8.1 Interpretation of results and implications for relevant theory
8.1.2 Perception of risk of depression compared with perception of risk of
of other illnesses
A comparative evaluation of the strength of the Weinstein (1980) and Moore and Rosenthal (1996) model of perception of risk as a series of semi-independent illness specific constructs was undertaken with respect to depression and three physical illnesses. An important finding from this study was that this model for understanding perception of risk most significantly predicted perception of risk of depression. Whilst demonstrating the highest level of statistical significance, this model accounted for a larger percentage of variance in depression than for any other illness examined. Despite extensive utilisation within the physical health literature (Davidson & Prkachin, 1997; Eiser et al., 1993; Harris & Middleton, 1994; Moore & Rosenthal, 1996; Van Der Velde et al., 1992; Weinstein, 1982, 1987) this model appeared more relevant to perception of risk of depression than to perception of risk of any of the illnesses examined by this study. The strength of this finding was unexpected and warrants further replication and expansion in future studies.
As anticipated different patterns of significance emerged for each illness. For HIV/AIDS the illness specific model was not a significant predictor for perception of risk. For Diabetes and Breast Cancer the illness specific model was found to significantly predict perception of risk of the illness. In addition to varying levels of significance, each illness revealed individual factors that were significantly predictive in their own right. Perceived control was demonstrated to be an important factor influencing perception of risk of HIV/AIDS. Perceived seriousness was found to be the most significant factor influencing perceived risk of diabetes. Finally, like
perception of risk for depression, the most important factors influencing perception of risk for breast cancer were perceived control and knowledge of the illness.
Perceived control and knowledge of the illness have already been discussed in relation to perception of risk of depression but also appear applicable to physical illness, particularly to HIV/AIDS and breast cancer. Given the relationship between increased control and decreased perceived risk discussed in relation to depression, this finding is perplexing. Perhaps it is not surprising that perception of risk of HIV/AIDS is bound with control given that generally individuals state that they can take measures to ‘control’ their risk of exposure, for example by practicing safe sex (Moore & Rosenthal, 1996). What is surprising is that perceived control was significant for breast cancer but not for diabetes. Breast cancer is not generally viewed as a ‘controllable’ illness with predisposing factors including sex and heritability (De Noouer et al., 2001). Therefore individuals are less able to control such non- modifiable risk factors and reduce their perceived risk. The interesting question arising from this finding is how is do women conceptualise perceived control in relation to their subjective risk of breast cancer? Such a question deserves rigorous investigation by future research. The other puzzling finding from this study was that perceived control was not significantly predictive of perceived risk of diabetes. This finding was unexpected, particularly in light of the fact that according to the physical health literature the community considers diabetes a ‘controllable’ illness (Haley et al., 2003). Most public health campaigns addressing diabetes target those attributes that individuals can modify to reduce their risk, including diet and exercise (AIHW, 2002). It appeared from this sample that women did not consider these ‘controllable’ factors important in their subjective risk assessments for diabetes. This finding needs
further investigate potentially with a larger sample to determine if this finding is representative of broader community attitudes and to explore the implications more thoroughly.
Interestingly this study did not find that stereotypical representations of sufferers of illness significantly predicted perception of risk of illness for any of the illnesses examined. Conversely this study did support the argument explored by Moore and Rosenthal (1996) that stereotypical representations of sufferers of illness exist. In this study a number of stereotypical sufferers were named for depression that differed from those named for other illnesses. As stated in section 3.2.4 in Chapter Three the follow-on effect of maintaining stereotypes for depression sufferers implies that individual’s who believe that they do not fit the mould supported by the stereotype are less vulnerable or less inclined to accurately identify symptoms (Harris & Middleton, 1994). The development and maintenance of stereotypes further enable individuals to cognitively distance themselves from the ‘typical sufferer’ by identifying differences rather than similarities between themselves and the ‘typical sufferer’. As previously discussed, cognitive distancing enables individual’s to reduce their anxiety about vulnerability / risk of illness. Future research is needed to examine this relationship to explore the process of cognitive distancing from stereotypes of sufferers.
The finding that knowledge of a sufferer did not predict perception of risk for any of the illnesses examined was unexpected. Research has suggested that knowledge of a sufferer gives individuals experience with symptoms that may induce them to act, or at least to recognise symptoms (De Noouer et al., 2001). Despite such research and the work of Moore and Rosenthal (1996) that indicated that knowledge of a sufferer
was an important influence on perception of risk of illness, the current study did not support this notion. Although Millstein and Halpern-Felsher (2002) reported that knowledge of a sufferer caused variation in perceived risk, the repercussions of this interaction are unknown within the present sample. However, this finding may provide additional support for the operation of an optimistic bias. That is, perhaps the women within this study were able to maintain their optimistic belief ‘that it won’t happen to me’ on the basis that they had evidence that it had happened to another.
Despite the notable finding that perception of risk for each illness was influenced by a variety of individual factors, the overall predictive validity of the Weinstein (1980) model with the additional Moore and Rosenthal (1996) factors should not be under- appreciated. The finding that this model was more significantly predictive of perception of risk of depression than for perception of risk of any of the physical illnesses examined is important. The value of this project was the strength of the perception of risk model for depression, particularly as the finding provided evidence for the usefulness of modelling perception of risk of depression to understand and subsequently predict behaviour (Rickwood & Braithwaite, 1994).
It makes sense conceptually that a process of modelling factors that has been successfully undertaken in physical health literature would also be valuable in mental health. In fact, models in physical health literature have extensively undertaken the task of linking perceptions and behaviours and have demonstrated that perceptions of high personal risk will increase the likelihood of precaution adoption (Ajzen & Fishbein, 1980; Jordon & Oei, 1989; Moore & Ohtsuka, 1999; Ratliff et al., 1999). Models such as TRA and HBM have had many applications in physical health
literature including, sunbathing and sunscreen use (Hillhouse, Stair & Adler, 1996), use of oral contraceptives (Doll & Orth, 1993), breast self-examination in older women (Lierman, Young, Kasprzyk & Benoliel, 1990), exercise (Gatch & Kendzierski, 1990), participation in cancer screening programs (DeVellis, Blalock & Sandler, 1990), AIDS related behaviours (Fisher, Fisher & Rye, 1995) and smoking in adolescence (Maher & Rickwood, 1997). Models such as these are able to provide direct links between perceptions and behaviour, as well as provide an indication of people’s intentions with respect to health behaviours including help seeking (Rickwood & Braithwaite, 1994). Millstein and Halpern-Felsher (2002) argued for the need to link outcomes with behaviour to enable people to make meaningful judgements about their risk. Modelling perceived risk in mental health areas allows such links to be investigated while further contributing to the understanding of the role of perceived risk in decision making about health behaviours.
The importance of understanding the similarities and differences between mental and physical illnesses cannot be underestimated. The World Health Organization (2000) argued that the relationship between physical health and mental health was under- investigated. This study provided an important insight into the comparative nature of women’s conceptualisations perceived risk of illness, in this instance mental illness (Depression) and physical illness (HIV/AIDS, Diabetes and Breast Cancer). The strength in this rationale was that this study was the first to consider the predictive strength of optimistic bias as a model for perception of risk across both mental and physical illness. The present study confirmed the applicability and validity of a model of understanding perceived risk, traditionally applied to physical health research to mental health research and specifically to depression. The model known as optimistic
bias was demonstrated to be more applicable to depression than to HIV/AIDS, Diabetes or Breast Cancer. Such findings have important implications about who to target for health promotion campaigns, as well as for the management and treatment of depression. Factors such as ‘perceived control’ and ‘knowledge of the illness’ appear to have crucial parts to play and would be valuable targets for public health policy. The results from this study suggest that knowledge of the illness was not enough to guarantee engagement in health and protective behaviours, a finding that should be borne in mind by policy makers and service providers.
A limitation of this study relates to the generalisability of the findings. Given that this study was a non-clinical population study, the sample size may be considered low for the generalisability of the findings to women at large or the wider community. Further, the small sample size raises the issue the ability to make comparisons with other research in the areas of depression, perception of risk and optimistic bias. The sample size also raises the question of statistical validity. However, it is usual with small samples for differences between sample and population parameters not to be detected when they are true; the Type II error rate increases. This means that statistical relationships emerging in this study therefore can be considered to be reliable. Nonetheless, where general trends were indicated, a statistically significant relationship may have been apparent if the sample had been larger, particularly in the case of correlations that failed to achieve statistical significance. Despite these limitations, the findings from this study appear robust with strong argument for replication and expansion in areas of mental health research.