Friday, October 9, 2015

Mammography Doesn't Reduce Breast Cancer Deaths and Computer Aided Diagnosis Doesn't Help

Today’s Managing Health Care Costs Number is $400 million

It’s not been a good month for mammography. Two important pieces of research were published that cast doubt on the overall  benefit of mammography, and on computer aided detection to augment digital mammography.

Researchers from Dartmouth and Harvard reported in  September’s JAMA Internal Medicine that counties with higher rate of mammography have more diagnosed breast cancer – but they have breast cancer mortality which is indistinguishable from counties with low rates of mammography.  Screening finds many more cancers, and many more cancers are treated, but that doesn’t appear to have any impact on death rtes.  There were many more small breast cancers found in the high screening counties (25%), but there was also a statistically significantly larger number of large breast cancers found (7%). 

This doesn’t prove that mammograms don’t sometimes save lives.  The study only covered ten years, and the benefit could have been later. It just shows that over the American population the amount of lives saved by mammography are very unlikely to be large.   Further, it casts doubt on the massive effort to increase mammography screening.    

Researchers in JAMA last week showed that mammograms augmented with computer aided detection (CAD) did not improve breast cancer diagnosis compared to mammograms without this additional service (and fee.).  CAD was associated with statistically insignificant lower sensitivity and lower sensitivity for invasive cancer. It did find significantly more cases of ductal carcinoma in situ; the treatment of DCIS is controversial and it’s likely that much of this is overdiagnosis.  CAD technology is used on 83% of all mammograms, and costs insurers $400 million. 

An accompanying editorial advocated that Medicare and insurers stop paying additional fees for CAD – but there are two problems.  It’s very hard to stop paying for something that’s been a covered service for 17 years – and we wouldn’t want to see radiologists billing women directly for these services. The other problem is that for radiologists mammography isn’t especially lucrative, so pulling this fee could decrease access.  On the other hand, research shows us that we should probably do fewer mammograms than were recommended in the past, so perhaps this is just fine.

From the JAMA editorial:

Congress should therefore rescind the Medicare benefit for CAD use. If we could curtail use of many similarly ineffective tests and interventions, we could significantly reduce US health care expenditures while augmenting resources for effective care or well-designed studies of promising innovations. Savings from a single year of CAD use, for example, could fund a much-needed clinical trial comparing screening outcomes with mammography vs mammography with breast tomosynthesis. The lesson of CAD is that broad societal investment in new medical technologies should occur only after large-sample evaluations prove their real-world effectiveness and justify their costs.

Thursday, October 8, 2015

Pharma Rent Seeking

Today’s Managing Health Care Costs Number is 538%

The blowback against some of the financial engineering pharmaceutical companies, which buy up “old” drugs and dramatically increase the prices, continues.

The New York Times had two pieces on this over the weekend.  The first article was an in-depth review of Valeant Pharmaceuticals, which has become a multi-billion company by purchasing existing companies, firing most of their staff, driving up prices, and watching the share price grow.   The article starts with a patient who has been on Cupramine, a drug that treats a copper overload syndrome, for 55 years.

Gretchen Morgenstern provides a few key statistics about Valeant
·         The company spends 3% of its revenue on research and development. The average across pharma is 15-20%
·         Recent acquisitions include
·         Medicis: Fired 750/790 employees
·         Bausch and Lomb – fired 3000/4100 employees
·         Salix-  intends to fire 420 of 977 employees

Valeant’s move to Canada and decreased taxes has helped it turbocharge its acquisitions;  even after losing  a third of its value since its peak earlier this year, the stock has returned 25x the return of the Standard and Poor’s Index. It paid its five top executives about half as much as it spent in total on research last year.

James Surowiecki in the New Yorker has a name for the pharmaceutical industry increasing the cost of old, cheap, generic medications --, rent seeking.  Essentially, the pharmas that do this are seeking to get payment when not providing any additional value.  He notes that there are two potential approaches short of government price controls:

·         Allow importation of foreign supplies manufactured in developed countries with robust regulatory infrastructure through regulatory reciprocity 
·         Prohibit closed distribution, where the manufacturing pharmaceutical company uses a very limited distribution network to make it difficult for a competing generic company to get samples of medication to do reverse engineering

Although all eyes have been on Turing Pharmaceuticals, where hedge fund maven Martin Shreki gained enough infamy to be parodied on Saturday Night Live by raising the price of a 62 year old generic antiparasite drug used in HIV treatment by 5000%, and Valeant.  But the overall problem of affordability in American health care is affordability, and the mainline pharmaceutical companies have also been vigorously raising unit prices of their medications to increase their profit margins.   

Sunday, October 4, 2015

In Defense of Fee For Service

Today’s Managing Health Care Costs Number is 90%

There’s a lot of momentum behind an effort to decrease our reliance on fee for service as a payment methodology.   Fee for service drives higher utilization, and high margins for high technology services (which have low variable costs) lead to sometimes-dramatic overuse of these services.   Anyone who says we don’t know how to embrace innovation in medical care hasn’t looked at the rapid adoption of PET scans or IMRT (intensity modulated radiation therapy).

I’m a big fan of Catalyze Payment Reform, an employer coalition that aims to push 20% of provider payment to value based payment by 2020. (More on “value based” in another post).  CMS has announced that it wants 90% of hospital payments to have a value based component by 2018. But even those reimbursements will largely be built on a foundation of fee for service. 

I want to devote today’s blog to a recitation of why we shouldn’t seek to banish fee for service.  It’s an important element of within a portfolio of payment methodologies – and there are good reasons to pay fee for service to some providers for some services.    Fee for service should diminish as the payment methodology in the US – but we shouldn’t seek to banish it.

Fee for service is the “better” method of payment for

·         Services which are high value but currently have low utilization.  This includes primary care services and immunizations and colonoscopy.  Does anyone think that including colonoscopies in a capitation makes it more likely that the affected population will get more colonoscopies?

·         Services which represent new and unanticipated advances – and which weren’t anticipated when a budget was created.    Hepatitis C treatment should be made readily available to those with Hep C.   Including this cost in a global budget would induce lower utilization, which is contrary to public health good.

·         Unusual medical care – like treatment of certain orphan diseases, or multiple organ transplants, should ideally be “carved out” of a capitation. They might be included in a bundle –so there is no obligation to pay for every gauze pad used or every microgram of a biologic medication.  But this type of care doesn’t fit well into budgets, and failure to carve these cases out is likely to lead to “dumping” of patients who can benefit from these services.

·         Care delivered in rural communities, where the population is not large enough to achieve actuarial stability.

·         Care delivered by fragmented providers.  Again, there can be value elements in payment (such as patient centered medical home payments, pay for performance, and even shared savings which must consider statistical validity.  But fee for service almost certainly has to underlie this value-based payment

·         Care delivered by physician organizations at academic medical centers.   These centers naturally attract an especially sick population, and risk adjustment just isn’t adequate to fully address this issue.  It’s possible to create a population budget based on past experience – but any population budget encourages the group not to welcome the sickest members into their risk pool.    But that’s why these physicians are working at an AMC in the first place!

Many non-fee-for-service payment methodologies require some fee for service element – and fee for service claims data helps us evaluate whether there is stinting on medical care.  Fee for service claims data is also still a major source of quality data.   

The top diagram above shows the desires of many health policy experts – to dramatically reduce FFS, or even eliminate it.   I believe that we instead want the second diagram, showing that we will re-equilibrate fee for service -  yet it will remain an important element of the portfolio of payment mechanisms.

Some past posts on this subject:

Thursday, October 1, 2015

Diesel Deception : Pollution costs lives and health care dollars

Today’s Managing Health Care Costs Number is $80 million

The scandal enveloping Volkswagen, AG, which has programmed its diesel autos to mislead emission checks and cause as much as 40 times more nitrogen oxide (NOX) pollution as advertised, is a reminder to us that environmental issues are themselves a driver of health care costs.  

Michael Greenstone of MIT calculated that this additional pollution was probably responsible for about $80 million in extra medical costs since 2009.  The New York Times estimated that the excess pollution led to about 106 extra deaths.  Neither the costs nor the death tolls are huge in the context of our economy and our population.  But, still.  People died who didn’t have to, and health care cost more than it should have.  

The Volkswagen story will stay in the news for years, as the company faces massive EPA fines and substantial liability from Volkswagen owners as well as shareholders.  Air pollution has been shown to increase asthma, obstructive lung disease, and heart attacks, and cleaner air has added as many as 5 years to the lives of newborns in regions of West Virginia.

There is a tragedy playing out in Flint, Michigan, that will get far less attention – but will gravely impact children there for the rest of their lives.   The city began to draw drinking water from the Flint River – and the water was corrosive.  It leached heavy metals from pipes and seams, and delivered high doses of lead to unsuspecting families.

Lead is an awful toxin – it leads to severe loss of intelligence in children, and brain lead concentration is probably responsible for a substantial amount of violence. Many  researchers believe that the dramatic drop in crime since the 1980s was largely due to a decrease in lead poisoning.  Prohibiting leaded gas meant that lead levels in the air decreased by 92%.   The new and totally preventable lead poisoning in Flint will cost millions in health care dollars, and will hurt residents and those around them for decades or longer.

Monday, September 28, 2015

CBO points out that defunding Planned Parenthood will be costly to feds

Today’s Managing Health Care Costs Number is $130 million

There’s very little within the medical care that saves money. Much of the care we deliver is cost effective – meaning that we get quality adjusted life years (QALYs) for a reasonable price.  We don’t get QALYs and cost savings for many medical services.

Immunizations are a shining counterexample.   Each dollar we spend on childhood vaccinations saves as many as $21 dollars in medical claims costs. The list of cost saving medical services besides immunizations is scant.  Statins are not cost saving; neither is cancer screening.  

The Congressional Budget Office has reviewed Republican plans to defund Planned Parenthood – and they’ve reported that the next impact of disallowing any federal or Medicaid funding of services at Planned Parenthood would be increased cost to the federal government of $130 million over ten years – as opposed to savings.  This is largely because more women denied access to effective contraception will have babies, and 45% of deliveries in the country are paid for by this federal/state program.

The CBO analysis understates the potential for increased costs, as it doesn’t consider costs borne by states or by individuals.  The CBO assumed that over time women would be able to find alternatives sites of care – which is optimistic.  Planned Parenthood provides services to 2.4 million – over a third of those who on government programs who receive contraception. There are simply not providers other than Planned Parenthood with the capacity to provide women’s health care to millions of Americans, especially in rural areas.  So the CBO estimate of increased costs is likely low. 

Federal funds that are directed to Planned Parenthood help us control the federal deficit!

H/T to Danielle Paquette who reported this for Wonkblog last week.

Thursday, September 24, 2015

Hillary Clinton’s Plan for Controlling Out of Pocket Costs: Hit and a Miss

Today's Managing Health Care Costs Number is 3

Kaiser Family Foundation and the Health Research Educational Trust released their annual report on employer sponsored health insurance on Tuesday - and the headline is that employee out of pocket costs continue to increase out of proportion to medical inflation.

·         Premiums for employer sponsored health plans increased by 4% - substantially more than workers’ income (+1.9%) and general inflation (-0.2%)
·         The portion of firms offering health insurance hasn’t changed significantly from last year.
·         Worker contribution to health care costs increased more than employer contribution, although employers continue to pick up the vast majority of health care costs.
·         63% of workers at small firms and 39% of all workers have single deductibles of $1000 or more;
·         13% have single deductiles of $6000 or more.

Hillary Clinton came up with a two-pronged plan in short order to address the issue of escalating out of pocket costs. The first addresses patients who individually have catastrophic out of pocket costs, and the second offers relieve to relatively healthier people who are never likely to reach their deductibles, but who hate to be paying first dollar for those first few non-preventive office visits.

Clinton proposes to offer a refundable tax credit (up to @$2500 for individuals and $5000 for families) to those who pay more than 5% of their income on out of pocket health care.  She would fund this by increasing taxes on the rich and on pharmaceutical companies.   As Jonathan Allen of Vox notes in an explainer, this would require Congressional action, which is unlikely even in the first term of a Hillary Clinton presidency barring an unexpected Democratic wave election.    From a public policy perspective -this offers the same type of out of pocket protection that the Affordable Care Act already affords to those with silver level exchange plans and income under 400% of the federal poverty level.   It's a good idea - and the benefit would predominately go to those with serious or chronic diseases.

Clinton also proposes that three non-preventive visits a year be covered in full before patients have hit their deductible.   This will impact a much larger portion of the insured population – but it’s more troubling. 

·         This will be hard to administer.   The cost of the office visit often pales compared to the cost of related diagnostic studies – and if the diagnostic studies are not included there will be many folks who are deeply disappointed.
·         This will increase costs – Clinton’s campaign says $100 per person per year although it could be substantially more than that
·         Health plans will have to strip out costs from the care of those with more substantial illnesses to stay within the actuarial value that they have promised
·         There is a timing issue- this needs to be done at last 18 months in the future, to allow integration into the medical cost and therefore premium
·         Employers facing the 40% excise tax will be especially vigilant about offsetting this new coverage (for pretty healthy people) by cuts in coverage (which will have to come from the coverage of those with higher medical claims).
·         Front end deductibles do lead more reluctance to seek care, and should lower the utilization of low value care.   Office visits have been going down in recent years, perhaps because of deductibles.  Making the first three visits exempt from deductibles will diminish

Rising out of pocket costs are a real problem for Americans with modest incomes– and it’s great that there is more attention to this issue. The Clinton plan might be a starting point, although the plan to offer a government subsidy for high out of pocket costs relative to total income makes good policy sense but is politically dead in the water.  The plan to add “free” sick visits is more likely to prevail in the political world, but it’s more problematic and could paradoxically lead to less rich coverage for those with serious illnesses, especially if the excise tax is not overturned. 

Tuesday, September 22, 2015

ACOs decrease low value services (a bit)

Today’s Managing Health Care Costs Number is 4.5%

Researchers from Harvard Medical School reported in JAMA Internal Medicine yesterday that Medicare Pioneer Accountable Care Organizations (ACOs), which accept the largest amount of financial risk for caing for their designated populations, decreased the use of low value services compared to colleagues in their geography who were not in Pioneer ACOs.   The 31 low value services were from the American College of Physicians “Choosing Wisely” campaign, from US Preventive Services Task Force “D” recommendations, Canadian health technology assessments, and the published literature.   All were by definition determinable by claims.

The researchers found that on average the Pioneer ACO patients received these services about 5% less than patients attributed to the non-ACO patients during the first year of the Pioneer ACOs (compared to just 4% a year earlier).   The research included 700,000 ACO member years and 17 million non ACO member years – and the researchers used a “diff in diff” approach to compare the ACO and control groups.    They adjusted for sociodemographics and  illness.   

This is good news – although by no means definitive proof that ACOs will add value.  The researchers did not evaluate the use of high value services, so we don’t know if the total value of care increased.  The cost decrease in the Pioneer ACO was lower in year two, but that data was not yet available to the researchers.  The big issue is that the actual change was pretty small.  I’ve recalculated from the provided data the use of low value services in the ACO and nonACO group for the graphic above. The researchers note that the cost of low value services was $256 per beneficiary – so a 1.9% difference in decline in low value care would yield under $5 bucks per beneficiary.  This could be a signal of high value care throughout other categories, in which case ACOs will prove hugely beneficial.  These claims codes are also somewhat “gamable,” so that if a low value care index was in widespread use, it’s likely that providers would choose to do claims coding to maximize adjustments and gain the best “score.”

In an accompanying editorial, Arnie Milstein of Stanford notes:

Comparisons of US health care spending with spending in other wealthy countries suggest that lowering population-wide health care spending and then perpetually slowing its growth without impairing quality of care will require more than decreasing low-value services. Decreasing spending will also require economically preventing costly health crises and lowering the cost of producing each unit of service.

We need efforts to lower the use of low value services – and this research shows that Pioneer ACOs appear to have done that in year one.  However, we’ll need to address unit cost and prevention of health care crises to obtain the kind of health care cost savings that we need.