Torhild Heggestad1, Solfrid E Lilleeng, Torleif Ruud. 1. Performance Data Unit, Department of Research and Development, Haukeland University Hospital, 5021 Bergen, Norway. torhild.heggestad@helse-bergen.no
Abstract
OBJECTIVES: Explore if a multi-dimensional analytic approach to routinely registered data provides a comprehensive way to characterise utilisation patterns, and to test if the patients' functional status is a predictor for the use of services. METHOD: We linked register contact data during a two-year period, including all types of specialised mental health services, in the population of a Norwegian county. Cox regression was applied in the models for prediction of admission and readmission. RESULTS: Great variability and complexity in patterns of utilisation were found, including multiple transitions between in-patient and out-patient statuses. The distribution of services was characterised by a small group of patients receiving a disproportionally large amount of resources. A majority of 77% appeared as out-patients only. Severity of symptoms as well as of dysfunction, as assessed by the split GAF-score, differentiated amongst utilisation groups. Both dimensions were significant predictors for admission. In contrast, only the severity of dysfunction predicted readmission. CONCLUSION: Multi-dimensional data architecture and analytical perspectives can be applied to routine data, and should be used to analyse the diverse patterns of utilisation. Risk populations could be predicted by routinely registered information on functional status.
OBJECTIVES: Explore if a multi-dimensional analytic approach to routinely registered data provides a comprehensive way to characterise utilisation patterns, and to test if the patients' functional status is a predictor for the use of services. METHOD: We linked register contact data during a two-year period, including all types of specialised mental health services, in the population of a Norwegian county. Cox regression was applied in the models for prediction of admission and readmission. RESULTS: Great variability and complexity in patterns of utilisation were found, including multiple transitions between in-patient and out-patient statuses. The distribution of services was characterised by a small group of patients receiving a disproportionally large amount of resources. A majority of 77% appeared as out-patients only. Severity of symptoms as well as of dysfunction, as assessed by the split GAF-score, differentiated amongst utilisation groups. Both dimensions were significant predictors for admission. In contrast, only the severity of dysfunction predicted readmission. CONCLUSION: Multi-dimensional data architecture and analytical perspectives can be applied to routine data, and should be used to analyse the diverse patterns of utilisation. Risk populations could be predicted by routinely registered information on functional status.
Authors: Julie Hallet; Elie Isenberg-Grzeda; Calvin H L Law; Victoria Barabash; Jesse Zuckerman; Simron Singh; Sten D Myrehaug; Angela Assal; Wing C Chan; Natalie G Coburn; Alyson L Mahar Journal: Support Care Cancer Date: 2022-10-05 Impact factor: 3.359