Health Affairs: Public hospital reporting needs to bulk up on data
Publicly reporting key process indicators may be clinically linked to improving hospital performance, lowering mortality and reducing length of stay, but alone they may provide too little information to be used as an indicator for healthcare quality, according to a study published in this month's Health Affairs.

“Low-quality care has been well documented in U.S. hospitals, where too often patients do not receive the recommended care for common conditions,” the authors wrote.

To incite improvements within hospital care, the Centers for Medicare & Medicaid Services (CMS) began making hospital performance public in April 2005 on the Hospital Compare website, part of the Hospital Quality Alliance initiative.

Rachel M. Werner, MD, PhD, and Eric T. Bradlow, PhD, of the University of Pennsylvania School of Medicine in Philadelphia, set out to investigate whether or not hospital performance changed during the first three years after the data were publicly reported and also whether these performance changes were linked to adjustments in mortality, length of stay and readmission rates.

Werner and Bradlow assessed hospital performance data of 3,476 acute-care, nonfederal U.S. hospitals who reported performance on the CMS Hospital Compare website between 2004 and 2006.

The researchers evaluated data every six months. These performance measures dealt with heart failure, MI and pneumonia including aspirin use at admission and discharge, beta blocker use at admission and discharge, left ventricular function assessments and timing of initial antibiotic therapy, among others.

In addition, they examined 30-day mortality, length of stay and 30-day readmission rates for all patients using CMS’ MedPAR file that included Medicare Part A claims data.

The researchers found that the rates of all process measures improved and hospitals that saw the lowest baseline performance in 2004 saw the greatest improvements between 2004 and 2006. “Most high-performing hospitals maintained their levels of performance over time,” they wrote.

Results showed that beta blocker use at hospital discharge increased from 90.2 percent in 2004 to 95 percent in 2006. The rate of hospitals administering ACE inhibitors for left ventricular dysfunction in HF patients also increased from 75.8 percent in 2004 to 85.3 percent in 2006.

The researchers also found that a 10-point increase in performance levels was linked to 0.6 percentage point decline in mortality rates, a 19-day decline in length of stay and a 0.5 percentage point decrease in 30-day readmission rates.

The results showed that increased performance in HF patients reduced readmission rates by 0.2 percentage points, while improved performance for pneumonia patients was associated with a 0.2 percentage point decrease in mortality, but also with a 13-day increased length of stay.

However, the researchers found that hospitals that exhibited the highest baseline performance showed little improvement in outcomes.

“The largest improvements in outcomes associated with a 10-point improvement in performance were seen among hospitals with mid-level baseline process performance, which suggests that outcome improvements may require certain minimal levels of process performance (that is, there may be a threshold effect),” the authors wrote.

The researchers said this variation could be attributed to different documentation strategies among the various hospitals or the fact that hospitals may exclude certain patients from its performance scores, “artificially inflating their measured performance,” even though CMS continuously audits these data.

While Werner and Bradlow said that most improvements were significant between 2004 and 2006, the “results do not prove conclusively whether or not public reporting caused an improvement in processes or outcomes,” and questioned their credibility as a stand-alone predictor of quality.

However, despite the large variations found in the results, Werner and Bradlow called the results “encouraging” for the future of quality improvements and public reporting efforts.

Werner and Bradlow suggested that improved access to care and coordination of care in addition to a broader set of performance measures could help make data more robust and further improve quality.

“Performance measurement, in combination with public reporting and pay-for-performance, is an important tool for quality improvement,” the authors concluded. “However, questions remain about whether public reporting causes these changes, and whether the documented improvements in outcomes are truly a response to process improvement, given the modest changes in outcomes that we documented in many areas.”