Alignment of Minimum Data Set 3.0 Ratings With Staff Care Partner Reports of Symptoms of Distress for Nursing Home Residents

Objectives The Minimum Data Set (MDS 3.0) for nursing home (NH) residents contains data collected at regular intervals that are vital to pragmatic research studies. However, there is evidence of discrepancies between MDS 3.0 data and staff perceptions of mood and behavioral symptoms. We examined the comparative endorsement of ratings of behavioral and emotional distress at one point in time for NH residents by staff care partners and on residents’ most recent MDS 3.0. Design Residents living with dementia who were enrolled in a psychosocial intervention program due to expressions of behavioral or emotional distress had distress recorded at baseline in 2 ways: (1) by the staff as a targeted symptom, and (2) extracted from the residents’ electronic medical records MDS 3.0 data. Setting and Participants The sample included 82 residents from 7 NHs owned or managed by a national long-term care provider. Methods To compare the endorsement of distress by direct staff care partners with endorsement of distress on the MDS 3.0 at baseline, we tabulated the count of residents who endorsed each symptom on each metric, the percentage of the sample for whom staff care partners endorsed a symptom of distress and this symptom was also endorsed on the MDS 3.0, and Cohen κ rating of agreement. Results The percentage of alignment between direct staff care partner report and MDS 3.0 ranged from 39% to 60%; Cohen κ ranged from 0.07 to 0.37. The lowest alignment was for wandering and the highest was for depressive symptoms. Average level of alignment across symptoms was 50%. Conclusions and Implications The discrepancy between staff perceptions and MDS data being no better than a coin toss creates major barriers to pragmatic research in NHs. Results are interpreted in terms of implications for designing and implementing pragmatic interventions.