Physician Compensation

Does an increase in family physicians (FPs) increase individual health? A naive research might believe that we could uncover whether or not this was true by comparing the average health levels of areas with lots of doctors with the average health levels of areas with few doctors. However, doctors often work where there are lots of sick people. Thus we can find a spurious correlation between FP quantity and individual health level.

Gravelle, Morris and Sutton (HSR 2008) aim to find the true relationship between FP quantity and individual health. They use data from the Health Survey for England and the General Medical Services (GMS) Statistics database. The authors note that family physicians may be attracted to areas where healthy patients live if these areas have attractive amenities not captured in the data. On the other hand, FPs might be attracted to areas where lots of sick people live since FP compensation is based on an age-related capitation payment. The authors claim that the following 2 instruments will be correlated with physician supply but uncorrelated with individual patient health: semidetached house prices and age-related capitation payments.

In the first stage, Gravelle and co-authors find that a larger age-related capitation payment attracts more FPs while increased house price decreases FP supply. The authors conclude the following:

“FP supply is positively associated with individual self-assessed health. If no allowance is made for the endogeneity of FP supply, the effect is not statistically significant…When FP supply is instrumented by age-related capitation, a 1 SD increase in FP supply increases the probability of reporting very good health by 4.1 percent. The results are robust to transformations of the health variable and to alternative specifications of the effects of individual age on health.”

• Gravelle, Morris and Sutton (2008) “Are Family Physicians Good for You? Endogenous Doctor Supply and Individual Health” HSR, 43(4) 1128-1144.

Appendix

In the paper, the authors use the ln(FP supply) as the key dependent variable. However, ln(FP supply) is an estimated figure. Thus, when the authors calculate the standard errors, they use a bootstrap methodology as follows:

1. Draw a sample of 351areas (local authorities) with replacement from the area-level data set.
2. Estimate the ln(FP supply) equations for each of the 3 years—1997, 1998, and 1999.
3. Estimate the individual-level health equation using predicted ln(FP supply) plus the individual and area-level covariates. The individual observations are weighted by the number of times the LA in which an individual lives was drawn in the first stage bootstrap sample.

The reported standard error of the instrumented ln FP supply coefficient is the standard deviation of the estimated coefficients on ln FP supply from 200 replications of this procedure.

In the N.Y. Times, Sandeep Jauhar discuss the problems with P4P.  

• Hat Tip: Cato-at-Liberty.

Yesterday I wrote about the problems with fragmented medical care in America.  Is a single payer system the only solution?  A Commonwealth Fund report shows that the single payer system is not the only path towards improved, more integrated care.

What we want

The report outlines six general improvements that need to be made to improve the quality of care in the U.S.:

1. Patients’ clinically relevant information is available to all providers at the point of care and to patients through electronic health record systems.
2. Patient care is coordinated among multiple providers, and transitions across care settings are actively managed.
3. Providers both within and across settings have accountability to each other, review each other’s work, and collaborate to reliably deliver high-quality, high-value care.
4. Patients have easy access to appropriate care and information, including after hours.
5. There is clear accountability for the total care of patients.
6. The system is continuously innovating and learning in order to improve the quality, value, and patients’ experiences of health care delivery.

These are high ideals that need to be translated into specific action-items for providers.  However, a variety of different organizational structures have been able to accomplish each of these six goals.

Models

There are four general types of structures and all can be successful in delivering high quality care.

• Integrated delivery systems such as Kaiser Permanente and Geisinger Health System.
• Large multi-speciality groups such as the Mayo Clinic and Partners Healthcare.  Both groups are nonprofits.  While the Mayo Clinic directly employs doctors on a salaried basis, Partners contracts with over 1000 PCPs and 3500 specialists to provide high quality care to patients.
• Private networks of independent providers, such as Hill Physicians Medical Group and Northland Health Alliance, generally receive a capiation payment from insurers for each practice but doctors are compensated on a FFS basis.  This is similar to RVU compenation.
• Government-facilitated networks of independent providers.  Here, the government does not directly provide care, but instead coordinates care between providers.  This has worked will for Medicaid patients in North Carolina. Denmark’s universal health care system coordinates physicians, who are paid via fee-for-service plus a fee for serving as the patients medical home.  Ninety-eight percent of physicians have paperless offices, prescriptions and lab tests.

Shih et al. (2008) “Organizing the U.S. Health Care Delivery System for High Performance“, Commonwealth Fund Report  no. 1155.

The U.S. healthcare system is one of the more fragmented systems in the world. Traditionally, economists believe that a splash of decentralized planning with a heap of free markets is a recipe for efficient outcomes. In the case of health care coordination, however, information sharing, and collaborative work are needed if quality is to improve and decentralization may not be the best option. Cebul et al. (2008) describe some of the problems with America’s fragmented system. For instance:

• Health insurance is a high turnover product. About one-fifth of health insurance policyholders cancel their plans in any given year. Most of these changes are due to i) employees switching jobs and ii) employers cancelling their group plans in favor of other plans. When insurers have short term relationships with their customers, it likely does not pay for them to invest in preventive care or chronic disease management programs.
• Having a fragmented insurance market can give insurers an incentive to lower quality. When adverse selection is present, offering high quality medical care will attract sicker individuals which will drive up insurance premiums. Thus, insurers often do not have an incentive to provide high quality care.
• The fragmented insurance system means that hospitals must spend more money paying administrators to collect claims. Woolhander et al. (2003) finds that hospitals in the U.S. spend $315 per capita on administration compared with $103 in Canada.
• The fact that physicians are rarely employed by the hospitals has lead to some perverse behavior by nurses. For instance at Stanford Hospital, “Nurses were harshly blamed by surgeons for instrumentation failures, but nurses who delivered clean instruments on time achieved ’star status’ among surgeons. In this setting, some operating room staff shared instruments between surgical suites. Some nurses kept critical instruments in their personal lockers. Some surgeons also took instruments with them when they left the hospital.”
• Further, physician heterogeneity hurts efficiency by making standard operating procedures nearly impossible to implement. Generally, hospitals allow doctor to gets what they want in order to attract physicians with large patient bases to their hospital. However, this creates an incredible amount of complexity and possibility for error in the health care system.
• When providers do consolidate, it is often not done in the best interest of the patient. While vertical integration could improve quality, consolidation is often done with the purpose of locking-in profitable referrals or increasing bargaining power.
• “[Medicare] patients with diabetes see a median of eight physicians in five distinct medical practices.”

In future posts, I will give some examples of organizations that have been able to overcome these problems, as well as policy prescriptions to improve the health of America’s medical system.

See also: Fragmented Medical Care II (The Models) and III (Policy Options).

Cebul RD, Reibitzer JB, Taylor LJ, Votruba M (2008) “Organizational Fragmentation and Care Quality in the U.S. Health Care System” NBER WP 14212.

The New York Times reports (”Doctor and Patient, Now at Odds“) that the doctor-patient relationship is suffering. Patients no longer place absolute trust in their doctor for a variety of reasons. On the physician side, patients know that doctors are pressured by insurance companies reimbursement mechanisms to have shorter office visits. Reports from the media on medical errors and physician kickbacks from drug companies are not helping either. On the patient side, the internet has opened up a vast new store of easily accessible information on diseases and communication with other patients who have the same disease. The prescience of Dr. Rich’s foretold of “the loss of this doctor-patient relationship” in his book, Fixing American Healthcare.

Hat tip: Dr. Jay Parkinson

According to U.S. Census projections, the number of individuals 65 and older will increase from 12.4% of the populations in 2000 to 20.7% of the population in 2050. With the expected incredible rise in the number of elderly in the U.S., one would expect a concurrent rise in the number of geriatricians.

NPR’s Marketplace, however, reports that there are too few geriatricians. Currently there are only 7000 geriatricians, a 22 percent decline eight years ago. Why isn’t the increased demand for geriatricians–due to the aging popualation–causing more medical student to choose to specialize in geriatircs?

Brandeis professor Dr. Stuart Altman reveals the reason: the insurance system is biased against doctors–like geriatricians–who concentrate on preventive medicine. “The way the health system pays the workers in it, it has a very strong bias in favor of high-tech services, highly specialized services and primarily services for acute care.” Time-intestive, non-procedure based primary care is not as highly compensated. In fact, geriatricians make a quarter to a third less than other specialists.

Some physicians who focus on primary care, however, are discovering that concierge practices give physicians more time to spend with patients and can be more lucrative as well. In my hometown of Milwaukee, Froedtert is opening its own concierge clinic.

Primary care physicians can be compensated in a number of ways. The most popular are capitation, fee-for-service, salary, or some mixture of the three. But how does the physician compensation method affect care levels? This is the question Gosden et al. (2000) try to answer in their Cochrane review. The authors search the literature for randomized trials or controlled before and after studies in order to see how changing physician compensation affects the quantity and quality of care.

A summary of the 4 papers which met Gosden et al.’s criteria is below.

The original four articles:

• Stephen M. Davidson, Larry M. Manheim, Mina M. Hohlen, Stephen M. Werner, Beth K. Yudkowsky, and Gretchen V. Fleming, “Prepayment with Office-Based Physicians in Publicly Funded Programs: Results from the Children’s Medicaid Program” PEDIATRICS Vol. 89 No. 4 April 1992, pp. 761-767.
• Gerald B. Hickson, William A. Altemeier, and James M. Perrin “Physician Reimbursement by Salary or Fee-for-Service: Effect on Physician Practice Behavior in a Randomized Prospective Study ” PEDIATRICS Vol. 80 No. 3 September 1987, pp. 344-350.
• B. Hutchison, S. Birch, J. Hurley, J. Lomas and F. Stratford-Devai (1996) “Do physician-payment mechanisms affect hospital utilization? A study of Health Service Organizations in Ontario” Canadian Medical Association Journal, Vol 154, Issue 5 653-661.
• A Krasnik, P P Groenewegen, P A Pedersen, P von Scholten, G Mooney, A Gottschau, H A Flierman, and M T Damsgaard “Changing remuneration systems: effects on activity in general practice.” BMJ. 1990 June 30; 300(6741): 1698–1701.

The controversy as to how much Medicare should pay doctors is continuing to brew (see N.Y. Times article).  Congress passed a law overriding a pay cut to Medicare doctors.  Although the president vetoed the bill, Congress garnered enough support to override the veto.

Dr. Rich of The Covert Rationing Blog claims that the Medicare reimbursement mechanism “is so fundamentally ridiculous that it can only be understood by recognizing that it is a fairly typical bureaucratic attempt to covertly ration healthcare.”

How would you enjoy having legislators determine your salary?

When you are sick and need a doctor, you need hope that you are given the best care possible. Most people assume that doctors will tailor their treatments to the individual patient needs. However, a paper by Frank and Zeckhauser (JHE 2007) explain that this may not be the case. The authors claim that there are four costs which may preclude physicians customizing treatments to individual patients.

• Communication costs: Whenever a physician prescribes a treatment outside of standard protocol, they will have to explain why they are doing this to the patient and this takes time. With patients armed with more information from direct-to-consumer advertising and the internet, communication costs have increased over time.
• Cognition costs: The authors claim that using brain power (cognition) has costs and there may be increasing marginal costs of cognition use. Thus, physicians may use heuristics to simplify the decision-making process.
• Coordination costs: As more and more physicians specialize, communication between physicians is increasingly important. Using standardized, less customized medical treatments makes communication between physicians regarding patient treatment much easier.
• Capability costs: Some doctors are trained to perform certain techniques. If a superior technique is developed, the physician may still decide to use the “old” technique since they have mastered the “old” technique and do not know how to preform the new, superior technique.

It is likely that customization of treatment varies significantly by treatment. For instance, in my “Operating on Commission” paper I find significant differences in surgery rates based on how physicians are compensated by insurance companies. Since this is a significant medical and financial decision by the doctor, one would expect there to be more customization than in other areas, since the benefits to surgery are so large relative to the costs outlined above.

The authors ennumerate how customization will vary accross patients as follows:

1. little is known about a patient and their responsiveness to various treatments
2. treatment is expected to be short-lived
3. there is little difference in the impact of different treatments on patients

On the other hand, Grand and Zeckhauser look at whether or not there is “norm-following behavior” in the length of office visits and physician prescribing behavior. They use data from the 2004 NAMCS and the Quality Improvement in Depression study. They find that physicians do customize treatment more for chronically ill patients than for patients with acute illnesses. Physicians do tend to spend more time in office visits with new patients, but the time spent with the patients does not vary by illness type or severity. Thus, the administrative and communication costs that new patients impose and not medical necessity seem to be dictating how the length of a visit varies. These results are similar to the ones found in Glied and Zivin (2002).

Thus, the authors conclude that some customization of prescribing practices and prescribing behavior does occur, but this behavior is not based on clinical factors.

• Frank RG, Zeckhauser RJ, (2007) “Custom-made versus ready-to-wear treatments: Behavioral propensities in physicians’ choices” Journal of Health Economics, v 26(6): 1101-1127.

Many studies have attempted to determine how the manner in which physicians are compensated by health insurance companies affects the quantity of medical care provided. Today I will summarize some seminal studies in this field.

Epstein, Begg and McNeil (NEJM 1986)

In this study, the authors examine whether or not there is a difference in the rate of ambulatory testing between physicians in a large fee-for-service (FFS) group and a large prepaid group. The physicians were internists who were caring for patients with uncomplicated hypertension. The authors found that 50% more electrocadriograms (EKGs) were obtained in the FFS group and 40% more chest radiographs were obtained in the FFS group [compared to the prepaid group]. There did not seem to be any difference in testing rates between FFS and prepaid doctors for blood counts and urinalyses. This is not surprising, however, because the profit physicians earned from EKGs and chest radiographs was significantly higher than the profit they made from each blood count or urinalysis. Further, the cost of preforming EKGs and chest radiographs was much higher than the cost to preform blood counts and urinalyses.

Thus, this study concludes that patient testing rates are different between FFS and prepaid doctors, but this effect is only observed for high cost and/or high margin ambulatory tests.

Newhouse and Marquis (JHR 1978)

Physicians may treat patients differently based on how the patient’s insurance company pays them. However, the patient base of most physicians includes a wide variety of health plans and thus it may be difficult to discriminate care levels by insurance type. The “norms hypothesis” predicts that physicians treat patients in accordance with the average or modal insurance coverage in a given metropolitan area. In other words, “the level of a community’s insurance coverage determines physician norms.” If this were true, it would imply that there would be significant variation in medical care quantities across geographic regions but very little variations by patient for each physician.

The authors use data from the RAND Health Insurance Study in Dayton, Ohio. The authors test whether or not the patients own insurance rate will affect hospital admissions, the length of a hospital stay or the number of physician office visits. The authors find no evidence that community-level coinsurance rate impact hospital admissions, while the patient’s own coinsurance rate significantly affects hospital admission. In the 1963 data, neither the community insurance variable nor individual insurance variable had any affect on the length of a hospital stay or the number of physician visits, however in the 1970 data, individual coinsurance rates had a large impact on the length of a hospital stay or the number of physician visits, while the community level coinsurance rate had no impact on these dependent variables.

Hellerstein (RAND J Econ 1998)

Most health policy wonks believe we could significantly reduce health care costs without sacrificing quality by using more generic drugs. Many states have passed “permissive substitution laws” which allow pharmacists to substitute generic drugs in place of name-brand ones unless the physician explicitly instructs them not to do so. Other states use a “two-line” prescription method. With these prescription pads, doctors can sign their name in one of two spots: the first indicates that generic substitution is allowed and the other indicates that the brand name option is medically necessary.

Hellerstein uses data from the 1989 NAMCS, and finds that “30% of the unobserved (residual) variance in the prescription choice is physician-specific, rather than patient specific.” Does an individual’s health insurance influence the physician’s prescribing behavior? This paper finds that an individual’s health insurance does not influence prescribing decisions. However, “conditional on a patient’s insurance status, a patient who switches to a physician with a marginally greater fraction of HMO patients is 10.12% more likely to receive a generically written prescription.”

Summary

Why does the composition of the physician’s patient base seem to matter for Hellerstein (1998), but not for Marquis and Newhouse (1978) or Epstein, Begg and McNeil (NEJM 1986)? The main reason is likely that Hellerstein examines medical care in which physician do not receive any compensation. Physicians receive no more or less revenue from prescribing name brand compared to generic drugs and thus have no incentive to find out how they are being paid by each individual’s insurance company. On the other hand, Marquis and Newhouse found that hospital admissions, the length of the hospital stay, and the number of office visits is affected by an indvidiual’s health insurance variables since this type of medical care is high margin and high cost. Similarly, Epstein, Begg and McNeil (1986) found that how an individual’s insurance company compensates physicians matters for high margin, high cost EKGs and chest radiographs, but does not seem to matter for low margin, low cost blood counts and urinalyses.

• Epstein AM, Begg CB, McNeil BJ. (1986) “The use of ambulatory testing in prepaid and fee-for-service group practices: Relation to perceived profitability.” N Engl J Med. 314(17):1089-94.
• Newhouse JP, Marquis MS. (1978) “The norms hypothesis and the demand for medical care.” J Hum Resour. 13 Suppl:159-82.
• Judith K. Hellerstein (1998) “The Importance of the Physician in the Generic versus Trade-Name Prescription Decision,” The RAND Journal of Economics, Vol. 29, No. 1, pp. 108-136 .

Many patients have an idealized view that physicians customize their treatments for each individual patient.  For instance, do physicians tailor prescription dosage based on individual characteristics and responses over time, or will they simple prescribe the standard dosage?

A paper by Frank and Zeckhuaser (JHE 2007) find that norm-following behavior (rather than patient-by-patient customization) is very prevalent.  The authors claim that there are 4 reasons why a physician would strictly adhere to a professional norm rather than customizing their treatments to maximize the quality of care for each individual patient.

1. Communication costs. In order to prescribe treatment outside of the norm, the physician must communicate their reasons for doing this to the patient.  If patients have preconceived notions of how they want to be treated, physicians may have to expend significant effort to convince the patient that this individualized treatment is the best way to proceed.
2. Cognition costs. As every person knows, exerting mental effort is costly.  Physicians can cut down on the cost of diagnosis by using heuristics.  These shortcuts may not be optimal for every patient, but they generally “do a reasonable job for a broad array of cases” and also cut down on the physicians mental computing costs.
3. Coordination costs.  Physicians often have to work with other physicians.  The more physicians customize their treatment, the more difficult it is to communicate this alteration in care levels with specialists and thus more difficult to coordinate care.
4. Capability costs.  Physicians are trained in certain treatments.  If a new, better treatment comes along, the physician has a choice of either 1) doing the old treatment, 2) learning the new treatment poorly and performing the new treatment, or 3) learning the new treatment well and performing the new treatment.  Choice (3) may be optimal from the patient’s point of view, but for the physician it may involve significant fixed costs involved in acquiring the human capital necessary to preform the new procedure.  If the physician decides not to incur the cost to learn the new technique well, it may in fact be optimal to choose option (1) over option (2) and thus old techniques will persist.

While these four costs will push physicians towards following norms, superior patient outcomes may compel doctors to customize their care.

Results

Frank and Zeckhuaser use data from the 2004 NAMCS to determine whether or not physicians customize the length of the patient visit or prescribing behavior.  The authors find that customization in prescribing behavior occurs most frequently for patients with chronic conditions.  This is likely because altering the “standard” treatment has more benefit for ‘repeat-visit’ patients than those with simply an acute illness.  However, race, gender, number of physician visits and insurance type do not affect prescribing behavior.

Regarding length of the office visit, the most important factor is whether or not the patient is a new patient.  Individuals who were self-pay had shorter visits while those with had Medicare insurance had longer visits, but these results were fairly small in magnitude.  Twenty-eight percent of the differences in the length of an office visit was due to physician specific factors.

Overall, the evidence shows that physicians often follow norms rather than customize care.  Also, it seems that the manner in which physicians are paid has no bearing on how they treat patients.  However, this is likely due to the fact that 1) it is very difficult to customize visit length especially when physicians are dealing with eleven managed care contracts on average [see other evidence in "Time Allocation" post on Tai-Seale et al. (2007) or the "Doctors Behave" post on Glied and Zivin (2002)], and 2) physicians do not receive compensation for pharmaceuticals and thus have no financial incentive to tailor treatment to patients based on their individual insurance.

My “Operating on Commission” paper does find that physicians tailor surgery treatment based on how they are compensated, but this is likely because 1) these are high margin procedures where it is worth the physicians time to find out how they are being paid, and 2) unlike pharmaceuticals, the surgeon is the one who receives the compensation directly.

• Richard G. Frank and Richard J. Zeckhauser (2007) “Custom-made versus ready-to-wear treatments: Behavioral propensities in physicians’ choices” Journal of Health Economics, Volume 26, Issue 6, Pages 1101-1127.
• Tai-Seale M, McGuire TG, Zhang, W (2007) “Time Allocation in Primary Care Office Visits” HSR vol 42(5), pp. 1871-1894.
• Sherry Glied and Joshua Graff Zivin (2002) “How do doctors behave when some (but not all) of their patients are in managed care?” Journal of Health Economics, Volume 21, Issue 2, Pages 337-353.

An Annals of Internal Medicine survey sheds some light on physicians opinions regarding universal health care. Overall 59% of physicians support national health insurance and 32% oppose it. Support for national health insurance increased 10 percentage points since 2002 (49%). Unsurprisingly, surgical subspecialties, anesthesiologists, and radiologists, were the only specialities where more than half of respondents did not support universal health care.

Any economist would not be surprised by these findings. Primary care is not highly compensated now and universal health care would likely not alter this. Further, primary care would likely simplify the world of primary care: there would either be one insurance company (as in the case of government provided care), or it would likely be clearer which treatments would be covered. Further, since there would be no uninsured, the primary care doctors would not have to provide any uncompensated care.

For specialists, however, it is likely that national health insurance will reduce compensation for physicians. Some procedures may not be covered, or will be reimbursed at lower rates. More referral restrictions and likely rationing of care would lead to lower profits for specialists.

Even physicians are divided about whether or not national health insurance is a good idea.

GoozNews has more information on the article.

A recent article in the Journal of Health Economics found that increasing Medicare reimbursement may have no meaningful effect on hospital use or patient outcomes.

“There is widespread concern about the quality of health care in the US, and the effect of provider payments on the quality of care is an important and unsettled issue in this debate. The critical question is whether changes in provider payments affect health. To date there is relatively little research on this question. Here, we present evidence of the effect of plausibly exogenous changes in Medicare reimbursement – caused by geographical reclassification – on hospital staffing (nurses) and patient outcomes. We find that changes in Medicare reimbursement levels of approximately 10% have no meaningful effect on hospital use of resources or patient outcomes.“ 

• R Kaestner, J Guardado, 2008. “Medicare reimbursement, nurse staffing, and patient outcomes” Journal of Health Economics, 27(2): 339-361.

David Williams of the Health Business Blog reviews an article from the Boston Globe (”Immigrants…“)  stating that immigrants reduce the cost of health care.  How can this be with so many immigrants relying on government programs and free clinics to receive their care?

While it is true that immigrants are consumers of medical care, they are also producers as well.  A study of the health care workforce in Massachusetts finds that 40% of pharmacists, 28% of physician assistants, 22% of opticians, 21% of licensed practical and vocational nurses, 17% of dentists and 14% of paramedics are foreign born workers.  Twenty-eight percent of physicians and surgeons are foreign born.

An increased supply of health care workers from foreign countries can help decrease labor costs for medical care.

The N.Y. Times (”…No Rhyme or Reason“) has an interesting essay about how doctors financial incentives pressure them to run too many tests on patients and refer them to too many specialists.

Doctors are usually reimbursed for whatever they bill. As reimbursement rates have declined in recent years, most doctors have adapted by increasing the quantity of services. If you cut the amount of air you take in per breath, the only way to maintain ventilation is to breathe faster.

The Healthcare Economist believes that money matters in medical matters.  My research regarding specialist compensation shows that financial compensation has a huge impact on surgery rates.

Musing on the modern propensity for physicians to overtreat their patients, one hospital executive said:

“The hospital is a great place to be when you are sick…[b]ut I don’t want my mother in here five minutes longer than she needs to be.”

Diarrheal disease is a leading cause of childhood mortality in many developing countries. The best treatment when diarrhea strikes is to give the patient Oral Rehydration Solution (ORS). Who provides better care for this disease, public or private providers?

A paper in Health Economics by Waters, Hatt and Black (2008) looks at data from the Living Standards Measurement Surveys (LSMS). This data set draw observations from Latin American and Caribbean countries.

The first item that Waters and co-authors note is that richer household are more likely to see the private physician. “Each additional quintile of household economic status is associated with an increase of 6.5 percentage points in the probability that a child with diarrhoea is taken to a private provider.”

But do these wealthier households receive better care? According to the authors, “treatments provided in the private sector are manifestly of worse quality than in the public sector. A total of 33.0% of children visiting a public provider received Oral Rehydration Solution, compared to 13.7% of those visiting a private provider. Conversely, children treated by a private provider are more likely to receive drugs, most commonly unnecessary antibiotics.”

Private physicians have an incentive to prescribe the more costly antibiotics because they make more money. The publicly financed providers do not have this incentive. The authors claim that prescribing antibiotics may also augment the provider’s reputation with patients. This would be analogous to an American physician preforming an MRI when it is not necessary. The physician makes money on the MRI, the patient believe they are getting high-tech care that is the best in the world, but the MRI may often be a waste of resources.

Winner: Public Providers

• Hugh R. Waters, Laurel E. Hatt, Robert E. Black (2008) “The role of private providers in treating child diarrhoea in Latin America” Health Economics, 17(1): 21-29.

Doctors are often perceived as benevolent professionals. They are hard-working individuals who extend their largesse by giving away free medical care to those in need. Studies by Cunningham and May (2006) and the American Hospital Association find that doctors provide uncompensated care equal to 6.3% or 5.6% of their cots annually respectively.

A recent Journal of Health Economics article by Gruber and Rodriguez concludes that these figures may be overstated. In fact, the study finds that physicians provide negative amounts of uncompensated care to the uninsured.

How is this possible? While it is true that doctors do give away free care to the uninsured and that many of those without insurance do not pay their bills, the uninsured patient often pay a large portion of the list price whereas those who have insurance receive a negotiated lower price. Thus, the authors find that “the majority of physicians actually make money, on net on their uninsured patients…12-14% of physicians found their uninsured patients patients more than twice as profitable as their insured patients; that is the net payments from the uninsured were more than twice the expected payments from the insured patients.”

Even when the authors ignore the higher list prices the uninsured pay, they still find that only about 1% of total revenues are given away as free care to the uninsured. Much of this amount, however, is due to non-payment by the patients rather than free care given away by the physicians.

Medicine may be a business after all.

• Gruber J, Rodriguez D. 2007. “How much uncompensated care do doctors provide?” Journal of Health Economics, 26(6): 1151-1169.

A paper by Alexander S. Preker and April Harding at the World Bank analyze the roles of the public and private sector in health care.  One of the more interesting portions discusses medical care financing using some examples from ancient history.

“Ideological views on the roles of the state and the private sector belong to a long list of false antitheses in the field of medicine and health care. Since the beginning of written history, the pendulum has swung back and forth between minimalist and heavy-handed state involvement in the health sector.

During antiquity, people used home remedies and private healers when they were ill. Yet, as early as the second millennium B.C., the papyri give fascinating evidence that Imhotep, archetypal physician, priest, and court official in ancient Egypt, introduced a system of publicly provided health care with healers who were paid by the community.

At best, financing, organization, and provision of health care was limited to the royal courts of kings, emperors, and other nobility who might have a physician for their personal use and for their troops at the time of battle. The masses got by with local healers, midwives, natural remedies, apothecaries, and quacks.” 

•  Alexander S. Preker and April Harding. 2000. “The Economics of Public and Private Roles in Health Care: Insights from Institutional Economics and Organizational Theory.” HNP Discussion Paper, World Bank.

Economists and health researchers have generally shown that when doctors are paid on a fee-for-service basis, they will advice the patient to undergo more medical procedures than when the doctor is paid on a capitation or salaried basis (see my own paper: “Operating on Commission“). Which payment method maximizes welfare has not been proven and is likely to vary based on the medical procedure in question.

To add to the confusion, Astrid Selder writes in “Physician Reimbursement and Technology Adoption” that physician reimbursement mechanism may also affect the rate at which new technologies are adopted.

Selder lays out the following simple theoretical structure in which the social planner maximizes the following:

• (1-π)U(Y-P)+πU(Y-P-acm-ε +G(m) -nm)
• s.t.: P + πacm-%pi;R=πpm>=0
• s.t.: R + (p-c)m>=0
• s.t.: m<=B

The first constraints represent the zero-profit conditions for the insurer and the physician respectively and the third is a technology constraint. The probability of falling ill is π. Individuals have income Y, an insurance premium of P, and a coinsurance rate of a to may for the marginal cost, c, of a given level of medical treatment, m. Non monetary treatment costs (e.g.: side effects, time costs) are represented by the parameter n, while the benefits of the treatment are represented by the function G(m). The doctor receives a capitation payment of R and a marginal revenue amount given by p-c. B represents the technology constraint.

It seems obvious, but Selder shows that an increase in monetary medical costs (c) or non-medical costs (n) are welfare destroying. Technological innovations, represented as an increase in B are welfare improving if the technological constraint is binding or have no impact on welfare is the technological constraint is not binding. This results is intuitive: higher costs are bad, technological innovation is good.

This result, however, only holds when there is perfect information. With asymmetrical information, technological innovations may be welfare enhancing or welfare destroying. The welfare enhancing argument is similar to above: better technology leads to better medical care. If there is a moral hazard problem, however, using more basic technologies may actually help to counteract the tendency to over-consume medical care in the presence of insurance. Also, in the presence of moral hazard, technological innovations will cause insurance companies to charge higher premiums in order to cover the additional cost of this care.

How does the technological constraint, B, affect welfare in different physician compensation schemes?

• “If B is a binding constraint and increases in B are welfare enhancing then a fee-for-service system should be used. Increases in B and reductions in [monetary medical costs] c are induced all of which are welfare improvements. Incentives with respect to [non-monetary costs] n cannot be improved upon.
• If B is a binding constraint and increases in B are welfare reducing then a system of cost sharing [e.g.: capitation] should be implemented which induces reductions in B and c and gives no clearcut incentives with respect to n. Incentives with respect to B and c are thus optimal, and incentives with respect to n cannot be improved upon.
• If B is not binding the optimal reimbursement system depends on the demand elasticities with respect to costs.”

In words:

“It has turned out that the state should classify diseases or treatments into subgroups which are reimbursed differently in order to achieve the desired welfare effects in each subgroup. Especially for the case of extremely severe illness shocks the introduction of a fee-for-service system may be socially desirable. It is these very severe diseases where the capitation system seems to induce negative welfare effects with respect to the technologically feasible boundary of treatment. Cost sharing/capitation systems are most suitable for treatments where a reduction of the amount of health care consumed is desirable.”

While the solution Selder proposes is logical, it does not take into account the cost to the state to collect information regarding these disease classes. Lobbying will likely influence whether a given treatment falls into the fee-for-service or cost-sharing grouping. Nevertheless, Selder’s general findings seems reasonable and applicable to the medical care finance design in the real world.

• Selder, Astrid. 2005. “Physician Reimbursement and Technology Adoption” Journal of Health Economics Volume 24, Issue 5, Pages 907-930

“On the biggest shopping weekend of the year, you’ll know the price of the big-screen TV and the Wii you’re about to buy - but you’ll likely be in the dark about the cost of any health care you need.”

Two Wisconsin state senators are trying to change this.  The Milwaukee Journal Sentinel (”…Health cost disclosure“) reports that “Sen. Jim Sullivan (D-Wauwatosa) and Rep. Steve Wieckert (R-Appleton) this week introduced a bill requiring health care providers and insurance companies to make available information about the cost of procedures or services to patients who ask for it.”

Here is one type of government intervention I support: the government is helping to reduce asymmetric information in the medical care market without fixing prices.  The bill would compel providers to state the cost of the 50 most common medical procedures.  These costs would be divided into 3 groups: the usual rate for the service (whatever that means); the rate paid by government programs; and the average rate for health plans.

Milwaukee pediatrician Carl Eisenberg, does note that the cost of a procedure can vary from patient to patient.  Nevertheless, the provider could charge a fee in which for some patients they lost money, but for some “easy” patients they made above average profits. 

There are other problems with the pricing structure.

Larry Rambo, chief executive officer for Humana’s Great Lakes region, said cost information might best come from an insurance company because it can encompass the “episode of care,” meaning all the costs that go into treatment.

“If you’re going in for knee surgery, you’re going to get a bill from the hospital, you’re going to get a bill from the surgeon, you’re going to get a bill from the anesthesiologist and you’re going to get a bill from the radiologist,” Rambo said.

Nevertheless, I am always in favor of more information.  As a supreme court jurist once wrote: “sunlight is the best disinfectant.”

I recently read a Health Affairs article analyzing a pay-for-performance (P4P) demonstration. The Local Initiative Rewarding Results (LIRR) demonstration in California involved seven Medicaid-focused health plans in California between 2003 and 2005. Here are some of my most recent thoughts on P4P:

• The article seemed to show that P4P worked best when there was much consensus over how doctors should treat patients. The irony of P4P is the following: P4P programs with high levels of physician consensus work, but if everyone physician is already acting appropriately P4P does not make a major change in behavior; larger behavioral changes can occur when physicians are greatly deviating from the optimal practice, but in this case it is very difficult to implement P4P.
• P4P plans are often implemented so those that do well gain income and those that preform poorly maintain their status quo income. When no one loses any money, P4P is supposed to gain more acceptance among physicians. In equilibrium, however, an economist would predict that those with low P4P scores will receive lower wages in the future (or wage increases below inflation) due to P4P.
• Unsurprisingly, the authors of the Health Affairs paper found that meager financial incentives do not significantly change physician behavior, while larger financial incentives have a greater impact. Also, good communication with physicians helps to increase physician compliance with P4P mandates.
• If P4P evaluates doctors on care levels which depend on patient compliance, health plans can help physicians to increase the quality of care. For instance in the above study, “many providers used the plan lists of members turning fifteen months old as the basis for outreach…”
• When P4P financial incentives vary across plans, it is difficult for physicians to focus on specific quality indicators due to complex reimbursement schemes in each plan.
• When HEDIS scores improve, some of this improvement is due to better quality while much of it may be attributed to better documentation in a patient’s medical records.

Suzanne Felt-Lisk, Gilbert Gimm and Stephanie Peterson (2007) “Making Pay-For-Performance Work In Medicaid” Health Affairs, v26(4) pp. w516-w527.

Arnold Kling of the EconLog site has some commentary on P4P when discussing Tim Hartford’s latest book (”The Logic of Life“).

I have a very different approach to compensation. I think that the key is to change compensation schemes frequently. The reason is that any scheme can be gamed, and the longer you wait to change any given scheme, the more effectively the participants will have gamed it. That is one reason I think that “Pay for Performance,” the newest miracle cure for health care costs, will fail miserably. The doctors will be able to run circles around the bureaucrats. In the U.K., they already have–all of a sudden, 91 percent of doctors were receiving bonuses for being above average.

I can’t verify this ‘91%’ figure.  Using pre-P4P measure as the measuring stick will lead to many physicians reaching the ‘above average’ mark in the post P4P mark.  Much of this increase is due to better record keeping; some of it is due to better care.

This is how a very interesting article (”P4P is changing me“) in Medical Economics begins. The essay won an award in the 2006 Doctors’ Writing Contest.

The author of the story is a physician who has been under much financial pressure of late; a divorce, med school loans, a mortgage, alimony and child support were all eating away at the author’s income. One day an 84-year old WWII vet walks into his office. He had a left renal mass, but did not want any further medical treatment for this problem. The author must contemplate how–or if–he should treat the patient.

“There’s very little I could do to improve [the patient's] life. I asked him once what I could do to help, and he requested merely that I talk with him. That’s what I’ve done, every three months, but I won’t get any extra pay for that. I could check his A1C, though—after all, he does have diabetes, and it has been more than a year since it’s been checked.“

The patient had been very diligent in maintaining the correct levels of blood pressures, heart rates, and blood sugars. The author believed that an A1C test was clinically unnecessary. But…

We talked for 30 minutes (10 minutes over the scheduled 20-minute appointment); I was then behind. How would I wrap this up? Should I make one last plea that he open himself up to counseling and antidepressants, and not suicide? Or should I tell him that I need him to go to the lab and have his blood work checked? Well, I doubted I could do much to improve his life, but I did need the $50 . . .

Should I start statins on the drooling demented to lower their LDL? Should I preach to paranoid schizophrenics that they must quit smoking? Doing so might help ease my burdens—will it ease theirs? Without a financial incentive, I treated practice guidelines as guidelines, and I treated patients as patients. With financial incentives, will the guidelines become my goal? Will I lose patience for patients who are just a means to my means?

• Mansfield, Richard J. (2007) “P4P is changing me,” Medical Economics, May 4, 2007.

“…purchasers typically reimburse health care providers on the basis of the volume and intensity of the services provided, rather than the quality or cost-effectiveness of those services. The result is a financing system akin to paying academics on the basis of the volume and intensity of footnotes.”

This website has blogged extensively on pay-for-performance schemes (see these articles). But what do other people think of P4P.

Michael Cannon of the Cato Institute gives his take in a 2007 article. Mr. Cannon speaks out against P4P in the Medicare setting since there is little room for experimentation or learning due to Medicare’s sheer size. Mr. Cannon writes:

“The current system of private P4P programs allows insurers and employers to conduct experiments and learn from each other’s successes. Competition to improve the quality of care in a cost-effective manner encourages private purchasers to experiment with P4P, and private control gives purchasers flexibility in designing and altering those experiments. As important, private P4P experiments confine any harmful failures to smaller populations.”

The author constantly mentions that errors which occur in publicly run P4P will harm many people, but avoids mentioning the flip-side that installing a successful P4P program in Medicare can also help the most people. Mr. Cannon’s point of using competition between insurers to allow each to experiment is wise assuming that insurers want the best care for their patients. As Dr. Fogoros notes in his GUTHealtcare website, patients generally do not pay for their health insurance, employers do. And for employers “As long as we don’t hear more than the average number of complaints from our employees, the health coverage we provide is, by definition, good enough.”

Still, one should take Mr. Cannon’s point seriously that at least there is some competition in the private sector health insurance while there is no competition in government run insurance plans. Competition leads to experimentation and experimentation leads to progress a la the Learning Economy model. Further, Medicare already has complex reimbursement rules and adding P4P schemes to the mix may only further increase physicians’ cost to serve Medicare patients. Finally, since Medicare is a political entity, it is inevitable that there will be significant lobbying and rent seeking in order to have Medicare’s P4P serve certain interest groups.

Another novel point the article makes is for insurance companies to focus on patient based P4P. “…Patients who receive recommended care (or who use providers known for delivering recommended care) would face lower out-of-pocket costs, while those who do not would face higher out-of-pocket costs. Patients would know sooner whether a provider was not adhering to the plan’s quality guidelines because that deviation would affect their pocketbooks.”

P4P Tradeoffs

The article also notes some tradeoffs provided by different types of P4P programs:

Problems with P4P

Poorly constructed P4P measures may lead physicians to methods of patient selection (i.e.: treating only healthy patients in order to increase outcome scores) as well as data manipulation. Physicians would never manipulate data…right? According to an article by Bogardus, Geist and Bradley (2004), as many as 50% of physicians admit they have manipulated third-party reimbursement rules to secure coverage of a particular treatment for a patient.

Another problem with P4P that Cannon wisely points out is that “A treatment’s overall beneficial effects may hide different effects on subgroups, including no effect or even harmful effects. For example, patients may respond differently to a given intervention as a result of multiple illnesses or interactions with treatment regimens for such co-morbidities. Financial incentives that encourage providers to treat such outliers according to what benefits the majority of patients may inadvertently encourage low-quality or even harmful care.”

• Cannon, Michael (2007) “Pay-for-Performance: Is Medicare a Good Candidate?,” Yale J. Health P. Law & Ethics, Vol. 7, issue 1: 1-38.

Many times on this blog, I have commented about pay-for-performance (P4P) programs. A healthcare economist, physician, or health services researcher may wonder where they can get data regarding P4P. In truth, getting P4P data is difficult; this is reinforced by the fact that for most physicians, P4P incentives make up a small portion of their salary and thus it may be difficult to detect major behavioral changes empirically.

Nevertheless, the Alliance for Health Reform has written a 2006 issue brief on P4P and cites some of the major programs.

• Agency for Healthcare Research and Quality
• Ambulatory Care Quality Alliance
• Bridges to Excellence
• California Report Card
• Integrated Healthcare Association
• Leapfrog Group
• MedPAC
• National Committee for Quality Assurance
• National Quality Forum
• Premier Hospital Quality Incentive Demonstration
• Robert Wood Johnson Foundation.

The largest P4P program is run by California’s Integrated Healthcare Associates (IHA). A 2006 report (”Advancing quality through collaboration“) chronicles IHA’s progress in working with the California Association of Physician Groups as well as Dr. Stephen Shortell to establish a P4P measures. The report has two very interesting summary tables: one describes how the IHA P4P measures have evolved over time, the other how payment methodologies vary by participating health plan.

Total payments to California physician groups for P4P programs totaled $37.4m in 2003 and $54m in 2004. Of the $54m, $26m went towards P4P clinical measures, $22m went to P4P based on patient experience, and the remaining $6m went for improvements in information technology. “Payments to individual physician groups ranged from 0 to $4.50 pmpm (per member per month). On average, the incentive payment was only about 1.5% of physician group income.

Through these efforts, the report finds that measured quality dimensions are improving but “‘breakthrough improvement’ has not been achieved.” Further, the report believes that much of the quality improvement is likely due to better data collection and record keeping rather than actual changes in care.

Where will IHA move P4P in the future? Recommendations include:

• increasing incentive payments up to 10% of total physician compensation by 2010 (if P4P payments are trivial, then there will likely be no behavioral change)
• incorporating better measures of risk adjustment into capitation payments (this will discourage physicians from refusing care to sicker patients for fear of reducing quality score measures)
• addition of an efficiency domain, including appropriate resource use measures (quality does not mean quality at unlimited cost to the payers)
• avoiding incentivizing physicians to ‘teach to the test’ (physicians may decrease quality care in dimensions not measured by the IHA’s P4P program).

The concept of pay-for-performance has been discussed repeatedly in this blog. But what about the other P4P: pay-for-participation?

In the pay for participation model, payers compensate physicians to add infrastructure or partake in data collection. For instance, a health plan may pay providers to implement an electronic medical records system. Providers may also compensate physicians to set up data collection systems where performance can be measured. One difference between pay for performance and pay for participation is that P4Performance measures are generally distributed to the public whereas P4Participation metrics are kept confidential between the payers and providers. While the lack of openness reduces the information the public has about quality, it also diminishes a hospital’s or physician’s incentive to game the system.

As an example of pay for participation, an article by Birkmeyer and Birkmeyer (NEJM 2006), looks at the case of Blue Cross -Blue Shield of Michigan (BCBSM). The BCBSM program targets cardiac surgery, bariatric surgery and other general and vascular surgery. “[H]ospitals are compensated for data collection and participation in improvement activities in the form of supplements to reimbursements according to the diagnosis-related group for surgical care.” Since the data collected are not publicly report, physicians have less of an incentive to ‘juke the stats.’ For instance, when P4P performance measures are publicly available, surgeons may decide to forgo surgery on the sickest patients in order to decrease their surgery mortality rate, even though it is just these types of patients who likely could benefit most from surgery. With their flaws not open to public scrutiny, physicians may be more open to admit errors or work with the payer to improve the quality of care.

The authors do note that pay for participation is difficult to implement and requires large amounts of organization and coordination. While the payers do cover the cost of implementing the data collection, neither hospitals nor surgeons “[profit] financially from the pay-for-participation program,” which may limit provider enthusiasm for P4Participation efforts.

• Birkmeyer, Nancy; Birkmeyer, John; 2006. “Strategies for improving surgical quality - Should payers reward excellence or effort?” NEJM, 354:8, pp. 864-870.

Pay-for-performance (P4P) is one of the latest hot topics in the health policy world. Yet it has not been conclusively answered whether or not P4P incentives affect the quality of care given. Meredith Rosenthal and Richard Frank review some of the more compelling empirical studies in their April 2006 paper in Medical Care Research & Review.

Hillman et al. (Am J Pub Health 1998). In this study, 52 capitated, non-exclusive primary care practices were randomly assigned to either a control group or treatment groups in order for the authors to determine the affect of financial incentives on cancer screening. The treatment group received a bonus equal to 20% of capitation payments if it was one of the top three performers and a payment of 10% if it was in the fourth, fifth or sixth best performer range. The study detected increased screening in both the control and treatment groups, but difference in screening rates between the control and treatment was negligible.

Hillman et al. (Pediatrics 1999). This was also a randomizd trial, but the focus was on childhood influenza immunization rates. The control group was compared against 2 treatment groups: one which received feedback on their performance and another which received feedback and a bonus payment if sufficient number of children were immunized. This study also found financial incentives had no effect on immunization rates.

• Rosenthal and Frank give a few explanations why Hillman and co-authors found financial incentives to be ineffectual in either of their studies. First, the authors had a small sample size and short time horizon (18 months). Secondly, the financial incentives only pertained to Medicaid capitation patients. If a physician contracted with many non-Medicaid health plans (e.g.: private HMOs, PPOs, etc.), the magnitude of the bonuses would have been small. Finally, the physicians may not have been aware of the program or were unaware of approximately what level of cancer screening/influenza immunization would be needed to receive a bonus.

Kouides et al. (1998). This study had a treatment and control group where the treatment group received between $0.80 to $1.60 per person immunized against influenza. The authors found a statistically significant 7% increase in immunization rates between the treatment and control. Since the two groups were not randomized, it is likely that those physicians who more aggressively immunize patients choose to be in the treatment group.  This selection effect likely biases the results. Further, the authors note that the treatment and control groups differed significantly in terms of practice size and specialty mix.

Fairbrother et al. (1999). This study also looked at the impact of financial incentives and performance feedback on childhood immunization rates. The framework of the study had one control group and three treatment groups. The first treatment group received feedback every 4 months for one year, the second group received feedback and a bonus for receiving a target immunization rate, and the third group received feedback and a bonus for timely immunization. “The rather sizable bonus did improve immunization rates, but this was primarily achieved through better documentation of immunizations children had received outside of the practice, suggesting no real gain in quality of care.”

One problems with the study is that each of the 4 groups (i.e.: control and the 3 treatment groups) only had a sample size of 15 physicians. Further, the 8-month time horizon may not have been sufficient to detect a significant change.

Roski et al. (2003).  This study examined how financial incentives affected smoking cessation care.  Forty clinics were randomized into three groups: 1) a control, 2) a financial incentive-only group, 3) a financial incentive group with a computerized patient registry linked with counseling over the phone.  The authors found that documentation of smoking status and documentation of advice to quit smoking increased significantly compared to the control group.  “Despite improved adherence to guidelines, however, there was no significant impact on smoking cessation rates. In addition, in a somewhat puzzling result, the clinics that were offered both the financial incentive and access to the patient registry and telephonic counseling system showed no improvement relative to the control group.”

Amundson et al. (2003).  This study looks at a study in the HealthPartners system in Minnesota in which physicians received a bonus of over 80% of patients were asked if they smoked and were subsequently counseled to quit smoking (i.e.: ask and advise rates).  The ask and advice rates increased significantly but since there was no control group, one can not determine if the increase is due to the financial incentives or a time trend.

Overall it seems that the studies cited by Rosenthal and Frank do not show that P4P caused large changes in care.  One issue is that most of the financial incentives were small relative to overall physicians pay.  Also, P4P may increase documentation of care–in order that the physicians can receive the financial bonus–but may not actually alter care levels.  It is also possible that the changes in care practices occur over a longer period of time (i.e.: more than 1 or two years), which would not be captured in these shorter term studies.  While I do not doubt that large financial incentives will cause physicians to perform more care on the dimensions measured, P4P may also cause physicians to substitute their effort levels away from un-measured care dimensions which may be equally important to patient health.

One can only conclude that significantly more research is needed to clarify the effects of P4P.

• Rosenthal, Meredith; Frank, Richard; 2006.  What Is the Empirical Basis for Paying for Quality in Health Care? Medical Care Research and Review, Vol. 63, No. 2, 135-157 (2006)
• Hillman, A. L., K. Ripley, N. Goldfarb, I. Nuamah, J. Weiner, and E. Lusk. 1998. Physician financial incentives and feedback: Failure to increase cancer screening in Medicaid managed care. American Journal of Public Health 88 (11): 1699-1701.[Abstract/Free Full Text]
• Hillman, A. L., K. Ripley, N. Goldfarb, J. Weiner, I. Nuamah, and E Lusk. 1999. The use of physician financial incentives and feedback to improve pediatric preventive care in Medicaid managed care. Pediatrics 104 (4): 931-935.[Abstract/Free Full Text]
• Kouides, R. W., N. M. Bennett, B. Lewis, and J. D. Cappuccio. 1998. Performance-based physician reimbursement and influenza immunization rates in the elderly. The primary-care physicians of Monroe County. American Journal of Preventive Medicine14 (2): 89-95.[Medline]
• Fairbrother, G., K. L. Hanson, S. Friedman, and G. C. Butts. 1999. The impact of physician bonuses, enhanced fees, and feedback on childhood immunization coverage rates. American Journal of Public Health 89 (2): 171-175.[Abstract/Free Full Text]
• Roski, J., R. Jeddeloh, L. An, H. Lando, P. Hannan, C. Hall, and S. H. Zhu. 2003. The impact of financial incentives and a patient registry on preventive care quality: Increasing provider adherence to evidence-based smoking cessation practice guidelines. Preventive Medicine 36 (3): 291-299.[Medline]
• Amundson, G., L. I. Solberg, M. Reed, E. M. Martini, and R. Carlson. 2003. Paying for quality improvement: Compliance with tobacco cessation guidelines. Joint Commission Journal on Quality and Safety 29 (2): 59-65.[Medline]

In a recent NBER working paper (”Tradeoffs“), authors Christopher Afendulis and Daniel Kessler, pose an interesting question: should the physician who is diagnosing you also be the one who provides treatment? On the one hand, a physician who both diagnoses and provides treatment has a financial incentive to recommend to the patient that they should have more aggressive (i.e.: read ‘expensive’) treatment. Similarly, car mechanics both diagnose the car’s problem and sell the appropriate remedies (i.e.: parts and labor). For this reason, car mechanics also are often stereotyped as recommending services you don’t need.

On the other hand, having the same physician (or mechanic) provide diagnosis and the treatment, may create efficiency gains. For example “the diagnostician may
have better information about how to treat the problem than he could (or would) provide
to an independent third party. Or, the diagnostician may be able to treat the problem
himself less expensively or more effectively (’half the cost is opening the engine block’).”

Data

To test how the diagnosis/treatment dynamic works in health care, Afendulis and Kessler use 1998-2000 data on Medicare patients with coronary artery disease. They compare type of treatment, spending and outcome measures when the patients were treated by the following 3 types of doctors:

• ‘Non-integrated’ Cardiologists - these physicians only diagnose the disease and provide non-surgical treatment (e.g.: pharmaceuticals). They do not conduct angioplasties.
• ‘Integrated’ Cardiologist - cardiologists who both diagnose and are able to preform surgical procedures (e.g.: angioplasties).
• Cardiac surgeons - unlike cardiologists, cardiac surgeons do not provide catheterization services for diagnostic purposes. These physicians preform the more complex heart surgeries (e.g.: bypass surgery).

Econometrics

The authors wish to identify how the type of physician impacts treatment types. The treatment types are angioplasty, bypass surgery, and drugs/non-surgical care. The authors use a multinomial logit framework but worry about selection bias. Selection bias would occur if there were “…unobserved differences in the health or preferences of patients treated by an interventional versus a non-interventional cardiologist.” To control for this, the authors employ predicted values of both physician type and hospital characteristics based on a patient’s three-digit-zip-code average rather than the patient’s actual choice of provider and hospital. This should increase the accuracy of the estimates, but the precision likely suffers.

Results

The authors conclude that treatment by an integrated cardiologists (rather than a non-integrated cardiologist) increases spending by $2800 per patient, but leads to no statistically significant change in health outcomes. However this is not the entire story. Some patients who would have not received surgery now receive angioplasties. On the other hand, some patients who would have been referred by the non-interventional to the surgeon for the more invasive, expensive bypass surgery, now have the less expensive, less invasive angioplasty preformed by the interventional cardiologist.

Looking at efficiency measures, diagnosis by an interventional cardiologist leads to higher spending on angioplasty patients with no health improvement; diagnosis by an interventional cardiologist leads to higher spending on bypass patients as well, but mortality drops significantly. Non surgical patients treated by the interventional cardiologist have similar spending levels, but higher mortality.

Conclusion

After analyzing the data, the authors make an interesting point regarding ‘kickback’ payments for referrals.

“Explicit ‘kickback’ payments from treating to diagnosing doctors are banned by law (for public purchasers such as Medicare and Medicaid) and by contract (for private purchasers like insurance companies and large employers). However, the principle underlying this ban is not generally applied to doctors’ decision to provide integrated diagnosis and treatment, even though integration can have the same effects on incentives and behavior as kickbacks do. In addition, allowing integration but banning kickbacks effectively allows rent capture by integrated but not non-integrated doctors, which can distort treatment decisions even further.”

• Afendulis, Christopher; Kessler, Daniel (2006) “Tradeoffs from Integrating Diagnosis and Treatment in Markets for Health Care” NBER Working Paper 12623.

An interesting post by Arnold Kling (”Doctors, Pharmaceuticals, and Statisticians“) reports on a randomized clinical trial which demonstrated that on average, angioplasties have no incremental health benefits once the patient is placed on multiple medications such as beta-blockers, ACE inhibitors, statins and blood thinners.

Dr. Kling writes: “Doctors think that they add value by giving advice on issues such as angioplasty. But the advice of statisticians may be better.  Doctors also think that drug companies earn too much money. But it may be the doctors who earn too much money.  I predict a collision between doctors and statisticians somewhere d