Clinical usefulness of ventilator associated events in predicting ventilator associated pneumonia

Corresponding author: Dr Amarja Ashok Havaldar MICU, St John’s Medical College, Kormangala, Bangalore -560034, India amarjahavaldar@rediffmail.com Abstract Ventilator-associated pneumonia (VAP) is the most common nosocomial infection in intensive care unit (ICU) patients. Traditionally the clinical pulmonary infection score (CPIS) has been used in the diagnosis of VAP. There is no firm agreement among critical care and infectious disease specialists regarding the diagnosis of VAP. The aim was to study the association between new conditions, ventilator-associated conditions (VAC) and infection related ventilatorassociated complications (IVAC) and VAP and whether if any of these precedes the diagnosis of VAP by CPIS.


Introduction
Ventilator-associated pneumonia (VAP) is one of the most serious and common nosocomial infections in the intensive care unit (ICU). 1,2 Development of VAP is associated with increased length of ICU stay and mortality. Timely diagnosis of VAP is important and essential in order to plan further management. [1][2][3][4] Traditionally the diagnosis of VAP has been made with the clinical pulmonary infection score (CPIS) criteria. 5 The sensitivity and specificity of CPIS has been questioned. It has been proposed that early diagnosis of VAP may reduce the unfavorable outcomes of patients with VAP in the critical care setting. Also, in certain countries such as the United States, reimbursements for VAP-related expenses are routinely denied by third party insurances as a punitive action. 6,7 The US Centers for Disease Control and Prevention (CDC) proposed a new surveillance system for VAP, which included ventilator-associated conditions (VAC), infection related ventilator-associated complications (IVAC), and possible and probable pneumonia. These surveillance definitions were suggested by CDC for public reporting and interfacility comparisons, and VAP was used only in internal quality improvement. 5,8,9 After the development of this new surveillance system there have been many studies trying to evaluate the utility of this newer system of reporting and comparing it with existing CPIS criteria. [10][11][12][13][14][15] There are certain changes as per the 2015 CDC update, in which possible and probable VAP are combined into one as against earlier CDC update. There is also specific mention about the list of pathogens excluded as a cause for VAP. Special mention of episodes on mechanical ventilation (EMV), ventilator free days and airway pressure release ventilation (APRV) days. 16,17 For analysis of data these changes were taken into account.
The aim of our study was to evaluate the association between diagnosis of VAC, IVAC and VAP in our subset of patients.

Materials and Methods
After obtaining institutional ethical committee approval (ref 2014/226A), this study was conducted in the medical ICU of a tertiary care center (Mazumdan Shaw Medical Centre, Narayana Health, Bangalore) for a period of 3 months (1st January 2014 to 31st March 2014). Inclusion criteria for the study were patients requiring invasive mechanical ventilation support for > 2 calendar days. Patients transferred from other healthcare facilities and patients unlikely to survive > 48 hours were excluded from the study.
VAC was defined as per the CDC guidelines as increase in FIO2 by 20% or increase in positive end expiratory pressure (PEEP) of 3 cm H2O after 2 calendar days of stability. Diagnostic criteria for IVAC included patients with VAC who developed temperature >38°C or <36°C, or WBC count >12000/mm3 or < 4000/mm3 and commencement of a new antimicrobial agent(s) and their continuation for >4 calendar days. VAP was diagnosed by modified CPIS score, which includes temperature, leucocyte count, tracheal secretions, oxygenation and chest radiograph findings. The modified CPIS score does not include microbiology or culture as against CPIS score. Patients with modified CPIS score > 6 were diagnosed as having VAP. In our study microbiological data were not included in CPIS calculations. Patients were followed up until ICU discharge.
Statistical analysis was done using STATA 13 (StataCorp, College Station, TX). Continuous variables were expressed as mean + SD (standard deviation) or median IQR (interquartile range). Continuous variables were analyzed by 't' test and categorical variables were analyzed by 'χ2' test. A standard 95% confidence limit and a p value at 0.05 were used for assessing statistical significance.

Results
One hundred consecutive patients were screened for the study. Out of these 13 patients were excluded (1 patient was transferred to another healthcare facility, 6 patients expired, and 6 patients got extubated within 48 h). Among the baseline characteristics, mean APACHE II score was 20 (SD 6.04) ( Table I).
Out of 87 patients, 15 satisfied criteria for diagnosis of VAC, 13 for IVAC, and 7 for VAP by CPIS. Figure 1 shows seven cases of VAP and the day on which VAC, IVAC and VAP diagnosis was made. Diagnosis of VAC not for citation purposes Clinical usefulness of ventilator associated events in predicting VAP Havaldar et al.  Figure 1. Baseline mortality was around 31%. Patients who developed VAC, IVAC or VAP had higher mortality and more ventilator days (Table I).
Among 13 patients with IVAC, 11 patients were classified as pneumonia. One patient who satisfied the criteria for IVAC was subsequently diagnosed as tumor lysis syndrome. In two patients diagnosed as IVAC by CDC classification, diagnosed as VAP as per the CPIS, lung infiltrates were not present on chest X ray and hence should accurately be diagnosed as ventilator associated tracheobronchitis (VAT) (Figure 2).
Fisher's exact test was performed to look for the association between VAC, IVAC and VAP. There was a statistically significant association among VAC and VAP, χ2 (1)  developed VAP (46.66%). In our study there was a higher incidence of VAC and IVAC as compared with the meta-analysis data of Boyer's study. 13,14 The diagnosis of VAP in the ICU can be quite challenging. Historically chest X-ray was used initially to diagnose VAP. The study by Wunderink et al showed that there was no good correlation between radiological signs of pneumonia with autopsy findings. 18 The major limitation for using X-ray is that it has nonspecific findings and abnormalities may not accurately reflect infection as a cause of Xray abnormalities. There can also be inter-observer variability. Diagnosis by clinical signs and symptoms such as fever, leukocytosis can also nonspecific. Evaluation of different scoring systems has not reached any consensus in the critical care field. 12 The routine CPIS score which comprises of different parameters such as fever, leukocytosis, radiological, tracheal aspirate, oxygenation and microbiology, is associated with variable sensitivity and specificity as reported in the different studies. 6,7 The modified CPIS score which excludes microbiological data is also shown to have limitations. 6,7 One study comparing CPIS vs. CDC criteria for different facets of VAP showed no clear advantage either in terms of reliability or specificity. Hence CDC criteria in the diagnosis of VAP has been used for surveillance purposes only. 11 A multi-center study showed high predictive accuracy for ventilator associated events (VAE) with respect to tracheobronchitis but not VAP. It also showed that a differentiation between tracheobronchitis and VAP is essential as there is a significant difference in the mortality between the two conditions. 15 Results of this study were consistent with the meta-analysis data by Yunzhou Fan et al., which showed VAE could not accurately and consistently detect cases of VAP. 13 In our study, two patients with VAP who had no new lung infiltrates can be classified as VAT who survived, as compared with five patients with VAP who died. In our study more severe VAP cases were captured with compromised oxygenation for > 24hr as reflected by serial CPIS.
There were certain limitations to our study. It was a single center study. Results of the statistical analysis on our data should be taken in the context of low number of patients studied. In the exclusion criteria, we excluded patients who died or extubated within 48hrs. Ideally the cut off should have been 96 hours to correctly identify VAE episodes. 13 However, the present study did find a statistically significant association between VAE and 28-day mortality. These results were in accordance with study by Boyer et al. 14 Possible reason for such results could be use of ultrasonography was already in place and that could have minimized interobserver variation in interpreting CXR findings, allthough use of ultrasonography was not a part of study methodology.
The CDC criteria for diagnosis of VAE could alert the possibility of developing IVAC.VAE is a useful tool for surveillance purposes but cannot be used for diagnosis of VAP. In our opinion the modified CPIS score is still useful in the diagnosis of VAP. Nonetheless it is important to recognize these

Conclusions
Our study has shown a statistically significant association between VAE and VAP. The results of our study need to be validated by larger multicenter studies. [20][21][22] Looking at recent studies use of VAE for surveillance itself is questionable and needs modifications.