Abstract
Background: Effective interventions to reduce unplanned admission are often expensive to implement. As such a score to predict potentially avoidable unplanned admission may help target patients who are at high risk and most likely to benefit. However, this have mostly been used using data from specialist care. The aim of this study was to develop a risk stratification model for unplanned admissions to hospital in Norway using register data from specialist, primary and social care
Methods: Data including demographics, primary and hospital healthcare use and comorbid conditions from 173,050 patients aged 19 and above were extracted from routine health registers. Multivariable logistic regression analysis was used to construct a predictive model. The main outcome measure was unplanned hospital admission.
Results: The results will be presented at the conferences. Preliminary analysis indicates that 11 variables from all sectors gives a C-value ROC of approximately 0.77. The variables in the preliminary analysis included diagnosis from both general practitioners, hospital and private specialists, in addition to demographics and use of specialist and social care.
Conclusion: The preliminary results indicates that routinely registered data from specialist, primary and social care are needed to build the best possible prediction models for unplanned hospital admissions.