The mediation of social influences on smoking cessation and awareness of the early signs of lung cancer

The possible influence of delays in diagnosis on survival and the risk factors for delay in patients with cancer have been the subject of considerable interest and controversy for many years [1]. Survival from cancer in the United Kingdom is poorer than that of other European countries, and it has been suggested that this can be attributed to a more advanced disease stage at presentation [2]. Cancers in the UK are diagnosed later and at more advanced stages than in other European countries with resulting low national survival rates [3]. The situation is particularly serious in the case of lung cancer, where for up to 80% of patients, their disease is found to be inoperable because it has been diagnosed too late [4]. In the UK, it appears that the picture is particularly bleak, with delays in the diagnosis of cancer being attributed to both patient and healthcare system factors [5]. Corner et al.[6] reported that delays of up to a year following the onset of worrying symptoms were not uncommon before patients decided to seek medical help. This is in contrast to much of Europe where delays can be far shorter, and survival rates correspondingly higher [7, 8]. In the USA too, it has been found that delays in seeking treatment can be shorter than in the UK, with Jensen et al.[9] for example, reporting that such delays can be less than six months. It is further evident that long-term smokers, particularly those with other conditions such as Chronic Obstructive Pulmonary Disease (COPD), and those living alone are at a higher risk of taking longer to consult with symptoms of lung cancer [10]. The reasons for this are a complex mix of individual and psychosocial factors [11, 12], but it has been established that one recurring feature in the case of lung cancer is that patients, regardless of how well progressed their disease was, failed to recognise the seriousness of their symptoms, and reported that they had noticed changes in their health status long before they sought medical help [13].

As yet there is no clear consensus on the potential for earlier diagnosis of lung cancer [14, 15], although research has suggested that the time between symptom onset and consultation can often be long enough to plausibly affect prognosis [10]. Moody et al. 2004 [16], for example, highlight that the most effective curative option for lung cancer is surgery, and although this modality has not changed significantly for many years [17], it can be highly effective for patients with early stage disease. It is far less so for those whose disease is more advanced. So it follows that a significant impact on survival rates could be achieved if a higher proportion of patients were diagnosed early enough to benefit from potentially curative surgery.

Current government initiatives in the UK and other countries, aimed at raising awareness of the early symptoms of lung cancer illustrate the importance being placed on this issue at a policy level. The 2005 guidance from the National Institute for Health and Clinical Excellence (NICE) on the diagnosis and treatment of lung cancer, for example, were recently updated to advise healthcare professionals of important advances in management and patient information strategies [18]. And research into public awareness of cancers in the UK has also lately been stimulated by the strategic input of the National Awareness and Early Diagnosis Initiative (NAEDI) [19]. This initiative, launched in 2008 aimed to involve public and third sector organisations to support activities that promote the early diagnosis of cancer and has been instrumental in establishing a number of local and national cancer awareness projects [20].

In this paper, we present an analysis of interview data collected as part of the Lung Symptom Interpretation and Diagnosis Study (LUSID).¹ The study aimed to investigate the social factors which influence symptom recognition and help-seeking behaviour, and focused on two distinct groups:1) patients with an existing diagnosis of lung cancer; 2) those at heightened risk of developing the disease. These groups were chosen in order to understand how perceptions and action in relation to help-seeking might change and develop across the illness trajectory.

This was a qualitative study which utilised the principles of grounded theory [21] and aimed to systematically interleave data collection, analysis, constant comparison and theorizing in an attempt to produce a plausible representation of the world view of participants that was rooted in their accounts, rather than being hypothesis driven. We also drew on existing material that emerged as relevant to our understanding as the analytical process developed.

Data were collected in 2012 and 2013, and we were able to included 25 participants from two groups; people with an existing diagnosis of lung cancer (n = 11); and people at a heightened risk of developing lung cancer (n = 14), i.e. long term smokers (that is, people who have smoked for over a year), and people with Chronic Obstructive Pulmonary Disease (COPD). The study was situated in two clinical settings in the North West of England. One was a dedicated specialist lung clinic in a large regional hospital, and the other was a local GP practice. All of the diagnosed participants were recruited from the lung clinic along with 3 ‘at heightened risk’. The remaining ‘at heightened risk’ patients came directly via their GP, or from responding to recruitment posters at the GPs clinic. Both the clinic and the GP practice were situated in an area of high economic deprivation on the outskirts of a major (UK) Northern industrial city, and a large proportion of patients in both settings lived locally. The mean age of the diagnosed group was 69 years, and 64 years for those ‘at heightened risk’. In the lung clinic setting, 3 patients who were approached by their consultant declined to take part; 4 ‘at heightened risk’ patients from the GP surgery declined. In all cases initial contact was made via a health care professional (either the person’s GP or their consultant) who obtained provisional consent, and passed their contact details on to the research team. A researcher then arranged to conduct a 30-60 minute semi-structured interview where written consent to participate was obtained. Participants were able to have a relative or carer present during their interview, and 4 participants took up this option. 6 participants who gave verbal consent to be contacted, subsequently chose not to go ahead with an interview and were not included in the study (4 from the ‘at risk’ group, and 2 who had been diagnosed). Resources permitted a fieldwork period of six months, after which time we closed recruitment. The study adheres to the RATS guidelines for reporting qualitative studies.

Interview guides for the in-depth interviews were generated from themes emerging from an initial literature review [22], formal and informal contacts with key informants, and input from PPI (Public and Patient Involvement) representatives. This comprised of formal consultations with an established PPI network at the University of Manchester (the PRIMER group).² PRIMER were able to provide input from an early stage of the project. Two representatives from this PPI group also joined the project steering committee and contributed to aspects of the analysis (see Acknowledgements). Ethical approval for the study was obtained from the regional NHS ethics committee (REC reference 12/NW/0592).

A striking feature of our data was the relatively limited interest that participants appeared to have in the signs and symptoms of lung cancer. This is perhaps understandable for those at heightened risk because even though it was technically accurate to describe them as such, this categorisation does not appear to have been a major feature of how they actually saw themselves on an everyday level. There may have been a degree of avoidance occurring too, so that people appeared reluctant to engage with the issue not because they weren’t interested but because to do so might be felt to be too pessimistic, or tempting fate in some way. That this was also evident in our diagnosed participants was perhaps surprising, but if thought of in terms of avoidance or distancing being used as a conscious or un-conscious coping strategy [24], it becomes more explicable. We did not undertake a specific measure of psychological distress, and so are not able to provide definitive evidence that this was in fact what our diagnosed participants were engaged in. However, a higher level of distancing (which reportedly allows the patient a greater capacity to gradually assimilate their situation, rather than having to deal with it head on) has been associated with lower levels of psychological stress in many forms of cancer [25, 26].

At a practical level, there have been some efforts to investigate the factors which influence how people judge whether or not changes in their health status are serious – and whether or not they decide to visit their doctor. Smith et al.[10] for example, undertook a qualitative synthesis of 32 research papers dealing with the help-seeking experiences of patients with 20 different types of cancer (>775 patients and carers). They found that there were strong similarities in patients across all cancer types, and that key concepts were the recognition and interpretation of symptoms, and fear of the consultation. Focusing specifically on lung cancer, Corner et al.[6], interviewed 22 patients who had recently been diagnosed with operable (early stage), and inoperable (late stage) lung cancer. They concluded that regardless of their disease stage, or their social background, these individuals failed to recognise symptoms that they had experienced over many months prior to their eventual diagnosis as particularly serious or warranting medical attention. The findings of our study resonate strongly with these two studies, but may further indicate that a tangible and deliberate lack of engagement with potentially serious symptoms (along with a clear underlying knowledge that they may occur) is also at work. The ‘downgrading’ of symptoms and the tendency of people to attribute them to less serious illnesses when they did occur was also evident in our data. This finding is broadly in line with the literature on health related help-seeking behaviour in general [27], but also registers strongly with research more specifically focused on cancer patients. A study by Mor et al.[28], for example, provided a content analysis of remarks made by a corpus of 625 lung, breast and colorectal patients, and concluded that patients suffering from lung or colorectal cancer were significantly more likely to blame their symptoms on other less serious illnesses, and thus delay seeking medical help. Mor et al.[28] provide no specific data on the nature of the health changes that patients failed to act upon, or chose to downgrade. However later work, including that of Tod et al.[4] and Braybrook, [29] is more definite, and outlines symptoms such as persistent or irritating coughs, changes in the weather, and ‘getting older’ as those that patients routinely describe retrospectively after diagnosis. Again, this is resonant with the accounts we collected. Essentially, as Chapple et al.[30] suggest, there may be a strong tendency for people to normalise symptoms.

An interesting detail to emerge was the tendency for some participants (particularly those with lower socio-economic status (SES) to bracket together any number of different medical conditions. Breathlessness, joint problems, heart problems and different types of cancer were all viewed as a kind of single homogenous condition. This may reflect the predominance of low health expectations in our sample, and an underlying acceptance of the inevitability of physical decline, but it may also be connected to a degree of ‘cancer fatalism’. This concept has been oporationalized in different ways in the literature, and has attracted a degree of philosophical as well as psychological attention (See, for example, [31, 32]). A number of studies have established a link between people who have a lower SES and higher levels of fatalism ([33‐35]). Further, one study of 2018 British adults found that people with low SES were less positive about early cancer detection, and more fearful about seeking help for a suspicious symptom [34]. While a population based survey, also conducted in the UK, reported a link between low SES, lower awareness of cancer warning signs and greater anticipated delay in seeking help [36]. In the context of this study we used the term in a very broad sense to incorporate not only the belief that an inevitable consequence of developing cancer will be death [23], but that various lesser degrees of resignation toward the trajectory of the illness can be observed.

To date there has been a limited body of work focussing directly on social networks and chronic illness, although some research has dealt with aspects of social support [37]. Distinctions have been drawn between the strong social ties which develop through close personal relationships such as those within families, versus weaker ties with civil organizations and other more formal groups. It has been suggested that groups with lower SES tend to depend more on strong ties whilst the middle classes are more adept at making use of weak ties [38, 39]. This did not appear to be the case, however. Although the importance given to wider networks per se by participants was broadly in line with studies which have identified the processes of social network engagement, it was really only very close family contacts that were reported to be influential in terms of cancer or other serious illness related issues. This echoes recent research highlighting that the greatest contribution to illness management within personal social networks of those with chronic illnesses, comes from partners and close family members [38]. Age may have been a factor here as well: A recent multi-cancer meta-analysis by Pinquart & Duberstein [40], for example, examined 87 studies on the associations of perceived social support, network size, and marital status, and although levels of engagement varied by cancer type, social network correlations were found to be generally stronger in younger patients. Our sample was predominantly comprised of older individuals, and our findings strongly support several other studies which have highlighted the role that very close family and spouses (as opposed to slightly wider community networks) play in prompting people to seek medical help when they notice potentially serious symptoms. Along with straightforward legitimisation of symptom severity – a theme which crops up in work on a variety of cancers (see, for example, [4, 29]) – such close networks can also perform the function of actually sanctioning a visit to the doctor. This ‘legitimisation of the appointment’ was found to be particularly important to male patients who were concerned that they would not be seen as wasting the doctor’s time with trivial symptoms, but the effect has been noted in some female patients too [10]. The issue of informal network legitimisation is particularly relevant in the context of our study because it suggests that people who have limited engagement with very close social networks may be at a disadvantage.

Early work by Goodwin et al.[41] noted significant connections between marital status and diagnostic delay in lung cancer, with those individuals who lived alone showing far more of a tendency to delay help-seeking than those in stable or long term relationships. More recently, Neal and Allgar [42] utilised the secondary analysis of patient-reported data from the ‘National Survey of NHS patients: Cancer’ to explore the relationship between socio-demographic factors and the components of diagnostic delay. And in line with the earlier findings of Goodwin et al, Neal & Allgar [42] reported that across six cancers (lung, breast, colorectal, ovarian, prostate, and non-Hodgkin’s lymphoma), it was single and separated/divorced people who had delayed longer than married people in seeking medical help.

Unusually, considering that recent data collated by Cancer Research UK indicates that smoking is directly responsible for 90% of lung cancers [43], a number of participants in our diagnosed group were either not currently smokers, or had never smoked. This gave us the opportunity to explore the issue of smoking related stigma – something which has been widely reported as presenting difficulties for patients who have never smoked [44]. Chapple et al. [30], for example, reported that non-smokers felt particularly stigmatised because of the association with smoking that lung cancer can have. In more recent work too, Sun et al.[45] describe how non-smokers with cancer often carry the stigma that their cancer is self-induced, and can feel ostracized by the public and abandoned by the oncology community. In our data, however, negative issues such as these were not strongly evident, even among those who had non-smoking related lung cancer. Work by De Nooijer et al. [46] has suggested that shame and embarrassment about symptoms actively hinders early presentation and diagnosis, but again, this was not the case in the people we interviewed. A possible reason for this may be linked to the social expectations, demographic makeup and age range in our corpus; none of our participants, for example, were under 57 years of age, and Menec & Perry [47] have suggested that stigma related to lung cancer is more likely to be associated with younger age groups.

The relationship between smoking and lung cancer risk is almost universally acknowledged. In a study based on data supplied by Cancer Research UK, Redeker et al.[48] reported a 90% awareness of this association. However, other recent work by Simon et al.[49] on the development of a robust cancer awareness measure clearly demonstrated that understanding of the symptoms and risk factors for lung cancer is very poor in the UK, with 38% of their large study sample unable to name a single symptom. There is a growing body of research focusing particularly on the psychological aspects of smoking cessation campaigns. Recent work by Borelli [50], for example, focused on the use of motivational interviews by healthcare practitioners at the individual and group level, but acknowledge that decreasing the prevalence of smoking will take multi-target, multi-channel and multi-method approaches. Armitage & Arden [51] were also concerned with the effects of motivational and volitional processes. In a study involving 350 smokers, they randomising participants to one of three groups who received either i) a smoking awareness questionnaire; ii) the questionnaire plus instruction to plan to quit; or iii) a questionnaire, plan to quit and help with forming an implementation plan. In line with the multi-method argument, Armitage & Arden proposed that harnessing both motivational and volitional processes would enhance the effectiveness of smoking cessation programmes, and they in fact found that considerably more people in the third group stopped smoking than in the first. Significantly too, individuals who were already motivated to stop smoking at the point of entering the study had most success in giving up. In a similar vein, recent qualitative work by Park et al.[52] concluded that although smokers often perceived smoking as a high-risk activity for lung cancer and smoking related diseases, this heightened concern in its self rarely motivated individuals to seek screening.

These findings are in line with reports from our participants over what would most influence them to stop smoking – the consensus from smokers in both groups being that individuals need to be self-motivated to stop smoking (for whatever reason), and that without this basic level of intention, any health promotion interventions are likely to be unsuccessful. The almost universal rejection of anti-smoking campaigns which are becoming ever more graphic and hard hitting, is an issue that also needs to be addressed if ‘hardened’ smokers are not to become alienated and unreachable.

The ambivalent attitude towards the dangers of smoking that we found in both our diagnosed and ‘at heightened risk’ groups, and a tendency for people to take a relatively perfunctory attitude towards the potential symptoms of lung cancer (or, if they were diagnosed, the details of their illness) may have been influenced by the particular social context of our corpus. Molassiotis et al. 2010 [5] outlined how certain socio-demographic, psychosocial and clinical characteristics can have an influence on the point at which a person first becomes aware of significant symptoms. In a study involving 75 patients diagnosed with a range of different cancer types, they found that older age, negative beliefs about cancer, fears about the consequences of having cancer, and reluctance to engage with the process of receiving bad news all came into play. Many of our participants were drawn from a relatively poor working class area, and in contrast to more affluent areas, smoking still carries far less of a general stigma than among some other groups [53]. These findings tend to support other work which has focused on the influence of social background and attitudes towards smoking and lung cancer. In particular, comparable trends have been reported by [6, 10, 49], while recent work by Beeken et al.[34] found that – partly because of a greater tendency towards fatalism over cancer that we outlined earlier in this article –people with lower socio-economic status saw it as less worthwhile to detect it and seek help early.

Against the backdrop of what is a significant issue in terms of health promotion, there is now a strong policy emphasis in the UK not only on helping people to give up smoking, but also on raising awareness of the early signs and symptoms of lung cancer. In this article we have outlined three main themes that emerged from the Lung Symptom Interpretation and Diagnosis Study.¹ These were: i) Fatalism and resignation in pathways to help-seeking and the process of diagnosis. ii) Awareness of smoking risk and response to cessation information and advice. iii) The role of social and other networks on help-seeking. We have tried to show that while our analysis confirms the findings of several other studies (notably those concerning the key role of close social networks, and the tendency of people to play down or attribute symptoms to other, less serious conditions), there are areas where our results indicate some divergence from extant work. In particular, we have described the phenomena whereby some respondents homogenised groups of unrelated illnesses, and the way in which this reflects a particular type of fatalistic perspective.

In terms of how our findings might inform future developments in health promotion and information provision in this area, we would suggest that a fruitful avenue to pursue is one that is able to utilise the strong influence of the close social networks we have described. Many of the high impact and high shock value campaigns that continue to be commissioned appear to have limited impact on certain key groups – particularly people who are hardened smokers. Campaigns that attempt to trigger help-seeking or smoking cessation behaviour by proxy in people close to those at heightened risk, rather than targeting the individual themselves may be more effective. Similarly, taking a ‘softer’ approach which does not immediately alienate users may be a better way to engage some groups. As our study has indicated, the present cycle of ever more graphic campaigns often cause those who may essentially be utilising mechanisms of denial or distancing (in the broadest sense) to ‘switch off’. Similarly, we would suggest that – particularly among lower SES communities where smoking is still extremely prevalent – it may be counter-productive to rely on the effect of smoking stigmatisation which has become a feature of more affluent sectors of society. There does appear to be potential to engage with people on a more conciliatory level, although it should be noted that early evaluations on the effect of the most recent ‘non-threatening’ or ‘soft’ national media campaigns linking smoking cessation and lung cancer awareness are at best, inconclusive (see, for example, Durkin et al. 2012) [54]. At root, it may be that, as Sweeny et al.[55] point out, although acquiring information can provide numerous benefits, people often opt to avoid engaging with it.