I think there are really two issues that need to be separated out. There is the area of precision medicine where the scientific evidence is pretty compelling and what to do is quite clear (aspirin for patients with heart attacks). The other area is cognitive medicine, where there really aren't quality measures, because this area isn't quite as clear or precise. The books the Innovator's Prescription as well as How Doctors Think are excellent in discussing cognitive medicine.

Dr. Ofri in her August 12, 2010, New England Journal of Medicine piece mistakenly lumps both types of medicine into one category. The metrics do matter for areas we can measure, do make a difference, and where the science is definitive. To imply that doctors are good because of bedside manner and nothing else matters is wrong. Increasing the time spent with patients, particularly if doctors don't believe the validity of these metrics, won't improve patient outcomes. There is plenty of research that has already demonstrated this – Shannon Brownlee's book – Overtreated – is an excellent introduction to this.

A study published in 2010 found that doctors are not correctly screening for colon cancer even though the guidelines are quite clear. http://davisliumd.blogspot.com/2010/04/seventy-five-percent-of-primary-care.html  Disappointing to say the least as it is the second leading cause of cancer death. The American Cancer Society estimates that 95% of colon cancers could be avoided if screening was done.

I believe all Americans deserve doctors who are consistently and systematically doing the right thing every time when the science is quite clear (precision medicine) as well as doctors skilled in areas not so clear (cognitive medicine) and who, incidentally, have great bedside manner. These skills are not mutual exclusive.

Now whether P4P actually helps get us there.  I have no idea.

I did a presentation on basically this same topic a while ago and can summarize a little my concerns:
1)P4P essentially uses a dichotomous value for a continuous variable (ie a HGA1c of 7.1 is bad but 6.9 is good) which is not rational
2)The performance measures are not based on good data – much evidence such as ACCORD, VADT show that tight blood sugar control or tight BP control in diabetics is not only not beneficial but may be harmful. Many of the guidelines are written by experts with potential conflicts of interest.
3)There are many patient factors that influence how a physician manages DM, or HTN that are beyond the physicians control such as adherence, patient preferences and psychosocial factors.
4)Most measures do not take into account the patients age, life expectancy and comrobid conditions. For example, treating a 95 year old bed bound demented patient to a Hga1c of 7 is ludicrous.
5)Physicians should have the right to ignore guidelines that they feel are based on poor data. when that ability is lost we lose the critical thinking that makes one a good physician. the example I give is that years ago a P4P could have included HRT for all postmenopausal women, and Vitamin E for all post MI patients. The guideline I personally ahve the most problem with is pneumovax. if you really look at the data behind it is amazing this was ever even approved let alone advocated for widespread use. But by including it in performance measures young physicians will automatically assume it is a given that this is well studied and beyond debate. yet it is essnetially being used because of one study done in the 70's in South african Gold miners that would not get published in a throw away journal if it were submitted today.
I could go on but will stop here 

I have nothing to add that hasn't been said earlier, but P4P has the feel of treating the labs, and not the patient.  As said earlier, an HgbA1C of 7.1 is bad but 6.9 is good.  There is no discussion of trending, either.  A patient with hyperlipidemia, no matter if your treatment resulted in a 25% reduction in total lipids, would still be bad.

The P4P movement is legislating morality in feel-good, engineered terminology.  The movement defines a good physician as one who hits guideline targets, and disregards other measures, such as patient perceived quality of life improvement. 
But, the mark of real harm that it causes is that P4P, while stating it is deeply couched in empiric, objective reality, is nothing more than faith-based proselytizing of a particular specialty's bailiwick being important uber-alles, trumping all other specialties and diseases.  When we consider that the human cells in the human body number in the 1X10^14 range, plus all of the other chaotic neutral inhabitants we have, we must embrace complexity, and we must embrace uncertainty.  P4P treats the human body and medicine as a solved problem, which is a dangerous simplification. 
At the best, P4P and guideline reproduction have caused no harm, only more work for physicians.  But, the mere fact of causing us to practice less efficiently, even without endangering safety, endangers safety through the spawning of fatigue and indifference.  It harms by distancing us from our patients and causing us to spend less time with them.

Continuing from PCB:
9.) Current established goals are hard targets and will not change
10.) There is proof quality can be measured meaningfully and objectively
11.) Rewards and punishments [financial or otherwise] motivate analytic and creative thinkers to "do the right thing".
12) Chaos theory can be ignored.

Dichotomous goals are often practical. For example, how much error is acceptable in a diagnostic assay before the error will cause clinicians to make incorrect decisions. The relationship between error and incorrect decisions is continuous, not dichotomous. The larger the error the more likely the incorrect decisions. But life is a series of dichotomous decisions (as in to treat no not to treat). Although there is a move afoot to at least chop this continuous line into 3 or 4 goals such as a Clarke error grid for glucose meters.