Results

DESCRIPTIVE DATA FOR THE TOTAL SAMPLE

Table 1 presents the data for the dependent variable and most independent variables for the total sample. The intentions to have a predictive test for HBC were mainly positive. Awareness of HBC was moderate, but only one fifth of the sample had heard about the predictive test for HBC and less than 10% had asked a medical doctor for information about these issues. As expected, the latter was more likely when breast cancer occurred in the family (χ²=22.40, p<0.01). The results of the three questions added up to the following awareness scores: 21% answered no to all three questions and got score 0; 57% answered yes to one question and obtained score 1; 19% obtained score 2 (two affirmative answers); 3% answered yes to all three questions and got score 3. The general attitude toward the development of a predictive test for HBC was favourable; the mean ratings for each of the four items were high, resulting in a mean general attitude score of 4.3. The mean perceived susceptibility score and the mean perceived seriousness score in our unselected sample were low to moderate. The mean importance ratings for the individual arguments for and against predictive testing for HBC are shown in table 2. The means for the benefits and the barriers subscales were 4.9 (SD=1.30) and 3.4 (SD=1.21), respectively. At test showed that the mean for the arguments in favour of testing was significantly higher than the mean for the arguments against (t=12.74, p<0.001).

VARIABLES ASSOCIATED WITH THE INTENTION TO HAVE A PREDICTIVE TEST FOR HBC IN THE TOTAL SAMPLE

Table 3 shows the univariate Pearson intercorrelations in the total sample between the dichotomised intention and the sociodemographic and independent variables. A chi-square test indicated that the dichotomised intention to have a predictive test for HBC was not associated with marital status (χ²=4.45, p=0.62). To identify factors having an independent association with the intention regarding the predictive test for HBC, a multivariate stepwise logistic regression analysis was performed, modelling uninterest. The results of the logistic regression in table 4 show that the perceived susceptibility score, which had an unexpected negative univariate association with the dichotomised intention (table 3), was not retained as a predictor in the regression. The benefits score was the first to enter the logistic regression with an odds ratio of 0.34. This means that when the benefits score increased by one unit, the odds to report a negative intention regarding the predictive test for HBC decreased by a factor of 0.34. Similarly, the odds ratio of 0.52 for the general attitude score indicates that with each increase in the attitude score, the odds to express uninterest decreased by a factor of 0.52. In other words, women who rated the arguments for a predictive test for HBC as more important (the benefits score) and women with a more positive attitude toward the development of such a test (the general attitude score) were less likely to report uninterest in the test. Higher odds of a negative intention were associated with a higher number of children and with the occurrence of breast cancer in the family. The value for the latter variable was quite high; the women in our sample who have (had) a relative with breast cancer had 3.61 higher odds of being uninterested in the predictive test for HBC than the women without a relative with the disease. The observation that women with breast cancer in their family were more doubtful about the predictive test is an unexpected and intriguing finding which was analysed further.

EXPLORING THE INFLUENCE OF THE OCCURRENCE OF BREAST CANCER IN THE FAMILY ON THE INTENTION TO HAVE THE PREDICTIVE TEST FOR HBC

To explore the influence of the occurrence of breast cancer in the family on the intention to have the predictive test for HBC, the sample was divided into two subgroups: a subgroup of women who have (had) at least one relative with breast cancer (the BC group, n=68) and a subgroup of women who have (had) no relatives with the disease (the No BC group, n=261). The results for the intention to have a predictive test for HBC for the separate subgroups are shown in table 5. Uninterest in a predictive test for HBC was observed much more often in the BC group (66% versus 25% in the No BC group, χ²=40.72, p=0.001). The spontaneous explanations for the reported intention that were given most frequently in both subgroups are also shown in table 5. The explanations involving preventive measures, knowledge, or certainty were more likely to be given by the No BC group, whereas the BC group was more inclined to use explanations referring to emotions like anxiety or to the emotional burden of being a carrier. Further, the following significant relationships with the occurrence of breast cancer in the family were observed (table 3, fifth row): the BC group was more aware of HBC and the predictive test and had a more negative attitude towards the development of such a test than the No BC group (the mean awareness scores were 1.3 and 1.0 respectively; the mean general attitude scores were 4.0 and 4.4 respectively), they perceived themselves as more susceptible to HBC (the mean susceptibility scores were 2.4 and 1.4 respectively), and they rated the arguments for having a predictive test for HBC as less important than the No BC group (the mean benefits scores are 4.1 and 5.1 respectively). The mean importance ratings of the arguments against were the same in the two subgroups (3.7 and 3.4 respectively). Subsequent within subgroup comparisons of the mean importance ratings show that the arguments for the predictive test were judged as more important than the arguments against in the No BC group (5.1>3.4, t=14.24, p<0.0001), while there was no difference in the BC group (4.1 and 3.7, t=1.54, p=0.13).

Separate stepwise logistic regression analyses explored whether the set of variables explaining a negative intention differed between the BC group and the No BC group. The results of these analyses are presented in table 6. In both subsamples, the benefits score was the first variable to enter the logistic regression; with each increase in the benefits score, the odds to express uninterest decreased by a factor of 0.40 in the No BC group and by a factor of 0.07 in the BC group, indicating that the explanatory value of the benefits score was higher in the No BC group. Age entered as an additional explanatory variable in the BC group; for each increase in age by one year, the odds to report a negative intention increased by a factor of 1.20. In the No BC group, the general attitude score and the number of children had an additional explanatory value; women without relatives with breast cancer were less likely to report uninterest in the predictive test for HBC when they had a more positive attitude towards the development of such a test, while they were more likely to report uninterest when they had more children.

Discussion

At first glance, our results confirm the strong interest in the predictive test for HBC found in most other studies. In the present study, the general level of interest was higher than the level observed in a study on the predictive test for Huntington's disease among a comparable community sample,37 in which about half the women expressed interest in the predictive test for this neurogenetic disease. This is not surprising given the crucial differences between breast cancer and Huntington's disease, for instance regarding the familiarity with the disorder and the possibilities for risk reduction and treatment. Also consistent with previous research,20 32 our sample considered the benefits of the predictive test for HBC on the whole as more important than its costs, and the possibilities of decreasing the breast cancer risk and to learn about one's children's risk were valuable arguments for testing. However, the contrast between women with versus without breast cancer in the family (the BC group versus the No BC group) forces a shift from this general level of discussion to a more specific one. The women in our sample who are related to a breast cancer patient (the BC group) were much less interested in the predictive test for HBC; a negative intention was expressed by two thirds of the BC group. This lack of interest is not only much larger than in previous studies among high risk women,13-15 20 but also more pronounced than in other studies among women with breast cancer in the family who have no increased risk for the disease.19 28 33 The discrepancy in our findings might be explained by cultural differences in dealing with risks and preventive options in the context of breast cancer38 or by specific characteristics of our sample. Being students at the Institute for Family Sciences, they might have been more interested in the relational and/or psychological issues in this context, rather than in technical information. Additionally, the kind of information that we provided on the predictive test for HBC, especially on the uncertainties and limitations involved, could have led to more reluctance for testing.39 Another explanation could be the method of recruitment of the study sample; our findings are based on a community sample without selection for a family history of breast cancer or breast related problems, while this kind of selection did take place in most other studies. The question about having a predictive test for HBC probably seems less hypothetical or unexpected in such selected samples, which might have resulted in less reluctant reactions than in our BC group. This may especially be the case when comparing with studies conducted on high risk families in a hereditary cancer registry and/or as part of a thorough research protocol. However, since the information on the number of relatives with breast cancer and/or the degree of kinship in our BC group was not always complete and could not be confirmed, more detailed comparisons with other studies are difficult.

Despite its lack of specification, the occurrence of breast cancer in the family had a strong independent association with uninterest in the predictive test for HBC. As shown by logistic regression analysis, women in the BC group were more than three times more likely to express uninterest in the test. To our knowledge, our study is the first to show such a strong negative relation between breast cancer in the family and interest in the predictive test. Shilohet al 33 observed a higher proportion of extreme rejection of the predictive test among at risk relatives of breast cancer patients (although their mean reported intention did not differ from the mean in the group of relatives without an increased risk). The greater uninterest among women who have breast cancer in the family in our study might have been the result of a stronger avoidance reaction in the BC group; because of their emotional involvement with breast cancer in the family and/or their higher perceived susceptibility to the disease, these women want to avoid the threatening possibility of detecting that they are carrying a HBC mutation. The stronger emotional involvement in the BC group is corroborated by the greater tendency in this group to give emotional explanations for the intention regarding predictive testing than in the No BC group. The operation of self-protective avoidance tendencies among at risk relatives of breast cancer patients was also suggested by Shiloh et al 33 and is in line with the idea that too much fear can lead to inaction (see also the conflict theory of Janis and Mann40). This idea receives further support from the negative univariate association between perceived susceptibility to breast cancer and the intention to have a predictive test for HBC observed in our study. Obviously, this contrasts with the predictions of the Health Belief Model. However, although the Health Belief Model has proven to be a valuable framework for understanding and predicting reactions to health threats, it has to be recognised that it is a rational, cognitive model which pays insufficient attention to emotional mechanisms. The importance of emotions in intentions to take a genetic test is illustrated in a study by Wroe et al 25; they found that emotional reasons were reported more frequently by people who had already actively thought about having a specific genetic test than by students for whom the test was hypothetical. Audrain et al 41 and Evers-Kiebooms et al 42 also stress the role of emotions in decisions about genetic testing. An example of an emotional factor that could have played a mediating role in not wanting a predictive test for HBC is self-blame or guilt about passing the disease onto the children.43 This might have especially been the case in our sample of students in Family Sciences, for whom family relations are likely to be of particular interest. Indirect support for this possibility stems from the logistic regression analyses on the total sample and on the No BC group; a higher number of children was significantly associated with higher odds of not wanting the test. However, this kind of independent association remained absent in the BC group.

An alternative explanation for the reaction of the BC group in our study against the predictive test is a more rational one. The experience with breast cancer in the family might have led to more awareness concerning the lack of effective preventive and/or therapeutic measures for the disease than in the No BC group, thereby reducing the perceived benefits of having a predictive test for HBC and/or increasing the perceived barriers or costs in the BC group. As such, the uninterest in the predictive test among the women in the BC group might have been the result of a rational decision based on the weighing up of their perception of costs and benefits. A similar systematic and rational process in the No BC group, with a different perception of barriers and benefits owing to the absence of familial experience with breast cancer, might have resulted in interest in the test. The differential perception of barriers and benefits of the predictive test for HBC according to the occurrence of breast cancer in the family is indeed supported by our findings: while the No BC group judged the benefits as more important than the barriers (a pattern that also emerged in previous studies14 20 28 32), the dominance of benefits disappeared in the BC group. On the other hand, as far as the explanatory value of these two Health Belief Model components is concerned, the results of the logistic regression analyses did not differ according to the occurrence of breast cancer in the family. In the BC group as well as in the No BC group, the perceived benefits were the first to enter the regression, while the perceived barriers remained absent. The failure of the perceived barriers component to enter the logistic regression can partly be explained by its substantial correlation with the other components of the Health Belief Model (table 3). However, this also suggests that the intentions in the No-BC group as well as in the BC group, despite the equivalence in the importance ratings of benefits and barriers in the latter group, were disproportionally affected by the perception of the benefits of the predictive test for HBC. A similar result was obtained by Lerman et al 32 in their multivariate analysis of BRCA1 test use. This underlines that providing comprehensive information on the complexities, the limitations, and the uncertainties involved in predictive testing for HBC (as well as on its benefits) is crucial for genetic counselling as well as public education purposes. Moreover, research has indicated that the intentions regarding genetic testing are not only influenced by the content of the information that is given, but also by the focus of attention.27 44Therefore, it is important to encourage and help people to consider all aspects of the test and all of its possible consequences, negative as well as positive ones. In a genetic counselling context, it was furthermore shown by Lerman et al 28 that a personalised approach extending beyond the standardised provision of information and giving room to the discussion of emotions can modify the evaluation of benefits and costs of the predictive test for HBC, perhaps through the more thorough processing of the information provided. Indeed, genetic counsellors should be sensitive to interpersonal differences in emotional and informational needs and priorities as well as in the diversity of reasons for predictive test requests.45 46 For instance, older women might be more concerned about their children's risks or about age related health problems than about their own genetic susceptibility to breast cancer.23 46 Our finding that older women in the BC group tend to be more reluctant about a predictive test for HBC for themselves is illustrative in this context. General concerns about the development of a predictive test for HBC as measured by the general attitude score seem not to be a priority in the BC group. Only for women who have no relatives with breast cancer does such a general measure play a role in their intention to have a test. When breast cancer does occur in the family, the influence of the general attitude score seems to be outweighed by the more personalised and instrumental measure concerning the perceived benefits of the predictive test. A similar finding concerning the influence of general attitudes towards health and medical testing has been reported by Shiloh et al.33

Overall, the results of our study show that the perceived benefits component is the most powerful Health Belief Model component in explaining interest in the predictive test for HBC. However, the absence of an independent association with perceived seriousness, perceived susceptibility, and perceived barriers does not mean that these factors are not important. They can influence other variables in the chain of the cognitive-emotional decision making process about the predictive test for HBC. Awareness of the possibility of their mediating role, as well as of the potential influence of emotions and/or more general concerns about the applications of the new genetic technology, is important for health care providers at different levels: at the level of genetic counselling, to facilitate free and informed decisions about predictive testing for HBC and at the level of community and clinical education programmes, to design targeted information addressing the specific needs and concerns of women at different risk levels. Although past experience with predictive testing for Huntington's disease47 and with carrier testing for cystic fibrosis48 has shown that reported interest in genetic testing cannot be equated with actual uptake, the results of our study and previous research on the predictive test for HBC are valuable in helping to reach these goals.