Too Smart to Forgo: Cognitive Ability and Subsidized Prescription D g IInsurance Drug Kate Bundorf Stanford University School of Medicine Collaborator: Helena Szrek AcademyHealth June 27, 2010 We gratefully acknowledge funding for this project from The National Institute on Aging through the Center for Advanced Decision Making on Aging at Stanford University (AG024957). We also thank Jon Baron, Jay Bhattacharya, Edward Cokely, Alan Garber, Abby King, and Kosali Simon for contributing to this project in various ways. 1 Plan Choice and Medicare Part D • Beneficiary choice among plans is a cornerstone of the Medicare Part D Prescription Drug Benefit. Benefit – Making a choice among plans is necessary to obtain a publicly subsidized benefit. benefit • Concern over the large number of choices people f face in i their th i d daily il li lives. – Over 40 Part D plans are available in every region of the country in 2010. 2010 • Questions regarding the desire and ability of older ld adults d lt tto make k choices h i iin thi this context. t t 2 Related Research • Many studies provide evidence that cognitive ability is positively correlated with performance on judgment and decision-making tasks. – C Cognitive iti ability bilit will ill nott b be correlated l t d with ith performance when decisions are too easy or too difficult (Stanovich and West 2008). 2008) • People delay decision-making decision making or are less satisfied with their decision when choice is extensive (Iyengar and Lepper 2000; Swartz 2004). 2004) 3 Research Questions • Does cognitive ability improve decision quality among older adults choosing whether to enroll in a Medicare Part D prescription drug plan? • Does the effect of choice set size on decision quality vary by cognitive ability? 4 Experiment • Web-based experiment in which people age 65 and over make hypothetical choices among Medicare Part D prescription drug plans. – Plans and information designed to closely resemble Medicare part D. • Administered on Knowledge Networks internet internetenabled panel (n=280, we use data for 240). • Respondents chose a plan from randomly assigned choice sets of different sizes (2, 5, 10 and 16)) and then answered a series of q questions about their choice and themselves, including cognition instruments. 5 Dependent Variable • Expected Enrollment: If presented with the choice of the above plans, plans how likely would you be to enroll in ANY plan (where the alternative is going without a plan)? – Response is based on a 7 point scale (1=Certain NOT to enroll 7=Certain to enroll) enroll, • Restrict sample to those for whom enrollment is normatively ti l optimal ti l (prescription ( i ti d drugs>=1). g 1) – For an individual taking one prescription drug regularly the expected benefits less the premium for regularly, the standard plan are equal to their expected costs if they do not enroll in a plan (Heiss, Winter and McFadden 2006). 6 Measures of Cognition • Numeracy (Lipkus et al., 2001 & Schwartz et al., 1997) – Imagine that we rolled a fair, six-sided die 1,000 times. Out of 1,000 rolls, how many times do you think the die would come up even (2, 4, or 6)? • Cognitive Reflection (Frederick, 2005) – A bat and a ball cost $1.10 in total. The bat costs $1.00 more th th than the b ball. ll H How much hd does th the b ball ll cost? t? • Memory – Individuals shown the choice set from which they made their second choice and were asked to recall the plan they chose. chose – Also asked how confident they were on the recall question. 7 Data Analysis • Ordered probit regressions with 3-category likelihood of enrollment as the dependent variable (1-3, 4, and 5-7). • Main M i effect ff off cognition ii – Separate models with cognition measures entered separately l and d jointly. j i l • Interact cognition measure with choice set size • Control variables include age, education, g taken household income, number of drugs regularly, gender, marital status, race, p y status. household size, employment 8 Change in Probab C bility of R Response e rela ative to Z Zero Corrrect Results: Numeracy and Enrollment ** * * ** 9 Change in Probab C bility of R Response e rela ative to Z Zero Corrrect Results: Interaction of Numeracy and Choice Set Size High Numeracy 0 30 0.30 * * 0.19 0.18 0.20 0.10 0.00 -0.10 Low Numeracy 0.30 0.20 0.10 * * Unlikely to Enroll -0.07 -0.07 -0.20 0 20 5 plans 10 plans 0.00 Likely to Enroll 16 plans -0.10 Unlikely to Enroll -0.20 0 20 5 plans 10 plans Likely to Enroll 16 plans 10 Conclusions • Cognitive ability is a strong predictor of expected enrollment. enrollment – Individuals with higher cognitive ability as measured by numeracy numeracy, cognitive reflection reflection, and memory are 16-27 percentage points more likely to enroll than those with lower cognitive ability. – Hypothetical responses are supported by actual enrollment decisions;; individuals that do not have drug insurance through Medicare or other sources (8% of our sample) have lower numeracy scores (mean of 1 39 vs. 1 1.39 1.57), 57) llower cognitive iti reflection fl ti scores (mean ( of 0.32 vs. 0.82), and lower recall. 11 Conclusions (con’t) • Our results are suggestive of an interaction between cognitive ability and choice set size – People with greater cognitive ability seem to be more sensitive to choice set size. size – Older adults with lower cognitive ability do not have the “mindware” mindware (Stanovich and West, West 2008) necessary for the decisions, and hence they are not sensitive to the decision frame ((choice set size). ) 12 Some Caveats • We only consider enrolling vs. not enrolling, which is only one level of good decision making. making • Our population is more educated than a representative i sample l off elderly ld l A Americans. i • Information was presented in an organized tabular format. • Choices made in the context of an experiment p may differ from those made in the real world. 13 Policy Implications • Older adults with low cognitive ability may be particularly vulnerable under the design of the Medicare Part D prescription drug benefit. – Th The problem bl may nott be b due d tto extensive t i choice h i ((as many critics argue) but because seniors are forced to make a decision in the first place • Policy approach should address the differential needs and capabilities of both groups. groups – Efforts to simplify the process, such as manipulating the number of choices or providing information, information most likely to help those with greater cognitive ability. – People with less cognitive ability may benefit from more paternalistic forms of assistance such as 14 automatic enrollment or expert advice.