OASIS Alert

Published on Fri Aug 29, 2003 PDF

The Centers for Medicare & Medicaid Services released Aug. 13 a long-awaited report, just in time for your beach reading - that is if you're a statistics enthusiast.

OASIS contractor Center for Health Services Research in Denver has provided 79 pages detailing the prediction model used to risk-adjust the outcomes in the outcome-based quality improvement reports that figure so prominently in HHAs' QI efforts.

It's a good thing to shine sunlight on the details of this model, especially since it's used for public reporting," says Brian Ellsworth with the Connecticut Home Care Association. Benchmarking firms and associations can build this data into benchmarking information agencies can use, he tells Eli.

Home health agencies have been waiting nearly a year for the nitty-gritty details about how HHA outcome reports are risk-adjusted. OBQI statistics are no longer used just to improve internal quality within agencies, but now affect interagency competition, certification and marketing, says former CMS official Bob Wardwell, now with the Visiting Nurse Associations of America. "There is some doubt that the level of risk-adjustment achieved at this point is appropriate for the latter use," he adds.

Agencies are concerned that this widely used outcome data doesn't fully represent the quality of the care they provide, and many believe risk-adjustment is one of the reasons for this, Wardwell explains. Now that the underlying model is available, researchers outside the government can analyze its validity "so it can be accepted, improved upon or, if necessary, debunked," he tells Eli.

But it will probably take a statistician to wade through the data contained in "Documentation of Prediction Models Used for Risk Adjustment of Home Health Agency Outcome Reports." It may be too much for HHAs. "My needs could have been met by CMS producing a document that fell someplace in between 'don't worry about it, just trust us' and 76 pages of detailed regression statistics," Wardwell says.