The Accuracy of Minimum Data Set Diagnoses in Describing Recent Hospitalization at Acute Care Facilities

Objectives

The Minimum Data Set (MDS) is the resident assessment instrument used to guide clinical care, reimbursement, and assess quality in long-term care facilities. This database has been used in many studies, although the accuracy of many data elements remains unknown. This study evaluated the accuracy of the MDS diagnosis variables with respect to the diagnoses for recent hospitalization from Medicare claims data.

Design

Retrospective cohort study.

Setting

945 skilled nursing facilities in Ohio.

Participants

17,294 residents admitted from an acute care facility during 2000.

Measurements

Eleven diagnoses listed in the MDS were compared with Medicare hospital discharge claims. Specifically, each MDS diagnosis was compared to the primary diagnosis, the list of secondary diagnoses, and the Diagnosis Related Group (DRG).

Results

Claims diagnoses were listed in the MDS with an average frequency of 79% (range: 31%–94%) for the primary diagnosis, 66% (range: 33%–90%) for any diagnosis, and 71% (range: 31%–94%) for the DRG. MDS diagnoses were listed as the primary diagnosis, any diagnosis, and DRG with an average frequency of 20% (range: 6%–81%), 62% (range: 41%–86%), and 19% (range: 7%–84%), respectively, with only hip fracture listed more than 80% of the time.

Conclusion

The sensitivity of the MDS for listing diagnoses from recent hospitalization appears good for most diagnoses. However, except for hip fracture, the MDS has poor predictive value with regard to the primary reason for the preceding hospitalization; this may have implications for resident care planning and the utility of this database in long-term care research.

Keywords:  Minimum Data Set , aging , long-term care , administrative data , epidemiology