Mortality measures: Back from the dead….

May 16, 2018 / By Richard Fuller, MS

Within the CMS proposed rule for FY2019 published at the end of April, there is a proposal to consolidate and improve the value-based purchasing (VBP) payment adjustment model. On the surface, the move should be lauded as one that removes duplicative effort contributing to the estimated $39 billion reporting costs associated with administrative regulatory compliance1. The proposed changes, to take payment effect in FY2021 (October 1, 2020), eliminate the safety domain as duplicative to the Hospital Acquired Condition Reduction Program (HACRP) and double the weight associated with the clinical outcomes in determining which hospital gets what from the $2 billion redistributed across hospitals by VBP.

The VBP clinical outcomes domain is comprised of four performance measures for which points are awarded for attainment (relative performance) and improvement. Three measures are for 30-day patient mortality for medical conditions (Acute Myocardial Infarction, Heart Failure and Pneumonia) and one measure associated with complications of care, Total Hip Arthroplasty (THA) and/or Total Knee Arthroplasty (TKA).

For those following along with CMS rule changes, until January 1, 2018 TKA was on the “inpatient only” list—effectively blocked from being provided on an outpatient basis. This means that when the baseline for the measure was constructed, the mix of patients—and the likelihood of complications—was markedly different from the pool likely to be seen in 2018 onwards. The baseline period from which improvement is measured precedes the starting payment adjustment period by seven years and three months2. Hospitals should not be too optimistic of getting “improvement points” for TKA in the near future, as the shift to outpatient (expected to result from its removal from the inpatient only list) will see less complex cases migrate towards outpatient and thereby outside of evaluation. Hence measurement will be conducted against a significantly less complex baseline. But this is actually less troubling than the heavy weight now placed upon the three medical mortality measures.

There are good reasons for hospitals to track their mortality rates. They should be reviewing the performance of physicians and nursing units, the efficacy of quality interventions and overall performance. This does not mean however that mortality rate comparisons across hospitals are a good idea for payment incentives. In our recent article on this topic, “Are We Confident of Across-Hospital Mortality Comparisons?3”, my colleagues and I highlight many of the issues and limitations that surround mortality measurement as it applies to making across-hospital comparisons. We wrote this in response to what we perceived as a push from some within the industry to place greater emphasis on mortality as an “unambiguous” measure of quality. With the proposed changes to VBP this has certainly come to pass.

So, what is wrong with mortality measurement? It basically comes down to preventability. For medical admissions such as those used in the VBP, the ratio of preventable to non-preventable death is too low to make a considered evaluation of performance variation across hospitals. This feature is compounded by the rise of hospital-based palliative care (and program variation across hospitals) for which no adjustment is made within the VBP measures. While we can make better inference of mortality rate variation for surgical admissions (for which we would hope survival is at least a preferred outcome), there are no surgical mortality rate comparisons in the VBP. As we argue in the article, let’s start by first having a system whereby patient mortality after discharge routinely results in notification to the hospital so that they can track their own performance and build from there.

And also, let’s not forget that we have been here before. Between 1986 and 1992 HCFA, with some fanfare, generated comparative in-hospital mortality reports and identified outlier hospitals with “death-lists” for a generally confused public (thanks to the reams of caveats required to interpret the reports). All this under the scrutiny of a generally skeptical and agitated hospital industry4,5. The limited reintroduction of mortality measures in payment buried within the clinical outcomes domain of VBP has so far drawn limited reaction and undoubtedly the science and available data has improved since the demise of the earlier program. However, with $2 billion of payment on the line and greater visibility of mortality rates afforded to both patients and hospitals the changes may not be the burden reducing industry pleaser they first appear.

Richard Fuller, MS, is an economist with 3M Clinical and Economic Research.


  1. American Hospital Association. Regulatory Overload: Assessing the Regulatory Burden on Health Systems, Hospitals and Post-Acute Care Providers.; 2017.
  2. Centers for Medicare & Medicaid Services. Hospital Value Based Purchasing Fact Sheet.; 2017. Hospital_VBPurchasing_Fact_Sheet_ICN907664.
  3. Fuller RL, Hughes JS, Goldfield NI, Atkinson G. Are We Confident of Across-Hospital Mortality Comparisons? Am J Med Qual. April 2018:1062860618771187. doi:10.1177/1062860618771187.
  4. Berwick DM, Wald DL. Hospital leaders’ opinions of the HCFA mortality data. JAMA. 1990;263(2):247-249. Accessed May 12, 2018.
  5. Cleves MA, Golden WE. Assessment of HCFA’s 1992 Medicare hospital information report of mortality following admission for hip arthroplasty. Health Serv Res. 1996;31(1):39-48. Accessed May 12, 2018.