Where Are the Health Care Entrepreneurs?


DAVID M. CUTLER

Where Are the Health Care Entrepreneurs?

The United States lacks a culture of organizational innovation in health care. As we begin to implement the new health care law, there’s a lot we can do to change that.

Health care in the United States is notorious for market imperfections. Costs are higher and outcomes worse than almost all analyses of the industry suggest are reasonable. Indeed, few other industries perform worse than health care in serving their consumers. In other industries characterized by inefficiency, efficient firms expand to take over the market, or new firms enter to eliminate inefficiencies. But such organizational innovation has been rare in health care. Two main barriers stand in the way: lack of good information on health care quality and the dominance of payment systems that reward volume of care rather than its value.

Recent reform legislation promotes changes in each of these areas. Whether the legislation addresses these problems sufficiently is something that only time will tell. Still, it is clear that there are a number of actions that can help to promote innovation and entrepreneurship and in the process improve the performance and lower the costs of the health care system.

Relative to the economy, health care spending has increased by a factor of four in the past half century. Of course, not all medical spending increases are problematic. A good share of rising costs is attributable to the development and diffusion of new technologies, which often bring significant value. But alongside valuable innovation in medical care is an enormous amount of waste. More than one-third of medical spending, upwards of $700 billion annually, is not associated with improved outcomes. Indeed, rising health care costs are the leading contributor to projected federal deficits during the next few decades and make expanding health insurance coverage difficult to afford.

In other industries characterized by inefficiency, efficient firms expand to take over the market, or new firms enter to eliminate inefficiencies. But such organizational innovation has been rare in health care.

Inefficient spending is an example of low productivity; more is spent than is needed to produce the output achieved (or equivalently, less output is produced than is possible given the inputs employed). One way to gauge the relative efficiency of health care over time is to compare its productivity growth with that of other industries.

Productivity growth is notoriously difficult to measure in health care. Accurate productivity assessment requires a good output measure. Health is difficult to measure and even harder to decompose into medical and nonmedical factors. As a result, official data are much better on productivity outside of health care than they are in health care. Still, official data do provide some insight. Overall productivity growth in the United States was low from the mid-1970s to the mid-1990s. Since the mid-1990s, however, productivity growth has increased rapidly. Productivity growth in private industry, for example, averaged 1.25% annually from 1987 to 1995 and 2.4% annually from 1995 to 2005. This resurgence of productivity growth is often attributed to greater use of information technology.

The most productive industries were durable-goods manufacturing (6.9% growth annually) and information technology (5.7% growth annually). These industries are fairly different from health care. There are some industries with high productivity growth that are more similar to health care, however. One example is the retail trade. In the past 15 years, productivity growth in retail trade averaged 4.3% annually. Productivity growth in health care during this period is estimated to have averaged –0.2% annually. This is almost surely an underestimate, for various reasons. But even so, the negative value is striking.

Sources of inefficiency

There are three key reasons for the inefficiency of medical care:

First, people receive too much care. Low patient cost-sharing combined with generous provider reimbursement means that neither patients nor providers have incentives to limit care. Thus, many people now receive more medical care than is appropriate for their condition, especially in acute settings.

Consider the treatment of localized prostate cancer. Almost all elderly men have cancer of the prostate. In many cases, however, the cancer grows slowly and the person will die of something else before the cancer becomes fatal or even clinically meaningful. Thus, “watchful waiting” is a common strategy. Yet only 42% of elderly men with prostate cancer receive watchful waiting. The rest receive alternative treatments that are far more expensive and can have adverse side effects. There is no evidence that the more expensive treatments have better outcomes. By some estimates, up to $3 billion annually could be saved by adopting less-invasive therapies.

The prostate cancer example is not unique. The PROMETHEUS payment model estimates that 14 to 70% of costs for common conditions in the elderly, such as joint replacements, heart attacks, congestive heart failure, and dia betes care, are avoidable.

Second, there is inadequate care coordination. For many medical conditions, people need to see generalist and specialist physicians, receive periodic lab tests, take medications, and modify their behaviors. This complex regimen is almost always left to the patient to plan and coordinate, although we know that many people are bad at this. Partly as a result, people receive too little chronic and preventive care, costing both lives and dollars. In the case of diabetes care, for example, only 43% of diabetics receive recommended therapy, and an even smaller share meet guidelines for risk factor control.

It is possible to do better, and a number of integrated provider systems show how. For example, HealthPartners, a health maintenance organization (HMO) in Minneapolis, began a program in the mid-1990s to improve diabetes outcomes. The organization worked with its physicians to identify diabetic patients who had not received recommended screening and provided nurse case managers to call the patients. Physicians were encouraged to start medication therapy in patients for whom diet and exercise were not sufficient. Patients, in turn, were reminded to take their medications and receive recommended screenings. Individual and group sessions developed mechanisms for people to manage their disease, and nurse case managers helped as needed. In the five years after this program was implemented, patients’ rates of high blood sugar fell by half and their diabetes was brought under much better control.

HealthPartners and other high-performing providers have three critical attributes that contribute to their success. They integrate care across different providers by having providers in the same physical or virtual organization; they pay physicians on a salary or productivity basis, not a fee-for-service basis; and they decentralize decisionmaking to encourage productivity.

The biggest problem for HealthPartners was that the economics did not work out well. The cost of the program was a few hundred dollars per diabetic patient per year. Better diabetes control translates into fewer adverse events, but that comes a few years down the road. The HMO feared that many patients would transfer to a new insurer before the benefits of prevention were noticeable. By one estimate, the plan’s return on investment would be favorable over a decade but not by anywhere near the social value of the program.

The lack of coordination that is endemic to chronic disease care is noticed by consumers. According to one survey, 25% of Americans with chronic disease have had the experience of having records unavailable when needed, and 20% have had a doctor order a repeat test. Overall, 35% of Americans felt their time was wasted because of poor organization.

Comparable data on perceptions of other industries are not available. That is not an accident; consumers are rarely as poorly served in other industries as in medical care. In making retirement savings decisions, for example, companies such as Fidelity and Vanguard automate the collection of money and its allocation. Airlines store flight information electronically for easy access throughout one’s trip. And specialty stores in retail bring together different products, so consumers do not have to physically compare products from different suppliers.

To be sure, the retail model of organization has imperfections. Electronics stores encourage people to buy more gadgets than they need and sell them overpriced insurance for what they buy. The fees collected by mutual funds are far higher than those that a perfectly competitive market would suggest. But still, these market organizers have gained enormous market share because of their service quality and low price.

The low level of service quality in health care is ironic given the enormous investment in nonclinical personnel. There are nine times more clerical workers in health care than there are physicians and twice as many clerical workers as registered nurses. But this investment has not paid off in superior outcomes or better customer service.

The third key reason for inefficiency in medical care is flawed production processes. Medical care providers are far less efficient than they should be. Wasted time is rampant. For example, physicians spend an average of 142 hours annually interacting with health plans, at an estimated cost to practices of nearly $70,000 per physician. Similarly, nurses in medical/surgical units of hospitals spend 35% of their time on documentation, considerably more than they spend on patient care. Doctors also routinely redo tests because the prior test results are not available or would require too much effort to obtain.

Better care models exist, generally involving the use of information technology and changes in workplace practices. For example, Kaiser Permanente found that use of information technology combined with organizational changes led to a 35-minute reduction in nursing overlap time associated with shift changes. Similarly, a variety of studies have shown that providing dedicated surgical suites for particular operations results in lower cost per surgery. Yet most full-service hospitals do not organize their operating suites in this fashion.

In some cases, this administrative complexity actually harms patients: 4% of hospitalized patients suffer an adverse event, of which one-third, or 1% of total hospital admissions, are a result of negligence. The Institute of Medicine estimated in a 2000 report that preventable medical errors resulted in between 44,000 and 98,000 deaths annually, making errors one of the top 10 leading causes of death. Errors are also expensive, costing the system about $30 billion annually.

There are many models for reducing medical errors. Adverse drug interactions can be virtually eliminated by computerized physician order entry systems, which cost roughly $8 million each. Yet only 4% of hospitals have fully adopted such systems. Similarly, surgical complications can be reduced through organizational innovations such as surgical checklists and other process steps, but the use of checklists remains relatively low.

All told, the costs of too much acute care, poor care management, and inefficient production are staggering. Excessive care accounts for as much as 30% of total medical spending. Administrative inefficiencies amount to about 10% more. And lack of prevention is associated with countless lost lives, with mixed data on costs. An efficient medical care system would save half or more of this amount, as much as $1 trillion per year.

How change occurs

Medical care is complex, and as in any industry where human action is important, there are bound to be mistakes. The failure of medical care is not so much that mistakes are made, but rather that the system has not developed or broadly adopted mechanisms to minimize those mistakes. In retail, in contrast, producers invest extensively in error reduction (Toyota, for example) and supply-chain management (Wal-Mart). In financial services, an entire set of firms is devoted to helping consumers save.

Some ways to improve production processes are clear. Primary care physicians, for example, could provide a “medical home” for patients with chronic disease. Similarly, multispecialty groups of physicians might combine into accountable care organizations to make sure patients do not fall through the cracks. Alternatively, payers for medical care (insurers or the employers they contract with) could push for coordination. Even further removed, a firm from outside medical care could enter health care and organize the care experience, as Amazon.com did with book sales and Expedia did with airline tickets.

Several health care organizations have become leaders in improving process design. For example, Virginia Mason Medical Center in Seattle committed itself to lean manufacturing principles in 2002. During the next several years, it focused on patient safety, care coordination, supply management, and nursing productivity. Among the returns have been greater patient volume, reduced capital expenditure, and less use of temporary and contract nurses. Similarly, Thedacare in northeastern Wisconsin cut costs by 5% in three years and improved quality by using tools of lean manufacturing. Perhaps the biggest transformation of all was the Veterans Administration (VA). Between 1995 and 1999, the VA handled 24% more patients despite a budget increase of only 10% (compared with a 30% increase in the health care system budget overall). The VA was able to do this through greater use of information technology, more local financial autonomy, and empowerment of regional managers to make decisions based on local conditions.

The VA and Virginia Mason examples suggest that a large amount of total hospital costs are unnecessary. If hospital costs alone could be reduced by one-quarter—an amount well in line with estimates of waste—total system savings would be about 8%.

Many of the investments needed to reduce the inefficiencies of health care and improve the quality of care must occur at the provider level. New information technology (IT) systems need to be introduced at the practice level, and workflow must be arranged for each provider. To understand the economics of provider-driven reform, consider the standard profit equation, in which profits are equal to total revenue (price of the service multiplied by quantity) minus cost. Thus, to be adopted, organizational innovation must positively affect the price or the quantity of services sold, or reduce costs.

The failure of medical care is not so much that mistakes are made, but rather that the system has not developed or broadly adopted mechanisms to minimize those mistakes.

Almost all of these interventions to improve provider-level productivity require upfront investment, either monetary or organizational. Computer systems to check for adverse drug interactions run into the millions of dollars, and changing surgical practices involves reorganizing care throughout the institution. Thus, provider groups need some return in order to make these investments.

Price increases are not a part of a favorable return. Hospitals are typically paid on a fixed-fee basis, independent of quality. For example, Medicare reimburses hospitals for a predetermined amount per stay, depending on the diagnosis of the patient and whether surgery was performed. A less good job earns as much as a better job. Building off Medicare, private insurers generally use per-stay or per-diem payments: A single payment is made for all services provided in that stay or during that day, again independent of quality. As a result, improved quality merits no higher price.

Quantity responses to quality improvements also are limited. One might imagine that more patients would choose to be operated on in hospitals with safety systems or more regular surgical times. But information about such forms of quality is not systematically available. Until very recently, there were no validated measures of provider quality that accurately accounted for differences in patient severity of disease. And even now, measures of clinical and service quality are extremely limited. As a result, hospital choice is based on reputation or uninformed recommendation more than actual data.

Thus, cost savings represent the only area where meaningful gains might offset investment costs. Many productivity innovations will reduce costs. For example, fewer errors means shorter hospital stays, which lower costs. It might also mean lower malpractice premiums. A full analysis of investment in more efficient production has not been undertaken, and it may be that providers should be investing in efficiency improvements on this basis alone. To date, however, the vast bulk of hospitals have concluded that the financial and organizational costs of transforming their institution are not matched by sufficient cost savings.

An example is telling. In the 1990s, Cincinnati Children’s Hospital decided it wanted to become a leader in quality of pediatric care. The hospital’s chief executive officer and its board of directors agreed with the plan. But the finance team saw quality improvement as harming the finances of the institution, which were based on admitting more children and treating them in a high-tech way. No payer reimbursed them more for higher-quality care; in fact, it was penalized.

In the end, the finance team was brought along, but only after someone pointed out an error in the team’s thinking: Having fewer medical errors meant more rapid discharges, which could be offset by admitting more patients from the queue. Thus, there would be no revenue loss from better care. After demonstrating that revenues would not be harmed, the staff at Cincinnati Children’s Hospital went ahead with the quality improvement efforts, and the hospital now is a model for other institutions.

Efforts to limit excessive care and better coordinate care also face financial difficulties. In each case, physicians or nurses must expend effort to make the system better. In the cases of prostate cancer and diabetes, for example, patients need to be counseled about treatment options and informed about the steps involved in good disease management. A successful intervention will probably lower downstream costs. But from the physicians’ perspective, the pricing of medical care makes the switch from invasive medical procedures to advising and counseling problematic. Most physicians are paid on a fee-for-service basis. In the case of Medicare, the service units are independent activities that a physician performs when seeing a patient: a routine office visit, a procedure, or an interpretation of an image. Quality is not a part of the calculated fee. Rather, the fee is based on intensity: Procedures are valued much more highly than is counseling. Further, many of the simple services that are involved in good care management are not reimbursed at all. There is no billing code for e-mail interaction, nor is there any payment for having a nurse place a reminder in the file to call a patient, and stressed providers focus on the quantity of services they bill to the exclusion of the outreach function.

Coupled with this financial disincentive is the traditional norm that separates the practice of medicine in a medical setting from social interventions. Doctors are trained to diagnose and treat patients. They are not trained to counsel or reach out to patients. Physicians can be made to see their job differently, but the incentives to change need to be very strong. In the current system, these incentives are weak, if present at all.

Payers not stepping in

Given the poor incentives transmitted to medical care providers, the obvious question is why payers—insurers, the employers who contract with them, and third-party firms that purchase and manage the care from individual providers—do not intervene. Payers have a number of options. They could require providers to adopt and use interoperable electronic medical records, and they could move to quality-based payment systems to provide incentives for more efficient care.

Why do they not do so? Four explanations have been proposed:

Network externalities. In this explanation, a single payer finds it difficult to have compensation arrangements or accreditation rules that are substantially different from those of other payers. Medicare and Medicaid together account for about 40% of acute care payments, and private insurance accounts for another 40%. (The remaining 20% is from other payers, including worker’s compensation, the VA, public health agencies, and individual consumers’ out-of-pocket payments.) Within the private insurance market, there might be three or four large insurers, for an average market share by each plan of about 10%. It is difficult for an insurer to fundamentally change the practice of medicine when it accounts for only 10% of the market. For example, even an insurer that put 20% of a physician’s revenue at risk for poor performance would affect only 2% of the typical provider’s income. Given the fixed cost associated with provider change, this incentive system is unlikely to do much good.

Further, because of the fixed costs of providers changing their practice, even if the insurer were able to change provider behavior, the savings would be realized by all payers. A primary care physician that responds to insurer incentives by hiring a nurse case manager to work with diabetic patients will have that nurse manager work with all patients, not just those of a particular insurer. Thus, the benefits of any insurer investing in better care extend well beyond that insurer.

Two solutions are generally available for solving the network problem. First, integrated firms may arise that provide both insurance and medical care and thus internalize all externalities. Kaiser Permanente is an example of such a firm, and it provides among the highest-quality chronic disease care. As with most high-quality firms in health care, Kaiser has walled itself off from the rest of the health system. Alternatively, providers could propose new contracts to insurers. For example, providers might suggest that cost savings that result from fewer hospital-based errors be shared between the innovating firm and the various insurance companies. The major problem here is Medicare. Medicare reimbursement has been fixed by law, making that part of revenue unalterable.

Lack of information. Within a market, lack of good-quality data means that consumers have a difficult time determining which providers are better and worse. And across markets, lack of good information means that firms with high quality in one geographic area will not necessarily be perceived to have high quality in other areas. The difficulty of measuring quality is a fundamental difference between health care and most other retail products. Retail stores can be virtually identical across the country, allowing firms to earn a national reputation for high (or low) quality relatively easily. In health care, national reputations are uncommon.

The information problem in health care is very much a public good. All insurers would like to have good data on physician quality, but no single insurer has an incentive to create such data, since quality information will rapidly disseminate across the market. Thus, some governmental involvement in information is needed.

Plan turnover. Suppose that an insurer decides to coordinate care on its own. It might hire nurse case managers, work directly with patients, and reconcile different physician recommendations. But investing in better care has up-front costs, whereas many of the savings occur only over time. For example, better diabetes care may lead to fewer complications, but only after 5 to 10 years. Because as many as 20% of people change plans annually, the insurer undertaking the original investment may not realize the savings. Thus, high turnover has been cited as a cause of low quality.

This explanation, however, is not entirely convincing. The high turnover in health insurance is partly endogenous: Customers feel little allegiance to a plan whose perceived quality is low and whose services are comparable to those of every other insurer. In plans with a reputation for good quality—Kaiser Permanente in California, for example—turnover is much lower.

The wrong customer. The issue of turnover raises a general question about who is the appropriate customer when payers consider care management. In the retail trade, the customer is the individual shopper. If Wal-Mart finds a way to save money, it can pass that along to consumers directly. In health care, in contrast, the situation is more complex, because patients do not pay much of the bill out of pocket. Rather, costs are passed from providers to insurers to employers (generally) and on to workers as a whole. If this process is efficient, the system will act as if the individual is the real customer, because that person is ultimately paying the bill. It may be, however, that the incentives get lost in the process, and efforts to innovate are not sufficiently rewarded.

What difference does selling to an employer or selling to an individual make? Even if insurers wrongly think that their customer is the employer purchasing insurance, that employer may still value improved quality. Many firms, for example, invest in wellness programs, which often involve attempts to coordinate care. If the cost savings or productivity benefits of improved health are sufficiently high, this is a natural step for employers.

Impact of recent legislation

Recent legislation has made a start at making health care more efficient by investing in better information and changing compensation practices for providers. On the information end, the key change was the HITECH Act of the American Recovery and Reinvestment Act of 2009. In that legislation, the federal government committed $30 billion over five years to finance a national system of electronic medical records. In addition, the Patient Protection and Affordable Care Act of 2010 mandates that Medicare data be made available to private parties, including insurers and employers, for purposes of forming quality measures. Thus, the nation may be on the verge of significantly reducing the information problems in medical care. In an otherwise contentious legislative process, changes allowing greater investment in health care IT were generally applauded.

Changing the payment system for medical care was more controversial. Broadly speaking, there are three approaches to payment reform. The first approach is to adopt a single-payer system in which physicians are salaried or paid on a fee-for-service basis within an overall budget target. Such a system is common in many countries and can be successful in reducing unnecessary care, assuming that physicians cut back on the appropriate services. The second approach is to turn health care into a market like other markets, where individuals are more in charge of their spending and service use. This would take the form of much higher deductibles in Medicare and incentives to purchase less generous policies in the under 65-market. The idea behind this model is that providers forced to compete for individuals would invest in higher quality, the same way that retail firms do. The third approach is to keep cost sharing as it is but to reform the way that Medicare payments operate, to incentivize value more than volume. The underlying theory is that changes in Medicare, integrated with changes in private reimbursement, will provide incentives for more efficient care delivery.

Following the third path, changes in Medicare reimbursement are a significant part of the recent Patient Protection and Affordable Care Act. Under that legislation, bundled payments will be made by Medicare to groups of providers who jointly agree to care for a patient with a particular condition; providers will split the overall amount and share the profits. Accountable care organizations go a step further, with groups of providers agreeing to accept a capitation payment in exchange for providing all services needed during a year. Pay for performance, or value-based purchasing, is a third payment reform, adjusting fee-for-service payments for primary care to reflect the quality of the care provided. Finally, care coordination and transition payments are introduced to provide support to nurses or primary care physicians who seek to manage care better.

Each of these payment systems has been the subject of experimentation, with some success in each case. Payment bundles are the best developed. Medicare’s Heart Bypass Center Demonstration Project in the 1990s bundled all care for coronary artery bypass graft surgeries; the program achieved savings of more than 15% per episode. Among care coordination efforts, the medical homes initiative of the Geisinger Health System, based in Pennsylvania, achieved a 7% total medical cost savings and a significant reduction in hospital admissions in the first year. In the program, each patient is assigned a health professional who acts as a coordinator, or “medical home,” for all of that patient’s care. Overall, care coordination efforts appear to be able to save about 15% of inpatient costs when they target populations with chronic illnesses.

How these various reforms play out will be interesting to watch. There is ample evidence that we can improve the performance and reduce the costs of today’s health care system. The challenge now is to take these metrics and the additional improvements identified as the health care reforms are implemented to build the fabric for tomorrow’s system.


David M. Cutler () is the Otto Eckstein Professor of Applied Economics at the Harvard Kennedy School of Government and a research associate at the National Bureau of Economic Research (NBER). This article is based on a paper presented at a NBER meeting on Innovation Policy and the Economy, held in May 2010, and will be published in a forthcoming NBER volume of papers prepared for the conference.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>