Who is studied in qualitative research?
For the most part, existing qualitative research has studied people considered to be at increased cancer risk, which has included people with: (i) symptoms or signs suggestive of increased risk (such as breast symptoms or abnormal pap smears); (ii) family history or genetic susceptibility (most commonly to breast and ovarian cancer); (iii) increased behavioural risk due to, for example, drinking alcohol, smoking or using sunbeds; (iv) increased environmental risk due to exposure; and (v) cancer survivors at risk of recurrence or second cancers. Interestingly, the exceptions, studies of ostensibly ‘healthy’ people, focused on specific age groups or ethnic groups, who were seen to be at higher-than-usual risk, either due to higher rates of risky behaviour or lower rates of participation in screening programmes: targeted groups included young women, African-Americans, African and Latin-American immigrants to the United States and Asian immigrants to the United States, Canada and Australia (the majority of studies of migrant groups took place in the United States).
What methods are used in qualitative cancer risk research?
Data was gathered either through in-depth interviews, focus groups or analysis of extant texts. Data analysis methods, where stated, were described as thematic analysis, discourse analysis, ethnography, thought unit analysis, grounded theory, life course analysis and content analysis. While several studies were positioned within sociological theories of risk, analyses were largely inductive and focused on for the most part on cognitive or psychological phenomena.
Analytic categories derived from the qualitative synthesis
Fifty-five analytic categories were developed from the line-by-line codes which were, in turn, synthesized into eight overriding analytic categories (Table 1): (i) perceptions of risk factors; (ii) process of risk perception; (iii) seeking control and taking responsibility (motivational factors); (iv) experiencing cancer directly; (v) constructing risk temporally; (vi) embodying risk; (vii) identifying with risk; and (viii) constructing risk in a social context. Categories 1, 2 and 3 seemed to us to have the most overlap with existing psychological research into risk perception. Hence, the names given to these categories would be familiar to quantitative psychologists. While these categories reflect broadly distinct domains, some degree of overlap was unavoidable and expected. Talk of fatalism, for example, was relevant to several of our categories (health beliefs, rationalization processes and the perception of control and moral responsibility), and the various means of cognitively processing risk information emerged not only an analytic category in its own right, but also as an important component of several other categories. There was also unavoidable overlap between identity, embodiment and time. We recognize that it would be possible to categorize these papers differently, with different emphases and with more or fewer divisions. We emphasize, therefore, that our goal in generating these eight categories was not to try to represent eight distinct and objective ‘realities’, but rather to generate an inductively derived categorization for use by clinicians and public health practitioners.
Perceptions of risk factors
Many of the qualitative studies of cancer risk explored people’s beliefs and attitudes regarding cancer risk factors and their management. Some of these studies focused on specific risk/protective factors, including diet, sexual activity, pesticide exposure, sunbed use, infection, heredity and genetic risk. Beliefs of different groups were often compared, including men and women, adults and adolescents, different ethnic groups, and lay people and health professionals. A number of culturally specific beliefs about risk were identified. For example, in a study of the influence of traditional Chinese beliefs on cancer screening behaviour among Chinese-Australian women, Kwok and Sullivan13 found that these women are heavily influenced by cultural traditions related to the lifecycle, and disease prevention, and tended to take a fatalistic view of cancer risk. Beliefs and attitudes were found to stem not only from cultural belief systems but also from personal health narratives. In a detailed narrative analysis of one woman with breast cancer, for example, Lawson14 made the connection between a deprived and traumatic life history and various ‘misperceptions’ about the cause of cancer, such as cancer being caused by trauma to the breast.
Only a few studies derived new risk typologies of beliefs about health risk inductively from participant’s accounts (something that can only be achieved through qualitative research). Chapman,15 for example, studied the perceived link between diet and breast cancer and inductively developed three perspectives on what kind of diet might be protective against cancer: a ‘traditional’ perspective (meat and potatoes); a ‘mainstream’ perspective (a balanced, low-fat diet); and an ‘alternative’ perspective (avoiding artificial additives and modifications).
Even where no direct comparison was made between the beliefs of lay people and health professionals, such a comparison was usually implicit in the interpretation of results. With only a few exceptions, whenever lay beliefs were found to differ from the biomedical model, these differences were considered to represent failures in knowledge, understanding and/or or memory or, at best, to represent a state of understandable uncertainty in the face of conflicting information.
Process of risk perception
Many qualitative studies focused on the idea that the process by which risks are conceptualized is not purely ‘rational’ or ‘objective’ but rather is (re)configured through a number of cognitive processes. These included simplifying and suppressing information, making various kinds of comparisons and rationalizing desired beliefs. In a study of reasons for sunbed use, for example, Murray and Turner16 found that people engage a number of strategies to justify their ongoing use including suppressing undesirable information about the riskiness of their behaviour and rationalizing their decision on the grounds that the short-term benefits outweighed long-term dangers. And in a study of women with a family history of breast cancer, Werner-Lin17 demonstrated the way in which women compared themselves to relatives, with the aim of finding phenotypic differences, so as to reduce their sense of being at risk.
In addition to focusing on cognition, many qualitative studies emphasized the way in which discussions of cancer risk and personal experience of cancer (particularly at times of uncertainty) evoked strong emotions including guilt at not protecting oneself or others; shame associated with an expectation of stigma; as well as the obvious sense of anxiety and ‘ontological uncertainty’. It was not surprising, therefore, that many studies found that a person’s emotional state could affect the meaning that they give to cancer, their construction of risk and their associated behaviour. McAllister,18 for example, found that people at increased genetic risk of colon cancer can be more or less ‘engaged’ (by which was meant that some people have greater ‘cognitive and emotional involvement’ with their increased genetic risk), and that those who were more ‘engaged’ were more likely to believe themselves to be carriers, irrespective of the risk estimate that they had been given by counsellors. Several studies attempted to link emotional responses to behaviour, and a strongly emotional response to risk was sometimes associated with the perceived need to engage in protective behaviour, and at other times with disavowal, denial or avoidance of risk.
Other studies showed that how the process of risk construction was shaped by intuitions. Chalmers and Thomson,19 for example, identified three methods by which women who had cared for relatives with cancer went about personalizing their own risk: reasoned (reflecting an objective and knowledgeable view of risk), intuitive (especially when information was incomplete or emotional impacts of cancer were unresolved) and variable (especially during early experiences with cancer when the sense of risk could vacillate strongly). They described intuitive risk perception as an instinctive, imagined, semi-stable perception of risk based upon an emotional interpretation of information and experience.
Seeking control and taking responsibility (motivational factors)
In many of the analysed qualitative studies, a sense of control and the related ability to take action (whether information-seeking, screening or preventive behaviour) enabled people to avoid fatalism or existential uncertainty, reduce anticipatory regret, put their minds at rest (or at least know what was to be expected), find a degree of security and ‘get on with their lives’. A number of studies demonstrated that people went to great lengths to construct their cancer risk, or other people’s cancer experiences, such that a sense of control was sustained. Sanders et al.,20 for example, observed that, in an effort to maintain a sense of control, some participants attempted to play down their genetic (and therefore unmodifiable) risks, even in the face of strong family histories of cancer. People had particular difficulty reconciling evidence that others had become ill despite engaging in protective behaviour, and needed to either come to terms with a fatalistic stance and accommodate uncertainty or (as discussed previously under ‘processing’) recalculate or rationalize their view of events in such a way that a belief in control could be sustained.
Information seeking was one of the major strategies by which people attempted to achieve control, but attitudes towards knowledge and surveillance were mixed and people derived varying degrees of certainty from available information. While some struggled with uncertainty and saw information as a ‘lifeline’, for others uncertainty itself was a psychological resource and fatalism and inaction seemed a reasonable, and even comforting, stance. Attitudes towards information were particularly ambivalent if this information was not associated with clear means of control or if it generated more uncertainty, too much information or too many competing options. Ryan and Sugg Skinner,21 for example, found that first degree relatives of women with breast cancer were ambivalent about risk counselling in the absence of preventive measures. The extent to which information and action were linked was particularly clear in a study by Phelps et al.22 which found that a significant subset of women undergoing genetic testing for breast cancer were paradoxically disappointed at being told that their risk was low or moderate, because this meant that immediate action was not warranted.
Several studies demonstrated an implicit link between taking control and a sense of moral responsibility. While many participants saw themselves as having a moral responsibility to take control and engage in personal advocacy (as, for example, in Chapple et al.’s23 study, which demonstrated the extent to which prostate cancer screening was construed as responsible behaviour), some resented the sense of personal responsibility and associated blame. These people used a lack of control (e.g. having ‘cancerous genes’ or mistrusting health professionals), a sense of not needing to take control (e.g. not having a genetic risk, so not having to worry), or a lack of certainty about risks, to absolve themselves of the moral responsibility to engage in self-protective or screening behaviour.
Experiencing cancer directly
Several qualitative studies emphasized that risk perceptions were strongly influenced by one’s own personal experiences of having cancer or of caring for a family member with cancer. In these cases, risk was first experienced, and subsequent biomedical screening or the development of cancer tended to simply confirm what was already experientially known. Several studies demonstrated the ways in which an experience of cancer heightened risk in people’s awareness and showed that the sense of being at risk was greatest if the experience was more direct and traumatic.
In various studies, people described themselves as ‘feeling’ at risk and developing a kind of ‘tangible’ knowledge about risk. This kind of personal, vicarious or experiential knowing seemed to override epidemiological evidence if, for example, one or one’s relative had developed cancer despite their being at a low risk epidemiologically. D’Agincourt-Canning,24 for example, studied people with a family history of breast and ovarian cancer, and observed the power of ‘empathetic’ knowledge about cancer which often took precedence over objective clinical estimates of risk. Indeed, familiarity with cancer, either in oneself or in a family member, could breed contempt for biomedical predictions and subsequent testing. The power of personal experience was also evident in studies which suggested that cancer education could not have an impact unless people were first made receptive by personal experience.
Constructing risk temporally
Closely related to experiencing cancer directly, several studies emphasized that people’s perceptions of cancer risk were related to the phases of life, and waxed and waned over time. Some studies focused on how people caring for relatives developed their own sense of risk, which tended to emerge only after the phase of ‘living through’ cancer with the relative. Chalmers and Thomson,19 for example, identified three phases of becoming an ‘at risk person’: living the cancer experience, developing risk perception and putting risk in its place. Other studies showed how a family history of cancer might become salient at a particular point in time. Werner-Lin17 observed that, for women at increased risk of breast cancer, the age at which their relative was diagnosed became a temporal ‘danger zone’ in their own lifecourse. Reaching the ‘danger zone’ led to a fairly abrupt onset of a sense of vulnerability and increased efforts to cope and achieve control.
The temporality of the at-risk experience was described in a particularly rich way by Kenen et al.,25 who observed that in its chronicity and variation over time, the state of being at risk had much in common with chronic illness.
A number of qualitative studies showed that people experienced risk as a ‘corporeal’ or ‘embodied’ phenomenon. For some people, personal risk was something that was at first ‘silent’ (and perhaps only detectable through screening), then ‘activated’ by the development of a symptom or sign. In their study of men previously diagnosed with prostate cancer, for example, Hedestig et al.26 noted that prostate-specific antigen (PSA) testing was seen by some men as a ‘message from the silent body’. This is in contrast to other studies which demonstrated that people assume that if they are asymptomatic, they cannot possibly be at risk, as evident for example in Weitzman et al.’s27 study of lay understandings of the risks of colorectal cancer.
An embodied sense of risk could have profound effects on the conceptualization and experience of risk. People with bodily changes simultaneously confronted current disease and the possibility of future disease. They also faced a threat from within themselves which became incorporated into their sense of self. At worst, people viewed their own bodies (e.g. those with precancerous cervical abnormalities28) as treacherous because these bodies were the source of cancer risk.
Identifying with risk
Qualitative research demonstrates clearly that cancer risk impacts upon identity. In many cases being ‘at risk’ demanded use of the health system, affected relationships, and led to biographical disruption (that is, it disturbed the lifecourse one had imagined for oneself). Scott et al.,29 for example, found that people who were at increased genetic risk of cancer tended to see themselves in a ‘liminal’ position, unable to identify with either the healthy or the sick, and that such individuals consequently sought recourse to systems of medical surveillance that could continuously monitor their state of health. In keeping with the findings we previously reported regarding control and responsibility, this sense of liminality could be more acute for those who were found to be at low genetic risk and were thus excluded from surveillance and care. In addition to disrupting one’s life course, the state of being at risk could also create a sense of a compressed life, a foreshortened future and a sense of urgency surrounding the achievement of life goals, such as having children.17
While being at risk is clearly disruptive to identity, several studies also showed that people could accommodate, integrate, or normalize risk (and associated surveillance) into their sense of self, even becoming redefined by and ‘living’ their risk of cancer. This was particularly the case when risk was experiential or embodied, and where it evolved over time. Hallowell et al.,30 for example, described the way in which a heightened bodily awareness of risk and a high level of monitoring eventually gave way to a state in which people had learned to ‘accommodate’ the risk of a recurrence within their lives. Identification with cancer was not, however, universal, and some studies demonstrated that people adopted a variety of stances to the information provided by mainstream medicine, including a variable willingness to take on the role of an ‘at risk’ person and engage in the related screening/preventive behaviours.
Qualitative research also demonstrates that the extent to which one incorporates risk into identity subsequently influences one’s perception of ongoing risk and one’s response to further risk-related information. Interestingly, for some participants, once cancer risk was incorporated into identity, further risk-related evidence seemed to cause relatively little further biographical disruption.30
Constructing risk in a social context
Several studies emphasized the effects of social context on risk construction. Some studies focused on the clinical context within which risk was constructed, and found that people expressed variable levels of trust in their health professionals and the health system (a position which was frequently based on previous care-seeking experiences). Other studies also focused on personal relationships and showed that these relationships could have a powerful role in shaping the experience of risk. While relationships could be helpful, they could also be a source of stigma and shame. Cramer Bertram and Magnussen,31 for example, identified the social stigma that could be associated with an at-risk status (in this case an abnormal pap smear due to infection with HPV). Intimate relationships could also make the experience of risk highly emotionally and morally charged (see also ‘control and responsibility’), particularly where children were involved. Indeed, Hallowell32 came to the conclusion that, for women at risk of ovarian cancer, managing risk was less an act of rational self management than the altruistic response of an emotional and relational self. Controlling risk was important in part because of the need to prevent others from suffering.
A few studies (although, interestingly not many) took a broader view of the social and examined the ways in which a person’s perception of risk might be affected by their embeddedness in particular communities with their collective memories (e.g. a shared sense of mistrust on the basis of previous victimization); specific socio-economic stressors (perhaps with a correspondingly weaker focus on individual risk prevention); shared understandings of disease and risk (community epidemiologies) and expectations with respect to what constitutes acceptable environmental exposure, and acceptable behaviour. Salant and Gehlert,33 for example, observed that, for African Americans, risk perception was shaped by their community’s shared nostalgia for a better, less risky, time; its sense of communal victimization; and a number of competing communal concerns. Communities also had specific risk narratives (like illness narratives) which determined appropriate risk-related behaviour, including the degree to which one was expected to take responsibility and be active in preventing illness. Wong and King,34 for example, observed that risk understandings were influenced by the dominant illness narrative of personal responsibility and restitution within Anglo-Western cultures. One participant in their study, for example, chose to dismiss public health statistics and recommendations about screening because they did not seem adequately proactive. Consequently, this participant was outraged when medical practitioners refused to perform a mammogram for her 28-year-old daughter. In keeping with the findings related to identity, these socio-politically derived ideas about risk (and appropriate risk-related roles and behaviour) were sometimes adopted wholeheartedly by individuals and sometimes resisted.