economics writing

ECON 317 The Economics of Canadian Health Care

Lecture 6: The physician as the patient’s agent

January 17th , 2020

Version 1.1 (Jan 17) – Added slide 10.

Required Reading

• Weinstein, M. C. (2001). Should physicians be gatekeepers of medical resources? Journal of Medical Ethics, 27, 268-274. Retrieved from http://jme.bmj.com/content/27/4/268.full

• This paper examines the tragedy of the medical commons, and the nature of physician responsibility to society as a whole vs their individual patients.

• You only need to read pages 271 to 273, plus the first paragraph on p. 274. Start with ‘The role of physicians’ on p. 271.

• Note: A QALY is a ‘quality adjusted life year’, a standard unit of measure of health gains. 1 QALY = 1 year in perfect health.

2

Optional Readings

• Mooney, G. & Ryan, M. (1993). Agency in health care: getting beyond first principles. Journal of Health Economics, 12, 125-133. Retrieved from http://www.sciencedirect.com/science/article/pii/0167629693900238

• An excellent summary of standard agency theory, and its limitations when applied to health care.

• Gafni, A., Charles, C. & Whelan, T. (1998). The Physician Patient Encounter: The Physician as a Perfect Agent for the Patient Versus the Informed Treatment Decision-Making Model. Social Science & Medicine, 47(3), 347-354. Retrieved from https://doi.org/10.1016/S0277-9536(98)00091-4

• Charles, C., Gafni, A. & Whelan, T. (1999). Decision-making in the physician-patient encounter: revisiting the shared treatment decision-making model. Social Science & Medicine, 49(5), 651-661. Retrieved from https://doi.org/10.1016/S0277-9536(99)00145-8

• These two papers investigate the nature of shared decision-making and what it takes to be a perfect agent.

• Labelle, R., Stoddart, G. & Rice, T. (1994). A re-examination of the meaning and importance of supplier- induced demand. Journal of Health Economics, 13(3), 347-368. Retrieved from https://doi.org/10.1016/0167-6296(94)90036-1

• An excellent article on the meaning and consequences of supplier-induced demand in health care. The basis of much of the second half of this lecture.

3

Learning objectives

• Gain an introductory understanding of principal-agent problems.

• Gain an introductory understanding of how incentive constraints are calculated.

• Gain an introductory understanding of the ways in which agency in health care deviates from standard agency theory.

4

The principal-agent problem

• A poorly-informed principal employs a well-informed agent…

• …to perform some duty in a way that will maximize the principal’s utility.

• The agent has her own, independent utility function.

• Because the principal and agent have different goals, and because the agent is better-informed, there is an incentive for the agent to cheat.

• The principal’s task is to come up with a contract that will ensure the agent acts in the principal’s best interest.

• Often, the agent only observes the outcome of the task, so the contract can only be conditioned on that outcome.

5

Moral hazard and physician agency

• For our purposes, the patient is the principal and the physician is the agent. The physician has superior information on health care, and the patient can (often) only observe the outcome of treatment.

• The agent can put High effort, H, or Low effort, L, into treatment. Effort is costly to the physician. Only the physician knows whether effort was H or L.

• Patients can be sick, S, very sick, V, or Terminal, T. S always recover, T always die, V recover only with effort H. Patient type is unknown to agent and principal.

• Each patient type is equally likely (1 in 3 chance)

• The physician has an incentive to always work with effort L, and in case of patient death, claim the patient was T - a moral hazard.

• (This drastic teaching example shares features with more realistic cases.)

6

Finding the right incentives

• In our simplistic example, the patient (or her estate!) can only observe the outcome of treatment: recovery or death.

• The patient would like the agent to employ H effort.

• Since effort can’t be observed directly (the patient only has the physician’s word for it), this must be done by rewarding the agent for patient survival.

• The contract, then, must take the form of an up-front payment P, plus a bonus payment B if the patient recovers after treatment.

7

The situation graphed

Agent Income,Y

Agent Utility U(L,Y)

U(H,Y)

Reservation Utility, U

Cost of effort, C

PL PH

• Given effort L, the agent is willing to take payment PL.

• To make effort H, the agent needs payment PH.

• If offered PH, the agent could earn an extra utility of C by only putting in L effort.

8

Calculating optimal payments • The physician will be paid P(rice) at the start of treatment, and B(onus) if the patient recovers.

• Suppose the chance of recovery is θH for high effort, and θL for low effort.

• Moreover, θH > θL.

• We must have expected utility from low effort be less than or equal to expected utility from high effort (the bonus payment allows this):

• 1 − θL U L,P + θLU(L,P + B) ≤ 1 − θH U H,P + θHU(H,P + B)

• To minimize costs, this should bind with equality.

• We also need expected utility to be at least equal to U. This constraint should also bind with equality for cost minimization.

• The end results will be that P < PL, but (P + B) > PH.

• Your turn: what’s the intuition for each of those?

• The second is a function of risk aversion in the way we drew the utility functions. Physicians need to be compensated for the risk: expected utility is less than the utility of the expected income.

• (Expected utility will be on the line connecting the two outcomes on the graph.)

9

The contract graphed

Agent Income,Y

Agent Utility U(L,Y)

U(H,Y)

Reservation Utility, U

PL PH

10

P P+B

1 − θH U H,P + θHU H,P + B = U

U(H,P + B)

U H,P

1 − θL U L,P + θLU(L,P + B) = U

U(L,P + B)

U L,P

What if the physician is risk neutral?

• If the agent is risk neutral, the situation is much easier.

• The principal can pass on all the risk to the agent, and the contract becomes:

• ‘Pay the agent the value of the outcome, minus a share reserved for the principal.’

• This is tricky to picture in health care, but not uncommon in agriculture:

• A landlord charges a fixed rent to the farmer, and the farmer is free to keep any extra income she makes from harvesting (risky, uncertain) crops.

11

Concepts of Agency in Health Care

• Patient: has private preferences

• Physician: has private knowledge

• View 1: a perfect agency would be one in which the physician transfers to the patient all necessary information, and the patient then makes the decision.

• Problem: what information? Health outcomes? More?

• View 2: a perfect agency would be one in which the patient communicates to the physician the entirety of her preferences, and the physician uses her knowledge to make a decision consistent with them.

12

Types of Information Exchange

• Recent models of patient/physician relationships have looked at shared decision making. Information flows in two directions:

• Medical knowledge, from the physician to the patient.

• Preferences, from the patient to the physician.

• Physician  Patient: makes sure all options are on the table and the consequences understood.

• Patient  Physician: makes sure the options are evaluated according to the patient’s unique (cultural, social, personal) context rather than assuming ‘one size fits all’.

13

What does it take to be a perfect agent? • This is trickier than it looks. To act as a perfect agent for her patients, a

physician must know the entirety of each patient’s utility function.

• (And utility functions can change with health status.)

• Simple questions aren’t enough to do this, especially with uncertainty involved. (e.g. Would a patient prefer pre-emptive chemo now or to live with a higher risk of cancer down the road?)

• One approach is to use decision trees. The physician fills them out with her knowledge of treatments and probabilities….

• …and the patient then attaches a valuation to each possible outcome.

• The tree is then ‘rolled back’ to find the most appropriate treatment option.

• We’ll learn how to do this later in the course.

14

A few more issues

• Not all patients have preferences compatible with expected utility theory.

• Not all patients understand or can be led to accept expected utility theory.

• Understanding the implications of the decision-tree approach to decision- making is non-trivial… we’re spending a few university lectures on it!

• Information flow in the other direction (medical information from the physician to the patient) is more manageable.

• In the past few decades, there have been great advances in decision boards and other visual aids.

15

Why don’t we see these schemes?

• In practice, we don’t often see complicated outcome-based payment schemes for medical professionals.

• Mostly fee-for-service, capitation, salary and so on. • (But see pay-for-performance, which we’ll cover later in the course.) • Elaborate incentives are needed in agency theory because we assume that

the utility functions of the principal and agent are separate. • If this is NOT the case, and patients and physician utility functions are

interconnected, then these incentives may not be needed. • More importantly… in many settings there is a SECOND ‘principal’

(government or insurance company) with the power to determine or strongly influence the methods of payment.

• This second principal may also impose other, non-monetary constraints…

16

The tragedy of the second principal

• The Tragedy of the Commons: Common pasture is over-grazed due to each farmer only looking to her own cattle.

• There exists a ‘medical commons’ of limited resources. Physicians, especially in a single-payer health care system, have a responsibility not just to their patients but to society as a whole.

• This implies that physicians can and should ration care to their patients. This goes against the expectation of perfect agency…

• …but is accepted and understood in settings such as emergency room triage and (regrettably scarce) organ donation.

• Society or the single-payer health care system can be thought of as a second principal in the agency problem. Just how the two principals’ conflicting objectives should be reconciled is an open, difficult and fascinating question.

17

Imperfect Agency

• In order to act as a perfect agent for her patients (and society):

• Maximize the patient’s health

• Maximize the patient’s utility

• Maximize health status or utility of society as a whole

• BUT physicians have their own utility functions:

• They may change their actions in response to financial (or other) incentives, even when this does not benefit the patient or society at large.

• To the extent physicians prioritize their own utility over their principals’, they are imperfect agents.

18