The ventilator care bundle and its impact on ventilator-associated pneumonia: a review of the evidence

LITERATURE REVIEW doi: 10.1111/j.1478-5153.2010.00430.x The ventilator care bundle and its impact on ventilator-associated pneumonia: a review of the evidence Petra Lawrence and Paul Fulbrook ABSTRACT Aims and objectives: The aim of this review was to critically analyse recent research that has investigated ventilator care bundle (VCB) use, with the objective of analysing its impact on ventilator-associated pneumonia (VAP) outcomes. Background: The VCB is a group of four evidence-based procedures, which when clustered together and implemented as an all or nothing strategy, may result in substantial clinical outcome improvement. VAP is a nosocomial lung infection associated with endotracheal tube use in ventilated patients. Since the VCB was introduced there have been several studies that have reported significant VAP rate reductions. Search strategy: A comprehensive search for research, published between 2004 and 2009, was conducted using Medline and PubMed. Key words were used to identify English language studies reporting VCB implementation within adult intensive care units (ICU) and associated clinical outcomes. Studies that implemented bundle variations that did not include all four elements were excluded. Conclusions: Because of the limitations of the observational designs used in the studies retrieved, a definitive causal relationship between VCB use and VAP reduction cannot be stated. However, the evidence to date is strongly indicative of a positive association. Several studies reported the use of additional VCB elements. In these cases it is difficult to establish which elements are related to the measured outcomes. Further research is recommended to establish baseline outcome measures using the four-element VCB, before adding further processes singly, as well as research investigating the effect of audit and feedback on VCB compliance and its effect on clinical outcomes. Relevance to clinical practice: A reduction in VAP is associated with VCB use. The evidence to date, whilst not at the highest experimental level, is at the highest ethically permissible level. In the absence of contradictory research, the current evidence suggests that use of the VCB represents best practice for all eligible adult ventilated patients in ICU. Key words: Microaspiration Nosocomial lung infection Ventilator-associated pneumonia Ventilator care bundle INTRODUCTION It is important for those working in critical care environments to examine their practices to ensure they are evidence based. The purpose of evidence-based practice is to apply existing research evidence (to practice). Evidence to inform practice comes in a variety of forms and there are many ways that evidence can be applied in a practice setting, for example, through the development of evidence-based protocols or care pathways (Fulbrook, 2003). One method that has been gaining Authors: P Lawrence, RN, BN, Research Assistant, Nursing Research & Practice Development Centre, The Prince Charles Hospital, Brisbane, Australia;PFulbrook,RN,PhD,MSc,PGDipEduc,BSc(Hons),Professorof Nursing, National Centre for Clinical Outcomes Research, Australian Catholic University, Brisbane, Australia; Nursing Director Research & Practice Development, Nursing Research & Practice Development Centre, The Prince Charles Hospital, Brisbane, Australia. Address for correspondence: P Fulbrook, The Prince Charles Hospital, Rode Road, Chermside, Brisbane 4032, Queensland, Australia E-mail: paul_fulbrook@health.qld.gov.au ground over the last decade is the care bundle. It was originally described in the critical care nursing literature by Fulbrook and Mooney (2003) as a group of several evidence-based practices: The idea is that several practices, when used in combination, or as acluster,all of the time, have a greater effect on the positive outcome of patients (p. 250). This holistic principle is familiar to nurses: the whole is greater than the sum of its parts. Care bundles provide a method for establishing best clinical practice, which in theory will improve clinical effectiveness of interventions (Fulbrook and Mooney, 2003). A variety of care bundles has been reported in the literature including those covering sepsis management, cardiovascular risk reduction, antibiotic use and management of central venous catheters (Cooke and Holmes, 2007; Fong et al., 2007; Poe et al., 2007; Smith, 2007). However, the most well known, which has been implemented widely, is the ventilator care bundle (VCB). The primary purpose of this paper is to present 222 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses Vol 16 No 5

a review of the evidence associated with use of the VCB and its effect on VAP rates. BACKGROUND Care bundles A bundle usually consists of three to five interventions which are grouped together into a single quality measure (Aboelela et al., 2007). The process measures relate to a common disease and together provide a more robust picture of quality than any single measure (Berenholtz et al., 2004). Berenholtz et al. (2002) conceptualized the idea of bundling evidence-based interventions, with each element of the bundle being well defined and based on strong evidence. Their systematic review of the literature between 1965 and 2000 identified all studies that provided a potential measure of quality of intensive care. Six outcome measures were identified: intensive care unit (ICU) mortality rate; ICU length of stay (LOS) greater than seven days; average ICU LOS; average days on mechanical ventilation; optimal pain management; and patient/family satisfaction. In general, processes are easier to measure than outcomes and can be used to provide immediate feedback to providers regarding their performance. Six process measures were also identified, and these were: effective assessment of pain; appropriate use of blood transfusions; prevention of VAP; assessment of sedation; appropriate gastric ulcer prevention; and deep vein thrombosis prophylaxis (DVTP). The latter four measures were grouped together by the Joint Commission on Accreditation of Healthcare Organisations (JCAHO) to form the VCB (see IHI, 2006): head of bed elevation; daily interruption of sedation and assess readiness to wean; gastric ulcer prevention; DVTP. The VCB clustered together four evidence-based interventions that were based on evidence that was sufficiently strong to suggest that they should be applied to all ventilated intensive care patients all the time. To test these process measures, the VCB was implemented in a 14-bed ICU for one year (Berenholtz et al., 2004). The results showed that increased compliance with all four elements of the VCB resulted in reduced mortality, ICU LOS, hospital days, and costs. Around this time, the Institute for Healthcare Improvement (IHI) launched the Saving 100 000 lives campaign and conducted a multi-centre ICU study in the USA and Canada from 2002 to 2004, to investigate the relationship between compliance with the VCB and its effects on clinical outcomes (O Keefe et al., 2008). This landmark study was significant because it demonstrated that the VCB not only reduced ventilator days and ICU LOS, but also dramatically reduced VAP rates. Since the VCB was first introduced it has been adopted in many ICUs throughout the world. Berenholtz et al. (2002) emphasized the importance of linking process with outcome measures, and several research studies have been conducted to evaluate the impact of the VCB on related measurable outcomes. To date, the research evidence suggests that the most significant impact of the VCB is its effect on reducing ventilator-associated pneumonia (VAP). VAP VAP is a nosocomial infection of the lung parenchyma that is associated with the use of endotracheal tubes in ventilated patients; however, the association is with the use of artificial airways as opposed to mechanical ventilation itself (Valencia and Torres, 2009). Patients are predisposed to VAP because the endotracheal tube mechanically breaches the body s normal defence mechanisms which resist the entry of pathogens into the lungs. Colonization of the oropharynx and the stomach with potentially pathogenic organisms precedes the development of VAP (Collard et al., 2003) and the artificial airway acts as a pathway for bacteria to enter the respiratory tract. Secretions which may harbour pathogenic bacteria accumulate below and above the endotracheal tube cuff, and may become dislodged and enter further into the lower lungs during ventilation flow, tube manipulation or suctioning (Tablan et al., 2003). The mucocilliary transport system is hindered because of the suppression of cough reflex by the use of sedatives and muscle relaxants, and the normal closure of the epiglottis is impaired, resulting in an incomplete seal of the laryngeal structures that protect airways (Berenholtz et al., 2004; Pruitt and Jacobs, 2006; O Keefe et al., 2008). In summary, the most important mechanism of VAP is the colonisation of the aero digestive tract with pathogenic bacteria and the subsequent aspiration of the contaminated secretions into the lower airways (Collard et al., 2003; Tablan et al., 2003; Shaw, 2005; Pruitt and Jacobs, 2006). Additional risk factors that predispose patients to VAP include pre-existing comorbidities, such as malnutrition, chronic obstructive pulmonary disease, poor oral hygiene, and chronic sinusitis (Shaw, 2005; O Keefe et al., 2008). A comprehensive review of 13 VAP studies revealed an ICU incidence of between 5% and 67% (mean 23 6%) with crude reported mortality of between 24% and 76% (mean 46 5%) (Chastre and Fagon, 2002). An earlier 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses 223

Canadian study found that VAP-attributable mortality of around 30% was variable depending on the population and infecting organism (Heyland et al., 1999). VAP may increase ICU patients LOS by 4 3 6 1days and hospitalization may be extended by 4 9 days; it is estimated that this represents an added cost of $40 000 (US) per patient per hospital admission (Tablan et al., 2003; American Thoracic Society, 2005). VAP can be difficult to diagnose, and there is still no consensus about the best microbiological diagnostic method (Bouza and Burillo, 2009). In the USA, the Centers for Disease Control (CDC, 2009) advises the use of a combination of radiological, clinical and laboratory criteria, which includes observation of the development of new or progressive pulmonary infiltrates, with fever, leukocytosis and purulent tracheobronchial secretions. Clinical criteria together with cultures of sputum or tracheal specimens may be sensitive for bacterial pathogens but they are highly non-specific, which makes diagnosis difficult (Lisboa and Rello, 2008). Diagnosis can also be problematic because of the difficulty of separating the infection from other lower respiratory infections such as tracheobronchitis (American Thoracic Society, 2005). Bronchoscopic techniques are therefore recommended to retrieve lung samples for histology review; however, this is invasive and can cause complications such as hypoxaemia, bleeding or arrhythmia (Tablan et al., 2003). Time of onset is an important variable for specific pathogens and outcomes (Shaw, 2005). Pneumonia is considered to be ventilator associated if the patient was intubated at the time or within 48 h of the onset of infection; there is no minimum period that the ventilator must be in place in order for the pneumonia to be considered ventilator associated (IHI, n.d.). The pathogen associated is usually a community-acquired pathogen such as Streptococcus pneumoniae, Haemophilus influenzae and Morazella catarrhalis, which has a better prognosis as these bacteria are usually antibiotic sensitive (Pruitt and Jacobs, 2006). Late-onset VAP is associated with increased morbidity and mortality and generally occurs 72 h or more after intubation. The development of VAP at this time is usually caused by antibiotic-resistant organisms such as Staphylococcus aureus, Acinetobacter baumannii, Pseudomonas aeruginosa, Klebsiella pneumonia and the enterobacter species (Pruitt and Jacobs, 2006; Westwell, 2008). These antibiotic-resistant pathogens also vary between hospitals and within hospitals (Tablan et al., 2003). METHOD Search strategy A comprehensive search for primary research articles was conducted using the Medline and PubMed databases, using the keywords ventilatorassociated pneumonia, VAP, ventilator bundle and care bundle, and combining searches using VAP with guideline, implementation, adherence, compliance, feedback and audit. The reference lists of retrieved articles were also examined as a potential source. Data were extracted by sample and setting, study design, type of intervention, methodological quality and reported outcomes. Inclusion criteria Primary studies were included if they met the following criteria: English language experimental research published between 2004 and 2009; studies which implemented the VCB (including those that used additional elements) in an ICU setting; clinical outcome measures were reported. Exclusion criteria Studies were excluded if clinical outcomes were not measured. For example, studies which examined the effectiveness of an education intervention targeted at health care personnel but did not investigate clinical outcomes associated with VCB use. Other exclusions included research conducted in paediatric settings, and other variations of the VCB such as those with less than four of the required elements. RESULTS The search strategy yielded 10 research studies, which are summarized in Table 1, and two literature reviews, which were excluded from this review. O Keefe et al. (2008) evaluated six studies but also included VCBs which did not incorporate all four elements of the VCB, and Zilberberg et al. (2009) evaluated four studies with all four elements of the VCB. Bundle elements The four elements of the VCB are based upon the highest level of evidence, i.e. systematic review of randomized controlled trials (RCT) and single RCTs (Berenholtz et al., 2002). Several of the studies reviewed employed additional elements (interventions) that were bundled with the original VCB (see Table 1). The VCB used by Papadimos et al. (2008) was part of 224 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses

Table 1 Summary of studies reviewed Authors (Country) Interventions Additional bundled elements Feedback Measures Outcomes Khorfan (2008) (USA) Daily checklist (every shift) and review Staff education Papadimos et al. (2008) (USA) DuBose et al. (2008) (USA) Youngquist et al. (2007) (USA) FASTHUG mnemonic (intensivist-led) Daily patient evaluation (a.m. and p.m.) Daily checklist Daily review of non-compliance with immediate correction Staff education Monthly review of deficiencies Visual reminders Involved in the IHI collaborative Daily checklist and review All or nothing approach to VCB Constant presence of nurse specialist for informal assessments Written communication Fact sheets, in services Visual reminders Oral care Out of bed Feeding evaluation Blood glucose monitoring Pain assessment Sharing results with staff VAP/1000 days From 1 7 to 0 27/1000 days (85% reduction) Morbidity/mortality Not reported but likely to result in decreasing LOS, morbidity and mortality and improving patient safety Comparisons to NNIS rates: 2004 Medical ICU: 0 27/1000 days NNIS 4 9/1000 days NHSN: not available NNIS rates: 2004 None described VAP/1000 days From 19 3 to 7 3/1000 days (p = 0 0004) (62% reduction) NNIS rates 2004 Patient demographics No difference Surgical ICU days No difference LOS No difference (although trend to significance p = 0 07) Severity of illness index Higher severity of illness index (p = 0 001) Medical diagnostic categories No difference None None described VAP/1000 days From 16 3 to 8 9/1000 days (45% reduction) Ventilation duration >24 h: pre 90 1%; 1 month 86 4%; 3 month 86 7% >48 h: pre 81 7%,;1 month 72 2%; 3 month 72 2% >72 h: pre 74 0%; 1 month 60 9%; 3 month 61 7% Self-extubation rate From 7 8 to 2 2/1000 days Compliance (process) Trend improvement HOB: 35 2 84 5% (p < 0 05) SEDHOL: 78 86% (p < 0 05) GUP: 76 2 92 3% (p < 0 05) DVTP: 91 2 92 8% (p = ns) None Monthly compliance audit and feedback VAP/1000 days Mercy: from 6 01 to 2 2/1000 days (63% reduction) Unity: from 2 66 to 0/1000 days (100% reduction) NNIS rates: 2004 ICU LOS Mercy: p = ns Unity: p = 0 02 Compliance (process) 100% 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses 225

Table 1 (Continued) Authors (Country) Interventions Additional bundled elements Feedback Measures Outcomes Unahalekhaka et al. (2007) (Thailand) Berriel-Cass et al. (2006) (USA) Hatler et al. (2006) (USA) Resar et al. (2005) (Canada and USA) Collaborative of 18 Thai hospitals Hospital directors signed approval Forms for data collection Staff education including national and regional workshops and face-to-face meetings Self-administered questionnaire Involved in the IHI collaborative Daily checklist and review Daily audit of DVTP and gastric ulcer prophylaxis with immediate follow-up Staff education Train the trainer approach HOTSPUD mnemonic Patient goals Staff education: newsletters, reports Visual reminders Cognitive theory of change Recognition activities Identification of barriers Collaborative for IHI Daily goals Staff education All or nothing approach to VCB ICU teams posted data monthly on a web-based extranet and submitted narrative descriptions describing the changes tested and strategies implemented Monthly review of data None None described VAP/1000 days From 13 3 to 8 8/1000 days (34% reduction) Oral care Hand-washing Monthly feedback VAP/1000 days From 8 2 to 3 3/1000 days (60% reduction) (p = 0 02) NNIS rates 2004 Average ventilator days Decreased ICU LOS Average: 8 0 4 9 days Oral care Poster of weekly results of compliance and VAP rates, improvements and transitions, and with efforts for next month VAP/1000 days 11 4 5 3/1000 days (54% reduction) ICU LOS Average: 4 4 3 59 days Compliance (process) 73 98 6% None VAP/1000 days 35 units overall: (VAP 45% reduction): 6 6 2 7/ 1000 days (59% reduction) (p < 0 01) Compliance (process) 21 units with 95% compliance ICU mortality Not reported ICU LOS Not reported 226 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses

Table 1 (Continued) Authors (Country) Interventions Additional bundled elements Feedback Measures Outcomes Hampton et al. (2005) (USA) Plan-do-check-act quality improvement methodology Nurse and physician champions Staff education: In services Visual reminders (posters) Compliance monitoring three times/week Implementation of one intervention at a time None Bulletin board with evidence, aims, goals and desired outcomes, and weekly quality data. Weekly graph compliance monitoring Monthly report card Ventilator days Average: 6 1 3 5 days (43% reduction) Number of re-intubations No increase ICU LOS Average: 4 0 3 59 days (18% reduction) Mortality Decreased more than 50% Compliance Near 100% Crunden et al. (2005) (UK) Daily checklist Two audits to assess compliance (baseline and after 7 months) Monthly feedback Data collected using WordWatcher database Education sessions None Ventilator days (survivors) Average: 10 8 6 1 days(p < 0 05) (44% reduction). Median ventilator days fell from 3 to 2 days ICU LOS Average: 13 7 days(sd 19 11) to 8 3 days (SD 10 21) (reduction of 39%) (p < 0 05) Mean duration 10 8 days (SD 15 58) to 6 1 days (SD 8 88) (43% reduction) Compliance (process) Overall compliance: 70 2 79 1% HOBE: 71 83% SEDHOL: 29 63% GUP: 100 100% DVTP: 81 71% DVTP, deep vein thrombosis prophylaxis; GUP, gastric ulcer prophylaxis; HOBE, head of bed elevation; ICU, intensive care unit; LOS, length of stay; NHSN, National Healthcare Safety Network; NNIS, National Nosocomial Infections Surveillance; SEDHOL, sedation hold; VAP, ventilator-associated pneumonia. 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses 227

the FASTHUG bundle (Vincent, 2005) and included feeding assessment, blood glucose monitoring and pain assessment elements. The evidence for these additional elements is not based on research of patients who are mechanically ventilated and therefore they may not be suitable as VCB components. Rather, they are better suited as an addition to a daily patient goals checklist, where the VCB is also a component (DuBose et al., 2008; Khorfan, 2008). Another addition to the VCB in several of the studies was oral care (Berriel-Cass et al., 2006; Hatler et al., 2006; Khorfan, 2008), which is supported by strong evidence. A metaanalysis of chlorhexidine oral care indicated a VAP reduction of 30%, with the benefit most marked in cardiac surgery patients (Chlebicki & Safdar, 2007). Berriel-Cass et al. (2006) included hand-washing as a part of their VCB; however, as reported by the authors, hand-washing was the hardest part of the bundle to measure. Different methods were used in an attempt to gather rigorous data, including peer observations, charge nurse observations, and sign in sheets. However, in practice, hand-washing fell to an honour system with auditing performed by a senior nurse. Implementation strategies The studies reviewed applied a variety of implementation strategies, which included using multidisciplinary teams to review ventilated patients, daily reviews, staff education and written education materials, visual reminders, and compliance audits (see Table 1). Most studies used the VCB protocol guidelines and measurement tools defined by IHI. Six studies included audit and feedback as part of their implementation strategy. Two studies used posters for weekly feedback, using graphs and other descriptions (Hampton et al., 2005; Hatler et al., 2006), and three other studies provided feedback monthly (Crunden et al., 2005; Berriel-Cass et al., 2006; Youngquist et al., 2007). Khorfan (2008) described sharing results but provided little detail about how it was done. Crunden et al. (2005) commented that monthly feedback was important because it reminded clinical staff about their accountability and helped to maintain a priority focus on the intervention. Hatler et al. (2006) presented a novel way to implement the VCB, focussing on behaviour change strategies. This approach enabled researchers to identify barriers and promote recognition activities such as thankyou notes and movie tickets to establish the new behaviour (the VCB) as the norm. However, the impact of this intervention was not investigated. Hampton et al. (2005) included nurse and physician champions to provide direction and influence but did not measure their effect, which may be related to their knowledge of local barriers and facilitation of behaviour change (Sinuff et al., 2008). Two studies used mnemonics to help to promote change. Hatler et al. (2006) used HOT- SPUD and Papadimos et al. (2008) used FASTHUG. The latter authors considered this an important aspect of the intervention, suggesting that use of the mnemonic created a heightened awareness across disciplines regarding the details of the interventions. VCB compliance Compliance with the VCB is an important factor, as it may take some time before a level of compliance is achieved that impacts on outcomes. Half of the studies reviewed reported VCB compliance rates. The large multi-centre study by Resar et al. (2005) demonstrated that increased compliance with the VCB resulted in greater reduction in VAP rates. ICUs with over 95% compliance rates had a 59% decrease of VAP, whereas across all 35 units, there was a 45% decrease. Unahalekhaka et al. (2007) also reported that increased compliance with the VCB decreased incidence of VAP. Two studies (Hatler et al., 2006; Youngquist et al., 2007) included compliance audit and feedback in their intervention; however, the effect of feedback was not measured. Youngquist et al. (2007) provided monthly compliance feedback to staff with a target compliance of 95%, implementing a variety of strategies to improve compliance, including education events. Compliance at two months was 93%, and at six months into the study 100% compliance was achieved. Papadimos et al. (2008) extended their data collection period for an extra year because of initially poor results. In their first data collection period the VAP rate fell from 19 3/1000 to 16 6/1000. However, during their extended data collection period, which also involved reinforcement of procedural interventions (which implied that compliance improved), the VAP rate fell significantly to 7 3/1000. Certain components of the VCB were harder to implement than others. Four studies had difficulty with head of bed elevation compliance but commented that extra education resulted in increased compliance (Hampton et al., 2005; Berriel-Cass et al., 2006; Youngquist et al., 2007; DuBose et al., 2008). DuBose et al. (2008) reported that at month 1, head of bed elevation education was their primary focus and with the use of visual reminders they were able to increase compliance from 32% to 84 5%. Head of bed elevation compliance was also problematic for Hampton et al. (2005). Near 100% compliance was not achieved for nearly a year and extra education was needed for staff, family members and support staff. Berriel-Cass et al. (2006) did not include compliance as one of their process measures, but 228 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses

commented that staff resisted head of bed elevation for reasons which included concerns about pressure ulcer development and blood pressure complications. Compliance with sedation holds was also difficult because of minimal physician support. Khorfan (2008) reported perfect compliance with both DVT and gastric ulcer prophylaxis, whereas Crunden et al. (2005) reported decreased compliance with their DVT protocol. VCB outcomes A variety of outcome measures was used across the 10 studies to indicate effects associated with use of the VCB (see Table 1). Those reported most frequently were VAP rates, ventilator days, ICU LOS, and mortality. Half of the studies also reported on VCB compliance (process measure). VAP rates VAP surveillance was reported in all but two studies (Crunden et al., 2005; Hampton et al., 2005). Overall, there was a trend in the reduction of VAP incidence per 1000 days with lowest and highest reductions being 34% (Unahalekhaka et al., 2007) and 85% (Khorfan, 2008). Of the eight studies reporting VAP rates, most used the CDC diagnosis guidance (Tablan et al., 2003; CDC, 2009). In the USA, both the National Nosocomial Infections Surveillance (NNIS, 2004) and National Healthcare Safety Network (NHSN) (Edwards et al., 2008) provide VAP rate data, according to ICU type (see Table 2), and several studies used these data to benchmark their findings. In three studies, VAP rates per 1000 days were higher than both NNIS and NHSN data (Hatler et al., 2006; Papadimos et al., 2008; Unahalekhaka et al., 2007). However, these studies still demonstrated VAP reductions of 54%, 62% and 34%, respectively. NHSN claims that a high rate of VAP does not necessarily define a problem but it does suggest an area for further investigation (Edwards et al., 2008). Also, a low VAP rate may be the result of inadequate infection surveillance (Edwards et al., 2008). Khorfan (2008) and Youngquist et al. (2007) both reported VAP rates/1000 days below both the NNIS and the NHSN VAP averages, with the former study reporting a VAP reduction of 85% and the latter reporting zero postintervention VAP incidence in one centre. Ventilator days Five studies (Crunden et al., 2005; Hampton et al., 2005; Berriel-Cass et al., 2006; DuBose et al., 2008; Khorfan, 2008) reported pre- and post-data about ventilator days, revealing an overall decrease. Hampton et al. (2005) reported a similar reduction of 42 6% to Crunden et al. s (2005) reported 43 5%. However, the latter study reported on surviving patients only. DuBose et al. (2008) reported a 3 7 13 1% decrease in duration of mechanical ventilation at month 1, but at month 3 found little or no further reductions. Khorfan (2008) reported the number of ventilator days, but did not report on the sample size, so average ventilator days was not recorded. Berriel-Cass et al. (2006) reported a decrease but did not quantify it. Table 2 NNIS 2002 2004 and NHSN 2006 2007 average VAP rates/1000 days: by type of ICU. In 2005, NNIS became amalgamated with other CDC surveillance departments to become NHSN Medical Medical/surgical: major teaching Medical/surgical: all others Surgical Trauma Pooled average NNIS: January 2002 June 2004 4 9 5 4 5 1 9 3 15 2 7 5 NHSN: January 2006 Dec 2007 2 5 3 3 2 3 5 3 9 3 5 1 Results Pre Post Pre Post Pre Post Pre Post Pre Post Pre Post Khorfan (2008) 1 7 0 27 Papadimos et al. (2008) 19 3 7 3 DuBose et al. (2008) 16 3 8 9 Youngquist et al. (2007) 6 01 2 2 2 66 0 0 Unahalekhaka et al. (2007) 13 3 8 8 Berriel-Cass et al. (2006) 8 2 3 3 Hatler et al. (2006) 11 4 5 3 Resar et al. (2005) 6 6 2 7 NHSN, National Healthcare Safety Network; NNIS, National Nosocomial Infections Surveillance; VAP, ventilator-associated pneumonia. Mercy. Unity. 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses 229

ICU LOS and mortality Six studies reported ICU LOS. Papadimos et al. (2008) found no difference, whereas the remaining five studies all reported reductions ranging from 15% to 39 2% (Crunden et al., 2005; Hampton et al., 2005; Berriel-Cass et al., 2006; Hatler et al., 2006; Youngquist et al., 2007). Only two studies provided specific information about mortality rates, indicating a trend in its reduction. Hampton et al. (2005) reported a large decrease in mortality of 50%, whereas Crunden et al. (2005) found a 9% reduction. However, in the latter study there was an increase in the number of ventilated patients during the study period. Khorfan (2008) indicated that one of the project goals was to reduce mortality; however the mortality data were not reported, suggesting somewhat vaguely that the VAP bundle was likely to result in...decreasing morbidity and mortality (p. 17). CRITIQUE Design The research design for all 10 studies was prospective, observational, pre- and post-intervention. All used non-randomized samples, with no concurrent controls for comparison. Resar et al. (2005) commented on their prospective observational design and stated that given its design and lack of bias controls, the research could not establish a causal link between the use of the bundle and a reduction in VAP. They noted that, historically, only experimental research study had been able to support the concept of causality. However, the IHI and JCAHO consider RCTs for this type of research to be unethical because the interventions have already been demonstrated to be effective (Papadimos et al., 2008). Nevertheless, well-controlled prospective observational studies are considerably stronger than retrospective studies and are best for examining causal relationships and to test a hypothesis, especially when the phenomenon is not amenable to randomized controlled experimental designs. In addition, samples for prospective designs may be more representative and investigators may be able to impose tighter controls for bias and confounding factors to rule out competing explanations of the results and to ensure the validity of the research (Polit and Beck, 2007). In an attempt to benchmark their data, Unahalekhaka et al. (2007) used a time-series approach with the pre-intervention period representing baseline data, whereas Khorfan (2008) used NNIS (2004) data for their baseline. A limitation of the latter form of benchmarking is the relevance of the compared data; the NNIS average VAP rate may be different to that of a particular ICU. Three studies gathered baseline data from retrospective chart reviews (Crunden et al., 2005; Hampton et al., 2005; Youngquist et al., 2007). Of these, only Youngquist et al. reported an attempt to control selection bias. In their study, the patients charts were reviewed by a clinical nurse specialist and a respiratory care advisory team, who attempted to benchmark performance using a revised set of CDC guidelines for nosocomial surveillance. The limitation of retrospective comparison is that the two groups are never likely to be totally comparable with respect to all factors, and there is often missing or incomplete information or inaccuracies in the data (Polit and Beck, 2007). Furthermore, there may be other variables affecting the outcome in one period that were not present in another. Samples Three of the studies were multi-centred ones; Youngquist et al. (2007) used two sites, Unahalekhaka et al. (2007) used 18 sites, with the largest study by Resar et al. (2005) collecting data from 35 sites. Of these, only Youngquist et al. identified its ICU sample group (medical/surgical) and reported on the number of patient days. The two largest multi-centre studies (Resar et al., 2005; Unahalekhaka et al., 2007) did not provide information about individual ICUs, such as the type of ICU or the size of the sample drawn from each. This information is important for benchmarking, as NNIS and NHSN aggregate data, across several years, reveal differences between different types of ICU (NNIS, 2004; Edwards et al., 2008). Theotherseven research settings were single ICUs ranging in size from 6 to 18 beds. Of the single-centre studies, five reported their sample size (Crunden et al., 2005; Hampton et al., 2005; Youngquist et al., 2007; DuBose et al., 2008; Khorfan, 2008). Prospective studies may require large samples if the variable of interest is rare. The VAP rate is normally described in terms of its incidence per 1000 ventilation days. Although the majority of studies reported this outcome, there was a large variance in the number of ventilator days for which data were recorded. For example, Youngquist et al. (2007) reported on 205 patient ventilator days in 2003 and 224 days in 2004, whereas DuBose et al. (2008) reported 810 ventilator days over the three months of their study and Khorfan et al. (2008) recorded 3681 ventilator days over three years. Data collection Although all 10 studies reported some information about data collection processes, the details were not uniform, comprehensive or complete. For example, 230 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses

Papadimos et al. (2008) reported that hand-washing compliance was aggressively enforced by infection control employees; however, they did not report on data collection of the other process measures such as VAP surveillance and patient demographics. Hampton et al. (2005) reported on compliance monitoring which was performed three times per week: once by the clinical nurse specialist, again by the unit manager and a third time by a manager from another unit. However, they did not detail how and by whom VAP surveillance, ICU LOS, and mortality data were collected. The tools used for data collection were also not comprehensively reported. Only three studies provided examples of their data collection tools (Hatler et al., 2006; DuBose et al., 2008; Khorfan, 2008). Two studies used a computer-based information system as a data source (Crunden et al., 2005; Resar et al., 2005) and seven studies stated that data were collected using either a daily checklist or daily goals sheet (Crunden et al., 2005; Berriel-Cass et al., 2006; Hatler et al., 2006; Youngquist et al., 2007; DuBose et al., 2008; Khorfan, 2008; Papadimos et al., 2008). DISCUSSION All of the 10 studies that were reviewed are limited by their research designs in that they did not have control groups. However, it is argued that this would be unethical because the VCB is known to improve outcomes. In addition, the studies employed various methods to collect data, of which several were subject to potential collector bias. If no specific control is imposed on data collection and recording, the possibility for incomplete data or inaccurate data is increased. Another limiting factor, in terms of the research design, was that the studies were unblinded. Although it is conceded that it would be very difficult to blind the VCB intervention, this can mean that people who are aware that they are part of an audited study alter their recording of events and their practices in such a way that it does not reflect their usual practice. This is especially the case with observational designs and can impact on the true nature of results. The four elements of the VCB are based on strong evidence, i.e. systematic reviews of RCTs and single RCTs (Berenholtz et al., 2002). Compared to when they are implemented individually, the premise is that when the four elements are bundled together as a suite of practices, it will lead to significantly improved outcomes for ventilated patients. The findings from this review support this: the measures employed in each of the studies demonstrated improved outcomes for patients. As noted above, there was a variety of outcome measures recorded. Nevertheless, there were demonstrably positive results in all studies, with reported decreases in VAP days per 1000 ventilator days, average ventilator days and average ICU LOS. Only one study reported mortality data, which demonstrated a significant reduction of 50%. Whilst this reduction was large, there are many other variables that may have impacted, and further multicentre research is needed to evaluate the impact of VCB implementation and compliance on mortality. A recent meta-analysis of audit and feedback found that despite mixed findings in previous reviews, they are reasonably effective tools for changing provider behaviour and quality of care, if designed correctly (Hysong, 2009). Improving compliance with the VCB has been shown to improve VAP outcomes. For example, Cocanour et al. (2006) demonstrated that daily audit of compliance with a modified VCB and weekly feedback reporting of results was associated with a statistically significant decrease in VAP rates. In a more recent multi-centre improvement study, increased VCB compliance from 50% to 82% was associated with a 41% reduction in VAP (Bonello et al., 2008). Several of the studies reviewed reported different rates of compliance with each of the VCB elements. However, it is important to note that compliance with the VCB is an all or nothing measure. It is defined by IHI (2006, p. 1) as The percentage of intensive care patients on mechanical ventilation for whom all four elements of the ventilator bundle are implemented and documented on the daily goals sheet and/or elsewhere in the medical record. Some studies reported their overall VCB compliance rates, and where these data were available, there was a clear trend to suggest that the better the adherence to the VCB, the greater the reduction in VAP incidence and ventilator days. Thus, if a relationship between process measures (i.e. VCB compliance) and improved outcome measures (e.g. VAP reduction) can be demonstrated, health care providers may be able to monitor performance using process measures rather than the more resource-intensive outcome measures (Resar et al., 2005). It is important to also note that if a bundle element is contra-indicated for a particular patient (for example, DVTP in a patient with a bleeding disorder) then that element should still be regarded as compliant (IHI, 2006). The value of the VCB as a cluster of interventions has been questioned in relation to VAP reduction, on the basis that only one element (head of bed elevation) is directly associated (Wip and Napolitano, 2009). Head of bed elevation between 30 and 45 decreases the risk of aspiration of gastric and oropharyngeal secretion plus aids ventilation and minimizes atelectasis. 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses 231

Its inclusion in the VCB is based on the research by Drakulovic et al. (1999) who showed that the semirecumbent body position reduced frequency and risk of nosocomial pneumonia in mechanically ventilated patients, especially those receiving enteral nutrition. However, if sedation hold and assessment of readiness to wean result in earlier extubation then the risk of VAP may be reduced. In several of the studies that were reviewed, additional elements were added to the VCB. This makes it difficult to determine what proportion of the outcome effect was attributable to the VCB itself. Although it is conceded by IHI that there are several adjunct elements which have been added to the VCB in many ICUs that are supported by strong evidence, e.g. subglottic drainage (Dezfulian et al., 2005), it does not wish to add further elements to the existing bundle because of its demonstrated effectiveness in achieving zero rates of VAP (IHI, 2008). It is argued that although there is a tendency for ICUs to implement all possible interventions, this may be unnecessary, as a select few is effective in minimizing risks. Furthermore, it is difficult to make comparisons between ICUs in which different elements have been added to the VCB. IHI (2008) encourages researchers to maximize implementation of the existing VCB before adding other components. Thus, when baseline measures are obtained with the VCB, further components can be added singly, and their additional effect on outcomes can be measured. There was a variety of implementation strategies used in the VCB studies reviewed. All used several methods to implement the VCB, including the use of staff education sessions, supportive written material, reminders and multi-disciplinary team meetings. A recent meta-analysis suggested a multifaceted approach to guideline implementation such as an active educational strategy combined with a reminder and audit, and feedback system is an effective approach for knowledge transfer in the ICU (Sinuff et al., 2008). However, the authors also stated that further research is needed to increase the understanding of effective strategies to improve clinical outcomes for critically ill patients. Significant gaps remain in the transfer of research into clinical practice and, although clinical guidelines may improve processes of care, their use alone is insufficient to change clinician behaviour (Sinuff et al., 2008). Although not specifically related to VCB implementation, some studies have demonstrated significant VAP reduction when structured multi-disciplinary education programs were implemented (Babcock et al., 2004; Salahuddin et al., 2004). Although use of the VCB has demonstrated reduced VAP rates, there are other interventions that have been shown to reduce VAP, and it should not be assumed that the VCB is the only appropriate intervention. Other evidence-based interventions have been comprehensively reported by the Canadian Clinical Trials Group (Muscedere et al., 2008). However, of particular interest is the degree of head of bed elevation, and further studies are warranted to test a more easily achievable 10 30 position (Wip and Napolitano, 2009). CONCLUSIONS Because of the limitations of the observational design used in the studies reviewed, a definitive causal link between the VCB and VAP incidence, length of ventilation, and ICU LOS cannot be proved. However, the weight of evidence in terms of the magnitude of difference in the measured outcomes especially with regard to VAP reduction suggests a strong relationship. As noted above, the association is considered to be so strong that it is considered by IHI and JCAHO to be unethical not to use the VCB. It should also be emphasized that, individually, each of the VCB elements is supported by its own experimental evidence at the highest level and, on this basis, each in its own right should be implemented for ventilated patients. In the absence of rigorous research evidence suggesting otherwise, standard practice should involve the use of the VCB for all adult ventilated ICU patients. A limited number of studies suggests that VCB compliance impacts on outcomes. Thus, when the VCB is used, it is important to audit compliance rates and accordingly implement processes that help ensure sustained outcomes. In terms of further research, it would be advisable to develop further prospective studies, which are controlled tightly in order to be able to measure VCB effects. In particular, further research is needed to demonstrate the effect of regular feedback on VCB compliance, and its subsequent effect on outcomes. In general, processes are easier to measure than outcomes and can be used to provide immediate feedback to providers regarding their performance. However, further research would be helpful to demonstrate which implementation and audit/feedback strategies are most effective in achieving clinician behaviour change that results in improved clinical outcomes. With respect to the VCB, rigorous research is needed to establish the supplementary benefit of other elements such as tight glucose control, which others have added. This requires the use of good baseline data with the VCB alone, followed by individual implementation and outcome monitoring of additional elements. Thorough evaluation strategies are needed to establish best practice and create a benchmark against which other process measures can be assessed. 232 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses

WHAT IS KNOWN ABOUT THIS TOPIC The VCB is a group of four evidence-based procedures, which when clustered together and implemented as an all or nothing strategy, may result in substantially greater clinical improvement. VAP is a nosocomial lung infection associated with microaspiration via an endotracheal tube in ventilated patients. Since the VCB was introduced there have been several studies that have reported significant VAP rate reductions associated with its implementation. WHAT THIS PAPER ADDS This paper has reviewed the evidence from studies that have investigated outcomes associated with VCB implementation. It is concluded that the VCB is associated with improved VAP outcomes. This review has identified that there is a lack of consistency in methodological approach in the various studies reviewed. This review has identified that some studies have used the VCB with added elements, which make it difficult to determine which elements are responsible for the measured outcomes. This review has identified that the level of compliance with the VCB impacts on outcomes, and therefore further research is required in this area. REFERENCES Aboelela SW, Stone PW, Larson EL. (2007). Effectiveness of bundled behavioural interventions to control healthcareassociated infections: a systematic review of the literature. Journal of Hospital Infection; 66: 101 108. American Thoracic Society. (2005). Guidelines for the management of adults with hospital-acquired, ventilator-associated, and healthcare-associated pneumonia. American Journal of Respiratory Critical Care Medicine; 171: 388 416. Babcock HM, Zack JE, Garrison T, Trovillion E, Jones M, Fraser VJ, Kollef MH. (2004). An educational intervention to reduce ventilator-associated pneumonia in an integrated health system. Chest; 125: 2224 2231. Berenholtz S, Dorman T, Ngo K, Provonost PJ. (2002). Qualitative review of intensive care quality indicators. Journal of Critical Care; 17: 1 12. Berenholtz SM, Milanovich S, Faircloth A, Prow DI, Earsing K, Lipsett P, Dorman T, Pronovost PJ. (2004). Improving care for the ventilated patient. Joint Commission Journal on Quality and Safety; 30: 195 204. Berriel-Cass D, Adkins FW, Jones P, Fakih MG. (2006). Eliminating nosocomial infections as Ascension Health. Joint Commission Journal on Quality and Patient Safety; 32: 612 620. Bonello RS, Fletcher CE, Becker WK, Clutter KL, Arjes SL, Cook JJ, Petzel RA. (2008). An intensive care unit quality improvement collaborative in nine Department of Veteran Affairs hospitals: reducing ventilator-associated pneumonia and catheter-related bloodstream infection rates. Joint Commission Journal on Quality and Safety; 34: 639 645. Bouza E, Burillo A. (2009). Advances in the prevention and management of ventilator-associated pneumonia. Current Opinion in Infectious Diseases; 22: 345 351. Centers for Disease Control (CDC). (2009). Ventilator-associated pneumonia (VAP) event. In: NHSN Manual: Patient Safety Component Protocols. [Online]. Available at: http://www.cdc. gov/nhsn/pdfs/pscmanual/6pscvapcurrent.pdf (accessed 20/04/10). Centers for Disease Control and Prevention, Atlanta GA, USA. Chastre J, Fagon J. (2002). Ventilator-associated pneumonia. American Journal of Respiratory and Critical Care Medicine; 165: 867 903. Chlebicki MP, Safdar N. (2007). Topical chlorhexidine for prevention of ventilator associated pneumonia: a meta analysis. Critical Care Medicine; 35: 595 602. Cocanour CS, Peninger M, Domonoske BD, Li T, Wright B, Valdivia A, Luther KM. (2006). Decreasing ventilator associated pneumonia in a trauma ICU. The Journal of Trauma, Injury, Infections, and Critical Care; 61: 122 130. Collard HR, Saint S, Matthay MA. (2003). Prevention of ventilatorassociated pneumonia: an evidence-based systematic review. Annals of Internal Medicine; 138: 494 501. Cooke FJ, Holmes AH. (2007). The missing care bundle: antibiotic prescribing in hospitals. International Journal of Antimicrobial Agents; 30: 25 29. Crunden E, Boyce C, Woodman H, Bray B. (2005). An evaluation of the impact of the ventilator care bundle. Nursing in Critical Care; 10: 242 246. Dezfulian C, Shojania K, Collard HR, Kim HM, Matthay MA, Saint S. (2005). Subglottic secretion drainage for preventing ventilator associated pneumonia: a meta analysis. The American Journal of Medicine; 118: 11 18. Drakulovic MB, Torres A, Bauer TT, Nicolas JM, Nogue S, Ferrer M. (1999). Supine body position as a risk factor for nosocomial pneumonia in mechanically ventilated patients: a randomised trial. Lancet; 354: 1851 1858. DuBose JJ, Inaba K, Shiflett A, Trankiem C, Teixeira PGR, Salim A, Rhee P, Demitriades D, Belzberg H. (2008). Measurable outcomes of quality improvement in the trauma intensive care unit: the impact of a daily quality rounding checklist. The Journal of Trauma, Injury, Infection and Critical Care; 64: 22 29. Edwards JR, Peterson KD, Andrus ML, Dudeck MA, Pollock DA, Horan TC. (2008). National Healthcare Safety Network (NHSN) Report, data summary for 2006 through 2007, issued November 2008. American Journal of Infection Control; 36: 609 626. Fong JJ, Cecere K, Unterborn J, Garpestad E, Klee M, Devlin JW. (2007). Factors influencing variability in compliance rates and clinical outcomes among three different severe sepsis bundles. The Annals of Pharmacotherapy; 41: 929 936. Fulbrook P. (2003). Developing best practice in critical care nursing: knowledge, evidence and practice. Nursing in Critical Care; 8: 96 102. 2011 The Authors. Nursing in Critical Care 2011 British Association of Critical Care Nurses 233