occult blood tests to triage referral for this group of people in NICE's guideline on suspected cancer, but heard from clinical specialists that this had not been widely adopted in primary care. It heard that people are typically assessed by taking into account their clinical history, their current symptoms and the results of any available tests to establish whether there is a clinical suspicion of colorectal cancer. If clinical suspicion is sufficiently raised, referral using a suspected cancer pathway for an appointment within 2 weeks is appropriate. The committee noted that in practice clinical history and current symptoms of people in this low-risk group were likely to be heterogeneous, and so management was likely to vary between individual patients and GPs. The committee concluded that current clinical practice is likely to vary because of the complexity of managing and investigating non‑specific symptoms in practice.
5.2 The committee discussed the potential benefits that may be associated with using faecal immunochemical tests in primary care. It heard from clinical specialists that faecal immunochemical tests are thought to be more accurate than guaiac-based faecal occult blood tests because they use immunochemical detection methods that are specific to human haemoglobin. They are also suitable for use with automated analysers, which allow high-throughput batch testing. It also heard from a patient specialist that faecal immunochemical tests often had sample collection devices that are easier to use than guaiac-based faecal occult blood tests and they need fewer samples, which makes them more acceptable to people and so may increase test uptake. The committee concluded that faecal immunochemical tests may have substantial analytical and practical advantages over guaiac-based faecal occult blood tests.
5.3 The committee discussed the evidence base for the clinical effectiveness of the quantitative faecal immunochemical tests for triaging people with symptoms who are at a low risk of colorectal cancer. It noted that there were 10 included studies that reported data for the OC sensor, HM‑JACKarc and FOB Gold assays. It also noted that none of the included studies reported data for the RIDASCREEN haemoglobin or the RIDASCREEN haemoglobin/haptoglobin assays. It therefore concluded that, in the absence of any data, the RIDASCREEN assays could not be considered further.
5.4 The committee questioned whether the data from the studies included in the clinical-effectiveness review were generalisable to the population in the decision problem for this assessment. It noted that many of the studies were done in secondary care and included people with higher-risk symptoms than those previously outlined in NICE's guideline on suspected cancer (see section 2.6). The committee heard from clinical specialists that although the populations in the studies were likely to have a higher prevalence of colorectal cancer, the data provide evidence that faecal immunochemical tests are likely to be highly sensitive for detecting haemoglobin in faecal samples. The committee concluded that although the differences in the populations introduced some uncertainty into the analysis it was reasonable to include the data in the review.
5.5 The committee discussed the diagnostic accuracy data in the included studies. Many reported diagnostic accuracy estimates at multiple thresholds and showed that there was a clear threshold effect; when lower thresholds were used the sensitivity of the test increased, but the specificity decreased. The committee noted that the external assessment group (EAG) had used the diagnostic accuracy data to establish the threshold for the best diagnostic performance. The committee heard from clinical specialists that they supported the EAG's conclusion that a threshold of 10 micrograms of haemoglobin (Hb)/g faeces seemed to give the best diagnostic performance for ruling out colorectal cancer. It also heard that using a threshold lower than 10 micrograms Hb/g faeces would create more variability in test performance because the assays are known to be more imprecise in their lower measuring range, and the availability of quality control materials to validate the detection of low levels of haemoglobin may also be limited. The committee concluded that using a threshold of 10 micrograms Hb/g faeces gave the test enough sensitivity to reliably rule out colorectal cancer in primary care.
5.6 The committee considered the consequences of false-positive faecal immunochemical test results in practice. It heard from clinical specialists that using faecal immunochemical tests in primary care to triage suspected colorectal cancer referrals could result in unnecessary colonoscopy referrals. Data in the clinical-effectiveness review suggested that around 200 of every 1,000 people tested do not have colorectal cancer but are referred for colonoscopy. The committee noted that the tests detect a marker of colorectal cancer (haemoglobin), which could also be associated with a range of other conditions. Data from studies reporting diagnostic accuracy for multiple target conditions in the same population suggested that up to 28.9% of people with a false-positive faecal immunochemical test result for colorectal cancer had bowel pathology, such as inflammatory bowel disease or high-risk adenoma. The committee concluded that it was plausible that the number of false-positive results that occur when using the tests to rule out colorectal cancer could be partially offset by detecting other treatable bowel pathology.
5.7 The committee noted that the review did not find any data that directly compared the OC Sensor, HM‑JACKarc or FOB Gold assays. Further, it noted that the review had found substantially less data for the FOB Gold assay so the EAG could not establish the optimal threshold for this test. The committee therefore examined whether the conclusions reached based on data from the OC Sensor and HM‑JACKarc assays could be extended to the FOB Gold assay. It heard from clinical specialists that the 3 assays were likely to perform similarly in detecting haemoglobin in faecal samples. It also heard that the FOB Gold assay was compatible with a range of clinical chemistry analysers, which may be an advantage for some laboratories. The committee concluded that there was more uncertainty in the clinical effectiveness of the FOB Gold assay but that the available data suggested that it was likely to perform similarly to the OC Sensor and HM‑JACKarc assays in practice.
5.8 The committee discussed the possibility of faecal immunochemical tests performing differently in the population subgroups outlined in the decision problem. It heard from clinical specialists that it is plausible that the diagnostic accuracy of faecal immunochemical tests will differ according to age and sex because it is thought that women have lower levels of faecal haemoglobin than men and that levels are higher in older people. The committee noted that only 1 study reported data for men and women separately; this study showed that faecal immunochemical testing was more sensitive in men, but more specific in women. No studies were found that showed data by age. It also heard that both faecal immunochemical tests and guaiac-based faecal occult blood tests may have high false-positive rates in people who are taking medicines that increase their risk of gastrointestinal bleeding, such as oral anticoagulants, antiplatelets or aspirin, but no data were available for this group. The committee concluded that there were insufficient data at present to determine whether different thresholds are needed for women and older people, or whether faecal immunochemical tests help with clinical decision-making when people are taking medicines known to cause gastrointestinal bleeding.
5.9 The committee considered the cost-effectiveness analyses for the OC Sensor, HM‑JACKarc and FOB Gold assays. It noted that 2 comparators had been included; guaiac-based faecal occult blood testing and no triage (that is, direct referral to colonoscopy). The committee discussed which comparator was the most appropriate. It heard from clinical specialists that guaiac-based faecal occult blood testing is no longer done by most clinical chemistry laboratories so primary care clinicians are not able to request the test. It also heard that people in the population outlined in the decision problem were unlikely to be directly referred for colonoscopy, and that a watch-and-wait strategy is most often used by primary care clinicians to monitor their condition. The committee heard from the EAG that a watch-and-wait strategy had not been included in the model because there were not enough data available to characterise the variations in clinical decision-making in this group. The committee concluded that, although the model did not fully capture current practice, the comparisons it made reflected the best available data for the population included in the assessment.
5.10 The committee discussed the likely consequences of false-negative results and whether they could affect a person's prognosis by delaying their diagnosis of colorectal cancer. It noted that the economic model assumed that all people with a false-negative result were subsequently diagnosed within 12 months. It heard from clinical and patient specialists that delayed diagnosis could lead to worse outcomes, but clinical specialists advised that if symptoms persisted a referral to secondary care would be made regardless of a previous negative test result. The committee concluded that the analysis had sufficiently captured the likely prognostic implications of false-negative test results.
5.11 The committee considered the assumptions made in the model when comparing faecal immunochemical tests with guaiac-based faecal occult blood tests. It noted that none of the studies in the clinical-effectiveness review compared faecal immunochemical tests and guaiac-based faecal occult blood tests, so indirect comparisons had to be modelled. The committee discussed the diagnostic accuracy estimates used in the model (see table 4) and noted that they suggested that the faecal immunochemical tests were more sensitive than guaiac-based faecal occult blood tests. It heard from clinical specialists that the conclusions drawn from the indirect comparisons were supported by direct comparative data from bowel cancer screening programmes, which have shown that faecal immunochemical tests are more sensitive than guaiac-based faecal occult blood tests. The committee concluded that the assumptions made about the accuracy of guaiac-based faecal occult blood tests were reasonable.
5.12 The committee considered the base-case analysis. It noted that the EAG had not included the FOB Gold assay in this analysis because no data were available for the accuracy of the assay at a threshold of 10 micrograms Hb/g faeces. In the comparison with the guaiac-based faecal occult blood test both the OC Sensor and the HM‑JACKarc assays were cost effective, with probabilistic ICERs of £5,040 and £14,600 per QALY respectively. Both assays were also cost effective when compared with no triage, with the HM‑JACKarc dominating (that is, it was more effective and less expensive) and the OC Sensor having an ICER of £2,580,000 saved per QALY lost. The fully incremental base-case analysis suggested that the OC Sensor was more cost effective than the HM‑JACKarc, but the committee noted that this comparison was driven by small differences in both costs and QALYs. The committee considered the scenario analyses that included the FOB Gold assay and noted that although there was more uncertainty in the clinical effectiveness of this technology it appeared to be cost effective, with an ICER of £15,700 per QALY gained when compared with guaiac-based faecal occult blood testing at a threshold of 6.8 micrograms Hb/g faeces. The committee concluded that the OC Sensor, HM‑JACKarc and FOB Gold assays had the potential to be cost-effective options for triaging referrals in primary care for people with symptoms but a low risk of colorectal cancer.
5.13 The committee considered the drivers behind the cost savings seen when the faecal immunochemical tests were compared with no triage and noted that a reduction in colonoscopy was a key parameter. It was aware that the comparison made between faecal immunochemical tests and no triage assumes that all people have either colonoscopy or CT colonography to investigate the cause of their symptoms. The committee discussed colonoscopy capacity and whether the cost savings seen in the analysis for this comparison would be realised in practice. It heard from clinical specialists that colonoscopy capacity is very limited in many areas, and in practice it would be unlikely that all people who are at a low risk of colorectal cancer would be referred for colonoscopy. The committee concluded that the cost savings seen in comparisons made between faecal immunochemical tests and no triage could not be considered robust.
5.14 The committee discussed the uncertainties in the cost-effectiveness analysis and noted that the cost effectiveness of faecal immunochemical tests was sensitive to the prevalence of colorectal cancer which influences the pre-test probability. This parameter drives the accuracy of the tests and so the costs and resource use. It noted the scenario analyses that used prevalence values of 3% and 5.4% compared with 1.5% in the base case. The results of these scenario analysis showed that when the prevalence was increased the faecal immunochemical tests became more cost effective when compared with guaiac-based faecal occult blood testing, and when the prevalence of colorectal cancer was decreased the faecal immunochemical tests became less cost effective. The committee therefore considered that if the faecal immunochemical tests are used in a wider population in practice, the prevalence of colorectal cancer will be reduced and the tests may no longer be cost effective. The committee concluded that the tests are likely to be cost effective when used alongside clinical judgement and the results of any other testing to guide referral for suspected colorectal cancer in people without rectal bleeding who have unexplained symptoms and are at low risk. Further, it recommended that where false-negative results are suspected, active monitoring (safety netting) should be used as recommended in NICE's guideline on suspected cancer.
5.15 The committee discussed the possible advantages of using quantitative faecal immunochemical tests. It heard that the ability of the tests to report the concentration of faecal haemoglobin instead of providing a semi-quantitative positive or negative result could have additional clinical uses. It heard from a clinical specialist that, in some areas of Scotland, quantitative faecal immunochemical tests have been adopted to triage referrals from primary care and that the faecal haemoglobin concentration is being used in secondary care to decide who should have a colonoscopy most urgently. The committee also heard that the prognostic-risk scoring tools, highlighted in the clinical-effectiveness review (section 4.20) are a growing area of research that aims to produce validated tools that can identify people at increased risk of colorectal cancer using variables such as age, sex and faecal haemoglobin concentration. The committee concluded that the development of risk-prediction rules may further refine the use of faecal immunochemical tests in primary care.
5.16 The committee noted that the NHS bowel cancer screening programme is adopting faecal immunochemical tests as a replacement for guaiac-based faecal occult blood tests. It considered that there may be instances when people, who have recently had a negative screening result, may get a positive result in primary care after reporting symptoms because of the use of different thresholds in the 2 clinical scenarios. The committee concluded that practices adopting faecal immunochemical tests as a triage tool should take this into account when developing their implementation plans, and ensure that information to explain the different thresholds and their consequences is available for people who have recently participated in the bowel cancer screening programme. Information about the different test thresholds should also be made available to people taking part in the NHS bowel cancer screening programme.