The Corpus Callosum

Thoughts About P4P

Pay-for-performance is one of the buzzwords in health care financing
these days.  I haven’t been following it much, but Dr. Gault
has, over at retired doc’s thoughts.
 His latest: AMA President nails P4P for what it is at
AMA Interim meeting
.  Needless to say, the President
of the AMA is skeptical, as are many physicians.  Dr. Reider,
at Family
Medicine Notes
, wonders if the President’s
statements reflect the overall position of the AMA.  There’s a
little discussion about it at The
Health Care Blog
, too, which has to do with how
much influence the AMA has or does not have in Congress at this point.
 Even more discussion can be found at DB’s Medical Rants.
 There are a few threads there about P4P, enough so that it is
not practical to link them all.  

The gist of it is this: P4P is a buzzword that refers to the concept of
linking physician pay to their reported conformity to certain practice
guidelines.  For example, if a certain percentage of patients
with diabetes get an HbA1C test every so many months, and the average
value is within certain limits, then there could be a little bonus.

What got me to write about it is the publication of an article on the
subject in the New York Times:

Links Doctors’ Pay to Practices

Published: December 12, 2006

WASHINGTON, Dec. 11 — After years of trying
to rein in the runaway cost of the Medicare program, Congress has
decided to use a carrot instead of a stick to change doctors’

Doctors had been fearing a pay cut under Medicare, the health care
program for 43 million elderly and disabled, but Congress instead has
offered doctors a small bonus with big strings attached. To get the
money, doctors will have to report how often they provide quality care,
as defined by the government.

Lawmakers approved the change as one of their final acts before
adjourning early Saturday morning, and proponents said it would improve
the quality of medical care.

But the plan immediately raised concerns among some doctors and
lawmakers who specialize in health issues. They said they worried that
it could be a step toward cookbook medicine and could erode the
professional autonomy of doctors…

There is a lot to be said about this, so I will try to stay focused on
a key point.  This is an elaboration on a point DB makes in this post.

P4P focuses on one aspect of quality as defined by
performance measures. Performance measures relate to patient
management. As the Donabedian quote that I added last week explains,
physician quality is multidimensional. If you emphasize one dimension,
perhaps other dimensions may suffer.

This gets to an aspect of P4P that is perhaps philosophical, but it is
absolutely critical to understand.  Let’s say we settle on a
very specific intervention that we want to promote.  Pick a
simple one, that is easy to measure.  Studies
that patients who have a heart attack do better if they
get aspirin within 24 hours.  Some providers
are very good at making sure this happens (good example of a bad graph
at that link).  Others are not.  

Now let’s say we decide to make this one of the standards for P4P.
 What have we just done?  We’ve taken a complex
system (the hospital’s MI protocol) and created a link between it and
another complex system (the reimbursement system).  Complex
systems are, well, complex.  When you link them, sometimes
unexpected things happen.  

Remember, it is one thing to show that MI patients who get aspirin
within 24 hours do better.  It is quite another thing to show
that altering the reimbursement process to promote this practice will
lead to better outcomes.  They are similar, yes, and it seems
to make sense that one would follow from the other.  But we
are dealing with complex systems here.  

It is entirely possible that getting health care providers to focus on
a few specific interventions could distract them from the task at hand.
 It could impose more paperwork, create another level of
complexity, and thus drive up costs.  So the outcome may not
be the intended one.  

Here, I am not saying that P4P is a bad idea.  Rather, I am
saying that it is important to not be swayed by ideological, conceptual
arguments.  It is necessary to stick close to the evidence.

Imposing P4P is a health care intervention.  In that way, it
is much like prescribing a drug.  There are potential good
effects, and potential adverse effects.  If you are going to
do it on a broad scale, you need to have a system in place to see it it
really does what you want, AND does it at acceptable cost AND does it
without causing more trouble than it fixes.  


  1. #1 Greg P
    December 13, 2006

    In the end, I think P4P is a sly way to reduce payments across the board, to create some pool of money to then redistribute with the performance concept. Said to be a carrot, but more likely a stick first, then the carrot.

    From what we’ve seen there is no reason to believe that P4P is going amount to creating new money sources to distribute.

  2. #2 Joseph j7uy5
    December 14, 2006

    What will happen, I think, is that if P4P appears to catch on, then a whole new industry of consultants and data processing companies will arise, offering to help hospitals and practices maximize their results. But it will only be worthwhile for the biggest players to do that. Solo folks won’t be able to play that game.

  3. #3 jayh
    December 14, 2006

    This is really problematic, as can be seen from countless other areas of human endeavor: introducing a metric distorts the process around the metric. In my software field, measuring output by code lines results in wordy inefficient routines. If this becomes widespread it can influence the nuances of how diagnoses are categorized, creating a temptation to push patients into a ‘pay track’ diagnosis.

    There is also the problem that only a small part of the general healt treatment spectrum can be categorized into a simple metric with a measurable response. The majority of real world conditions will not fit into these neat categories.

    Then there is the issue of change. Good doctors keep up with current news, and may change their recommendations frequently as better information becomes available, based on his own professional judgement. Regulations are far more rigid. The need to revisit regulations regularly, and adequatly quantify the need to make a change.

  4. #4 Scott Simmons
    December 15, 2006

    “This is really problematic, as can be seen from countless other areas of human endeavor: introducing a metric distorts the process around the metric.”

    I learned something about this this past month. I’m on a project at my company that’s re-designing the automated phone flow to provide better service. The VP in charge is having trouble selling the success metrics she’s using for the re-design, because the higher-ups have gotten comfortable with the traditional metric they use for all of the IVR redesigns–fallout rate. If the customer ends up requesting to talk to a live representative, then the automated system obviously hasn’t met their needs, right? So if more calls complete in the IVR rather than dropping to a CSR, then the IVR flow is more successful.

    Sounds fine, right? Until the VP points out that you can also reduce your fallout by making it hard to get out of the automated flow & reach a person. This ‘improves performance’ under the success metric, while actually degrading service to the customers (and annoying the holy hell out of them while it’s about it).

    So now you know why you can never figure out how to reach an actual person from those automated phone systems. 🙂

New comments have been disabled.