Assessment of Hospital Reporting of Present on Admission Indicators on Medicare Claims

WHY WE DID THIS STUDY

Previous OIG work found that an estimated 13.5 percent of Medicare beneficiaries hospitalized in October 2008 experienced adverse events, defined as serious harm from medical care resulting in prolonged hospitalization, permanent disability, life-sustaining intervention, or death. An additional 13.5 percent experienced temporary harm events, defined as requiring intervention but not resulting in lasting harm. To determine these rates of adverse and temporary harm events, we examined medical records for a nationally representative sample of 780 hospitalized Medicare beneficiaries in October 2008. This memorandum supplements our prior work by providing national estimates for the extent to which hospital coding staff misreported indicators that identified conditions as present on admission (POA). While hospitals may have improved POA indicator reporting since the 2008 timeframe, little is known about POA indicator accuracy and the nature of any miscoding. Additionally, CMS officials expressed continued interest in the accuracy of POA indicators.

HOW WE DID THIS STUDY

We contracted with certified coders to review the claims and associated medical records for the same sample of beneficiaries examined in the adverse events study. The coders analyzed POA indicators for all valid diagnosis codes within the sample using a standardized protocol. We based review standards on the guidelines in place in October 2008. The coders documented all misreported POA indicators and described circumstances that may have contributed to the errors. We made appropriate statistical adjustments and projected error rates to the population of claims.

WHAT WE FOUND

Hospital coders incorrectly reported 3 percent of the 5,491 POA indicators reviewed, resulting in the presence of at least one incorrect indicator on 129 claims (18 percent). By dividing the POA indicator errors into three groups based on noted similarities, we determined that 21 percent related to the assessment of developing or chronic conditions, 32 percent involved errors in assigning POA indicators to exempted conditions, and 47 percent involved other reporting errors not associated with developing or chronic conditions or with exemptions.

WHAT WE RECOMMEND

This memorandum report does not contain recommendations. The 3 percent national POA indicator error rate is relatively low, particularly given that our review assessed claims submitted early in implementation of the POA reporting requirement. However, POA indicators provide an opportunity for monitoring hospital quality of care and are critical to CMS's efforts to link payment to quality; they must be accurate to serve these purposes. Encouraging hospitals to assess POA reporting practices related to developing conditions and exemption codes, and to retrain staff as needed, could help to ensure accuracy.