Understanding Variation in Practice

I read with great interest the article by Miller and colleagues¹ on understanding variation in practice. I agree completely with their thesis that improving the quality of care for our patients will not, and cannot, come simply by standardizing how that care is provided. One key point to address is the difference between variation in the structure and process of patient care as opposed to variation in the content of care.

It is certainly clear that practices vary in how they are structured and how they take care of people. The differences between small 1 or 2 physician practices and large groups (to simplify the issue) offer patients options with regard to the type of environment in which they receive health care. Some physicians are very efficient and are always on time, while others may run late but always give patients the time they need. Eliminating this type of variation would be difficult if not impossible to accomplish, and would likely add little to the quality of care.

Variation in content of care is probably more important in addressing quality concerns. One model divides what we do into 3 categories.² One category includes those tests and treatments for which scientific evidence suggests a clear benefit (eg, beta blockers after myocardial infarction, some immunizations, simple counseling regarding tobacco cessation). In these cases, the intervention should be offered to all appropriate patients, and any variation from this approach would likely decrease quality. The second group is made up of those medical interventions for which evidence is not clear (eg, treatment of benign prostatic hypertrophy or the best approach to colon cancer screening). In these cases, a shared decision–making approach, presenting patients with information about the alternatives, and having patients actively involved in the treatment, is probably the way to improve quality. This approach may lead to variation if there are regional differences in patient values and preferences. The final group includes medical interventions for which there is good evidence that more is not better (intensive end-of-life care, antibiotics for most respiratory infections) In these cases, variation in the direction of doing more should be minimized to improve overall quality.

To the extent that we see our role as providing the best information we have to our patients and then working with them to make decisions about care that are consistent both with our scientific knowledge and with their values and beliefs, the rich variations that Miller and colleagues point to will not be lost in a standardized medical world. Our patients, our staff, and our own individuality will continue to contribute to creative variations in how we do what we do.

Neil Korsen, MD
Maine Medical Center
Family Practice Residency
Portland

REFERENCE

1. Miller WL, McDaniel RR Jr, Crabtree BF, Stange KC. Practice jazz: understanding variation in family practices using complexity science. J Fam Pract 2001; 50:872-8.
2. Wennberg JE, Fisher ES, and Skinner JS. Geography and the debate over medicare reform. Health Affairs 2002; February 13. (https://www.healthaffairs.org/WebExclusives/Wennberg_Web_Excl_021302.htm)

Drs Miller, Crabtree, and Stange respond:

Thank you, Dr. Korsen; we agree! The distinction between the structure and process of care and the content of care may have different implications for understanding variations in care. If only it were so simple. Variation is good, but….

Variation in structure and process is essential if family practices are to remain responsive and adaptive to local assets, needs, and values. We suggest that trying to eliminate this type of variation would not only add little to improving quality, but could worsen it. On the other hand, unfortunately, there are too many family practices with structures and processes that are not responsive to their communities. Process errors, inadequate accessibility, and poor documentation are just some examples of this troublesome variation.

The 3-category classification of content-of-care quality concerns is a helpful one. We agree that greater variation in those areas where there is no scientific evidence of clear benefit is useful guidance and an excellent starting point. The suggestion that variation is more problematic when there is scientific evidence of clear benefit or good evidence that “more is not better” is less certain. Some variation in these areas may also be good. The commonly accepted standard for “scientific evidence” is the randomized controlled trial (RCT), which is a practical but flawed standard. Most RCTs use disease-based outcomes as their criteria of benefit. Application of RCT results to populations in their ecological context assumes that maximizing control of each individual disease will result in better overall health. There is no evidence to support this assumption, and some evidence to challenge it.