Table 1 Summary of content from stakeholder meetings regarding development of personalized decision aids to guide genomics-informed treatment decisions for radiation therapy after prostatectomy
Stakeholder Group | Thematic Category | |||
|---|---|---|---|---|
Empathy | Privacy Concerns | Formatting | Location/Access | |
Patients (n = 5) | - This is a time of high anxiety, so it is hard to absorb data. A few bullet points to help address anxiety important. - There is appreciation that patient has to make decision. This helps guide discussion. - This helps frame decision. There are some slow deciders, and this helps them be better at managing anxiety. - From genomics standpoint, there is always fear regarding genetics and insurance companies—need to address how this differs from genetics. - This program makes the genomic test more human. - For mortality risk input, be sure not to scare patients. Consider putting behind the scenes. Just focus on prostate cancer for patients, hide the rest. | - For websites, concerns about privacy. Anonymizing the process would help. - Offering a CD or thumb drive might help with feeling of privacy. - I have no concerns about genomics and privacy, since this is different than genetic risks. | - Need summary page for main points, plus option of seeing more details. - Recommend showing “spread disease” as an image to teach what metastasis is. - Average value comparison (vs. individualized results) not important, doesn’t belong in main document for patients. - Life expectancy predictions should include more elements than just age. - Be sure to spell out acronyms. - Graphs and numbers both important, because patients decide differently. - Suggest consider showing QALYs vertically and include width representation of percentage of cases where this recommendation is preferred by the model. So, the overall area of the box shows how much “better” the option is. - Like the idea of using graphs to present personalized outcome. - Need to make this interactive with ability to change inputs/assumptions. Want to be able to make thoughtful adjustments to personalized inputs. - Utilities for prostate cancer outcomes may not be the same for surgery patients as for all patients. - Would like ability to change variables as much as possible in program to see the effects of each number. - Recommend combining charts together as much as possible—maybe include the average results bar overlaid within same graph. - Area chart looks good, or consider stacked bar graph since that is height only - This program helps to show factors that affect decision; more info better always - Make the distinction between human genome and cancer genome. - Consider adding exercise and lifestyle factors as option to influence mortality estimates, as this could be another opportunity to influence lifestyle choices | - Information should be available in office for review with physicians. -This should be available WHEREVER THE PATIENT WANTS IT. - Be sure to categorize and thematically link all information so it is easy to understand. - I like to be pre-informed, so would want this tool first prior to urology appointment. Then, can bring questions/prepare. - Really need to do this early before the meeting with doctor. This is helpful, but would be overwhelming without time to prepare. - Would want MD to present this tool, but then give some space to consider the information. |
Urologists (n = 5) | - May be of interest to some patients; this will reassure some anxious patients. | None. | - Patients like numbers and figures, so this will help to facilitate discussion. - Urinary, fecal and sexual side effects important for patients, so should try to include estimated risks. - Really have to explain QALYs and life years well. - Need to communication rate of developing metastatic disease, and that that is linked to receiving ADT. - Data incorporated is similar to nomograms in use in clinic, so there is precedent. - Simple graphical display of survival and complications curves best. - Data regarding impact of predictive data on treatment outcomes are needed to best inform the model. - The most important drivers of clinical decision are data regarding ART versus very early SRT with ultrasensitive PSA, and there are limited such data to drive the model for this decision aid. - Genomic information plus risks of therapy are the most important drivers for the clinical decision | - This data should be delivered in clinic with doctor. - Potential stumbling blocks for this comes down to execution and usability/access of the program. - Genomics seems more like black box to many, compared to other clinical factors, so patients will depend on urologists for information. - This provides a good framework for the clinical decision. - Recommend focusing on iPad/iPhone format for access and portability. - Reality is that most community docs refer to larger center, where this would be discussed. - Most patients (80%) would like this. - This is most useful in intermediate risk patients. - This should be delivered 1:1 with patients. This may fit well in multidisciplinary clinic where it will help build consensus over time. |
Radiation Oncologists (n = 5) | None. | None. | - This decision aid will have value in clinic. For physicians and patients alike, a web based tool would be attractive similar to the MSKCC nomogram, or Adjuvant! Online sites. It will be important to cater this to patients rather than physicians, since physicians will have stronger biases, I think. - This needs to simplified, as even QALYs may be too complex. People might rather see the difference in recurrence risk and % complications, maybe, shown on a time line/graph? This will depict the matter of QALYs perhaps. - Calculations of comorbidity will be important for the model since the results are so dependent on life expectancy, this could be a difficult part of having success with the model. - Need to explain QALYs and utilities to patients, will be challenging. - Suggest plotting life expectancy and risks of side effects. - For genomics test, the field is kind of split regarding whether we should consider these genomics tests to be valid at this point. - Showing the base case is confusing to patients, don’t show the average. - This is reminiscent of Talcott’s work, since disease recurrence prevention has QoL benefits. - QALYs will be a foreign concept to patients, should also show clinical outcomes on graphs. - Simple graphs are best for communication, no need to compare to average. - Think about adding Charlson score to estimating mortality from other causes. - Think about comparing rates of death from prostate cancer to death from other causes. | - I am excited about the Decipher test too but is it ready to be incorporated as a required tool in the model? It has not been proven prospectively to be useful in the adjuvant/salvage decision. - Since it takes a month to get results, it is not practical to incorporate for a first time consult, face to face discussion. - I think it’s too complex to integrate with integrate with EMR at this point. - This could be very useful clinically, with a dual audience (patients and providers). - This should be available direct to patient. - This is definitely something patients would like. There should be a patient web portal, and then the patient can review it with doctors after studying it. It may help to have clinicians present this to patient first. |
Biomedical Informatics (n = 2) | None. | None. | - The EMR is challenging for collecting and incorporating individual data. - Design: do not need to show comparison of average values to patients, this would be too confusing. - A lot of EMRs are trying to allow add-ons to EPIC. If that is done, this could interact with MyChart. - Self-rated health question could be used to estimate mortality. | - Can pull age and medical comorbidities from EMR. - Need to figure out how report gets entered into EMR. - Genetic counselor could input report into EMR. - The doctor/patient pair in an office is a good way to deliver information. - This needs to be accessible with communication, so need to promote this to patients. - Midlevels (e.g., NPs) in urology could discuss and input data for this. |
Genomic Assay Industry Representatives (n = 5) | - Using a test ID code could be used to de-identify this. - Would need to ensure PHI secure, may be able to use test codes that don’t link to PHI. | - This is useful for patients to see it plotted graphically. Would be nice to give patients control to play around with inputs and numbers. - Both graphs and numbers are important. - Would want to be able to print both graph and numbers. - Really need to keep individual factors simple, and make sure it’s easy to enter the Decipher score. - For a given test score, this show’s what is risk given variety of treatment scenarios. - A random code can be assigned to a patient’s report to input Decipher score in web portal. - Important that model is accurate; need to continually update inputs based on evidence (like MSKCC nomograms). | - It’s important that this be discussed with physician and patient together. - EMR linkage may help control accuracy. May want to make this accessible both patient-controlled and also in EMR. - Need to create an independent website and include disclaimer that this is for research use only (like nomograms). - Can potentially store information in one place and link to research database. - This might be a commercially viable type of system—selling point for businesses to use this; could do per-click or contract with companies. - Offices who order Decipher send letter to patients. Some people request copies directly to patient. - Web-based approach seems important. | |