The Role of Consumer Knowledge on the Demand for Preventive Health Care Among the Elderly

1
The Role of Consumer Knowledge on the Demand for
Preventive Health Care Among the Elderly

• Stephen T. Parente, Ph.D., Project HOPE Center
  for Health
• Affairs and Johns Hopkins University
• David S. Salkever, Ph.D., Johns Hopkins
  University and NBER
• Joan DaVanzo, Ph.D., The Lewin Group
• (Formerly of the Barents Group, LLC)
• Research sponsored by the Health Care Financing
  Administration,
• (Contract 500-95-0057/Task Order 2)

2
Presentation Overview

• Study Objectives
• Supporting Literature
• Modeling Approach
• Estimation Results
• Policy Implications
• Next Steps

3
Study Objectives

• Develop an economic model to of demand to
  estimate the direct effect of information on the
  preventive service utilization.
• Identify the effect of income, education and
  insurance coverage, controlling for information
  effects.
• Estimate the benefits of a fully informed
  non-institutionalized population.
• Identify characteristics of a knowledgeable
  Medicare beneficiary .

4
Literature Supporting ResearchRole of consumer
information on medical care demand

• Arrow (1963) Patients rely on providers and
  insurers to act as their agents.
• Pauly Satterthwaite (1981) First empirical
  analysis of the role of consumer informations
  indirect effect on demand.
• Kenkel (1990) Estimated the direct effect of
  information on medical visit demand.
• Hsieh and Lin (1997) Estimate effect of
  information on preventive services in Taiwan.

5
Literature Supporting ResearchKnowledge of
Health Services and Health Insurance Benefits

• McCall et al (1986) Medicare beneficiaries in
  six States have a low level of knowledge of
  Medicare benefits and supplemental policies..
• Issacs (1996) Blendon et al, (1997) Significant
  misunderstanding among the general population
  regarding managed care and entitlement programs.
• Barents/Project HOPE/Westat (1998) General
  probability sample of 1996 Medicare
  non-institutionalized beneficiaries suggests
  people learning principally by experience.

6
Unexplored Areas

• The direct effect of knowledge (not information)
  on the demand for Preventive Services in the
  Medicare population.
• The impact of recent change in preventive care
  benefit design on the use of preventive care.
• The direct effect of Medicare program knowledge
  on curative medical care utilization and cost.
• The 120 Million (Medicare Choice) question
  Will better informed beneficiaries have different
  in their health care use, cost, health status,
  and satisfaction than uninformed beneficiaries?

7
Modeling Approach

• Model the demand for influenza vaccination and
  mammography screening among Medicare
  beneficiaries (both were covered benefits as of
  1992).
• Use a two part model of demand
• Treat beneficiary knowledge of the preventive
  care benefit as endogenous.
• Use IV estimation with instrument(s) for
  beneficiary knowledge of the benefit.

8
Two Part Demand Model for Preventive Care

• Eq. 1 Ki a0 a1Xi ui
• Eq. 2 Di b0 b1Yi b2Ki vi
• Ki knowledge of beneficiary I
• Xi vector of patient characteristics relevant
  to obtaining K
• Di preventive service demand (0,1)
• Yi beneficiary attributes
• ui and vi random error terms
• Eq. 2 provides a reduced form demand expression

9
Data Medicare Current Beneficiary Survey (MCBS)

• Panel survey of (14,000) beneficiaries offered
  every year since 1991.
• Household interview that occurs in three rounds
  annually.
• Self-report survey data is linked to Medicare
  claims data for the non-Medicare risk HMO
  population.
• Provides cost and use information (including
  drug use), SES, health status, functional status.

10
Data Used for this analysis -1

• 1996 calendar year MCBS population
• Three rounds of combined data
• Round 16 Cost and Use, Medicare claims for all
  1996 attached.
• Round 17 Includes supplemental questionnaire
  regarding beneficiaries information needs.
• Round 18 Includes supplemental questionnaire on
  specific Medicare benefit and program knowledge.

11
Data Used for this Analysis -2

• Selected a subset of non-institutionalized
  elderly. (N9735 all, 5601 women)
• Dependent Variables
• Mammography screening use (for women) 1yes,
  0no
• Flu shot 1yes, 0 no
• Explanatory Variables (e.g., benefit knowledge)
• True (1) or false (0), Medicare covers flu shot
  /mammography?
• Control Variables (demand shifters)
• Income, Education, Network effects
• Insurance (supplemental, dual eligibility)
• SES, Health status, functional status

12
Data Used for this Analysis -3

• Key Instrumental Variable for Benefit Knowledge
• Combined score of Medicare program quiz
  questions
• A physician who accepts assignment can not
  charge whatever they want?
• You have the right to appeal a decision made by
  Medicare regarding a covered benefit?
• Physical exams are a covered Medicare expense?
• 21.5 got all three right.
• Second Instrument
• Correct Knowledge of the other preventive care
  benefit.

13
Descriptive Statistics of Key Variables

• Flu Shot Mean STD
• Use 64.9 (0.477)
• Correct Knowledge 76.4 (0.425)
• Mammogram (women only)
• Use 39.4 (0.489)
• Correct Knowledge 66.2 (0.473)
• Program Rules 21.5 (0.411)
• women only 19.9 (0.399)

14
First Stage Results-1Flu Shot Benefit Knowledge
15
First Stage Results-2Mammography Benefit
Knowledge
16
Second Stage Results-1Flu Shot Use
17
Second Stage Results-2Mammography Screening Use
18
Key Findings

• Knowledge of a preventive care benefit appears to
  have the greatest impact on demand.
• Income, supplemental coverage and education level
  have a positive impact on the demand for
  services.
• Mixed results regarding race. Black
  beneficiaries less likely to get a flu shot, but
  there is not a similar finding for mammography.
  May reflect an interaction between race and
  gender.

19
Policy Implications

• If all non-institutionalized beneficiaries were
  knowledgeable about prevent care benefits
• 1.5 million for beneficiaries immunized (9.1
  increase)
• 1 million more females screened (14.7 increase)
• Per beneficiary annual cost-saving for flu shot
  is 1.64 to 9.03 depending on health status.
• Medicare will spend 3 per year on beneficiaries
  for consumer education.
• Medicares goal of increasing preventive service
  use further will depend on the marginal cost of
  reaching one more beneficiary.

20
Next Steps

• Control for impact of prior use of services on
  knowledge.
• Compare results from 2SLS to bivariate probit
  results.
• Examine effect of knowledge on health care
  utilization and cost.
• Develop new instruments for Medicare Choice
  analysis using the MCBS.

21
Solicitation for Future Instruments to be
included in MCBS Supplement

• Current list includes
• Most people covered by Medicare can choose among
  different kinds of health insurance plans within
  Medicare (T/F).
• Medicare by itself doesnt pay for all of
  your/(SPs) health care expenses.
• The government is trying to force some people on
  Medicare to join HMOs even if they dont want to.
• People can complain to Medicare about their HMOs
  or supplemental plans if they are not satisfied
  with them.
• HMOs that cover people on Medicare often cover
  more health services, like prescribed medicines,
  than Medicare by itself.
• If (you/SP) join(s) an HMO that covers people on
  Medicare, (you/SP) will have fewer benefits than
  if (you/SP) just had Medicare by itself.
• If (you/SP) join(s) an HMO that covers people on
  Medicare, (you are/SP is) allowed to drop out
  only during certain times of the year and still
  be covered by Medicare.