Friday, April 18, 2014

Mammography and Numeracy

Today’s Managing Health Care Costs

Number is 1

Here’s a radiologist’s letter to the editor in the Boston Globe from Wednesday:

Mammograms have stood, and aced, test of time
RECENTLY, MUCH has been written, both pro and con, about routine screening mammography in women over 40 and its role in cancer detection and survival (“Doctors may oversell benefits of mammograms,” April 2). When it comes to all cancers there are two truths with no disagreement: First, the earlier a cancer is found, the greater the likelihood for cure. Second, the standard method to evaluate treatment success or failure is the five-year survival rate.

In 1975, when I started performing screening mammograms, and before mammography came into wide use, the five-year survival rate for breast cancer had been stuck at 35 percent. This means that for every 100 women diagnosed with breast cancer, only 35 were alive at the end of five years.

Mammography is capable of finding breast cancer when it is about the size of a grain of rice. Before mammography, palpation was the only way to find breast cancer. A breast lump had to be at least 1 centimeter before it could be felt. A 2 to 3 mm cancer had to grow by a factor of 4 before the diagnosis was made.

By 2005, the five-year survival rate for breast cancer had risen to 90 percent, and only a fraction of this remarkable increase in survival can be attributed to drug therapy and treatment changes.

Screening mammography works, and, statistically, results in 55 more women alive at the end of five years than in the pre-screening years.

I rest my case.


This drives me crazy.  Five year survival rates are terribly misleading, because “lead time bias” rules.   If two patient each die of the same type of cancer with similar lethality on the same day in 2020 – but one had a cancer diagnosed through screening in 2010 and the other in 2017, the first survived more than five years with her cancer.   But the earlier diagnosis didn’t save a minute of her life – and might have led to more years of therapy that diminished quality of life.   And that’s saying nothing of the many women who might have required followup tests to rule out cancer, or who were “overdiagnosed” and treated for cancers or precancers that would not have caused them harm.

Further, there is strong evidence in the medical literature that a substantial portion of improved real (not just apparent) survival rates from breast cancer are from advances in chemotherapy and hormonal therapy.   The rate of women dying of breast cancer, though, has not changed a whole lot in the era of mammography.  We are diagnosing many more early stage cancers, but we should only celebrate this if it genuinely increases women’s survival chances.
The figure shows the numbers of 50-year-old women in the United States expected to be alive, to die from breast cancer, or to die from other causes if they are invited to undergo regular mammography every 2 years over a 10-year period, as compared with women who do not undergo mammography.  Source

The New England Journal published a perspective article from Switzerland regarding mammography yesterday.  An ethicist and an epidemiologist who participated in a recent panel to make recommendations about screening for breast cancer point out that it was “nonobvious…that the benefits of screening mammography outweighed the harms.”   Among the panel’s conclusions:

1. It is not recommended that systematic mammography screening programs be introduced.
2. A time limit is to be set on existing systematic mammography screening programs

I especially like the graphic that the authors presented (above), which shows the subjective sense (from surveys) of likelihood that mammography will save lives compared to the literature-based likelihood.   Mammography might save one life per thousand people screened – although it appears to have little if any impact on overall population mortality.  I’m hoping a picture is worth a lot of verbage – and might be one way to combat our innumeracy.

Here's a graphic from Esserman, JAMA 2009 showing results of epidemiologic mammographic screening in the US

Note that detection of regional and metastatic disease has not decreased - we're finding an enormous amount of localized cancer.  Over decades, this should lead to a lower rate of detection of invasive disease.  Source 

Thursday, April 17, 2014

Sweet Success: Decreasing Diabetes Complications

Today’s Managing Health Care Costs
Number is 68%

Today’s New England Journal has a heartening report showing dramatic decreases in serious complications from diabetes over the last two decades.  The press coverage has been swooning.

From 1990 to 2010, heart attacks were down (68%), strokes were down (53%) amputations were down (51%), End stage renal disease was down (28%).   It’s striking that the only one of these complications that decreased in the nondiabetic population was heart attacks.  This is likely through a combination of less adult smoking, better diet, statin medications, and improved heart disease treatment. 

Diabetic care has changed dramatically over these two decades.  We now fully appreciate the benefit of “tight control,” and diabetics routinely check their blood sugars multiple times a day.   More use insulin pumps that can improve blood sugar regulation.  There are 4 new categories of antidiabetes drugs, each of which can lower haemoglobin AIc, and some of which actually cause weight loss rather than weight gain.  Insulins made from animal protein and generic insulins have disappeared.  The statin drugs lower cholesterol and also decrease inflammation which diminishes large blood vessel disease.   The cost of diabetic drug treatment has skyrocketed.   Diabetics remain at vastly higher risk for these complications than nondiabetics (relative risk 4 for heart attacks, 3 for strokes, 19 for amputations, and 14 for end stage renal disease.)

This is great news, and good indication that a combination of public health measures (smoking cessation and focus on exercise and nutrition) and pharmaceutical therapies has made a real difference in quality of life for those with diabetes.    All is not happy, though.  The authors note that the increasing prevalence of diabetes and the aging of the population  “suggest that the total burden, or absolute number of cases of complications, will probably continue to increase in the coming decades.” 


Wednesday, April 16, 2014

Will health care inflation come roaring back?

Today’s Managing Health Care Costs

Number is 5.6%

We’ve all been pretty excited about the decrease in medical inflation – even if we can’t say we understand the reason why.   This moderating trend is why the CBO has lowered the projected cost of the Affordable Care Act.  Lower health care cost trends should free up government billions for worthwhile projects like infrastructure and education,  and should finally push employers to increase real take home wages.

The decrease in medical trend coincided with the Great Recession – and the debate has been vigorous and partisan about whether the change is structural, at least in part due to the ACA, or whether it is purely macroeconomic, in which case a return to prosperity will lead to a return to high medical inflation rates.

Yesterday’s IMS report showing an increase in utilization throws some cold water on the celebration of the end of runaway medical costs.   Office visits were up by 2.7% (driven by a 4.9% increase in specialist visits); hospital services were up by 13 million (2.6%) driven by a whopping 10.5% increase in scheduled inpatient services, and prescription use went up 1.6%    

IMS began as a data company in the prescription drug space, and their data is most robust for prescriptions. There were 36 new drugs on the market in 2013 .  24 were orphan drugs, and specialty and oncology drugs represented 60% of the new drug spending.  Patient out of pocket costs for the 2.3% of prescriptions with patient charges of over $70 represented 30% of the total out-of-pocket cost – with an average per prescription cost to patients of $145.  ACA full coverage of oral contraceptives decreased out of pocket spending by $483 million.   (A side note - this likely decreased overall medical costs.)  About 10% of prescriptions written were not filled

Sarah Kliff, now of Vox, has a review of various sources showing increasing utilization and cost, including CVS Caremark and the Altarum Institute. The Bureau of Economic Indicators shows that health care costs increased by 5.6% (annualized) in the last quarter of 2013 – the highest inflation rate since 2004, and CMS has estimated that health care costs will go up 6.1% in 2014, largely due to increased coverage.

David Cutler is optimistic, and says that the prolonged decrease in medical cost inflation will mean that rates will not return to their previous heights.  I suspect that we’ll need continued structural changes in health care to keep the rate of inflation down.   We have certainly not conquered this problem.

Selected pharma-related screenshots from the IMS report below:

Note the difference between increased costs and per capita costs. Important to note distinction as we cut the number of uninsured in half over the coming years

LOE is loss of patent protection.  Even costs of new expensive drugs are not overcoming the pharma companies' income loss from generics, which are up to 86% of prescriptions. 

BUT in chronic disease medications, brand pharmas HAVE put in place price increases that overwhelm the generic savings 

New cholesterol recommendations cut push up #2 here

Tuesday, April 15, 2014

CBO finds the Affordable Care Act more affordable

Today’s Managing Health Care Costs
Number is $105 billion

The Congressional Budget Office released its revised estimates of the cost of the Affordable Care Act – and there are two pieces of excellent news. The first is that the total cost to the taxpayers is estimated at $5 billion less for 2014 compared to just two months ago, leading to a savings of $104 billion over the next decade.   The second is that the ACA is expected to cover a million more Americans (even with this reduced price.)

Why are there such lower cost estimates?   The overall rate of health care inflation has moderated substantially – and premiums for exchange health plans have come in lower than expected, decreasing the cost of federal subsidies. 

Interestingly, even the increase in federal savings from the excise tax on high cost health plans is NOT from increased estimate of federal excise tax collected. It’s from expected lower employer health insurance costs that are projected to lead to more employee compensation in the form of income (which is taxed) rather than benefits (which are not taxed). 

There are many false arguments that have been marshaled against the ACA, including “death panels” and “government takeover of health care.”  Opponents have also labeled Obamacare as “job killing.”  The CBO report helps us understand that if we can lower the rate of inflation within health care this health insurance expansion can be affordable, and can even free up more income for workers.  This is especially nice to see on the eve of tax day.  

Sunday, April 13, 2014

Rural hospitals close, leaving their communities more impoverished

Today’s Managing Health Care Costs
Number is 1.6%

Public radio’s Marketplace had a story from rural Sparta, Georgia last week. The Hancock Regional Hospital in Sparta closed in 2001 in part because of the terrible area economy.  The economy dragged the hospital down, but now having no hospital in the community constrains economic growth.

Hospitals represent great employers – they offer among the highest salaries, even to many without advanced degrees. They offer rich benefits.  They sustain nearby businesses, including fast food joints. They give business to local electricians, plumbers and other workmen.  

The Chair of the County Commission said that potential industrial employers don’t want to locate in a county with no hospital.  Studies confirm that communities that lose their hospitals suffer bad economic damage: 

University of North Carolina professor Mark Holmes studied the economic impact of 140 rural hospital closures nationwide.  He found that three years out, losing a hospital costs a community, on average, “about 1.6 percentage points in unemployment, about $700 in per capita income, and that was in [year] 2000 dollars so that’d be probably about $1,000 currently."

We have something like this going in in Western Massachusetts, too. The North Adams Regional Hospital closed this month with less than a week’s notice.   530 employees were left jobless, and less than a third were rehired by Berkshire Medical Center, which intends to offer urgent care and hospice services in North Adams.   Eastern Massachusetts has the vibrant tech centers in Boston and Cambridge, but the mills and plastic factories in Western Massachusetts have long since gone dark. 

Areas with declining populations have a hard time sustaining hospitals, and difficulty recruiting medical professionals. Small and rural hospitals lack the capital for conversion to electronic medical records and other clinical and technology advances.  Communities with more poverty have “poor payer mix,” and low payments from Medicaid programs and larger writeoffs for the uninsured lead to low margins, disinvestment, and a downward spiral of lower utilization.  The Massachusetts hospital that just closed was only 20% occupied.   Hospitals have very high fixed costs, so loss of volume is a death knell.

This is a good illustration of how hard it is to control health care costs; lowering health care costs causes economic pain in virtually every community.  But spending too much money on health care leaves us without the resources to educate our kids and fix our roads.  We need to figure out how to centralize cost-effective delivery of high tech health care, and offer less intensive care over the wide geographic areas which will not themselves have hospitals at some point in the future.  

We will need better use of virtualization technology, since there are few counties in the entire country that can support a pediatric endocrinologist!   We also will need more care delivered by pharmacists and nurse practitioners who can have sustainable businesses in smaller communities.   Efforts to shorten hospital stays and streamline diagnostic workups can save money – and they will also help keep clinical care more accessible to residents of communities that have lost their local hospitals.

Wednesday, April 9, 2014

Medical Leaders on Pharmaceutical Company Boards

Today’s Managing Care Number is $310,000

Last week’s JAMA has an article showing that 94% of US pharmaceutical companies have an academic medical leader on their boards.  This includes 2 university presidents, 6 deans, 6 hospital or health system executive officers, and 7 clinical department chairs or center directors.  The average reimbursement for this board service is $310,000 a year.  

The Atlantic has a commentary on its web site:

How can academic medical centers that prohibit faculty members and trainees from accepting even an inexpensive pen or notebook from a pharmaceutical company—let alone cash—allow their leaders to serve on the boards of firms that pay them on average over $300,000 per year? Is it reasonable to suppose that, if even a small gift represents an unacceptable conflict of interest, a large payment that substantially exceeds the average annual compensation of a faculty physician might do so, as well?

CMS Releases Doc Pay. Let the Sun Shine!

Today’s Managing Health Care Costs

Number is 880,000

Source Wall Street Journal 

The Centers for Medicare and Medicaid Services (CMS) released data on Medicare payments to physicians today – it’s an watershed moment for transparency, and the Wall Street Journal played an important role in court as well as in the newspaper in making this change happen.

There’s good reporting on this in both the WSJ and New York Times. Both the Times and the WSJ well-executed query functions allowing readers to search for Medicare payments to individual physicians. (The Times includes billed amounts, while the WSJ does not).

It’s no surprise that a few specialties represent a large portion of the highly reimbursed physicians.   The elderly represent a disproportionate share of patients of ophthalmologists and cancer specialists – so more of their income is from Medicare than from employer-sponsored health plans. 

This data will be useful to find cases of fraud, and will be useful to infer volume for many physicians.  Commercial health insurance plans can use this to get a fuller picture of the services provided by their contracted providers.   Developers of apps can make this data more convenient, and perhaps “outlier” physicians with unusual billing profiles will change their habits.   I looked up my PCP – who saw about 140 Medicare patients for problem-oriented office visits, and he billed about 70% more level 4 (more complex) than level 3 (less complex) office visits, which seems about right to me.  

There are some caveats to this data. The American Medical Association spokesperson pointed out that physicians hadn’t reviewed their data – but they cashed the checks, and I think that’s enough.  I’m sure that there are mistakes in the specialty attribution, as credentialing databases are notoriously.  Here are my caveats:

·      This is Medicare fee for service data only.  It doesn’t include Medicaid, and it doesn’t include the quarter of Medicare beneficiaries who have chosen private plans (Medicare Advantage)
·      The disclosed data represents billing and CMS payments – and doesn’t consider resource cost.  Many ophthalmologists use Lucentis (instead of much-less-expensive Avastin) for macular degeneration injections; this can cost $2000 per injection – so the high reimbursement to physicians is passed directly to the pharmaceutical company (Roche in this instance).
·      Data is only provided for physicians who had more than 10 claims for a particular service – so it will be harder to get a sense of the work done by lower-volume physicians
·      Medicare uses a fee schedule, so the “billed” amount is essentially fiction – although always disheartening to see what those without insurance be charged.
·       There could be some errors – where an office bills multiple providers under a single provider’s number, for instance.   This public disclosure will be a good way to convince these offices to improve their procedures!  Many corrupt datasets only get fixed when they “matter!”  Data gets better when it is used.
·      This is 2012 data – and might not reflect changes in clinical or billing practices since then.

The next news cycle will be about the Florida ophthalmologist who collected about $21 million from Medicare, and the few dozen physicians who collected over $4 million.  Lucentis treatment for macular degeneration cost Medicare over $1 billion in 2012.  That’s important news – and I think this released data set will ultimately nourish valuable broader efforts to give patients information about the performance of their physicians.  Sunshine is good!