Cadth 2015 a4 regier cadth bias(1)
Download as PPTX, PDF1 like230 views
1) There are two problems with not involving the public in healthcare decision making: public values are viewed as biased, and this can lead to misallocation of resources. 2) Involving the public is important for both normative democratic reasons and pragmatic reasons like improving comparative effectiveness research. Barriers include a perception that public input is biased. 3) A study tested whether a video introduction could better engage respondents in a stated preference study on incidental genomic findings and mitigate hypothetical bias compared to a text only introduction. Preliminary results found lower willingness to pay values in the video group, suggesting it may reduce bias.
Cadth 2015 a4 regier cadth bias(1)
- 1. Advancing Health Economics, Services, Policy and Ethics 2015 CADTH Symposium Saskatoon, Saskatchewan Dean A Regier, PhD Cancer Control Research, BC Cancer Agency Assistant Professor, School of Population and Public Health, University of British Columbia
- 2. Problem #1 • Input from the public is not routinely pursued in health-care decision-making • Public values viewed as biased Problem #2 • Public values are (probably) biased • Leads to misallocation of scarce resources Public Engagement & Value 2
- 3. Involving the public in policy-forming activities • Public includes patient/lay public Normative & pragmatic motivations • Democratic ideals; economic theory • Comparative-effectiveness Why Public Engagement? 3
- 4. 4*Regier DA, Bentley C, Mitton C, et al. Public Engagement in Priority-Setting: Results from a pan-Canadian Survey of Decision-Makers in Cancer Control. Social Science & Medicine; 2014: 122:130-139. Relative to clinical effectiveness and cost • Input from the public is rarely pursued Barriers • Implies public input is biased
- 5. Stated preference elicitation of utility • Non-market valuation of goods Hypothetical bias • Benefit over-valuation leads to investing in goods that cost too much in terms of available alternatives Mitigating hypothetical bias • Rationality tests; cheap-talk; oath Public Engagement & Bias 5
- 6. Communication theory The medium is the message - McLuhan, 1964 • The medium delivers change separate from content 6 Hypothesis: a video introduction to a stated preference study will differently engage respondents and mitigate hypothetical bias
- 7. Next generation genomic sequencing • Predictive therapy, prognostic therapy, hereditary cause of disease Potential of incidental findings • Information on diseases not related to current diagnosis Background 7
- 8. Published list of incidental findings (Green et al, 2013) • High-penetrance & clinical utility • List of 56 genes, 24 disorders • Labs look for mutations, IF’s returned to patient, through managing physician Controversial • Patients not offered a choice • (Public not consulted) ACMG Recommendations 8
- 9. Objective • Personal utility for the return incidental findings • Discrete choice experiment (two choice + opt-out) Respondent Sample • General public in Canada (N=1200) • English and French language versions Objective & Sample 9
- 10. Define Attributes/levels • Cognitive interviews (n=6)/ 2 focus groups (n=12) Experimental design • D-efficient design with informative priors Statistical Analysis • Mixed Logit Model (preference heterogeneity) Welfare analysis • Willingness to pay (compensating variation) Methods Approach 10
- 11. Evaluate difference in welfare estimates D1 D2 D3 Text Introduction Only Study design Video Introduction & Text Intro English-speaking Respondents D4 randomized randomized 11
- 12. Choice task example 12 Option A Option B No information Diseases with a 80% lifetime risk or higher Diseases with a 5% lifetime risk or higher No information Recommended effective medical treatment and lifestyle change Recommended effective medical treatment only No information Mild health consequences Moderate health consequences No information Does not provide information on carrier status Information on if your family members could be affected No information $425 $1500 $ 0 Option A Option B No Information Disease Risk More disease will be identified if the lifetime risk is lower Disease Treatability Disease Severity Health consequences of the diseases you may develop You m Carrier Status Disease risk not affecting you but can affect your family Cost to you
- 13. 13 Video+Text Version Text Version Scenario 1 Medical treatment , 80% or greater risk, severe QOL $420 95% CI 191-528 $515 95% CI 417-778 Scenario 2 (vs Scenario 1) Medical & No treatment , 80% or greater risk, severe QOL $235 95% CI 195-275 $320 95% CI 225-371 *t-test (unequal variances)=-1.66, p-val=0.11 • Lower WTP values in video version • Potential to mitigate hypothetical bias Welfare Analysis
- 14. 1. Is it necessary for decision-makers to consult the public for each health care investment/disinvestment decisions? 2. Willingness to pay (and utility) is often biased, is there a role for this metric in decision-making? • Focus on naturalistic units? 3. Do researchers need do more with how the public is engaged? Questions 14
- 15. Advancing Health Economics, Services, Policy and Ethics Thank-you • Acknowledgements: Stuart Peacock, Reka Pataky, Kimberly van der Hoek, Gail Jarvik, Jeffrey Hoch, David Veenstra 15 • Funding for this research obtained from the Canadian Centre for Applied Research in Cancer Control (ARCC); ARCC is funded by the Canadian Cancer Society Research Institute grant #019789, #703549