DISCUSSION In this manuscript, we presented data generated by the VAP surveillance module of the Dutch PREZIES network. The VAP rates reported by the hospitals participating in the surveillance (0‐20.1/1000 ventilation days) are in the expected range[1‐3]. The observed variation between hospitals could be partly due to the inter‐observer reliability of a VAP diagnosis, which is known to be low[31, 32]. However, this applies less to intrahospital comparisons where only one intensivist or radiologist was usually dedicated to surveillance. Four of the five hospitals that participated ≥2 years demonstrated a reduction in VAP, two of which significantly so. Apart from a possible effect of the surveillance itself[33], various changes during the follow‐up could have caused this reduction. Several hospitals (C‐G) implemented interventions (SOD and/or ivSDD, Evac cuffs, closed suctioning system, VAP bundle), sometimes temporarily. Although the increase in VAP incidence in hospital F was duly investigated at that time, no cause was found. Hospital A did not introduce any interventions. Hospital B, with zero VAPs, used a complete SDD regimen during the surveillance. Therefore, while hospitals appear to be able to reduce VAP, either by introducing an intervention or by surveillance alone, success does not seem to be guaranteed. The observed effects of patient and treatment characteristics vary among studies[9, 11‐18], resulting from differences in the other measured covariates and case mix, and, frequently, low statistical power. In our results, Apache II score, specialty, intubation site, length of hospital/ICU stay before ventilation, postsurgical ventilation, inhalation trauma, stress ulcer prophylaxis, corticosteroids, neutropenia, intestinal prophylaxis, and feeding method were not significantly associated with VAP. Although the overall study population in our study was relatively large, low numbers of patients in certain categories could explain failures to detect associations. In our data, a higher Ramsay score was associated with an increased VAP risk, which corresponds with the increased risk associated with coma or increased Glasgow coma scale[9‐12, 16, 17]. Systemic antibiotics (not for ivSDD) appeared to lower the VAP risk, as identified by others[9, 10]. In most analyses where antibiotic use was not analysed as a time‐dependent variable, ‘prior antibiotic use’ (or variations) was not found to be associated with VAP[12, 16, 17]. Our results demonstrated that both ivSDD and SOD were associated with a VAP reduction. De Smet et al, performing a multicenter randomized clinical trial, concluded that both SOD and SDD (full regimen) led to a reduction in respiratory tract colonisation with highly resistant microorganisms[21]. As intestinal prophylaxis partly coincided with ivSDD and SOD, we evaluated a model without ivSDD/SOD. Intestinal prophylaxis was not significantly associated with VAP in this model either. 4 77 Risk factors for VAP using flexible methods
RkJQdWJsaXNoZXIy MTk4NDMw