The illusion of cost savings from quality improvements

Short piece in the latest NEJM claiming that clinical quality improvements will typically fail to result in bottom line cost savings due to the cost structure of health care.

The explanation lies in the cost structure of the typical health care setting. Its management and organization create a rigid cost structure that is relatively insensitive to small changes in patient volume, resource use, or the severity of patients’ health conditions.

They lay out 4 cost layers and discuss the result of reduced use of resources in that layer that are due to clinical quality improvements (fewer readmissions, lessened volume to investigate a given condition, small improvements in patient health, etc).

Layer 1 “truly variable costs” (not using something that can be used on another patient) can reduce costs in the short run, and improved efficiency of administrative functions (fixed costs not associated with patient care, layer 4) can do so in the long run. However, there are bounds to how much savings due to quality improvements in these two layers can be realized.

Reductions in use in the other two layers (semivariable costs, like nursing and other labor) and (semifixed, like expensive technology) will not typically result in bottom line cost savings due to quality improvements per the piece, primarily because of the timing and cost structure of health care provision. If you don’t order another scan, the machine still must be paid for and the nurse and the tech to run the machine can’t be shifted into another activity easily, or at all. Layers 2 and 3 seem to be inherent attributes of health care (labor intensive and lots of expensive equipment) and thus hard to change, at least in the short run, which is the essence of their claims that it will be difficult to realize cost savings from clinical quality improvements that reduce certain types of health care use.

I need to think a bit more about this piece, but it is an interesting breakdown of “cost layers.” We know that we have quality as well as cost problems and both need to be addressed. If improving quality could also lower costs that would be great, but this piece claims that outcome will be very hard to realize due to the current cost structure of health care in the U.S.

update: Jeff Levin-Scherz also has a post on this paper from a few days ago.


Stephen S. Rauh, M.B.A., C.F.A., Eric B. Wadsworth, Ph.D., C.P.A., William B. Weeks, M.D., M.B.A., and James N. Weinstein, D.O. The Savings Illusion — Why Clinical Quality Improvement Fails to Deliver Bottom-Line Results. NEJM 2011;111:1662.


End of life savings will likely disappoint, ctd.

Several people have asked for more basic information about health care costs at the end of life which I will provide over several posts. Brenda Spillman and James Lubitz have done some of the classic work (a bit dated, but a model for clarity) on the relationship between age at death and cost. The first figure shows the relationship between age at death and costs in the last 2 years of life (1996 dollars). Basically, Medicare costs get lower the older a person experiences death, while nursing home costs get higher, with total costs rising with older age at death.

Note the time frame is last 2 years of life, so is more telescoped forward from death than many end of life studies, but the retrospective nature of the identification is the same.

The second figure shows cumulative costs from age 65 until death, by age at death (1996 dollars). Of course cumulative costs rise with older age at death since persons were at “risk” of health care expenditures for a longer period of time. Cumulative costs are slightly higher for women than men by age of death (fig. 3 in the paper; not shown here) which is an effect separate from a gender-specific survival difference. The authors speculate this to be because diseases progress to death faster in men than women, and it is likely exacerbated by the increased chance that a wife outlives her husband.

These are high level descriptions of the relationship between cost and the age at death. The increase in nursing home costs in the two years prior to death with older death age is easy to explain; older age leads to increased disability and an increased chance of outliving your spouse which increases the risk of institutionalization.

The decrease in the Medicare cost the last 2 years of life with older age at death is a bit trickier to explain. This phenomenon coexists with the general finding that 1 in 4 Medicare dollars are spent in the last year of someones life; but the declining Medicare cost in the last two years of life with older age at death suggests there is a moderation of expenditure the older a person gets. In other words, if this phenomenon didn’t exist, the accumulation of costs relatively near death would be even greater in the Medicare program.

What cannot be answered by analyses such as this one (or really any claims based approach) are whether such decreases in the propensity to spend with older age at death are driven by explicit wishes of patients, or more implicit decisions/interactions between patients, family members and providers throughout the health care system. These are key questions.


Brenda C. Spillman and James Lubitz. New England Journal of Medicine 2000; 342:1409-1415

Constant Tinkering

Rudolf Klein with an interesting discussion of the most recent round of reforms in the NHS, and how there is nothing new under the sun in the constant tinkering with the health system.

As an aside, reading this post took me back to when I was a post-doc at the University of Manchester (U.K.) in 1995-96, and got a chance to meet Dr. Klein on a trip to the University of Bath. He graciously talked with me about health policy and my research at the time. The discussion went to the role of culture in shaping what is politically feasible in terms of cost control. He made an observation comparing the U.S. and the U.K. along these lines: the U.K. is an original sin nation; they expect things to be bad and to likely get worse, be it the weather, waiting lists for knee replacements, or England’s chance of winning the World Cup. In the U.S., it is the opposite. We believe in the perfectability of humankind, and expect that with enough effort, money and energy, we should be able to fix anything.

Can we control costs without choosing between Ryan v. Obama?

Gene Steuerle argues yes, that budget caps placed on total federal spending (Medicare, Medicaid, tax subsidy of private insurance) is the way to go. For one thing, he says we will never completely decide between the various approaches to cost control. Instead, we will maintain a muddled system with many mid-course corrections, likely driven by election victories of the various parties, fueled at least in part by opposition to efforts at health care cost control. And Steuerle notes that for all the heat between the sides and talk of fundamental differences, three basic questions loom over any approach to cost control.

  1. How should budget constraints be applied?
  2. Should automatic budget growth for health care programs (particularly, Medicare) finally be reined in?
  3. Should government health program budgets be limited even if neither side gets its way on question 1?

He notes that question 1 is the focus of most debate (think IPAB v. voucher), and any answer to this question will be politically hard (again, see comments on IPAB and town hall meetings on Ryan budget). This drives both sides to needing budget caps on federal health spending if we are to control health care costs. He says if we wait for political agreement on how to actually implement cost control (the last step, again IPAB v. vouchers for simplicity), we will never control costs, because we will never decide once and for all. So, the answer to question #3 must be yes, if we are to address health care costs. He believes budget caps that provide flexibility to implement different answers to question 1 is the way to go (and he makes a convincing case).

I especially like the reply that Steuerle makes in the comments section of his piece in a reply to Joe Antos (comments in parentheses are mine):

….I asked, but he (Joe Antos) did not answer, what one does if there is no permanent resolution of the health care debate.  Right now Joe is arguing with his partner (the Dems) over how they should raise their kid, but the kid is playing in the street.  To those who say we don’t know what to do, or we can’t do anything unless done my way—often those on the other side of the issue from Joe—I say we do know what to do: bring the kid in off the street.Common ground doesn’t mean agreement on everything.  Removal of open-ended budgets represents the common ground element of both the Obama and Ryan plans. [emphasis mine]

The bigger difference between the Obama v. Ryan approach is that the ACA aims to address costs while expanding insurance coverage, whereas the Ryan budget does not have a strategy to expand insurance coverage. Perhaps budget caps to control federal costs and fighting out the next election (and the next, etc) over the coverage issue and the mechanism of applying the agreed upon caps is the best we can hope for at this time.

What Does Value for Money Mean?

Daniel Callahan with an interesting essay searching for a practical meaning of the phrase ‘value for money’ in health care, and wondering whether we can use it to control health care costs.

Three different deployments of the concept can be discerned: in contexts where circumscribed comparisons of like with like can be made, where it is a stand-alone principle, and where it is meant to trump short-term cost control in the name of future benefits.

He focuses on the third use of the term, trading off short-term cost control in light of future benefits, which he views as the essence of our current health care situation.
No doubt, there are many other areas of health care spending where the upfront costs are high but the long-term gains are worth it. Unfortunately, we have a cost problem here and now, one that will only worsen if we listen to the siren song of eventual payoffs for more spending now…. 

The sad consequence of our cost escalation is that the upfront costs for later benefits must be cut now, and threats of harm to patients have to be put aside. That is what serious cost control entails if we swear off all evasions and rationalizations. If we fail to do so, our future health care system will have fewer and fewer of any kind of benefits to offer.
There are probably areas in which the costs are high and the benefits are small, non-existent or even negative. I am not sure how much of this type of care there may be, but we need to find out.  To do so we need to systematically ask the questions: does it improve quality of life? and does it extend life?  If no, then we shouldn’t be doing it (there is of course uncertainty, and a distribution of effect that must be taken into account).  These questions need to become cultural ones, that each of us ask when thinking about health care choices, and not just technical health policy questions.  If we get to the point that the answer is yes that all care is beneficial and we still need savings, then we are stuck where Callahan leaves us, trading off the need to slow costs to sacrifice some benefits.  This would be hard.  Of course we could then also decide to increase the Medicare payroll tax….The first step in figuring this out is learning to ask these questions; then we can move on to answering them.