Article Text
PDF
Abstract
Background Guidelines recommend urgent chest X-ray for newly presenting dyspnoea or haemoptysis but there is little evidence about their implementation.
Methods We analysed linked primary care and hospital imaging data for patients aged 30+ years newly presenting with dyspnoea or haemoptysis in primary care during April 2012 to March 2017. We examined guideline-concordant management, defined as General Practitioner-ordered chest X-ray/CT carried out within 2 weeks of symptomatic presentation, and variation by sociodemographic characteristic and relevant medical history using logistic regression. Additionally, among patients diagnosed with cancer we described time to diagnosis, diagnostic route and stage at diagnosis by guideline-concordant status.
Results In total, 22 560/162 161 (13.9%) patients with dyspnoea and 4022/8120 (49.5%) patients with haemoptysis received guideline-concordant imaging within the recommended 2-week period. Patients with recent chest imaging pre-presentation were much less likely to receive imaging (adjusted OR 0.16, 95% CI 0.14–0.18 for dyspnoea, and adjusted OR 0.09, 95% CI 0.06–0.11 for haemoptysis). History of chronic obstructive pulmonary disease/asthma was also associated with lower odds of guideline concordance (dyspnoea: OR 0.234, 95% CI 0.225–0.242 and haemoptysis: 0.88, 0.79–0.97). Guideline-concordant imaging was lower among dyspnoea presenters with prior heart failure; current or ex-smokers; and those in more socioeconomically disadvantaged groups.
The likelihood of lung cancer diagnosis within 12 months was greater among the guideline-concordant imaging group (dyspnoea: 1.1% vs 0.6%; haemoptysis: 3.5% vs 2.7%).
Conclusion The likelihood of receiving urgent imaging concords with the risk of subsequent cancer diagnosis. Nevertheless, large proportions of dyspnoea and haemoptysis presenters do not receive prompt chest imaging despite being eligible, indicating opportunities for earlier lung cancer diagnosis.
- Lung Cancer
- Imaging/CT MRI etc
- Clinical Epidemiology
Data availability statement
Restrictions apply to the availability of these data, which were used under licence for the current study, and so are not publicly available. For reusing these data, an application must be made directly to the Clinical Practice Research Datalink (CPRD; www.cprd.com).
This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Guidelines recommend prompt investigation of dyspnoea and haemoptysis in order to support early diagnosis of lung cancer but there is currently little evidence regarding how these guidelines are implemented.
WHAT THIS STUDY ADDS
Substantial proportions of patients newly presenting with dyspnoea or haemoptysis did not receive prompt imaging as recommended by clinical guidelines.
Guideline-concordant imaging was more likely in those later diagnosed with cancer, and less likely in patients who had recently had chest imaging and those with existing respiratory morbidities.
Among dyspnoea presenters, those with prior heart failure; current or ex-smokers; and those in more socioeconomically disadvantaged groups were also less likely to have guideline-concordant imaging.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Guideline-concordant imaging patterns are suggestive of appropriate clinical reasoning.
However, certain groups at higher risk of lung cancer were less likely to have urgent imaging.
Additionally, large proportions of individuals later diagnosed with cancer had not received urgent imaging indicating possible opportunities to improve earlier detection of lung cancer.
Introduction
International and regional variation in cancer outcomes indicates the need for improvement in lung cancer diagnosis.1–3 Lung cancer screening for high-risk individuals offers promise,4–8 but most patients are diagnosed with cancer via symptomatic pathways, typically starting in primary care.9 Nonetheless, achieving timely diagnosis of symptomatic lung cancer is challenging. The presenting symptoms of lung cancer are often non-specific, and other diagnoses such as chronic obstructive pulmonary disease (COPD), pneumonia and cardiac conditions may offer plausible alternative explanations.
In England and other countries, clinical guidelines have been developed to encourage the recognition and investigation of symptomatic individuals for suspected cancer in primary care.10–13 The guidelines recommend urgent referral (in England, via the ‘two-week wait’ fast-track referral pathway) or urgent primary care-led investigation for ‘red-flag’ symptoms with relatively high positive predictive value for cancer. For patients presenting with haemoptysis or persistent dyspnoea, urgent chest imaging is recommended.14 15
There is currently limited evidence about how primary care referral guidelines for suspected cancer operate in practice. A recent study found that three-fifths of patients with certain alarm symptoms (not including respiratory symptoms) were not referred in spite of guideline recommendations.16 Understanding guideline implementation could inform the development of quality indicators to improve the diagnostic process.17 With this goal in mind, we examined the proportion of patients with newly presenting dyspnoea or haemoptysis that received urgent chest imaging concordant with clinical recommendations and related variation by patient-level factors. Additionally, we examined patients diagnosed with cancer in the 12 months following symptomatic presentation by guideline-concordant status.
Methods
Study design and population
We conducted an observational cohort study using anonymous electronic patient records collected between 1 April 2012 and 15 March 2017. Primary care data from the Clinical Practice Research Datalink (CPRD) GOLD were linked with cancer registration data collated by the National Cancer Registry Analysis Service (NCRAS) and secondary care data, including the English Hospital Episode Statistics Diagnostic Imaging Dataset (HES-DID).
For individuals presenting with haemoptysis or ‘unexplained or persistent (longer than 3 weeks)’ dyspnoea, the 2005 National Institute for Health and Care Excellence (NICE) guidelines recommend that an urgent chest X-ray is carried out within 2 weeks.10 We defined two cohorts (each for dyspnoea and haemoptysis) using Read code lists to include individuals aged 30+ years if they had presented with either symptom at least 12 months following practice registration (see online supplemental appendix 1).18 19 Patients were excluded if their outcome status could not be confirmed, namely if they presented on or after 16 March 2017 (ie, within 2 weeks from the last reliable date in the available DID data); had left their CPRD practice or had died within 2 weeks of presentation; or if their practice had left CPRD within 2 weeks from their presentation.
Supplemental material
It was not possible to distinguish between patients who had experienced dyspnoea for 3 weeks or longer before consulting, and those consulting for new-onset dyspnoea as information on symptom duration was not available. Therefore, the first recorded occurrence in primary care was assumed to represent the first presentation of dyspnoea or haemoptysis, respectively.
Outcome of interest
A guideline-concordant imaging event was defined by applying the following three criteria to the linked DID data (see online supplemental appendix 2 for further details):
Imaging modality and body region: A chest X-ray or CT scan of the lung or chest using previously published National Interim Clinical Imaging Procedure (NICIP) and Systematised NOmenclature of MEDicine (SNOMED) code lists for such imaging investigations.20
Source of imaging referral: Imaging events ordered by a General Practitioner (GP) and/or from primary care.
Time from symptomatic presentation: Chest imaging that took place 0–14 days following symptomatic presentation was assumed to be relevant to the symptom.
Covariates of interest
Sex (male or female) and age group (10-year age bands from 30 to 39 years to 80+ years) were based on information in CPRD, and socioeconomic status (Index of Multiple Deprivation 2015 score quintiles based on patient postcode of residence) from linked national data. Ethnicity was categorised using information in the order of preference from HES inpatient, HES outpatient and HES-DID files and categorised as white, non-white or missing. Cancer diagnoses recorded in the 12 months following symptomatic presentation were based on national cancer registration data. Individuals were categorised as having been diagnosed with lung cancer; non-lung cancer excluding non-melanoma skin cancers (C44) and non-malignant tumours (D-code and in situ tumours); or no cancer. For patients with multiple tumours with the same diagnosis date, lung cancers and tumours with non-missing stage were prioritised over non-lung cancers and tumours with missing stage, respectively.
Each individual was categorised as a non-smoker, ex-smoker or current smoker based on primary care records prior to the date of symptomatic presentation by collating previously published Read code lists for smoking status and smoking cessation product codes and using the last observation carried forward approach to impute values closest to the index date where possible.21–25
Certain pre-existing conditions could serve as alternative explanations of dyspnoea presentation (COPD or asthma, and heart failure) and haemoptysis, thereby influencing the likelihood of receiving guideline-concordant imaging. Therefore, diagnoses of respiratory disease and heart failure recorded in primary care between 78 and 6 months prior to symptomatic presentation were used to categorise patients as having no morbidities; respiratory conditions only; heart failure only; respiratory conditions or heart failure; and both respiratory conditions and heart failure.23 26
Finally, we examined recent prior imaging as another possible explanation for guideline discordance, defined as receipt of primary care-ordered chest imaging up to 6 weeks prior to symptomatic presentation (specifically during a period from −42 days to −1 day from the index date).
Statistical analyses
The two patient-based symptom cohorts were examined independently of each other. Descriptive statistics followed by crude and adjusted logistic regression models were used to examine variation in guideline-concordant imaging by patient-level covariates (sex, age group, ethnicity, smoking status and pre-existing respiratory disease and heart failure). We considered two adjusted models: the first excluded cancer diagnosis status (as this is not known at the time of presentation when decision-making for using imaging takes place), while the second included cancer diagnosis as a covariate (as while this occurs after presentation, it may act as a marker of other unmeasured characteristics of individuals with lung cancer, eg, symptom severity and other features that may have been present at presentation and taken into account by the primary care physician).
Subsequently, among the subgroups of patients in each cohort who were diagnosed with lung cancer, descriptive statistics were used to compare stage at diagnosis (tumour, node, metastases (TNM) stages 1/2, 3/4 or missing), route to diagnosis (one of eight routes as established by NCRAS27) and time from symptomatic presentation to diagnosis (the diagnostic interval28) by imaging status, using χ2 tests for significance.
Supplementary analyses
We undertook the following supplementary analyses, which are reported in the online supplemental appendix:
Considering an imaging interval of 0–28 days from symptomatic presentation (instead of 14 days, as in the main analysis) (online supplemental appendix 3).
Considering imaging ordered from any source within 14 days of symptomatic presentation (instead of just primary care-ordered imaging, as in the main analysis) (online supplemental appendix 4).
Results
Study population
A total of 162 161 individuals with newly presenting dyspnoea and 8120 individuals with newly presenting haemoptysis were included (table 1). The majority of individuals in both symptom cohorts were 60 years or older, white and either ex-smokers or current smokers as opposed to non-smokers.
Composition of the dyspnoea cohort (n=162 161) and haemoptysis cohort (n=8120)
Guideline-concordant imaging
A total of 22 560/162 161 (13.9%) patients with dyspnoea and 4022/8120 (49.5%) patients with haemoptysis received guideline-concordant imaging, namely primary care-ordered chest imaging within 2 weeks of presentation (tables 2 and 3).
Receipt of guideline-concordant imaging within 2 weeks from presentation among patients with newly presenting dyspnoea (n=162 161)
Receipt of guideline-concordant imaging within 2 weeks from presentation among patients with haemoptysis (n=8120)
Among both cohorts, women, the youngest age groups (30–39 and 40–49 year-olds) and those with missing ethnicity were less likely to receive guideline-concordant imaging compared with men, 50–59 year-olds and white individuals, respectively. Individuals who had had chest imaging in the 6 weeks prior to symptomatic presentation were much less likely to receive guideline-concordant imaging (adjusted OR (aOR) 0.16, 95% CI 0.14–0.18 in dyspnoea cohort; aOR 0.09, 95% CI 0.06–0.11 in haemoptysis cohort). History of COPD/asthma was also associated with much lower odds of guideline concordance in dyspnoea presenters (aOR 0.23, 95% CI 0.23–0.24) with a much weaker though similar in direction association among haemoptysis presenters (0.88, 0.79–0.97).
In the dyspnoea cohort, individuals with lower socioeconomic status, current/ex-smokers and individuals with morbidities were less likely to receive imaging. Individuals with pre-existing COPD/asthma or heart failure were much less likely to receive imaging for dyspnoea, with the lowest odds of imaging seen among patients with both morbidity types (0.20, 0.17–0.24).
Adjustment for cancer made no material difference to the associations between sociodemographic variables and guideline-concordant imaging. Individuals who were diagnosed with lung cancer in the year post-presentation were more likely to have received urgent imaging for newly presenting dyspnoea or haemoptysis (2.07, 1.78–2.41 and 1.39, 1.06–1.83, respectively). A similar association was also observed among patients later diagnosed with other cancer types following a dyspnoea presentation (1.33, 1.20–1.48), but without such evidence for haemoptysis.
Lung cancer outcomes by imaging status
The proportion of patients subsequently diagnosed with lung cancer among those who received guideline-concordant imaging was twice as high compared with those not promptly imaged (1.1% vs 0.6%, p<0.001) for dyspnoea presenters, and a third higher for haemoptysis presenters (3.6% vs 2.7%, p=0.076) (table 4). However, the majority (854/1103, 70%) of dyspnoea presenters and (110/253, 43%) of haemoptysis presenters subsequently diagnosed with lung cancer did not receive guideline-concordant imaging.
Cancer diagnoses in the year following symptomatic presentation by imaging status
Compared with those who did not receive guideline-concordant imaging, a slightly higher proportion of those who received imaging were diagnosed with advanced stage (TNM stages III–IV) among both dyspnoea and haemoptysis cohorts though this may have been a chance finding (table 5).
Lung cancer-related outcomes by imaging status
There was substantial variation in diagnostic route. Patients who received guideline-concordant imaging following dyspnoea or haemoptysis presentation were more likely to have been diagnosed via the two-week wait pathway for suspected cancer (43% vs 27% for dyspnoea; 59% vs 35% for haemoptysis), and less likely to be diagnosed via an emergency compared with those who did not receive guideline-concordant imaging, particularly for haemoptysis presenters (30% vs 38% for dyspnoea; 9% vs 30% for haemoptysis; p<0.001 for overall variation by concordant imaging status by diagnostic routes in both cohorts, table 5).
Diagnostic interval among patients with lung cancer
Patients with dyspnoea who were subsequently diagnosed with lung cancer had a median (IQR) diagnostic interval of 83 (28–205) days, while among patients with haemoptysis this was 39 (21–71) days.
Patients who received imaging following dyspnoea presentation had a shorter diagnostic interval than those who did not (median: 34 vs 114 days, nearly a fourfold difference). In comparison, there was little difference in the distribution of time to cancer diagnosis by imaging status in the smaller haemoptysis cohort (median: 39 days for both groups) (figure 1).
Time from symptomatic presentation to lung cancer diagnosis by imaging status in 1103 patients with dyspnoea (A) and 253 patients with haemoptysis (B). The horizontal green line represents the median diagnostic interval in all patients with lung cancer in the dyspnoea cohort (83 days) and the haemoptysis cohort (39 days), respectively.
Supplementary analyses
Additional analyses considering a 4-week interval in which imaging took place (online supplemental appendix 3), or imaging ordered from secondary care and other sources in addition to imaging ordered by GPs (online supplemental appendix 4) identified a greater number of patients who received guideline-concordant imaging but patterns of variation by patient factors remained largely unchanged. One exception was in the haemoptysis cohort, where current smokers and patients subsequently diagnosed with lung cancer were more likely to have receive prompt imaging (whereas there was no such evidence in the main analysis).
Discussion
Summary of findings
Less than one in seven patients with newly presenting dyspnoea and one in two patients with newly presenting haemoptysis received primary care-ordered chest imaging within 2 weeks of presentation in line with national guidelines. Women, younger patients, individuals who had received chest imaging before presentation and those with pre-existing COPD or asthma were less likely to receive guideline-concordant imaging, while individuals diagnosed with lung cancer in the year following presentation were more likely to have been promptly imaged. Of those who subsequently received a lung cancer diagnosis, most patients presenting with dyspnoea (854/1103, 70%) and many patients presenting with haemoptysis (110/253, 43%) had not received timely imaging.
In both cohorts, those who did not receive prompt imaging before lung cancer diagnosis were more likely to be diagnosed as emergencies. Among patients with lung cancer initially presenting with dyspnoea, prompt imaging was associated with shorter intervals to diagnosis.
Comparison with prior literature
A study examining urgent primary care referrals for six ‘red-flag’ cancer symptoms found that only 40% of eligible patients received an urgent referral within 14 days of presentation, with substantial variation by symptom16; in our study, the corresponding figures for prompt chest imaging being 14% and 50% for dyspnoea and haemoptysis, respectively. The proportion of symptomatic patients who were subsequently diagnosed with cancer without having received an urgent or fast-track referral ranged from 2.8% to 9.5% by symptom; in comparison, we found the respective proportions for lung cancer to be 0.6% of dyspnoea presenters and 2.7% of haemoptysis presenters. Our results are also aligned to a US study that identified 38% of patients with lung cancer had had a missed opportunity,29 and a study using English electronic health records data that found 65% of patients with lung cancer who had a chest X-ray had received it 2 weeks or longer after presentation.30
Our findings indicate that patients with lung cancer who received guideline-concordant imaging had shorter diagnostic intervals but a higher proportion were diagnosed with advanced stage; this concords with previous research on patient populations with lung cancer that reported shorter diagnostic intervals among those with late-stage versus early-stage cancer.31 32 This may also reflect confounding by indication and the waiting time paradox which has been previously described.33
Strengths and limitations
We analysed nationally representative linked primary care data. The HES-DID and NCRAS data sets represent gold standard sources of information on ascertaining imaging investigations34 and cancer diagnoses,35 respectively.
We acknowledge several limitations. First, factors beyond clinical decision-making in primary care such as imaging capacity and patients’ ability to attend for the ordered investigations may have influenced whether or not guideline-concordant imaging occurred. Nevertheless, when we examined a longer period for the imaging to occur (online supplemental appendix 3) or included imaging ordered by other sources (online supplemental appendix 4), there remained substantial numbers of eligible individuals who did not receive prompt imaging. Furthermore, CT imaging may be subject to longer waiting times due to capacity constraints in comparison to X-rays. However, the vast majority (99%) of imaging events conducted in the chest region in both symptom cohorts (within 2 weeks from presentation or with no time constraint) were chest X-rays not CT imaging (data not shown).
The 2005 NICE guidelines indicate that patients with persistent dyspnoea should be ordered urgent chest imaging, defined as lasting 3 weeks or more.10 Some of the individuals in our dyspnoea cohort may have presented with dyspnoea less than 3 weeks after onset, leading to the underestimation of the true proportion of clinically eligible individuals receiving guideline-concordant imaging. Free text primary care records data could have captured this kind of detail, but are not available for research purposes due to resource constraints in ensuring non-disclosivity of the data.36
We assumed that the chest imaging events identified following presentation were related and not incidental to the dyspnoea or haemoptysis, which could have led to the overestimation of guideline concordance. Similarly, we examined lung cancers that were diagnosed in the 12 months following symptomatic presentation: some of these cancers could also have been unrelated to the coded symptom. However, most imaging was conducted within the first 1–2 days from presentation, and the majority of lung cancers diagnosed in the 12 months after presentation were identified in the first 6 months (71% and 93% for dyspnoea and haemoptysis, respectively), supporting the validity of our assumptions (data not shown).
Implications
Individuals who presented with dyspnoea or haemoptysis and were later diagnosed with lung cancer were more likely to have received guideline-concordant imaging than symptomatic individuals who were not diagnosed with lung cancer. This suggests appropriate clinical decision-making took place for these individuals, though we must acknowledge there are additional patient, doctor and system-level factors contributing to urgent imaging taking place following presentation.
Some of the observed variation in guideline-concordant imaging may have a plausible explanation. Individuals with pre-existing respiratory disease and/or heart failure were less likely to receive urgent imaging, possibly because the symptoms were attributed to those conditions.37 Women, younger patients and those who had had imaging prior to presentation were also less likely to receive guideline-concordant care, which may reflect appropriate assessment of the lower prior risk of lung cancer in these groups (compared with men, older patients and individuals who had not been ordered chest imaging recently, respectively).
Nevertheless, other patterns of variation are harder to explain: current or ex-smokers and patients residing in poorer neighbourhoods were less likely to receive prompt imaging for dyspnoea despite being at relatively higher risk of lung cancer compared with non-smokers or more affluent patients.38 These associations require further elucidation, including through qualitative studies.
Substantial proportions of the individuals later diagnosed with cancer did not receive guideline-concordant imaging, potentially representing missed opportunities for earlier lung cancer diagnosis. Clinical case note review could enhance our understanding of the reasons for guideline discordance and missed opportunities.39 40 Nevertheless, the findings demonstrate the potential for investigation or referral activity captured in electronic health record systems to be used as a diagnostic quality indicator. Additionally, examining provider-level variability in guideline-concordant care could be informative,41 noting that guideline concordance increased when we examined imaging ordered by all sources and not just primary care.
A critical consideration prior to improving guideline adherence is imaging access and capacity. Chest X-rays may miss 20% of symptomatic lung cancers42 while CT capacity has been insufficient in England since before the emergence of COVID-19.43 The recently launched community diagnostic centres (aiming to improve access to diagnostic tests outside of hospital settings) could form part of service redesigns aimed at improving lung cancer diagnosis pathways.44
Conclusion
In the context of cancer diagnosis, primary care-ordered urgent imaging patterns broadly accord with clinical risk. However, large proportions of dyspnoea or haemoptysis presenters do not receive prompt imaging, likely representing missed opportunities for more timely lung cancer diagnosis, especially in patients with haemoptysis. Developing quality metrics based on guideline concordance for prompt chest imaging could improve the quality and equity of urgent imaging in primary care.
Data availability statement
Restrictions apply to the availability of these data, which were used under licence for the current study, and so are not publicly available. For reusing these data, an application must be made directly to the Clinical Practice Research Datalink (CPRD; www.cprd.com).
Ethics statements
Patient consent for publication
Acknowledgments
This study is based in part on data from the Clinical Practice Research Datalink obtained under licence from the UK Medicines and Healthcare products Regulatory Agency. The data is provided by patients and collected by the NHS as part of their care and support. The interpretation and conclusions contained in this study are those of the author/s alone. The study was approved by the MHRA (UK) Independent Scientific Advisory Committee, protocol number 18_299RMnA5, under Section 251 (NHS Social Care Act 2006).
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @mmkoo12, @LukeMounce, @DrMeenaRafiq, @CallisterMat, @HardeepSinghMD, @garyabel, @glyratzopoulos
Contributors MMK, GL and GAA conceptualised and designed the study. MMK undertook data analysis with support and input from LTAM and GA. MR, HS and MEJC provided clinical commentary on methods and interpretation. MMK drafted the initial manuscript, with critical revision of the article from all authors. MMK is responsible for the overall content as guarantor.
Funding The research was supported by the Gordon and Betty Moore Foundation (GBMF 8838). Additonally, GL acknowledges funding from a Cancer Research UK Clinician Advanced Scientist Fellowship (C18081/A18180); HS reports receiving grants from the Houston Veterans Administration (VA) Health Services Research and Development (HSR&D) Center for Innovations in Quality, Effectiveness, and Safety (CIN13-413), the VA HSR&D Service (IIR17-127), the VA National Center for Patient Safety, the Agency for Healthcare Research and Quality (R18 HS029347 and R01HS27363), and the Gordon and Betty Moore Foundation and serving as co-Chair of The Leapfrog Group’s National Advisory Group for Recognizing Excellence in Diagnosis; to end of 2022, GL was Associate Director, HS and GAA were Senior Faculty members, and MMK and MR were Faculty members of the CanTest Collaborative, funded by Cancer Research UK (C8640/A23385).
Disclaimer The interpretation and conclusions contained in this study are those of the authors alone.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.