INTRODUCTION
The disappointing performance of cancer outcomes compared with other high-income countries has been a focus of health policy in the UK for decades.1 Lung cancer is the leading cause of cancer death both worldwide and in the UK. Although improvements have been achieved in recent years, with the UK 5-year net survival for lung cancer increasing from 7.2% (95% confidence interval [CI] = 7.0% to 7.3%) in the period 1995–1999 to 14.7% (95% CI = 14.5% to 15.0%) in the period 2010–2014, these remain the poorest outcomes among comparable countries studied in an International Cancer Benchmarking Partnership (ICBP) study.2 Possible explanations for this disparity of the UK include adverse comorbidity and deprivation,3 longer durations before which patients seek assessment for symptoms (patient interval),4 and a greater reluctance among clinicians to organise investigations for symptoms.5 An additional possible factor is lower availability of computed tomography (CT) and greater reliance on the less sensitive chest X-ray (CXR), which might contribute to later-stage diagnosis in the UK.6,7
Guidance from the National Institute for Health and Care Excellence (NICE) advises GPs to investigate all patients with potential lung cancer with a CXR,8 other than those aged over 40 years with unexplained haemoptysis, for whom immediate referral on an urgent suspected cancer pathway is advised. In the UK, screening using CT to investigate asymptomatic patients aged 55–74 years identified as high risk because of smoking history has recently been approved by the National Screening Committee.9 Some have questioned whether symptomatic patients should have to rely on CXR given that the test may not identify approximately 20% of lung cancers, and whether CT should be made available as a first-line test for those presenting with symptoms, as well as a screening test.10,11
While we have a limited understanding of the performance of CT for detection of lung cancer in symptomatic patients compared with CXR,12 this is known for asymptomatic populations from the National Lung Screening Trial. The trial demonstrated a sensitivity and specificity of 93.8% and 73.4%, respectively, for CT compared with 73.5% and 91.3% for CXR.13 The performance of CT in symptomatic patients may differ from the screening context because of differences in lung cancer prevalence and severity of disease presentations. It should also be noted that these sensitivities were determined with reference to cancer diagnosis within 1 year. Therefore they could overestimate sensitivity, for example, if cancers were present at time of investigation but were not diagnosed within 1 year.
IS THE UK AN OUTLIER IN RELYING ON CXR?
We identified primary care guidelines on first-line investigation for suspected lung cancer with the help of colleagues worldwide and through guideline databases including the National Guideline Clearinghouse, Trip, and the World Health Organization (Table 1). Where possible we included guidelines from national bodies equivalent to NICE. Where none were identified, we included guidance from jurisdictions from within countries and guidance issued from primary care professional associations or consensus statements. Guidelines issued from April 2012 to April 2022 were considered.
The guidelines identified were remarkably consistent in advocating CXR as the initial investigation, aside from Danish guidance (Table 1). These included countries that utilise CT more readily than the UK, such as Australia and the US.14
Australian15 and Canadian (British Columbia)16 guidelines encourage immediate CT if there is a strong clinical suspicion of lung cancer. These guidelines, along with the US guidance,17 also advocate follow-up investigation with CT if symptoms persist or are unexplained following the CXR. Australian and New Zealand guidelines also make provision for repeat CXR if symptoms persist, for example, beyond 6 weeks. Australian guidance in particular confers a great deal of autonomy to GPs in the selection of appropriate testing strategies based on clinical intuition that includes CT, and CXR with or without repeat imaging of either modality.
By contrast, NICE guidance, which pertains to England and Wales, does not outline further steps to be taken by GPs following an unremarkable CXR in the context of persistent or concerning symptoms. However, Scottish guidelines do advise that, ‘Referral to the respiratory team is required if risk factors and symptoms raise the possibility of cancer even if the chest X-ray is normal.’18,19 A recent report has called for NICE to issue additional safety-netting advice to GPs that ‘should make clearer what should be offered to patients who have ongoing, unexplained symptoms after a negative chest X-ray’.11
None of the guidelines addressed how decisions about investigation for symptoms might, or should, be influenced by participation in lung cancer screening with CT. For example, if a patient has had a recent negative CT screen and subsequently develops symptoms, how long, if at all, should a clinician wait until organising repeated investigation and with which modality?
CONCLUSION
Despite concerns that the UK strategy for symptomatic detection of lung cancer is excessively reliant on CXR, there is a remarkable degree of concordance in guidelines internationally. While guidelines are one factor informing practice, adherence is likely to vary within and between jurisdictions. The responsibility of GPs to exercise their own clinical judgement in addition to the result of CXR is explicitly referenced in several guidelines. Which presentations should provoke a high suspicion of lung cancer, aside from patients with haemoptysis, is not precisely defined within any of the guidelines identified. Deploying clinical intuition to identify those patients who warrant additional investigation is a core competency of GPs that cannot be replaced by protocols or algorithms. However, the challenge this presents is arguably greater in settings where access to CT is limited, thereby demanding careful stewardship of available radiology resource.
Although there is a surprising degree of consistency in the initial imaging recommended in the guidelines identified, it is likely the availability of CT affects investigation practice. Countries with fewer resource constraints, and with the ability to request investigations directly, permit GPs in those settings to undertake more liberal imaging strategies, in contrast to the UK where access is much more limited.
It is plausible that the propensity of doctors to arrange imaging for patients with possible lung cancer symptoms could affect outcomes. While no such evidence to our knowledge is available regarding CT, there is an equivocal evidence base with respect to CXR. O’Dowd et al found practices that used CXR most frequently had higher risk of death within 90 days of lung cancer diagnosis, while Kennedy et al did observe a beneficial stage shift corresponding to a symptom awareness campaign that led to increased utilisation of CXRs.20,21 If increased CXR utilisation could help expedite lung cancer diagnosis, the success of such a strategy would require those tested to have some degree of risk, as manifested by symptoms, rather than being deployed indiscriminately.
The implementation of lung cancer screening in several countries, including the UK, will lead to earlier detection for some patients. However, it is crucial that patients and GPs recognise that asymptomatic screening will not lessen the need for prompt investigations to be arranged for those with symptoms. Even where screening programmes have been introduced, the majority of diagnoses are likely to continue to follow symptomatic British Journal of General Practice, February 2023 presentation. This is because a large proportion of those with the disease would not have been eligible for screening, and only around half of those who are eligible participate in screening.22,23
The extent to which alternatives to CXR, such as CT, are accessible to GPs is likely to be a crucial determinant of the extent to which these are used. Whether the wider use of CT in symptomatic patients will yield sufficient benefits in terms of earlier diagnoses to offset the potential harms, such as overdiagnosis, and whether such policies would prove cost-effective remain unknown. This is an important policy question that warrants exploration in a clinical trial in symptomatic populations with comprehensive health economic analysis.
Acknowledgments
We are indebted to Brian Nicholson who suggested the idea for this article. We would like to thank Ambika Seth, Charles Helsper, Ole Hilberg, Simon Gordon, Sam Merriel, and Lucas Helal who provided advice on international guidelines on lung cancer diagnosis.
Notes
Funding
None.
Provenance
Freely submitted; externally peer reviewed.
Competing interests
Bobby SK Bhartia and Stephen H Bradley have applied to a funding call from NIHR Health Technology Assessment (HTA) for research on low-dose computed tomography compared with chest X-ray for lung cancer diagnosis in patients with symptoms (reference: NIHR HTA 22/98).
- © British Journal of General Practice 2023