Introduction
Recent advances in genetic research have enabled us to identify individuals at risk for a wide variety of medical conditions due to their genetic makeup (Collins et al.
2003). At the same time, these advances have created the need to educate and guide these individuals (Lerman et al.
2002). Informing them of their hereditary risk and of the options for how to deal with this risk is the primary aim of genetic services (Wang et al.
2004). Genetic services involve both genetic counseling and genetic testing; of these, genetic counseling in particular aims to enable at-risk individuals to accurately identify, understand and adaptively cope with their genetic risk (Biesecker
2001; Pilnick & Dingwall
2001).
The National Society of Genetic Counselors’ (NSGC) Task Force defines genetic counseling as “the process of helping people understand and adapt to medical, psychological, and familial applications of genetic contributions to disease” (Resta et al.
2006, p. 79). As such, genetic counselors are faced with three important tasks: (1) to interpret family and medical histories to enable risk assessment, (2) to educate counselees about issues related to heredity, preventive options (e.g., genetic testing), and personal risk, and (3) to facilitate informed decisions and adaptation to personal risk (cf. Trepanier et al.
2004). The latter task may be considered the “core” (i.e., the desired outcome) of genetic counseling, with the former tasks in service of its fulfillment.
Informed decision making and adaptation to personal risk, however, are abstract concepts that cannot easily be assessed. As such, several measures have been developed to assess the efficacy of genetic counseling. Kasparian, Wakefield and Meiser (
2007) summarized 23 available measurement scales which include satisfaction, knowledge, psychological adjustment, and risk perception measures. Although each of these measures significantly contributes to our understanding of the effect of genetic counseling, risk perception measures (and especially risk perception accuracy) may be regarded as one central concept. Indeed, several influential models of health behavior, such as the Health Belief Model (Janz & Becker
1984), the Protection Motivation Theory (Rogers
1983), and the Extended Parallel Process Model (Witte
1992), posit that adequate risk perception acts as a motivator to take (preventive) action and, as such, is a prerequisite of preventive behavior. Moreover, risk perception and risk perception accuracy have been shown to be related to several other important outcomes of genetic counseling, such as coping (Nordin et al.
2002), worry (Hopwood et al.
2001), and anxiety (Meiser et al.
2001).
The effect of genetic counseling on risk perception has been heavily examined during the past two decades, from early research into reproductive genetic counseling (e.g., Humphreys & Berkeley
1987) to recent studies into genetic predispositions to cancer (e.g., Bjorvatn et al.
2007). While these studies are valuable in their own right, few have investigated the effect of genetic counseling on risk perception
accuracy. Indeed, to facilitate informed decision making and adaptation to personal risk, counselees must have
accurate risk perceptions.
In their 2002 meta-analysis, Meiser and Halliday (
2002) identified only six studies that assessed the effects of genetic counseling on risk perception accuracy. Their meta-analysis showed that individuals at risk for breast cancer significantly perceive their own risk more accurately after genetic counseling. In particular, they observed an average increase of 24.3% of the participants who accurately estimated their personal risk after counseling. A systematic review by Butow and colleagues (
2003) 1 year later confirmed the positive impact of genetic counseling in breast cancer risk perception accuracy, although 22–50% continued to overestimate their risk even after counseling.
Research thus suggests that genetic counseling may indeed improve risk perception accuracy in some individuals. However, Meiser and Halliday (
2002) and Butow et al. (
2003) only included studies examining breast cancer risk. To date, there is no systematic review or meta-analysis which examines the effect of genetic counseling on perception of genetic risks in general. Thus, the purpose of the present review is twofold: (1) to provide an updated overview of the impact of genetic counseling on risk perception accuracy in papers published between January 2000 and February 2007, and (2) to extend the results of Meiser and Halliday’s (
2002) meta-analysis and Butow et al.’s (2003) systematic review to other genetic conditions.
Discussion
The purposes of this review were (1) to provide an updated overview of the impact of genetic counseling on risk perception accuracy from January 2000 until February 2007, and (2) to extend the Meiser and Halliday (
2002) meta-analysis and the Butow et al. (
2003) systematic review to other genetic conditions. Overall, the studies showed that an increased proportion of individuals correctly perceived their risk after counseling rather than before, and those who did not had smaller deviations from their objective risk than before counseling. These positive effects were sustained even at follow-up 1 year later. Some studies, however, observed no positive effect of genetic counseling, or only for low-risk individuals. These results are in line with those reported in the 2002 Meiser and Halliday meta-analysis and the 2003 systematic review conducted by Butow and colleagues.
The research in the present review may shed some light on why some studies observe positive effects of genetic counseling on risk perception accuracy and others do not. First, one study (Codori et al.
2005) that observed no effect explicitly mentioned that personal risk information was not communicated during the relevant counseling session. Second, the provision of information about the role of family history, as recommended by the NSGC Task Force, may provide an appropriate context in which counselees can make sense of the risk information (cf. Codori et al.
2005), resulting in accurate risk perceptions. Third, some counselors may go to great lengths to explain risk information in terms the counselees can understand (cf. Kent et al.
2000). Unfortunately, research has shown that verbal and numerical risk estimates often do not coincide. That is, verbal risk information results in more variability in risk perception than does numerical information (Gurmankin et al.
2004b). Bjorvatn et al. (
2007), for example, observed incongruence between numerical and verbal measures of risk perception. Similarly, Hopwood et al. (
2003) observed that counselees included a wide range of numerical risk estimates within the same verbal category. The significance of this is discussed below, where we present the implications of our study for clinical practice. Finally, several studies (Pieterse et al.
2006; Rothemund et al.
2001) that observed no effect of genetic counseling on risk perception accuracy had small sample sizes, and thus may not have observed a significant effect due to power limitations.
The present review has several important implications for future research. First, we selected a large number of studies assessing risk perception changes as a result of genetic counseling. However, we had to exclude 19 of these studies because they did not explicitly link risk perception to an objective risk figure. Assuming that researchers are aware of these objective risk figures, future studies should link risk perception changes to objective risk figures to assess changes in risk perception accuracy.
A second implication concerns the definition of risk perception accuracy, which differs between studies. For instance, in several studies accurate risk perception is defined as falling within a certain category (e.g., Bjorvatn et al.
2007; Kelly et al.
2003; Lidén et al.
2003) or within 50% of the counseled risk (e.g., Pieterse et al.
2006; Rothemund et al.
2001), while the majority define it as the correct counseled risk estimate (e.g., Bowen et al.
2006; Hopwood et al.
2003; Tercyak et al.
2001). Additionally, the reviewed studies based the counseled risk estimate on different methods, such as family history assessment (Huiart et al.
2002), Gail’s score (Bowen et al.
2006), or the BRCAPRO procedure (Kelly et al.
2003). These issues reduce our ability to compare the results of the studies, thereby lessening their value. Future researchers should define risk perception accuracy as correct counseled risk, and base their risk estimate on generally accepted and applied methods to allow for better interpretation of the results and comparison between studies.
A third, related issue concerns the type of outcome measure used: several studies report changes in the proportion of individuals who correctly perceive their risk, while others report the degree of overestimation or underestimation as a measure of risk perception accuracy. Researchers are advised to include both measures in their studies, as both provide valuable information about the effect of genetic counseling on risk perception accuracy.
Further, we observed that the quality of the genetic counseling descriptions (in those descriptions that were present) was poor. Although the counseling sessions were labeled as standardized, they were described in general terms, such as “discussion about the risk” and “information was given about how hereditary factors contribute to disease.” These general descriptions leave room for substantial differences between counseling sessions. This is especially problematic given that perceptions of genetic risks before genetic counseling can determine the content of the counseling session (Julian Reynier et al.
1995), which tends to alter patient outcomes (Lobb et al.
2004).
Differences in the quality of the counseling session content may well explain the fact that not all studies in the present review observed a positive effect on risk perception accuracy. Future studies should therefore try to link the content of the counseling session to risk perception to determine which feature of the session actually contributes to improved risk perception accuracy (cf. Pieterse et al.
2006, or Shiloh et al.
2006). The present review provides some insight into how the content of the counseling session relates to risk perception accuracy. Indeed, the provision of information on the role of family history was observed to positively impact risk perception accuracy, perhaps because it creates a context in which the counselee can understand the information. Additionally, forcing numerical risk estimates to fit lay terms to aid counselees’ understanding may lead to inaccurate risk perceptions (Kent et al.,
2000). A possible avenue for further research may be to link effectiveness to certain sociodemographic variables. We could then examine the influence of known psychological differences between certain groups, which is a more complex process and should thus occur later in time. By associating these psychological differences to the effectiveness of genetic counseling, we may be able to identify the processes responsible for the positive effect of genetic counseling on risk perception accuracy. Knowledge of such processes will enable us to match the session’s content to these processes and thus to increase the session’s effectiveness.
Finally, we observed a relative lack of diversity in research on genetic counseling and genetic test result disclosure in terms of the genetic disorder under consideration. Although genetic counseling and testing can be effective for a variety of disorders (Biesecker
2001; Lerman et al.
2002; Pilnick & Dingwall
2001), most recent studies focus on their impact on cancer risk perception, particularly breast cancer. Although genetic counseling on cancer has been shown to positively affect risk perception accuracy, this does not guarantee it will do the same for other genetic conditions. Extensive research is needed to assess whether genetic counseling also effectively enhances risk perceptions for other genetic predispositions.
Based on the results, we have formulated some implications for practice. First, in accordance with the recommendations of the NSGC Task Force, we again strongly urge genetic counselors to discuss the role of family history and perform a family history assessment. We suggest that this information is an important factor in accurate risk perception because it may provide the necessary context in which counselees can understand the risk information. Indeed, the results seem to suggest that the provision of such information is positively related to risk perception accuracy. While this implication may seem redundant as it repeats the earlier recommendations by the NSGC Task Force, we nonetheless repeat it here since several studies in this review did not mention communicating this information to the counselee (Codori et al.
2005; Meiser et al.,
2001; Kaiser et al.
2004; Kent et al.
2000; Rothemund et al.
2001).
Second, while explaining risk information in lay terms seems to be a useful strategy to help counselees to better understand their risk (cf. Trepanier et al.
2004), the one study that explicitly mentioned doing so did not observe a significant effect on risk perception accuracy (Kent et al.
2000). Moreover, there appears to be incongruency between verbal and numerical risk estimates (e.g., Bjorvatn et al.
2007; Hopwood et al.
2003). Both types of risk estimates, however, possess qualities that would make them especially suited for counseling. Compared to verbal risk estimates, numerical risk estimates have been shown to increase trust in (Gurmankin et al.
2004a) and satisfaction with (Berry et al.
2004) the information. On the other hand, individuals have been shown to more readily use verbal information when describing their risk to others (Erev & Cohen
1990) and when deciding on treatment (Teigen & Brun
2003). We therefore advise genetic counselors to present numerical risk estimates first, as they are accurate, objective information. The patient may then be asked what that risk estimate means to him or her. The patient’s verbal response will provide an opportunity for further discussion of the meaning and impact of the risk information. Genetic counselors should, however, be aware of the disadvantages of verbal information in accurately communicating risk information.
A third, related implication concerns the presentation of numerical risk information. Research has shown that visual presentation of risk information (e.g., odds or percentages) may be better understood than written presentation formats. Indeed, there seems to be general agreement that graphical formats, in comparison with textual information, are better able to accurately communicate risk information (Schapira et al.
2001; Timmermans et al.
2004) although contradictory evidence has also been published (Parrot et al.
2005). Furthermore, graphical information seems to have a larger impact on risk-avoiding behavior than textual information (Chua et al.
2006). We therefore advise genetic counselors to use visual aids when communicating numerical risk information (cf. Tercyak et al.
2001).