1 INTRODUCTION

Lung cancer is the leading cause of cancer death in the world.1 Patients who have never smoked represent approximately 14% of UK lung cancer cases,2 while worldwide this varies from 10% to 25%.3 To put this into perspective, when measured as a separate cancer, lung cancer in never-smokers is the seventh most prevalent cancer in the world4; and the eighth most prevalent cause of cancer-related death in the UK; higher than cervical cancer, ovarian cancer, leukaemia, and lymphoma.2, 5

Low net-survival of lung cancer is often attributed to late-stage detection; treatment can offer encouraging prognosis when lung cancer is detected at an earlier stage.6, 7 However, the majority of patients are still diagnosed when their lung cancer has advanced to stage III/IV6 where one-year net-survival is poorest.8, 9

There are a number of potential differences between lung cancer patients who have never smoked (hereafter referred to as ‘never-smokers’) and those who are currently or have previously smoked (‘ever-smokers’). First, there are biologically distinct pathways towards lung cancer caused by tobacco smoking compared with other exposures or genes. Tobacco smoke damages the DNA in lung epithelial cells, leading to tumour development and progression.10, 11 In contrast, never-smokers’ cancers are more likely to be caused by environmental substances (e.g. pollution), occupational substances (e.g. carcinogenic chemicals) or genetic predisposition.12 These differences in aetiology contribute to different forms of cancer. Ever-smokers have higher levels of squamous cell lung cancers that grow in the centre of the lungs (bronchi) compared to never smokers who are more likely to have adenocarcinomas that grow in the outer part of the lung.13, 14 This can mean that never-smokers are less likely to experience noticeable symptoms at an early stage of disease, which is likely to contribute to delays in diagnosis.

Second, never-smokers may assume that they are not at-risk of lung cancer and have an amplified tendency to attribute symptoms to other acute conditions.15 For example, they are less likely to recognise breathlessness as a potential symptom of lung cancer compared to those with a smoking history.16 This may be due to international public health efforts to reduce the burden of lung cancer, primarily targeted via anti-smoking educational campaigns.17, 18 This has resulted in widespread public awareness of the link between lung cancer and tobacco exposure, as well as stereotypical views of who is likely to get lung cancer, but potentially obscured the fact that never smokers can get lung cancer too.

Third, healthcare professionals in the diagnostic pathway may display a detection bias against pursuing a diagnosis of lung cancer in never-smokers until other diagnoses have been excluded.19 Although never-smokers make up a significant proportion of lung cancer cases, not much is known about this population.

Responding to an urgent need for information about the experience of lung cancer patients who have never smoked, we designed the PEARL study (Patient Experience of symptoms, help-seeking And Risk factors in Lung cancer in never smokers). The first part of the study is this rapid review, synthesizing evidence relating to experiences of the pathway to diagnosis for patients with- and without a smoking history. This will inform a qualitative study that will aim to generate targeted recommendations to reduce delays in diagnosis of lung cancer in patients who have never smoked.

The Model of Pathways to Treatment (MPT) is a framework that can be used to understand intervals and structure research findings across the cancer patient pathway: including Symptom Appraisal, Help-seeking, Diagnosis and Pre-treatment (see Figure 2).20 This framework promotes greater consistency (e.g. in terms of defining intervals) across early diagnosis research, allowing comparisons to be made with existing literature, as well as ensuring consideration of range of patient (e.g. comorbidities), healthcare system (e.g. access) and disease factors (e.g. tumour type).

The MPT20 was used in this study to categorise the experiences of patients into chronological order using the intervals, with a particular focus on findings that may explain or contribute towards delays in the pathway to diagnosis for never-smokers.

1.1 Research aims

To our knowledge, there has been no review of literature investigating the pre-diagnostic experiences of lung cancer patients who have never smoked. The current review aims to provide a narrative synthesis comparing the experiences of never-smokers and ever-smokers. Additionally, the review will examine how any experiences unique to never-smokers may impact on, or introduce, delays in the pathway to diagnosis of lung cancer.

The research questions guiding this review were:

What are the symptom appraisal and help-seeking experiences of patients diagnosed with lung cancer, who have never smoked?

How does this experience compare with the experience of patients with lung cancer who have smoked?

How do the social and life histories of people prior to the development of lung cancer, particularly for never smokers, impact on their presentation and diagnosis?

2 METHODS 2.1 Rapid review

A rapid review methodology was chosen, as this is a systematic approach to synthesise current literature, which can provide a timely and conclusive answer in relation to the direction and strength of current evidence.21 This will provide a foundation for the urgent subsequent planned work that will investigate experiences of the pathway to lung cancer diagnosis of patients who have never smoked.

The review followed the Preferred Reporting Items for Systematic Reviews guidelines (PRISMA)22 and practical recommendations by Tricco et al. 201723 on how to a conduct rapid a review. To ensure transparency, the review protocol was registered with PROSPERO (CRD42020191563). The review was made rapid by streamlining the search process through: 2.2 Search strategy Three electronic databases (MEDLINE, PsychINFO and Google Scholar) were searched on 28th June 2020, including results from 1st January 2010 onwards. The following electronic search strategy was employed using Boolean operators: –

((“lung cancer” or “lung neoplasms” or “lung carcinoma”)

–

AND (“smok* status” or “never smok*” or “cigarette smoking” or “tobacco smoking”)

–

AND (“symptom appraisal” or “symptom assessment” or “help-seeking” or “delayed diagnosis” or “missed diagnosis” or “early diagnosis” or “diagnos* pathway” or experience or symptom or presentation))

–

Search results were limited by language (English) and date (2010–2020).

2.3 Eligibility criteria

Both qualitative and quantitative studies that met each of the inclusion criteria displayed in Table 1 were included in this review.

TABLE 1. Eligibility criteria Inclusion Exclusion Qualitative/quantitative data Systematic/Scoping/Rapid review, editorials, books, dissertations and commentaries Published & peer-reviewed Full-text unavailable Sample of patients diagnosed with lung cancer Community sample Assessed lung cancer specific experiences of the pathways to treatment (appraisal, help-seeking, diagnosis, pre-treatment). Explicit focus on experiences post-treatment Sample includes both ever and never-smokers. Sample of ever-smokers only Published in English Published in languages other than English Any country/healthcare system – Published in 2010–2020 Published before 2010

The review includes studies from any country/healthcare system, and studies that looked at actual as well as hypothetical symptom appraisal and help seeking. In terms of smoking history, we only selected studies that had samples that included both never-smokers and those with a smoking history, with a particular focus on differentiating the experiences of lung cancer patients by smoking status, or directly comparing experiences of never-smokers and ever-smokers.

Studies were screened by abstract and full text for eligibility (see Table 1) by two reviewers (AS and SvO) independently; disagreements regarding final study selections were resolved by discussion.

2.4 Quality appraisal

Two reviewers (AS and SvO) then assessed the methodological quality of the included studies independently using the Mixed Methods Appraisal Tool (MMAT),24 which can concomitantly assess qualitative, quantitative and mixed-methods studies. Criteria assessed studies' methodological quality, analysis and interpretation of results using a simple ‘Yes’, ‘No’ or ‘Can't tell’ rating system. Differences were resolved by consensus with separate criteria for qualitative and quantitative studies. All evidence from studies that were included in the final analysis was treated equally.

2.5 Data charting and analysis

The MPT20 was used to categorise the experiences of patients into chronological order using the intervals. Data were extracted from articles that identified experiences unique to never-smokers or provided a direct comparison of experiences between smoking statuses, across any of the MPT intervals. This included: patient appraisal of lung cancer symptoms, interactions with healthcare professionals at any stage of the pathway, experiences of primary and secondary care, and experiences of stigmatisation. Data extraction prioritised any findings that may explain or contribute towards delays in the pathway to diagnosis for never-smokers.

Due to the wide range of research designs included in the final analyses, and the rapid design of this review, a meta-analysis was not appropriate for data synthesis. Instead we carried out a narrative synthesis, a common alternative for the reporting of findings used in systematic reviews.25 To limit the influence of reviewer bias, guidance outlined by the UK Economic and Social Research Council26 was used throughout to direct data synthesis.

3 RESULTS

Figure 1 outlines the process followed to identify relevant articles for the review. Searches completed in MEDLINE, PsychINFO and Google Scholar identified 262 records. Articles were exported into EndNote and after removing duplications 246 articles remained. Initial screening of titles and abstracts excluded 213 irrelevant studies (e.g. smoking cessation research), leaving 33 potentially relevant articles. Back-chaining of retrieved articles identified another 12 articles. After a full-text review of these 45 articles, 35 articles were excluded leaving seven quantitative studies and three qualitative studies. Seven27-33 were conducted in the USA, one in Australia,34 one in North India,35 and one across three European countries (Denmark, Sweden, UK).36 Full study details, including participant characteristics, data collection and findings, can be found in Tables 2 and 3.

Preferred reporting items for systematic reviews guidelines flow diagram of study selection

TABLE 2. Summary of quantitative studies Study Recruitment Sample/smoking status (%) Primary outcomes Results Conclusion Study limitations Age (SD) Country Carter-harris (2015)28 Patients diagnosed with non-small cell lung cancer (primary) were recruited from an academic thoracic-oncology clinic and a community hospital-based outpatient radiation centre. N = 93 Lung cancer stigma scale No significant difference (p < 0.05) in total perceived stigma scores between smoking groups. Perceived lung cancer stigma can act as a critical obstacle to medical help-seeking. However, there was no difference in stigma scores between those with a smoking history and those without. Recall bias may have influenced patient data in relation to symptom awareness and medical help-seeking, although key-event mapping was used to reduce the risk of recall bias. Current: 30 (32.3%) Timing of medical help-seeking Patients with higher levels of perceived stigma sought medical help later (r = 0.27, p = 0.01) Small sample size (n = 93) resulting in limited statistical power and ability to generalise results. Former: 33 (35.5%) Never: 30 (32.2%) Age: 62 (8.7) USA Carter-harris (2014)27 Relates to same data set as Carter-Harris (2015) reported above. N = 93 CLCSS No significant difference in mean time to medical help–seeking between those with and without a smoking history (p = 0.222). Results revealed that lung cancer stigma does not correlate with medical help-seeking. Small sample size (n = 93) resulting in limited statistical power and ability to generalise results. Current: 30 (32.3%) Help-seeking: Previous Lung cancer stigma only significant predictor of variance in time from symptom onset to medical help-seeking (p < 0.01) An unexpected finding was that smoking did not correlate with time from symptom onset to medical help-seeking. Former: 33 (35.5%) Health care system distrust Never: 30 (32.2%) Social desirability Age: 62 (8.7) USA Choi (2019)31 Patients being investigated for actual or suspected lung cancer were opportunistically recruited from a midwestern university comprehensive cancer center. N = 52 State & trait anxiety Ever-smokers reported greater anxiety/worry, lower cognitive functioning and more negative perceptions of their health-environment. No significant differences between smoking groups in anxiety/worry, cognitive function and perceptions of the health environment at the time of diagnosis. Only cancer-related worry was significantly higher amongst smokers. Small sample size, particularly never-smokers (n = 3), with low statistical power, low re-producibility of results and a reduced chance to detect a true effect. Current = 12 (23.1%) Worry Former = 37 (71.2%) Cancer-related worry Never = 3 (5.77%) Attentional function Age: 64 (11.6) Health environment perceptions USA Weiss (2017)32 Participants diagnosed with any stage lung cancer were recruited through J. Edwin Brown and US data corporation targeted lists provided by Thomson Reuters. N = 174 Themes surrounding lung cancer: (1) Support Never-smokers reported significantly lower self-blame compared to former/current-smokers (mean 0.86 vs. 2.2, p < 0.001). Most participants did not feel personally stigmatised, although self-blame and personalized were significantly greater amongst smokers. Those without a smoking history reported the lowest satisfaction with care. The questionnaire was not designed for academic research at inception and lacked psychometric testing. Current: 33 (19%) (2) Amount of research Never smokers reported lower satisfaction with care compared to ever smokers (2.6 vs. 3.2; p = 0.01) controlling for stigma and blame did not modify this Former: 119 (68.4%) (3) The cause Never: 22 (12.6%) (4) Treatment by society & healthcare providers Modal age: 60–69 (5) Attitudes & experiences USA Criswell (2016)30 Participants diagnosed with lung cancer within the last 6 months were recruited from Loma Linda University medical center (LLUMC) and city of hope medical center (COH) through questionnaire packs received in mail. N = 213 Psychological adjustment, Never-smokers reported lower personal responsibility and lower regret than former/current-smokers; medical stigma was not different between smoking groups. Never-smokers differed to those with a smoking history in several of their associations between CRRS factors and adjustment variables. Specifically, never-smokers had stronger associations between all CRRS factors and depressive symptoms. Sub-sample sizes for current and never-smokers were small compared to former-smokers. Current: 38 (17.8%) Depressive symptoms, distress, For never-smokers, PR, regret and medical stigma all had moderate positive associations with depressive symptoms, healthcare dissatisfaction, avoidance coping strategies and psychological needs. CRRS did not include a measure of perceived (non-medical) social stigma. Age: 64.4 (9.8) Physical health-related adjustment, Former: 141 (66.2%) Symptom bother, Age = 68.2 (10.2 Satisfaction with healthcare, Never: 34 (16%) Supportive care needs coping strategies (CRRS): Age = 63.9 (12.2) - Personal responsibility USA - Regret - Medical stigma Singh (2012)35 Newly diagnosed lung cancer patients were recruited from a tertiary level referral health care facility in North India. N = 974 Smoking index Stage-IV disease at presentation was higher in never-smokers compared to heavy smokers (67.4% vs. 39.1%). Clinician ‘detection bias’ was suggested to be a key reason for later diagnosis in patients who had never smoked. Bidis (hand-rolled tobacco) were considered equivalent to one cigarette. Current/Former: 503 (76.9%) Staging of lung cancer (TNM classification) The heavy smoking group was overwhelmingly male (97.9% vs. 57.7% in the never-smoker group). It is likely that late diagnosis of never-smokers was (partially) due to the gender inequality in access to healthcare in India. Never:151 (23.1%) Age = 58.1 (10.8) North India Williamson (2020)29 Patients diagnosed with lung cancer in the last 12 months were recruited in thoracic oncology clinics within two National cancer Institutes (NCI) in the southern and Northeastern regions of the USA. N = 266 Lung cancer stigma Clinically meaningful lung cancer stigma was experienced by current (93.9%), former (85.4%), and never-smokers (60.0%). Constrained disclosure was similar across all groups. Current/former-smokers reported higher total, internalised and clinically meaningful lung cancer stigma than never-smokers. Lung cancer stigma is experienced by patients in all smoking status groups, although it is more commonly reported by those with a smoking history. However, patients reported similar levels of discomfort in disclosing their diagnosis, irrespective of smoking status. Data were cross-sectional so cannot infer causality. Current: 49 (18.4%) Depressive symptoms Former: 154 (57.9%) Never: 60 (22.6%) Age: 63.3 (10.8) USA TABLE 3. Summary of qualitative studies Study Recruitment Sample/Smoking split (%) Methods/Aim of interview Results Conclusion Study limitations Hajdarevic (2018)36 Patients diagnosed with lung cancer in the last 6 months were recruited through purposive sampling, snowball sampling and social media. N = 72 Semi-structured interviews aimed to assess how (and if) smoking was discussed by patients. The topic of smoking was never introduced by the interviewer. Feeling responsible for cancer was most common in England. And amongst English participants, feeling responsible for the cancer was more prominent in smokers. English participants almost always discussed their lung cancer with reference to smoking. Smoking plays a dominant role in the lung cancer narrative in England, more so than in the other countries. Does not identify smoking status split within sample, limiting the ability to draw conclusions in relation to differences between never-smokers and those with a smoking history. Denmark: 22 90% of English participants discussed smoking during their interview (Denmark 73%, Sweden 60%). England: 20 Never-smokers used not-smoking as an explanation for not anticipating their diagnosis. Sweden: 30 No detail on smoking status of the full sample, detail of smoking status of participants who are quoted in the paper is included in the results. Age range (England): 51–70 Scott (2015)34 Patients with a lung cancer diagnosis and GPs were recruited from lung cancer support networks across NSW, Australia. Patients N = 20 Open-ended interviews examined symptom recall and medical help-seeking behaviour. Never-smokers often feel the need to clarify their non-smoking status (including to clinicians). This study has highlighted the stigma that is felt by lung cancer patients, regardless of smoking status. The study also highlights that GP's often rule out lung cancer in patients who do not smoke. Interviews were only completed with one group, meaning other themes may have been missed. Former: 13 (65%) Interviews with GPs assessed perceptions of lung cancer symptoms. Patients felt anti-smoking messaging encourages stigma and could increase hesitation to seek help. Never: 7 (35%) Delays in diagnosis may occur when GP's (and patients) do not consider non-smokers at risk of lung cancer. GP's N = 10 Age unknown Australia Hamann (2013)33 Patients with a confirmed lung cancer diagnosis were identified by treating oncologists from the UT Southwestern's Harold C. Simmons cancer center Interviews An interview framework guide was developed to: (i) Describe experiences of stigma; (ii) explore smoking history's role in stigma (iii) investigate stigma's impact on treatment-related behaviour. 95% of interviewees discussed perceived stigma. Perceived stigma is experienced by patients with and without a smoking history. However, internalised stigma was more prevalent amongst those with a smoking history. Researchers believed sample lacked hispanic participants due to language barriers. N = 42 Focus groups intended to provide feedback on the conceptual model from interviews. ∼50% of participants reported negative responses from HCP's. Never-smokers perceived judgmental smoking-related assumptions from HCP's. Never-smokers reduce the negative impact of stigma by disclosing their never-smoker status. Current: 10 (24%) Some never-smokers described how they ‘pre-empt’ stigmatising reactions by disclosing their non-smoking status. Recent: 11 (26%) Internalised stigma was discussed more by current-smokers than never-smokers. Former: 10 (24%) Never: 5 (12%) Focus groups N = 23 Age unknown USA 3.1 Study quality

All 10 studies passed the MMAT screening questions, and were taken through the full MMAT quality assessment (see Supporting Information for full MMAT results).

3.1.1 Quantitative studies (n = 7)

Six quantitative studies27-32 had samples with a mean age above 60, predominantly white ethnicity (78-95%) and a gender split ranging from 35%–64% female. Singh et al. (2012)35 recruited a sample that was younger (mean age: 58) and mostly male (83%), and Choi et al. (2019)31 included only a small proportion of never-smokers (5.7% of total sample). We were unable to examine non-response bias for three studies29, 31, 32 as they did not report characteristics of non-responders.

3.1.2 Qualitative studies (n = 3)

Only Hajdarevic, et al. (2018)36 reported participants' age, however this study did not report the proportion of never-smokers in their sample. All studies sufficiently substantiated their findings using data and there was coherence between sources, data collection and analysis. However, Scot et al. (2015)34 did not provide details of how themes were developed. Interviewer-bias and sample size justification were only addressed by Hamann, et al. (2014).33

3.2 Narrative synthesis

Once extracted, findings were categorised into the MPT intervals (Figure 2) and themes were constructed. These themes are presented in this results section organised by MPT interval.

Summary of review findings in relation to model of pathways to treatment

3.2.1 MPT appraisal interval

Theme 1: Never-smokers are less likely to perceive themselves at risk of lung cancer.

Of the seven quantitative articles, none reported on symptom awareness of never-smoker patients. However, a lack of lung cancer symptom awareness amongst never-smokers was a frequent theme in the qualitative studies reviewed.