More surprisingly, because unlike other studies where the risk for older patients was similar or increased[3, 11, 17, 18] younger patients (16‐40 years) appeared to be at increased risk. This may have resulted from an overrepresentation of young patients in neurosurgery and traumatology, both specialties with high VAP rates. This association remained borderline significant when adjusted for specialty and Apache II score or when these two specialties were excluded from the analysis. When excluding one center, where 11 of the 32 patients aged 16‐40 developed pneumonia, the association was still present (HR 3.2 (0.49 – 21.2), although not significant anymore. While COPD is generally found to be either unassociated with VAP[3, 16], or to increase the VAP risk[11, 17], here COPD appeared to be associated with a lower risk. Our model did not detect a significant interaction with systemic antibiotics. In previous studies, COPD patients had longer ventilation durations relative to patients without COPD[3, 34]. In our study, the first ventilation periods for patients with and without COPD were similar in duration (average of 8.2 and 8.4 days, respectively), as was the total ventilated duration (9.3 and 8.7 days) and the proportion of ventilation periods lasting 28 days(1.4 and 2.8%, respectively), suggesting that the similarities in ventilated duration are not subject to censoring bias. Furthermore, mortality data, available from five hospitals, showed that mortality was comparable (40.4% and 38.5%, respectively). The lower risk in our study may reflect differences in criteria for COPD diagnosis[35], but also improved care. For example, Funk analysed the outcomes of COPD patients admitted in ICU, for the period 1998–2008, and found that risk‐adjusted mortality had improved[36]. The use of jet nebulizers demonstrated a slightly reduced risk compared to no inhalation therapy, whereas MDI did not. Appropriate delivery of medication is reported to be more challenging with MDI than with jet nebulizers. In one study no difference in VAP risk was found between both methods, but the population size was small[37]. For the daily measured ivAB, ivSDD, and inhalation therapy, the WCE approach explained the associated VAP risk better than the standard Cox regression analysis. Additionally, this method provides insight into the cumulative effect over time and the relevant retrospective timeframe in which exposures are associated with a VAP. IvSDD affected this hazard for 24 days. It must be noted that in one hospital many patients received ivSDD throughout the entire ventilation period instead of the currently recommended first four days only. Although differences in the AIC were usually small (Table C in S1 File) the WCE approach identified ivSDD as an independent factor affecting the risk to develop VAP whereas standard Cox regression of current values or of one or two days earlier did not. The WCE approach is worth considering in future analysis of time‐dependent risk factors of VAP, and for other device‐associated infections. 78 Chapter 4
RkJQdWJsaXNoZXIy MTk4NDMw