We found that only a minority of patients who had VAP also met the diagnosis of VAE. The concordance between two surveillance criteria was poor. The incidence of VAP was 13.66 per 1000 ventilator days in ICU which was similar with a recent multicenter study that reported a prevalence of 22 per 1000 ventilator days (15). However, VAE rate as 6.16 per 1000 ventilator days was lower than the previous studies which reported VAE rate as 12.5–17 per 1000 ventilator days (16–18). VAP and VAE definitions are focused on different targets and the correlation between both classifications was shown to be poor in previous studies (19, 20). Persistent deterioration of oxygenation is the first criterion to detect VAC. A possible explanation of the low concordance between VAP and VAE is that some cases of VAP may not be severe enough to cause sufficient deterioration in mechanical ventilator settings to reach thresholds for the diagnosis of VAC. A dedicated team work is provided at ICU led by an experienced intensivist, continuously. Moreover, an infectious disease team round at the ICUs daily. Improvement in the quality of care of the ICU patients by infectious diseases consultation was recently reported from our center including the pandemic period (21). This approach might have an impact to prevent significant deterioration in oxygenation with early intervention and treatment, and then we may not be able to detect these cases by using VAE surveillance.
The diagnosis of VAE requires a dynamic monitorization of respiratory functions. Previously we observed a high variance of VAP rates according to the surveillance criteria in surgical ICUs where intensivists were not in charge. However, the rate of VAP in the medical ICU did not change according to the surveillance criteria at consecutive periods (22). Patients were placed at single bed rooms with negative pressure and HEPA filters in ICU during pandemic. We were able to allow the junior doctors who worked in ICU to rest for 48 hours after 12 hours shift, but this was not possible for intensive care fellows who were primary responsible for the respiratory care. Although we had a certain ventilation protocol, it is not easy to rule out some deviations from this protocol during the night shifts when intensive care fellows were under heavy workload. This can be a possible limitation for the diagnosis of VAE.
The most common bacteria that caused VAP was A. baumannii in ICU (Fig. 2) where we experienced a significant increase in the rate of A. baumannii infections during the second wave of COVID-19 in third quarter of 2020 (23). The findings of this study showed that we continued to face with difficulties to prevent the spread of A. baumannii in ICU, however, the team including intensivist and infectious diseases specialist were ready to treat the infectious complications before they effect the oxygenation permanently. Although the mortality rate of the patients with VAP was as high as 50%, there was no statistically significant differences in the mortality rates and length of stay at ICU between patients with VAP and patients without VAP (Table 2).
Despite VAE detected a lower rate of infectious complications of MV (Table 4), this approach enables to detect other causes of respiratory failure. ARDS was found to be the cause of VAEs in more than 50% of patients with COVID-19 who developed VAE (17). Similarly in our study, ARDS was the cause of the majority of ventilator-associated non-infectious respiratory complications in these patients. A meta-analysis of 18 studies from 8 countries showed that pooled sensitivity and positive predictive value of each VAE type for VAP detection did not exceed 50% which points out that VAE misses a high rate of patients with VAP (24). In hospital mortality rates were higher in VAE in this meta-analysis. However, 30-day mortality rates (40% vs 51.5, p = 0.48) and in hospital mortality (80% vs 84.8, p = 0.68) rates were not statistically different when the VAE episodes were compared with VAP episodes in our study. The median duration of ICU stay was 25 days for VAP and 28 days for VAE (Table 2, 3). These findings raised the question about the impact of a missing diagnosis of IVAC or PVAP according to VAE which did not influence the outcome. If the surveillance on multidrug resistant bacteria just relies on the invasive device related infection rates, the missed cases by using VAE criteria can cause a significant disinformation about the epidemiology of antibiotic resistance in ICU.
Table 4
Relationship between VAP, VAC and IVAC
| VAP as The Comparison Standard (n = 33) |
Condition as Compared with VAP | Presence of Both VAP and VAC or IVAC | Sensitivity, % | Specificity, % | Positive Predictive Value, % | Negative predictive Value, % |
VAC (n = 15) | 11 of 33 (33.3%) | 33 | 90 | 73 | 64 |
IVAC (n = 10) | 8 of 33 (24.2%) | 24 | 95 | 80 | 63 |
VAP: Ventilator associated pneumonia, VAC: Ventilator associated condition, IVAC: Infectious ventilator associated complication |
There were several limitations of this study. It was a single center observational study included limited number of patients. Microbiological sampling depended on the request of the physician in charge for the patient and an endotracheal aspirate culture was not requested from two patients with IVAC and seven patients with VAP. While we did not detect any difference in the rates of VAP and VAE between prospective and retrospectively investigated patients, there were 34 patients with missing information who were excluded form the study. The risk of subjective interpretation of plain radiography was solved by enabling two independent researchers for radiological evaluation which is not an easy task for daily practice.
As a conclusion, despite the enhanced capacity of VAE to detect non-infectious MV associated problems, it has a potential to underestimate the infectious complications of MV. NHSN/VAP criteria detected earlier and higher rates of infectious complication of MV, but it is time consuming and requires an experienced work force to avoid subjectivity. In the light of current study, we decided to continue VAE surveillance to detect major problems in MV with a continued surveillance of incidence of multidrug resistant bacteria from respiratory samples regardless of IVAC and VAC diagnosis. We believe that this approach will enable to keep our infection prevention team at bedside instead of standing by a computer to line up dozens of portable plain graphs to compare. The role of infection control nurses to encourage the staff on compliance for basic infection prevention precautions such as hand hygiene and mechanical ventilation care bundles is critical (25). Considering the well-known saying “One-size does not fit all”, each ICU has unique features and priorities, so, feasibility and usefulness of VAP or VAE criteria in COVID-19 patients should be assessed individually.