CLINICAL DECISION MAKING
* Finalists for the Lee B. Lusted Student Prize
Method: Both members of 37 couples completed surveys at 3 time points: prior to a first consultation with a reproductive specialist, one week after the consultation, and one week after receiving results from fertility-related testing. At each time, participants were prompted to identify the fertility treatment approach they most preferred at that time, and then to complete the Decisional Conflict Scale (DCS) considering that preferred approach. We analyzed differences by gender and changes across time in the DCS and its subscales. Scores above 37.5 are associated with delaying decisions; scores below 25 are associated with implementing decisions.
Result: Before the first consult, both men and women had mean decisional conflict scores above 37.5 (48.53 and 39.24, respectively). Men’s scores were highest on the Informed subscale (measures feeling informed about decision alternatives, benefits, and risks), while women’s scores were highest on the Uncertainty subscale (measures uncertainty in choosing options). Decisional conflict declined post-consult (men to 32.68, p<0.001; women to 30.41, p<0.001) and again post-test-results (men to 28.13, p=0.085; women to 22.74, p=0.001). With one exception, men’s scores were higher than women’s on the total scale and all subscales at all time points. Differences between men and women were largest on the Informed subscale (pre-consult: men=60.30, women=46.15, p=0.013; post-consult: men=40.95, women=31.62, p=0.035; post-test-results: men=30.30, women=20.72, p=0.009).
Conclusion: Prior to a first consultation with a reproductive specialist, both men and women experienced high levels of decisional conflict about which reproductive treatment options to pursue, with men reporting higher scores than women, on average. Decisional conflict tended to decline after couples met with the reproductive specialist and again after receiving test-results, but men’s scores did not reach the cutoff associated with implementing decisions. We will continue to track couples’ decisional conflict and decision making over the first year after meeting with a specialist.
Method: This decision improvement intervention focused on a surgical group practice consisting of 50 surgeons representing 9 specialties practicing at an 800-bed community teaching hospital. Over a 5 year period all physicians engaged in a common training experience consisting of individual chart audits, reporting of actual and benchmark performance measures, and group review of individual, division and overall group improvement following semi-annual training sessions. Best practice decision process for patient information collection, recording and sharing, specifically after-hours call documentation and informed consent, were established following an audit of existing physician compliance and a simultaneous review of malpractice claims conducted by the groups liability carrier. Open reporting of audit results was mandated with pre and post changes reviewed at division and group meetings. Training sessions used actual malpractice claims (appropriately redacted) experienced by the surgeons in order to increase salience and included active physician presentation. Pre and post testing was done through chart audits conducted by outside nurse reviewers using written protocols for chart review. On-call communication requests were recorded separately and then matched to chart documentation by date and time.
Result: This intervention produced a 67% reduction in claims frequency over the five year time period, a 74% reduction in average claims paid per year, a 36% reduction in premiums in the last two years of the program, 98% provider satisfaction, improvements in re-audit results of informed consent and after-hours documentation of between 50-75% (depending on individual physician and/or surgical sub-specialty) with sustained results meeting or exceeding benchmarks.
Conclusion: Improved clinical decisions and a reduction in PAEs based on improved recording and transfer of relevant clinical information is achievable, cost-beneficial and sustainable.
Method: MyPSYCKES is a bilingual web-based application developed by the New York State Office of Mental Health (NYS OMH) to strengthen patients' ability to participate in clinical decision-making. Before each medication appointment, patients use the MyPSYCKES program in their clinic with the assistance of peer staff, answering questions about their use of wellness activities and medications, symptoms and status, medication concerns, personalized health risks, and goals for treatment. MyPSYCKES produces a one-page report that synthesizes their answers and aggregates data from previous visits to show trends over time. The patient and physician use this report in session and develop a shared decision about next steps in treatment. After conducting usability testing and developing robust training and implementation protocols with our two pilot sites, we expanded the program to ten additional sites statewide.
Result: Guided by a diverse Stakeholder Advisory Committee that included patients, clinicians, researchers, policymakers, and payers of care, we developed an extensive evaluation protocol for MyPSYCKES. The research involves several methods of data collection to address the questions of interest identified by stakeholders, including the following priority outcomes: 1) patient empowerment, activation and health outcomes, 2) changes in clinical discussions and decision-making processes, 3) greater inclusion of patient preferences in records, 4) program use by different groups, and 5) implementation best practices.
Conclusion: The MyPSYCKES evaluation will provide important information about the usability of such a system for individuals with SMI and other vulnerable subgroups, as well as critical information on the value and impact of patient reported outcomes and shared decision-making on patient activation, engagement in care, and health outcomes. These data will help guide patients, families, providers, payers, and systems of care on investment decisions for future practices methodologies.
Method: We conducted a randomized controlled trial using the Child Health Improvement through Computer Automation (CHICA) system. CHICA is a clinical decision support system implemented in four primary care pediatric clinics in Indianapolis, IN. It provides physicians with reminder prompts which are algorithmically derived and based on: 1) general pediatric healthcare guidelines and 2) responses parents provide about their children on a pre-visit screening form. Physicians place a check-mark in a box next to each prompt to indicate whether they responded to it. We selected eight prompts for randomization. Four of these prompts were yellow-highlighted when presented to physicians in two of the clinics. The other four prompts were yellow-highlighted when presented to physicians in the remaining two clinics. Each of the pairs of clinics served as the other pair's control. Analyses compared physician response rates to the highlighted versus non-highlighted prompts. Additionally, four prompts deemed “high-priority” (e.g., household violence) were yellow-highlighted during the study period at all of the four clinic sites. Physician response rates to these high-priority highlighted prompts were compared to response rates for those same prompts during the year before the study period, when they were not highlighted. Data analyses included binary logistic regression and chi-square analysis, including Bonferroni corrections for multiple comparisons.
Result: There was no difference in physicians’ response rates to highlighted compared to non-highlighted prompts (OR=1.056, p=0.259, ns). Similarly, physicians were no more likely to respond to highlighted high-priority prompts, compared to the previous year when the prompts were not highlighted (χ2=0.067, p=0.796, ns).
Conclusion: We did not find evidence that yellow-highlighting clinical decisions support reminders is an effective strategy to increase physician responsiveness. Alternative possibilities for combating alert-fatigue and increasing responsiveness to prompts should be explored.
Method: Physician-patient pairs were recruited from our outpatient clinics. Patients submitted DNA samples for broad clinical pharmacogenomic testing. Results were provided to physicians at outpatient visits via an electronic genomic prescribing decision-support system. Patient and physician surveys were administered to examine attitudes and perceptions surrounding receipt of results and medication decisions.
Result: Over 980 patients seeing 16 different physicians have enrolled, and since October 2012, >800 unique outpatient clinic visits at which pharmacogenomic results were available were analyzed. Physicians accessed patients’ pharmacogenomic results via the genomic delivery system at 84% of encounters. In total, 221 different medication changes occurred during the evaluable visits, and importantly, physicians reported that 24% of all medication changes were influenced by the available pharmacogenomic results. Patients taking current medications with cautionary or high-risk pharmacogenomic results had a 2.6 times higher odds of their physician changing the medication at their visit than patients with a favorable pharmacogenomic result. When using the decision-support system during medication changes, 61% of the time physicians described pharmacogenomic information as enabling a more informed therapeutic decision; in 39% of these instances physicians said pharmacogenomic information helped to choose between multiple medication options; and for 28% of drug changes physicians said the information reduced the likelihood of an adverse reaction. At 94% of visits after receiving pharmacogenomic decision support, physicians said they would be ‘very likely’ to enroll other patients in the program. Physicians reported significant increase in their awareness about pharmacogenomic information compared to before availability of results (P=0.009). Patients overwhelmingly reported that they believed genetics influence therapeutic decisions: 83% said they ‘agree strongly’ or ‘agree somewhat’ with the idea; and 91% said they believe their physician practices personalized medicine.
Conclusion: Physician-patient pairs in an institutional program examining the clinical impact of delivering pharmacogenomic results with decision-support report robust utilization and relevance to prescribing decision-making. Dissemination to a greater number of physician-patient pairs to examine the impact of prescribing behavior changes on downstream outcomes like medication adherence, adverse events, and drug efficacies, is warranted.
The purpose of this study was to pilot test perspective taking instructions as a potential medical education tool to increase student empathy and reduce cognitive biases in medical decision-making. Perspective taking instructions are an established empathy induction tool in non-medical basic research contexts, in which they have been demonstrated to increase empathic emotions and decrease forms of social cognitive bias in helpers.
Perspective taking instructions were adapted for a clinician-patient encounter to be piloted via web-based software. The procedure was pretested on a sample of physicians and refined in response to their feedback.
Eighteen fourth year medical students participated in the pilot test. Each was randomly assigned to receive imagine-self (n = 7), imagine-other (n = 4), or objective (n = 7) perspective instructions. The students then read a patient case of a middle-aged female patient who described ongoing symptoms of chronic pain and fatigue. Dependent measures included students’ empathic feelings for the patient, and the degrees to which students believed the patient’s condition to be attributable to (a) psychological and (b) physiological causes. Students were also assessed for trait empathy levels.
One-way analysis of variance with planned contrast coefficients replicated prior findings that the imagine-other perspective led to greater levels of empathic concern for the patient than the imagine-self, t(16) = 1.95, p = .03. In addition, both the imagine-other perspective (b = .523, p = .007) and trait perspective taking (b = .521, p = .006) were retained as significant predictors of experienced empathic concern in a stepwise multiple regression analysis that included trait measures of empathy, F(2, 17) = 10.87, p = .001, R2 = .592, R2 = .592, adjusted R2 = .54.
Also consistent with prior research, students assigned to perspective take (either imagine-self or imagine-other) placed greater weight on physiological relative to psychological factors as causes of the patient’s complaint, t(16) = 2.36, p =.03, 95% CI [.21, 3.90], compared to objective students.
These initial data support the translational potential of perspective taking instructions for medical education curricula to help students maintain empathic functioning in clinical contexts, and to minimize cognitive biases that could detract from medical decision making.