e-dokumen.com

Visualizing uncertainty in health care:present needs and future directionsPaul Han, MD, MA, MPHCenter for Outcomes Research and EvaluationMaine Medical CenterTufts University School of Medicine

Objectives Identify key uncertainties in health care that need to be communicatedto patients Describe recent efforts to develop novel representations for visualizinguncertainty in clinical risk prediction Outline potential directions for future uncertainty visualization effortsin health care

Communicating uncertainty to patients: a growing need Growth of evidence-based medicine (EBM) “The conscientious, explicit, and judicious use of current best evidence in makingdecisions about individual patients.” Increasing visibility of medical controversies Rise of shared decision making (SDM) movementEthical justification: respect for patient autonomy Idea that patients need to understand uncertainty to make well-informed decisions Growing need communicate uncertainty not only to physicians but to patients:what do they need to know?

UncertaintyMain Entry: un·cer·tain·tyPronunciation: \-tən-tē\Function: nounDate: 14th century1 : the quality or state of being uncertain : doubt2 : something that is uncertainsynonyms uncertainty, doubt, dubiety, skepticism, suspicion, mistrust mean lack of surenessabout someone or something. uncertainty may range from a falling short of certainty to analmost complete lack of conviction or knowledge especially about an outcome or result assumed the role of manager without hesitation or uncertainty . doubt suggests bothuncertainty and inability to make a decision plagued by doubts as to what to do . dubietystresses a wavering between conclusions felt some dubiety about its practicality . skepticismimplies unwillingness to believe without conclusive evidence an economic forecast greetedwith skepticism . suspicion stresses lack of faith in the truth, reality, fairness, or reliability ofsomething or someone regarded the stranger with suspicion . mistrust implies a genuine doubtbased upon suspicion had a great mistrust of doctors . A metacognition: the conscious awareness of ignorance Multiple varieties in health care

Uncertainty in health care: domains Prevention and early detection Disease risk estimatesRisks and benefits of preventive interventionsPerformance characteristics of screening tests Diagnosis Interpretation of symptomsPerformance characteristics of screening tests Treatment Risks and benefits of therapeutic, palliative interventionsPrognostic estimates

Uncertainty in health care: ed)TreatmentrecommendationsStructures ofcareProcesses ------PATIENT-CENTEREDMalignant vs.benignLife expectancy,response to treatmentCancer risk factors,carcinogenic eventsEfficacy and safety ofcancer treatmentIdentity, competenceof health care providerRequired actions foraccessing health careEffects of treatment onpersonal relationshipsExamples of specific uncertainty issues: cancer treatmentHan PKJ, Klein WMP, Arora NK. Varieties of uncertainty in health care: a conceptual taxonomy. Med DecisMaking. 2011; Jan 18. [Epub ahead of print]Effects of illness onsense of meaning in life

Uncertainty in health care: sources Probability: indeterminacy of future outcomes, 1st order, “aleatory” Ambiguity: indeterminacy of knowledge, 2nd order, “epistemic”uncertainty Complexity: incomprehensibility of informationHan PKJ, Klein WMP, Arora NK. Varieties of uncertainty in health care: a conceptual taxonomy. Med DecisMaking. 2011; Jan 18. [Epub ahead of print]

Probability Formal language of uncertainty Expression of indeterminacy/randomness Alternative interpretationsObjective (frequentist) interpretation Subjective (Bayesian) interpretation

Ambiguity Decision theory construct (Ellsberg*) A specific type of uncertainty: “second order” vs. “1st order risk”,“epistemic” vs. “aleatory” Lack of “reliability, credibility, adequacy”Incomplete / missing information Amount or quality of available evidence Questionable precision or accuracy Wide confidence intervals Questionable reliability Inconsistent findings, reproducibility Conflicting expert opinion *Ellsberg D. (1961) Risk, ambiguity and the Savage axioms. Quart J Econ, 75

Complexity Features of information that make it difficult to understand Conditional probabilities, multiple risk factors, attributes, outcomes

Sources of uncertainty in health careUNCERTAINTYProbabilityAmbiguity20% probability of benefit10-30% probability of benefitfrom treatment (Indeterminacy from treatment (Imprecision)of future outcome)Expert disagreement aboutbenefits of treatment(Conflicting opinion/evidence)Insufficient scientific evidence ofbenefit (Lack of information)Complexity20% probability of long-termremission from treatment inpatients with localized diseaseand who are HER2/neupositive, estrogen-receptorpositive, pre-menopausal, andhave no other comorbidities(Multiplicity of causal factorsand interpretive cues)Examples and representations of different sources of uncertaintypertaining to breast cancer treatment outcomesHan PKJ, Klein WMP, Arora NK. Varieties of uncertainty in health care: a conceptual taxonomy. Med DecisMaking. 2011; Jan 18. [Epub ahead of print]

Challenges in communicating uncertainty to patients Topic-related Multiplicity of sources, issues Conceptual complexity User-relatedInnumeracy Cultural barriers Individual preferences, tolerance of uncertainty Potential adverse effects: “ambiguity aversion” Method-relatedOptimal representational methods unknown Unclear outcomes: acceptability, understanding, adverse effects (?)

Can communicating uncertainty be bad? “Ambiguity aversion”: propensity to choose against ambiguousoptions, outcome probabilities being otherwise equal Underlying cognitive process: pessimistic bias in theinterpretation of ambiguous risk information Psychological consequences: Heightened perceptions of riskDiminished expectations of benefitIndecision / inaction Greater complexity, potential for confusion*Ellsberg, D. (1961) Risk, ambiguity and the Savage axioms. Quart J Econ, 75

Uncertainty visualization for patients: present needs Need for effective representational methodsPromote understanding Minimize potential adverse effects Promising work on visual approaches Can aid comprehension particularly in low numerate individuals Initial work, more research needed

Clinical prediction models (CPMs) Statistical models to predict future health outcomes“ provide the evidence-based input for shared decision making, by providingestimates of the individual probabilities of risks and benefits combine a numberof characteristics (e.g., related to the patient, the disease, or treatment) to predict adiagnostic or therapeutic outcome.” “Individualized” risk estimates used increasingly for clinical decision making Numerous uncertainties in risk estimates, but not often communicated topatientsSteyerberg E. Clinical Prediction Models: a Practical Approach to Development, Validation, and Updating. NewYork: Springer; 2010.

Uncertainty in CPMs: multiple varieties, levelsFigure 1: Five levels of uncertainty. While the first three form anatural hierarchy, Levels 4 and 5 apply to the entire modellingprocess and may exist even if there is little uncertainty expressedwithin the modelling framework.Spiegelhalter D and Riesch H, 2011. Phil Trans Roy Soc A (in press).

Visualizing uncertainty in risk estimates: past effortshttp://www.yourdiseaserisk.wustl.edu/

Visualizing uncertainty in risk estimates: past y.aspx#Figure 1

Visualizing uncertainty in risk estimates: past efforts No attention to fundamental uncertaintiesAleatory (first-order): indeterminacy/randomness Epistemic (second-order): ambiguity Mental visualization, understanding of users is assumed Need to better represent these uncertainties Users need to understand, but do have problems How should effectiveness be evaluated?Risk perceptions Affective response Decision making Understanding / insight

Visualizing uncertainty in risk estimates: new sk/Default.aspx

Communicating uncertainty in cancer risk estimates: effects NCI Colorectal Cancer Risk Assessment Tool (CCRAT) – Freedman et al2008 Effort to develop patient-centered communication tool using visualrepresentations of uncertainty:Ambiguity (imprecision): model misspecification, error Randomness: indeterminacy Use of new visualization methods: blurring, disarraying Mixed-methods study examining effects of different representationalformatsHan PKJ, Klein WMP, Lehman TC, Massett H, Lee SC, Freedman AN. 2009. Laypersons’ responses to the communication of uncertainty regardingcancer risk estimates. Medical Decision Making 29(3): 391-403.Han PKJ, Lehman TC, Massett H, Lee SC, Klein WMP, Freedman AN. 2009. Conceptual problems in laypersons’ understanding of individualized cancerrisk: a qualitative study. Health Expectations 12 (1): 4-17Han PKJ, Klein WMP, Lehman TC, Killam B, Massett H, Freedman AN. 2010. Communication of uncertainty regarding individualized cancer riskestimates: effects and influential factors. Medical Decision Making 2011 Mar-Apr;31(2):354-66. Epub 2010 Jul 29.Han PKJ, Klein WMP, Lehman TC, Killam B, Massett H, Freedman AN. 2011. Representing randomness in the communication of individualized cancerrisk estimates: effects on cancer risk perceptions, worry, and subjective uncertainty about risk. Patient Educ Counseling 2011 Mar 3. [Epub ahead ofprint]

Visual representations of uncertainty: imprecisionTextual, ambiguity absentTextual, ambiguity presentVisual, ambiguity absentVisual, ambiguity present

Visual representations of uncertainty: imprecisionTextual onlyIntegrated textual visual:Solid barIntegrated textual visual:Blurred bar

Effects of visual representations of imprecision: experimentalevaluationTotal participants(N 135)Total participants(N 240)Ambiguity Absent(Point Estimate)(n 120)Textual(n 60)Visual(n 60)Ambiguity Present(Range)(n 120)Textual(n 60)Visual(n 60)Text-only(n 45)Text solid bar graph(n 45)Text blurred bar graph(n 45)Experiment 2: 3-condition design testing effectsof enhanced representations of imprecisionExperiment 1: 2 x 2 x 2 design testing effects of ambiguity(absent vs. present), representational format (textual vs. visual).Additional test of comparative risk information (pre- / post-)Han PKJ, Klein WMP, Lehman TC, Killam B, Massett H, Freedman AN. 2010. Communication of uncertainty regarding individualizedcancer risk estimates: effects and influential factors. Medical Decision Making 2011 Mar-Apr;31(2):354-66. Epub 2010 Jul 29.