The multiple risks and uncertainty of the risks, e.g. because of the incomplete penetrance of BRCA1/2 mutations, introduces complexity into risk communication regarding hereditary and familial breast cancer. Based on these complex risk contexts, decisions have to be made such as whether and when to start mammography screening, and, assuming eligibility, whether to test for mutations in BRCA1/2, and subsequent decisions in the case of positive or negative genetic test-outcome. Understanding risks is considered to be crucial for informed decision-making [7], whereas inappropriately high (or low) perceptions of risk can lead to various and potentially harmful decisions, for example overscreening due to overestimation of risk [8]. Overestimation of risk is also shown to be positively correlated with breast cancer worry, which may affect psychological well-being negatively [9]. Previous studies found a high percentage of unaffected women with a family history of breast cancer overestimating their breast cancer risk, although underestimating also occurred (see review [10]). In general, genetic counselling decreases generalized anxiety [11], and significantly improves the accuracy of women's perceptions of their breast cancer risk [11,12], but not necessarily to the correct level [11], or for the long term [12]. Therefore, it is important to identify strategies that will optimize understanding of risk.

Risk communication in genetic counselling is difficult, not only due to the complexity of the risks involved, but also due to problems that counsellors themselves experience in conveying these risks in an understandable way [13]. In addition, the counsellee's understanding and retention of the information given may be problematic. Personal experiences with breast cancer and counsellee's personal characteristics could have mediating effects on the efficacy of risk communication. For instance, the impact of a woman's family history can influence her risk perception prior to counselling through phenomena such as "anchoring" or availability heuristics [14]. Additionally, people with a high need for cognition potentially process risk information more thoroughly and have a higher need for additional risk information compared with people with a low need for cognition [15]. Moreover, women who face difficult decisions with respect to, for instance, genetic testing, may cope differently as a result of their dispositional decision-coping patterns [16].

Regarding the risk communication process, research has shown that the context and format in which genetic risks are presented affect people's risk perception, and their subsequent decisions [17,18]. Although risk presented in numeric terms are precise and preferred by healthy counsellees [19] and patients [20], risks in verbal terms (e.g. "a high risk") are generally more familiar and easier to understand. However, the inherent vagueness of verbal risks leads to substantial variation in their interpretation [21,22]. Risks presented frequentistically (e.g. "1 in 10 cases"), may result in more accurate risk perception and better-founded decisions compared with risks presented in percentages [23]. When frequencies are used, rates (i.e. X in 100) are easier to understand than proportions (i.e. 1 in X) when comparing risks [24]. In sum, both numerical and verbal formats have their pros and cons (see also Lipkus [25]). Several recommendations have been made directed at the use of graphical displays in risk communication [25,26], which may provide helpful support, particularly for persons with low cognitive capacity such as low levels of numeracy [26,27]. Baty et al., for example, reported that graphical displays enabled women to quickly understand their risks of getting breast cancer [28]. However, others have shown that risks presented as icons or pictographs (e.g. population figures in 10 rows of 10) did not result in a better understanding, while having a higher affective impact, and were perceived as larger compared to numerical formats [18,29].

Furthermore, the time horizon in which risks are presented is relevant. Time horizons may raise confusion, especially for long-term cumulative risks such as lifetime breast cancer risks [30]. Short-term risks such as risks for the next ten years may therefore be more meaningful and immediate [30], and may result in less overestimation. Research showed that for the presentation of breast cancer risks, women preferred risk information framed over multiple time horizons (10-year-, 20-year- and lifetime horizons) to risk information framed at one time point [31]. Until now, most studies used hypothetical scenarios in exploring patients' preference for the manner in which risk information is given. However, it is important to discover the best format for communicating cancer risk information in practice tailored to individual counsellees [26].

Several dimensions or outcomes in evaluating the effectiveness of the risk communication process have been distinguished [25,32,33]. Among these dimensions are, firstly, a cognitive evaluation, which refers to risk perception, and understanding and recalling the information provided. Acquisition of knowledge in the context of familial breast cancer comprises of understanding the nature of (familial) breast cancer, knowledge of personal risk factors, understanding the probability of developing breast cancer when having a genetic predisposition, and understanding what actions can be taken to reduce the risk. A second dimension is psychological well-being, including affective reactions and concern, for example, anxiety, (breast cancer) worries, and intrusive thoughts. A third dimension is the decision-making process including a behavioural evaluation, for example with regard to adherence to recommended behaviours (e.g. monthly breast self-examinations, mammography screening) and other decisions such as informing relatives who may be at risk. Finally, another dimension is how counsellees evaluate the information that is provided, which includes the extent to which the risk information is regarded as helpful and clear by counsellees.

Currently, protocols in genetic counselling for breast cancer contain no guidelines for the optimal format of risk communication. Therefore, counsellors present risks in ways that seem right to them despite a lack of evidence to support this [13]. Observational research shows that in breast cancer counselling, risks are given at least half of the time in verbal terms [34,35], and in 25% of the consultations in percentages [36].

In the Netherlands, current information booklets on hereditary breast cancer to support genetic counselling generally present risks, e.g. chance of developing breast cancer, as absolute percentages with a long-term time horizon ('lifetime risk'). In genetic counselling, generally the same formats are used, although one study recently revealed that a time horizon is not often stated in breast cancer counselling [35]. In addition, age-related risks regarding breast cancer counselling are usually not communicated, although in other countries, such as the UK, age-related risks are more commonly used [19].

The BRISC study is a field study; in other words, a clinical trial which offers a unique opportunity to evaluate effects of different formats of communicating risks to counsellees facing decisions such as informing relatives who may also be at risk, whether and when to start mammography screening, and, assuming eligibility, whether to test for mutations in BRCA1/2.

Quantitative outcome measures are collected using questionnaires at four points in time. The follow-up questionnaire is sent 6 months after the intervention, thereby enabling medium-term effects to be detected. With this comprehensive collected data, it will be possible to clarify the various effects of different formats on four dimensions evaluating the effectiveness of the risk communication process. It is not inconceivable to expect that some formats have positive effects on one of the dimensions and negative effects on another dimension. Additionally, the outcomes of this project will give a better understanding of who, for example, as a function of cognitive capacity, might or might not benefit from different formats.

Two control conditions are introduced in the BRISC study: condition 5 in which the risk information is communicated in percentages as is commonly done in usual care, and a second control condition (condition 6) in which no intervention is given. With these two control conditions, it is possible to distinguish between the effect of repetition and the effect of giving the risks in a structured way on the one hand, and between the effects of giving the risks in a structured way in different formats on the other.

Because the genetic counsellor involved has no role in the inclusion of participants, selection bias is restricted, other than from self-selection. Cluster-randomisation is applied as it is too complex to randomise at the level of counsellees because in that case a risk counsellor is asked to give different risk formats of genetic risks to different counselees on the same day. An exception is made for participants who are close family members.

The risk counsellors who give the intervention are not involved in usual care. This will overcome concerns of contamination bias. All risk counsellors are women, eliminating bias due to gender differences. The Latin Square design and the control group, which is constructed in addition to all condition rounds, both serve to eliminate bias due to time. To achieve uniform risk communication in the genetic counselling sessions, several precautions have been taken (e.g. striving for a standard counselling procedure, "counselling checklist", "checklist after standard counselling"), minimising confounding.

Some limitations of the BRISC study need to be addressed. This study aims at improving women's understanding of the risks, and we acknowledge that there are factors other than risk that can influence the decision-making process. In addition, although several precautions have been taken to achieve an optimal standardisation in counselling and procedures, interactive features must be acknowledged. Another limitation of this study is that it is not possible to blind the genetic counsellors completely because of the nature of the intervention. This applies to the intervention allocation and to the age-related formats due to anticipation of questions by the counsellee. Participants are informed that the BRISC study concerned different kinds of risk counselling and are blinded. Furthermore, although we would have preferred to test an integrated intervention in standard counselling sessions, it is not considered feasible, either ethically or practically, to have the intervention done in a structured way by the genetic counsellors themselves. If one of the formats proves to be most effective, this format could be integrated into the standard counselling session. Since members of the research team provided the additional consultations (interventions), implementation of a format in daily practice will need to receive further attention.