The structure and demographic correlates of cancer fear

Cancer occupies an almost unique position among diseases in terms of the fear it engenders. The word ‘cancer’ was once considered unacceptable in the public sphere, and even today, euphemisms such as ‘the Big C’ are common. In the 1950s, the British Empire Cancer Campaign concluded that education about early symptoms of cancer in Britain would create mass panic [1]; and similar issues have been raised in connection with campaigns to promote self-examination for early signs of testicular cancer in the UK [2]. Polls in the US and Europe find that at least half the population say they fear cancer more than any other disease [3–5], and around a third to a fifth say they fear cancer more than other potential catastrophes, such as violent crime, debt, and losing a job [3, 6].

Fear is an unpleasant emotion and the pervasiveness of cancer fear in the population may have implications for quality of life. In addition, cancer fear has been shown to be associated with screening uptake and presentation of suspicious symptoms, although both motivating and deterrent associations have been found (for an overview, see [7, 8]). In the light of the frequency of public statements about cancer fear, it is clearly a societally important matter. Most research to date has examined the behavioural consequences [7–10], and studies that have focused on cancer fear itself are mostly qualitative (e.g. [11]) or done in undergraduate student samples for whom the threat of cancer is less relevant due to their young age [12, 13]. A better understanding of the nature of cancer fear may help identify those who suffer from maladaptive and undue fears, and help explain why the behavioural responses seem to vary.

Fear as an emotion has a complex architecture, with cognitive, physiological and affective components that may be only loosely interconnected. These components are often not distinguished in the cancer context, and terms such as ‘cancer fear’ and ‘cancer worry’ are sometimes used synonymously (for example, see [7, 8, 14]). A failure to distinguish between different fear components may have contributed to the apparent inconsistencies in the behavioural effects of fear. For example, appraising cancer as uniquely frightening may lead to avoidance of the fear stimulus, while worry may encourage behaviours that will result in reassurance. Appraising cancer as uniquely frightening cannot therefore be assumed to translate into high levels of worry or avoidance [13] or show the same behavioural outcomes. To date, no studies have examined the associations between the appraisal of cancer as frightening, discomfort when thinking about cancer, and worry about cancer, nor investigated whether demographic and psychosocial correlates are consistent across the different components of cancer fear. Large datasets that include different indicators of cancer fear are needed to examine the interconnections between different fear components.

Some previous studies have explored associations between different aspects of fear, although these were mainly conducted in the context of specific cancer screening programmes. Consedine et al. [15] explored three aspects of fear: trait anxiety, prostate cancer worry, and screening fear, in a sample of older men in the US. All three were associated with lower income and education, and prostate cancer worry and screening fear, but not trait anxiety, was higher in Black men. Another study from the US found strong associations between cancer worry and cancer-related discomfort among women with and without a family history of breast cancer, but only moderate associations between cancer worry and trait anxiety [13]. Another examined the contributions of cancer worry and cancer-related distress to breast screening uptake in women at an increased familial risk, but did not report the associations between the different fear components [16]. No large studies on inter-relationships between the components of cancer fear have been conducted outside the US.

Little is known about the demographic correlates of individual fear components. Polls in the US and Europe that show cancer to be uniquely frightening have not explored whether certain subgroups are more likely to endorse this view, although a French survey noted that more women than men viewed cancer as their ‘number one’ fear [4]. The 2003 Health Information National Trends Survey (HINTS; [17]) and the Pittsburgh Lung Screening study [18] both showed higher cancer worry in women. Lower socioeconomic status (SES) was associated with cancer worry in both these studies, and in the UK Flexible Sigmoidoscopy Trial [19]. Ethnic minority status has been linked with higher cancer worry in studies in the US and UK [15, 20–22], although the association has differed by type of cancer and specific ethnic background [17, 23]. The reason that so little is known about the correlates of general cancer fear is partly that much previous work measured single components of cancer fear and focussed specifically on associations with screening uptake, without exploring the population distribution of fear (e.g. [15, 16, 21]).

An important potential confounder in studies of demographic variation in cancer fear is general anxiety. Anxiety tends to be higher among women and more socially disadvantaged groups [24, 25], so might explain sex or education differences in cancer fear. Results have been more varied in relation to ethnicity. African American men showed lower trait anxiety than White Americans despite higher prostate cancer worry in one study [15]. In the HINTS results, controlling for psychological distress reduced both gender and ethnic differences in cancer worry, although multiple other behavioural factors were also included as control variables, making it difficult to identify whether psychological distress was the key confounder. High trait anxiety has also been shown to increase the effect of media breast cancer messages on breast cancer fear [26].

The present study aimed to examine associations between three indicators of cancer fear that represent different components (having cancer as greatest health fear, discomfort thinking about cancer, and cancer worry) and associations between all three and general anxiety. It also explored the demographic correlates of the three components and examined whether effects were explained by differences in general anxiety and self-rated health. There is no prima facie reason to believe that the architecture of cancer fear would be different across cultures, but the socio-demographic correlates may vary between countries because of differences in healthcare provision, public knowledge of cancer, or beliefs about cancer prevention. Few previous studies of cancer fear have been conducted in the UK, a country that has a well-organised health care system, but also a tradition of the ‘stiff upper lip’, and a history of reluctance among health professionals to provide much public information about cancer for fear of scaring the public.

The questionnaire was mailed to 13,351 adults in the eligible age range in participating General Practices. The return rate was 59.7% (n = 7,971), of which 6,527 (82% of responses) had complete data on all variables. There were slightly more women (53%) than men (47%). More than half the respondents had no educational qualifications (63%), and the majority were of White ethnic origin (96%) and married or cohabiting (75%). The mean STAI score was 10.6, which is comparable to other community-based studies of older adults [34, 35]. Most respondents (70%) rated their health as good or excellent.

More than half this large, community-based sample of 55–64 year-olds in the UK had cancer as greatest health fear and felt uncomfortable thinking about it, and a quarter said they worried ‘a lot’ about cancer. The three indicators were moderately inter-correlated, suggesting some commonality between the three facets of cancer fear. This was supported by finding similar demographic correlates, with all three fear indicators being higher in women and respondents with lower levels of education, and none being associated with age or marital status. Ethnicity was the only demographic variable to show differential associations by fear indicator, with higher worry in non-White groups but no differences in the other indicators. As expected, general anxiety was associated with all three indicators, although the moderate size of the correlations is consistent with cancer fear being distinct from general anxiety. Controlling for general anxiety did not materially change the associations between the sociodemographic predictors and the cancer fear indicators.

The endorsement rate for having cancer as greatest health fear (59%) in this UK sample was similar to previous population surveys conducted in the US, UK, and France, which have found rates of between 35% and 62% [3, 4, 36]. Similar to findings in a French survey [4], more women than men in our study expressed having cancer as greatest health fear. The rate of cancer worry (25%) was also similar to previous studies. General cancer worry was reported in a quarter of UK adults [37], while studies about specific types of cancer showed worry about colorectal cancer in 13% to 23% of community-based samples in the US and UK [34, 38, 39], and worry about lung cancer in about 22% in the US [38]. Worry about breast cancer tends to be higher; around a third of women in the US, UK and Norway reported frequent or considerable breast cancer worry [40–42]. This could be due to the emblematic nature of breast cancer [43], but also to the generally higher rates of cancer worry in women. Similar to US based studies [17, 18], we found that rates of cancer worry tended to be higher in women and people with lower education. Ethnic differences in cancer worry are more difficult to compare across countries, because of the different ethnic minority groups. Overall, comparing our findings with the results of previous studies suggests that gender and education differences in cancer fear may be fairly consistent across Western countries.

The modest inter-correlation between the cancer fear indicators, and the fact that the number of people who identified cancer as their greatest health fear or experienced discomfort thinking about cancer was twice the number of people who experienced cancer worry, suggests that the items used in the current study reflect different aspects of the ‘cancer fear’ construct. This supports suggestions made by other authors that there could be distinct cognitive and affective components of what is often referred to as ‘cancer worry’ [8], and that these components may need to be distinguished to understand the role of cancer fear in cancer-related behaviours [7, 8]. Cancer worry has been associated with higher rates of cancer screening in some studies [7, 44], although this effect has not been entirely consistent (e.g. [20, 21]). But cancer fear may also promote avoidance of the fear stimulus, and has indeed sometimes been mentioned as a barrier to screening [45] and shown to impede cancer screening uptake [39, 42]. The present findings support observations made by other authors that the variation in measurement strategies in studies of cancer fear may have hampered our understanding of its behavioural effects, and that a better understanding of the construct is needed, including an exploration of whether or not it is a multi-dimensional construct [7, 8]. Although the components included in the current study may not be the only relevant ones, our findings give some support to the idea that distinguishing between different components of cancer fear could contribute to understanding of the concept. Further research is needed to determine whether the behavioural effects – for example on screening uptake - also vary by the specific cancer fear component, what additional fear components need to be distinguished, and how all of them could be measured more accurately. Understanding the effects of different fear components may also have implications for the evaluation of public health interventions, which may need to include multiple indicators of fear to accurately assess their effects [for an example of a public health intervention evaluation using multiple fear indicators, see [35]].

The moderate inter-correlations and differential endorsement rates of the three items used in the present study may also suggest a mechanism of protection against worry. High fear states seldom persist unregulated [7], and people who are uncomfortable when thinking about cancer may deploy strategies to reduce their daily worry about cancer. This would be consistent with other common fears, where discussion of the fear object can cause distress but emotional reactions do not necessarily intrude in daily life; as distinct from true cancerophobia [46]. That said, a quarter of the population worrying a lot about cancer, and more than half of the population experiencing discomfort about it might be seen as important issues for quality of life; and public health authorities may be rightly cautious about magnifying cancer fears. However, given that cancer rates are rising, and that there may be a motivating effect of cancer worry on screening uptake, three-quarters of the population not worrying about cancer may also be considered a problem. The difficulty of identifying the ‘right’ level of fear for potentially modifiable risks is a general problem in modern societies, and research is needed to get a better understanding of the balance.

The impact of cancer fear on national healthcare systems may be considerable. High cancer worry may motivate more frequent consultations with healthcare professionals to obtain reassurance [47–49]. Alternatively, for individuals who cope using denial or avoidance, discomfort thinking or talking about cancer could lead to delay in help-seeking for potential cancer symptoms [9, 50, 51] and interfere with cancer screening uptake [40, 52, 53]. It could also affect the success of public education on cancer. Miles et al. [10] showed that people with higher levels of cancer fear were more likely to avoid cancer information, including information on the benefits of early detection, thus potentially perpetuating negative beliefs about the scope to reduce cancer risk.

Previous research in an undergraduate sample found moderate correlations between three indicators of cancer fear and dispositional worry, suggesting that cancer fear may be partly due to, but is also distinct from, general anxiety [12]. A study of prostate cancer worry also found only moderate correlations with trait anxiety [15]. The results of the present study are important because they indicate a similar pattern for general cancer fear in a community sample at an age when the threat of cancer is more relevant.

This study has several limitations. First, it was part of larger study that was not designed primarily to investigate cancer fear, and so the selection of predictor variables may not have been optimal. Nonetheless, the large sample size was an advantage. Additional predictors could be considered in future studies, including personal or family history of cancer and perceived personal risk. Participants were aware that it was a survey about cancer and non-responders may have been even more afraid of cancer than responders. A larger proportion of our sample (63%) than the national average of those born between 1936 and 1945 (45%; [54]) reported not having any educational qualifications, although this is unlikely to have influenced the associations with cancer fear that were found in this study. The large proportion of participants without educational qualifications may be due to the location of the General Practices through which they were recruited, which were in more deprived areas of the country. Consistent with the proportion of ethnic minorities in the older British population [55], the majority (96%) of respondents in our sample were from a White background, which limited the power of the study to detect ethnic differences and made it difficult to interpret our findings about the influence of ethnicity on cancer fear. In addition, some evidence suggests that cancer fear is generally lower in those who are older [8], but investigation of age effects was restricted by the narrow age-range of the sample. Lastly, the three components of cancer fear were each measured with single items to reduce participant burden in the main study. There are validity problems associated with single item measures including limited reliability and a limit on the maximum size of any associations, although the effects are offset to some extent by the large sample size.

Cancer’s highly feared status was endorsed by the majority of a large community sample of 55–64 year-olds, and this was associated both with discomfort in thinking about cancer and frequent worry. Women and people with less education or from ethnic minority backgrounds are disproportionately affected, independent of health status and general anxiety. Because cancer fear is an unpleasant emotion, as well as potentially influencing cancer-protective behaviours, it is important to gain a better understanding of its origins and consequences, and find ways to minimise its impact on quality of life without undermining participation in cancer prevention.