The 36th Annual Meeting of the Society for Medical Decision Making

October 18 - 22, 2014

Purpose:

Decision aids (DA) have been shown to increase patients’ knowledge and involvement in treatment decisions, and reduce decisional conflict in randomized controlled trials. However, less is known about the impact of these tools in routine care. The aim of this study was to assess the effectiveness of a prostate cancer screening DA when used in primary care.

Method:

All providers in the 18 primary care practices affiliated with Massachusetts General Hospital are able to order DAs through the electronic medical record. The DA is then mailed to the patient. A questionnaire was sent with all prostate cancer screening (PSA) DAs that asked about knowledge, goals, and screening intentions before and after watching the program.

Result:

From March 2010-November 2013, 2,588 PSA DAs were mailed and 616 completed questionnaires (response rate 23.8%) were received. The respondents were 83.4% White, mean age 58.4 years old (SD 8.1 years), and 55.3% reported a college education or more. Most respondents watched all or most of the program (75.8%), slightly fewer read all or most of the booklet (60.4%).

Respondents’ mean knowledge score was 89.4%, (SD 21.7%). A majority understood that most men with prostate cancer die of something other than prostate cancer (84.4%). Most respondents found the DA very or extremely useful in helping them understand what a PSA test is (77.5%), deciding whether or not to have a PSA test (78.2%), and preparing to talk with their provider about the PSA test (71.3%).

After viewing the DA, fewer men remained unsure about testing (33.6% pre-viewing vs. 17% post viewing, p≤.001) and more men did not intend to have a PSA test (32% pre-viewing vs. 51% post viewing, p<0.001). Men who were leaning toward having a PSA test felt it was more important to find cancer early (8.4 vs. 5.0 out of 10, p<0.001) and to have peace of mind from a negative result (8.2 vs. 4.9 out of 10, p<0.001) compared to men who were leaning toward not having PSA test.

Conclusion:

The use of the DA in routine care helped men clarify preferences about PSA test, be well informed, and prepare to discuss PSA screening with primary care providers. These results provide evidence of the effectiveness of DAs in supporting shared decision making in clinical practice.

Purpose: As mental health practice continues to anchor itself in patient centered methods of care, it is increasing valuable to understand patients’ preferences and perceptions of the treatments they engage in. Recent studies suggest that patients’ use of mental health medication is often intermittent, and in parallel, their use of alternative forms of medicine is rising.  However little data is available about the extent to which these topics are included in traditional medical dialogues. MyPSYCKES is a web-based shared decision making program that patients use before their appointment. We investigated patients’ answers to MyPSYCKES questions about medication concerns and medication alternatives in particular. 

Method: We implemented MyPSYCKES in eight diverse mental health clinics in New York State and extracted response data for each patient’s first time use of the program. We then investigated patterns of concern around patients who endorsed an interest in exploring alternative therapies or strategies in comparison with other medication concerns. 

Result: A total of 963 patients used the MyPSYCKES application. 28.45% (N=274) of patients reported that they were exploring alternative strategies.  Among those, 30.29% explicitly cited exploring natural or cultural healing methods, 21.90% cited an interest in stopping their medication, and 10.95% cited exploring only taking medication when they are having a difficult time. Among this subset of patients (N=274), 74.82% also answered that they had concerns about how helpful their medications were. 65.33% had concerns about side effects, and 66.42% were concerned about how their medications were affecting their health. Additionally, 43.80% of patients cited concerning fears, 38.32% of patients cited concerning beliefs, and 44.16% cited trouble finding motivation to take medications in addition to their exploration of alternatives. 

Conclusion: 28.45% of patients stated that they were exploring alternative strategies to taking their medication. The large majority of these patients were also concerned about the effectiveness of their medications, which validates previous findings that predict use of alternative therapies, and were also twice as likely to endorse fears (43.80% vs. 21.60%), conflicting beliefs (38.32% vs. 15.37%), or trouble finding motivation (44.16% vs. 22.64%) as concerns that impacted their use of mental health medication. In order to support strong therapeutic alliances and patient centered care, it will be important for practitioners to understand patients’ reasoning and preferences for exploring alternatives.

Purpose: People typically overestimate the unpleasantness of medical experiences and may avoid important screenings (Dillard et al., 2010) or medical procedures with long-term health benefits (Angott et al., 2013). The purpose of this research is to determine whether experience narratives can reduce these mispredictions or ‘affective forecasting errors.’

Method: In Study 1 (N=366), participants were surveyed about their experiences with 10 common medical events (e.g. Pap test, donating blood). Those who had never experienced the event provided ratings of predicted discomfort and those who had experienced the event provided ratings of actual discomfort. Participants making predictions were randomly assigned to either the control condition (no narratives) or the narrative condition (4 narratives describing experiences with the medical event provided by participants in a pilot test) before making predictions. To evaluate affective forecasting errors, we compared predicted experiences with actual experiences for each medical event in a between-subjects comparison.

In Study 2, participants (N=203) made predictions about the discomfort associated with the cold pressor task (keeping your hand in ice water 1-2°C for up to 2 minutes). Participants were randomly assigned to one of three conditions: 1) control (no narratives), 2) positive narratives (2 stories describing the task as not painful), or 3) negative narratives (2 stories describing the task as painful). Narratives were selected from an earlier cold pressor study. All participants completed the cold pressor task and then immediately provided ratings of the discomfort experienced. To evaluate affective forecasting errors, we compared participants’ predictions of discomfort to their ratings of the discomfort experienced.

Result: In Study 1, affective forecasting errors were observed for 6 of the 10 medical events; specifically, predicted discomfort was significantly greater than experienced discomfort, p<.05. However, experience narratives did not improve affective forecasting errors and, counterintuitively, narratives made predictions worse for 2 of the medical events, p<.05. In Study 2, affective forecasting errors were also observed, however this time predicted discomfort was significantly less than experienced discomfort, p<.05. Further, the use of negative experience narratives successfully reduced the bias in predictions; see Figure 1. 

Conclusion: Affective forecasting errors can be improved with the use of narratives that emphasize unanticipated elements of the experience. However, narratives must be carefully selected because less focused stories may actually increase bias.

Purpose: The inability to understand and use numerical information in health care decisions has been independently associated with poor patient decision making and worse health outcomes. Research has also found the presentation format can affect patient numerical comprehension. Here, we extend this line of investigation to examine whether decision support tools can be tailored to improve women’s understanding of hereditary breast cancer risks based on women’s informed preferences for presentation format.

Methods: Women at increased risk for breast cancer were recruited through the Cancer Genetics Network. Participants were initially asked for their preferred format of risk communication including: percentages, frequencies, bar graphs, pictograms, and comparison to average risks. They were then presented with a statement about risks associated with BRCA mutations using various risk communication formats and asked to answer a question that required them to interpret the risk information presented. After using the formats, participants were again asked to indicate their preferred format. 

Results: 380 women enrolled and completed the study. Although pictograms and bar graphs are considered to be more effective, almost no participants initially preferred these (Table 1). Instead, 54% preferred percentages, 25% preferred a comparison to others, and 17% preferred frequencies. Presentation format had little influence on the ability to interpret risk information: 72-75% identified the correct interpretations in each format. However, only 53% interpreted all formats correctly, and 27% correctly interpreted less than half of the formats. Once the participants had an opportunity to use each format, 58% changed their preferences (Table 1).  Among those who shifted their preferences to bar graphs or pictograms, the accuracy of interpretation was 65%, compared to 45% among those who preferred these graphs initially.

Conclusion: Women at risk for breast cancer appear to change their preference for receiving risk information once they have had experience with the different formats. Ability to correctly interpret risk information was also related to shift towards graphical representations. Informed preference regarding risk communication format may better predict risk perception abilities for some. While these results need further validation they suggest that interactive approaches to tailoring presentation of risk information may improve risk perception.

Format	Initial Preference	Final preference
Percentages	54%	32%
Frequencies	17%	9%
Bar graphs	2%	41%
Pictograms	0%	9%
Comparison to other women	25%	8%

Purpose:  People with serious mental illness are at high risk for obesity and related medical problems, and die 10 to 20 years prematurely, most commonly from cardiovascular disease.  Current guidelines recommend specialized, in-person weight management interventions, and these result in weight loss in efficacy research.  In clinical practice, these interventions are typically not provided, and, when provided, client enrollment and retention is low.  Barriers could be addressed using computerized provision of diet and exercise education and decision support, combined with motivation and support from peer coaches.

Method:  234 patients with serious mental illness were recruited from a Veterans medical center, and 54 from a county mental health clinic.  We randomized patients to 1) online weight management with peer coaching, 2) in-person clinician-led services, or 3) to continue with treatment as usual. Online weight management included 30 modules plus weekly telephonic peer coaching.  The online system could be accessed from clinic kiosks, or anywhere there is internet access.  It provided simultaneous audio and text-based education, video, pedometer tracking, goal setting and homework, diet plans, and quizzes to ensure learning.  Coaching was delivered by individuals with lived experience with mental illness, was phone-based, and utilized motivational interviewing principles.  In-person weight management included 24 sessions of a weight management intervention, and had the same curriculum as the online program.  At 6 months, patient outcomes were assessed and semi-structured interviews conducted with participants.

Result:  A mixed measures repeated model predicted Body Mass Index at 6 months.  In VA clients, there was a significant group by time interaction (F=3.1, p=.05).  The online and peer coaching group had weight reduction averaging 0.5 BMI points (5.2 pounds, p=.03), while neither treatment as usual (p=.29) nor in-person services (p=.86) had a substantial change.  Groups at the county clinic had similar effects in the same direction.  42% of clients completed the on-line program compared to 0% completing all in-person groups (Chi-Sq=16.4; p<.0001).  On-line services and peer coaching were well received.

Conclusion:  On-line weight management with peer supports is feasible, and well received.  It allows the provision of educational content and decision support that is tailored to individual patients, convenient, and patient-centered.  This produces weight loss, and may have greater effectiveness than clinician-led services.  Marginal costs are low, and this approach is amenable to broad dissemination.