Introduction

Ventilator-associated pneumonia (VAP) is one of the common nosocomial infections in intensive care units (ICUs). It is reported that 6–52 % of mechanically ventilated patients develop VAP [1–4]. VAP is associated with prolonged ICU and hospital stays [5, 6] and mortality [7, 8]. The annual cost for VAP is considerable and approximated $3.0 billion USD [9]. Thus, the prevention of VAP has substantial merits from the clinical and societal perspectives.

Currently, two types of endotracheal suctioning systems are available: closed tracheal suction systems (CTSS) and open tracheal suction systems (OTSS). CTSS allow multiple episodes of endotracheal suctioning without disconnecting the patient from the ventilator. Their suggested advantages compared with OTSS use include limited environmental and personnel contamination [10], maintained positive end expiratory pressure, lung volume, and oxygenation [11, 12], and fewer physiologic disturbances during suctioning such as decreased arterial deoxygenation, increased heart rate, and increased mean arterial pressure [13, 14].

Another potential advantage of CTSS use is the prevention of VAP. Previous systematic reviews and meta-analyses have concluded that CTSS use has no benefit over OTSS use in preventing VAP [13, 15–20]. However, they were based on a relatively small number of trials, and some even suggested a potential publication bias [13, 20]. Trials published in some non-English languages were not included.

Therefore, we conducted a systematic review and meta-analysis to reassess the efficacy of CTSS use to prevent VAP in mechanically ventilated adult patients, in comparison with OTSS use. Recent trials and trials published in non-English languages were included, and trial sequential analysis was conducted to challenge the robustness of the available evidence.

Materials and methods

Study selection

The study protocol was pre-specified as a protocol and followed the preferred reporting items for systematic reviews and meta-analyses (PRISMA) Statement for reporting on systematic reviews [21]. We searched PubMed, the Cochrane Central Register of Controlled Trials, and the Web of Science. The search strategy was listed in the protocol (Supplementary File). Clinicaltrial.gov, Google Scholar, and the references of retrieved articles and previous systematic reviews were reviewed for potentially relevant trials. No language restrictions were placed on the search. The last search was conducted on October 27, 2014.

Eligibility criteria

Randomized controlled trials that compared CTSS and OTSS use in adult patients (aged 18 years or over) on mechanical ventilation in ICUs were included. Trials published in abstracts were also included if pertinent data on patients’ characteristics and outcomes of interest were available. Crossover trials and trials conducted on neonates and infants were excluded.

Data abstraction and risk of bias assessment

At least two of the reviewers (A.K., N.U., and J.F.) independently extracted the following information in duplicate: (1) study characteristics (the types of ICUs, sample size, inclusion or exclusion of patients with pneumonia at admission to ICUs, and definitions of VAP); (2) participants’ demographics (age and sex); (3) interventions (CTSS brand names, the cycle of CTSS exchange, industry sponsorship); and (4) outcomes of interest. Attempts were made to contact the original authors for more details. The authors were considered unresponsive, when three e-mails were sent and no reply was obtained. At least two of the authors (A.K., N.U., and J.F.) independently assessed the risk of bias, using the Risk of Bias tool recommended by the Cochrane Collaborations [22], as well as the sponsorship. Any disagreement was resolved through discussion.

Statistical analysis

The primary outcome was the incidence of VAP. Any definitions of VAP were allowed. Secondary outcomes were mortality and length of mechanical ventilation. When trials had more than one arm for the intervention, the data were pooled into a single group [22]. When trials had zero events in either arm, continuity corrections were applied with addition of 0.5 to each cell of 2 × 2 tables from the trial [23]. Dichotomous outcomes were combined and presented as risk ratios (RRs) with associated 95 % confidence intervals (CIs). Length of mechanical ventilation was combined using weighted mean differences (WMD). We a priori knew that the included populations would be clinically heterogenous, and thus data were pooled using the DerSimonian and Laird random-effects model [24]. Statistical heterogeneity was assessed with the I 2 and Q statistics [25]. When significant heterogeneity was identified (I 2 ≥ 50 % or p < 0.1), meta-regression analysis was conducted to investigate the potential sources of heterogeneity. We performed subgroup analysis stratified by the ICU population, as done in the Cochrane review [19] and on our hypothesis that there is no difference in the effect size among subgroups. A subgroup analysis stratified by the cycle of exchanging CTSS and meta-regression analysis to examine the relationship between the cycle and the effect size were also conducted. The test-of-interaction was performed in subgroup analyses [26]. Sensitivity analyses were conducted by excluding trials that included patients with pneumonia at admission, and by excluding trials with unclear or high risk of bias in any domain of sequence generation, allocation concealment, or blinding of outcome assessors. Publication bias was tested using Egger’s method [27]. All these analyses were conducted with Stata v.11.2 (Stata, College Station, TX, USA).

A meta-analysis may suffer from the type I error due to an increased risk of random error due to sparse data, and due to repeated significance testing when meta-analyses are updated with new trials [28]. Sensitivity analysis with trial sequential analysis (TSA) on our primary outcome (VAP) was additionally performed to adjust for random error and repetitive testing. Meta-analysis monitoring boundaries and required information size (cumulated sample size of included trials) were calculated, with D2 (diversity adjusted information size) and adjusted 95 % CIs. The idea behind TSA is that if the cumulative Z-curve crosses the boundary, a sufficient level of evidence is reached and no further trials are needed. If the Z-curve does not cross the boundary and the required information size is not been reached, evidence is insufficient to allow investigators to conduct further trials for a conclusion. We conducted TSA to maintain a type I error of 5 %, and calculated the required information size with an anticipated intervention size of 20 % relative risk reduction (RRR), at a power of 80 % [29]. We conducted TSA with TSA 0.9 (The Copenhagen Trial Unit, Copenhagen, Denmark).

Results

The search produced 400 articles (Supplementary Fig. 1). After application of inclusion and exclusion criteria, 16 randomized, controlled trials that compared CTSS and OTSS were identified [14, 30–44]. Two major CTSS sales companies were contacted, but no new information was obtained. A total of 1,929 mechanically-ventilated patients was included in the analysis (Table 1). The mean age of participants was 48.3 years, and 29 % was women. The median sample size was 74. The follow-up periods were reported in six trials, and ranged from 7 to 31 days [30, 33, 34, 37, 41, 44]. Four trials were conducted in medical ICUs [32, 35, 36, 40], four in surgical ICUs [14, 30, 31, 38], and six in mixed (medico-surgical) ICUs [33, 34, 37, 39, 43, 44]; the remainder was unclear. Six trials used Trach Care, two Steri Cath, two Hi Care, and one Ty Care as the CTSS; the CTSS brand was unclear in five trials. The cycle of CTSS exchange varied considerably; nine trials exchanged CTSS every 24 h [14, 30, 31, 34, 35, 38, 40, 42, 44], one every 72 h [37], one every 168 h [32], and one trial included groups that exchanged CTSS every 24 and 48 h [43], respectively. All trials except two excluded patients with pneumonia at the onset of the studies [32, 41]. Fourteen trials set VAP as the primary outcome [14, 31–40, 42–44], four of which a priori calculated the sample size [32–34, 37]. Twelve trials were reported in English, two in Chinese [43, 44], one in Korean [37], and one in Arabic [42].

Table 1 Characteristic of the included trials
Full size table

Study quality

Overall, four trials (25 %) had adequate sequence generation, whereas two (13 %) had adequate concealed allocation (Supplementary Table 1). Outcome assessors were judged to be adequately blinded in four trials (25 %). Only one trial was assessed as overall low risk of bias [32]. Two studies (13 %) disclosed the involvement of industry sponsorship; CTSS was provided for one trial, and a support grant was given for the other trial.

Primary outcome analysis

Use of CTSS was associated with a reduced incidence of VAP compared with OTSS (RR 0.69; 95 % CI 0.54–0.87; Q = 26.14; df = 14; I 2 = 46.4 %, p = 0.03) (Fig. 1). The point estimates of the effect size were similar across the subgroups (test-of-interaction p = 0.95) (Table 2). No publication bias was evident (p = 0.13). The pooled data had moderate heterogeneity. Meta-regression analysis showed that sample size (p = 0.06), age (p = 0.35), sex (p = 0.50), publication date (p = 0.53), publication type (p = 0.65), single- or multicenter study (p = 0.75) and any risk of bias were not the source of the heterogeneity. Compared to OTSS use, the use of CTSS was not associated a reduced incidence of VAP in the only one trial with overall low risk of bias (RR 0.56; 95 % CI 0.27–1.14), whereas pooled results from the trials with overall unclear or high risk of bias showed beneficial effect (RR 0.70; 95 % CI 0.54–0.90; Q = 25.51; df = 13; I 2 = 49.0 %, p = 0.02).

Fig. 1
figure 1

Comparison of CTSS and OTSS on the incidence of ventilator-associated pneumonia. CTSS closed tracheal suctioning systems, OTSS closed tracheal suctioning systems

Full size image
Table 2 Subgroup analyses by the type of ICU populations
Full size table

Secondary outcome analysis

Compared with use of OTSS, use of CTSS was not associated with reduction of mortality (RR 0.96; 95 % CI 0.83–1.12; Q = 2.27; df = 6; I 2 = 0.0 %; p = 0.89) (Supplementary Fig. 2) or a shorter length of mechanical ventilation (WMD −0.45 days; 95 % CI −1.25 to 0.36; Q = 6.37; df = 6; I 2 = 5.8 %; p = 0.38) (Supplementary Fig. 3). No publication bias was evident in mortality (p = 0.94) or length of mechanical ventilation (p = 0.90).

Subgroup analysis stratified by the cycle of CTSS exchange

An analysis of each outcome stratified by the cycle of CTSS exchange was conducted (Table 3). Use of CTSS was associated with a reduced incidence of VAP in the subgroup with 24-h and 72-h cycles, and reduced length of mechanical ventilation in the subgroup with a 48-h cycle. No significant differences were found in the other subgroups. Meta-regression analyses showed that there was no relationship between the cycle of CTSS exchange and VAP (p = 0.50), mortality (p = 0.33), or length of mechanical ventilation (p = 0.78).

Table 3 Summary of pooled outcomes by the cycle of changing the closed tracheal suctioning systems
Full size table

Sensitivity analyses

Of the 16 trials, 2 trials included 252 patients with pneumonia at the initiation of the studies [32, 41]. When these two trials were excluded, use of CTSS was associated with a reduced incidence of VAP (RR 0.71; 95 % CI 0.54–0.94; Q = 24.08; df = 12; I 2 = 50.2 %; p = 0.02), but it was not associated with reduced mortality (RR 1.03; 95 % CI 0.86–1.23) or shorter length of mechanical ventilation (WMD −0.47 days; 95 % CI −1.43 to 0.50). Meta-regression analysis failed to find the source of heterogeneity in the pooled outcome of VAP.

We additionally conducted sensitivity analyses on all outcomes (Supplementary Table 2). Use of CTSS tended to lower the incidence of VAP in all sensitivity analyses, but the statistical significance disappeared due to a limited number of trials and less statistical power. The results of sensitivity analyses on mortality and the length of mechanical ventilation were consistent with the secondary outcome analysis.

We reanalyzed the data using TSA. The required diversity (D 2 = 53 %, model variance-based) adjusted information size of 4331 participants was calculated, based on 26 % events in the control group, a type I error of 5 %, a power of 80 %, and an RRR of 20 %. The cumulative Z-curve did not cross the trial sequential monitoring boundary for benefit, or the required information size was not reached (Fig. 2). This implies the lack of evidence of CTSS use on the incidence of VAP.

Fig. 2
figure 2

Trial sequential analysis comparing CTSS and OTSS on the incidence of ventilator-associated pneumonia. CTSS closed tracheal suctioning systems, OTSS closed tracheal suctioning systems, RRR relative risk reduction

Full size image

Discussion

Our traditional meta-analysis showed that use of CTSS was associated with a 30 % reduction in VAP development, compared with use of OTSS. While each subgroup by the type of ICU included a small number of trials, and thus had a wide confidential interval and moderate heterogeneity, the point estimate of the effect size was similar across the subgroups. There was no significant difference in mortality or length of mechanical ventilation between CTSS and OTSS use. Use of CTSS tended to lower the incidence of VAP, but the statistical significance disappeared owing to less statistical power due to a limited number of trials included in the sensitivity analyses. Trials with high risk of bias might have overestimated the intervention effect of CTSS in the traditional meta-analysis, and thus the results should be interpreted cautiously. TSA as sensitivity analysis also revealed that the evidence is lacking to suggest that use of CTSS is associated with a lower incidence of VAP, and further research is needed.

Exogenous bacteria travel to the lower respiratory tract in mechanically ventilated patients, from the oropharynx along the external surface of the tube to the trachea, or by inadvertent cross-contamination during the disconnection of the respiratory circuit for suction, and they may result in VAP development [35]. Endotracheal tubes with subglottic secretion might block the former route [45]. Use of CTSS might reduce the chance of the latter mechanism, thereby reducing VAP development. However, prior evidence has been inconclusive with respect to this hypothesis.

Previous systematic reviews and meta-analyses of randomized trials have concluded that use of CTSS was not associated with a reduced incidence of VAP compared with use of OTSS [13, 15–20]. Some also suggested a potential publication bias. The present traditional meta-analysis showed that the incidence of VAP was significantly reduced by CTSS use. The differences in the findings between previous reviews and the present review might be attributed to two factors: (1) the number of included trials was greater, and (2) all newly included trials favored CTSS use over OTSS use. There was no evidence of publication bias in the present analysis. However, TSA indicated that the current evidence as to whether the use of CTSS is superior to the use of OTSS in preventing VAP was still lacking.

Earlier guidelines for the prevention of VAP were inconclusive about the effectiveness of CTSS as a VAP prevention measure [46, 47]. Some guidelines have favored CTSS over OTSS use for cost and safety considerations, despite the scarcity of favorable evidence supporting CTSS use for the prevention of VAP [48–51]. The present analysis showed that the number needed to prevent (NNP) with CTSS for reducing one incidence of VAP was 8.85 (95 % CI 5.62–21.27). However, scarcity of high-quality trials as well as a lack of firm beneficial evidence on CTSS rendered this issue inconclusive.

Our study triggered uncertainty regarding the optimal timing of CTSS exchange. Trials included in the present review exchanged the CTSS at 24, 48, 72, and 168 h, respectively. Meta-regression analysis suggested that there was no relationship between the cycle and the incidence of VAP. One randomized trial suggested that the incidence of VAP was not different between the 24-h and 48-h cycles of CTSS exchange, but it was underpowered [52]. Another randomized trial compared daily and no routine exchange of in-line catheters of CTSS [53]. It suggested that, while the incidence of VAP and hospital mortality were similar for the two methods, no routine exchange of in-line catheters reduced the cost. Currently, most manufacturers recommend daily CTSS exchange, but no direct evidence supports this. Sufficiently powered trials are needed to determine the optimal cycle of CTSS exchange. A cost-effectiveness analysis concerning the cycle of CTSS exchange, incidence of VAP, and related costs should subsequently be considered.

Our study has some strengths. First, a comprehensive search for trials was conducted. During the search, three Chinese trials that favored CTSS use as a prevention measure against nosocomial pneumonia were not included, because the diagnostic criteria for nosocomial pneumonia in mechanically ventilated patients were unclear. Nevertheless, four trials reported in non-English languages were included, and the total number of trials was the largest to date. Second, appropriate and relevant subgroup and meta-regression analyses were conducted due to the large number of trials. This made the analysis more rigorous. Third, the study protocol was preplanned, as recommended by the Cochrane Collaboration [22]. Finally, the introduction of TSA as well as sensitivity analyses using the traditional meta-analytic methods led to our cautious interpretation of the current evidence about CTSS, which otherwise supported the use of CTSS as a preventive measure against VAP.

The present study has limitations. First, patients’ characteristics, study protocols including prophylactic antibiotics and oral care, and the risk of VAP were not uniform across trials. The pooled analysis of VAP showed significant heterogeneity in the subgroup of mixed ICUs. However, a lack of details about these factors precluded investigating this heterogeneity. Second, many trials underreported their methodology. Of the 16 trials, 13 were published after 1996, when the CONSORT (Consolidated Standards of Reporting Trials) statement was developed to enhance researchers’ complete, clear, and transparent reporting of randomized trials [54]. More than half of these trials lacked information on sequence generation, allocation concealment, and blinding of outcome assessors. Meta-regression analysis showed that these risks of bias did not affect the pooled outcomes. However, high-quality trials are still warranted. Third, most trials included in our analysis was relatively small. It is known that most large treatment effects emerge from small-sized trials [55], while meta-regression analysis in our study barely failed to show that smaller trials tended to show favorable effectiveness of CTSS (p = 0.06). Thus, when future trials are conducted with larger sample sizes, the results may not support our study.

Conclusion

Our traditional meta-analysis suggested that CTSS use was associated with a reduced incidence of VAP compared with OTSS use. CTSS use showed no differences in terms of mortality and length of mechanical ventilation compared with OTSS use. However, sensitivity analyses including TSA suggested the scarcity of high-quality trials and the lack of firm evidence for the benefit of CTSS use compared with OTSS use in reducing VAP. This does not yet support the use of CTSS as a VAP prevention measure, which was advocated in some current guidelines. High-quality trials with a better reporting of trial results are still needed.