Are Readmissions Really an Indicator of Poor Quality?

A Canadian study reported in the Canadian Medical Association Journal should give significant pause to CMS regarding the design and implementation of its Medicare hospital readmisssions penalty and incentive program.  For some time, CMS has required hospitals to report readmission rates in regard to several high cost conditions which see a fair number of readmissions.  Beginning next year, CMS will begin to collect data to be used to in essence rank hospitals as either above or below average in regard to readmissions for index admissions of heart failure, pneumonia and heart attack, and to penalize/reward the hospitals by as much as 3% of a DRG payment on all Medicare payments.  The CMS program is being picked up by Medicaid and private payers as well.  A very significant amount of money is at stake, but the program may be flawed.

The Canadian study prospectively looked at all unplanned, urgent readmissions in 11 hospitals for all adults for the time period October 2002 to July 2006.   (CMJ Article) The readmissions were reviewed by at least four physicians in a standardized process to identify which of the readmissions might have been avoidable.  There were 4812 index admissions and 649 urgent readmissions related to those index admissions.  Of the readmissions, 104 or 16% were deemed avoidable.  Factors which made the readmission avoidable included management error in 48% of the admissions, surgical complications in 38%, medication errors in 33%, nosocomial infection in 18%, system error in 15% and diagnostic error in 11%.  About half of all the readmissions occurred within a month after discharge, but almost 70% of the avoidable ones were in that time period.

There was significant variation across the eleven hospitals in overall readmission rates, but there was very limited variation in the proportion of readmissions considered avoidable.  In other words, there was almost no correlation between a hospital’s overall readmission rate and its rate of avoidable readmissions.  Therefore, using an overall readmission rate as a likely indicator of a corresponding rate of avoidable readmissions for quality performance programs is an erroneous approach.   It should also be noted that less than 20% of the readmissions were deemed avoidable, indicating that the potential savings in eliminating those may not be as great as anticipated.   The study also clearly indicates that clinical review is far and away the only accurate method for determining inappropriateness of a readmission.

The theory of the CMS readmissions program appears to be that excessive readmissions are due to poor quality.  But CMS is counting almost all readmissions, whether related to the initial diagnosis and whether appropriate or not, in its formula, which it must do in part because it is relying on claims data, not an actual clinical review of the case.  In addition, CMS is not currently proposing to reward improvement, only absolute performance.  If the Canadian study’s results are accurate, and it is a well-designed study, CMS’ formula is not going to distinguish good from poor hospital performance on readmissions, and therefore hospitals will be punished or rewarded almost randomly.  Hospitals must spend a great deal of money to comply with the program and to try to improve readmission rates.  But because of the CMS formula, those efforts may be in vain, since quality improvement would presumably only reduce inappropriate readmissions, which the study found were uncorrelated with the rate of overall readmissions.  Given the cost of compliance and amount of money involved, CMS should change the program’s design to require clinical review of readmissions and should simply not pay for inappropriate ones.  That would likely not cost any more and would be fairer.