Baicker and Chandra go to the Federal Reserve

Today’s Managing Health Care Costs Indicator is $247,000

Katherine Baicker and Amitabh Chandra, both of Harvard, gave a paper to the Federal Reserve meeting in the Rockies last week.  The paper got a reasonable amount of press – but most of the focus was on the two pages where they challenge the conventional wisdom that accountable care organizations will necessarily lower health care costs.    

That coverage was accurate – but the paper was dramatically richer.

The title, “Aspirin, Angioplasty, And Proton Beam Therapy: The Economics Of Smarter Health Care Spending”  is a good place to start. Baicker and Chandra make the important point that we are purchasing high tech expensive medical care (like angioplasty and proton beam therapy), often when they haven’t even been shown to improve care.  On the other hand, it’s hard to get us to embrace inexpensive low technology innovations like aspirin to prevent heart attacks, or handwashing to prevent surgical infections.

They point out graphically that small incremental investments in low technology (aspirin and handwashing) could have huge health care benefits, while large incremental spending on high technology (angioplasty and proton beam therapy) would have only small benefits. A 1990s evaluation suggested that medical advances leave us currently paying about $247,000 per quality adjusted life year saved.

Curve A below represents appropriate productivity efficiency in health care, where investments are first made in low tech high return items like handwashing and aspirin.   This is a conventional economics efficiency frontier – each dollar is promoting further value, but the value declines with more investment as the marginal returns diminish.   Curve B represents an economists nightmare – where investments are prioritized to high technology which itself is either unproven or not shown to be of huge value, and later investments are made for the high value (but inexpensive) interventions like proton beam therapy for prostate cancer.  As you can see, each additional dollar does yield more social benefit – but we end up allocating extra dollars to health care, and we neglect schools or roads or other social needs.

Click image to enlarge.

Other key points from this paper:

• Expert opinion health care often cites that 30% of health care spending is % waste, but it’s hard to remove that waste
• The federal government’s tab is $250b annually to provide tax subsidies for employer sponsored insurance
• Americans have historically had first dollar coverage, which leads to more moral hazard and can lead to overuse of less valuable care.  Of course, we’ll see how this changes with the advance of high deductible health plans.
• Health insurance is “social insurance” which redistributes from the healthy to the sick.  For all the talk about accountability, we really don’t want to disrupt this redistribution.
• Income tax rates would have to increase by 70% to fully fund the cost of health care if it continues to increase at a rate 1% greater than overall inflation.  This type of income tax increase could lead to reductions of 3-14% in GDP. I found this number especially sobering.
• The authors point out that as long as Medicare and the FDA cannot consider cost when they determine coverage and approval, we will purchase lower value health care. 
• Information is a public good, and will require government investment to subsidize comparative effectiveness research.

Baicker and Chandra conclude that there are a few important steps to take to encourage smarter spending on health care

-        Public payers (Medicare and Medicaid) should bundle provider payments  

-        Patients should have more cost-sharing – but it should be nuanced to encourage more attention to the value of care

-        We should provide patients with far better information about the cost and quality of the care that they could receive.

Provider Payment Reform is Key to Encourage Disruptive Innovation in Health Care

There is a lot of great innovation in health care – but not nearly as much disruptive innovation as we’d expect in an industry that represents a sixth of the American economy.  Disruptive innovation can bring huge social benefit – in terms of better good or services at lower prices available to more and more people.  However, disruptive innovation can be terrible for incumbents – and it’s only nurtured in environments where those making purchasing decisions are intensely price conscious. 

So – if we want more disruptive innovation, we need more price-consciousness in health care.

And we’re getting it, too.  As Drew Altman notes in his most recent Kaiser Family Foundation column, high deductible health plans are growing rapidly in the US – and it’s been almost under the radar.  Those with high deductible health plans use far less resources – sometimes skipping valuable care as well as discretionary care. 

Higher patient exposure to health care costs is likely to lead to more fertile territory for disruptive innovation.  For instance, if all office visits cost a $15 copayment at the point of service, patients would not flock to retail clinics, with their limited menu of services.  When patients are paying the first $2,000 or more for their care, and then 20% for care beyond the deductible, they’re much more likely to “shop” for better bargains – which is great for disruptive innovation.  Hospital emergency departments offer many services that most people with earaches don’t need, and higher member cost-share discourages overuse of expensive sites of care.

But patient cost share only works to encourage more prudent purchasing behavior if

-        The health care service is elective, where the patient has a choice. 

o       It’s pretty hard to convince an ambulance driver to take you to the most cost-effective hospital if you feel an elephant sitting on your chest, are out of breath, and are drenched with sweat!

-        The patient has some idea of the cost. 

o       Most physicians have no idea what various health care services cost, and even  health plans often aren’t sure of the actual cost of a service.  This is even worse because services are not bundled in ways patients find intuitive.  Even if I knew that my physician would charge $150 for a level 4 office visit, I’m not sure that she would bill that level of office visit, and she might order laboratory tests that would increase the cost dramatically

-        There is some signal available about quality

o       No one would recommend going for the cheapest angioplasty if the cardiologist had a history of bad outcomes.

-        There is choice

o       In many rural areas, there is a single hospital and often a single group in each specialty.

The biggest problem is that a small portion of patients represent the vast majority of health care costs (10% of patients represent about 60% of all costs). I recently saw data where those in the top 50% represented 93% of total costs (averaging over $3000 per person).  These patients will almost always exceed the annual deductible, so all this extra “skin in the game” might dishearten those with severe illness, but it’s not so likely to change their behavior.  If the extra member cost-share decreased utilization among this half of the population by a factor of half, it would only lower overall health care costs by 3.5% - less than the inflation rate for a single year! Much of the care received by the low cost half of the population is preventive, so we really don’t want to decrease that anyway!

In many instances, the purchaser of health care services is frankly not the patient – but is the health care provider. For example, patients are not generally shopping for the highest value hip prosthesis.  They “shop” for an orthopedist, and the orthopedist recommends what artificial hip should be implanted.  

That’s why I believe that bundled payments and provider risk are the real keys to increasing disruptive innovation in health care. 

Bundled payments have been very effective at increasing disruptive innovation in the provider community in the past.  For instance, when Medicare switched to the diagnosis related group (DRG) payment methodology, hospital stays got dramatically shorter, and total days in the hospital plummeted.  Essentially, the physicians who demanded longer hospital stays under a “cost plus” payment system figured out how to cut hospital stays when they were asked to be prudent purchasers of hospital services.  Independent practice associations and other provider organizations taking “risk” or capitation have put in place innovative programs to diminish hospital admissions and readmissions, and dramatically increase use of generic medications. 

Disruptive innovation is nourished by value-based purchasing, and government can do two things to increase value-based purchasing.

The first is to help us know what really works, and what doesn’t.  That’s why it’s critical to increase our investment in comparative effectiveness research – whatever we want to call it. We’re not talking about rationing or death panels – we’re talking about collecting the data we need to know what’s worth paying for.   

The second thing is, can we please allow the FDA to make determinations about approvals based on value rather than just effectiveness?  As long as we approve a new medication or medical device that is ten times as expensive as existing medicines, and is merely “noninferior,” pharmaceutical and medical device companies will focus their research efforts on innovations that will increase the cost of care, rather than those that could substantially increase value, and sometimes even lower costs.  

The Managing Health Care Costs Indicator will return next week

Tamiflu: Less Effective and More Dangerous Than Initially Believed

Today’s Managing Health Care Cost Indicator is $10 billion

That’s how much the world spent on influenza preparedness in 2009; about $4 billion was spent in the US alone.  Two billion dollars were spent on purchase of Tamiflu (Olseltamavir).

However, there is increasing evidence that Tamiflu has more serious neuropsychiatric side effects than initially thought, and the evidence of its effectiveness appears to be overstated. There was a well-referenced review of this in the current issue of New York Review of Books. 

The initial approval of Tamiflu was based on a manufacturer-sponsored metaanalysis of ten studies, 

only two of which had been published in peer-reviewed journals.  Researchers associated with the Cochrane Collaboration initially published a positive review of the medicine, but have more recently been rebuffed in their efforts to obtain the primary data from those three studies.  The author of the metaanalysis has ‘lost’ the data, and  the pharmaceutical company which markets Tamiflu in the US has only been willing to share summary data. 

There are powerful reasons why initial small studies tend to overstate the effectiveness of a medication. That’s why we need more comparative effective research, which must be continued after a drug is approved for marketing. 

Keeping us safe from influenza, either avian or swine, is pretty important.  The FDA needs to do more to be sure that the drugs we spend billions on are both safe and effective. 

CT Screening For Lung Cancer: Right For SOME Smokers

Today’s Managing Health Care Costs Indicator is 20%

The National Cancer Institute  announced a few weeks ago that spiral CT scan for smokers with at least 30 pack years (Number of packs per day times number of years smoked) lowered risk of death from lung cancer by 20%.  

This is genuinely good news. There is a long history of trials of screening smokers and former smokers for lung cancer showing that the risk of death was higher in the screened group than the nonscreened group, presumably because of the harm caused by biopsies and surgeries in a group with limited lung capacity due to the ravages of smoking.  The current study is huge (53,000 patients), and it is well-designed.  It was not funded by any party with a vested interest in a positive result. I’m surprised at the results – but good science doesn’t always confirm our prior beliefs.

Here’s bad news, though.  One imaging center in California put out a press release the same day saying “"It is clear that in patients at risk, particularly those who have smoked for over 10 years, this is an indispensable part of your annual examination.”   

This misstates the results of the trial significantly, and would lead to a large number or spiral CT scans, with much expense and radiation exposure. Screening can be effective with a high prevalence of disease.  When the same screening is applied to a lower risk group, the risk of false positives climbs dramatically.  At some point, the cancer from excess radiation outweighs the benefit of potential earlier cancer detection –although we don’t know what that point is. The New York Times had a well-balanced article on this. 

The medical director of that center said “We know from our own anecdotes that we have saved a lot of lives.”   That’s a highly unscientific statement.  Anecdotes don’t tell us we’ve saved lives – well designed trials do.  We need more comparative effectiveness research.  Once we’ve done the research, we should provide coverage for effective diagnostics or therapy.  We should NOT pay for services that have not yet proven to be effective except as part of a trial to determine whether these services bring benefits to patients.

Disconnect between knowledge and clinical practice

Two articles published this past week demonstrate that there is a striking disconnect between publication and dissemination of knowledge.

Last week, the Wall Street Journal had an article on angioplasty with stents.   The COURAGE study in the New England Journal (2007)  showed that angioplasty (WSJ estimated cost $15,000) gives slightly quicker relief from chest pain of angina, but does not lower the risk of heart attack or death.  In fact, the stock price of Boston Scientific went down by 23% the month the study was published.   However, the rate of angioplasty has continued to increase after a brief dip.   The evidence was in – but this did not lead to a change in practice.

The Wall Street Journal conclusion is that comparative effectiveness doesn’t work.  I think this shows it didn’t work – not that it can’t work.

Yesterday’s New York Times  has an article about the rapid adoption of the daVinci robot to do prostate cancer surgery.  In fact, the only study done shows that those getting laparoscopic or robotic surgery appear to have more incontinence and erectile dysfunction than those who have traditional “radical” prostatectomy.   (This study combines laparoscopic and robotic surgery, and infers complications from claims) The urologists focus on a 40 year-old policeman who was able to have sex a few days after his procedure. Men facing prostate cancer find that heartening, but this is an unusual prostatectomy patients whose experience is not generalizable to most such patients.  This anecdote is certainly not enough to be the foundation for public policy.

What gives?

Periodically, we hear complaints that it’s difficult to disseminate innovation in health care.  I’d say that this is the wrong diagnosis.  Innovation is speedy when it leads to higher profits and more margin opportunity. That’s the case study of the daVinci robot for prostatectomy.  Dissemination of innovation and knowledge is painfully if that knowledge leads to lower margins and less profit opportunity.  Hence, the message that angioplasties with stents don’t offer that much benefit to those with stable coronary disease has little influence on clinical practice.

Should we give up on comparative effectiveness research?

Absolutely not.

It’s critical that we have evidence to determine what is best practice.   We also need to align payment with evidence-based medicine.   We need to decrease the profit margins of procedures with unclear incremental benefit. This is not easy to do, of course.  Cardiologists,hospitals and medical device makers don’t want to lose margin of angioplasties, and they will argue forcefully that angioplasties are far better than medical therapy for a select group of patients – those with unstable angina or acute evolving heart attack.  That argument is correct, and perhaps we need different fees based on diagnosis since the value is different based on diagnosis.

Similarly, urologists and hospitals will argue, correctly, that for surgeons with a long track record of performing robotic surgery – that method is indeed likely to be better, and can even decrease resource cost by saving OR time.  (Most surgeons, with less than a hundred of cases behind them, take more time with robotic surgery. It’s likely that the ‘learning curve’ is one of the reasons that the robotic surgery study results were so disappointing.)  Some might suggest differential payments for surgeons based on their volume – but frankly that’s complicated and might encourage aggressive surgery recommendations.  My preference would be a bundled rate –and if the hospital and physician feel that a high-capital-cost item will be worth it – they can spring for it. 

The issue of high fixed costs and low marginal costs with new technology also looms large.  Once a hospital has a daVinci robot, there are very few incremental costs associated with increased volume. Therefore, once technology is in place, it is highly likely to be used.

We need to do more research on comparative effectiveness, disseminate the results quickly and effectively, and consider results was we design payment and incentive systems to drive more value for our patients and for health care purchasers.

By the way, the governor of Massachusetts has announced a wide-ranging plan to regulate health care cost increases. I'll be planning to blog on that later in the week. 

Comparative Effectiveness: Discordant Drumbeats

There is increasing interest in comparative effectiveness research – and the Annals of Internal Medicine has an editorial this week pointing to the importance of knowing the real value of what we are paying for.   (Harvard Link) The stimulus package and the 2010 budget proposal both envision a large federal investment in such research, and the Congressional Budget Officehas even suggested that such research will save health care dollars (although perhaps not as many dollars as the research will cost).   On the other hand, a heartfelt opinion piece in the Boston Globe last week by the CEO of the Society for Women’s Health Research, points out that what is good for a population might not be good for all individuals.   Phyllis Greenberger says:

 

As the American comparative effectiveness agency is assembled in the coming months, administrators must take into account the personal needs of individual patients. If the council were to primarily focus on cost effectiveness, it would likely only consider the "average" patient. But in medicine, every patient is unique.

 

So – here’s a dilemma.  It will be difficult (or impossible) to make cost-saving decisions that will not make anyone feel like they were given every chance.   See a previous post on the woes of the National Institute for Clinical Excellence NICE in the UK.  l Lowering health care costs means standardization and sometimes making tough choices and tradeoffs.

 

Another dilemma raised by opponents of using comparative effectiveness research to determine what should be covered is that costs decrease when there are competitors for effective innovations. There is an article in this week’s Annals of Internal Medicine (Harvard Link)  reviewing cost-effectiveness of cholesterol-lowering therapy to prevent heart attacks.  Cost-effectiveness is hugely dependent upon price.  Before it became available generically, Zocor cost over $3 per pill, meaning that it was not cost-effective to use on even a small portion of the “at risk” population.  On the other hand, generic simvastatin can now be obtained for only 10 cents a day,  making it cost-saving (not merely cost-effective) for all patients with LDLs over 130 (a majori ty of the population).  If we had not approved the use of Zocor at $3-$4 a pill a dozen years ago, we would not now have available one of the few cost-saving interventions in adult medical care.

 3/9/09 Update: Good column in Newsweek today noting the irony of a physician-legislator's opposition to science guiding medicine. 

WSJ op-ed assails comparative effectiveness research

The January 20 Wall Street Journal included an op-ed from the American Enterprise Institute decrying the potential for a government-funded comparative effectiveness institute.

In this opinion piece, Scott Gottlieb suggests that such a government-funded comparative effectiveness institute
(1) Will not save money
(2) Will use poor scientific methods
(3) Would do research that would be effectively done by the private sector if only the FDA would allow private companies to publicize their comparative effectiveness findings.

My response:
(1) Here are the CBO comments from December, 2007.

Generating additional information about comparative effectiveness and making corresponding changes in incentives would seem likely to reduce health care spending over time—potentially to a significant degree. The precise impact, however, depends on several factors and is difficult to predict. Given the time necessary to conduct the research, to alter incentives in a manner reflecting the results, and to affect behavior through those changes, any potential for substantial cost savings from new research would probably take a decade or more to materialize. Even so, generating additional information comparing treatments would tend to reduce federal health spending somewhat in the near term—but that effect may not be large enough to offset the full costs of conducting the research over that same time period


(2) There is some thought that we might have to settle for research that wouldn’t merit publication in the New England Journal. Question – isn’t some information even if imperfect better than the current state of utter lack of information?

(3) The CBO points out that private parties just don’t have the right incentives to do good comparative effectiveness research. Pharmaceutical and device makers are not likely to be impartial enough (Here’s an example. This article showing that a medicine was ineffective was published in Annals of Internal Medicine 8 years after the completion of data collection. The publication was delayed until long after fluconazole, the drug in question, lost its patent protection). . On the payer side, there is no single health plan (except perhaps Medicare) representing enough of the market to take on this cost.

Comparative effectiveness research is expensive and takes a long time. The UK's National Institute of Clinical Effectiveness faces serious opposition to its efforts to restrict coverage to more cost-effective therapies. (See my previous post on this issue). Doing good comparative effectiveness research could help allocate precious (and not limitless) resources. This research won't happen without government participation, and probably won't have much impact as long as government payers are prohibited from using this information in coverage decisions. I believe we should fund this research through the Agency for Health Research and Quality, and governmental and nongovernmental payers should be able to use this information when designing coverage.

Thanks to Ben Geisler from our class for sending me a link to this article.