Abstract
Severe sepsis is one of the most common causes of acute lung injury (ALI) and is associated with high mortality. The aim of the study was to see whether a protective strategy based approach with a plateau pressure <30 cmH2O was associated with lower mortality in septic patients with ALI in the Surviving Sepsis Campaign international database.
A retrospective analysis of an international multicentric database of 15,022 septic patients from 165 intensive care units was used.
Septic patients with ALI and mechanical ventilation (n=1,738) had more accompanying organ dysfunction and a higher mortality rate (48.3% versus 33.0%, p<0.001) than septic patients without ALI (n=13,284). In patients with ALI and mechanical ventilation, the use of inspiratory plateau pressures maintained at <30 cmH2O was associated with lower mortality by Chi-squared test (46.4% versus 55.1%, p<0.001) and by Kaplan–Meier and log-rank test (p<0.001). In a multivariable random-effects Cox regression, plateau pressure <30 cmH2O was significantly associated with lower mortality (hazard ratio 0.84, 95% CI 0.72–0.99; p=0.038).
ALI in sepsis was associated with higher mortality, especially when an inspiratory pressure-limited mechanical ventilation approach was not implemented.
Acute lung injury (ALI) is linked to multiple causes [1]. The highest incidence of ALI occurrs in patients with sepsis [2] and sepsis-associated ALI carries the highest mortality rates [3]. The exact incidence today has not been reported after the worldwide implementation of lung-protective strategies [4]. Death usually results from multisystem organ rather than lung failure alone and patients who develop sepsis-associated ALI have a worse outcome than those without sepsis [5].
Over the past two decades, several studies have reported better outcome in patients with ALI/ARDS, based on lung-protective ventilation using low tidal volumes and pressure limitation being one of the biggest advances in the application of mechanical ventilation [4, 6]. Yet, studies continue to demonstrate low adoption of these lung protective ventilation strategies. The Surviving Sepsis Campaign (SSC) developed a quality improvement programme for the bedside management of severely septic and septic shock patients that included achieving inspiratory plateau pressures maintained at <30 cmH2O for mechanically ventilated patients [7].
The aim of the present study was an epidemiological analysis of the impact of ALI in a large cohort of patients with sepsis, and the effect of mechanical ventilation if inspiratory plateau pressures maintained at <30 cmH2O were associated with better outcomes in patients with and without ALI.
MATERIAL AND METHODS
Study design
We conducted a retrospective analysis from a multicentric study of a cohort of 1,738 mechanically ventilated patients with ALI of a total of 15,022 septic patients from 165 intensive care units (ICUs). The analysis set was constructed from the subjects entered into the SSC database from its launch in January 2005 to March 2008. The a priori data analysis plan limited inclusion to sites with ≥20 subjects and ≥3 months of subject enrolment. The analysis presented here was limited to the first 2 yrs of subjects at each site. Hospitals contributed data for a mean duration of 15.6 months (median 14 months). Subjects were characterised by baseline severe sepsis information: sepsis admission source (emergency department, ICU or ward); site of infection (pulmonary, urinary tract, abdominal, central nervous system, skin, bone, wound, catheter, cardiac, device or other); and acute organ dysfunction (cardiovascular, pulmonary, renal, hepatic or haematological). Data were entered into the SSC database locally at individual hospitals into pre-established, unmodifiable fields documenting performance data and the time of specific actions and findings. Data on the local database contained private health information (PHI) that enabled individual sites to audit and review local practice and compliance as well as provide feedback to clinicians involved in the initiative. Data stripped of PHI were submitted every 30 days to the secure master SSC server at the Society of Critical Care Medicine (Mount Prospect, IL, USA). Further details of this observational study can be found elsewhere [7].
Data collection and definitions
To be enrolled, a subject had to have a suspected site of infection, and meet two or more systemic inflammatory response syndrome criteria [8] and one or more organ dysfunction criteria. Clinical and demographic characteristics and time of presentation with severe sepsis criteria were collected for analysis of time-based measures. Patient clinical characteristics, compliance with the quality indicators, use of a protective mechanical ventilation strategy based approach (defined as plateau pressure <30 cmH2O) and hospital mortality was analysed in patients with and without ALI.
ALI was defined as an arterial oxygen tension (Pa,O2)/inspiratory oxygen fraction (FI,O2) ratio <300 mmHg with bilateral infiltrates plus mechanical ventilation. The worst Pa,O2/FI,O2 at the time of diagnosis of severe sepsis, entry into the database and median plateau pressure during the first 24 h after ICU admission were recorded. When a patient was included in the study and mechanically ventilated, the inspiratory plateau pressures maintained at <30 cmH2O had to be accomplished as soon as possible and scored over the first 24 h. If a patient was not ventilated at the time of admission but was eventually ventilated, the inspiratory plateau pressure was maintained at <30 cmH2O in the first 24 h. Quality indicators were divided into two sets of performance measures: a resuscitation bundle to be accomplished within 6 h of presentation with severe sepsis and/or septic shock, and a second set, the management bundle, to be accomplished within 24 h.
Statistical analysis
Summary statistics of patient clinical characteristics are presented as percentages for categorical variables and medians with their associated interquartile ranges for continuous variables. Pearson’s Chi-squared test was used to test differences in categorical variables and the Wilcoxon rank-sum was used to test differences for continuous variables. Pearson’s Chi-squared test was used to test differences between the groups for the 12 key quality indicators. The log-rank test was used to compare hospital survival curves between patients who maintained a mechanically ventilated plateau pressure of <30 cmH2O and those who did not. A random-effects Cox proportional hazard regression (shared gamma frailty model) was used to estimate the hazard of death for those with a plateau pressure <30 cmH2O compared with those whose pressure was ≥30 cmH2O for patients with ALI who were mechanically ventilated. Random-effects regression is used due to the hierarchical nature of the data. Subjects were nested within specific ICUs; thus, the within-ICU and between-ICU variability were used to estimate the standard errors that were used to test the model coefficients. Both the unadjusted results and the adjusted results are presented. A risk factor modelling approach was used to determine which covariates were to be included in the model. Only covariates that acted either as a confounder or as an effect modifier were included. The following covariates were evaluated: region (North America, Europe or South America); sepsis admission source (emergency department, ward or ICU); infection location (urinary tract infection, abdominal, meningitis, skin, bone, wound, catheter, endocarditis, device or other); broad-spectrum antibiotic treatment; fluids; vasopressor use; drotrecogin alfa administration within 24 h; and steroids within 24 h. The method of fractional polynomials was used to determine whether the continuous variables were linear in the log hazard. The proportional hazard assumption was checked in the final Cox model. All analyses were performed using Stata 10.1 (Stata Corporation, College Station, TX, USA).
RESULTS
A total of 15,022 septic patients were included in the SSC database. From these, 7,877 were mechanically ventilated and ALI was present in 1,738 (22.1%) patients.
In patients who presented with ALI and mechanical ventilation (ALI/MV), the most frequent source of infection was pneumonia; episodes of sepsis were more frequently identified in patients admitted in the ICU than the emergency department. Patients from the ICU suffered a significantly higher number of organ failures than patients from the ward and emergency department (mean±sd 2.02±1.04, 1.99±1.04 and 1.89±0.97, respectively; p<0.001). Table 1 shows that patients who developed ALI/MV presented with a subsequent organ dysfunction to sepsis more frequently when compared with the rest of the patients. Patients who developed ALI/MV had a significantly higher mortality than those who did not (48.3% versus 33.0%, p<0.001) (fig. 1). When only the 7,877 patients under mechanical ventilation were compared, mortality remained different in patients with or without ALI (48.3% versus 45.7%, p=0.054).
a) Hospital mortality for patients with acute lung injury (ALI) and mechanical ventilation versus the rest of the cohort. b) Hospital mortality based on the effect of the protective strategy in mechanically ventilated ALI patients. Patients under mechanical ventilation, n=15,022; patients with ALI, n=1,738. ***: p<0.001.
Table 2 illustrates that of the 1,738 patients with ALI/MV, 390 (22.4%) patients did not receive a protective mechanical ventilation strategy based approach (i.e. their plateau pressure was ≥30 cmH2O). Patients with a nonprotective mechanical ventilation strategy had a higher mortality (55.1% versus 46.4%, p=0.0021) (fig. 1). No significant differences were observed between patients in term of severity manifested by the number of organ failures mean (2.92 versus 2.93 for plateau pressure ≥30 and <30 cmH2O, respectively; p=0.910).
Tables 3 and 4 present the description of SSC quality indicators of the mechanically ventilated patients with and without ALI. Median inspiratory plateau pressure <30 cmH2O over the first 24 h after presentation was significantly associated with survival in the ALI and non-ALI populations (p=0.010 and p<0.001, respectively).
The effect of a protected strategy (plateau pressure <30 cmH2O) in mechanical ventilation was assessed over time from ICU admission. A survival benefit trend was observed using a Kaplan–Meier hospital survival curve (fig. 2) with a log-rank p-value of 0.075. When the results were adjusted for confounding factors (region, sepsis admission source, drotrecogin alfa administered within 24 h, vasopressor use and site quarter) using a random-effects Cox proportional hazard regression model, the survival benefit of protective mechanical ventilation strategy was significant (hazard ratio (HR) 0.84, 95% CI 0.72–0.99; p=0.038) (table 5). In addition, the same beneficial effect was noted in a cohort of patients who were mechanically ventilated but did not have ALI (HR 0.77, 95% CI 0.70–0.85; p<0.001) (table 6).
Kaplan–Meier hospital survival analysis for acute lung injury septic patients with or without a protective strategy in mechanical ventilation (censored to 120 days). ICU: intensive care unit.
DISCUSSION
The main finding of the present study was that the presence of ALI was associated with an increased mortality in mechanically ventilated septic patients whereas a protective strategy based approach using a plateau pressure <30 cmH2O was associated with an increased survival not only in patients with ALI but also in those who were mechanically ventilated but did not have ALI.
The Severe Sepsis Bundles based on the recommendations published by the SSC in 2004 [9] and consistent with the revised publication in 2008 [10] were designed in a manageable format for use at most institutions to create customised protocols and pathways specific to their institutions. The implementation of SCC guideline recommendations, as evidenced by improvement in compliance with bundle targets, in patients with severe sepsis/septic shock has been shown to be associated with improved outcomes [7]. The implementation of the severe sepsis performance improvement programme has been extensively reported and targeted to assure aggressive haemodynamic resuscitation in other settings and countries [11, 12]. Murphy et al. [13] reported from 212 patients with ALI complicating septic shock that early and late conservative late fluid management was associated with better patient outcomes, and Sakr et al. [14] reported from 3,147 patients included in the SOAP (Sepsis Occurrence in Acutely Ill Patients) study that high tidal volumes and positive fluid balance were associated with a worse outcome from ALI/acute respiratory distress syndrome (ARDS). However, outcome data on protective lung strategies based on limiting inspiratory plateau pressures in severe sepsis/septic shock patients are still lacking.
Lung-protective ventilation, a strategy that targets lower tidal volumes and limits plateau pressure to <30 cmH2O, may be applied with either volume- or pressure-controlled ventilation in order to avoid acute parenchymal lung injury and a release of cytokines into alveoli and the systemic circulation [15] that may contribute to multiple organ dysfunction. Lung-protective ventilation strategies attempt to avoid these consequences by limiting peak lung distension and preventing end-expiratory collapse [16]. An important question might be to determine whether the severity of the respiratory failure is related to a lower compliance that results in higher plateau pressures. In some clinical settings, it is not clear whether a higher mortality might be due to a higher plateau pressure or a higher severity, or based on the different clinical management provided. In the present study, as a parameter for disease severity on admission, no differences were found in patients with or without a plateau pressure <30 cmH2O were observed in the number of organ dysfunction. In addition, lung-protective ventilation strategies have been recommended in order to prevent tidal alveolar collapse and over-distension in patients with ALI/ARDS, and are the only clinical interventions to demonstrate a mortality benefit in published studies [4, 17, 18]. In addition, lung-protective strategy represents a physiologically sound and proven ventilation mode in order to avoid ventilator-induced lung injury (VILI) [19]. VILI injury caused by mechanical ventilation can lead to volutrauma, barotrauma and atelectrauma [20]. All of these mechanisms may initiate biophysical and biochemical injuries that increase alveolar–capillary permeability and a release of inflammatory mediators, and has been implicated in the concept of “biotrauma” [21]. In patients with sepsis and ALI, the release of mediators into the systemic circulation with subsequent systemic inflammation may contribute to the multiple organ failure and subsequent death of some patients [22, 23]. Lung-protective ventilation strategies have been associated with a more rapid attenuation of the inflammatory response in patients with ALI [24]. Lung-protective ventilation strategies might, in part, attenuate this overwhelming systemic inflammatory response syndrome triggered by a highly virulent pathogen. In addition, between 37% and 40% of the patients received protective ventilation while in the ICU if their sepsis was discovered in the emergency department or the ward. If the patient’s sepsis was discovered while in the ICU, only 22% received protective ventilation. We suspect the reason for this is based on the nature of the patient admitted to each setting. Patients from ICU suffered a significantly higher number of organ failure than patients from the ward and emergency department. Based on the data extracted from the results of the SCC study [7], the mortality of the patients that developed sepsis in the emergency department was 27.6% compared with 41.3% in the ICU. The multivariable mortality prediction model also demonstrated that the ICU, compared with the emergency department as the admission source, was independently associated with a worse outcome (OR 1.37, 95% CI 1.27–1.48; p<0.0001).
An important point to consider based on our results is that patients mechanically ventilated patients without ALI had also a better outcome when a protective mechanical ventilation strategy based approach using a plateau pressure <30 cmH2O was implemented. Gajic et al. [25] found that from 332 patients who did not have acute lung injury from the outset that the main risk factors associated with the development of ALI were the use of a large tidal volume. Along the same line, Determann et al. [26] performed a randomised controlled unblinded preventive trial comparing mechanical ventilation with conventional versus lower tidal volume in critically ill patients without ALI at the onset of mechanical ventilation. Baseline plasma interleukin (IL)-6 levels were comparable in both study groups (median 50 (interquartile range 21–122) versus 51 (interquartile range 20–182) ng·mL−1 in the conventional and lower tidal volume groups, respectively; p=0.74). In the conventional tidal volume group, plasma IL-6 levels decreased after 4 days (21 (9–99) ng·mL−1), but the decrease over time was more pronounced in the lower tidal volume group (11 (5–20) ng·mL−1) (p=0.01). Interestingly, the trial was stopped prematurely for safety reasons because the development of lung injury was higher in the conventional tidal volume group as compared with the lower tidal volume group (13.5% versus 2.6%, p=0.01).
This study has several strengths. First, the analysis was conducted in a large number of patients. In addition, the cohort comprised a homogenous population with severe sepsis. Secondly, the patients included in the analysis represented a population that fairly represents a systematic but realistic approach that differs from randomised clinical trials with a probable selection bias [24, 26, 27]. Thirdly, the present study has assessed not only the impact of an inspiratory pressure-limited approach to mechanical ventilation in septic patients alone but also the beneficial effect of management with the compliance with applicable elements of sepsis management bundles.
The present study has several potential limitations that should be addressed. The first is inherent in its observational design. Patients included in the present study were based on a voluntary, self-selected (by institution) basis and not randomised; therefore, the possibility of unforeseen and unmeasured biases that could affect the results are possible. Secondly, some of the patients might be not accurately included in the ALI group since only a clinical assessment was used to determine the presence of noncardiac pulmonary oedema and other characteristics, such as the number of lung quadrants involved on chest radiography, were not recorded. Strengthening this clinical assessment was the exclusion of cardiac postoperative, COPD and pre-existing lung disease patients in order to improve accuracy of diagnosis. Thirdly, several confounding factors, potentially justifying the use of higher tidal volume, were not recorded (and thus not taken into account). Severe acidosis or associated brain injury may require higher tidal volume; increased abdominal pressure may lead to increased plateau pressure for a given tidal volume. Improved understanding of the mechanisms underlying VILI and barotrauma makes it imperative to control higher tidal volume in order to prevent alveolar overdistension. There still exits a reasonable controversy regarding the optimal tidal volume for ventilation and the best parameter to monitor its effect (i.e. plateau pressure). Current practice involves tidal volumes that are lower than those used in the past. The use of low tidal volume and limiting plateau pressure might represent “two faces of the same coin”. Fourthly, because the level of positive end-expiratory pressure (PEEP) was not recorded, the distension pressure was unknown. However, a recent meta-analysis [28] of trials comparing higher versus lower levels of PEEP in adults with ARDS revealed that a significant statistical difference in hospital mortality was not observed between groups. However, higher levels were associated with improved survival among the subgroup of patients with ARDS, the answer in patients with ALI is still unknown. Finally, in the present study and based on the epidemiological approach, several factors influencing mechanical ventilation setting and outcome have not been considered (the period of time during which the plateau pressure was >30 cmH2O, modalities of mechanical ventilation, the use of neuromuscular blockade, protocols of sedation, etc.).
In summary, the present study shows that the development of ALI in sepsis was associated with higher mortality. A protective mechanical ventilation strategy based approach has previously demonstrated a better outcome in heterogeneous populations of patients affected by ALI/ARDS. The implementation of inspiratory pressure-limited mechanical ventilation approach strategies in patients with severe sepsis with or without ALI was associated with improved survival. Our data suggest that this strategy should be strongly incorporated in performance metrics for management of septic patients with or without ALI.
Footnotes
Statement of Interest
None declared.
- Received December 16, 2011.
- Accepted April 2, 2012.
- ©ERS 2013
REFERENCES
