Have large increases in fast track referrals improved bowel cancer outcomes in UK?BMJ 2020; 371 doi: https://doi.org/10.1136/bmj.m3273 (Published 10 November 2020) Cite this as: BMJ 2020;371:m3273
- Michael Thompson, consultant colorectal surgeon1,
- Daniel O’Leary, consultant colorectal surgeon1,
- Iona Heath, retired general practitioner2,
- Lynn Faulds Wood, patient representative2,
- Brian Ellis, general practitioner1,
- Karen Flashman1,
- database manager Portsmouth outpatient database,
- Neil Smart, associate professor of colorectal surgery3,
- John Nicholls, emeritus consultant surgeon professor of colorectal surgery4,
- Neil Mortensen, president of the Royal College of Surgeons of England5,
- Paul Finan, professor of colorectal surgery6,
- Asha Senapati, consultant colorectal surgeon1,
- Robert Steele, chair UK National Screening Committee7,
- Peter Dawson, senior lecturer of colorectal surgery4,
- James Hill, professor of colorectal surgery8,
- Brendan Moran, consultant colorectal surgeon9
- 1Queen Alexandra Hospital, Portsmouth, UK
- 2London, UK
- 3University of Exeter Medical School, Exeter, UK
- 4Imperial College, London, UK
- 5John Radcliffe Hospital, Oxford, UK
- 6St James’s University Hospital, Leeds, UK
- 7University of Dundee, Dundee, UK
- 8Manchester University, Manchester, UK
- 9North Hampshire Hospital, Basingstoke, UK
- Correspondence to: M R Thompson
UK Department of Health policies to improve survival from bowel cancer through GP referral guidelines and public awareness campaigns have increased urgent referrals to hospitals. This has led to an unsustainable demand for colonoscopy and CT colonography without evidence of significant clinical benefit. These policies could be improved by more precise stratification of the risk of having bowel cancer to achieve prompt, rather than earlier, diagnosis while avoiding over-referral and investigation of patients with transient symptoms from benign conditions.
Development of GP referral guidelines and fast track clinics
The first UK cancer plan was developed in 2000 to improve outcomes for patients. The plan introduced the concept of the “two week wait,” from urgent GP referral to the first outpatient appointment. In 2000, the committee developing the referral criteria for bowel cancer1 advised that only those with symptoms persisting for 6 weeks should be referred. The 2005 National Institute for Health and Care Excellence (NICE) review of the guidelines largely endorsed the referral criteria, which identified nine out of 10 bowel cancers (box 1) and had a 9-14% predictive value for cancer34 (table 1). Fast track clinics reduced time to diagnosis but not to treatment.34 Over the following nine years the number of referrals increased by 45% (table 1) while the predictive value for cancer decreased from 14% to 8%34567 (table 1). A review of several articles was carried out in 2001-04, with one study reporting a 14% predictive value for cancer and another, by 2009, before the introduction of the public awareness campaign, reporting a predictive value of 7.9%.
Seven typical characteristics of bowel cancer2
These characteristics identified over 92% of bowel cancers referred to outpatient departments.
Adding two age thresholds: >40 years for the first symptom combination and >60 years for the second and third combinations still identified over 90% of cancers.
>50% of all patients with bowel cancer present with rectal bleeding in combination with a persistent and unremitting change in bowel habit to increased frequency of defecation and looser stools (odds ratio 5.4*)
>20% have a similar change in bowel habit without rectal bleeding (odds ratio 2.1-2.5*)
>12% have persistent rectal bleeding without anal symptoms without a change in bowel habit (odds ratio 2.9*)
<2% have abdominal pain (odds ratio 0.6†) as a single symptom, which is always provoked by eating, causing weight loss
25% have a rectal mass (odds ratio 31.5)
12% have an abdominal mass (odds ratio 1.8-8.5*)
8% have an iron deficiency anaemia with or without bowel symptoms (odds ratio 4.4-8.4*).
*Odds ratio on multivariate analysis.
†Odds ratio of abdominal pain on univariate analysis without defining whether it is always associated with eating and weight loss.
Because survival from cancer was still lower in the UK than in most developed countries,11 public awareness campaigns were introduced in 201112 and NICE issued new guidelines in 2015,13 liberalising the referral criteria with a 3% threshold for risk of cancer. It was felt that this change would neither overwhelm clinical services, nor greatly increase the possible harm to patients from overinvestigation.14 These changes were applied to bowel cancer, even though a review,115 published before the introduction of the original guidelines, had found little evidence of benefit from earlier diagnosis of symptomatic bowel cancer. The public awareness campaigns resulted in a 62-77% increase in referrals,567 and falls in predictive values for cancer to 4-9%56789 (table 1). The new NICE referral criteria, as intended,1314 resulted in a further 78-100% increase in referrals89 with reductions in positive predictive values8910 to 3% (table 1).
Some subgroups of patients who are referred have a risk of cancer well below the 3% threshold, including those whose symptoms have resolved or have had a previously normal colonic investigation,10 and as expected by previous studies,516 those presenting with single symptoms without a mass or iron deficiency anaemia.
Effect on general practice
Public awareness campaigns, by not fully taking into account the concept of the “symptom iceberg,”17 have greatly increased the numbers of the worried well seeing their GPs in primary care.18 It is important to understand that the number of people seeing their GP with bowel related symptoms is only a small fraction of all those who have these symptoms. Thus even small increases in the proportion of people seeking advice could result in a greatly increased workload for GPs. This factor, together with the view, encouraged by the two healthcare policies,1213 that even modest delays in referral affect survival,19 makes it more difficult for GPs to manage people at low risk of having cancer.1820
Effect on hospital services
To cope with the increase in fast track referrals, hospitals initially provided additional outpatient clinics. However, capacity was soon overwhelmed, leading to strategies designed to reduce face-to-face consultations, including telephone triage, nurse led clinics, and booking people straight to test.821 Such approaches only transferred the increased workload from clinics to radiology and endoscopy departments.9 Increases in requests for CT colonography, which now exceed requests for colonoscopy (2:1 ratio in Portsmouth), require outsourcing and training of advanced practitioner radiographers.22
The 3% risk threshold means that a clinician would need to see 32 patients to diagnose one cancer. A standard of eight patients per clinic would require four clinics taking a clinician 20 hours, not including the additional time needed for patients having whole colonic imaging.
Numerous audits presented at the British Society of Gastroenterology and Association of Coloproctology of Great Britain and Ireland attest to a worsening of these problems.
Effect on bowel cancer outcomes
The 2000 and 2005 NICE guidelines, increased referrals but did not lead to significant improvements in cancer stage,3423 or survival.624 The 2015 NICE 3% risk threshold was introduced for a range of cancers to widen the net, on the basis that this would lead to a greater number of early referrals and facilitate diagnosis of cancer at an earlier, more curable, stage.131425 The NICE guideline development group states that “The lower the threshold could reasonably be set, the more patients with cancer would have expedited diagnoses, with accompanying improvements in mortality and morbidity.”14
The new guidance was mainly based on observational studies,25 which suggested that reducing diagnostic delay in primary care improved survival. The evidence for bowel cancer as compared with breast and lung cancer was less clear,1925 however, and it was accepted that there was no evidence that improvements in survival would be cost effective.25
Bowel cancer screening in the UK by faecal occult blood testing26 and flexible sigmoidoscopy is offered to patients aged between 50 and 74. A positive test in bowel cancer screening, with a 10% yield of cancer achieves a relatively small 15% cost effective improvement in survival.26 It is unlikely that investigation of more patients with a 3% risk of having cancer at later stages in their natural history,27 when more will have become incurable, will result in similar benefits.
Five years after broadening the guidelines, a study on 14 026 patients seen in fast track clinics before and after the introduction of the 2015 NICE guidelines showed no improvement in the stage of cancer.28
The review undertaken in 2000,15 before the introduction of two week clinics, included a prospective study of 5173 patients, which showed that almost half of all deaths from bowel cancer occurred in those treated without delay, and of the 60% who survived, many had had long delays. This suggests that the critical point at which most bowel cancers become incurable is either before, or long after, the onset of symptoms, not in the first few weeks.27 Thirty nine other studies were also reviewed,15 of which 16 showed worse survival after earlier diagnosis, five better survival, and 18 no difference. The review also noted that delays to treatment are the same in the UK as in countries reporting better five year survival,15 making it unlikely that delay is a major cause of worse survival in the UK. Delay from the onset of symptoms to treatment is shorter in patients with bowel cancer who are referred as an emergency rather than to outpatient departments (2.1 v 7.2 months),15 which supports the view that the biological behaviour of a cancer is the major factor in determining time to diagnosis1929 and survival.2930 Although there has been an overall reduction in emergency admissions since 2013,31 the 36% also requiring emergency surgery32 has not decreased, and as most other patients already have shorter delays to treatment15 it is unlikely that diverting these patients to two week clinics will significantly improve their chance of survival.33
Effective and efficient management of symptomatic patients
Diagnosis of symptomatic bowel cancer must be efficient as well as effective.34 Precise stratification of the risk of bowel cancer is essential to identify those at higher risk, with one of its seven typical characteristics (box 1) for prompt referral and investigation. It is also as important to identify people at low risk, without one of these characteristics, for longer treatment in primary care, with later referral if symptoms persist.
Many studies have identified the typical presentations of symptomatic bowel cancer, which have not changed over many years.1 Combinations of symptoms have greater predictive value than a single symptom.135 Studies in primary1636 and secondary care,2 based on a structured history, show that over 85% of bowel cancers present with one of four age/symptom combinations (box 1). Fifty per cent present with a combination of rectal bleeding and a persistent change in bowel habit to increased frequency of defecation and looser stools.21636 Over 90% of patients present with one of three age/symptom combinations and an abdominal or rectal mass or iron deficiency anaemia.2 Only 135 (1%) of 12 605 patients referred to hospital without these characteristics had bowel cancer.2 These studies show it is now possible to stratify patients according to risk simply on the basis of a structured history, clinical examination, and the results of a blood test for iron deficiency anaemia, in both primary1636 and secondary care.2
Investigations must be used appropriately. The ideal investigation is whole colonic imaging. Outpatient flexible sigmoidoscopy identifies around 85% of bowel cancers37 and as most of the remaining 15% of cancers have an abdominal mass, iron deficiency anaemia, or early symptoms of colonic obstruction, the diagnostic yield of whole colonic imaging in patients without these characteristics can be as low as 0.1% (6/4132).2 Iatrogenic harm is likely to exceed any small benefit from immediate whole colonic imaging in these patients. Flexible sigmoidoscopy, with highly selective whole colonic imaging and later referral only if symptoms persist, could safely reduce the need for these investigations.
Many people in the community with symptoms associated with cancer may benefit from more nuanced information, which can only be given by internet decision support tools. In primary and secondary care, internet decision support based on referral guidelines that aim to identify or exclude the seven red flag symptoms and signs of bowel cancer, could help health professionals to confidently identify and advise patients at low risk of having cancer that it is safer to “treat, watch, and wait” and accept some degree of uncertainty38 rather than to have immediate referral for invasive investigations that might cause harm. Many patients’ symptoms resolve within six weeks.1 If symptoms do not resolve and remain low risk, patients, particularly those having previous normal colonic imaging,10 should be referred to routine gastroenterological or surgical clinics for reassurance and specialist advice on the treatment of their symptoms. Giving written information to patients may reinforce messages about provision of a safety net20 and make clear the need to seek further advice if their symptoms persist or recur.
Another approach to stratification for risk in symptomatic patients, which is rapidly becoming more accepted, is the use of the faecal immunochemical test. NICE diagnostic guidance 3039 recommends the use of the faecal immunochemical test, reported at a threshold of less than 10 µg haemoglobin/g faeces, for patients without rectal bleeding and unexplained symptoms who do not meet the criteria for referral for suspected cancer.39 Recent work suggests that the faecal immunochemical test can effectively rule out cancer, even in patients with higher risk symptoms who do meet referral criteria.40 A large observational study showed a reduction in referrals to secondary care of 15% in the first year of triaging with a faecal immunochemical test, with a sensitivity for cancer of 90.5% and a specificity of 48.6%.41 As a small number of patients with cancer have a false negative test, it is still necessary to provide a safety net for these patients.20 A universal protocol for this has not yet been designed and more evidence of the safety and cost effectiveness of the faecal immunochemical test is required before its routine use in primary care.
It is important that patients and the public are involved in developing safe diagnostic strategies for those at low risk of having cancer. People, particularly those who have had bowel cancer, can contribute to the development of referral guidelines and internet decision support for others in the community.
Increasing diagnosis of asymptomatic cancers
Bowel cancer typically has a long asymptomatic phase. Many cancers become incurable before the patient develops any symptoms.27 In contrast, diagnosis of cancers by screening26 has been shown to detect early stage, more curable cancers. If the relatively poor uptake of screening and the sensitivity of the test could be increased—for example, by the faecal immunochemical test, this could greatly reduce death from bowel cancer.
UK public awareness campaigns in 2011, and revision of the NICE GP referral guidelines in 2015, have substantially increased referrals to fast track clinics. These referrals are overwhelming hospital resources without producing the expected increases in survival. The policies did not take into account the high prevalence of bowel symptoms in the absence of underlying cancer, and failed to ensure that only those with persistently higher risk symptoms and signs were fast tracked for urgent investigation in hospital.
A range of approaches could be used in primary and secondary care to avoid excessive delays in the diagnosis of bowel cancer while safely reducing the number of people without cancer being referred to hospital and undergoing unnecessary investigations. The first approach is to improve risk stratification on the basis of a structured clinical history and examination, the second is internet decision support for people in the community as well as for doctors in primary and secondary care. In the future faecal immunochemical testing may further improve selection of patients. Validated internet decision support tools need to be developed and a trial of faecal immunochemical testing is under way.
Screening identifies patients with bowel cancer before they develop symptoms and represents higher value clinical activity.
Forty eight years ago Cochrane34 pointed out the importance of efficient as well as effective healthcare. Now more than ever, the right test for the right patient at the right time is an important goal for all.
The continuing increase in referrals to fast track, bowel cancer clinics and the consequent requests for whole colonic imaging are becoming unsustainable
Evidence that fast track clinics lead to diagnosis of earlier stage bowel cancer and improve survival is weak
Many people without bowel cancer worry unnecessarily and have potentially harmful investigations
Improved stratification of the risk of having bowel cancer could help to solve these problems and enable resources to be redirected to higher value clinical activity such as screening
Contributors and sources: All the authors have been involved in the development of services for patients with bowel cancer in the UK to improve outcomes and several have taken part in national and international meetings on bowel cancer over the past 40 years. LFW was cofounder/chair of the European Cancer Patient Coalition. She died before publication of the paper. MRT, IH, BE, LFW were on the committee that developed the first set of GP referral guidelines in 2000. IH, BE were former general practitioners in busy clinical practices. IH was former president of the Royal College of General Practice. BE had extensive experience in diagnosing bowel cancer in primary and secondary care by flexible sigmoidoscopy. JN, NM, MRT, PF, AS, RS, PD, JH, and BM are former presidents of the Association of Coloproctology of Great Britain and Ireland, a professional organisation whose members are responsible for the diagnosis of 80% of bowel cancers in the UK. NS is editor in chief, colorectal disease. DO is chair of the Colorectal Site Specific Group, South Central Cancer Network 2005-10. MT acts as guarantor.
Competing interest statement: We have read and understood BMJ policy on declaration of interests and declare that we have no competing interests.
Patient and public involvement: LFW was patient representative for the 2000 GP Referral Guidelines Committee, and for the Bowel Cancer Services Cost Benefits Report of 2007, and has also been president of the European Cancer Patient Coalition 2003-09.