reports

Development and evaluation of the Johns Hopkins University risk adjustment models for Medicare+Choice plan payment

Final report to Centers for Medicare and Medicaid Services, US Department of Health and Human Services. Baltimore, MD, USA: Johns Hopkins University.

Johns Hopkins University, Baltimore, MD, USA

In October 2000, the US Health Care Financing Administration (HCFA) (now the Centers for Medicare and Medicaid Services – CMS) awarded a contract to the Health Services Research and Development Center at the Johns Hopkins University (JHU). The goal of the project was to provide a potential risk adjustment method for payment to Medicare M+C plans. At the time the contract was awarded, the primary tasks were to: revise the hospital dominant (HOSDOM) variable; update and test the “ADG/HOSDOM” model using more recent data; refine the model for the disabled and the frail; develop a categorical “ACG” model; and produce software that CMS and its contractors could use to test the validity and applicability of the models.
JHU’s development agenda has taken place against the backdrop of CMS’s implementation of diagnosis-based risk adjustment for the Medicare+Choice program. Starting January 1, 2000 all Medicare+Choice HMOs and CMPs have had 10 percent of their monthly capitation revenue adjusted for the health risk of the members enrolled in their plan based on diagnoses found on hospital inpatient claims. The method used for this process is the Principal Inpatient Diagnosis Cost Group (PIP-DCG) approach.
On October 1, 2000, CMS began collecting physician encounter data for date of service on or after July 1, 2000. Outpatient data started April 1, 2001. These data were intended to support a prospective risk adjustment model based on a comprehensive clinical profile of each Medicare+Choice enrollee. However, in May 2001, Secretary of Health and Human Services Thompson suspended collection of data from ambulatory settings, citing the reported burden on M+C plans trying to comply with the data submission regulations.
New data collection requirements, stating that data collection would resume October 1, 2002 for date of service on or after July 1, 2002, were issued in May 2002. In addition to a dramatic reduction in the number of data elements to be collected, CMS also limited the diseases/conditions that plans had to report to those that CMS believed were relevant for adjusted payments under a risk adjustment formula. Plans will only have to report a maximum of 3,136 diagnosis codes (out of more than 15,000), organized into 61 disease groups. These 61 disease groups have been labeled the selected significant diseases model.
Although CMS decided to adopt a limited-diagnosis 61-condition model, JHU’s progress on this CMS sponsored risk adjuster development remains potentially useful to CMS or the health plans for many other purposes. JHU risk adjusters, based on comprehensive diagnostic information (not just a more limited subset of diagnoses used for payment purposes) are desirable for concurrent applications such as provider profiling and detection of fraud, waste and abuse within the Medicare program, quality of care evaluation, or care management.