Expensive Hospitals: The Enemy Within


Everyone agrees that health care is bankrupting the nation. The prevailing winds have carried the argument that a system that pays per unit of health care delivered and thus favors volume over value is responsible. The problem, you see, was the doctors. They were just incentivized to do too much. This incontrovertible fact was the basis for changes in the healthcare system that favored hospital employment and have made the salaried physician the new normal. Yet, health care costs remain ascendant.


It turns out overutilization in the US healthcare system isn’t what its cracked up to be.

utilizationFigure 1. Utilization rates in different health care systems

A recent analysis (Figure 1) by Papanicolas et al., in JAMA demonstrates that while the United States is no slouch with regards to volume of imaging and procedures in a variety of different categories, it does not explain a health care system twice as expensive as its nearest competitor. The problem turns out not to be volume, rather its the unit price of healthcare in the United States.

Health Care Costs and Glass Houses

There are many stones cast by all the various players in healthcare when it comes to cost, and of course, everyone bears some degree of responsibility, but it’s also clear that some folks live in larger glass houses than others. The most beautiful of all the glass houses are those built by hospitals. From 1996 to 2013, it was not population growth, health status, doctors visits, or prescription drugs that drove spending increases. Sixty-three percent of the increase in cost over an almost 20-year time span can be attributed to hospital stays and testing during doctor visits. Consider that the average hospital stay in the US costs $18,142, and lasts 4.9 days compared to other industrialized countries where average hospital stays last 7.7 days, and cost $6,222. But despite these exorbitant prices hospital systems in the United States complain they barely stay afloat.

Physicians blame administrators, administrators blame physicians, woke liberal venture capitalists blame the Koch Brothers, and economists blame everyone. Ambling through hospitals for the better part of the last two decades suggests real culpability strikes disturbingly close to home. Physicians work in a hospital climate where they are utterly oblivious to the cost of the health care they deliver. No one has any idea what an X-ray, a bag of saline, or an antibiotic costs. Insulation from cost isn’t all bad – it’s nice to know physician decisions are being made irrespective of the ability to pay, but the flip side of this is that there is precious little incentive to be efficient about health care delivery. Physicians ask for the moon and usually get it. Hospital administrators poorly versed in clinical matters respond to high costs by doing what they do best: pushing doctors to see as many patients as possible, and rewarding physician rainmakers that bring in the most clinical revenue. So, hospital administrators don’t choose to run inefficient enterprises, they just happen to be constrained by clinical partners with no incentive to be remotely efficient.

Consider the change over time in how hospitals are staffed. It used to be that hospitals were staging grounds for physicians who owned and operated their practices, managing busy outpatient clinics, and managing patients in the hospital when the need arose. In this setup, the hospital had no role in paying physicians directly, rather physicians paid themselves out of the clinical revenue generated from seeing the patient in the hospital. Driven by a system that moved to pay hospitals per diagnosis rather than by a number of days, and thus put a premium on shortening the length of stay, the inpatient hospitalist was born. Hospitalists are physicians who specialize in the care of hospitalized patients and are almost always employed by hospitals. It soon became standard for outpatient physicians to stay in clinics, and have hospitalists take care of inpatients. In theory, this would allow for better care of inpatients, and allow primary care physicians to see more outpatients. In the real world, as the outpatient practice of medicine was made more onerous and less lucrative, doctors flocked to do hospital shift work upon graduating.

The results in hospitals were wondrous to behold. The same small hospital I frequented that a decade prior operated with a skeleton full-time physician crew now employs over 20 full-time hospitalists. Length of stays are certainly shorter, documentation of inpatient stays is robust, but of course, there is no evidence that real outcomes for patients are any better. Hospitalists initially were focused within general internal medicine, but have now quickly metastasized to specialty care as well. Innovation in hospitals seems to consist of creating a fresh class of hospitalists for every organ system. Soon there will be hospitalists that specialize only in the big toe. The documentation promises to be fantastic. Factoring in medical malpractice and benefits, the average hospitalist in a big city market costs a quarter million dollars yearly today, and there are 50,000 hospitalists now working in the United States. All of this contributes to hospital salt water that is the most expensive in human history. To be absolutely clear, I am not suggesting there is no value to hospitalists. Some of the best physicians I know belong to this group. I am suggesting that this is an utterly inefficient, fragmented way to deliver care supported by health care prices outside the bounds of reality.

Using the Evidence to Compound Health Care Costs
And this is only the tip of the iceberg. Hospitals are masterful in their ability to expend vast resources to achieve illusory goals. In all fairness, this is driven by a research industrial complex constantly publishing the next great therapies that may happen to reach statistical significance, but in reality, have at most Lilliputian effect sizes.
Consider a recent multidisciplinary ‘teaching’ conference I happened to attend that was focused on the management of pulmonary embolism (clots in the lung). The most important thing to know about pulmonary embolisms is what was not mentioned during the conference – the case fatality rate for the diagnosis has not changed in a quarter century. The chart below does demonstrate a doubling of the incidence of PE, but this would be because of ever more sensitive ways to image the pulmonary arteries that reveal clinically irrelevant pathology since there has been no change in deaths from PE over the same time span!

PE-incidenceNonetheless, this particular academic conference has interventional radiologists, cardiologists, pulmonary critical care doctors, vascular medicine physicians, and yes even hospitalists impressively arrayed to focus on improving the management of Pulmonary Embolism. The propose of this conference is to make a rare problem even rarer by creating a hospital Avengers team know as PERT (Pulmonary Embolism Response Team). The goal of the meeting is to convince the wider medical community to activate the PERT team to evaluate pulmonary embolisms for new therapies for PE.
Traditionally PE’s are treated with blood thinners (Heparin/low molecular weight heparin/Factor Xa inhibitors) that prevent propagation of clot or clot-busting drugs that actually break down the clot. A very small number of patients don’t respond to these therapies because the burden of the clot in the lungs is too great, and the thrombus needs to be surgically removed in a gory, but life-saving maneuver called a pulmonary thrombectomy.

More recently, interventional radiologists have developed techniques to thread catheters directly into the pulmonary arteries to dissolve clots – a technique known as catheter-directed thrombolysis (CDT). The clear message at the conference was that your next patient with a pulmonary embolism should be evaluated by the PE response team for advanced therapies such as these. The basis for all of this is sold as evidence-based commandments, but the reality is far from it. (Readers are advised to skip the next two paragraphs if they would like to just take my word for it)

The Evidence-Based House of Cards

The results of this approach are somewhat predictable and are encapsulated in an interesting paper described a single institutions experience with PERT. The study looked at a 20 month time window after PERT came to a New York Hospital. There were a total of 124 PERT activations – 43 in the first ten months, 81 in the last ten months. 21/25 patients taken for CDT were taken because of something called submassive PE. Submassive PE refers to strain seen on the right side of the heart as it attempts to pump blood around the clot. It used to be that right-sided strain was diagnosed based on a heart ultrasound, but recently a more sensitive, but less specific technique measuring proteins released by the straining right heart was included as part of the diagnostic criteria. The problem is that this biomarker elevation without sonographic evidence of right ventricular dysfunction isn’t a high-risk marker in isolation because any stretch of the ventricular chambers will cause some leakage of biomarkers. Only 13-14% who had an abnormal test in a case series ended up dying. While this is not a small number, it does argue against this test in isolation being able to discriminate a population that would significantly benefit from an intervention. Even if you add in the presence of sonographic right ventricular dysfunction, it is unclear that using clot-busting medications up front is beneficial. A 2003 study that compared a clot-busting drug to blood thinning was ‘positive’ only because more people in the blood-thinning arm eventually ended up getting a clot-busting drug. The PEITHO trial in 2014 showed similar results. More patients clinically decompensated in the blood-thinning only group (5.6% vs 2.6%) but overall survival was no different between the two groups, though there were more bleeds into the brain (2% vs 0.2%) with a clot-busting drug. So, in general, a strategy of blood thinning first with the addition of a clot buster in the case of clinical decompensation is a safe yet effective option.

CDT sought to capture the small benefit of systemic thrombolysis while reducing the risk of bleeding into the brain by infusing clot-busting drug directly into the artery where the clot resides. The SEATTLE II trial examine this strategy by taking 150 patients with central PE and performing CDT. Statistically, significant differences were seen before and after CDT with regards to pulmonary artery pressures and left to right chamber size ratio. I don’t mind relevant surrogate endpoints, but the problem here is that the clinical meaningfulness of these surrogate endpoints is questionable. Pulmonary artery pressures, as well as right ventricular chamber size, are notoriously dynamic in nature. Transient elevations in pulmonary artery pressures cause the thin-walled, compliant right ventricle to dilate frequently. Both parameters may quickly return to normal once the underlying cause of high pulmonary pressures has been treated. There was also no control arm to see how CDT performed relative to conventional blood thinning. The trial certainly did demonstrate safety but was far from convincing in demonstrating efficacy.

To be clear, I’m a fan of procedures with a large upside and little downside in patients in extremis. I don’t care that much about negative randomized control trials in this space because I don’t really care about average yardage gained when throwing a hail mary. But we all should care when protocols are put in place to trigger algorithms in patients that aren’t in extremis. This is usually what happens when a panel comes together to design protocols. In order for them to be practical, they are usually simple flow diagrams based on some easy to measure variable. You can’t have a protocol that says “if the patient looks sick” because that would be subjective and introduce that horror of all horrors: variability. Judgment is deferred to the PERT team that is more likely than not to be comprised of someone making a decision based on reading some other disconnected from the case physician report of an echocardiogram. That this is state-of-the-art medical care in 2018 would make Osler weep.

The change in practice pattern after PERT institution are remarkably predictable. If you set up a protocol to call a barber, you get a lot of haircuts. In the 20 months prior to PERT, there was 1 CDT. In the 20 months after PERT inception, there were 25. Recall the problem we are attempting to solve (Death related to PE) has a rare case fatality rate that has not changed in two decades. The majority of these PERT activations were due solely to the presence of a biomarker elevation, which almost surely means most of these interventions took place in patients that would likely have been fine with conventional anticoagulant therapy.

My bias and unwillingness to trust the surrogate endpoint improvement relates to a large number of patients who develop transient RV dysfunction during an acute event that normalizes after conventional therapy. There are a small minority of individuals who develop a chronic clot and elevated pressures, but it is impossible with what we know now to predict who that is, and we don’t have a good sense either mechanistically or in a trial setting if CDT would prevent this. So while it is wonderful to know there is another option for the decompensating patient with a PE, the PERT program appears designed to funnel more patients to a lab to get a procedure with questionable benefit.

The resources to marshall all of this are not cheap. A separate layer of PERT consult physicians are added to the mix, catheters to do the procedure need to be stocked, and staffing to accommodate 24/7 interventional radiology coverage for this procedure are marshaled. Outside community hospitals in the network even start to send patients to the hub just to have this procedure.  Costs spiral, physicians pat themselves on the back, patients thank the heavens their lives were saved, and hospital executives earnestly proclaim to politicians and those who pay for healthcare that providing state of the art healthcare in the United States is an expensive endeavor they can do nothing about.

Physicians are wont to point everywhere but themselves whenever high prices in health care are raised. I know all the arguments well, but the hard truth lurking beneath the covers is that physicians lie at the heart of all of this. Fixing this mess from the bottom up will require physicians to take ownership of a situation we have created lest we wish the solution be dictated to us by people without a clue.

The cartoonist Walt Kelley may have said it best: “We have met the enemy and he is us”.

Anish Koka is a cardiologist in private practice in Philadelphia.  He is fond of living in glass houses and can be followed on twitter @anish_koka

Spread the love

18 replies »

  1. Academicians, think tanks, most economists, central planners and common sense will tell you that the price a hospital, or any business for that matter, charges for a service is determined by their costs of production. But as any good Misesian will argue, all these folks have the cause-effect equation backwards: it is the price payers are willing to pay that determines the costs of production. All we hospital administrators do is simply bid up factors of production (physicians, nurses, accountants, marketing, drugs, other supplies) and capital (brick and mortar, technology, etc.) in response to expected profits (i.e., the expected future prices minus expected costs). So why are hospital costs high is the wrong question. The right question is how do Medicare, Medicaid and private insurance companies determine the prices they pay for the services hospitals provide. Have fun finding a logical answer. Medicare, for example, uses cost information that hospitals submit, the byzantine “cost report,” to set initial operating and capital payment base rates that a set of efficient providers are supposed to incur. Then all of that data is pushed through mathematical models and moderated by a variety of levers, including things such as geographical wage indexes, inflation, etc. plus intense lobbying, to come up with a price. Can anyone in the industry really say with a straight face that Medicare produces meaningful pricing information to rationally direct a hospital’s decision regarding allocation of capital and factors of production? I don’t think so. If hospital costs in the US are too high, and the data in this study suggests that they are, there are two ways and only two ways to solve this conundrum. One is “government rationing” via cost reports, DRG’s, Value Base Payments, capitation, denials, precertification, HACS, ACO’s, next generation ACO’s, etc., etc. I say government because inevitably, more and more of the healthcare dollar is spent through government programs such as Medicare, Medicaid, the VA, “public-private partnerships,” etc., and more and more private insurance companies mirror Medicare, for example by paying a fraction or a multiple of Medicare. This has lead us towards what Austrian School economists call the socialist calculation problem: that under conditions of central planning, rational economic calculation to efficiently allocate scarce resources is practically and theoretically impossible. What you instead get is a gross misallocation of capital and factors of production, which in turn leads to high costs, cueing, high variation in quality, poor outcomes, etc. The other solution to the conundrum is “market rationing,“ where the price of a health service is derived from the voluntary exchange between the consumer (patients) and producer (hospitals and physicians), in other words, the free market, like it is determined in most other industries that actually work, say food, auto, electronics or housing. There are no “third ways.” Given the ideological and political environment in the US for the last 100 years or so, to paraphrase Mises, we will continue to grope in the dark.

  2. We are hypocrites. Docs* shouldn’t even talk about costs because we know nothing. Eg no doctor knows why the PBMs keep 10-20% of the manufacturers rebates to the insurers or why the hospital is getting rebates at all. ? Or why the patient often benefits little from the rebates? Or why all these figures are quite hidden from everyone? Or why the total rebates are so huge? 89 billion for one year according to CNN Money. Before we yap about costs, we need to be sophisticated. We need to take agency for our patients seriously, as per Allan.

    GPOs (Group Purchasing Organizations) for hospitals are just as opaque. There are rebates from these back to the hospital. Do we even sit down with our administrators and ask what is happening to all thus dough? Is it benefitting patients?

    *Barry Carol excepted.

    • Actually Dr. Adam Fein recently reported on his Drug Channels blog that CVS’ PBM, Caremark, only kept about 3% of the rebate money it received last year, not 10%-20%. The rest was passed on to self-funded employer clients and insurer clients which was used mainly to reduce premiums and employee contributions toward the cost of health insurance.

      Starting in January 2019, United will change its system so that insured members only pay the net price of the drug within the deductible, not the artificial full list price. That’s a big step in the right direction. It will reduce the amount of rebate money available to cut premiums but it’s the right thing to do in my opinion.

      Rebates are volume based payments designed to move market share through favorable formulary placement. They are not just paid for the fun of it. Formulary placement with tiered copays is the main mechanism payers have to steer patients toward the least expensive and most cost-effective drug for their condition. The other alternative is to not cover the drug at all if deemed too expensive vs. alternatives. Other approaches to limit the use of the most expensive drugs within a therapeutic class include prior authorization and quantity limits.

      For the record as I’ve noted before, I’m not a doctor and have never worked in the medical field. My career was in the investment and money management field.

      • Vizient, Premier Inc., HealthTrust and Intalere are giant purchasers for supplies for hospitals trying to use oligopsonic purchasing to get good deals from manufacturers. They should be paid by hospitals and clinics, not manufactures. They were allowed by Congress in the mid-80’s to get paid rebates from the manufacturers. What has happened is that they, in fact, are taking money from both sides, from hospitals to get good prices through large purchasing (oligopsony) and from manufactures to get exclusionary supply contracts. It is an obvious conflict of interest. They even share kickbacks back with their members and “owners”, the hospitals. Many hospital executives of the Premier operation rely on money from Premier as a part of their annual compensation.

        The whole mess is shrouded with secrecy and the Emergency Physicians ( Dr. Blum) are mad because this increases prices, and causes shortages, they believe, of sterile injectible products….epinephrine, chemo drugs, anesthetic agents, painkillers, antibiotics, sterile water.

        See WSJ “Where Does the Law Against Kickbacks Not Apply?” by Phillip Zweig and Frederick Blum” …about 2 months ago.

        • I don’t know that much about how GPO’s actually work in the medical field. If I were a hospital, I would want to know how the price it pays for supplies compares to the price the GPO paid the manufacturer. Some spread would be appropriate in order to cover the GPO’s operating overhead.

          If I’m a manufacturer, it’s pretty routine to offer a better price for larger orders and exclusive sourcing contracts. When Walmart or Home Depot buy paint by the truckload, they are going to get a better price than the local hardware store that just needs one or two cases.

  3. I wonder about two things. First, how much more testing is there in the U.S. both before admission to the hospital and while in the hospital for any given procedure vs. the same procedure in other developed countries? Second, how do the costs of operating U.S. academic medical centers and community hospitals per licensed or occupied bed compare to similar hospitals in other first world countries?

    Costs can be driven higher in the U.S. by such factors as staff compensation, stricter building codes, higher nurse staffing ratios, and administrative complexity around billing among other factors. How could we have Medicare for all at existing Medicare reimbursement rates if most hospitals and other providers as well claim that Medicare doesn’t pay them enough to cover their full costs?

    • taking the data points about utilization as some sort of basis, suggest we lead the world in amount of testing ordered. Its just not enough I think to explain a doubling of cost compared to our compadres.

      Medicare for all at current medicare prices would mean hospitals as currently imagined would cease to exist. Which would also mean the army of people the hospital is currently able to employ would disappear..

      • So what would healthcare actually look like under Medicare for all at current Medicare prices vs. what it looks like now? What would the impact on patients be on everything from wait times to end of life care to the availability and access to cutting edge specialty drugs and devices?

  4. We all want prices to go up. It makes us richer.

    We could stop this tomorrow if we billed the patient and had claim money go to the patient first.

    I.e. returned to indemity.

    • No question – the system has unleashed vast resources to be consumed by healthcare. Some of that has been very good. Much of that has been not so good. Would a smarter approach be possible – I think so.. but requires physicians to get their heads out of the sand

    • Bill, I totally agree with you but it is not indemnity that does the job of cutting costs rather patient involvement in the process. The US Government involvement has been too pervasive in the wrong places. In order to correct its errors it has tried to create artificial marketplaces and failed miserably.

      • I should have added it is the physician’s job to do his best for his patient. The system is playing the blame game and that doesn’t work either. When the patient is in control the blame game can stop for the price of care and insurance has to fall to the levels the patient accepts. He then indirectly chooses the type of care available based on the insurance plan he is willing to buy.

  5. “Sixty-three percent of the increase in cost over an almost 20-year time span can be attributed to hospital stays and testing during doctor visits.”

    Do you have a citation for this? The literature with which I am familiar suggests that it is outpatient costs which have been the bigger driver.

    “Fixing this mess from the bottom up will require physicians to take ownership of a situation we have created lest we wish the solution be dictated to us by people without a clue.”

    Amen. In general, we need much more physician involvement and much better leadership. I think this should start somewhere in training, but it needs to be nurtured when people are out in practice.


    • I sort of assume you used this JAMA paper since it uses the same dates you quote. How many papers are going to look only at changes from 1996 to 2013.


      It shows total spending increasing by $931 billion between 1996 and 2013. Of that, $324 billion was from ambulatory care cost increases, and $258 billion is attributed to inpatient costs. The remainder is scattered across ofter areas, but the increase in inpatient spending, while important, was not the primary driver of increased health care spending during that time, according to this paper.


      • Think you’re misreading steve – hospital stay, and testing during visits to the doctor – that is both inpatient and outpatient. My claim that hospitals are the biggest glass houses of course is that 1/3rd of all hcare costs are hospital based.

        • Insurers will tell you that the actual medical claims they pay for fall into three buckets which break down as follows: 40% for hospital based care, inpatient and outpatient combined; 40% for physician fees and clinical services including imaging, labs, physical therapy, etc. and 20% for prescription drugs.

          Medicare spending breaks down roughly as follows. For 2017, the numbers were 42% for Part A, 44% for Part B and 14% for Part D. Of the Part B spending, three percentage points was for specialty drugs to treat cancer and certain other diseases. Reallocating that to prescription drugs would make the breakdown 42% for Part A, 41% for Part B and 17% for prescription drugs.

        • Why the emphasis on inpatient care when outpatient spending is a larger expense and has been growing faster than inpatient costs? Your article is titled “Expensive Hospitals” but it certainly looks like you are including outpatient costs that are not necessarily coming from a hospital. Using this article I don’t see how you determine that, unless your reference was something different.


Leave a Reply

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