described in chapter 2. But when including all patients in this surveillance, i.e. those with and without an invasive device, developing nosocomial sepsis or ≥ 2 infections was associated with increased mortality. Adjustment for confounders was limited, as in many studies. Large‐scale randomised clinical trials, such as those on selective decontamination of the digestive tract (SDD) and selective oropharyngeal prophylaxis (SOD), when well balanced, do not have this limitation and have demonstrated a reduced mortality by preventing bacteremia [67]. The results of the HAI‐Net mortality review study demonstrated that treating physicians were of the opinion that in 39% of the assessed cases, HAI definitely contributed to the death of the patient. The WHO‐based score reveals that HAI were most often considered as part of the causal sequence (56% of treating physicians) and rarely viewed as the sole cause (9%). It is imaginable that when mortality review in this form is implemented into daily practice, HAI will be found to contribute less to mortality. For one reason, in cases with more than one HAI present, the most severe HAI was selected for the mortality review. For another, the on‐site investigators that participated in this study probably were convinced of the relevance of HAI. On the other hand, the treating physicians that co‐reviewed the patients were not ‘selected,’ and [their] agreement on pneumonia and BSI was quite reasonable. So, HAI are considered relevant with respect to this ultimate patient outcome, but to what extent are they preventable? As these two concepts are closely related, at least for [clinical] laymen, clinicians and hospital management sometimes fear legal consequences when explicitly evaluating the contribution of HAI to the death of patients (chapter 6). Therefore, it is no surprise that there are very few relevant published studies. Decoster and colleagues combined their review of the contribution of HAI to death with an assessment of preventability and concluded that death was preventable in 35 of the 182 (19%) patients (McCabe score 0 and 1 only) in whom death was attributable to HAI [68]. In the 35 cases, the HAI itself could have been prevented and/or the following death. Baines and colleagues evaluated adverse events in Dutch hospitals and found that in 31% of the 186 deceased patients with a HAI as adverse event, the adverse event had been preventable [69]. Dantes et al. recently evaluated hospital‐onset bacteraemia and fungemia as a potential measure of HAI in a pilot study of 60 patients in three hospitals. Two‐thirds of all HOB events and half of non‐skin‐commensal HOB events were judged as potentially preventable [63]. The preventability of HAI has been studied more on the system level, using data on achieved reductions in improvement studies and regression analysis on risk factors. The landmark SENIC study concluded that 32% of HAI were preventable by well‐organised and highly effective infection control programmes [7]. Harbarth et al. evaluated the 10 245 General Discussion
RkJQdWJsaXNoZXIy MTk4NDMw