The epidemiology and clinical outcomes of ventilator-associated events among 20,769 mechanically ventilated patients at intensive care units: an observational study

Most critically ill patients admitted to intensive care units (ICUs) require life-saving mechanical ventilation (MV), despite the multiple complications associated with it [1‐3]. Ventilator-associated pneumonia (VAP) is one such complication and is the most common hospital-acquired infection (HAI) in ICUs [4, 5]. It has been reported that VAP is associated with longer duration of MV, prolonged hospital and ICU stays, long-term disability, higher mortality, and increased hospital costs [1, 2, 4, 6, 7]. Therefore, the surveillance of VAP for the incidence estimates is particularly important for understanding the epidemiology and risk management of VAP.

Traditional definitions of VAP, however, were subjective, complex, and poor both for sensitivity and specificity, which made it difficult to implement [3, 8‐13]. In January 2013, a new approach to VAP surveillance—ventilator-associated events (VAEs)—was proposed by a working group convened by the Centers for Disease Control and Prevention (CDC) and comprised members of several stakeholder organizations [3]. The following three definition tiers are embedded within the VAE algorithm: (1) ventilator-associated condition (VAC), (2) infection-related ventilator-associated complication (IVAC), and (3) possible VAP (PVAP). In contrast to traditional VAP, VAE surveillance definitions were objective, streamlined, and potentially automatable that can identify a broad range of conditions and complications, more than VAP, only occurring in mechanically ventilated adult patients [8, 13‐15].

Although it has been more than five years since the VAE surveillance was proposed and has replaced the traditional VAP surveillance, its adoption outside the USA seems limited [16]. Several published studies have reported the incidence of VAE. However, the rates varied markedly, ranging from 3 to 77% (6–107 per 1000 ventilator-days) [1, 2, 4, 7, 17‐36]. Explanations for the discrepancy may be various study settings and different patient populations used as the denominator for the calculation of VAE rate. These studies were mainly from the USA, followed by Europe, while the population used as the denominator included all MV patients or patients with at least a certain number of ventilator-days. Besides, the majority of these studies were retrospective, in that the completeness and accuracy of data had a great impact on case identification. In China, the surveillance was mainly limited to traditional VAP. Studies on VAE incidence are scarce [7, 21], which may further limit the understanding of comprehensive complications among mechanically ventilated patients in China.

In 2015, a routinely active monitoring module for VAE was established and embedded in an existing ICU-HAI system (an HAI monitoring system in ICU units) in West China Hospital (WCH), China. To our knowledge, this is a unique system carrying out routine surveillance for VAE in China. By linking the HAI system with the electronic medical record (EMR) and ICU systems, we have developed an ICU-HAI registry. We undertook an observational study based on this registry with the aim to evaluate the following: (1) the clinical characteristics and outcomes between patients with and without VAEs; (2) the incident rates of different VAE tiers; and (3) the pathogen distribution in patients with PVAP.

This study was approved by the Ethical Committee of West China Hospital in 2018 (WCH2018-409), and the need for patient consent was waived.

The VAE rates varied significantly in previous studies, mainly because the denominator used to calculate VAE rate differed among individual studies. For instance, some studies only recruited patients with at least 48 h (7–40.8 per 1000 ventilator-days) [1, 2, 19, 25, 27, 30, 31, 38], 4 days (6–13.8 per 1000 ventilator-days) [21, 22, 24, 27, 35] or 5 days (107 per 1000 ventilator-days) [4] on a ventilator, while some studies included all mechanically ventilated patients (6.3–14.4 per 1000 ventilator-days) [7, 17, 23, 26, 28, 33, 34, 36]. The rates of VAC-plus in our study were slightly higher than in studies with the same VAE definition and all MV episodes as the denominator. This is likely because VAE cases were prospectively and actively identified in our study. The values of PEEP and FiO2 were recorded on an hourly basis. A threshold-based warning system derived from these records was implemented. Once PEEP or FiO2 reached the threshold, an alarm would be triggered, and infection control practitioners would immediately check the patient. The system provided a useful approach to identify all potential cases, leading to a higher rate than that of other retrospective studies. Moreover, as a national critical care center in Western China, ICU patients at WCH had relatively serious illnesses and thus may be more prone to developing VAE. The rates in studies that restricted eligibility to patients with at least 2 ventilator-days were almost higher than 20 per 1000 ventilator-days [4, 19, 25]. In addition, differences in rates depended on ICU type as shown in our study and previous studies [17]. We found that the VAC-plus rates were 23.7, 21.4, and 7.3 per 1000 ventilator-days in the SICU, NICU, and TICU, respectively, whereas Klomps et al. reported 16.0, 9.8, and 12.9 per 1000 ventilator-days in these three units, respectively [17]. Zhu and Magill et al. also found that the VAE rates of the major teaching hospitals were much higher than those in non-major teaching municipal hospitals [7, 28]. In the three tiers of VAE definition, PVAP is a closer proxy for traditional VAP, and rates were 0.9 per EMV and 2.2 per 1000 ventilator-days in our cohort, which was similar to some previous studies [7, 17, 21, 38].

Most patients may be more prone to VAEs early in the ventilation episode. The median time from the start of ventilation to the first event onset in this study was 5 days for both the VAC-plus and IVAC-plus groups, while it was 6 days for the PVAP group, consistent with other reports [1, 2, 4, 17, 19]. In this study, most VAEs occurred within the first two weeks from the initiation of MV. The mean rate in the first week was more than three times that in the second week and dropped sharply to less than one per 100 patients after 14 days. Similar trends were shown in the study by Klomps et al., wherein patients were more prone to VAEs early in the course of MV as a consequence of the acute interventions performed to stabilize the patient’s presenting illness [17].

Due to the poor correlation between VAE and the traditional definition of VAP, the new approach has not been widely implemented outside the USA and is mainly used for surveillance [16]. However, VAE is not designed to be a proxy for VAP, rather it was intentionally to broaden the surveillance from targeting pneumonia only to include the complications associated with ventilation in healthcare settings [13]. Therefore, many VAEs may be conditions other than pneumonia. Nevertheless, surveillance aims to identify patients with severe complications and to track the impact of prevention strategies. With shifting the focus on all major causes of respiratory deterioration in ventilated patients, VAE surveillance imposes a severity threshold to identify a subset of patients with severe disease. Indeed, previous studies have found that patients without impaired gas exchange are associated with a relatively benign clinical course [39]. Compared to those with traditionally defined VAP, patients with VAE had a higher mortality rate as demonstrated by our study group [7] and that others [2, 40]. This confirms the clinical importance of VAE surveillance. In addition, traditional ventilator bundles were primarily based on subjectively and nonspecifically traditional VAP definition, but some components proved to be harmful [41‐47]. VAE surveillance, which is defined based on objective ventilator data, is less vulnerable to misattributing benefit to neutral or negative interventions [13]. Several potential strategies for preventing VAEs have been proposed including minimizing sedation, paired daily spontaneous awakening and breathing trials, and conservative fluid management [44]. Studies have suggested that the VAE prevention bundles were associated not only with lower risks of VAE, but also with less time to extubation and shorter LOS in hospital [2, 45, 48‐51]. In general, VAE is a relatively new algorithm to address complications associated with ventilation including, but not restricted to, VAP. This is quite different from the traditional approach to address VAP alone and therefore brings challenges for the long-standing thinking pattern and practice routines in patient care. It may take time for such a relatively new algorithm to be adopted widely in clinical practice. More studies are warranted to further demonstrate the preventability of VAE and its clinical significance on patient outcomes in different countries and among different patient populations. These studies will be likely to generate more high-quality, convincing, and patient care-focused (in addition to the surveillance purpose) evidence to further demonstrate the clinical relevance of VAE. Meanwhile, as VAE is simple and can be easily applied, we recommend clinicians to raise awareness towards VAE and to consider performing observations of the incidence and impact of VAE for their patients.

This study was based on a routinely prospective surveillance system, which is so far unique in China. Compared with retrospective investigations, it is less likely to miss VAE cases, and the diagnosis of VAE has been validated as relatively accurate. However, our study has some limitations. Our findings were based on data from one tertiary hospital, which may not be generalizable to other settings. The preferential practices may vary remarkably in ICU patients with MV across hospitals with different levels or in different countries. Second, as we did not monitor traditional VAP, we were unable to compare the differences between traditional VAP and PVAP.

The majority of ICU patients required MV. VAEs were common among those with ≥ 4 ventilator-days and occurred early in the course of MV. All tiers of VAE were highly correlated with poor clinical outcomes, including longer stays in the hospital and ICU and an increased risk of mortality. These findings suggest the importance of VAE surveillance and of developing strategies to prevent VAEs. However, uniformed inclusion criteria and patient populations used as the denominator for calculation of VAE rate should be carefully addressed in future studies.