ORAL ABSTRACTS: DECISION AIDS AND DECISION SUPPORT
Michael Pignone, MD, MPH
University of North Carolina at Chapel Hill
Method: The CHOICES DA was developed based on literature, telephone surveys of 1100 Florida cancer patients, and semi-structured interviews with cancer survivors. Following beta testing with 64 minority cancer survivors, a pilot randomized controlled trial (RCT) comparing the CHOICES DA vs. the NIH clinical trials webpage was performed with 104 newly diagnosed cancer patients. We assessed the effect of these two interventions on patient knowledge and preparedness to make a decision about participation in a trial at baseline (pre), immediately after viewing a website (post), and 2 week follow up.
Result: In the RCT, there were no differences in age, gender, education or race/ethnicity of participants in the CHOICES DA (n=48) and NCI website (n=56) groups. Both intervention groups had significant increases from pre to post intervention in measures of self-reported knowledge, decision preparedness, and values clarification related to CCTs. However, there were few differences across intervention arms for these measures. Finally, objectively measured knowledge about CCTs increased significantly in both intervention arms, but did not differ between the arms.
Conclusion: In this study, we found significant increases in knowledge and decision preparedness from pre to post intervention, but few differences between the two arms of the intervention. In contrast, a RCT of the CHOICES DA and the Washington University Siteman cancer website in St. Louis, MO showed significantly greater improvement in decision preparedness and knowledge among cancer patients randomized to the CHOICES DA. Reasons for this disparity in effects should be explored, and the effect of the CHOICES DA on actual participation in CCTs needs to be established. However, the CHOICES DA shows substantial promise for improving patient decision-making about participation in CCT, and thus for enhancing patient engagement and autonomy in decision making.
Effective communication of the uncertainty surrounding the expected benefits and harms of medical interventions is an important aspect of informed decision making. Patient decision support interventions (PDSI) aim to improve patients’ knowledge and yield medical choices more aligned with patients’ preferences. The International Patient Decision Aid Standards (IPDAS) group, which aims to provide criteria for improving decision aids, recommend that uncertainty in evidence be described. This study sought to understand how uncertainty is being described in PDSIs.
We reviewed all uncertainty statements around risks and benefits within all of the PDSIs available in three registries (Ottawa Hospital Research Institute, Choosing Wisely, Option Grid Collaborative). We developed a framework by which we could classify these statements by their presentation of first (the randomness of future events (i.e. the risk)) and second order uncertainty (the imprecision in risk estimates). All statements were extracted and classified by two reviewers (MB, LS), any discrepancies were arbitrated by a third reviewer (MH or NB).
461 decision aids were identified and included in the review. The decision aids we reviewed ranged in the number of options they presented to people (2 to 16), and in the number of harms (0 to 31) and benefits (0 to 16) that they included. The number of harms described outweighed the benefits. We identified 8729 statements as describing an uncertain risk or benefit. The predominant method of presenting both first order and second order uncertainty was qualitative, but the majority of statements did not convey second order uncertainty. When first and second order uncertainty was combined there were many different ways (n=26) of describing uncertainty.
There is considerable heterogeneity in the methods that PDSI developers are using to convey uncertainty around risks and benefits to patients to support shared decision making. This heterogeneity is a product of a lack of evidence about how best to communicate risk and uncertainty to patients. There is an urgent need for more research into the evaluation of methods for communicating uncertainty to patients and its impact on patient preferences.
Purpose: Theory-based models predict superior results from cultural targeting of patient decisions aids (PDAs). However, these models remain untested. Three studies in two disease conditions were conducted to examine effects of culturally targeted versus non-targeted PDAs.
Method: Researchers ran two experiments in breast cancer and diabetes contexts. A third experiment retested the first breast cancer study using a different design. Online panels of self-identified Hispanic women in the U.S. read a disease scenario and imagined a recent diagnosis. They were randomly assigned to a culturally targeted (ethnic imagery, linguistics, values, narratives) or non-targeted PDA prior to consultation. An expert on Hispanic health communication helped develop targeted versions and focus groups of Hispanic women reviewed culturally relevant content. Standard outcome measures associated with PDA evaluation research were collected. Manipulation checks showed high awareness of cultural information only in the targeted PDA group in each study. Respondents spent eight minutes or more on the task and passed attention filters. Thought listings indicated engagement.
Results: Sample demographics were similar in all three studies (e.g., Hispanic women, age 40-55; English capable; modal categories: lower to middle income, some college education, and homemaker). Despite large samples and strong manipulations, only one significant difference in studies 1-2 was detected (see Table). To rule out possible confounds from group-interdependence measures in studies 1-2, study 3 premeasured group-interdependence and a week later, invited participants to take the main part of the breast cancer experiment. Again, non-significant results were obtained. Potential moderators such as group-interdependence, ethnic identification strength, and years in the U.S. did not interact with PDA treatment in any study.
Conclusions: Recent cultural psychology research sheds light on possible reasons for null PDA targeting effects across two disease contexts and three studies. First, culture appears to exert its greatest influence when individuals are “thinking fast” or constrained on resources such as time. Other evidence suggests that culture is most influential when processing objectives are less clear and task-specific schema/norms aren't primed. When individuals deliberate and/or are guided by relatively clear processing goals, individual differences and task-related schema/norms are more likely to dominate “shared” cultural values. Three studies suggest that cultural targeting in serious disease contexts, while not harmful, may not enhance pre-consultation objectives as expected with ethnically adapted PDAs.
Methods: 58 patients followed at Tufts Medical Center Primary Care were randomized in a 2x2 factorial design to financial incentives or decision aid (vs. usual care). The Tufts Diabetes Statin Decision Aid included values clarification, individualized UKPDS stroke and myocardial infarction risk, statin benefit, costs and harms, and patient intended action. Financial incentives for patients consisted of $20 for testing and $40 for goal achievement. Comparisons were made with Chi-square or Fisher’s Exact Test.
Results: Participants (47% women; 17% Blacks; 10% smokers) had a mean±SD age of 58±12 years, systolic blood pressure of 127±14, hemoglobin A1c of 7.7%±1.7, total cholesterol of 183±49 and HDL of 49±21 mg/dL. At baseline, the mean estimated 10-year risk of stroke or myocardial infarction was 25% if not taking a statin (67% of the cohort) and 18% if taking a statin. 10 participants received usual care only (information sheet, no incentives), 15 decision aid only, 16 financial incentives only and 17 both decision aid and financial incentives.
In the calendar year of the intervention, 30 of 32 (94%) decision aid vs. 23 of 26 (88%) usual care patients had testing (p=0.65), and 30 of 33 (91%) financial incentive vs. 24 of 25 (96%) usual care patients had testing (p=0.63). 25 of 32 (78%) decision aid vs. 13 of 26 (50%) usual care patients achieved the outcome measure of LDL<100 (p=0.049), and 22 of 33 (66%) financial incentive vs. 16 of 25 (64%) usual care patients achieved this goal (p=0.95).
In the following calendar year, 56% decision aid vs. 35% usual care patients met both outcomes, i.e., underwent testing and achieved the LDL goal (p=0.17), and 48% financial incentive vs. 44% usual care patients met both outcomes (p=0.94).
Conclusions: In this factorial randomized controlled trial, neither the decision aid nor financial incentives improved cholesterol testing. The decision aid significantly improved LDL goal achievement, but the improvement attenuated and became non-significant in the following year.
Title: DEVELOPMENT OF A PATIENT DECISION AID FOR SYNCOPE IN THE EMERGENCY DEPARTMENT: THE SYNDA TOOL
Probst MA, Hess EP, Breslin M, Sun BC, Richardson LD.
Purpose: We sought to develop a patient decision aid to promote shared decision-making for patients who present to the emergency department (ED) with syncope (i.e. transient loss of consciousness) and have no serious diagnosis identified.
Methods: Using input from thought leaders in the field of syncope and shared decision-making, we created a prototype of a paper-based decision aid to engage patients in the disposition decision (admission vs. discharge) after a negative ED work-up for syncope. We then conducted 1-on-1 interviews with 8 emergency physicians, 5 cardiologists, and 8 ED syncope patients to get detailed feedback on the content and design of the decision aid.
Results: We iteratively modified the decision aid after every interview using detailed feedback from each interviewee. The evidence-based decision aid, named SynDA, is written at an 8th-grade reading level and includes 4 basic sections: 1) Explanation of syncope, 2) Explanation of future risks, 3) Personalized 30-day risk estimate, and 4) Disposition options. The communication of 30-day risk is stated in natural frequency (e.g. 2 patients out of 100) and graphically displayed using a color-coded, 100-man pictogram. At the end of the interview process, both patient and physician participants expressed satisfaction with the clarity and usability of the decision aid.
Conclusion: Using interviews with patients and physicians, we developed an evidence-based, easy-to-use, patient decision aid, named SynDA, to facilitate shared decision-making for ED syncope patients with a negative diagnostic evaluation. We intend to evaluate this decision aid in a single-center randomized controlled trial.
Method: Eligible hip and knee OA patients at 2 hospitals were randomly assigned to receive a short or long DA before their visit with a surgeon. The short DA is a 16 page booklet which includes a knowledge test and values clarification exercise. The long DA is a 40 page booklet and 40 minute DVD which includes patient testimonials and detailed data on options and outcomes. Participants completed a short survey, after receiving the DA and before their visit, assessing treatment preferences, knowledge, DA usage, health literacy, and quality of life (EQ-5D). We compared the two DAs on (1) amount of use, (2) knowledge, and (3) treatment preference using t-tests and Pearson chi-square.
Overall response rate was 73.4%, 25/37 for the long DA arm and 33/42 for the short DA arm (67.6% vs. 78.6%, p=0.32). There were no significant differences in age, gender, joint, or education between groups. The long DA group had higher EQ-5D scores (0.72 vs. 0.61, p=0.016).
The short DA group reported higher rates of reviewing all the DA compared to the long DA group (64.5% short vs. 24.0% long, p=0.003). Patients with lower literacy were as likely to review all of the long and short DAs as those with high literacy (p=0.6). Knowledge scores were not significantly different between arms (55.2% short vs. 48.8% long, p=0.4). In the long DA arm, patients who reviewed all the DA had higher knowledge scores than those who did not (80.0% vs. 38.9%, p=0.004), whereas in the short DA arm, the knowledge scores were similar (62.0% for those who viewed all DA vs. 52.7% those who did not, p=0.27). A similar percentage of each group preferred surgery (54.5% short vs. 52.0% long, p=0.81).
Conclusion: Despite having different content and formats, the two DAs had similar overall effectiveness. Patients were more likely to review the short DA; however, patients who reviewed all the long DA had the highest knowledge scores. Health literacy was not related to use of either DA. A larger study should examine the use and impact of the tools and determine whether patients are more likely to receive preferred treatments based on their DA assignment.