Public Health Insurance and Physicians Behavior in Japan

1
Public Health Insurance and Physicians Behavior
in Japan

• Atsushi Yoshida, Univ. Of Tsukuba
• Shingo Takagi, Hokkaido University

2
Contents

• Brief Comparison of Figures Relevant to Public
  Health Systems
• Introduction of Public Health Systems in Japan
• Empirical Study on Physician Behavior Using
  Insurance Claims from Hospital to Insurers

3
(No Transcript)
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
Characteristics of Public Health Systems in Japan

• The ratio of medical expenditure to GDP is
  moderate
• The ratio is growing fast. The reason is alleged
  to be higher aging rate.
• Japan is similar to UK in health expenditure
• The role of private health insurance is
  negligible in Japan
• Japan is outlier in terms of outpatient visits to
  doctor and inpatient stay in hospital
• Japan is urged to reform the systems to reduce
  frequent visits and longer stay and to improve
  the quality of health services.

8
Health Insurance Systems in Japan
Fee Schedule
Malpractice Insurance
Private Insurance Company
Government
Fee-per-item
Physician
Free Access
Fee-for-service
Health Service
Reseputo (health insurance claims)
Private Insurance
reimbursement
Copayment
Insurer
Insured
Insurance
Premium
Health Insurance Society National Health
Insurance Health Services Systems for the
Aged (for over-70-year-old persons)
Beneficiary head dependents aged
9
Background of the Research

• Change of Cost-sharing Policies for the Elderly
  Health Care
• 2000 starts fixed copayment to proportional
  copayment both in inpatient and outpatient
• Outpatient copayment before the reform
• fixed copayment per visit \530 nearly 3 (up to
  four times visits \2060 nearly 12 per month)
• Outpatient copayment after the reform
• 10 copayment rate for total medical cost (up to
  nearly 12 per month)
• Objectives
• To suppress public spending for the elderly
  health care by discouraging the visits to
  physicians.

10
Prospected Effects of the Reform

• Characteristics of the elderly health care
• chronic but not serious disease
• the demand may be inelastic with respect to the
  copayment rate
• Two types of patients
• Effects of the reform is not uniform for
  patients
• One type copayment increases
• The other type copayment decreases
• Is the demand for health services by the elderly
  suppressed?
• Does the reform has no effect on the treatment by
  physicians?

11
Characteristics of the elderly health care systems

• In general,
• Fee-for-service (Fee-per-item) reimbusement
  scheme
• No distinction between GP and specialist
• No doctor fee nor hospital fee
• A physician determines the treatment
• free access
• No gate keeper
• Particularly in the elderly health care,
• A physician is likely to know the demand
  function of a patient because the elderly patient
  with chronic disease does not easily change the
  physician.
• Then a physician may change her treatment after
  the reform

12
Purpose of the research

• To propose a simple behavior model of patients
  and physicians under elderly health care systems
  in Japan
• Assuming a physician maximizes her payoff
• Assuming a physician acts as a Stackelberg leader
  to the elderly patients.
• To investigate the change of physicians
  treatment behavior after the reform with the data
• Physician may provide more treatment to patients
  whose copayment decreases, while less to patients
  whose copayment increases.

13
Previous Literature

• Demand Inducement physician as the
  decision-maker
• Fee-for-service scheme
• change of the treatment associated with the
  change of relative fees of medical services in
  fee-schedule (e.g. the Medicare Fee Schedule)
• Moral Hazard patient as the decision maker
• change of the cost-sharing rule (proportional or
  fixed copayment )
• patients choice of health insurances whose
  coverage or copayment is different among them
• Optimal health insurance with respect to
  cost-sharing policy and reimbursement scheme
  (principal-agent approach)
• Fee-for-service or capitation
• Gate keeping or non gate keeping

14
Demand Inducement physician as the decision-maker

• McGuire and Pauly (1991) Theoretical
• Gruber and Owings (1996) Empirical
• Cesarean section delivery
• Giuffrida and Gravelle (2001) Empirical
• night visits in primary care
• Yip (1998) Empirical
• Physicians whose incomes were reduced the most by
  Medicare fee cuts performed higher volumes of
  CABGs.

15
Moral Hazard Patients behavior

• Ludin (2000) Empirical
• Patients having to pay large sums out-of-pocket
  are less likely to have trade-name versions
  prescribed than patients getting most of their
  costs reimbursed.
• Cockx and Brasseur (2003) Empirical
• Increase of copayment rate in Belgium
• Chiappori, Durand and Geoffard (1998) Empirical
• Increase of copayment rate in France
• Yoshida and Takagi (2002) Empirical
• Increase of copayment rate in Japan

16
Optimal Insurance Contract cost-sharing policy
and reimbursement scheme

• Mariñoso and Jelovac (2003) Theoretical
• Gps behavior of referral under gate-keeping and
  non-gate-keeping systems
• McGuire (2000) Theoretical
• monopolistic competition
• Blomqvist (1991) Theoretical
• A physician act as double agent of both the
  patient and insurer. Under fee-for-service
  payment scheme, she acts more in interest of the
  patient, while she acts more in interest of the
  insurer under HMO contracts.
• Blomqvist and Léger (2005) Theoretical
• Incorporating both physicians and patients
  incentives
• Discuss second-best optimal insurance in four
  cases of whether or not asymmetric information
  between patient and doctor and whether or not
  fee-for-service or capitation schemes.
• Altman, Culter and Zeckhauser (2003) Empirical
• Contrary to conventional wisdom, indemnity plans
  do not have greater treatment intensity.

17
Patients Behavior

• Free access to providers
• Patients keep visiting a physician as long as her
  utility is not below the reservation utility
• Patients can control the number of visits
• Patients understand the effects of a treatment to
  their health
• Fixed Copayment Proportional Copayment
• Parameter t is opportunity cost of a visit.

18
Patients Visit

• Equilibrium
• Before the reform (fixed copayment)
• After the reform (proportional copayment)

19
Change of the Visits

• Visits increase when
• copayment plus opportunity cost of a visit
  (cost of a visit) becomes lower after the reform
• Visits decrease when
• copayment plus opportunity cost of a visit
  (cost of a visit) becomes higher after the reform

20
Physicians Behavior

• Physician acts as a Stackelberg leader to patient
• She maximizes her payoff given a patients
  reaction function.
• The claims (reseputo) from a physician to the
  insurer is minitored by the insurer.
• Every sort of treatments has its upper limit.

21
Profit function of a provider

• providers profit per patient under fixed
  copayment
• providers profit per patient under proportional
  copayment

22
Providers strategy

• Assumptions
• There are only two types of treatment (x1 and
  x2)
• x1 laboratory tests and diagnostic imaging
• x2 consultations (general and specific) , etc.
• type-2 treatment contributes health but type-1
  not at all
• the rate of difference between the reimbursement
  and cost, rate of payoff, is larger in x1 than in
  x2 (?f-?)
• It is difficult to change x2 rather than x1.

23
Strategy before the reform

• As long as patients optimal visit is positive
  and the difference is positive, the optimal
  intensities of the treatment for x1 and x2 are
  their upper limits.
• However, x1 or x2 do not seem to attain the upper
  limits because of the monitoring.

24
Strategy after the reform

• Marginal effect of type-1 treatment to
  providers profit
• The first term is negative since
• When (pfxt) is large, the marginal effect is
  likely to negative so that a physician decreases
  x1 to make her payoff bigger and vice versa.

25
Cost of a visit increases( )

• The post-reform copayment rates will be higher
  than the pre-reform rate (f/fx).
• Patients may decrease visits that leads to less
  payoff of a provider.
• Case of (pfxt) being large
• It is difficult to change x2 drastically.
• A physician decreases x1 to maximize the payoff.

26
Cost of a visit decreases ( )

• The post-reform copayment rates will be lower
  than the pre-reform rate (f/fx).
• Patients may increase visits that leads to more
  profit of a provider.
• Case of (pfxt) being small
• It is difficult to change x2 drastically.
• A physician increases x1 to maximize the payoff.

27
Data Source

• Reseputo (health insurance claims) data from two
  hospitals, one located in Tokyo (Hospital A) and
  the other in Osaka (Hospital B)
• Hospitals are nearly the same size
• About 400 beds
• Nearly the same specialties, e.g., internal
  medicine, surgery, ophthalmology
• Separation of medical practice and drug
  dispensation is incomplete in Japan
• Hospital A not separated Hospital B separated.
• The averaged claim is larger in Hospital A
  because of prescription drug cost
• Outpatients
• The over-70-year-old aged who visits at least
  once in both pre- and post-reform periods
• Data period 12 months before the reform and 6
  months after the reform

28
Changes of Visits, Costs, Copayment and Copayment
Rate (Table1Hospital A)

• Two types of patients
• Type 1 3621(84) samples of the copayment-rate
  increased
• Type 2 690(16) samples of the copayment-rate
  decreased
• Characteristics of two types
• The type-1 visits the hospital more frequently
  and their costs are higher than the type-2.
• The type-1s points per visit is 1335, whereas
  Type-2s 368.
• The type-1s points per visit or per capita
  expected cost per month did not change, while the
  type-2s increased.
• The type-1s copayment increased, while the
  type-2s does not change.
• This implies a physician changes treatments for
  the type-2 after the reform.

29
(No Transcript)
30
Changes of Visits, Costs, Copayment and Copayment
Rate (Table1Hospital B)

• Two types of patients
• Type 1 3621(57) samples of the copayment-rate
  increased
• Type 2 690(43) samples of the copayment-rate
  decreased
• Characteristics of two types
• The type-1 visits the hospital more frequently
  and their costs are higher than the type-2.
• The type-1s points per visit is 1054, whereas
  Type-2s 303.
• The type-1s points per visit or per capita
  expected cost per month decreased, while the
  type-2s increased.
• The type-1s copayment increased, while the
  type-2s increased.
• This implies a physician changes treatments for
  the type-2 after the reform.

31
(No Transcript)
32
Changes of Medical Costs per Visit by pre-reform
cost groups (Table 2)

• Stratifying samples into 10 groups by decile
  points of the medical costs per visit before the
  reform
• The medical cost per visit increases in the lower
  pre-reform cost groups whereas it decreases in
  the higher groups both of hospital A and B.
• But the higher pre-reform cost groups of hospital
  A decrease the cost slightly.

33
(No Transcript)
34
Transition Matrix of Medical Cost (Table3)

• In general, samples of lower cost-band tend to
  move higher band and vice versa in both hospital
  A and B.
• Samples in hospital A tends to stay in the same
  band after reform, whereas they tend to move in
  hospital B.
• These facts imply that a physician changes
  treatment to the patients in accordance with
  their pre-reform cost band.

35
(No Transcript)
36
Category of Treatments

• Ten categories of treatment in the
  fee-for-service scheme
• General consultation
• Specific consultation
• Doctor-visit cure
• Medication
• Injection
• Simple cure
• Operation
• Laboratory test
• Diagnostic imaging
• Others

37
Changes of Medical Cost by Treatment Categories
and by Cost-bands (Table 4)

• Costs of laboratory test and diagnostic imaging
  increase in the lower pre-reform cost bands,
  while they decrease in the higher pre-reform cost
  band.
• The findings is consistent to the predictions of
  the model.
• It is hard for the PID hypothesis to explain why
  a physician increases the medical cost of the
  lower pre-reform-cost-band patients by using the
  test and diagnostic imaging more.

38
(No Transcript)
39
(No Transcript)
40
Nonparametric Estimation Results

• Conducting statistical inference about the facts
  of table 3 and 4
• Nonparametric Estimation of CDF of Medical Costs
• How does the cdf of cost per visit shift after
  the reform?
• Nonparametric Estimation of the Differences of
  Costs by Treatment Categories

41
(No Transcript)
42
(No Transcript)
43
Nonparametric Estimation of CDFs of Medical Costs

• Fig. 1 and 2 Smoothed CDF of claimed medical
  cost per visit
• The 25th percentile and median shifts to left but
  75th percentile shift to right. The medical cost
  per visit become more dispersed after the reform.
  
• In hospital A, the post-reform CDF crossed the
  pre-reform at about 1000 points.
• In hospital B, the post-reform CDF crossed the
  pre-reform at about 600 points.

44
(No Transcript)
45
(No Transcript)
46
(No Transcript)
47
Nonparametric Estimation of the Differences of
Costs by Treatment Categories (Hospital A)

• Figures of Hospital A
• The differences of total medical cost per visit
  and laboratory tests are positive in lower
  pre-reform cost range but negative in higher
  range.
• There is not so large difference in diagnostic
  imaging.
• A physician changes the treatment after the
  reform.

48
(No Transcript)
49
(No Transcript)
50
(No Transcript)
51
Nonparametric Estimation of the Differences of
Costs by Treatment Categories (Hospital B)

• Figures of Hospital B
• The differences of total medical cost per visit,
  laboratory tests and diagnostic imaging are
  positive in lower pre-reform cost range but
  negative in higher range.
• The seemingly vertical line means the cases where
  only laboratory tests are additionally conducted
  after the reform.
• The seemingly inclined line means the case where
  only laboratory tests are excluded from the
  treatment.

52
Conclusion

• We find that a physician changed the treatment
  after the reform.
• A patient of decreased copayment rate receives
  more laboratory test and/or diagnostic imaging,
  while a patient of increased copayment rate
  receives less.
• We propose a model that a physician behaves as a
  Stackelberg leader to a patient. This model can
  explain the facts.