27 PHYSICAL FUNCTIONING AFTER ICU-ADMISSION 2 adjusted reference value. Spearman’s rank correlation coefficient (rho) was computed and tested for significance. Finally, a binomial regression analysis was used to identify independent risk factors at ICU-discharge for non-recovery. Throughout the analysis, a 2-sided p-value of 0.05 was considered statistically significant. All statistical analyses were conducted using SPSS Statistics for Windows, Version 24.0 (IBM, Chicago, IL, USA). Before conducting a multivariate analysis, data were inspected for missing values. The main reason for missing data was related to the delayed implementation of physical functioning measurement in the ICU and in the specialised post-ICU clinic. Variables with ten percent or more missing values were supplemented using multiple imputations (S1 Table). The assumption was made that the majority of the missing data was missing completely at random, as this was mainly caused by a delayed implementation of the physical measurements. Five imputed datasets were created using predictive mean matching in order to correct for non-normality. Binomial regression analysis was applied using the backward elimination enter method. 2.3 RESULTS 2.3.1 Identification of R-group and NR-group A total of 250 patients visited the standard care post-ICU clinic at 3-months after ICU-discharge and completed the PF-domain of the RAND-36 questionnaire at 3 and 12-months. PF-domain scores at 12-months were compared to age-adjusted reference values;13 median age was 67 [58–74]. Concurrently, 110 patients (44%) were assigned to the NR-group (PF-subscale score at 12-months: 35 [20–50]) and 140 (56%) to the R-group (PF-subscale score at 12-months: 85 [75–95]). In an univariate analysis, NR-patients had a significantly lower PF-domain score at 3-months compared to the R-patients (PF-subscale score at 3-months: 38 [25–55], 75 [60–85], respectively. p < 0.001). Additionally, PF did not change significantly over time in the NR-group. Contrarily, PF-domain scores of R-patients were higher at 3-months after discharge and increased significantly over time (p < 0.001) (Figure 2). 2.3.2 Baseline & ICU-characteristics Further univariate comparison of baseline characteristics comparing NR to R-patients revealed significant differences in demographic factors, comorbidities, disease aetiology, and ICU-morbidity. NR-patients had a higher age (p < 0.05), were more often of the female sex (p < 0.05), and had a higher incidence of
RkJQdWJsaXNoZXIy MjY0ODMw