Continuous control of tracheal cuff pressure for VAP prevention: a collaborative meta-analysis of individual participant data

Background Underinflation of tracheal cuff is a risk factor for microaspiration of contaminated secretions and subsequent ventilator-associated pneumonia (VAP). The aim of this collaborative meta-analysis of individual participant data is to determine the impact of continuous control of Pcuff on the incidence of VAP. Methods Studies were identified by searching PubMed and references of relevant articles. Data from 3 prospective controlled trials (two randomized and one quasi-randomized), which evaluated the impact of continuous control of Pcuff on the incidence of VAP, were obtained and pooled together. Three different devices were used to continuously control Pcuff. VAP was diagnosed using clinical, radiologic, and quantitative microbiological criteria. The impact of continuous control of Pcuff on VAP was assessed by Cox regression analysis, stratified on trial. Results 263 (48.4 %) patients received continuous control of Pcuff, and 280 (51.5 %) patients received routine control of Pcuff using a manometer. 36 (13.6 %) VAP were diagnosed in continuous control group, and 72 (25.7 %) in routine care group (HR 0.47, 95 % CI 0.31–0.71, p < 0.001). However, heterogeneity was apparent in continuous control effect size across trials (I2 = 58 %, p = 0.085). The number of patients needed to treat to prevent one VAP episode was 8. No significant impact of continuous control of Pcuff was found on duration of mechanical ventilation, ICU length of stay, or mortality. Conclusion Continuous control of Pcuff might be beneficial in reducing the risk for VAP. However, no significant impact of continuous control of Pcuff was found on duration of mechanical ventilation, ICU length of stay, or mortality.


Background
Prevention of ventilator-associated pneumonia (VAP) is an important strategy to improve the quality of care provided to critically ill patients [1][2][3]. While VAPattributable mortality is still a matter for debate [4], this ICU-acquired infection is associated with increased antimicrobial use and duration of mechanical ventilation [5,6]. Important progress has been achieved during the last two decades in the understanding of pathophysiology of VAP, resulting in improvement in prevention strategies and reduced incidence of VAP [7,8].
Microaspiration of contaminated oropharyngeal and gastric secretions is the main mechanism of entry of bacteria into the lower respiratory tract [9,10]. Local and general host defenses play an important role in the progression from colonization to VAP [11]. The quantity and virulence of bacteria are also important factors in this process [12]. Several measures have been studied for prevention of microaspiration in intubated critically ill patient. These could be classified into enteral nutrition, mechanical ventilation, tracheal tube, and patient-related factors [13]. With regards to tracheal tube, several recent studies investigated how sealing could be improved, in order to avoid microaspiration. Subglottic secretion drainage has been shown to significantly reduce VAP incidence, antimicrobial use, and duration of mechanical ventilation [14,15]. Other preventive measures related to tracheal tube, such as polyurethane cuff and conical cuff shape have been suggested. However, a recent randomized controlled multicenter study did not report any significant impact of these measures regarding the rate of tracheobronchial colonization, or VAP [16].
Underinflation of tracheal cuff (<20 cmH 2 O) is a wellknown risk factor for microaspiration and VAP [17]. Therefore, it is recommended to adjust cuff pressure (P cuff ) around 25 cmH 2 O using a manometer, to prevent complications related to underinflation and overinflation of tracheal cuff [18]. However, in spite of routine control of P cuff using a manometer, underinflation and overinflation are very common in intubated patients [19,20]. Continuous control of P cuff has been suggested to improve tracheal sealing and to prevent VAP. Three prospective trials evaluated the impact of continuous control of P cuff on the incidence of VAP [21][22][23]. However, all these studies were performed in single centers and reported different results. Therefore, we performed this collaborative-pooled meta-analysis to determine the impact of continuous control of P cuff on the incidence of VAP in critically ill adult patients.

Methods
We established a collaboration to undertake this metaanalysis of individual patient data. We included all prospective trials, which evaluated the impact of continuous control of P cuff on the incidence of VAP.

Ethical aspects
The three studies used for this pooled analysis have been approved by local institutional regulatory boards. Informed consent was obtained from all patients, or from their next of kin.

Search for eligible trials
We identified clinical prospective clinical trials that compared the continuous control of P cuff and routine care regarding the incidence of VAP. We searched PubMed (from January 1995 through June 2015), using the terms "continuous control of tracheal cuff pressure, " "continuous control of endotracheal cuff pressure, " and the term "ventilator-associated pneumonia. " We also searched references of relevant articles. Studies comparing continuous control of P cuff and another intervention versus routine care were excluded, because it is impossible to determine the exact impact of continuous control of P cuff on VAP rate in these studies.

Outcomes
The primary outcome was the incidence of VAP. Secondary outcomes included duration of mechanical ventilation, mechanical ventilation-free days, duration of antimicrobial treatment, length of ICU stay, and ICU mortality.

Collected data
The rationale for choice of factors was based on prior association with outcome. At ICU admission: age, acute physiology and chronic health evaluation (APACHE) II score, sequential organ failure assessment (SOFA) score, male gender, cause for ICU admission, type of admission, diabetes mellitus, chronic obstructive pulmonary disease (COPD), chronic heart failure, cirrhosis, chronic renal failure, and immunosuppression. During ICU stay: SOFA score at randomization, subglottic secretion drainage, duration of mechanical ventilation before randomization, sucralfate, proton-pump inhibitor, or H2 receptor antagonist use, reintubation, mean P cuff , percentage of P cuff measurements <20 cmH 2 O, underinflation of P cuff (<20 cmH 2 O), overinflation of P cuff (>30 cmH 2 O), mean positive end expiratory pressure (PEEP) level, sedation, Ramsay score, head of bed elevation, paralytic agent use, red blood cell transfusion, enteral nutrition, ventilatorassociated tracheobronchitis (VAT) [24], and tracheostomy. At VAP diagnosis: polymicrobial VAP and type of microorganisms.

Statistical analysis
Quantitative variables are expressed as mean [±standard deviation (SD)] in case of Gaussian distribution, or median [interquartile range (IQR)] otherwise. Normality was examined using histograms and Shapiro-Wilk test. Qualitative variables are expressed as numbers (percentage). Patient characteristics at ICU admission and during ICU stay; and secondary outcomes were compared between the two study groups using Student t-test for quantitative variables (Mann-Whitney U test was used for non-Gaussian distribution) and Chi-square test for qualitative variables (Fisher's exact was used when the expected cell frequency was <5).
The probability of VAP occurrence over time was compared between the two study group using a Cox proportional hazard model stratified on trial. Heterogeneity across trials was examined by formal interaction test and quantified by calculating the I 2 . Patients were censored at the time of death or extubation. We performed an exclusion sensitivity analysis to evaluate the contribution of individual studies to the overall pooled estimate. A sensitivity analysis excluding patients receiving subglottic secretion drainage was also performed.
In both groups pooled together, we used a Cox proportional hazard model stratified on trial to identify factors associated with the occurrence of VAP. All variables with a p value < 0.20 were included into a backward-stepwise Cox regression analysis.
Statistical analysis was done at the 2-tailed α level of 0.05, except tests for the homogeneity in which an α level of 0.10 was chosen. Data were analyzed with the SAS software package, version 9.3 (SAS Institute, Cary, NC, USA).

Study characteristics
Among the 30 identified studies, 23 studies were directly excluded (reviews 9, duplicates 5, out of scope 5, others 4). Among the 7 remaining studies, 2 were excluded because P cuff control was not continuous, and 2 because other concomitant preventive measures were used in the intervention group (Fig. 1).
Data from 543 patients were obtained from the three included trials, representing 99 % of all included patients (5 patients were excluded for missing data). Two studies were randomized controlled and one was quasi-randomized controlled (patients who were admitted to an odd-numbered ICU cubicle received continuous control of P cuff , and those admitted to an even-numbered ICU cubicle received routine care). Three different devices (Mallinckrodt Pressure Control ® , VBM Medizintechnik GmbH, Sulz am Neckar, Germany; Nosten ® , Leved, Saint Maur, France; and an electronic artisanal device) were used to continuously control P cuff . Subglottic secretion drainage was used in some patients included in one trial. All tracheal tubes were polyvinyl chloride (PVC), standard shape-cuffed. Characteristics of the three studies are presented in Table 1.

Patient characteristics
Two hundred and sixty-three (48 %) patients received continuous control of P cuff and 280 (52 %) received routine care using a manometer. No significant difference was found in patient characteristics at ICU admission between patients who received continuous control of P cuff , and those who received routine care (Table 2). While mean P cuff was significantly higher in patients who received continuous control of P cuff compared with those who received routine care, rate of patients with underinflation of P cuff , with overinflation of P cuff , and percentage of P cuff measurements <20 cmH 2 O were significantly lower in patients with continuous control of P cuff compared with those who received routine care. Other patient characteristics during ICU stay were similar in the two groups ( Table 3).

Impact of continuous control of P cuff on outcomes
Continuous control of P cuff was associated with significantly reduced incidence of VAP, with a HR of 0.47 (95 % CI 0.31-0.71) (Fig. 2). However, heterogeneity in continuous control effect size across trial was apparent (I 2 = 58 %, p = 0.085). The effect of continuous control of P cuff to reduce the incidence of VAP remained significant in exclusion sensitivity analysis, with a lower effect after exclusion of the study of Nseir and colleagues (HR 0.58; 95 % CI 0.37-0.92, p = 0.019). In further sensitivity analysis, excluding patients receiving subglottic secretion drainage, the effect of continuous control of P cuff on VAP occurrence was not modified with a HR of 0.52 (95 % CI 0.33-0.79).
In multivariate cox regression analysis, continuous control of P cuff remained significantly associated with decreased risk for VAP [HR 0.42, 95 % CI (0.27-0.64), p < 0.0001], independently of subglottic secretion drainage, chronic renal failure, respiratory, and digestive failures at ICU admission.
No significant difference was found in VAT rate between patients with continuous control of P cuff and patients with routine care [13 of 263 (5 %) patients vs 22 of 280 (8 %), p = 0.227]. No significant impact of this procedure was found on duration of mechanical ventilation, mechanical ventilation-free days, antimicrobial treatment, length of ICU stay, or ICU mortality (Table 5).

Safety data and cost-effectiveness
Similar rates of tracheostomy and reintubation were found in patients who received in patients with continuous control of P cuff compared with patients with routine care (Table 3). No significant difference was found in tracheal ischemic lesions between the two groups (n = 96).
The number of patients needed to treat to prevent one VAP episode was 8.

Discussion
Our results suggest that continuous control of P cuff might be beneficial in reducing the incidence of VAP. However, continuous control of P cuff had no significant impact on secondary outcomes such as duration of mechanical Out of scope (n= 5) Others (n=4)

Fig. 1 Studies evaluated for inclusion in this analysis
ventilation, mechanical ventilation-free days, antimicrobial treatment, ICU stay, or ICU mortality.

Strengths and limitations
The major strengths of this study are the large number of included patients (n = 543) and the collaborative-pooled design. The similarity of the trials, including intervention, disease, and study population, provided a strong rationale to pool individual patient-level data. However, some limitations of our study should be acknowledged. First, the number of included studies was small, one study was quasi-randomized, and different devices were used to continuously control P cuff in these trials, which could probably explain the observed heterogeneity in estimates. Nevertheless, percentage of measurements of P cuff < 20 cmH 2 O in continuous control of P cuff group was quite similar in the three studies. Other differences between the three studies, including the lower rate of VAP in control group, and the higher percentage of patients with chronic respiratory disorders in one study [21] compared with the two others, could also explain the high heterogeneity. Second, no significant impact of continuous control of P cuff was found on secondary outcomes, such as duration of mechanical ventilation, mechanical ventilation-free days, antimicrobial treatment, or ICU mortality. This could be explained by the fact that a larger study sample is required to demonstrate such an effect. A posteriori calculation, based on the results of the current meta-analysis, of number of patients required to demonstrate a significant impact of continuous control of P cuff on outcomes indicates that 348 patients are required to demonstrate a reduction of VAP incidence of 12 % (from 26 to 14 %, p = 0.05, power 80 %), and 1132 patients are required to demonstrate a reduction of mechanical ventilation duration of 2 days (mean 14 vs 12 days, standard deviation 12, p = 0.05, power 80 %). Previous well-conducted randomized controlled studies aiming at evaluating a preventive measure of VAP also suffered from this limitation [14,26,27]. For example, several randomized controlled studies comparing subglottic secretion drainage to routine care demonstrated a significant reduction in VAP rate, but all failed to show any significant reduction in mortality rate or duration of mechanical ventilation. However, a large meta-analysis [15] performed on 2442 patients showed a significant reduction in mechanical ventilation duration in patients with subglottic secretion drainage compared to those with routine care. Third, subglottic secretion drainage was used in some patients of one trial, and might have influenced the results. However, the rate of patients who received subglottic secretion drainage was similar in patients who received a continuous control of P cuff and those who received routine care. In addition, sensitivity analysis excluding patients who received subglottic secretion drainage did not modify the protective effect of continuous control of P cuff on VAP incidence. Fourth, cost-effectiveness analysis could not be performed. However, the number of patients needed to treat to prevent one VAP episode was 8. Fifth, the safety of continuous control of P cuff , regarding tracheal ischemic lesions, was evaluated in only one study. Sixth, the under-reporting of negative results could have biased our results. However, we have checked the abstract of the major international critical care congresses and did not find any additional study on the impact of continuous control of P cuff on VAP incidence.

Explanations for study results
Rello et al. [17] have previously reported that underinflation of P cuff was independently associated with VAP in a subgroup of patients without antimicrobial treatment. In addition, one of the three studies included in this analysis [22] has investigated the impact of continuous control of P cuff on microaspiration of gastric content and tracheobronchial colonization. This study reported a significant (27 %) reduction of microaspiration of gastric contents, defined as the presence of pepsin in >65 % of tracheal aspirates, in patients who received continuous control of P cuff , compared with those who received routine care. Further, a significant reduction in bacterial concentration in tracheal aspirates was also observed. Several studies reported that routine care using a manometer was not efficient in continuously controlling P cuff , and that risk factors for underinflation and overinflation of tracheal cuff were not modifiable [19,28,29]. Other studies clearly showed the efficiency of some devices in continuously controlling P cuff , suggesting that their use might be beneficial in preventing intubationrelated complications [30,31].
Two randomized studies, excluded from this analysis, reported a beneficial effect of continuous control of P cuff on VAP incidence [32,33]. However, in these studies continuous control of P cuff was not the only tested preventive measure. While, low-volume low-pressure cuff was used in intervention group, PVC-standard cuff was used in routine care group. Therefore, it is impossible to determine whether the reduced incidence of VAP is related to  continuous control of P cuff or to low-volume low-pressure cuff use. This is the reason why these studies were not included in our analysis. Similarly, two other studies showed a beneficial effect of implementing a bundle for VAP prevention, including routine care for tracheal cuff [34,35]. However, whether this beneficial effect is related to P cuff control is unknown.

Different devices for continuous control of P cuff
The heterogeneity between the three trials was high, probably reflecting the use of different devices for P cuff control. An experimental study reported that electronic P cuff controllers with rapid pressure correction interfere with the self-sealing mechanism of high-volume, low-pressure PVC-cuffed tracheal tubes and reduce their sealing characteristics [36]. Further, a recent prospective crossover study compared the efficiency of a pneumatic with an electronic device aiming at continuously controlling P cuff [37]. The authors found underinflation of tracheal cuff to be more frequent using the electronic device than the pneumatic device (7 vs 0 %, respectively) and attributed this result to the over compensation of any elevated P cuff .

Future studies
Future randomized controlled multicenter studies are required to confirm the beneficial effect of continuous control of P cuff on VAP incidence, before recommending its routine use. In addition, the efficiency of different available devices should also be compared in critically ill patients.