Quality Metrics: Outcomes Versus Process

An area that has caught my interest lately is the concept of quality metrics. In the third year of medical school, we get a two-week block on health policy, including one discussion section on quality metrics. In this section, we discussed some of the public reporting data from the New York CABG studies. In short, yes outcome reporting increases quality improvement activities at the hospital-level and in some cases decreases overall mortality rates. However, what was even more interesting were the unintended consequences reported in this data. In many instances, the sicker cardiac patients were selectively referred out of state and were less likely to receive not only CABG surgery, but also percutaneous coronary intervention (a more common and less invasive treatment) than patients in Michigan where there was not public reporting.

Although physician and future physicians including myself may like to think we are above incentives, it is human nature to respond to financial and peer pressure. Does it make sense to give physicians the incentive to risk adjust their patient populations and penalize those providers that treat the highest risk patients? We tried this incentive with health insurers and saw skimping of benefits, limits on preexisting conditions and underwriting. Health insurance is composed of insurance risk based on the patient’s demographics, adherence and genetics and performance risk based on the provider’s care. Outcome measures place the responsibility for both insurance and performance risk squarely on the shoulders of the provider.

There are two solutions I see to this problem. One would be to develop a comprehensive method for risk adjustment, which is not only difficult to develop for a certain population, but relatively impossible to extrapolate to a broad, diverse patient population. Personally, I believe process metrics make more sense—we should compare providers on whether they follow evidence-based best practice standards thus holding them accountable for their performance risk and not for the patient’s inherent risk factors. For instance, instead of measuring diabetic hemoglobin A1C at 6 or 12 months, we would check to see if the provider ordered this lab test to be done at least twice per year, according to the American Diabetes Association guidelines.

Some say that following evidence-based guidelines produces “cookie-cutter” medicine, where all patients receive average, standardized care. However, if we have proven the best algorithm for a specific condition why wouldn’t we provide this medicine where applicable and individualize therapy when necessary? I don’t think we should be threatened by the standardization of guidelines and best practices. There is no algorithm that can replace the doctor patient relationship, the professional instinct to know when the guidelines don’t apply and the educational and advisory role of the physician as the patient’s partner in medical care.

Fung, C., et al (2008). Systematic Review: The Evidence That Publishing Patient Care Performance Data Improves Quality of Care. Annals of Internal Medicine, 148 (2) 111-123.