Ventilator-associated pneumonia (VAP) is a nosocomial infection that occurs at least 48 hours after intubation in mechanically ventilated adult patients when respiratory and gastric fluid containing micro-organisms is aspirated into the lower respiratory tract and lung field, enabled by the presence of an endotracheal tube (Figure 1). Intubation interrupts the mechanical immune defences of the oropharynx and allows highly antibiotic-resistant infection into the pulmonary system.
Higher rates of morbidity and mortality, increased ventilation time and length of hospital stay are all associated with VAP, and thus the harm to patients and the cost to society are undeniable. Although data for the UK are not currently available, the World Health Organization (WHO) (2010) has suggested that VAP accounts for a 27.5% increase in mortality in critically ill adults across the developing world; hence, it is a global problem. Further data reveal that VAP occurs in 9–27% of intubated patients and extends stays in intensive care units (ICUs) by an average of 6 days (National Institute for Health and Care Excellence (NICE), 2007), signifying that a reduction in its prevalence is imperative.
There are numerous causes of VAP in adults, and a reduction in any one of these could lead to a significant decrease in incidents. Oral colonisation is a known pathogenesis of VAP and was previously treated with intra-oral antibiotics. However, an increase in antibiotic resistance has resulted in its use being discouraged (Centers for Disease Control and Prevention (CDC), 2003) and consideration of alternative medications.
Chlorhexidine is a worldwide healthcare essential broad-spectrum antiseptic (WHO, 2017), which acts by disrupting the cell membrane of both Gram-negative and Gram-positive organisms, causing metabolic change within the cell that effectively destroys it. It is commonly used as a mouthwash that reduces dental bacteria and particularly as a means of managing oral colonisations (Ellepola and Samaranayake, 2000; Ellepola et al, 2013). Saliva is important for oral health because of its lubricating, antimicrobial and buffering properties. However, these will be inhibited in ventilated patients (Labeau, 2011).
While there is no specific recommendation for the use of chlorhexidine in this population (Tablan et al, 2004), bacterial colonisation of the oral cavity occurs more rapidly in ventilated patients (Sona et al, 2009; Messika et al, 2018) and may explain why personal experience has seen it used commonly in ICUs. Nevertheless, there appears to be no standardised practice, with differences in solution strength and type, frequency of application and, in some cases, the use of alternative medications or therapies observed. Known potential adverse effects of chlorhexidine (Price et al, 2014) have been reported (Table 1) and may have led to this inconsistency. As such, alternative or combined interventions have been widely researched.
| Adverse effects reported | Source |
|---|---|
|
|
National Institute for Health and Care Excellence, 2019 |
|
|
Naiktari et al, 2014 |
|
|
Pemberton and Gibson, 2012 |
Combined therapies and the cardiac surgery patient
The combination of chlorhexidine and tooth brushing have been investigated; clinical trials have shown that a reduction in dental plaque, which acts as host to intra-oral colonising bacteria, also reduces VAP (Fourrier et al, 2000; Lansford et al, 2007; Munro et al, 2009). Furthermore, it has been suggested that dental surgeon intervention, paired with chlorhexidine use, can reduce the occurrence of VAP (Bellissimo-Rodrigues et al, 2014). In addition, Shitrit et al (2015) and Sona et al (2009) found that care bundles which include using different approaches to intra-oral care, including chlorhexidine decolonisation, are an effective way to reduce VAP—this may explain why they are widely used. In Shitrit et al's (2015) study, this involved raising the head of the bed, hand hygiene, the use of chlorhexidine, maintaining cannula balloon pressure and measuring nasogastric (NG) tube food remnants prior to every NG tube feed. In Sona et al's (2009) study, it involved combining toothpaste, brushing and the use of chlorhexidine.
Other studies, however, have recommended alternative antiseptics to reduce the rate of VAP (D'Amico et al, 1998; Kolahi and Soolari, 2006; Klarin et al, 2008). One study used hydrogen peroxide as part of a care bundle instead of chlorhexidine, and found a reduction in VAP (Lev et al, 2015). In consideration of these findings, the effect of chlorhexidine as a single component is unclear, suggesting the need to review the literature surrounding the use of chlorhexidine alone.
This article intends to complement other reviews on the subject, such as the one undertaken by Klompas et al (2014), who performed a systematic review and meta-analysis to explore the effects of chlorhexidine on the prevalence of VAP. The study conducted a comprehensive literature search, inclusive of patient populations across several specialty ICUs, with only randomised controlled trials (RCTs) eligible for inclusion. The results indicated that in cardiac-surgery patients VAP was reduced with the use of chlorhexidine only, with the recommendation made to stop using it in other patient groups. An in-depth exploration of the study, however, showed it to have numerous weaknesses, raising questions about its reliability. There was little blinding, substandard reporting of quality, and more than half the subjects were cardiac-surgery patients. The results were consequently more significant in these patients than any other group.
The studies described in the current review were selected following comprehensive database searches of CINAHL and Medline and studies with the following variables were excluded to allow the effect of chlorhexidine to be reviewed as a single component:
Four RCTs emerged, all of which were absent from the review by Klompas et al (2014). This could provide additional insight into the effectiveness of chlorhexidine exclusive of the cardiac-surgery patient group.
Study 1
Koeman et al (2006) conducted an RCT with a pre-estimated population of adults across five general and surgical ICUs. The study had three arms: placebo (n=130), chlorhexidine (n=127) and chlorhexidine colistine (n= 128). Nurses were trained in administration technique to ensure continuity of application and both randomisation and double blinding were implemented. Patients who had immunosuppression, impairment preventing physical application, or who were pregnant were excluded. VAP was measured as the primary outcome using clinical assessment of four criteria following the presence of infiltrate on chest X-ray, which included positive culture of potential respiratory pathogens, visual examination of tracheal aspirate, leukocytosis and temperature. The results were as follows:
Study 2
Özçaka et al (2012) carried out an RCT studying 66 adults on a respiratory ICU, with 32 in the chlorhexidine group and 34 controls. Administration consisted of chlorhexidine or saline and there was no staff or patient blinding. Randomisation did take place, however, and nurses were trained in administration technique. Exclusion criteria included witnessed aspiration and lung cancer, thrombocytopenia, pregnancy, and patients with a history of adverse reactions to chlorhexidine. VAP was measured with a diagnosis made using culture results only. The rate at which VAP occurred was found to be very high, perhaps reflective of this limited diagnostic measurement and selection of a high-risk respiratory patient group. Statistics showed a VAP occurrence as 41% in the chlorhexidine group and 69% in the saline group.
Study 3
Scannapieco et al (2009) conducted an RCT involving 175 patients in a multi-specialty ICU. The study took a three-arm approach: 59 patients were allocated to placebo twice a day; 58 to chlorhexidine once a day and placebo once a day; and 58 to chlorhexidine twice a day. Nurses were trained in administration technique and observed periodically to ensure continuity of application was maintained.
Patients were excluded if they had lung cancer, there was evidenced aspiration, thrombocytopenia, immunosuppression, pregnancy, patients sensitive to chlorhexidine, and individuals legally imprisoned. Clinical measurement of VAP was based on the widely used Clinical Pulmonary Infection Score (CPIS) tool (Pugin et al, 1991) (Table 2) and the presence of potential respiratory pathogens in pleural fluid; clinician opinion was not sought. A total of 29 subjects were excluded, 19 of whom were extubated or had died prior to sampling, and data collection was incomplete for a further 10, leaving a total of 146. The results showed that prevalence of VAP was 14% in the chlorhexidine group, 15% in the chlorhexidine/placebo group, and 24% in the placebo group.
| Score | 0 | 1 | 2 |
|---|---|---|---|
| Temperature | 36.5–38.4 | 38.5–39.0 | <36.0 or >39.0 |
| Leukocytosis | 4000–11 000 | 11 000–17 000 | >17 000 |
| New chest X-ray infiltrate | None | Patchy | Localised |
| Endotracheal secretions | None–minimal | Moderate | Large amount |
| Oxygenation PaO2/FiO2 mmHg | >330 | – | <330 |
Study 4
Finally, Tantipong et al (2008) undertook an RCT with 207 patients across a general, a medical and a surgical ICU. Randomisation was based on sex. The nurses were not blinded and there was neither the use of a placebo, nor any education regarding technique; however, the investigators were blinded. Patients with pneumonia or chlorhexidine sensitivity were excluded.
Clinicians diagnosed VAP by, first, identifying infiltrate on chest X-ray and then establishing the presence of positive tracheal cultures for potential respiratory pathogens, pyrexia, leukocytosis and purulent tracheal aspirate. No scoring system was used. The study found that VAP occurred in 5% of patients in the chlorhexidine group compared with 12% of the controls.
Participation and physiology
All four studies differed in terms of ICU setting. Consequently, each patient group had varying baseline characteristics that were compared using the Acute Physiology and Chronic Health Evaluation II (APACHE II) disease severity scoring tool (Knaus et al, 1985) (Table 3). APACHE II records patient age, 12 physiological parameters and consciousness level, using the Glasgow Coma Scale. Participants in three of the studies (1, 3 and 4), which each included a mixture of surgical, medical and general ICUs, had comparable baseline characteristics enabling the chlorhexidine to be the independent variable. It could therefore be argued that this increased the chance of accurate results.
| Physiologic variable | High abnormal range | Low abnormal range | |||||||
|---|---|---|---|---|---|---|---|---|---|
| +4 | +3 | +2 | +1 | 0 | +1 | +2 | +3 | +4 | |
| Rectal temperature (°C) | ≥41 | 39–40.9 | 38.5–38.9 | 36–38.4 | 34–35.9 | 32-33.9 | 30-31.9 | ≤29.9 | |
| Mean arterial pressure (mmHg) | ≥160 | 130–159 | 110–129 | 70–109 | 50-69 | ≤49 | |||
| Heart rate (beats per minute) | ≥180 | 140–179 | 110–139 | 70–109 | 50-69 | 40-54 | ≤39 | ||
| Respiratory rate (breaths per minute) | ≥50 | 35–49 | 25-34 | 12–24 | 10–11 | 6-9 | ≤5 | ||
| Oxygenation |
≥500 | 350-499 | 200–349 | <200 |
PO2 61–70 | PO2 56–60 | PO2<55 | ||
| Arterial pH | ≥7.7 | 7.6–7.69 | 7.5–7.59 | 7.33–7.49 | 7.25–7.32 | 7.15–7.24 | <7.15 | ||
| HCO3 (mEq/litre) | ≥52 | 41–51.9 | 32–40.9 | 22–31.9 | 18–21.9 | 15–17.9 | <15 | ||
| Potassium (K) (mEq/litre) | ≥7 | 6–6.9 | 5.5–5.9 | 3.5–5.4 | 3-3.4 | 2.5–2.9 | <2.5 | ||
| Sodium (Na) (mEq/litre) | ≥180 | 160–179 | 155–159 | 150–154 | 130–149 | 120–129 | 111–119 | ≤110 | |
| Serum creatinine (mqm/dl) | ≥3.5 | 2-3.4 | 1.5–1.9 | 0.6–1.4 | <0.6 | ||||
| Hematocrit (%) | ≥60 | 50–59.9 | 46–49.9 | 30–45.9 | 20–29.9 | <20 | |||
| Thin-layer chromatography (103/cc) | ≥40 | 20–39.9 | 15–19.9 | 3–14.9 | 1–2.9 | <1 | |||
In contrast, however, study 2 claimed similarity in baseline characteristics, but failed to account for significant differences in age and comorbidities, both of which were higher in the control group—it could be argued that these increase the risk of VAP. Comorbidities can be a great risk factor for VAP, but the extent of this varies greatly, depending upon the condition. Chronic respiratory diseases pose a much higher risk of VAP than a great number of other comorbidities, but the data from this study failed to identify these patients and therefore the influence that such diseases may have had on the results is unclear. However, the prevalence of respiratory conditions on a respiratory ICU can be assumed to be great, which would mean that the control group had a higher number of such patients. The authors did, however, identify the limitations due to the small population size, substantially the smallest in this review (n=66), with high-risk subjects, having conducted the trial on a respiratory ICU. Increased risk of VAP in the control group and poor population size may explain why the difference in VAP between the two groups was significantly larger than in any of the other studies.
In addition, researchers in the second smallest study (study 3), on recognising that their population size was too small, identified the reduction of VAP as a secondary outcome. This study had the second largest difference in VAP between the groups after study 2, a result the researchers found to be statistically insignificant. This suggests that both small population size and contrasting baseline characteristics may have led to overestimates of the effect of chlorhexidine on VAP occurrence. As a result, the researchers recommended that a study with a larger patient population be undertaken.
Environmental influences
Three of the four studies (1, 2 and 3) adopted the inclusion criteria to investigate only patients expected to require mechanical ventilation for longer than 48 hours. Clinicians were responsible for making a subjective assessment and predicting which candidates fulfilled the criteria, prompting questions about reliability. Due to the nature of critical illness, it can be assumed that some patients who were predicted to be likely candidates for extubation within 48 hours were in fact not extubated. Enrolment, therefore, relied on physicians assessing patients within time constraints, with the consequence that not all eligible patients were included, which throws doubt on the generalisability of the studies.
In contrast, study 4 adopted a different tactic: subjects were not assessed for estimated length of intubation, enrolling all patients admitted to the participating ICUs instead. This ensured that all eligible candidates were included. However, what this study failed to account for was the rate of extubation within 48 hours. As a result, almost half the subjects' data was irrelevant and excluded from the final analysis, leaving insufficient numbers and questionable results. If the researchers had anticipated such an attrition rate, they could have considered enrolling a much larger population to ensure they had an adequate amount of data, which would have enabled them to draw reliable conclusions.
There were further contrasts, with studies 1 and 3 using placebo to disguise the solutions in order to blind both staff and patients and preventing behavioural bias. In contrast, studies 2 and 4 used saline as the control, failing to eliminate bias, which may have influenced the results and made them unreliable. Despite this weakness, study 2 did employ double blinding; however, study 4 blinded only the investigators, which also left room for inaccurate results. However, there are no comparative trends in the data to suggest that this variation in technique had any significant effect on the results.
To further prevent bias, three studies (1, 2, and 3) used computer software or subject identification numbers to randomise patients to each of its study groups. However, study 4 randomised subjects according to sex without providing a rationale for this. There is no link between VAP and chlorhexidine and sex, and therefore should not have been used as a criterion for inclusion/exclusion.
Furthermore, 10% of participants in study 4 experienced adverse reactions compared with none in the other three studies. It has been previously reported that chlorhexidine is incompatible with certain toothpastes (NICE, 2019) and can cause adverse reactions. However, it is difficult to determine the role that toothpaste played in study 4 because there is no mention of this by the authors. As illustrated in Table 4, the solution strength, administration time and amount of solution used in each study varied widely, which could be another possible explanation for adverse reactions in study 4.
| Strength of chlorhexidine | Frequency of application | Method of application | Control group solution | Subglottic aspiration | Education and training given | Adverse effects observed | |
|---|---|---|---|---|---|---|---|
| Study 1 (Koeman et al, 2006) | 2% | 6 hourly | 2 cm applied to buccal cavity | Placebo | Not performed | Yes | No |
| Study 2 (Özçaka et al, 2012) | 0.2% | Four times a day | 30 ml applied to teeth and intraoral soft tissues for 1 minute | Saline | 6 hourly and after position changes | Yes | No |
| Study 3 (Scannapieco et al, 2009) | 0.12% | Once or twice a day to determine minimum frequency required | 1 oz applied to teeth and all intra-oral soft tissues | Placebo | 12 hourly and after position changes | Yes | No |
| Study 4 (Tantipong et al, 2008) | 2% | Four times a day | 15 ml applied to oropharyngeal mucosa | Saline | Not performed | No | Yes |
Study 4 used a strong solution type of 2%; however, study 1 also used chlorhexidine at 2% strength and reported no adverse effects. In contrast, it is clear that researchers in studies 1, 2 and 3 standardised their chlorhexidine administration technique to ensure there was little difference between patient treatments. In these studies, the researchers also periodically observed the administration technique to maintain high standards of practice. Once adverse reactions had occurred in study 4, the researchers recognised the consequences that a lack of teaching may have had and provided staff with education and training. It is of note that no further adverse reactions were observed and oral conditions improved. This therefore indicates that chlorhexidine had been incorrectly administered up until this point, possibly for a large proportion of subjects and thus it is contentious to suggest that there was a direct relationship between VAP and the use or non-use of chlorhexidine in the patients in study 4. This study had the lowest reduction rate between the intervention groups, so it could be suggested that this was due to incorrect administration of solutions.
The method and frequency of chlorhexidine application also differed greatly between studies (Table 4). Studies 1, 2 and 4 applied chlorhexidine four times per day or 6 hourly, whereas study 3 had two chlorhexidine arms: chlorhexidine once daily and twice daily and, while the results showed a higher rate of VAP in the placebo group, there was no significant difference between both chlorhexidine groups, which could be explained by the small population size of the study.
Studies 1 and 4 implemented semirecumbent positioning for all participants, an intervention that has been shown to reduce the rate of VAP by preventing the transmission of micro-organisms via gastric fluids entering the pulmonary system (Torres et al, 1992; Drakulovic et al, 1999; Shitrit et al, 2015). Positioning as a factor is not mentioned in studies 2 and 3. Similarly, subglottic suctioning reduces VAP in the same way (Dodek et al, 2004) and, while this was performed in studies 2 and 3, it was omitted in studies 1 and 4, without justification as to why. Each study applied chlorhexidine to different areas of the oral cavity, including the buccal mucosa; chlorhexidine, however, was applied directly to the teeth in only two studies (studies 2 and 3)—this is an area that Heo et al (2008) found acted as a reservoir for potential respiratory pathogens. The wide range of interventional methods used across the studies makes it unclear how these variables affected the occurrence of VAP, leading to the conclusion that further research into application methods may be beneficial.
VAP pathogenesis and diagnosis
There were further differences in the exclusion criteria of the four papers. Studies 2 and 3 had similar strict criteria, which excluded those at risk of harm, patients with pre-existing conditions that could influence the results, giving incorrect diagnosis of VAP, and those whose clinical history would put them at a greater risk of VAP. This allowed chlorhexidine to be the independent variable so its effect on VAP could be clearly observed. Similarly, study 4 excluded patients in whom chlorhexidine posed a risk or those with a clinical history that increased the risk of VAP. It failed however to identify patients whose medical conditions would affect blood results, thus giving a false diagnosis of VAP. This may have led to incorrect diagnosis. Study 1 chose to include patients at higher risk of VAP, including those who had lung cancer and witnessed aspiration which, it could be argued, is a common event for a large number of ICU admissions due to the nature of injury, particularly in those with brain injuries who vomit due to raised intracranial pressure and traumatic injury patients. Exclusion of these patients may have led the studies to conclude that chlorhexidine may not prevent VAP in these high-risk patients, but there is no evidence to suggest this. Furthermore, not all these patients develop pneumonia and therefore it can be argued that they should be given the same treatment as everyone else. This suggests that only the results from study 1 are generalisable to the complete ICU population.
Finally, the diagnostic criteria used to measure VAP differed greatly between the studies (Table 5). Study 1 had a strong technique, which measured 5 criteria and presented results blindly to three of the hospital's treating intensivists to review and, only when they concurred, was a diagnosis made. In current practice, consideration of such criteria and discussion between clinicians within the ICU enables a diagnosis to be reached (Allman and Wilson, 2016), which supports the plausibility of these results. Study 4 used the same criteria but failed to recognise the importance of expertise and the investigators, whose ICU experience is unknown, determined diagnosis of VAP instead; there is no reference to the number of investigators involved in this decision-making process either, making it difficult to determine how much scrutiny there was in ensuring accuracy of results.
| Treating Intensivist expertise | Clinical pulmonary infection score tool | Infiltration on chest X-ray | Presence of potential respiratory pathogenss | Leukocytosis | Pyrexia | Tracheal aspirate examination | |
|---|---|---|---|---|---|---|---|
| Study 1 (Koeman et al, 2006) | X | X | X | X | X | X | X |
| Study 2 (Özçaka et al, 2012) | X | ||||||
| Study 3 (Scannapieco et al, 2009) | X | X | X | X | X | X | |
| Study 4 (Tantipong et al, 2008) | X | X | X | X | X |
In study 3, diagnosis depended on achieving a score of 6 or greater using the CPIS (a tool that is commonly used by intensive care professionals to aid diagnosis of VAP in practice and is well accredited) and the detection of potential respiratory pathogens in aspirate. The CPIS calculates the severity of 5 criteria, which are the same as those used by studies 1 and 4 to diagnose VAP. In addition, study 1 used the CPIS tool on a daily basis to guide decision making. In contrast, study 2 instituted a poor technique and VAP was diagnosed using the presence of potential respiratory pathogens in cultures only, a criterion considered as only one of five markers by the other papers. In practice, this indicator is not used in isolation to diagnose VAP and, in fact, microbiologists often recommend withholding antibiotics with a positive culture until there is clinical indication (Rojo, 2006). This invalid measurement of VAP calls into question all of the results from study 2 and may explain why antibiotic usage and leukocytes, both an indication that these subjects had infection, were higher in the VAP negative chlorhexidine group than those believed to have VAP.
Currently, a gold standard for diagnosing VAP has not been established (Lisboa and Rello, 2008; Grgurich et al, 2013; Kalanuria et al, 2014), which may explain why the diagnostic criteria selected across the four studies varied so greatly and why there was no discussion to justify their selection.
Discussion
Following an in-depth critique of current research papers addressing the effect of chlorhexidine on the occurrence of VAP, the evident weaknesses in studies 2 and 4 of study design, methods and measurement leads to the conclusion that their results are not reliable or plausible enough to influence practice. The failure of study 4's researchers to recognise its limitations led them to overestimate the significance of their results and a recommendation for practice within the studied hospital was made. In contrast, both studies 1 and 3 had many strengths including population, design and methods, indicating that the results are valid and reliable. It can be determined from this that intra-oral application of chlorhexidine reduces VAP across the critically unwell patient group inclusive of all specialties. This challenges the literature, which indicates that a reduction in VAP occurs within the cardiac surgery patient group only (Chlebicki and Safdar, 2007; Klompas et al, 2014) and therefore its use should be implemented in current practice across all sectors of the ICU population.
Resistance to chlorhexidine is very low (WHO, 2017), which strengthens the advantage of its use in practice, but it is widely known that pathogens have the ability to adapt and build resistance (Gould and Brooker, 2008). This may make it necessary to study the benefit of chlorhexidine in comparison with other antiseptics in the future, which up until this point has mainly shown chlorhexidine to be the most effective treatment (Labeau et al, 2011); however, there is limited literature to prove the comparison definitively.
All four studies measured the patients' days on mechanical ventilation, length of ICU stay and rate of mortality as secondary outcomes, but found that none of them were affected by chlorhexidine application; this is similar to findings found in a recent review (Messika et al, 2018). However, this review considered both adults and children and, considering the known relationship between VAP and each of these components, this is perhaps surprising, and a clinical trial with a much larger population examining each of these components separately in adults alone may be necessary to understand the greater implications of chlorhexidine use.
Conclusion
There is evidence taken from quality research to show that chlorhexidine reduces VAP in critically ill, mechanically ventilated adults. A recommendation for implementing the use of chlorhexidine for all mechanically ventilated patients either twice daily or four times a day can be made. However, further research into the effect of administration frequency and method would be valuable to prevent overexposure, which may exacerbate pathogenic resistance.