Central Planning vs. Medical Knowledge

Why Medicare’s Latest Idea, Accountable Care Organizations, Will Not Work

Accountable Care Organizations (ACO’s) are Medicare partnerships set to start in January 2012 in which relatively large number of local physicians and other local providers of healthcare services will team with local hospitals and Medicare and will attempt to improve quality of care and reduce costs.  The model calls for incentive payments by Medicare based on numeric indicators on quality, clinical processes, patient satisfaction, utilization, costs and outcomes. Medicare would assign a minimum of 5,000 local patients to an ACO and will track these indicators over time. Medicare will develop a baseline consisting of the prior 3 years’ cost information of those patients assigned to an ACO and will adjust it for risk, that is, how sick its patients are. Medicare will also develop a cost reduction target for the population.  An ACO’s performance will then be compared to cost reduction targets and if the yearly per-capita beneficiary costs are below the target specified by Medicare, then the ACO will be entitled to a share of the savings which will be distributed to the members of the ACO. 

ACO’s are a response to the regional variations in Medicare expenditures per enrollee documented by the Dartmouth Atlas group, variations that range from $6,000 (Dubuque, Iowa) to $17,000 (Miami, Florida). Dartmouth Atlas finds that these variations in expenditures are caused by differences in the amount of care given to the population and that the amount of care depends on the supply of medical services available in a particular region. These variations are supported by a Medicare payment system that keeps the supply of medical services fully employed regardless of their capital intensity or their cost to Medicare.  The higher the proportion of hospital beds and specialists in a region, the higher the number of hospital admissions and referrals to specialists and the higher the Medicare total payments in a region. Simply put, the higher the supply, the higher the demand.  

The hypothesis behind ACO’s is that if spending could be safely reduced to the level of the low spending regions, overall savings for the US healthcare system would be in the order of 30%, and that since according to the Dartmouth Atlas data large clinically integrated health systems are able to provide care at lower costs, then ACO’s that emulate these large healthcare systems could serve as the vehicle for provider integration in regions that lack such large scale integration and, if the proper financial incentives from Medicare payment systems are in place, ACO’s will form and will achieve those savings. 

The policy implications of ACO’s are clear: to reduce costs, Medicare will need large organizations of physicians and hospitals with more primary care physicians, fewer specialists, fewer hospitals, fewer hospital beds, fewer capital intensive diagnostic and treatment technologies, primary care physicians who do not refer to specialists and finally, specialists who do not use hospitals or capital intensive equipment.

But Medicare’s and Dartmouth Atlas’ logic is based two intellectual mistakes. Their first mistake is to assume that Medicare’s current payment methods (fee for service) reinforce an already poorly coordinated and expensive medical care and that supply causes demand. The cause and effect relation moves exactly in the opposite direction: we have poorly coordinated and expensive medical care because Medicare’s payment method interferes with the formation of free market prices and we have higher supply of services because under the current sytem, we have an unlimited unchecked demand for services. As Ludwig von Mises pointed out, without prices economic calculation is impossible so managers and entrepreneurs are unable to allocate capital, labor and supplies to the highest valued use and without the constraints on cost to themselves, consumers of healthcare face no limit to their demand. We have explored this argument here. 

Their second intellectual mistake is to assume that the healthcare system can be made more efficient by encouraging and forcing integration. This mistake is caused by Medicare's failure to understand how complex societies transmit knowledge and information. F.A. Hayek answered this question in 1945 in his essay The Use of Knowledge in Society. Sociologist Charles Bosk supports Hayek’s answer in his classic 1974 field study of residency medical education titled Forgive and Remember. The rest of this article follows the logic behind these two works and applies them to the healthcare reform debate. 

The knowledge necessary to improve coordination and reduce costs in our healthcare system is never and can never be concentrated in a central planning organ, be it at the national level with Medicare, or at the local level, with an ACO. To the contrary, medical knowledge is always dispersed in incomplete and contradictory form scattered among million of patients and thousands of health care providers. Medical knowledge is only made present in the individual interaction between a particular patient and a particular health care provider in a particular physical space and time. The problem that Medicare is trying to solve is not what is the best way to allocate scarce healthcare resources in general. Medicare’s problem is infinitely more complex; Medicare is attempting to find out how to allocate scarce health resources when only the individual patient and his physician know the relative importance that those scarce resources have when compared to other alternatives.  

If Medicare could concentrate all the knowledge which is initially dispersed and then convey that knowledge back to individual patients and their physicians so they can make the right decisions, then ACO’s (and central planning indeed) could be successful. But this depends on the type of knowledge that needs to be communicated. If what Medicare is communicating is scientific knowledge, perhaps a body of experts could be assembled to decide on the right diagnostic and treatment protocols that ought to be used--though Medicare would then have to address the obvious questions of who selects the experts, what criteria is to be used for that selection and how those protocols will be enforced across space and time. Indeed Congress has already allocated $10 billion to Medicare to establish a center for innovation in January of 2011 charged with develop innovative models to slow growth and improve quality. 

But Hayek points out that scientific knowledge, is not comprehensive knowledge.  There is knowledge that is unorganized rather than scientific, particular rather than general, bound by time rather than timeless and bound by location and circumstance rather than universal. This is the knowledge that a patient has about his specific symptoms, his history, his physical condition, his emotional state, his subjective level of pain, his spiritual inclination, etc., that can only be voluntarily disclosed if it is to be used by his physician to diagnose and treat the condition that is causing the symptoms. It is also the knowledge that a hospital or medical practice manager has:  knowledge about relative surplus, scarcity and quality of supplies, weather or other conditions that may interrupt the flow supplies, labor and patients, best locations to rent physician offices or build hospitals, how and when to consolidate or expand services or nursing units within a hospital, how to cut costs in this particular department at this particular time without sacrificing quality, when to improve quality even if costs increase, key stakeholder analysis, accountability systems, etc.  

This type of practical knowledge is also characteristic of the clinical knowledge that an experienced physician has. In his field study of medical education Forgive and Remember, sociologist Charles Bosk follows a team of medical residents and attending physicians over the course of 18 months to observe how medical knowledge is transmitted. Specifically, he concentrates on the nature and types of medical errors. He points out that a misunderstanding about two types of medical errors (a reasonable treatment option that is later proven wrong versus a treatment option that is an indefensible mistake) and how physicians make the distinction between the two, leads to incorrect policy recommendations about how the medical profession ought to be controlled. He argues that medical errors are an indeterminate category, similar to beauty or mercy (p.24). This indeterminateness is caused by the dual nature of medical knowledge itself: on the one hand, there is scientific knowledge (or “general knowledge” in Hayek’s terminology) and on the other, there is clinical expertise (or “local knowledge” in both Hayek’s and Bosk’s terminology). Bosk found that when physicians make clinical decisions,  “arguments based on clinical expertise override those based on scientific evidence” (p.85). Local medical knowledge, therefore,  trumps general knowledge, exactly the same conclusion that Hayek reaches. 

Residency programs for medical students are apprenticeships in which an experienced attending physician nurtures and develops through a series of almost choreographed rituals, humor, unwritten rules, stereotypes, hierarchical structure, tension, drama, survival-of-the-fittest character, public display of student’s and attanding’s mistakes and accomplishments, compassion, self-deprecation, team effort, individualism, camaraderie, horror stories, etc., in the inexperienced resident the practical judgment to know when to follow scientific knowledge and when to follow clinical experience. This is the most critical element of medical education.  Residents are expected to study and know the general rules. Attending physicians are expected to teach residents to recognize the exceptions to those general rules. That clinical acumen or clinical eye characteristic of an experienced attending physician, his clinical expertise, “is a charismatic possession, a gift of grace; its exact nature is a mystery“ (p.92). Medical education has evolved over hundreds of years and is structured today as a way for medical residents to appreciate that mystery.  

Bosk points out that the assumption by managed care, and I would add Medicare, Dartmouth Atlas and all medical central planners, that the healthcare system can be made more efficient through centralized decision making far from the bedside clashes with the very system that  clinicians use to grapple with medical errors, understand their cause and implement remedies, and teach future physicians, all part of a process that is “an extreme example of local knowledge” (Bosk’s own terminology in p. xxiii). Bosk concludes that medical “errors are not events that can be counted. Their existence needs to be debated; the discourse over precisely what is and is not an error is necessary to the formation of a sense of professional responsibility.” 

The patient’s knowledge about his condition, the manager’s knowledge about the local conditions of the market, the physician’s clinical knowledge about exceptions to the rule are all knowledge that cannot be counted, aggregated and expressed in statistical format. It cannot be conveyed to Medicare planners. Aggregation destroys the minor differences that are the “bread and butter” of the day to day actions of physicians and healthcare managers at the local level. And even if it were possible to aggregate, this knowledge could not be disaggregated and transmitted back to individual patients and providers of care in time to have any useful meaning to guide action at the local level, especially given the rapid rate with which change at a local level takes place.  

Bosk doesn’t offer any solutions to the quandary facing healthcare central planners, but his study reveals what is the result of centuries of development in the medical education field: an apprenticeship environment at the local level can best impart the practical, local and non-theoretical clinical knowledge necessary for medical students to become effective clinicians. This supports Hayek’s insight that only by decentralizing can coordination be improved.

The challenge faced by Medicare is how to allow decentralization and at the same time impart the additional knowledge to local patients, local physicians and local hospitals that will allow them to make decisions in a way that the overall objectives of Medicare can be met. In this regard, Medicare faces not just a difficult problem, but an impossible task. Almost anything that happens in the world can and does have an effect on the decisions made at the local level. 

The paradox of simultaneously having to decentralize and coordinate in order to achieve common general objectives is solved by the market price system. Hayek states that when free market prices are allowed to function, they coordinate the separate actions of different people in the same way that an individual’s subjective values assist him in coordinating the parts of his own plan, without having to know all the details of the problem to be solved or having to develop all the ramifications of each option. He concludes:  “The mere fact that there is one price for any commodity...brings about the solution which might have been arrived at by one single mind possessing all the information which is in fact dispersed among all the people involved in the process.”

The price system communicates information critical to achieving coordination at the local level and achieves general objectives as if an omniscient central planner instantaneously and continuously provided guidance to everyone. In fact, Hayek states, the marked does it so efficiently that its participants need to know nothing about the causes and ramifications of events affecting a particular local process in order to react quickly and through trial and error approximate the correct action that a central planner would recommend if he were omniscient. 

If Medicare aims to have local patients, physicians and hospitals to reduce costs, improve quality and adapt quickly to local changes, it would seem that the best and indeed only way to make that happen is to allow local providers who are most familiar with those local conditions, local resources and local changes to make those adaptations. Without a market and its price system, however, this is impossible to achieve. Centralization and price controls can only slow down adaptation, reduce coordination, increase misallocation of capital, supplies, and labor and ultimately reduce quality and increase costs. 

__________

¹ Israel Kirzner’s Competition and Entrepreneurship,  Chapter 6 Competition, Welfare and Coordination, helps clarify and augments Hayek’s argument.

Medicare can’t play market

The Journal of the American Medical Association (JAMA) reported June 24, 2010 that Medicare’s program requiring hospitals to reach certain numeric goals that measure the Surgical Care Improvement Process (SCIP) in care of surgery patients do not result in better outcomes: 

“Despite this evidence, hospital-level performance on individual SCIP measures is publicly reported "to assist patients in selecting centers of excellence" for receipt of their surgical care. This publicized use implies that reported adherence on these measures is directly related to improved outcomes. Our findings are unable to support this assertion.”

This is a significant blow to Medicare’s “Value-Based Purchasing.” This initiative is part of a trend that the tax-supported payor program for seniors started more than a decade ago to mimic the free market. The method that has been selected attempts to derive numeric scores in three quality domains: process of care, patient satisfaction and outcomes. From these scores, Medicare will apply a payment formula that will redistribute a pool of funds created by a 2% withholding of hospital Medicare payments. This is designed to provide the necessary financial incentives for hospitals to comply with the program. Hospitals with scores in the top quartile will receive 100% of their contributions into the pool. Hospitals in the bottom quartile will not receive any of their contributions back, and the middle 50% will be paid according to a formula that goes from 100% to 0%. These payment formulas will be phased-in from 2013 to 2017. In an era in which Moody’s “A” rated hospitals have operating margins of 2.2% and Medicare represents over half of that margin, giving up 2% of Medicare payments is the difference between keeping the hospital open or closing it. 

The study from JAMA looked at a subset of measures for the process of care in surgery known as Surgical Care Improvement Project (SCIP). This project was started by Centers for Medicare and Medicaid Services in 2002 to decrease complications and mortality associated with postoperative care in hospitals. Measures include antibiotics given one hour before incision, clipping rather than shaving hair on the surgical site, keeping blood sugar controlled between certain limits, treatment to prevent blood clots, etc., but there are other non-surgical process measures such as pneumonia patients receiving pneumococcal vaccination. Satisfaction measures include doctors and nurses always communicate well, rooms were always kept clean and quiet at night, etc. Outcome measures include mortality and re-admission rates.  Hospitals started reporting in 2004 on 10 measures. Currently they must report on 46 measures. This number is scheduled to grow to 55 in October 2010 and to 90 by 2012.   

Medicare’s hypothesis for Value Based Purchasing seems straight forward: if hospitals attain these goals in process, satisfaction, and outcomes, the quality of patient care will improve, and costly interventions to correct negative medical outcomes, such as additional surgeries, antibiotics, vascular procedures, additional days in the hospital, etc., will be avoided. Moreover, if hospital performance in reaching these goals is reported to the public, consumers will chose hospitals with better scores (and lower costs), hence overall healthcare costs for the program will drop. 

Using the Misesian critique of socialist calculation we have argued in prior articles (The Health Czar Can’t Calculate” and “Why Obamacare Can’t Work”) that it is impossible for Medicare to give any coordinating content to its mandates, that it is also impossible for healthcare agents to rationally allocate their scarce labor and capital resources under Medicare’s cost-based payment methods and that therefore, at best, Medicare will not be able to reign-in healthcare costs. 

The empirical results of this JAMA study point out that even if the socialist calculation problem did not exist, Medicare cannot know with certainty which hospitals offer high quality care and which don’t.  First, in reviewing other studies, the authors of the JAMA article found weak or no statistical significance between adherence to individual process of care measures and improvements in patient outcomes. Second, they argue that without better cause and effect relationships between process and outcomes, it is difficult to justify the massive investment in time and resources needed to track all these measures. Third, complications stemming from hospital care are influenced by a myriad of factors that are independent from the process of care measures. Fourth, public reporting of performance in process of care measures which compare hospitals do not infer quality differences between hospitals. And finally, that implementing financial incentives do not improve hospital quality. 

These empirical findings by JAMA are fully supported by the Misesian critique of the use of the method proper to the natural sciences to the social sciences. One could make the argument that medicine can use the method of the natural sciences effectively. However, medicine is as much science as it is art. Writing a progress note in a medical record, for example, one that successfully communicates to other clinicians the unique medical history, symptoms, diagnosis and treatment for a patient cannot be fully encapsulated in a statistical function derived from the reaction of the same types of patients in clinical trials or in the experience of the physician. But even granting such a concession, when the method of the natural sciences is applied beyond the narrow confines of very specific predictions about patients and their care to the world of hospital management in which social relationships between thousands of human beings such as patients, care-givers, support staff,  management, bond-holders, philanthropists, employers, third party payors, suppliers and competitors influence a myriad of cause and effect relationships that directly and indirectly affect nature of the care given, then the natural sciences method cannot make statistically significant predictions about that care. This is because complex relationships between individuals with their own uniqueness and freedom influence the outcomes that are being measured, and, more importantly, the actions of these individuals many times provoke the very events that Medicare is trying to prevent. 

If the cause and effect relationship between Medicare’s proposed goals for hospitals and overall quality is statistically weak or non-existent, then their connection to cost reduction is also weak or non-existent. Medicare’s Value-Based Purchasing initiatives will not cause a reduction in costs. This, in conjunction with the socialist calculation problem, will continue to confuse Medicare’s central planners as they attempt to encourage patients and healthcare providers to simulate the results of competition.