Tuesday, December 1, 2009

Penalizing Hospitals for Readmissions: Unintended Consequences?


Medicare recently started publicly reporting hospital readmission rates as a measure of quality, and there are proposals to financially penalize hospitals with high rates of readmission. A recent article in the Annals of Internal Medicine makes a compelling case that an unintended consequence of such a proposal could be to worsen health inequalities by inappropriately penalizing hospitals that care for our most vulnerable underserved patients. I found the article convincing. I don't think Medicare should implement any plans to penalize hospitals based on readmission rates without answering the concerns of this article. Further, the article raises serious questions about viewing readmission rates as a measure of quality. The authors of the article were Drs. Rohit Bhalla and Gary Kalkut of Montefiore Medical Center in the Bronx, NY and the Albert Einstein College of Medicine.

It is certainly true that encouraging hospitals to reduce readmission rates in older patients is a good thing. There is a strong basis for viewing readmission rates in the elderly as a measure of quality. While Medicare may be primarily concerned with the cost of rehospitalization, it is also the case that repeated hospitalizations in older persons can have very deletious complications, including exacerbating disability and accelerating cognitive decline. The need for readmission can be an indicator that coordination of care was poor.

However, there are many things beyond the hospitals control that may impact rates of readmission. Independent of the quality of care, patients in poor communities are probably at higher risk of readmission. Factors that contribute to this include lower levels of social support, less access to outpatient care, and lower education. While Medicare readmission models adjust for disease diagnoses, they don't currently adjust for any of these other factors that may also be important determinants of readmission. Therefore, comparing readmission rates may unjustly make hospitals caring for vulnerable patients look bad. It would certainly be even more unjust if our safety net hospitals suffer inappropriate financial penalties for readmissions related soley to their higher risk populations.

In addition to these concerns that were raised in the article, it should also be noted that readmission models do not adjust for factors like frailty, disabilty, and cogntive impairment. As a result, hospitals that care for large numbers of frail elders are also at risk of being unjustly labelled as having worse quality based on readmission rates.

Hopefully, Medicare will take the issues raised by Bhalla and Kalkut seriously.

4 comments:

Dan Matlock said...

I have mixed feelings on this; we don't want to throw the baby out with the bathwater.

I must admit, because of the risk adjustment, I have been excited about the 30 day rehospitalization measure. I think it is a huge leap forward in quality measurement to have an outcome measure (not a process or surrogate measure) that is risk adjusted.

The authors of the Annals piece make a strong argument that this measure may widen the gap and penalize hospitals that care for the underserved. It is a solid argument and is certainly something that I had never thought about before. That said, I could still see such a measure improving the care of the patients seen in these hospitals. Indeed, the incentive for these hospitals to coordinate care would be even stronger.

Perhaps all that needs to be done is an inclusion of the the SES of the hospital's patient population into the adjustment model? (easier said then done).

Good points also about frailty and CI not being included but such is geriatrics.

ken covinsky said...

Thanks Dan. I think there is an interesting underlying issue here that pertains to these comparative quality measures more generally. It is posed by the following question:

If the net impact of a quality measure on patients is positive, is the measure a good thing even if it is not completely accurate or unfairly labels some providers and facilities as providing bad care?

There are a couple of ways a quality measure can be inaccurate and still cause net good for patients (ie, the total good exceeds the total harm).

a--Reimbursement is lowered for hospitals with high readmission rates. They close down. A few of these hospitals that closed down were good hospitals. The reason for their high readmission rates was not related to bad care. They just happened to have a patient mix at very high risk of readmission. But most of the hospitals that close really were bad. Closing a few good hospitals and a lot of bad hospitals makes for an overall better health system.

b. The published readmission rate measure may not be a very accurate measure of quality. However, in response to the measure hospitals do good things to reduce readmission rates that improve quality of care.

Within some limits, I think it is reasonable to argue that quality measures that improve outcomes for our patients are probably OK even if they are not completely fair to providers and facilities. After all, the interests of patients should trump the interests of providers.

However, it is possible that what is bad for facilities is also bad for patients. In the case of safety net hospitals, many of their patients can not easily go to another facility. Unfairly weakening such a facilty could have negative impact on the most vulnerable patients.

One of the tables in the Bhalla article suggests the mislabeling of safety net hospitals may be a real problem. This table shows that on at least one measure (providing discharge intstructions), hospitals in the Bronx (more disadvantaged patients) do Better than hospitals in Fairfield, CT (very wealthy region). Yet, in terms of readmission rates, hospitals in the Bronx look Worse.

Bhalla and colleagues do suggest some adjustments, such as adjusting for the SES mix of each hospitals patients, that may result in a more accurate comparative measure of readmission.

Dan Matlock said...

That is a good point - we certainly need to protect the hospitals that care for the most vulnerable.

sandy said...

good post i love to see this