Specialists

Specialists have a unique position in the health system as they provide health-care services to patients and so are agents of patients, but in addition, they provide patients to, and order services from, other health-care providers (e.g., hospitals) and so are agents of other health-care providers as well. In the economics literature this situation is known as common agency. These two agency relationships are characterized by asymmetric information, in that patients have limited medical knowledge about the type of specialist to visit, the quality of any particular specialist, or the appropriate treatment, whereas hospitals have limited knowledge about the preferences of specialists over income and patient welfare.

In addition to having a unique position in the health-care system, specialists operate in a unique market. Patients’ health-care expenditures are often covered by public or private insurance and so the demand for specialists’ services is often not rationed by price. The payment that specialists receive for their services is either regulated by the public purchaser, the private insurer, or a private purchaser such as a Preferred Provider Organization or a Health Maintenance Organization. As a result, the quality and the quantity of services specialists provide depend very much on the incentives contained in their payment schemes.

In this environment, some interesting questions are: (1) what factors lead to particular specialty choices by doctors, (2) how are patients allocated between general practitioners and specialists, (3) is the quality and quantity of services provided by specialists efficient, and (4) how are specialists allocated between hospitals. These questions are addressed below.

The Allocation Of Doctors

An important result in economics is the First Fundamental Theorem of Welfare Economics that states that competitive markets are efficient. A fundamental characteristic of competitive markets is free entry. However, entry into the medical profession is not free as places at medical schools are restricted by governments and entry into specialties is restricted by various professional bodies. As a result, the allocation of physicians between specialties is often inefficient. For example, in many countries, the market for specialties is characterized by excess supply, whereas the market for general practitioners is characterized by excess demand.

These inefficiencies have led researchers to investigate what factors are important in the specialty choice decision of doctors as these factors can then be manipulated to achieve a more efficient outcome. A number of US and Canadian studies have found that graduates of medical schools are more likely to choose specialties with less demanding workloads and higher expected income. In the US, the income elasticity of a particular specialty choice has been estimated to be 1.4, where the income elasticity of a particular specialty choice was defined as the percentage increase in the number of medical students who rank a particular specialty first associated with a 1% increase in expected income. In Canada, the income elasticity of any specialty choice (relative to general practice) has been estimated to be 0.2, where the income elasticity of any specialty choice was defined as the percentage increase in the number of physicians who chose any specialty (relative to general practice) associated with a 1% increase in the fee-perconsultation. Specialty choice is quite responsive to expected income. Therefore, policy makers can influence specialty and general practitioner choice through income differentials and by changing the work environment.

The Allocation Of Patients

Patients with a medical condition have limited medical knowledge and so do not know whether a general practitioner (GP) or a specialist is needed to treat the condition. Even if patients knew they need to see a specialist, they do not know which type, for example, does a patient with chest pain need to see a cardiologist or a thoracic surgeon? To analyze the first situation, consider a model in which treatment by a specialist is more expensive than by a GP, but for some conditions treatment is only successful if completed by a specialist. In this latter case, if the GP treats the patient, the patient suffers a waiting loss as specialist treatment is delayed. To determine which type of condition the patient has required the GP to expend effort. In this environment, if specialist costs and patient waiting costs are high, then a gate-keeping system, in which patients must get a referral from a GP before seeing a specialist, is more efficient than a system in which referrals are not needed. Under gate-keeping, a GP payment scheme, which involves a bonus for nonreferral and another bonus for not providing treatment provides GPs with an incentive to undertake effort and only refer those patients with the condition that only specialists can treat. This saves on patient waiting costs and ensures that patients with the condition which is treatable by GPs are not referred to specialists. GP fund-holding, whereby GPs are given a budget from which they pay for a range of elective procedures for their patients, also provides incentives for GPs not to refer patients to specialists and hospitals as it acts like a bonus for nonreferral. However, this is at the cost of not referring patients who should be referred.

The second situation is similar to the first. In a gate-keeping system the GP makes a diagnosis and makes an informed decision about which type of specialist the patient should see. However, in a non gate-keeping system, the uninformed patient chooses which type of specialist to see and if they make an incorrect decision, they waste scarce specialist time and delay appropriate treatment. Where specialist costs are high and delay in treatment is costly, a gate-keeping system dominates a non gate-keeping system.

Specialist Quality And Quantity Of Care

The benefit a patient, who seeks diagnosis or treatment from a specialist, receives depends on the quality of care provided by the specialist, Q, the quantity of medical services delivered by the specialist, q, and the quantity of medical services the specialist orders from other health-care providers such as hospitals, q0. Let this benefit be given by B=B(Q,q,q₀), where benefit is increasing in all three variables. Patient benefit also depends on the quality of the services provided by other health-care providers, but this is outside the domain of the specialist. The cost of the specialist providing services is given by C =C(Q,q) and the cost of the services provided by other care providers is C₀=C₀(q₀). Both these costs are increasing in all variables. The efficient provision of quality occurs where the extra patient benefits a unit of quality generates equals the extra cost of providing that unit. A similar condition applies for the efficient provision of the quantity of specialist services and the quantity of services ordered from other health-care providers.

Illness is an uncertain event and potential patients are often covered by either private or public health insurance for any health expenditures they incur. The terms of these insurance contracts affect the decision to seek care from a specialist and then the specialist determines the quality and quantity of care once care has been sought. Therefore, the payment scheme that the public or private insurer offers to specialists and the incentives it contains will be a major determinant of the quality and quantity of specialist care.

If quality and quantity were observable and verifiable by courts, a contract between the public or private purchaser and the specialist could be written in which payment was conditional on a particular quality and quantity of service. Competition between private purchasers or maximization of welfare by the public purchaser would then ensure the efficient provision of quality and quantity of health care. However, although the quantity of services is observable and verifiable the same is not true of quality. The quality of a medical service has many dimensions, for example, the appropriate diagnosis of a patient’s condition, the appropriate treatment given diagnosis, and appropriate pain management. These dimensions are likely to vary from condition to condition and from patient to patient and so are not easy to specify in a contract in a verifiable manner. Therefore, efficiency cannot be achieved by the writing of conditional payment contracts.

Although quality is not verifiable in courts, it nevertheless maybe be observable to patients. If patients can observe the quality of provider care and the demand for a particular provider is increasing in quality, it is well known that a prospective payment (a payment per patient) coupled with fee-for-service (a payment per service) can achieve efficient provision of provider quality and quantity of service. Essentially the fee-for-service is set to achieve the efficient provision of quantity and the prospective payment is set to achieve the efficient quality as providers compete for valuable patients.

For specialist services, the assumption that patients can observe quality is problematic. Patients in general have little or no information concerning what is the appropriate treatment. In addition, their relationship with specialists is often once-off and so they cannot use past experiences as an indicator of quality as they might do with other health-care providers with whom they have long-term relationships. As competition for patients cannot be relied upon to ensure efficient quality what institutions or avenues exist which might?

Institutions

Specialists need to be licensed to practice, but given evidence from the US that current specialist licensing arrangement guarantee nothing more than minimum specialist expertise and little specialist learning postmedical school, licensing does not provide an indication of specialist quality. In addition, specialist quality is not ensured through self-regulation as review boards fail to respond to patient complaints, and when they do, they rarely impose serious disciplinary sanctions.

Tort law is a vehicle through which specialists who are negligent are liable for any damage that results from not exercising due care. This gives specialists an incentive to maintain quality, but not a very strong one as evidence suggests patients are not very good at detecting negligence when it occurs.

Unlike patients, gate-keeping GPs do form long-term relationships with specialists. As a result, over time, they are able to monitor many patient outcomes from treatment by particular specialists and base referral decisions on these outcomes. This is an imperfect mechanism for ensuring specialist quality as GP referral decisions are not only based on past patient outcomes, but on other factors, for example, friendship. Furthermore, where GPs perfectly observe specialist quality and base their referrals only on quality, specialists have an incentive to overprovide quality to receive more referrals. This incentive to overprovide quality is mitigated if specialists are paid by a combination of a prospective payment and fee-for-service.

In summary, licensing, tort law, and GP gate-keeping provide only limited incentives for specialists to provide quality efficiently.

Specialist Preferences

An avenue for the efficient provision of quality and quantity exists if specialists not only value income, but also the welfare of their patients. Such altruism on the part of providers is thought to be an important characteristic of medical services. Therefore, it is assumed that specialist utility is given by U=aB(Q,q,q₀)+I-C(Q,q), where a is the weight the specialist attaches to patient welfare and I is income. Under the assumption that specialist payments are regulated in most countries by public or private purchasers, specialist income is assumed to consists of a prospective payment, P, and cost reimbursement of r∙C(Q,q), where r is the proportion of cost that is reimbursed, that is, I=P +r ∙C(Q,q). It is assumed that cost is observable.

If the payment scheme involves only full-cost reimbursement, so that P=0 and r=1, specialist utility is U=aB (Q,q,q₀). The specialist maximizes utility by choosing quality so that the extra benefit the patient gets from an extra unit of quality is zero. The specialist is not concerned with the extra cost involved in providing an extra unit of quality as costs are fully reimbursed. Therefore, relative to the efficient quality, the specialist provides ‘too much’ quality. Similarly, the specialist provides ‘too many’ services and orders ‘too many’ services from other providers relative to the efficient quantities.

However, if the payment scheme involves only a prospective payment, so that P>0 and r=0, specialist utility is U= aB(Q,q,q₀)+P –C(Q,q). The specialist maximizes utility by choosing quality so that the extra benefit the patient gets from an extra unit of quality weighted by a equals the extra cost of providing that unit of quality. If the specialist values patient benefit and income equally, that is, if a=1, then the specialist provides the efficient quality. However, if the specialist values patient benefit less than income, a<1, then the specialist provides ‘too little’ quality relative to the efficient quality. Similarly, the specialist provides ‘too few’ services relative to efficient provision, but orders ‘too many’ services from other providers relative to efficient provision as the specialist bears none of the costs of the other providers’ services.

Finally, if the payment scheme involves both a prospective payment and some cost reimbursement, so that P>0 and r<1, specialist utility is U=aB(Q,q,q₀)+P –(1- r)C(Q,q). The specialist maximizes utility by choosing quality so that the extra benefit the patient gets from an extra unit of quality weighted by a equals the extra cost of providing that unit of quality weighted by 1-r. If r is chosen so that r =1-a, then the specialist provides the efficient quality and also the efficient quantity of services. It is still the case that ‘too many’ services are ordered from other health-care providers.

So even though quality is not observed, a payment scheme, which is a mix of a prospective payment and partial cost reimbursement (supply-side cost sharing) is able to induce the specialist to provide the efficient quality and quantity of own services. Having patient welfare in the specialist utility function provides an incentive for the specialist to provide quality and reimbursing some of the cost of quality provision reinforces this incentive.

The costs of the specialist not only depend on quality and quantity, but also on the amount of unobservable cost reducing effort the specialist expends. In this case, specialist costs are given by C(Q,q,e), where costs are decreasing in cost reducing effort e. The cost of this effort to the specialist is v=v€ and is increasing in effort. If the specialist payment scheme depends only on q and contains no cost reimbursement, specialist utility is U=aB(Q,q,q₀)+P(q)-C(Q,q,e)-v(e) and the specialist has strong incentives for cost reduction, but weak incentives for quality provision. However, if the specialist payment scheme involves some cost reimbursement, then specialist utility is U=aB(Q,q,q₀)+P(q)-(1-r)C(Q,q,e)-v(e), and the specialist has weak incentives for cost reduction, but strong incentives for quality provision. The optimal payment scheme involves some cost reimbursement as long as a<1.

Different specialists have different degrees of altruism, different a’s. In this case, it is optimal to offer a menu of payment schemes to specialists and allow each specialist to choose the one that is best for them. These payment schemes involve a fixed payment component and a cost reimbursement component with a greater fixed payment component being associated with a smaller cost reimbursement component. Specialists who are more altruistic choose a scheme with a greater fixed component and a smaller cost reimbursement component. This has intuitive appeal as more altruistic specialists have a greater incentive to provide quality and so are given strong incentives for cost reduction.

The consensus from this theoretical literature is that quality provision above some minimum amount ensured by licensing requirements, malpractice liability, or GP referral decisions requires specialists to value the welfare of patients. If specialists value patient welfare equivalently to their own income, then a prospective payment with no cost reimbursement attains efficient provision of quality and quantity. However, if specialists value patient welfare less than their own income, then regardless of whether costs depend on cost-reducing effort or not, the optimal payment scheme involves some degree of cost reimbursement. Full-cost reimbursement is not optimal as ‘too much’ quality is provided and ‘too little’ cost-reducing effort is undertaken.

Given the specialist values patient benefit, the specialist orders ‘too many’ services from other health providers. This is because the specialist bears none of the cost of these services. However, if the specialist values the welfare of patients and the other health provider’s profit, then efficiency can be achieved. To help clarify the exposition, let the other health provider be a hospital. Assume the hospital is paid with a prospective payment and some cost reimbursement, then hospital profit is П=P-(1-r)C₀(q₀). Specialist utility is U=aB(Q,q,q₀)+ П. If a<1, and if the proportion of cost reimbursement is chosen so that r=1-a, then the specialist orders the efficient quantity of services from the hospital. Here it is not how the specialist is paid that is fundamental, but rather how the hospital is paid. Once again, some cost reimbursement is optimal.

It should be noted that little has been written about an environment in which insured patients do not know specialist quality and specialist payments are unregulated.

Empirical Evidence Concerning Specialist Payment Schemes

Although the theory above suggested that efficiency of quality provision required at least some cost reimbursement, this is not how specialists are usually paid. In practise, specialists are paid in a number of ways including salary, fee-for-service, and combinations of both salary and fee-for-service. Salary is similar to a prospective payment in that it does not depend on the quantity of services delivered to a patient. Theory suggests it will induce ‘too little’ quality and ‘too few’ services relative to efficient provision. Fee-for-service involves a payment for each service delivered so depending on the level of the fee ‘too little’ or ‘too many’ services will be provided relative to efficient provision. If the fee is set greater than the extra benefit of an additional service at efficient provision weighted by (1-a), then too many services relative to efficient provision will be provided. Fee-for-service will also induce ‘too little quality.’

It is useful to think of Q and q as the amount of time the specialist devotes to quality and quantity provision, respectively, where quality is increasing in the time spent gaining and retaining expertise and quantity is increasing in the time spent delivering services. That is, the specialist devotes time to two tasks. The time devoted to both tasks in total is Q+ q. Salary provides ‘weak incentives’ for quality and quantity provision, whereas fee-for-service provides strong incentives for quantity. Relative to payment by salary, with payment by fee-for-service the specialist provides a higher quantity as it is strongly rewarded and a lower quality as its marginal cost of provision has increased. Because quantity is higher and quality is lower, it is not clear whether the patient is better-off.

A number of empirical studies have examined quantity choices under various payment schemes. The usual finding is that the quantity of services provided by specialists is greater under payment by fee-for-service than under payment by capitation (a prospective payment, where the specialist receives a fixed payment per patient) or salary.

The predictions of the multitasking framework have also been examined empirically in an environment in which specialist have a choice between being paid by fee-for-service or by a mixed payment scheme (salary plus fee-for-service). As the mixed scheme has a smaller fee-for-service component, multitasking theory suggests specialists who choose it will provide less services (lower quantity), but the services will be of a higher quality. Empirical evidence indicates that specialists who switched from being paid by fee-for-service to the mixed scheme reduced their volume of services by 6.2%, but increased the average time spent with patients by 3.8%. This indicates a substitution from quantity to quality as theory suggests. The welfare implications of the policy change are unclear.

Theory and the empirical evidence suggest that pay for performance initiatives should be viewed with caution if not all aspects of performance can be measured and rewarded. This is because specialists will substitute into tasks that are measurable and rewarded and out of task that are not measurable and not rewarded. In such an environment ‘weak incentives’ such as those provided by salary or a mixed payment schemes can be optimal.

Specialist And Patient Selection

If specialists are remunerated through a prospective payment, they have an incentive to refuse treatment to patients who require many services as these services are costly to provide and these costs are not reimbursed. That is, specialists have an incentive to ‘dump’ high-cost patients. In addition, specialists have an incentive to attract low-cost patients and they do this by overproviding services. That is, specialists have an incentive to ‘cream skim’ low-costs patients.

In addition to specialists providing services to patients, specialists provide patients to hospitals and order hospital services for these patients. If hospitals are paid full-cost reimbursement plus a profit margin per service, then they have an incentive to employ specialists that value patient welfare highly (a high a) as they will order many services for their patients. However, if hospitals are paid a prospective payment, then they have an incentive to employ specialists who do not value patient welfare highly. Therefore, the manner in which hospitals are paid has implications for the type of specialists hospitals employ.

It turns out that the way specialists are paid also has implications for the type of specialists hospitals employ and the type of patients they service. Assume there are two types of patients who differ in their length of stay in hospital. There are two types of hospitals, private and public. Private hospital profit increases with length of stay though at a decreasing rate, that is, the second day a patient stays in hospital generates less profit than the first day. Therefore, the private hospital makes more profit with many short stays than with fewer long stays and so prefers to treat short length of stay patients. The public hospital is benevolent and is indifferent between the type of patients it treats. All specialists value income, but differ in their preference for fairness, where fairness is defined as treating all patients equally regardless of type.

In this setting, if private hospitals pay specialists by fee-for-service and public hospitals pay specialists by salary, then (1) specialists who place a relatively high value on income work in private hospitals and treat short stay patients as this maximizes the number of patients they treat, maximizes their utility, and also maximizes the profit of the private hospital and (2) specialists who place a relatively high value on fairness work in public hospitals as this maximizes their utility and the utility of the benevolent public hospital. As healthy short stay patients are treated in private hospitals, public hospitals treat unhealthy long stay patients. Hospitals have selected the type of specialists that they employ through the payment scheme offered and this payment scheme induces these specialists to select the type of patient the hospital prefers. The result is consistent with the empirical observation that specialist physicians are paid by either salary or fee-for-service, with salary being more common in the public sector than the private sector.

References:

1. Brekke, K. R., Nuscheler, R. and Straume, O. R. (2007). Gatekeeping in health care. Journal of Health Economics 26, 149–170.
2. Chalkley, M. and Malcomson, J. M. (1998a). Contracting for health services with unmonitored quality. Economic Journal 108, 1093–1110.
3. Chalkley, M. and Malcomson, J. M. (1998b). Contracting for health services when patient demand does not reflect quality. Journal of Health Economics 17, 1–19.
4. Chalkley, J. M. and Malcomson M. (2000). Government purchasing of health services. In Culyer, A. J. and Newhouse, J. P. (eds.) Handbook of health economics, vol. 1, ch. 15, pp. 847–890. Amsterdam: Elsevier.
5. Cooper, R. A. and Aiken, L. H. (2001). Human inputs: The healthcare workforce and medical markets. Journal of Health Politics, Policy and Law 26, 925–938.
6. Dumont, E., Fortin, B., Jacquemet, N. and Shearer, B. (2008). Physicians’ multitasking and incentives: Empirical evidence from a natural experiment. Journal of Health Economics 27, 1436–1450.
7. Ellis, R. P. (1998). Creaming, skimping, and dumping: Provider competition on the intensive and extensive margins. Journal of Health Economics 17, 537–555.
8. Ellis, R. P. and McGuire, G. (1986). Provider behaviour under prospective reimbursement. Journal of Health Economics 5, 129–151.
9. Gagne, R. and Leger, P. T. (2005). Determinants of physicians’ decisions to specialize. Health Economics 14, 721–735.
10. Gawande, A. (2002). Complications: A surgeon’s notes on an imperfect science. New York: Henry Holt and Company.
11. Holmstrom, B. and Milgrom, P. (1991). Multitask principal-agent analyses: Incentive contracts, asset ownership, and job design. Journal of Law, Economics, and Organization 7, 24–52, (Special Issue, January).
12. Jack, W. (2005). Purchasing health care services from providers with unknown altruism. Journal of Health Economics 24, 73–93.
13. Ma, C. -T. A. (1994). Health care payment systems: Cost and quality incentives. Journal of Economics and Management Strategy 3, 93–112.
14. Marinosos, B. G. and Jelovac, I. (2003). GPs’ payment contracts and their referral practice. Journal of Health Economics 22, 617–635.
15. Newhouse, J. P. (2002). Pricing the priceless: A health care conundrum. Cambridge, MA: MIT Press.
16. Nicholson, S. (2002). Physician specialty choice under uncertainty. Journal of Law and Economics 20, 816–847.
17. Sage, W. S. (2002). Putting the patient in patient safety. Journal of the American Medical Association 287, 3003–3005.
18. Simoens, S. and Giuffrida, A. (2004). The impact of physician payment methods on raising the efficiency of the healthcare system. Applied Health Economics and Health Policy 3, 39–46.
19. Shafrin, J. (2010). Operating on commission: Analyzing how physicians financial incentives affect surgery rates. Health Economics 19, 562–580.
20. Sloan, F. A., Mergenhagen, P. M., Burfield, W. B., Bovbjerg, R. R. and Hassan, M. (1989). Medical malpractice experience of physicians: Predictable or haphazard? Journal of the American Medical Association 262(23), 3291–3297.
21. Thornton, J. (2000). Physician choice of medical specialty: Do economic incentives matter? Applied Economics 32, 1419–1428.
22. Weiler, C., Hiatt, H., Newhouse, P., et al. (1993). A measure of malpractice: Medical injury, malpractice litigation, and patient compensation. Cambridge: Harvard University Press.
23. Wright, J. (2007). Specialist payment schemes and patient selection in private and public hospitals. Journal of Health Economics 26, 1014–1026.