Evaluation of a Decision Aid for Incidental Genomic Findings

BACKGROUND: Health care providers are increasingly using GS to diagnose, prognose and treat diseases. GS offers increased sensitivity over classic genetic tests, decreasing time-consuming and costly diagnostic cascades. However, GS may also incidentally reveal inherited risks for many other cancers and diseases. Guidelines recommend doctors inform patients of their incidental GS results. Yet there are limited tools to communicate the scope and implications of the thousands incidental results available to help guide patients' decisions about which results they wish to learn.

Gaps: Decision aids (DAs) are best suited to meet this challenge, but no DA exists to guide patients' decisions about incidental GS results.

Rationale: It is not feasible to counsel patients on the thousands of incidental findings available to make informed choices about which incidental results they wish to receive because of the limited genomics expertise and capacity among oncologists, and the long wait times for genetic counseling. Our DA fills this critical care and translational gap by improving the quality of patients' decisions and saving oncologists time counseling patients on incidental findings.

Preliminary data: 1) DA development: We created an interactive online DA. It begins with a professional whiteboard video (by Dr. Mike Evans) that conveys the key concepts, risks and benefits of learning about incidental GS results to educate patients. It then prepares patients for decision-making using a values clarification exercise (with feedback of their preferences) and a knowledge questionnaire (with correct answers provided after). It ends by asking participants to select result categories they want to learn using a menu tool. 2) Usability testing: We also evaluated the DA's usability with 15 patients in 2 rounds. Interviews demonstrated strong face validity and content comprehension. Most patients found the amount of information 'just right' (11/15), clear (12/15) and balanced (14/15). All patients felt that the information was sufficient to reach a decision, that the DA was easy to use and would recommend it.

OBJECTIVES

1. Evaluate the efficacy of the DA compared to standard genetic counseling (GC)
2. Understand the decision-making patients' use regarding GS and selecting incidental findings.

METHODS

Phase 1 - RCT to evaluate the DA:

Methods: We will evaluate the efficacy of the DA in reducing decisional conflict compared to standard genetic counseling (GC) using a superiority trial.

Population: We will recruit adult cancer patients who are eligible to have GS (i.e., tested negative for the classic gene mutation associated with their cancer - e.g., BRCA1/2, MLH, MSH, PMS, APC, MUTYH) from genetics clinics at Mount Sinai Hospital, Princess Margaret Hospital and Sunnybrook Hospital in Toronto, ON Canada. We will include adults who speak and read English and exclude patients with metastatic/recurrent disease as incidental results are less consequential to this population.

Sample size: TThe primary outcome is decisional conflict; the study requires 64 patients/arm to detect the minimal clinically important difference (MCID) of 0.3 using the Decisional Conflict Scale (DCS) (Appendix 3), assuming a standard deviation of 0.6, an alpha of 0.05 (two-sided) and power of 0.815,16. In the last 3 months, 244 patients with a family history of breast and colon cancer tested negative for their associated classic mutations (BRCA1/2, MLH, MSH, PMS, APC, MUTYH) most of who would be eligible for GS. Extrapolating this over the next 9 months we estimate that there would be 732 eligible patients. It is highly feasible to reach our target of 128 patients.

Participants will be consecutively randomized and allocated from an existing list of eligible subjects using a computer-generated randomization in a 1:1 ratio with random permuted blocks of varying sizes. Patients from each clinic will be randomized separately to ensure we have an even distribution of this population in both arms of the study.

Intervention arm: Participants will view the online DA and then complete the online self-administered measures (below) in one sitting within 14 days of recruitment. Next, they will speak with a GC over the telephone after the DA, using a standardized script. They will then complete the same online measures again after speaking to the GC.

Control arm: The GC will conduct the GC session over the telephone within 14 days of recruitment. A topics script will be used to standardize GC discussions covering standard educational content to enable patients to select incidental GS results (participants will not view the DA nor the video). Participants will complete the online self-administered measures after speaking with the GC.

Outcome: Consistent with the Ottawa Decisional Support Framework, our primary outcome is decisional conflict. Secondary outcomes are: knowledge of GS, satisfaction with decision, preparation for decision-making and anxiety.

Measures: We will use validated scales to assess decisional conflict, knowledge, anxiety, satisfaction with the decision and preparation for decision-making. We will develop a standardized topic script for the GC in each arm, as well as a questionnaire to collect intervention fidelity (e.g., usage statistics, duration of counseling sessions), demographic and clinical characteristics (e.g., cancer status and genetic testing).

Analysis: Consistent with the Ottawa Decision Support Framework, our primary outcome is decisional conflict, assessed via the validated Decisional Conflict Scale (DCS). Knowledge is the secondary outcome, will be measured by the Cliseq genomic sequencing questionnaire and a set of internal developed knowledge questions. Satisfaction and anxiety with also be assessed. Satisfaction will be measured using the Satisfaction with Decision scale (SWD) and the Preparation for Decision Making scale (PrepDM). Anxiety will be measured using the state subscale of the State-Trait Anxiety Inventory (STAI). We will also include a demographics and cancer history questionnaire.

The analysis of outcomes will follow the intention-to-treat (ITT) approach. Mean DCS, SWD, PrepDM and STAI scores will be compared using a t-test. Knowledge scores will be assessed by summing the number of correct responses to the questions, and compared using t-tests. Linear regression will be used in a secondary analysis to account for known predictors for decisional outcomes such as education. Secondary analyses will compare the mean DCS, knowledge, SWD, PrepDM and STAI scores before and after GC in the intervention arm to explore the additional benefit of GC after the DA. Un/adjusted mean differences and 95% confidence intervals will be reported. We will use descriptive statistics to report participants' characteristics.

Phase 2 - qualitative study of decision making for incidental results:

A subset of study participants will be asked to take part in a qualitative interview about their decision-making regarding selecting incidental findings. These semi-structured interviews will take place over the phone with a total of 40 participants. For the qualitative component a purposeful sample of study participants will be used. We will target a mix of participants across ages, cancer type and stage, gender, and study group to assess the varying approaches to decision-making. At total of 40 participants will take part in the qualitative component.

Analysis: Qualitative data analysis will draw on grounded theory methodology. We will sort the data by searching for themes/patterns and variations within and across interviews using HypeRESEARCH. Coding, which is the first stage in the analysis process, will involve 'labeling' the data with descriptive codes. Two team members will independently code each transcript. Consensus on coding will be reached through comparison and discussion among these members. The second stage will involve constant comparison, where codes and their content will be compared across interviews to discern common and divergent themes and issues across them. The final stage is selective coding, which integrates all the codes under a central phenomenon to build a theory. Validation methods include triangulation and member checking.