32 CHAPTER 2 the majority of patients did not return to the age-adjusted physical functioning SF-36-score at one-year post-ICU.21 In a large cohort of long-stay ICU-patients with sepsis as the primary reason for ICU-admission, the mean SF-36 physical functioning at 12-months was 41 ± 35 percent.22 Furthermore, in a small subset of acute respiratory distress syndrome (ARDS) survivors, 43 percent had self-reported functional limitations 12-months after ICU-admission.23 The overall picture that emerges from these data is that a substantial amount of patients do not recover to the fullest extent. Our data confirmed the presence of comorbidities as an independent risk-factor for the absence of full physical recovery, as reported by others.24 However, in contrast to the general perception that severity of illness score and age are risk-factors for reduced HRQoL as well, the APACHE III score and age were not independently associated in our multivariate model.25 In a general ICU-population with a median APACHE II score of 11, HRQoL after one-year was not different either between patients with high and low Sequential Organ Failure Assessment (SOFA) scores.26 In contrast, in a large cohort of patients with pneumonia and/or sepsis, the Simplified Acute Physiology Score (SAPS) II was an independent predictor for reduced HRQoL. However, this effect was completely attributable to the difference between Q1 (lowest SAPS-score) and the rest of the group, whereas there was no significant difference between Q2 to Q4. Furthermore, the presence of sepsis and age were not independently associated with HRQoL in this study, which is similar to the results in the present study.27 In a recent retrospective study, several multivariate models were constructed to assess influential factors on one-year HRQoL, expressed as the EQ-5D-derived utility index. Although the APACHE II and SOFA score, as well as age, were independently associated, the authors underlined that the models at best explained 20–40 percent of the variability in utility index score.6 The limitations of the available predictive models for outcome are generally well-recognised, since they are designed to predict hospital mortality instead of long-term outcome and HRQoL.28 The most important distinction between previous models and our approach was related to the inclusion of HRQoL and physical status data at 3-months, which was done in order to predict HRQoL one-year post-ICU. At first glance, it might be confusing to incorporate baseline characteristics from the moment of ICU-admission as well as post-ICU parameters. However, the results of the present study justify the notion that the net result of long-term outcome is not only determined by the severity of illness but also by the individual response of the patient to the insult. According to the Nagelkerke R2, not only was our model
RkJQdWJsaXNoZXIy MjY0ODMw