The Center Effect
The End Stage Renal Disease (ESRD) Program has often been a laboratory for testing ideas about how CMS can assure the quality of care for Medicare beneficiaries. As part of the Program each dialysis unit is under the oversight of geographically based ESRD Network. The Network is charged with insuring quality of care and has access to clinical information about every patient (and every unit) in the Network. In 1994 McClellan and Soucie identified consistent trends where, despite adjustment for patient-specific risk factors, some units had consistently higher than expected mortality rates and others lower, whereas one would expect random variation around the median. They termed this the “center effect.”
Critics of this concept pointed out that it is difficult to account for all known confounders. Consider the problem of congestive heart failure, which is both common and important in dialysis patients. However, a moment’s reflection suggests that the clinical significance of the diagnosis varies. Diastolic dysfunction, a term beloved by my cardiology consultants, is often applied to a dialysis patient presenting the Emergency Room with pulmonary edema after binging on fluids over the weekend even though the heart may be normal. On the other hand, a dialysis patient presenting to the ER with pulmonary edema, arterial hypotension, and a known ejection fraction of 20%, has a much worse prognosis. The medical evidence form used to report co-morbidities to the Network does not include this level of specificity. Naturally, units with an adverse center effect argue “my patients are sicker,” or “we are under-coding,” or “the measures are not perfect.”
Despite these arguments, the center effect has proven robust and has been found to exist in many other studies, albeit with different estimates of the magnitude of the effect. If we concede the effect is real, what are possible mediators?
The principles of dialysis therapy are reasonably standardized and can be taught to new staff (and renal fellows.) Application of these principles to individual patients, though, requires empiric knowledge. In other words, providing an effective dialysis treatment requires the application of both explicit and tacit knowledge. Having served as a dialysis unit medical director for more than thirty years, I have used tacit knowledge to opine that it takes the average dialysis nurse about a year to get comfortable with the process and about three years to become truly expert. The learning curve for this process is both steep and long.
Second, dialysis units are expected to deliver quality dialysis care, of course, but are also expected to provide effective medical care in managing the complications of kidney failure and its associated co-morbidities. This represents yet another layer of both explicit and tacit knowledge. The complexities involved are more than any one nurse can master, so a team-based approach is needed to attain adequate results. The explicit knowledge encoded in the unit’s policies and procedures has to be combined with the tacit knowledge of which staff person is best suited to address particular issues.
It follows, then, that outcomes of dialysis care depend upon the formal training processes, the policies and procedures used for normal operations, the experience of the individual staff person, and the function of the dialysis unit staff as a team. Units vary in all of these factors, even in the same organization, and none of them is routinely measured. It should be no surpise there is a center effect.
Dialysis units are a specific clinical microsystem. I have argued that hospitals would be more effective in they were seen as collections of clinical microsystems. So, if the center effect is real for dialysis units, it follows it should be real for hospitals, too. Yet we are already hearing the same arguments used by dialysis units—our patients are sicker, the measures are bad, etc.—as Medicare begins to use these sorts of data on the hospital side. Maybe it would be more productive for patients if we acknowledge the center effect is real, and focus instead on developing highly functioning teams composed of well-trained, experienced individuals who were used to working with each other to apply care plans and protocols in a compassionate, patient-centered way.
If we are to do this, we must recognize the central problem in this formulation—you have to invest time and money in staff development, and they have to stay. High staff turnover, which is the norm in hospitals, has to be seen as the Achilles’ heel for producing high quality, safe medical care. Addressing the issues is complex and expensive. But I am reminded of a slide I saw which read:
CFO: What if we invest all this money in the staff and they leave?
CEO: What if we don’t invest all this money and they stay?
16 October 2015
 McClellan WM, Flanders WD, Gutman RA. Variable Mortality Rates Among Dialysis Treatment Centers. Ann Intern Med 1992;1774(4):332-336.
Confronting The Quality Paradox - Part 1
Confronting The Quality Paradox - Part 2
Accounting is not simply a matter of recording reality objectively, it makes things up and changes the definition of what really matters.
Confronting The Quality Paradox - Part 3
Confronting The Quality Paradox - Part 4
There will never be authentic quality within healthcare unless the word explicitly accommodates the truth that a human being is simultaneously both a subject and an object.
Confronting The Quality Paradox - Part 5