Will Medicare Bother to Learn Anything from Its Own Demonstrations? – Part I

This past week, the Congressional Budget Office (CBO) released two reports summarizing the results of 14 major demonstration projects aimed at improving the “quality” and/or reducing the cost of care provided to Medicare patients.

These fell into two broad categories: disease management and care coordination demonstrations, and “value-based payment” projects.  Both types of projects are of enormous importance to the future of the Affordable Care (“ObamaCare”) Act and the healthcare policies of the current administration.  This is because ObamaCare greatly expanded insurance coverage – and therefore total U.S. healthcare costs – without a proven, sustainable plan for reducing the financial burden of all this new care.  Without some cost-saving miracle, the result will be well over a trillion dollars of new healthcare spending by the federal government alone over the next ten years, (and even more if you disallow the bogus “savings” that would have been produced by the now-defunct CLASS program).  The innovations behind these Medicare demonstrations were to have provided that miracle.

Unfortunately, things aren’t looking so good on the miracle front  Here’s the short version from the CBO:

“The evaluations show that most programs have not reduced Medicare spending. Programs in which care managers had substantial direct interaction with physicians and significant in-person interaction with patients were more likely to reduce Medicare spending than other programs, but on average even those programs did not achieve enough savings to offset their fees. Results from demonstrations of value-based payment systems were mixed. In one of the four demonstrations examined, Medicare made bundled payments that covered all hospital and physician services for heart bypass surgeries; Medicare’s spending for those services was reduced by about 10 percent under the demonstration. Other demonstrations of value-based payment appear to have produced little or no savings for Medicare.”

But we really shouldn’t stop there.   The long version is well worth going through in a little detail.  So much is riding on these results that it would be a crime to simply acknowledge the findings with a headline and blithely launch into yet another series of demonstrations.  After all, our national objective should be to find and disseminate what works, and discard what does not work.  So what have we really learned thus far?  Let’s look at the “value-based payment” demonstrations first.  As shown in the table below, there are four of these.

Click to Enlarge

The philosophy behind of all of them is the same: since doctors, hospitals and other healthcare providers are directly responsible for incurring healthcare costs as a result of the care they provide, all we have to do to save money is reward or penalize their incomes based upon how much they spend to take care of their patients.  Want to curb spending?   Just dangle a carrot or wield a stick in the direction of their bank accounts.  Greed will do the rest, wiping out the inefficiency that the doctors themselves have built into the system.  The CBO’s Lyle Nelson describes the structure and results of all four of these demonstrations in some detail.

We’ve written about the Physician Group Practice (PGP) demonstration in a previous post.  This study is of special significance because it serves as a dress rehearsal for the highly acclaimed Accountable Care Organization (ACO) approach to saving healthcare dollars.  ACOs are supposed to combine groups of doctors and hospitals together into a single economic unit for the purposes of taking care of large groups of Medicare patients.  The financial incentives and penalties involved are fairly intricate and require that each participating group gathers large amounts of data.  Calculating them involves computing the actual money spent on thousands of Medicare beneficiaries, estimates of how much would have been spent in the absence of the ACO, how many “quality-of-care” measures fell into a specific range, whether the calculated savings exceeded a certain threshold, which year of the demonstration it happens to be, and so on.  When all is said and done, the PGP’s are supposed to get a cut of the “savings” that were generated by their newly efficient quality practices.

According to the CBO, here’s what happened:

“After accounting for the bonuses paid to the PGPs, the estimated net savings for the Medicare program for the second year of the demonstration was about $1.6 million.  That represents a net savings of about $7 per beneficiary across the 10 participating PGPs—a net reduction in Medicare spending of about 0.1 percent. Estimated savings were lower for the first year…”

But the real savings probably weren’t even that large.

“There are two reasons why that estimate might overstate the amount of any net Medicare program savings. First, it appears that some PGPs changed their diagnostic coding practices under the demonstration in a way that increased the risk scores of their patients relative to those of the comparison group…  A second factor that might have led to an overestimate of Medicare program savings is that, for the four PGPs that received bonuses in the second year of the demonstration, Medicare expenditures per patient were growing more slowly than were those of the local comparison groups before the demonstration.”

As for quality, the CBO estimates that the proportion of patients receiving recommended tests increased by 1-5% in the demonstration group.

Although the net financial cost to Medicare was negligible, this was hardly the case for the participating PGPs.  On average, each of these groups had to install an additional $1.7 million in infrastructure in the first year alone.  That was an average of about extra $737 per provider.  You can do the math.  Just to participate in the project, the PGPs themselves spent ten times the amount of money that Medicare “saved” over the entire five years.  But this is almost certainly just the tip of the true-cost iceberg.  Any program of this type requires substantial numbers of additional administrators – personnel whose only jobs are to track what’s going on, hold meetings, remind participants of what’s going on and generate reports.  Although neither Medicare nor the CBO provide any numbers on what these administrative costs entail, it’s easy to come up with a rough approximation.  Let’s assume that each PGP hired just one single additional administrator for this initiative at a fully burdened cost of $50,000 per year.  This is almost certainly an underestimate.  $50,000 per year x 5 years x 10 PGPs = $2.5 million.  This sum alone is over 50% more than the reported savings to Medicare.  Adding in the original $17 million investment for all ten PGPs, and we have a total net cost of this demonstration to the U.S. healthcare system of about $18 million over 5 years, in exchange for a minimal increase in the number of “recommended tests” performed.  In essence, the physician groups involved and privately insured patients are directly subsidizing Medicare and impoverishing rate-payers in order to achieve a savings of about one-tenth the amount that’s being spent.

It seems pretty clear that while repeating the PGP demonstration a bad business idea, expanding it in the form of ACOs is nothing short of financially and medically irresponsible.  Every dollar “saved” by Medicare translates into more than $10 in higher premiums and costs borne by private business and individuals; hardly an approach to “affordable” healthcare.  That’s the kind of program you’d pay to do without.  Let’s see how the other value-based projects fared.

The Premier Hospital Quality Demonstration was actually an unsolicited proposal to Medicare from Premier, a company owned by 200 non-profit hospitals and health systems.  Under this program hospitals reported data on the quality of care provided to patients in five clinical areas: acute myocardial infarction (AMI, or in lay terms, a “heart attack”), congestive heart failure, pneumonia, coronary artery bypass surgery and hip or knee replacements.  The information was used to compute a “composite quality score” for each hospital in each clinical area.  The top 10% of hospitals in each clinical area were initially paid a bonus by Medicare of 2% of their Medicare payments for those areas.  Later in the program, this formula was changed to reward those hospitals whose scores were above the median of the prior two years, or if they had improved markedly compared to their own previous scores.

Why should any of this save money?  The inherent assumption is that it’s “quality” as measured by: (1) adherence to various process measures, (such as whether and when to anti-coagulate people); and, to a lesser extent (2) real indicators of morbidity and mortality, that determines how costly it is to hospitalize people.  Presumably good quality leads to low costs, while bad quality leads to high costs, longer hospitals stays and frequent re-admissions.  Pay people based upon quality and you’ll optimize care and reduce costs at the same time.  There is, however, also a second assumption embedded in this study.  That is that the quality that people put forth is directly related to how you pay them.  Pay them based upon quality measures, and those quality measures will improve.

Simple, eh?  How did that turn out?  We’ll let the CBO tell the story:

“The composite quality scores for the demonstration hospitals rose from the first quarter of the demonstration to the twelfth quarter for each of the five clinical areas.  On average, the increase ranged from 7 percentage points for AMI (the average score rose from 90 percent to 97 percent) to 22 percentage points for CHF (the average score rose from 67 percent to 89 percent). Those increases in composite quality scores were primarily the result of improvements in the process scores. There was little room for improvement in the outcome scores, which were very high at the start of the demonstration.”

So part of our second assumption was right.  Pay people to increase their process scores, and they’ll do just that.  Or is it?

“Not all of the increases in quality scores for the demonstration hospitals can be attributed to the effects of the demonstration, because quality scores were improving nationwide. The best available evidence indicates that the demonstration was responsible for small increases in quality of care and that most of the increases in quality that occurred at the participating hospitals would have occurred in the absence of the demonstration. The CMS-funded evaluation reported that the effect of the demonstration was to raise quality scores during the three-year period by an average of about 1 to 4 percentage points.”

This suggests that hospitals that are sufficiently motivated to participate in a study like this are already doing just about everything possible to maximize the quality of care as measured by the composite score yardstick.  So in a sense, Medicare learned nothing from this wasted six-year effort except perhaps that preaching to the converted is a waste of time.  Since the “value-based” payment mechanism didn’t really change anything in terms of the way these hospitals behaved, one would predict that no money was saved as a result.  And this is exactly what happened:

“The demonstration did not affect Medicare expenditures for inpatient hospital care…  The analysis also showed that the demonstration did not affect the total number of days that patients were hospitalized during the episodes of care or the number of days they were hospitalized for re-admissions.”

So as was the case for the PGP study, if we add in the cost of any extra administrative overhead required to achieve these results, we end up with a net loss of dollars to the overall healthcare system.

Two down and two to go.  We’ll look at the remaining two “value-based purchasing” demonstrations, and talk about what this means so far in our next post.