Assessment of a training programme for the prevention of ventilator-associated pneumonia

Background Ventilator-associated pneumonia (VAP) is the most frequent nosocomial infection in intensive care units (ICUs). Most published studies have analysed nurses' theoretical knowledge about a specific procedure; however, the transfer of this knowledge to the practice has received little attention. Aim To assess the impact of training session on nurses' knowledge regarding VAP, compliance with VAP preventive measures, VAP incidence and determining whether nursing workload affects compliance. Method A prospective, quasiexperimental, pre- and post-study of the nursing team in a 16-bed medical/surgical ICU. Pre-intervention phase: a questionnaire to assess nurses' knowledge of VAP prevention measures, direct observation and review of clinical records to assess compliance. Intervention phase: eight training sessions for nurses. The post-intervention phase mirrored the pre-intervention phase. Findings Nurses answered more questions correctly on the post-intervention questionnaire than on the pre-intervention (17·87 ± 2·69 versus 15·91 ± 2·68, p = 0·002). Compliance with the following measures was better during the post-intervention period (p = 0·001): use of the smallest possible nasogastric tube, controlled aspiration of subglottic secretions and endotracheal tube cuff pressure, use of oral chlorhexidine and recording the endotracheal tube fixation number. VAP incidence remained unchanged throughout the study. However, a trend towards lower incidence of late (>4 days after intubation) VAP was observed (4·6 versus 3·1 episodes/1000 ventilation days, p = 0·37). Conclusion The programme improved both knowledge of and compliance with VAP preventive measures, although improved knowledge did not always result in improved compliance.

Many studies have analysed nurses' theoretical knowledge regarding specific procedures (Fulbrook, 2003;Blot et al., 2007). However, the application of this knowledge to practice has received little attention (Chang et al., 2002). Moreover, the instruments used in these studies (e.g. questionnaires) were mainly limited or not entirely appropriate (Cormack and Benton, 1996). Questionnaires may not be reliable measure of compliance as answers may not reflect performance. Thus, it is important to observe nursing practices in situ. Few studies have used direct observation to study how nurses apply non-drug-based VAP preventive measures (Day et al., 2002;Montial et al., 2005;Williams et al., 2008). In their study on bed headboard positioning in critical patients, Montial et al. (2005) observed that only 25% of nurses placed the headboard according to guidelines (>30 • ).
The Disease Control Centres (DCCs) consider training a key strategy in reducing VAP incidence and care costs (Thompson, 2000;Zack et al., 2002;Hilary et al., 2004).
The main aim of this study was to assess a training programme to improve nurses' compliance with VAP preventive measures. The secondary objectives were to (a) determine programme impact on nurses' theoretical knowledge of and compliance with the measures, (b) analyse the relationship between workload and compliance and (c) measure programme impact on VAP incidence.

MATERIAL AND METHODS
A prospective, quasi-experimental, pre-and post-study was carried out in a 16-bed medical-surgical ICU with a mean patient/nurse ratio of 1:2·28 from January 2008 to May 2009. All the ICU nursing staff took part in the study (n = 58), excluding the research team. To ensure intervention quality, nursing staff rotation was limited to 15% during the study period. The study was approved by the hospital ethics committee and participation was voluntary. The variables studied were as follows: • Nurses' theoretical knowledge: An ad hoc questionnaire of 22 multiple choice questions was designed following 14 non-drug preventive measures related to nursing care selected from the European Task Force (Torres and Carlet, 2001) and Centers for Disease Control and Prevention (CDC) (Tablan et al., 2004) guidelines. These were grouped into five categories (Table 1). To prevent bias, nurses were asked not to comment on their answers until the last questionnaire had been collected.  The study was divided into three phases: -Pre-intervention phase: The following aspects were assessed: (a) nurses' basic knowledge using the questionnaire; (b) aspiration of endotracheal secretions during 180 min of non-participatory observation consisting of a systematic sample of 77 observations randomized by day of the week, shift and patient's room; (c) headboard positioning, assessed daily in all three shifts through direct observation and (d) the eight remaining preventive measures on patients' clinical charts, daily in all three shifts. -Intervention phase: The research team designed and held a series of 60-min training sessions covering VAP definition, problem epidemiology and scope, risk factors, aetiology, risk reducing methods and endotracheal secretion aspiration procedure. The session was held eight times to ensure maximum attendance of the three nursing shifts. The session included theoretical training and practical exercises to identify possible errors. In addition, informative posters were displayed in the ICU, a consultation protocol was drawn up and each nurse received a leaflet summarizing the guidelines. -Post-intervention phase: Nurses were assessed identically to the pre-intervention phase.

Data analysis
Results are presented as mean and standard deviations for quantitative variables, likewise as frequencies and percentages for qualitative variables. The results of the pre-and post-intervention phases were compared using Student's t-test for quantitative variables and Pearson's chi-square or Fisher's exact

FINDINGS Impact on knowledge
The questionnaire was completed by 48 (82·7%) professionals in the pre-intervention phase and 31 (64·5%) in the post-intervention phase. The mean overall number of correct responses in the post-intervention questionnaire was higher than that in the pre-intervention (17·87 ± 2·69 versus 15·91 ± 2·68; p = 0·002). Knowledge about how to prevent microaspiration of subglottic secretions and oropharyngeal hygiene improved significantly (Table 2).

Impact on actual performance
For each phase of the study, a chart was drawn up indicating the day, shift and patient. These two charts included 77 randomly selected observations to determine nurses' adherence to guidelines during aspiration of endotracheal secretions. This procedure was observed 67 times in the pre-intervention phase and 111 times in the post-intervention phase. The difference in the number of observations may be due to the procedure not being performed or being performed more than once during the observation period.
Headboard positioning and the other eight preventive measures were checked at the same time on 1145 occasions in the pre-intervention phase and on 1147 occasions in the post-intervention phase ( Table 3). The discrepancy in number of observations is due to illegible handwriting on the charts.

Workload dependent compliance
Worse compliance was observed with recommendations for the narrowest nasogastric tube, correct endotracheal cuff pressure, gastric retention control, oral hygiene and endotracheal tube fixation in patients requiring more care, measured by NEMS (Table 4).

Impact on knowledge and compliance
In general terms, the knowledge nurses have acquired regarding VAP preventive measures has significantly increased post-intervention. In particular, the results obtained show that this educational intervention improved ICU nurses' scientific knowledge about measures related to microaspiration of subglottic secretions and oral hygiene, corroborating reports (Zack et al., 2002;Kelleher and Andrews, 2008) that pre-intervention praxis is based more on routine than on scientific evidence. In addition, nurses' compliance with controlling and reducing gastric reflux, preventing microaspiration, oral hygiene and endotracheal tube fixation improved.
As observed in other studies, although oral hygiene procedures were performed more than once a day before the training programme, most nurses did not use chlorhexidine solution (DeRiso et al., 1996;Tablan et al., 2004;Koeman et al., 2006). Cason et al. (2007) also found that only 26% of nurses used chlorhexidine. Compliance with recommendations for oropharyngeal hygiene improved after the training programme.
Compliance with both measures related to gastric reflux control (headboard positioning and nasogastric tube diameter) improved. Headboard angle after the intervention was >30 • in 82·6% of observations. This percentage is higher than that reported by other authors. Curtis et al. (2006), who also compared headboard angle using a template, found the angle was >28 • in only 23% of observations. Williams et al. (2008) obtained 72% adherence after incorporating a device to demonstrate headboard angle. Lyerla et al. (2010), after a training session, found correct headboard positioning in 67% of observations. Our data corroborate previous findings (McMullin et al., 2002;Montial et al., 2005) that nurses tend to overestimate the headboard angle. Interestingly, nurses knew some VAP preventive measures yet did not apply them (Ricart et al., 2003). In particular, the importance of hand washing pre-and post-secretion aspiration was well known in the preintervention phase, as might be expected for a preventive measure with IA evidence (Tablan et al., 2004). In the post-intervention period, a slight improvement was observed. However, adherence remained insufficient considering the rate of morbimortality associated with nosocomial infections in critical patients. Several publications have reported the low adherence to this crucial guideline (Day et al., 2002;Boyce and Pittet, 2003;Creedon, 2005;Williams et al., 2008). Recently, Kelleher and Andrews (2008) have reported 31% adherence to hand washing before aspirating endotracheal secretions. As suggested elsewhere (Boyce and Pittet, 2003), low adherence to this guideline in this study might be due to heavy workload. Cho et al. (2003) concluded that a heavy nursing workload contributes to the failure to wash hands and isolate patients with multiresistant organisms. Following international campaigns for patient safety (Pittet et al., 2006), we have introduced more efficient alcohol-based solutions (Pittet et al., 2000;Girou et al., 2002).
Compliance with other recommendations for aspirating endotracheal secretions was better than that referred by other authors: sterile catheters were used in 98% (compared with 59% reported by Kelleher and Andrews, 2008), and as with other studies (Ania et al., 2004;Kelleher and Andrews, 2008), sterile gloves were used in over 95%. The results obtained are not statistically significant, because compliance with these practices was very high before the training intervention and was difficult to improve.
Failure to wash hands when aspirating secretions suggests the persistence of the misconception that wearing gloves makes hand washing unnecessary (Pratt et al., 2001).

Workload-dependent compliance
Failure to apply guidelines derives from resistance to change, difficulties in accessing literature, lack of resources, costs of the interventions and misinterpretation (Needleman et al., 2002;Rello et al., 2002;Ricart et al., 2003).
The results obtained in this study coincide with the findings of Hugonnet et al. (2007), who describe work overload as a reason for not following the guidelines. Thus, compliance with preventive measures decreases as NEMS increases. Six measures were affected: using the narrowest nasogastric tube, gastric reflux control, oral hygiene, aspiration of subglottic secretions, checking the patency of subglottic tubes and correct endotracheal tube fixation. Apparently, when faced with an increased workload, nurses ignore measures they consider less important, such as replacing a nasogastric tube or completing charts thoroughly. Failure to record an action does not mean it was not performed; nursing registers are seldom complete (Alconero et al., 1999). Moreno and Miranda (1998) propose classifying NEMS results into three groups: group 1 (light workload): NEMS < 21; group 2 (moderate workload): NEMS 21-30 and group 3 (heavy workload): NEMS > 30. All the patients studied were in group 3, i.e. unstable patients requiring a lot of care (Table 4). Despite the small workload differences generated by patients, significant differences were observed in compliance with some preventive measures, probably due to the large number of determinations recorded.

Impact on VAP
Unlike other studies, which report decreases in VAP incidence after improved compliance with these measures (Zack et al., 2002), we observed no differences in overall VAP incidence pre-and post-intervention (9·9 versus 9·3 episodes/1000 ventilation days). However, late VAP (>4 days of MV) tended to decrease after the intervention (4·6 versus 3·1 episodes/1000 ventilation days). Nurses' preventive measures have a greater impact on late VAP than on early VAP.
The Spanish National Surveillance Study of Nosocomial Infection in the ICU (ENVIN, 2008(ENVIN, -2009) showed a decrease in VAP incidence in Spain from 14·95 to 11·44 episodes/1000 days MV in 2008and 2009, respectively (ENVIN, 2008-2009. Overall VAP incidence was difficult to improve applying only non-drug-related preventive measures because it was already low before the intervention.

Study limitations
The observational method has inherent limitations. Most importantly, the effect of the 'observer' on the 'observed'. Participants who are aware they are being observed may change their behaviour, introducing the Hawthorne effect bias. Another limitation is that many subjects who completed the first questionnaire did not participate in the second phase because they were no longer working in the ICU. However, we consider this drawback less important, because 64·5% of all possible subjects were enrolled.
Another possible limitation of this study is that the questionnaire used to evaluate nurses' knowledge was not validated. Although there is no standardized way of assessing knowledge, we trust that the results obtained do objectively assess nurses' knowledge on VAP prevention, as the questions included in the questionnaire were based on the guidelines established by the CDC (Tablan et al., 2004) and the European Task Force (Torres and Carlet, 2001).
As this was not a randomized study, other concomitant factors may have influenced the results. Nevertheless, during the study, the antibiotics policy remained constant and changes in the nursing staff were minimal.
As this is a single-centre study, the results obtained cannot be extrapolated to centres with different antibiotic policies or nurse-patient ratios.
One should bear in mind that the results obtained regarding the level of knowledge acquired cannot be exclusively attributed to the educational programme. This fact may be influenced by the memory effect (memory bias), i.e. remembered from completion of the first questionnaire. Nevertheless, this aspect would not explain the statistically significant differences found; therefore, the impact of the teaching programme should be assessed positively. Lastly, due to implementation of the teaching programme, the results cannot be explained as a whole, because the design did not consider periodical series being carried out.

CONCLUSIONS
The positive results obtained in this study lend support to the CDC's recommendations to reinforce training to improve adherence to VAP preventive strategies. Training activities and evidence-based protocols aimed at ICU nurses, improving the care quality and narrowing the gap between scientific knowledge and actual performance.
The training programme improved ICU nurses' theoretical knowledge and adherence to VAP preventive measures.
The results yielded show that the training programme carried out improved nurses' knowledge and clinical practice regarding VAP preventive strategies. It should be pointed out that information obtained from the two questionnaires clearly shows that nurses' scientific knowledge is not necessarily applied in daily practice, which justifies the need of training strategies to reinforce adherence to preventive measures against VAP.
A new line of research should look into the reasons why ICU nurses do not put into clinical practice the measures they know are important. A change in professional practice will only be possible through indepth knowledge of the reasons for non-adherence to these guidelines.

WHAT IS KNOWN ABOUT THIS TOPIC
• VAP, the main nosocomial infection in ICUs, increases duration of hospital stay, cost and patient mortality.
• Nurses are responsible for applying non-pharmacological measures to prevent VAP; however, implementation thereof is not standard.
• The main reasons for the incomplete implementation of non-adherence to preventive measures may be due to heavy workload, a deficient theoretical background and a clinical practice based on established routines as opposed to scientific evidence.

WHAT THIS PAPER ADDS
• Training programmes improve nurses' adherence to non-pharmacological VAP prevention protocols.
• The direct, passive observation method reveals the current degree of non-pharmacological measures by nurses.
• The study opens a new line for specifically assessing the factors involved in the non-implementation of theoretical knowledge into clinical practice.