Abstract
BACKGROUND: Rehospitalisation rates in patients with cardiovascular diseases are high. Routine data - including nursing data - might help identify patients at risk. OBJECTIVE: To evaluate the potential predictive value of routinely collected inpatient data and nursing assessment (ePA-AC) scores to identify cardiovascular inpatients at risk of unplanned 30- and 180-day all-cause rehospitalisation. METHODS: This retrospective cohort study included patients hospitalised ≥48 h in the cardiology department from December 2012 - June 2022. The sample was divided into derivation and validation sets based on time of first hospitalisation. Logistic regression and multiple Cox proportional hazards regression analyses were applied. RESULTS: The derivation dataset included 6049 patients, the validation set 1005. Of these 7054 patients, 505 (7.2 %) experienced unplanned all-cause rehospitalisation within 30 days and 1186 (16.81 %) within 180 days post-discharge. The ROC's area under the curve (AUC) values for the 30-day logistic regression model and 180-day Cox regression model were respectively 0.61 and 0.65. Both models identified two key risk factors: ≥1 emergency department visit in the past year (OR 1.49, 95 % CI 1.18-1.86, HR 1.74, 95 % CI 1.52-2.00); and use of coumarin (OR 1.47, 95 %-CI 1.12-1.90, HR 1.27, 95 % CI 1.08-1.50). From the ePA-AC, chronic pain (HR 1.48, 95 %-CI 1.14-1.91) and acute breathing problems (HR 1.41, 95 %-CI 1.03-1.94) were risk factors for 180-day but not 30-day rehospitalisation. CONCLUSION: Both models demonstrated low to moderate predictive value. From the ePA-AC variables, only pain and breathing problems were predictive for unplanned all-cause 180-day rehospitalisations.