Multiple genetic markers for colorectal cancer (CRC) risk have been identified, and while practical tests for CRC risk are not currently available, such tests could be used in the future by VA primary care clinicians to tailor cancer screening regimens to each individual's personal risk. Patient acceptance of CRC-specific genetic testing initiatives within VA may depend on communication techniques used by clinicians to discuss test results.
We experimentally evaluated methods of communicating CRC-predictive genetic test results to VA patients to identify approaches that (a) optimize comprehension, (b) generate consistent risk perceptions, (c) evoke the greatest trust in the validity of the test and (d) result in guideline-adherent CRC screening intentions. We also sought to identify what properties of a CRC-predictive genetic test and justifications for genetic testing influence VA patient expectations, acceptance of genetic testing, and/or reactions to test results. Lastly, we assessed how VA clinicians feel about communicating genetic test results for CRC.
The initial phase involved 4 iterative surveys of veterans and others who were in two Mid-West VA Primary Care waiting areas to identify which communication factors and test attributes most influence understanding and reactions to genetic test results. Results from these surveys informed the design of a national mailed survey of veterans and VA primary care clinicians.
A total of 4006 participants completed 1 of the 4 clinic-based surveys. The national mailed sample was sent to 2000 VA patients and 1000 VA providers, and we received completed surveys from 767 patients and 383 providers (response rates of 39.4% and 41.4% respectively after adjusting for ineligible and unreachable addresses.
Clinic-based surveys (Surveys 1-4): In Survey 1 genetic test results were presented as a risk percentage (pictograph/icon array), a risk score (scale with risk categories) or a table of genetic mutations. Participants were less worried (p<0.001) and thought they had a lower likelihood of colon cancer (p<0.001) when they saw test results as a risk percentage versus a risk score. In Survey 2 and Survey 3, we focused on effects of family history. We found that participants who were told to imagine they had a family history of colon cancer were not reassured by average genetic test results. In the follow-up study, we found this failure to reassure to be resilient to several debiasing efforts, including reading a fact sheet before viewing test results, getting information about test error rates, or reading explanations of how people with family history might still have average genetic risk. In Survey 4, we varied the format of test results to try to increase their ability to reassure people with average risk results that they can wait 10 years before a repeat colonoscopy. These manipulations had little effect. However, there appeared to be differences in behavior between white and non-white participants even after controlling for education and numeracy. Younger (>50 years old), black participants were less likely to want to wait the recommended 10 years for their repeat colonoscopy compared to white participants (OR = 0.89, 95%CI [0.41, 1.09]).
Mailed surveys: In our mailed patient survey factors included 1). Receiving additional information regarding the typical progression of colon cancer and 2.) Seeing a flyer about altruism (suggesting that if patients get colonoscopies before genetic test result says to, they might be taking an appointment from another veteran who has a greater need.) Veterans' willingness to wait to get a repeat colonoscopy so that higher risk patients could be screened first was influenced by receiving the altruism manipulation (p < 0.01), extra information on colon cancer (p < 0.01) and self-reported racial/ethnic group (white vs all other; p < 0.01). Participants were also asked whether they were OK with being told that their average risk genetic test result meant that they would not be allowed to have a colonoscopy more frequently than every 10 years. Here we found a significant main effect for extra information on colon cancer (p = 0.03) and a significant interaction effect between altruism and extra information on colon cancer (p = 0.02). Specifically, adding extra information made people more willing to wait up to 2 more years (after 5 years) with having to wait 10 years so that higher risk patients could be screened first only when they did not receive the altruism flyer.
In our mailed clinician survey, when presented with a scenario about a patient without any additional information about family history or genetic test results, only 9% of clinicians wanted this patient to return for a repeat colonoscopy in less than 10 years. Yet, after being given a nearly identical scenario in which the patient had a family history of colon cancer but had a recent clean colonoscopy and an average genetic test (and should therefore be screened as a patient with an average risk of colon cancer), nearly half (49%) of clinicians wanted the patient to return for a repeat colonoscopy earlier than the guideline recommended 10 years. In a logistic regression model that included years of experience, practice setting, and the average number of patients seen in a week, none of these factors predicted likelihood of recommending early colonoscopies.
Our overall conclusion is that both patients and clinicians find it difficult to rely upon genetic test results that indicate average risk of cancer when the patient has a family history of disease. This tendency is consistent across people and appears resilient to interventions designed to mitigate it.
Rigorous and prospective research is needed to anticipate barriers to effective communication about genetic tests; for example, disbelief of average genetic risk results by veterans with family history of disease may be a significant barrier to future VA programs that might seek to use genetic screening tests to optimize screening regimens and minimize overuse of screening resources. This project provided some answers to these questions but further research is needed.
- Zikmund-Fisher BJ, Exe NL, Lipkus IM, Roberts JS, Tarini BA, Knight SJ, Saini SD, Schoenfeld P, Tremblay AS, Fagerlin A. Mutations vs. Risk Score: The Format of Genetic Test Results Can Alter Patients' Reactions and Risk Perceptions. Paper presented at: VA HSR&D National Meeting; 2011 Feb 18; Washington, DC.