Be careful what you read.

Sandy at Junfood Science is one of my blogging heroes. She writes eloquently and loves to bash the whole "obesity epidemic" bandwagon. Her website is a very worthwhile daily read.

However, I find that I disagree with the overall slant of her recent post on Pay for Performance (P4P).

Some Arguments Against P4P

The gist of the post is that P4P is fraught with problems including:

1. Doctors avoiding care of sick patients as they are penalized for noncompliant patients.

Many doctors have said that a system which gives poorer grades to doctors whose patients are sicker is perverse. They’re concerned that the financial incentives could mean the sickest or most difficult to treat patients would be less likely to be accepted into doctors’ practices and might have a harder time getting care. Last month, the American Medical News reported that 82% of doctors said the quality measures would have unintended consequences by their avoidance of high risk patients and that it would hurt minorities and the poor. As one doctor wrote: “The poor, unmotivated, obese and noncompliant would all have to find new physicians.”

2. Care measures may not be appropriate measures of real quality

If the evidence behind the clinical guidelines were strong and clear, there wouldn’t be so much disagreement to them within the medical community, controversies many consumers may not be aware of. Annual physicals for everyone, for instance, are being compelled by many insurers, but an expert committee sponsored by the Agency for Healthcare Research and Quality in 2003 found little benefit in many of the tests and suggested that they only serve to increase the cost of healthcare, while exposing patients to unnecessary risks. The other measures, such as weight management, adherence to lipid-lowering (“cholesterol”) and anti-hypertension (high blood pressure) medications, and diabetes management are equally controversial.

3. Even appropriate care measures may not be appropriate in all patients

A study led by Dr. Leonard M. Pogach, M.D., MBA, at the Health Services Research and Development Center for Healthcare Knowledge Management Research, Department of Veterans Affairs New Jersey Healthcare System, examined one controversial P4P measure. Under the clinical practice guideline of the American Diabetes Association, the target for glycosylated hemoglobin (A1C) level measurements are less than 7%. The doctors found that this arbitrary threshold had been achieved in clinical trials with resources not usually available in real-life clinical practices and that they are not appropriate for certain types of patients, namely older ones or those with comorbidities (other health problems). Examining the data on patients from 144 Veterans medical centers during 1999 and 2000, they found that more than one-third of the diabetic patients would never benefit from meeting the A1C values. Instead, the aggressive control of blood sugars necessary to achieve the low A1C levels brings dangers such as of hypoglycemia, that could increase their health risks and be dangerous or even fatal for certain patients.

4. Overly focusing on performance measures will leave less time to care for the patients' real needs

Because these P4P measures emphasize certain selected measures, doctors could be coerced to focus on them and ignore care issues that might be of greater need or importance to individual patients. Doctors are concerned these P4P may result in patient care focused on meeting the measures, rather than what’s best for each individual patient.

Last month, Dr. Westby G. Fisher, M.D., FACC, a board certified internist, cardiologist, and cardiac electrophysiologist in Evanston, IL, wrote a commentary on his blog in the form of a imaginary letter to his patients:

I regret to inform you that I will be spending less time focusing on your heart problem because I have decided to focus on the heart and medical problems that Medicare deems important to assure I get paid. They call this initiative “Pay for Performance (P4P)." You see they published a list of 74 criteria that will be measured to see if I give good care, so I will get paid appropriately....

So, dear patient, I'm sorry if you have pericarditis or heart block. I'm gonna need to focus on my heart attack and heart failure patients a bit more to make sure my office staff can still return your calls. I hope you understand.

Problems with these arguments

1. The current system discourages already caring for complex patients.

The current system rewards volume over content. We are paid to document, not care for patients.

I can see 40 sinus infections, sore throats, and ear infections in a day and bill a level 3 (99123) for each of them at around $65. The documentation and visit will take from 5-10 minutes maximum. This means that I can bill $2600 for 5 hours' worth of work (7.5 min per visit).

If I work to educate these patients regarding the need to not use antibiotics in every case, I add time for which I am not reimbursed. If I note that a woman needs a mammogram or that a diabetic needs a pneumococcal vaccine, I earn no more than if I don't do these things, I simply spend more time with each visit and my hourly wage goes down. Any attempt at improving the quality of these visits will result in a decrease in income or an increase in my time working for the same income.

A diabetic with hypertension and hyperlipidemia who comes in for a recheck (with other complaints - a typical complex adult patient) will take a minimum of 15 minutes (if straightforward) and often 30 minutes to do the visit and documentation. I generally bill a level 4 for these visits, charging approximately $90 for this type of visit. This means that I see 26 diabetics in the same 5 hours (if the average visit is 22.5 minutes long) and can bill $2340.

Any focus on quality (foot checks, scheduling eye visits, ordering labs, etc.) will take time, as will the follow-up on the results of these tests and consults. None of these measures are reimbursed, so I lose money if I focus any time on them.

The best way for a practice to succeed in the current system is to see simple patients for simple problems and spend as little time, ordering as few tests as possible.

2. There are many well-documented care measures that have clearly positive outcomes.

If point #2 was really true, then I would quit primary care. I am a preventive medicine physician. I believe that it is better to have good diabetic control than poor control. People who smoke, have high blood pressure, have poorly controlled diabetes, and have high cholesterol are worse off than those who don't have these things. I spend my day working on these things and many groups have documented the evidence behind the pursuit of these measures.

From the American Diabetes Association's 2007 Guidelines:

Glycemic control is fundamental to the management of diabetes. The goal of therapy is to achieve an A1C as close to normal as possible (representing normal fasting and postprandial glucose concentrations) in the absence of hypoglycemia. However, this goal is difficult to achieve with present therapies (26). Prospective, randomized, clinical trials in type 1 diabetes such as the DCCT (27,28) have shown that improved glycemic control is associated with sustained decreased rates of microvascular (retinopathy and nephropathy), macrovascular, and neuropathic complications (28–31).

In type 2 diabetes, the U.K. Prospective Diabetes Study (UKPDS) demonstrated significant reductions in microvascular and neuropathic complications with intensive therapy (32–34). The potential of intensive glycemic control to reduce CVD in type 2 diabetes is supported by epidemiological studies (32–34) and a recent meta-analysis (35), but this potential benefit on CVD events has not been demonstrated in a randomized clinical trial.

In each of these large randomized prospective clinical trials, treatment regimens that reduced average A1C to ~7% (~1% above the upper limits of normal) were associated with fewer long-term microvascular complications; however, intensive control was found to increase the risk of severe hypoglycemia and weight gain (31,34).

Recommended glycemic goals for nonpregnant individuals are shown in Table 6. A major limitation to the available data is that they do not identify the optimum level of control for particular patients, as there are individual differences in the risks of hypoglycemia, weight gain, and other adverse effects. Furthermore, with multifactorial interventions, it is unclear how different components (e.g., educational interventions, glycemic targets, lifestyle changes, pharmacological agents) contribute to the reduction of complications. There are no clinical trial data available for the effects of glycemic control in patients with advanced complications, the elderly (≥65 years of age), or young children (<13 years of age). Less stringent treatment goals may be appropriate for patients with limited life expectancies, in the very young or older adults, and in individuals with comorbid conditions. Severe or frequent hypoglycemia is an indication for the modification of treatment regimens, including setting higher glycemic goals.

More stringent goals (i.e., a normal A1C, <6%) should be considered in individual patients based on epidemiological analyses suggesting that there is no lower limit of A1C at which further lowering does not reduce the risk of complications, at the risk of increased hypoglycemia (particularly in those with type 1 diabetes). However, the absolute risks and benefits of lower targets are unknown. The risks and benefits of an A1C goal of <6% are currently being tested in an ongoing study (ACCORD [Action to Control Cardiovascular Risk in Diabetes]) of type 2 diabetes.

From the Annals of Internal Medicine, April 2006:

Currently, 1 in 5 people with diabetes (2.2 million people) has poor glycemic control (hemoglobin A1c > 9%), 2 in 5 people with diabetes (3.6 million people) have poor LDL cholesterol level control (LDL cholesterol level 3.4 mmol/L [130 mg/dL]), 1 in 3 people with diabetes (3.5 million people) has poor blood pressure control (140/90 mm Hg), and 1 in 3 people with diabetes has not received annual eye (3.2 million people) or foot examinations (3.1 million people).

The bottom line is that there are many sources for quality measures that are supported by a preponderance of evidence. The problem is not that these measures aren't out there or aren't clear, it is that they are not being followed closely enough.

3. P4P measures are population-based, not individual patient based.

When our practice went for NCQA certification (which we achieved), we measured the quality of diabetes care in a number of categories: A1c, Eye Exam, Foot Exam, Microalbumin Screening, Smoking Cessation, Lipid Measurement and control, and Blood Pressure Measurement and Control. Very few other physicians have qualified for these guidelines - mainly due to the fact that the collection of the data is extremely labor intensive if you do not have a readily accessible database. Our quality measures greatly exceeded the NCQA threshold in many of the measurements (9% of my diabetics have an A1c over 9, where the threshold was to have less than 20% to get credit).

I have never in my 12 years of practice discharged a patient for noncompliance. Yet my quality numbers were high enough to nearly get a perfect score on the NCQA certification (my partner actually did get a perfect score). The threshold for qualification is 20% over A1c over 9. That means that one-fifth of your patients can have very bad diabetic control and you can still get credit for high quality! The current P4P standards are no more strict than this (that I am aware of). Bridges to Excellence relies on NCQA certification for qualification in their P4P program.

The Medicare 1.5% reimbursement (which is, as I have said in the past, not going to motivate anyone), is not measuring the level of A1c in diabetics. It is simply measuring if you addressed the issue in your encounter with the patient. Of the 74 criteria from medicare, you need to submit 3 to get the reimbursement - we are going to report 6 of them, of which Medicare will take the top 3.

4. Too much focus on procedures and volume has lead to physicians ignoring prevention and disease management.

In the current system I could easily write the following letter:

Dear Mrs. Jones:

I am sorry to hear about your breast cancer. Yes, it would have been good for me to make sure you had your mammograms regularly - especially with your mother having had breast cancer - but I was too focused on sinus infections and making sure my documentation of 99214 visits was in line so I did not get turned in for Medicare fraud. I hope the pain from the metastases in your back are not too bad. You should have called to remind us you were due. We don't have time to keep track of such things, as we are keeping our schedule full to cover our overhead.

I am also sorry to hear about your husband's heart attack. We never did get around to checking his cholesterol when he came in for his diabetes, as he always would talk about farming and his back pain. The discussion about cholesterol management and good diabetic control takes about 20 minutes of non-reimbursed time and I can't afford that. I am sure you understand. I hope his leg is doing well from the amputation in January.

This is simply a ridiculous argument: that there is no time for sick care if you are busy doing prevention and standard-of-care medicine. The bottom line in this argument is that we should not reimburse both prevention and acute care because the acute care will suffer for the attention on prevention. I must be missing something.


It may sound like I have "drunk the Kool-Aid" for P4P. I understand the shortcomings of the current P4P systems, I just do not think that the argument about the changes needed in our system is being carried out with much thought. It is easy to criticize the ideas of others; it is far harder to defend your own ideas. All I hear from physicians on the whole P4P discussion is criticism and very few ideas.

Here are my thoughts on this whole issue:

  • The current system is terrible. It is unfairly paying physicians - rewarding bad care and penalizing those who try to do the right thing. The current system needs changing.
  • Pay for performance is not "the answer" for the problem, but any solution must take quality of care into the reimbursement formula. We need to stop penalizing physicians for doing good work and rewarding those who do bad work. No other industry rewards poor quality so highly.
  • If P4P is going to work, it must not be run by insurance companies. Insurance companies are publicly-held companies whose moral obligation is to make as much profit as possible for their shareholders. To put them in charge of designing a fair reimbursement scheme is like letting the criminals design the security system for a bank. There is very little chance for it to succeed.
  • If P4P is based on clinical data instead of claims data, it will put the integrity of the data in the hands of the physicians, not the payers. The biggest flaw of the new Medicare P4P (aside from the pitiful 1.5% bonus) is that it is based on claims data rather than physician-reported data. Physicians must be willing and able to show the quality of care they offer. I once got a "quality report" from a payer and found it to be only 50% accurate. I should be able to submit my more accurate data from our patient database.
  • If physicians simply sit back and take shots at P4P, then we will get what others design for us. We need to be at the table making sure that our interests are put forth. The ultimate transaction in medicine is between patient and provider. Those are the two with the most to gain and most to lose from the way the system gets changed. Those are the two who must be pushing for the right kind of changes.
  • There is a necessary link between information systems and P4P. The only way for physicians to truly be able to report their own quality is for there to be an information system in place to measure and report that data with.

Physicians will stay on the periphery of the discussion on healthcare reform if we simply take shots at new ideas. We should not accept them without question (especially if the ideas come from the insurance industry), but we should find ways to right the current wrongs and do what is best for our profession and for our patients.