37th Annual Meeting of the Society for Medical Decision Making

October 18 - 21, 2015

Purpose: Colorectal cancer (CRC) screening is recommended for persons aged 50-75 who are at average risk of CRC. However, compliance with this recommendation has been historically low. In light of the introduction of a new stool-based CRC screening test, this study sought to understand the impact of educating patients on CRC screening test options on self-reported intent-to-screen within the next year.

Method: A web-based survey was developed for persons aged 50-75, who were at average risk of CRC, but who have not yet undergone screening. Respondents were asked a series of questions to gauge their perspectives on CRC screening, their knowledge of CRC screening tests, and, using a 5-point Likert-type scale, their baseline intent-to-screen in the next year (1=definitely not, 3=maybe, 5=definitely). Respondents were then introduced to the profiles of 5 available CRC screening tests: colonoscopy, flexible sigmoidoscopy (FS), fecal occult blood testing (FOBT), fecal immunochemical testing (FIT), and stool DNA testing. Each profile contained a brief description about the test, and information on dietary preparation, time requirements, physical discomfort, complication risk, frequency of testing, accuracy, and follow up. After reviewing the test profiles, respondents were asked another series of questions to elicit their preferred CRC screening option and the test attributes influencing their choice. They were then asked again, on the same Likert-type scale, their intent to undergo CRC screening in the next year.

Result: To date, 415 eligible persons have completed the survey, representing a diverse population in terms of gender, race/ethnicity, education, and income. Prior to reviewing the test profiles, the majority of respondents were unaware of CRC screening options other than colonoscopy. Following their introduction to various CRC screening tests, self-reported intent-to-screen increased significantly (p<0.05), using a one-tailed, paired t-test, from a mean (SD) of 2.89 (SD=1.20) to 3.61 (SD=1.19). 67% of patients indicated they had never discussed CRC screening with their physicians in the past; of the 33% of patients who had been counseled on CRC screening, most were recommended only colonoscopy.

Conclusion: Awareness of CRC screening options other than colonoscopy may increase patients’ intentions to undergo screening in the next year. These findings affirm the need to educate patients on the importance of CRC screening and to provide alternate test options to patients unwilling or unable to undergo colonoscopy.

Purpose: Whilst total knee arthroplasty (TKA) is the most common joint replacement surgery in Canada, published research indicates dissatisfaction post-surgery in up to 20% of patients. This statistic is troubling, especially as the underlying problems and solutions remain unclear, but also offers potential for service improvement and efficiency gains. Our research team is currently investigating this question through a longitudinal mixed methods cohort study: Patient Experience of Arthroplasty of the Knee (PEAK).

    Method: Our project explores patient satisfaction, experience and outcomes quantitatively (using survey data collected at 6 and 12 months, and administrative data) and qualitatively (interviews conducted at 7 and 13 months post-surgery). A cohort of 515 patients has been established­—57 of which were purposefully sampled for the qualitative portion—with recruitment from all regions of British Columbia. A highly engaged patient cohort has been achieved, evidenced through very high response rates to our postal surveys (91% at 6 months, 88% at 12 months). To explain variation in survey-reported satisfaction, we used bivariate and multivariate ordered logistic regression using two-level (patient and health region) random intercept proportional odds models. The mixed methods frame for this project resulted from a team commitment to interdisciplinarity. Quantitative survey data have been used to inform sampling for the qualitative component, and qualitative data were used to support the quantitative analysis and interpretation. All team members are involved in regular qualitative and quantitative data discussions.

    Result: Our survey data indicate a dissatisfaction rate of approximately 15% at both 6 and 12 months. Key drivers of variation in survey-reported dissatisfaction include: pre-surgery patient expectations and mental health (particularly depression); and post-surgery health outcomes, most notably pain and functional limitations (e.g., stiffness, mobility, usual activities, etc.). The qualitative data are supportive and complementary to our quantitative findings, indicating the importance of personal and clinical support, particularly post-surgery. In addition, patients reported dissatisfaction with not being sufficiently forewarned about post-surgical pain and having insufficient interaction with their surgeon and the health care system post-surgery.

    Conclusion: These results indicate where the TKA process and the health care system might be able to provide better patient-centered care.  Areas highlighted include patient selection, and post-operative care and support, particularly challenging the boundaries of where the health care system ends its relationship with the patient.

Purpose: Current treatments for refractory overactive bladder (OAB) differ considerably: sacral neuromodulation (SNM) involves implanting a device in the buttock; onabotulinumtoxinA involves injections into the bladder; and percutaneous tibial nerve stimulation (PTNS) involves inserting a needle into the ankle. We sought to use best-worst scaling (BWS) to assess patient preferences for differentiating attributes of these treatments and compare direct preference elicitation with the BWS scores versus traditional Likert responses. 

Methods: Qualitative interviews with OAB patients and clinicians informed the development of an online survey incorporating several stated-preference methods. Best-worst scaling (Case 1) was used to assess 13 treatment features. Across 13 tasks presenting subsets of attributes, respondents identified the best and worst. A score ranging from -1.0 (worst) to 1.0 (best) were calculated based on the rates each attribute was chosen. Attitudes toward the attributes were assessed via like/dislike Likert scales, and patients were asked their percentage likelihood (0-100%) of trying each treatment, based on standardized treatment descriptions.    

Results: 245 OAB patients (118 US, 127 UK) completed the survey (79% female; mean age of 50 + 7.8.  ‘Lasting improvement’ (0.82), and ‘minimal side effects’ (0.67)’ were rated most favorably, and ‘implant complications’ (-0.65), and ‘Be willing to self-catheterize’ (-0.53) were rated worst. The percentage likelihood estimates for trying one of the three treatments were significantly correlated with the BWS scores.  Specifically, the likelihood of trying SNM was correlated with ‘implanted device’ and ‘sends signals’, and negatively correlated with ‘repeated visits’, ‘needle in ankle’, and ‘minimal side effects’.  The likelihood of trying onabotulinumtoxinA was correlated with ‘Botox (botulinum toxin) treatment’, ‘self-catheterize’, ‘treatment via urethra’, and ‘minor procedure’, and was negatively correlated with ‘needle in ankle’, ‘implant complications’, ‘repeated visits’, and ‘implanted device’.   The likelihood of trying PTNS was correlated with ‘needle in ankle’ and ‘sends signals’, and was negatively correlated with ‘minor procedure’ and ‘Botox (botulinum toxin) treatment’.  In contrast, all the attribute like/dislike Likert scores only were positively correlated with willingness to try treatment, thus disliking attributes was not associated with willingness to try an alternative treatment. 

Conclusions: BWS was successful in assessing the magnitude of patient preferences for attributes associated with different refractory OAB therapies.  Compared to Likert items, BWS may be more sensitive in capturing both positive and negative attributes driving treatment selection.

Purpose: More than 4 million patients in the United States are newly evaluated each year for ischemic heart disease (IHD), and both policymakers and researchers have raised concerns about high variability in patterns of subsequent medical management and referral to follow-up cardiovascular tests and procedures, with reports of both under-treatment and over-treatment. Little is known about how patients’ beliefs and preferences influence subsequent clinical decisions, but a better understanding of this relationship is needed to effectively improve medical management and reduce unnecessary tests and procedures.  

Method: We surveyed 280 randomly selected patients without prior diagnosis of IHD who were newly evaluated with stress echocardiography, stress myocardial perfusion imaging (MPI), or cardiac computed tomography angiography (CCTA) between November 1, 2013 and February 28, 2015 within Geisinger Health System. We assessed how important patients felt their initial test was to their health, whether they believed their initial test result was abnormal, and how likely they were to complete any recommended follow-up. Using electronic health records, we constructed logistic regression models of follow-up testing and procedures within 90 days while adjusting for anginal symptoms, initial test results, and other clinical and sociodemographic characteristics. Analyses accounted for survey design and nonresponse.

Result: Of 280 patients (mean age=59 years), 46% were men, 16% had diabetes, and 36% reported experiencing angina. The initial stress test or CCTA was abnormal in 8% and subsequent tests or procedures were performed on 11% (6% after normal test, 64% after abnormal test). Most patients felt their initial test was important (70%), had an accurate understanding of their test result (80%), and had strong preferences for completing recommended follow-up (64%). While cardiac risk factors and a test abnormality were associated with subsequent testing, patients’ beliefs about test importance (OR 0.9 [95% CI, 0.4–1.7]), understanding of initial results (OR 1.2 [95% CI, 0.6–2.7]), and preferences for follow-up (OR 0.9 [95% CI, 0.5–1.7]) were not.

Conclusion: Among patients with suspected ischemic heart disease, variation in subsequent testing and procedures cannot be explained by patient beliefs or preferences. Variation that leads to over-treatment or under-treatment may be better addressed by focusing on physician factors.

Purpose: Although improvements in childhood cancer treatments have resulted in remarkable survival rate increases, survivors of childhood cancer face very high risks for serious chronic illness and premature death from late-effects in adulthood, including secondary cancers and cardiac disease. To date, utility weights are lacking for this unique, complex population, limiting comparative effectiveness modeling. We sought to understand the impact of late-effects on quality of life in childhood cancer survivors using health utility scores.

Methods: We calculated SF-6D health utility scores for childhood cancer survivors using SF-36v1 data (n=7105) from the NCI-sponsored Childhood Cancer Survivor Study (CCSS), a multi-institutional study of 5-year survivors of childhood and adolescent cancer, and the general population using SF-12v2 data  (n=12,803) from the Medical Expenditures Panel Survey (MEPS). We calculated SF-6D scores for the overall cohort (age 18-49) and for sex- and age-strata (ages 18-29, 30-39, 40-49). We also compared SF-6D scores among survivor subgroups (e.g., based upon original cancer diagnosis, cancer treatment, and number of chronic conditions). We defined a Minimally Important Difference (MID) as a 0.03 point difference in SF-6D score and statistical significance at the P<0.05 level.

Results: Based on CCSS SF-36 data, we found that SF-6D scores for survivors were statistically lower than MEPS general population estimates (males, 0.787 (SD, 0.118) vs. 0.827 (SD, 0.170); females, 0.751 (SD, 0.124) vs. 0.790 (SD, 0.173)). This was consistent across age-stratum, with lower SF-6D scores for older ages (e.g., males age 40-49, 0.772 (SD, 0.122) vs. 0.807 (SD, 0.152); females age 40-49, 0.735 (SD, 0.130) vs. 0.776 (SD, 0.151). Within CCSS responders, SF-6D score differences did not reach MID when comparing across original cancer diagnosis groups, age at diagnosis, and treatment subgroups. SF-6D scores were lower in survivors who reported chronic conditions; when compared to those who reported no conditions (0.81 (SD, 0.107), SF-6D scores were MID and statistically significantly lower in survivors who reported 2 (0.773 (SD, 0.118) or ≥3 conditions (0.735 (SD, 0.128), regardless of severity grade. Multivariate linear regression models found that age, sex, and most chronic conditions were associated with statistically significant SF-6D score decrements (p<0.03).

Conclusions: Health utility weights for childhood cancer survivors are consistently lower than for the general population, largely attributable to the multiple chronic conditions that develop after initial cancer cure.

Purpose: Disease-specific instruments are frequently used in HIV quality of life research; however, scores generated from these instruments are not suited for cost-effectiveness analyses as they do not assign utility values to health states. Our objective was to estimate, validate, and compare two mapping algorithms to predict HUI3 and EQ-5D health state values from a disease specific instrument based on the Medical Outcomes Study (MOS-HIV) in patients with HIV.

Method: We developed and validated mapping algorithms using data from two separate HIV clinical trials. We divided data from the first trial into estimation (n=294 patients) and internal validation (n=73) datasets; data from the second trial formed the external validation dataset (n=168). We compared ordinary least squares (OLS) with the more flexible beta regression method; the 10 MOS-HIV domain scores served as predictor variables. We assessed model performance using mean absolute error (MAE) and root mean square error (RMSE).

Result: Both the OLS and beta regression models accurately predicted the mean HUI3 and EQ-5D scores in the external validation sample. The mean observed HUI3 score was 0.837, while the predicted score from the OLS model was 0.817; the mean observed EQ-5D score was 0.897, while the OLS-predicted score was 0.883. Model fit, in terms of mean absolute error values in the external validation sample, ranged from 0.068 to 0.104. Both the OLS and beta regression models predicted HUI3 and EQ-5D values that were too high for patients in poor health. For the sickest tertile, the mean observed HUI3 was 0.13; both the OLS and beta regression models predicted a mean of 0.38.

Conclusion: The proposed mapping algorithms can be used to predict HUI3 and EQ-5D health state values from the MOS-HIV, with the caveat that overprediction may pose a problem in samples where a substantial proportion of patients are in poor health. These algorithms may be useful for estimating health state values for cost-effectiveness studies when HUI3 or EQ-5D data are not available.