Readers know my disdain for P4P.  I often feel like the child in that famous story.

Why not P4P?  Here is my short list

1. Read Daniel Pink's Drive – this form of "motivation" does not work
2. We measure what is measurable, rather than what is important
3. P4P has negative externalities, i.e. when you pay for 1 thing something else goes wrong
4. Quality is much more complex than what we measure

Please suggest more arguments against P4P

I love saying "I told you so!"  P4P does not have a data driven rationale.  Too many embraced it because "it made sense".  Evidence in other fields documents that it works only in very specific situations.  Medical care does not fit those specifics (read Drive by Daniel Pink).

Financial Rewards for a Doctor’s Care

But a new study published in Tuesday’s BMJ, the online version of the old British Medical Journal, casts serious doubt on whether pay-for-performance programs are an effective way of improving the quality of care.

“The assumption that throwing money at doctors is going to improve quality is really questionable at best,” said Brian Serumaga, a policy researcher at the University of Nottingham Medical School and one of the study’s authors.

The researchers looked at a pay-for-performance program, begun in Britain in 2004, that financially rewarded primary care doctors for better hypertension care. The researchers looked at what happened to nearly 500,000 patients from 2000 to 2007 as a way of trying to judge the impact of the initiative.

“Pay for performance had no discernible effects on processes of care or on hypertension-related clinical outcomes,” the authors concluded. “Generous financial incentives, as designed in the U.K. pay-for-performance policy, may not be sufficient to improve quality of care and outcomes for hypertension and other common chronic conditions.”

Stephen B. Soumerai, a health policy researcher at Harvard who is also one of the study’s authors, says policy makers should try to evaluate the evidence before they invest substantial sums of money in quality improvement programs that may not work.

Pay for performance (P4P) – Goodhart's law aside is it unethical?

Dr. Edmund Blum, an internist from Brooklyn makes the argument that pay for performance (P4P) involves a "irresolvable conflict " with the ethical standards of the medical profession.( American Medical News,Nov. 6,2006 issue in their "Professional Issue Section.) My bolding.

He says that P4P rests on 3 flawed premises or fallacies the most important of which is that P4P is consistent with medical ethics. He argues that it is not. (The other 2 fallacies are:P4P rests on a valid statistical foundation and P4P will improve the safety and quality of patient care). To those I would add a 4th namely that Goodhart's law would not be operable in the medical care setting.It has definitely been shown to operate there as well.

Readers know that I have major reservations about P4P – not just in medicine but as a way to improve quality generally.  The ethical argument adds a layer to my concerns.  As Dr. Gaulte makes clear in his excellent rant, the presence of P4P changes how we practice, and not always in positive ways.

The same is true about education and many other fields.

Not being an ethicist, I can only respond with my gut reactions.  P4P detracts from professionalism.  It leads to unintended consequences.  It does not address the breadth of medicine.

Tomorrow I will post a patient story that illustrates my point.

Sometimes I feel that I am the only one out there screaming about P4P, but that is not true.  Actually, many bloggers understand this issue better than the suits do.

I highly recommend these 2 blog posts – Quality programs have social and medical consequences

All of these problems might not matter so much if such “quality” programs were costless or without social and medical consequences, but they are clearly not. Ironically, it is practically impossible to determine what the true cost of these programs really is. No good estimates of the total spending for government and/or private “quality” programs seems to exist. Cost-benefit analyses do not seem to be a common component of these programs. What is certain is that physician-insurer interactions already cost medical practices between $23 billion to $31 billion per year, and consume an average of over one-half work day per provider per week.

Stop the Phony Quality Measures, Ctd

And so if the goal of these measures is to improve the quality of care, they’re missing the mark. Doctors will tell you that the biggest challenge they face in delivering quality care is the fact that they have to see 40 patients a day, spending 15 minutes with each one. Checklists and report cards continue the systematic devaluation of the thinking, reflecting and deciding aspects of medicine. So instead, why not start with what really matters: the time doctors spend with their patients. If you ask Dr. Ofri – what would definitely improve the quality of your patients’ care, she will tell you: “an hour-long visit instead of 15 minutes.”

Direct primary care has started to document this last point.  Direct Primary Care Practice Model Eyed to Trim Health Care Spending

For example, the affiliated medical practice, Qliance Medical Group, analyzed internal data from 2009 and found that its direct primary care model lowered emergency department visits by 62% and hospital days by 26% for patients on its plan, when compared with regional averages for the same year, Dr. Bliss said. The group also cut specialist referrals by 55% and advanced radiology services by 48%, compared with regional averages.

“Insurance really doesn’t work for primary care; it’s meant to protect you from catastrophic events. It just adds a whole lot of administrative cost, not only on the doctors’ office side but also on the insurance side,” Dr. Bliss said in an interview. “Primary care is 90% of what people need, and we can provide that.”

The direct primary care practice model resembles the concierge medical practice model, but – unlike concierge practices – direct primary care practices cater to a lower-income demographic and attempt to provide almost all necessary care for a flat monthly fee. Many concierge practices charge a monthly or annual retainer, but also bill for services provided to the patient.

The direct primary care practice model also follows some of the same principles as the patient-centered medical home model. However, the direct primary care model does not team physicians with other health care professionals, such as social workers and pharmacists.

We must think outside the bureaucratic box.  We can do better.

We must value the intelligence and knowledge that primary care PHYSICIANS bring to the doctor patient interaction.  Until we do, we will have expensive ineffective care.

Uwe Reinhardt, noted Princeton health economist, has written 2 very interesting pieces on P4P in the NY Times.  Last week featured this post – The Uncertainties of Pay-for-Performance  .  This post has some interesting insights.

Why is it so difficult to apply the intuitively appealing idea of P4P in practice?

The answer is that in health care, performance can no more be easily defined and measured than it is in the corporate setting. There, methodological problems and boardroom politics have frequently made a hash of what is regarded as performance-based compensation of executives — especially in the financial sector.

In the health-care setting, P4P is difficult enough at the methodological level. But as the recent debate on health care has shown, it must also navigate the shoals of political demagoguery before even reaching the rocky shores of our tort system. A serious and wide application of P4P in health care is likely to usher in a new legal specialty devoted to attacking specific quality metrics in the courtroom. That always happens when serious money is at stake.

Reinhardt does an excellent job of defining many of the problems of P4P.  Some of the comments provide even more fuel for my fire.  I particularly like these observations by disappointed:

Pay for performance metrics are necessitated by overly large bureaucracies. The further one is from the action, the greater the necessity to utilize standardized metrics to assess performance. Management experts have known for some time that overly centralized management has severe limitations and that many decisions and judgements are better left to those closer to the ground. Standardized metrics have their place, but with pay for performance, the standard is too often how to game the metrics system rather than how to deliver quality services. Finally, these type of metrics too often focus on short-term results.

Reinhardt, and others, often ignore the widespread failure of P4P in other fields.  This 7-year-old HBR article makes important observations – Pay-for-Performance Doesn’t Always Pay Off

In practice, however, the process of connecting pay to performance may be far trickier that it at first appears, according to HBS professor Michael Beer.

As he discovered when he examined programs in pay-for-performance that were discontinued at Hewlett-Packard, these programs may indeed have an upside—but there is a potential downside lurking, too. The HP experience was eye-opening as well as sobering. Thirteen separate units of the company—at different types of sites, in different states—launched pay-for-performance plans in the early 1990s. Within three years, all had dropped them.

Yet this author still believes that P4P can work. 

P4P just sounds so fair and logical that accumulated evidence of its flaws are explained away.  Reinhardt and many other colleagues persist in using flawed heuristics.  We know that those heuristics lead to missed medical diagnoses.  I contend that those who favor P4P are blinded by their religious belief that P4P just must work.

Reinhardt followed his initial piece with a follow-up this week – Basing Pay-for-Performance on Outcomes

He continues to explain why P4P is not yet ready for prime time:

All of these experiments with pay-for-performance are fledgling efforts, because the science of outcome measurement has yet to scale many methodological hurdles. Furthermore, in practice the approach will work only if the providers of health care find them sufficiently accurate and fair to sign on. And economic theory tells us that to make the approach effective, it must be backed with significant financial incentives. So far, in many instances, the sums of money at stake have been rather small.

But one should not underrate the sentinel effect of pay-for-performance. Mere talk of the coming of pay-for-performance is likely to draw the attention of the providers of health care — notably the boards of hospitals — to the importance of patient safety and quality. Beyond talk, some research shows that carefully designed pay-for-performance demonstrably appears to enhance the quality of care.

Reinhardt really wants P4P to work.  However, he fails to grasp the totality of medical interactions.  Adopting P4P might help standardize care for a few common diseases.  He links to an article on diabetes management.

But few physicians care for one disease only, not even sub-sub-specialists.  Very few patients have one disease only.

P4P will have implementation difficulties because we cannot define medical care with a numeric outcome.  The multidimensional factors that might describe physician quality make measurement almost impossible. 

How do we measure diagnostic accuracy?  What standard can we use?  Do we credit physicians for treating the wrong diagnosis with the proper treatment for that missed diagnosis?

How do we measure the physician patient interaction?  This interaction includes history taking, patient education and patient motivation.  Some suggest we use patient satisfaction scores, but those have major flaws and suffer from intense grade inflation.

How do we measure an appropriate balancing for the management of 5 (or more) diseases?  How do we assess the appropriate prioritization of medications?  How do we value decreasing polypharmacy by not treating every performance indicator to its fullest?

How do we value appropriate referrals to palliative care?  How do we put a number on excellent comfort care?

We might be able to use complex models to assess large hospitals, but I doubt that any model will work for individual physicians.

What can we measure?  Instead of calling them never events, we should report these events and set an acceptable standard.  When we show that processes can decrease central line infections (Provonost's checklist) then we can penalize hospitals with unacceptable central line infection rates. 

We must thank Reinhardt for showing that P4P is not yet ready for prime time.  I hope his 2 pieces will further the discussion and discourage premature adoption of a process that likely will have negative externalities.

I know that title shocks all my readers.  Can you imagine that I would attack P4P?  Pauline Chen points out the problems nicely in the NY Times this week – Paying Doctors for Patient Performance

Drawing on the experiences of more than 125,000 patients, the researchers first ranked the doctors based on criteria commonly used in pay-for-performance reimbursement plans. While they found that doctors who took care of older or sicker patients tended to rank higher, presumably because of more frequent patient follow-up, the researchers also discovered that primary care practitioners who cared for underinsured, minority and non-English-speaking patients tended to have lower quality rankings than their counterparts.

Using statistical modeling, the researchers then attempted to rerank all the doctors after adjusting for differences in patient characteristics. When they factored patient race, ethnicity, primary language and insurance status into their physician evaluations, many of the original rankings changed, with doctors who worked in community centers — and therefore with more minority and non-English-speaking patients — being more likely to improve in ranking, often by more than 10 percentile points.

Despite showing the flaws of P4P, the lead author of a recent JAMA paper says this:

“Pay-for-performance can work,” said Dr. Clemens S. Hong, lead author and a general internist at the Massachusetts General Hospital, “but we need more sophisticated measures to make sure we are actually measuring physician quality.”

I do not understand the fascination with P4P.  The accumulating data shows all the flaws of P4P.  We should recall Onora O'Neil's words:

Yet faith in performance indicators is hard to dislodge. Every time one performance indicator is shown to be inaccurate, shown to encourage perverse behaviour, or shown to mislead the public, eager people imagine that they will find other performance indicators free of such adverse effects. Experience suggests that they are as mistaken as those who produced the last lot of indicators.

The author of the JAMA paper has done a great service, but still has a delusion that if we just could find the right measures …  We cannot find the right measures through audit.  Perhaps we could use expert observation, but the bean counters would not like that.  We want to achieve and judge excellence, but the task just is not amenable to audit data.  We should drop that fantasy.  P4P in medicine is, and will be  FLAWED.

More Evidence That Pay For Performance Is Working

DrRich with his expected wisdom and sarcasm tells a wonderful story about the impact of P4P:

This week, at Digestive Disease Week (the year’s major scientific gathering of gastroenterologists), doctors from Johns Hopkins will present a paper demonstrating that pay-for-performance reimbursement schemes create financial incentives for surgeons to shun obese patients.

Under this species of pay-for-performance, surgeons are “rewarded” (i.e., not punished) for meeting specified quality standards which have to do with certain patient outcomes. (For pay-for-performance to occasionally equate quality with outcomes is a particularly useful formulation, since expressing reservations about such pay-for-performance measures immediately brands one as being against good medical outcomes, in the same way that being concerned about illegal immigration brands one as being against immigrants, or having reservations about certain of President Obama’s policies brands one as being a racist.)

The Johns Hopkins researchers have found that performing surgical procedures on obese patients results in substantially more complications than performing the same surgical procedures on non-obese patients. For instance, fat people had 27% more complications after gall bladder surgery, and 11% more complications after appendectomy, than thinner people. They also had substantially longer hospital stays, and generated much larger medical bills. The researchers conclude that surgeons (some of whom are literate and understand rudimentary statistics, and therefore not only have access to this kind of information, but are also capable of processing it to at least some extent) can only conclude that, in order to maintain a viable surgical practice, they will need to avoid operating on obese patients. At the very least, they will need to avoid doing elective surgery on fat people, waiting instead until they are in extremis, and require emergency surgery (since at least some effort is made to “adjust” the expected outcomes in these situations).

This result, of course, is similar to the result DrRich reported regarding the publication of Physician Report Cards. Namely, thanks to publicly-available report cards, cardiologists in the state of New York have been more reluctant than cardiologists in other states to aggressively treat patients with severe heart attacks, and as a result (while the report cards are cleaner) the mortality of these patients is higher in New York.

Thus, P4P makes it more difficult for more complex patients to have surgery.  The paradox tells us that P4P then discriminates against physicians who do not discriminate!  As a physician we should care for the patients who present to us and provide the best care for that patient.  P4P can encourage us to "cherry pick".  Adopting DrRich's logic, should we not declare P4P discriminatory and therefore illegal. 

And oh, by the way, P4P in other fields leads to cheating – Study: Employees More Likely to Cheat with Bonus-Based Incentives Programs

The best way to lower health care costs and improve outcomes is to spend more on excellent primary care.  The data are consistent, and the concept has great face validity.

This article is intriguing – For Diabetes, P4P Improves Patient Care, Outcomes

As one reads the article, skepticism about the title enters.

High-quality care — defined as receiving at least two tests for glycated hemoglobin (HbA1c) and one for LDL cholesterol during a given year — was delivered 16% more often by physicians in the pay-for-performance system (rate ratio 1.16, 95% CI 1.11 to 1.22), the researchers also reported online in the American Journal of Managed Care.

"This study showed a robust, consistent, significant, and positive association between increased receipt of appropriate laboratory monitoring of A1c and LDL cholesterol levels and decreased hospitalization rates," Chen and colleagues declared.

On the other hand, the researchers also found that quality of care diminished when patients saw multiple primary care physicians during a given year.

"This finding supports the hypothesis that patients have better outcomes when they have a medical home," Chen and colleagues indicated.

The problem with these studies is the problem of multiple confounders. At least the performance measures used here were ones that we can endorse as not being overly directive – they did not specific the goals of treatment, only the measurement.

What I find more important in this article is that admissions decrease with primary care consistency.  This is the big point that Congress does not understand.  This is the big point that the CBO will always refuse to use in their calculations.

The best way to lower health care costs and improve outcomes is to spend more on excellent primary care.  The data are consistent, and the concept has great face validity.

Subspecialists argue this point, because they are protecting their turf.  They are no disingenuous, rather they see the world through subspecialty eyes, focusing on their diseases.  Consistent primary care more often focuses on the patient than the diseases.  Excellent primary care physicians treat the diseases in the context of the other diseases and the patient's desires. 

We can only improve care if we invest in better primary care.  That will require making the job a better one – less patients per day. less administrative hassles and (I believe) a salary rather than fee for service billing.

I just finished reading A Whole New Mind by Dan Pink.  The book has its strengths and weaknesses, but it does make some very good points about right brain function.

Medicine is not a left brain function solely.  If we just see trees then we miss the patient in the forest.  Too often I see patients on too many meds, with medication side effects, because a physician (or 2 or 3 or more) was treating a problem rather than the patient.

Too often the patient is given the correct prescription, but the physician does not explain the medication appropriately.  Too often the patient cannot afford the medication, but the physician does not even ask.

Patients need our empathy in addition to our checkbox medicine.  We have bean counters, only examining the left brain function of patient care.  But patients are not widgets.  Every day I interact with patients, and try to understand how best to improve there quality of life.  I try to provide high quality care, but that does not always fit onto a checklist.  Many important things that we do are not included in checklists.

P4P is (as I said earlier) immoral and I believe unethical.  It is not scientifically sound, because it has not passed any scientific testing!

Pay for Performance in Primary Care in England and California: Comparison of Unintended Consequences

Somehow I missed this one.  This article has parallel important analyses.

Unintended consequences reported by physicians varied according to the incentive program. English physicians were much more likely to report that the program changed the nature of the office visit. This change was linked to a larger number of performance measures and heavy reliance on electronic medical records, with computer prompts to facilitate the delivery of performance measures. Californian physicians were more likely to express resentment about pay for performance and appeared less motivated to act on financial incentives, even in the program with the highest rewards. The inability of Californian physicians to exclude individual patients from performance calculations caused frustration, and some physicians reported such undesirable behaviors as forced disenrollment of noncompliant patients. English physicians are assessed using data extracted from their own medical records, whereas in California assessment mostly relies on data collected by multiple third parties that may have different quality targets. Assessing performance based on these data contributes to feelings of resentment, lack of understanding, and lack of ownership reported by Californian physicians.

The use of P4P is, in my opinion, immoral.  P4P changes physician behavior, and it is unlikely that that change is an improvement.

The underpinning idea of P4P is that it will improve patient outcomes.  We have no evidence that this theory will work.  If I introduce a new drug, I must jump through many hoops showing both efficacy and lack of significant harm.  The gurus of EBM insist on evidence for treatments and diagnostic tests – and they are right.

We need an outcry about P4P, which has no evidence behind the concept.  It is immoral because the predictable unintended consequences are not good for patient care.

Our study findings suggest that unintended consequences of incentive programs relate to the way in which these programs are designed and implemented. Although unintended, these consequences are not necessarily unpredictable. When designing incentive schemes, more attention needs to be paid to factors likely to produce unintended consequences.

Nice study but they clearly do not understand motivation.  Incentive schemes are bound to fail. Physicians want to do the right thing for their patients.  We can motivate better with incentive schemes.  We should ban this practice in the US.