Reducing Ventilator-Associated Pneumonia in Intensive Care
Reducing Ventilator-Associated Pneumonia in Intensive Care
Objectives: Ventilator-associated pneumonia is the most common intensive care unit-acquired infection. Although there is widespread consensus that evidenced-based interventions reduce the risk of ventilator-associated pneumonia, controversy has surrounded the importance of implementing them as a "bundle" of care. This study aimed to determine the effects of implementing such a bundle while controlling for potential confounding variables seen in similar studies.
Design: A before-and-after study conducted within the context of an existing, independent, infection surveillance program.
Setting: An 18-bed, mixed medical–surgical teaching hospital intensive care unit.
Patients: All patients admitted to intensive care for 48 hrs or more during the periods before and after intervention.
Interventions: A four-element ventilator-associated pneumonia prevention bundle, consisting of head-of-bed elevation, oral chlorhexidine gel, sedation holds, and a weaning protocol implemented as part of the Scottish Patient Safety Program using Institute of Health Care Improvement methods.
Measurements and Main Results: Compliance with head-of-bed elevation and chlorhexidine gel were 95%–100%; documented compliance with "wake and wean" elements was 70%, giving overall bundle compliance rates of 70%. Compared to the preintervention period, there was a significant reduction in ventilator-associated pneumonia in the postintervention period (32 cases per 1,000 ventilator days to 12 cases per 1,000 ventilator days; p < .001). Statistical process control charts showed the decrease was most marked after bundle implementation. Patient cohorts staying ≥6 and ≥14 days had greater reduction in ventilator-associated pneumonia acquisition and also had reduced antibiotic use (reduced by 1 and 3 days; p = .008/.007, respectively). Rates of methicillin-resistant Staphylococcus aureus acquisition also decreased (10% to 3.6%; p < .001).
Conclusions: Implementation of a ventilator-associated pneumonia prevention bundle was associated with a statistically significant reduction in ventilator-associated pneumonia, which had not been achieved with earlier ad hoc ventilator-associated pneumonia prevention guidelines in our unit. This occurred despite an inability to meet bundle compliance targets of 95% for all elements. Our data support the systematic approach to achieving high rates of process compliance and suggest systematic introduction can decrease both infection incidence and antibiotic use, especially for patients requiring longer duration of ventilation.
The United States-based Institute for Healthcare Improvement (IHI) has championed the "bundle" approach to improving practice and emphasizes the need to achieve high overall compliance rates (preferably 95% compliance with all bundle elements). Implementing care bundles has been strongly advocated in ventilated intensive care patients, who are at risk for ventilator-associated pneumonia (VAP). VAP is associated with prolonged mechanical ventilation, intensive care unit (ICU) and hospital stay, greater illness costs, and possibly higher mortality. Several clinical interventions have been found to reduce the incidence of VAP, including elevation of the head of the bed, a daily sedation break, a daily trial of ventilator weaning, and topical oral chlorhexidine. The quality of evidence supporting the effectiveness of each intervention and the relative importance of each has been questioned. VAP incidence has become a quality indicator in many healthcare systems, leading to comparisons between ICU and, in the United States, may influence reimbursement rates by insurers. The "VAP prevention bundle" is a central component of most critical care patient safety programs.
An important issue relating to reporting VAP incidence is the method of reporting, because this requires a clinical suspicion and the recognition of clinical signs, which introduces the potential for reporting bias in unblinded trials and quality-improvement studies. Diagnostic method can also vary; patients may meet clinical criteria alone or have a microbiologically confirmed diagnosis, which can be made using various approaches that vary in sensitivity and specificity. A recent systematic review of care bundles for ventilated patients concluded that a lack of methodologic rigor in published studies meant it was not possible to conclude that the bundles were either clinically effective or cost-effective. Previous studies have universally reported VAP rates, but reporting of more patient-centered outcomes such as antibiotic use and length of stay have been more variable, thus making it more difficult to estimate the impact of any reduction in VAP on patients and ICUs.
This study set out to establish the effects of systematically implementing a VAP prevention bundle using IHI methodology on VAP rates, antibiotic use, and duration of ICU stay in an ICU with clinical and research interests in VAP and preexisting VAP prevention guidelines.
This study was conducted in an 18-bed, mixed medical–surgical ICU in a Scottish teaching hospital admitting >1000 patients per year, of whom ≈80% require support of at least two organs or invasive ventilation, or both, and 50% stay for 48 hrs or longer. We have been studying VAP since 2005, when we established independent ICU infection surveillance using the Hospitals in Europe Linked for Infection Control through Surveillance methodology. Screening, data collection and reporting was undertaken by a trained, dedicated, full-time nurse and diagnosis of VAP was made independently by the treating clinical team. Chest radiograph interpretation was undertaken "off-line" and by clinicians who were independent of the treating team. For VAP diagnosis, Hospitals in Europe Linked for Infection Control through Surveillance has a two-stage definition: first, clinically suspected VAP based on clinical criteria; and second, microbiologically confirmed VAP based on further investigations. The details of this process are shown in Figure 1. Antibiotic use was also recorded prospectively for all patients on a daily basis. We used statistical process control methodology to report rates and before implementing IHI methodology had made attempts to decrease VAP incidence by implementing unit guidelines and protocols including head-of-bed elevation, a nurse-led weaning protocol, and hourly clinical sedation scoring linked to a protocol. Intermittent audits of compliance with these interventions showed variable performance, but we were not recording these processes or feeding them back systematically. From 2005 to 2008, we were unable to decrease the incidence of clinically diagnosed VAP but we had shown that increasing the use of quantitative analysis of bronchoalveolar lavage fluid for microbiological diagnosis resulted in a decrease in the reported incidence of microbiologically confirmed VAP, which was explained by superior test specificity compared with analysis of tracheal aspirates.
(Enlarge Image)
Figure 1.
Diagnostic criteria for clinical and microbiologically confirmed ventilator-associated pneumonia (VAP) used in this study, based on the criteria set out by the Hospitals in Europe Linked for Infection Control through Surveillance project (17). BAL, bronchoalveolar lavage; PCR, polymerase chain reaction.
This study used a before-and-after structure to analyze the effects of implementing a VAP prevention bundle. In 2007 National Health Service Scotland launched the Scottish Patient Safety Programme in collaboration with the IHI as the first nationwide patient safety program. One key aim was to reduce VAP incidence in Scottish ICUs, and a VAP prevention bundle was developed by the Scottish Intensive Care Society Audit Group for the Scottish Patient Safety Programme. The bundle has five elements: (1) a daily sedation hold; (2) daily trial of ventilator weaning for suitable patients; (3) head-up position; (4) chlorhexidine mouth care; and (5) subglottic secretion drainage using specialized endotracheal tubes. The only elements of the bundle that we had not attempted to implement already were chlorhexidine mouth care and subglottic suction endotracheal tubes, but we had not used the systematic approach to practice change promoted by the IHI for any of the bundle components and were unsure if the Scottish Patient Safety Programme would produce clinically important benefits.
We chose to implement four elements of the VAP prevention bundle, namely a daily sedation hold and trial of ventilator weaning for suitable patients (which combined were called "wake and wean"), head-up position, and chlorhexidine mouth care. We did not implement subglottic secretion drainage using specialized endotracheal tubes, because this required frequent endotracheal tube changes and most of our patients were admitted intubated from either the operating theater or the emergency department. The additional cost of specialized tubes meant this was also the most costly bundle element.
The wake and wean algorithm was based on that used in a recent randomized controlled trial. In our unit, the nurse at the bed space was required to assess the patient's suitability for a sedation break and weaning of ventilatory support every morning. Bundle introduction was accompanied by nurse education relating to wake and wean practice. Nurses were asked to follow a checklist each morning that included an assessment of whether they considered the patient suitable for a wake and wean trial and whether they performed the trial. This "process measure" was completed as a checklist on the patient chart each morning. Compliance was defined as whether the checklist was completed without individually assessing whether medical staff agreed with the decision made. When in doubt, the nurse was asked to consult medical staff. Unless there were clinical indications for supine nursing, all patients were placed in a position of at least 30 degrees head-up tilt. Topical buccal 1% chlorhexidine gel was prescribed four times per day for all ventilated patients unless the patients either were allergic to the gel or refused the gel (2% gel is unavailable in the United Kingdom).
Bundle elements were applied using Plan, Do, Study, Act cycles until a method of implementation was identified that maximized compliance. Methods used during implementation included nurse and medical champions, teaching materials, education sessions, bedside cues, changing the 24-hr observation charts, and feedback of compliance at meetings, by e-mail, and with posters. All modifications to improve compliance were led by bedside nursing and medical staff.
Compliance of all patients was audited weekly at unannounced and variable times by an independent, nonclinically trained audit clerk who audited charts from the previous day to minimize observation bias. Compliance was defined as clear documentation either confirming compliance or deciding to avoid a bundle element for clinical reasons. Failure to complete bundle documentation was coded as a lack of compliance. Throughout the study period, standard infection control precautions were in place and were not altered. These included removal of outdoor clothing/white coats by visiting clinicians, single-use aprons for each bedside, bedside alcohol gel, and access to sinks at each bed space for hand decontamination. Regular audit of adherence to these measures showed no major changes or trends throughout the period of study.
We were interested in whether the introduction of the Scottish Intensive Care Society Audit Group VAP bundle using the IHI approach would decrease the incidence of both clinically diagnosed and microbiologically confirmed VAP, and whether it changed antibiotic use. We also wanted to measure its effect on rates of methicillin-resistant Staphylococcus aureus (MRSA) acquisition, another key quality indicator. MRSA acquisition was defined as MRSA cultured from either surveillance swabs or samples taken during investigation of suspected sepsis in a patient who was MRSA-negative at ICU admission based on admission screening swabs or other pre-ICU cultures. All patients were screened at ICU discharge to enable acquisition status to be determined.
We chose duration of mechanical ventilation, ICU stay, and ICU mortality as important patient outcomes that also have important cost implications as secondary end points. We recognized that patients with longer ICU stays are at greater risk for VAP, so we reported outcomes for several patient subcohorts, namely all patients at risk for VAP (ICU stay ≥48 hrs), patients with ICU stay ≥6 days, and patients with ICU stay ≥14 days. Our ICU also reports standardized mortality ratios on an annual basis as part of the Scottish Intensive Care Society Audit Group national audit, which includes all patients admitted to the ICU and compares ultimate hospital mortality to the Acute Physiology and Chronic Health Evaluation II prediction model. We report trends in these over the periods covered by the study as an overall indicator of ICU performance.
Elements of the Scottish Intensive Care Society Audit Group VAP prevention bundle were introduced progressively between February and September 2008. We used our surveillance data from January 2005 to February 2008 as baseline data, from February 2008 to September 2008 as the "run-in" period, and from September 2008 to August 2009 as the "post-VAP prevention bundle implementation" period. We compared the periods before and after the intervention using conventional statistical methods for comparing.
Summary statistics using Z-test to compare proportions, Mann-Whitney U test to compare median values (Prism; Graphpad, LA Jolla, CA), and Poisson regression for incident densities (PASW version 18.0; IBM, Armonk, NY). Changes were also assessed using statistical process control charts using monthly data throughout the period of observation. Patient gender, age, and admission Acute Physiology and Chronic Health Evaluation II score were compared as potential confounders to a before-and-after evaluation. As a service evaluation, ethics committee approval was not required.
Abstract and Introduction
Abstract
Objectives: Ventilator-associated pneumonia is the most common intensive care unit-acquired infection. Although there is widespread consensus that evidenced-based interventions reduce the risk of ventilator-associated pneumonia, controversy has surrounded the importance of implementing them as a "bundle" of care. This study aimed to determine the effects of implementing such a bundle while controlling for potential confounding variables seen in similar studies.
Design: A before-and-after study conducted within the context of an existing, independent, infection surveillance program.
Setting: An 18-bed, mixed medical–surgical teaching hospital intensive care unit.
Patients: All patients admitted to intensive care for 48 hrs or more during the periods before and after intervention.
Interventions: A four-element ventilator-associated pneumonia prevention bundle, consisting of head-of-bed elevation, oral chlorhexidine gel, sedation holds, and a weaning protocol implemented as part of the Scottish Patient Safety Program using Institute of Health Care Improvement methods.
Measurements and Main Results: Compliance with head-of-bed elevation and chlorhexidine gel were 95%–100%; documented compliance with "wake and wean" elements was 70%, giving overall bundle compliance rates of 70%. Compared to the preintervention period, there was a significant reduction in ventilator-associated pneumonia in the postintervention period (32 cases per 1,000 ventilator days to 12 cases per 1,000 ventilator days; p < .001). Statistical process control charts showed the decrease was most marked after bundle implementation. Patient cohorts staying ≥6 and ≥14 days had greater reduction in ventilator-associated pneumonia acquisition and also had reduced antibiotic use (reduced by 1 and 3 days; p = .008/.007, respectively). Rates of methicillin-resistant Staphylococcus aureus acquisition also decreased (10% to 3.6%; p < .001).
Conclusions: Implementation of a ventilator-associated pneumonia prevention bundle was associated with a statistically significant reduction in ventilator-associated pneumonia, which had not been achieved with earlier ad hoc ventilator-associated pneumonia prevention guidelines in our unit. This occurred despite an inability to meet bundle compliance targets of 95% for all elements. Our data support the systematic approach to achieving high rates of process compliance and suggest systematic introduction can decrease both infection incidence and antibiotic use, especially for patients requiring longer duration of ventilation.
Introduction
The United States-based Institute for Healthcare Improvement (IHI) has championed the "bundle" approach to improving practice and emphasizes the need to achieve high overall compliance rates (preferably 95% compliance with all bundle elements). Implementing care bundles has been strongly advocated in ventilated intensive care patients, who are at risk for ventilator-associated pneumonia (VAP). VAP is associated with prolonged mechanical ventilation, intensive care unit (ICU) and hospital stay, greater illness costs, and possibly higher mortality. Several clinical interventions have been found to reduce the incidence of VAP, including elevation of the head of the bed, a daily sedation break, a daily trial of ventilator weaning, and topical oral chlorhexidine. The quality of evidence supporting the effectiveness of each intervention and the relative importance of each has been questioned. VAP incidence has become a quality indicator in many healthcare systems, leading to comparisons between ICU and, in the United States, may influence reimbursement rates by insurers. The "VAP prevention bundle" is a central component of most critical care patient safety programs.
An important issue relating to reporting VAP incidence is the method of reporting, because this requires a clinical suspicion and the recognition of clinical signs, which introduces the potential for reporting bias in unblinded trials and quality-improvement studies. Diagnostic method can also vary; patients may meet clinical criteria alone or have a microbiologically confirmed diagnosis, which can be made using various approaches that vary in sensitivity and specificity. A recent systematic review of care bundles for ventilated patients concluded that a lack of methodologic rigor in published studies meant it was not possible to conclude that the bundles were either clinically effective or cost-effective. Previous studies have universally reported VAP rates, but reporting of more patient-centered outcomes such as antibiotic use and length of stay have been more variable, thus making it more difficult to estimate the impact of any reduction in VAP on patients and ICUs.
This study set out to establish the effects of systematically implementing a VAP prevention bundle using IHI methodology on VAP rates, antibiotic use, and duration of ICU stay in an ICU with clinical and research interests in VAP and preexisting VAP prevention guidelines.
Setting
This study was conducted in an 18-bed, mixed medical–surgical ICU in a Scottish teaching hospital admitting >1000 patients per year, of whom ≈80% require support of at least two organs or invasive ventilation, or both, and 50% stay for 48 hrs or longer. We have been studying VAP since 2005, when we established independent ICU infection surveillance using the Hospitals in Europe Linked for Infection Control through Surveillance methodology. Screening, data collection and reporting was undertaken by a trained, dedicated, full-time nurse and diagnosis of VAP was made independently by the treating clinical team. Chest radiograph interpretation was undertaken "off-line" and by clinicians who were independent of the treating team. For VAP diagnosis, Hospitals in Europe Linked for Infection Control through Surveillance has a two-stage definition: first, clinically suspected VAP based on clinical criteria; and second, microbiologically confirmed VAP based on further investigations. The details of this process are shown in Figure 1. Antibiotic use was also recorded prospectively for all patients on a daily basis. We used statistical process control methodology to report rates and before implementing IHI methodology had made attempts to decrease VAP incidence by implementing unit guidelines and protocols including head-of-bed elevation, a nurse-led weaning protocol, and hourly clinical sedation scoring linked to a protocol. Intermittent audits of compliance with these interventions showed variable performance, but we were not recording these processes or feeding them back systematically. From 2005 to 2008, we were unable to decrease the incidence of clinically diagnosed VAP but we had shown that increasing the use of quantitative analysis of bronchoalveolar lavage fluid for microbiological diagnosis resulted in a decrease in the reported incidence of microbiologically confirmed VAP, which was explained by superior test specificity compared with analysis of tracheal aspirates.
(Enlarge Image)
Figure 1.
Diagnostic criteria for clinical and microbiologically confirmed ventilator-associated pneumonia (VAP) used in this study, based on the criteria set out by the Hospitals in Europe Linked for Infection Control through Surveillance project (17). BAL, bronchoalveolar lavage; PCR, polymerase chain reaction.
Methods and Bundle Description
This study used a before-and-after structure to analyze the effects of implementing a VAP prevention bundle. In 2007 National Health Service Scotland launched the Scottish Patient Safety Programme in collaboration with the IHI as the first nationwide patient safety program. One key aim was to reduce VAP incidence in Scottish ICUs, and a VAP prevention bundle was developed by the Scottish Intensive Care Society Audit Group for the Scottish Patient Safety Programme. The bundle has five elements: (1) a daily sedation hold; (2) daily trial of ventilator weaning for suitable patients; (3) head-up position; (4) chlorhexidine mouth care; and (5) subglottic secretion drainage using specialized endotracheal tubes. The only elements of the bundle that we had not attempted to implement already were chlorhexidine mouth care and subglottic suction endotracheal tubes, but we had not used the systematic approach to practice change promoted by the IHI for any of the bundle components and were unsure if the Scottish Patient Safety Programme would produce clinically important benefits.
We chose to implement four elements of the VAP prevention bundle, namely a daily sedation hold and trial of ventilator weaning for suitable patients (which combined were called "wake and wean"), head-up position, and chlorhexidine mouth care. We did not implement subglottic secretion drainage using specialized endotracheal tubes, because this required frequent endotracheal tube changes and most of our patients were admitted intubated from either the operating theater or the emergency department. The additional cost of specialized tubes meant this was also the most costly bundle element.
The wake and wean algorithm was based on that used in a recent randomized controlled trial. In our unit, the nurse at the bed space was required to assess the patient's suitability for a sedation break and weaning of ventilatory support every morning. Bundle introduction was accompanied by nurse education relating to wake and wean practice. Nurses were asked to follow a checklist each morning that included an assessment of whether they considered the patient suitable for a wake and wean trial and whether they performed the trial. This "process measure" was completed as a checklist on the patient chart each morning. Compliance was defined as whether the checklist was completed without individually assessing whether medical staff agreed with the decision made. When in doubt, the nurse was asked to consult medical staff. Unless there were clinical indications for supine nursing, all patients were placed in a position of at least 30 degrees head-up tilt. Topical buccal 1% chlorhexidine gel was prescribed four times per day for all ventilated patients unless the patients either were allergic to the gel or refused the gel (2% gel is unavailable in the United Kingdom).
Bundle elements were applied using Plan, Do, Study, Act cycles until a method of implementation was identified that maximized compliance. Methods used during implementation included nurse and medical champions, teaching materials, education sessions, bedside cues, changing the 24-hr observation charts, and feedback of compliance at meetings, by e-mail, and with posters. All modifications to improve compliance were led by bedside nursing and medical staff.
Compliance of all patients was audited weekly at unannounced and variable times by an independent, nonclinically trained audit clerk who audited charts from the previous day to minimize observation bias. Compliance was defined as clear documentation either confirming compliance or deciding to avoid a bundle element for clinical reasons. Failure to complete bundle documentation was coded as a lack of compliance. Throughout the study period, standard infection control precautions were in place and were not altered. These included removal of outdoor clothing/white coats by visiting clinicians, single-use aprons for each bedside, bedside alcohol gel, and access to sinks at each bed space for hand decontamination. Regular audit of adherence to these measures showed no major changes or trends throughout the period of study.
Outcome Measures and Analysis Plan
We were interested in whether the introduction of the Scottish Intensive Care Society Audit Group VAP bundle using the IHI approach would decrease the incidence of both clinically diagnosed and microbiologically confirmed VAP, and whether it changed antibiotic use. We also wanted to measure its effect on rates of methicillin-resistant Staphylococcus aureus (MRSA) acquisition, another key quality indicator. MRSA acquisition was defined as MRSA cultured from either surveillance swabs or samples taken during investigation of suspected sepsis in a patient who was MRSA-negative at ICU admission based on admission screening swabs or other pre-ICU cultures. All patients were screened at ICU discharge to enable acquisition status to be determined.
We chose duration of mechanical ventilation, ICU stay, and ICU mortality as important patient outcomes that also have important cost implications as secondary end points. We recognized that patients with longer ICU stays are at greater risk for VAP, so we reported outcomes for several patient subcohorts, namely all patients at risk for VAP (ICU stay ≥48 hrs), patients with ICU stay ≥6 days, and patients with ICU stay ≥14 days. Our ICU also reports standardized mortality ratios on an annual basis as part of the Scottish Intensive Care Society Audit Group national audit, which includes all patients admitted to the ICU and compares ultimate hospital mortality to the Acute Physiology and Chronic Health Evaluation II prediction model. We report trends in these over the periods covered by the study as an overall indicator of ICU performance.
Elements of the Scottish Intensive Care Society Audit Group VAP prevention bundle were introduced progressively between February and September 2008. We used our surveillance data from January 2005 to February 2008 as baseline data, from February 2008 to September 2008 as the "run-in" period, and from September 2008 to August 2009 as the "post-VAP prevention bundle implementation" period. We compared the periods before and after the intervention using conventional statistical methods for comparing.
Summary statistics using Z-test to compare proportions, Mann-Whitney U test to compare median values (Prism; Graphpad, LA Jolla, CA), and Poisson regression for incident densities (PASW version 18.0; IBM, Armonk, NY). Changes were also assessed using statistical process control charts using monthly data throughout the period of observation. Patient gender, age, and admission Acute Physiology and Chronic Health Evaluation II score were compared as potential confounders to a before-and-after evaluation. As a service evaluation, ethics committee approval was not required.
Source...