Abstract
The aim of the present study was to estimate the impact of methicillin resistance on mortality in ventilator-associated pneumonia (VAP) due to Staphylococcus aureus.
PubMed, Scopus and the bibliographies of the eligible studies were searched. The DerSimonian-Laird random effects model was used to determine the effect of methicillin resistance on mortality.
Eight articles were included. Crude in-hospital mortality was higher in patients with VAP due to methicillin-resistant S. aureus (MRSA) than in those with VAP due to methicillin-sensitive S. aureus (MSSA). This was also the case for crude intensive care unit mortality. However, three of the selected studies, which adjusted for potential confounding factors, including adequacy of empirical treatment and severity of illness, demonstrated no difference in in-hospital mortality between patients with MRSA and MSSA VAP. This was not the case for another eligible study that also made adjustment, but for confounders other than those shown above.
The limited available evidence seems to suggest that methicillin resistance is associated with death among persons acquiring Staphylococcus aureus ventilator-associated pneumonia. However, although supported by even more limited data, adjustment for risk factors suggests that this association may not be causal, but probably due to confounders, such as the adequacy of empirical treatment and severity of illness.
- Antimicrobial resistance
- critical illness
- healthcare-associated pneumonia
- morbidity
- nosocomial pneumonia
- survival
Most experts appear to agree that infections due to drug-resistant pathogens result in poorer clinical and economic outcomes as opposed to those due to sensitive strains 1–3. With regard to Staphylococcus aureus infections, this opinion relies mainly on the findings of a recent meta-analysis of studies published during 1980–2000 comparing the mortality of patients with methicillin-resistant S. aureus (MRSA) and methicillin-sensitive S. aureus (MSSA) bacteraemia 2. By accumulating data from 31 studies, this analysis revealed that mortality was higher in patients with MRSA bacteraemia than in those with MSSA bacteraemia, with a pooled odds ratio (OR) of 1.93 (95% confidence interval (CI) 1.54–2.42) 2. Interestingly, a number of more recent studies have confirmed these findings 4, 5.
However, methodological flaws in the studies included in the above meta-analysis 2 may limit its conclusions. Indeed, the authors of the analysis clearly stated that the individual studies did not provide specific data regarding the adequacy of the initially administered antimicrobial regimen, which probably confounds the relationship between methicillin resistance and fatality 6, 7. In addition, they considered as questionable the quality of adjustment for severity of illness of bacteraemic patients in the selected studies, mainly because of the lack of a validated method for this purpose. Thus scepticism remains as to whether methicillin resistance influences patient outcome, a scepticism further enhanced by the paucity, to date, of investigations establishing a definitive association between antimicrobial resistance and S. aureus virulence 8, 9.
The impact of methicillin resistance on mortality has also been addressed in patients with S. aureus ventilator-associated pneumonia (VAP). Studies exploring this issue have produced conflicting results, thereby further fuelling the controversy. VAP is the most common infection in the intensive care unit (ICU) setting, accounting for almost a third of total nosocomial infections 10, 11. S. aureus has been recognised as one of the most common pathogens isolated in patients with VAP, and the prevalence of methicillin-resistant strains is increasing 12. Thus elucidation of the impact of resistance on patient outcome may be essential in defining the prognosis of the individual patient with S. aureus VAP.
Thus attempts were made to exploit the available evidence in order to quantify the effect, if any, of methicillin resistance on mortality in patients with VAP due to S. aureus. A systematic review and analysis of the accumulated data were performed for this purpose.
METHODS
Data sources
A systematic literature search of PubMed and Scopus was conducted (during February 2007) to trace relevant studies. The search terms used were: (“ventilator-associated pneumonia” OR “ventilator associated pneumonia” OR “nosocomial pneumonia” AND “methicillin”. The search was also repeated using the term “oxacillin” instead of “methicillin”. In an attempt to increase completeness, the bibliographies of the articles initially retrieved were also hand-searched. Abstracts of conference proceedings were not sought and no restriction was set on publication date or language.
Study selection
The literature search was performed independently by two investigators (Z. Athanassa and I.I. Siempos). Both observational and interventional studies were considered eligible if they provided data on the mortality of patients with MRSA and MSSA VAP. Reports investigating nosocomial pneumonia were included in the present analysis only if ≥75% of the study population had received mechanical ventilation.
Data extraction
The studies were independently reviewed by two investigators (Z. Athanassa and I.I. Siempos). The following data were extracted: first author, year of publication, country, study design, type of ICU, study population, number of patients with S. aureus VAP, and types of culture used for confirmation of the S. aureus VAP diagnosis. In addition, data were collected on the number of patients, mortality and length of ICU stay after VAP onset in the MRSA and MSSA VAP groups. Factors thought to act as potential confounders (mean age, sex, underlying diseases, severity of illness on ICU admission, adequacy of empirical treatment, length of hospital stay and mechanical ventilation before VAP onset, presence of bacteraemia and presence of an outbreak of MRSA infection) were also recorded 2. Finally, information dealing with adjusted mortality, methods of adjustment and factors for which mortality had been adjusted were gathered.
Definitions
Outcomes of the systematic review
Crude in-hospital mortality was the main outcome of the present systematic review. Adjusted in-hospital mortality and crude ICU mortality, as well as length of ICU stay after VAP onset, in patients with MRSA and MSSA VAP also served as outcomes of the present study.
Ventilator-associated pneumonia
Pneumonia was defined by clinical, laboratory and imaging findings attributed by the authors of the individual selected articles to this infection. When the above findings occurred in patients who had received mechanical ventilation for ≥48 h, the pneumonia was regarded as ventilator-associated.
Data analysis and statistical methods
The number of MRSA (versus MSSA) VAP patients who died was divided by the total number of MRSA (versus MSSA) VAP patients. Pooled ORs and 95% CIs were calculated using a conservative DerSimonian–Laird random effects model. Heterogeneity among the results of the included studies was tested using the Chi-squared test; a p-value <0.10 was defined as denoting significance in the analysis of heterogeneity.
RESULTS
Selected studies
The selection process applied to capture the studies included in the present systematic review is depicted in figure 1⇓. The PubMed search yielded 246 potentially relevant reports; 239 of them were excluded for the reasons detailed in figure 1⇓. Two studies 13, 14 enrolled, in large proportion, the same study population (i.e. patients included in the PneumA trial, a multicentric randomised controlled trial comparing 8 versus 15 days of antimicrobial treatment for VAP 15); thus, one 13 of these two studies was omitted. Three additional potentially relevant reports were found by hand-searching the references of the articles initially retrieved. Two 16, 17 of these three studies were excluded because <75% of the patients enrolled had received mechanical ventilation 16 or because they did not provide specific data on MRSA VAP 17. Search employing the Scopus database did not reveal any additional reports fulfilling the present inclusion criteria. This was also the case for the search carried out replacing the term “methicillin” with “oxacillin”. Thus eight articles 14, 18–24 were included in the present systematic review.
Flow diagram detailing reviewed articles and exclusion. MRSA: methicillin-resistant Staphylococcus aureus; MSSA: methicillin-sensitive Staphylococcus aureus; VAP: ventilator-associated pneumonia.
Characteristics of the selected studies
The characteristics of the eight selected articles 14, 18–24, representing 2,814 patients with S. aureus VAP, are listed in table 1⇓. Their mean (range) cohort size was 154 patients (49–1,851). All were designed as prospective (four out of eight; 50%) 19, 22–24 or retrospective (50%) observational studies 14, 18, 20, 21. One 14 of the included studies was partly based on the 401 patients included in the previously mentioned PneumA trial 15.
Characteristics of the studies ofStaphylococcus aureus ventilator-associated pneumonia (VAP) included in the systematic review
The VAP diagnosis was assessed using clinical, laboratory and imaging findings in three 18, 21, 24 of the eight eligible reports. In contrast, in the remaining five studies 14, 19, 20, 22, 23, VAP diagnostic criteria included microbiological confirmation by qualitative or quantitative culture of respiratory specimens (i.e. protected brush or catheter specimen or bronchoalveolar lavage fluid) and/or blood (with the same microorganism isolated from blood and sputum specimens).
Study population
The characteristics of the patients of the studies included in the systematic review are depicted in table 2⇓. Compared with patients with MSSA VAP, patients with MRSA VAP were older in six studies 14, 18, 19, 21–23 and their ages did not differ in one 20. The other study 24 did not give relevant data. The MRSA and MSSA groups were similar regarding sex in five 14, 18, 19, 22, 23 of the eight eligible reports 14, 18–24 that provided such data.
Outcome data from the included systematic review studies: patients with ventilator-associated pneumonia (VAP) due to methicillin-resistant (MRSA) versus methicillin-sensitive Staphylococcus aureus (MSSA)
Regarding the incidence of underlying diseases (namely cardiac, respiratory and neurological failure, as well as malignancy and diabetes mellitus) in the patients with pneumonia, data were available for six 14, 18–20, 22, 23 of the eight eligible reports 14, 18–24. In two studies 18, 23, cardiac 18 and respiratory failure 23 were more common among patients with MRSA (compared with MSSA) VAP, whereas, in two other reports 19, 22, neurological failure was more prominent in patients with MSSA (compared with MRSA) VAP. Severity of illness on admission was assessed in five 14, 18–20, 22 of the eligible studies; no differences were shown between comparators. As opposed to patients with MSSA VAP, those with MRSA had a more prolonged duration of hospitalisation and mechanical ventilation in the included reports giving such data (table 2⇑).
Mortality
The outcome data of the studies included in the systematic review are presented in table 2⇑. Among 2,814 patients with S. aureus VAP, 757 (27%) patients had MRSA VAP. With respect to the follow-up period for mortality, one 21 of the selected studies did not provide relevant information, whereas another 19 reported on both in-hospital and ICU mortality. In the remaining articles, this period was extended up to 6 weeks after VAP onset 24 or ICU discharge 22, 23 or hospital discharge 14, 18, 20.
Crude in-hospital and ICU mortality
The crude in-hospital mortality was higher in patients with VAP due to MRSA as opposed to MSSA (OR 1.79 (95% CI 1.210–2.65); data from eight reports 14, 18–24; table 3⇓ and fig. 2⇓). This was also the case for crude ICU mortality (OR 2.49 (95% CI 1.54–4.06); data from three reports 19, 22, 23).
Forest plot comparing crude in-hospital mortality rates in studies of methicillin resistance versus sensitivity in patients with ventilator-associated pneumonia due to Staphylococcus aureus and the pooled analysis. Data are presented as odds ratio (OR; ▪ (size reflects weighting)) and 95% confidence interval (CI: horizontal bars). The centre of the diamond represents the pooled OR and its extremities the CI. …… : line of no difference. If OR >1, methicillin-sensitive S. aureus (MSSA) is favoured (i.e. mortality is lower); if OR <1, methicillin-resistant S. aureus (MRSA) is favoured. In testing for heterogeneity of the pooled results that contributed towards the overall mean result, Χ2 = 18.87 (7 degrees of freedom; p = 0.009) and I2 = 62.9%. In testing for overall effect, z = 2.91 (p = 0.004).
Characteristics of the studies ofStaphylococcus aureus ventilator-associated pneumonia included in the systematic review
Subgroup analysis was performed after exclusion of the study that provided the largest number of patients, and that did not provide certain essential clinical (such as mortality during the follow-up period) and microbiological data (i.e. types of culture, if any, used for confirmation of the VAP diagnosis) 21. Even after the exclusion of the above study, the finding of the main analysis did not change substantially (OR for crude in-hospital mortality 1.62 (95% CI 1.06–2.48), with data from seven reports 14, 18–20, 22–24).
Adjusted in-hospital mortality
In four 14, 19–21 of the eight reports included in the present review, mortality was adjusted for potential confounding factors, such as the adequacy of the empirical treatment 14, 19, severity of illness 14, 19, age 14, 19, 21, presence of bacteraemia 14 and duration of mechanical ventilation before 14 (table 2⇑). One 20 of these four reports did not give specific data regarding the confounders.
Of the four studies that were adjusted 14, 19, 20, 21, three 14, 19, 20 failed to reveal a difference in in-hospital mortality between patients with MRSA and MSSA VAP (table 2⇑). In contrast, in the remaining study 21, the mortality of patients with MRSA VAP remained higher than for those with MSSA VAP, even after adjustment (OR 2.62 (95% CI 1.69–4.02)). However, it should be noted that the latter study 21 differed from the above in that it was adjusted for factors other than the adequacy of the initial treatment and severity of illness.
Length of ICU stay after VAP onset
Data on the length of ICU stay after VAP onset were available for four 14, 18–20 of the studies included in the present review. In three studies 14, 19, 20, there was no difference regarding length of stay between the compared groups, whereas, in the remaining study 18, length of ICU stay after VAP onset was more prolonged in the MRSA than in the MSSA group (table 2⇑).
DISCUSSION
The findings of the present systematic review suggest that both the in-hospital and ICU mortality of patients with VAP due to S. aureus were higher in the presence than in the absence of methicillin resistance. Thus they are in line with those of a recent meta-analysis inferring that methicillin resistance is associated with increased mortality in patients with S. aureus bacteraemia 2.
However, the results of the current contribution contradict those of two clinical studies on the same issue that did not fulfil the present inclusion criteria 25, 26. The fact that these two studies did not establish any association between methicillin resistance and excess mortality may not be surprising. Indeed, in the first study 25, >50% of MSSA patients were treated with vancomycin (which was found to be associated with a higher mortality compared with cloxacillin 25); this could account for their increased mortality and, thereby, the lack of difference between the MRSA and MSSA groups regarding this outcome. It is noteworthy that several investigators have reported that patients with MSSA infections experience worse clinical outcomes when treated with vancomycin rather than a β-lactam 27–29. In the other study 26, several methodological issues (such as the enrolment of patients not receiving mechanical ventilation, as well as the absence of an accurate microbiological confirmation of the pneumonia diagnosis based on quantitative cultures) might have substantially influenced its findings. Likewise, three 18, 20, 24 of the eight articles included in the present review also demonstrate no relationship between methicillin resistance and mortality in patients with S. aureus VAP. This may also be explained by their small sample size 20 and the considerable proportion of enrolled patients with MSSA nosocomial pneumonia who were treated with vancomycin 20, as well as the missing data regarding the duration of follow-up 18 and microbiological confirmation of S. aureus VAP 18, 24.
The mortality of patients with S. aureus VAP varied widely among the individual selected studies, a fact presumably explained by several considerations. For instance, there were differences in the mean age of patients enrolled in the selected studies. Given that age is an independent risk factor for mortality in patients with S. aureus VAP 14, 21, it seems plausible that studies enrolling young patients show low mortality. In addition, data, when available, dealing with the adequacy of empirical treatment and the severity of illness revealed substantial differences among the selected studies. It is also noteworthy that evaluation of the observed differences in severity of illness among the eligible reports could not be accurately performed, due to application of different disease severity indices. Finally, the follow-up periods were not identical among the selected reports; it might be argued that studies with more extended follow-up periods may also report higher mortality.
Another interesting finding of the present contribution is the possible attenuation of the effect of methicillin resistance on mortality when adjusting for adequacy of empirical treatment and severity of illness. It should be emphasised that this is based on only two reports 14, 19 including 305 patients. Even so, this finding may lead to several considerations. The fact that MRSA VAP causes increased mortality, compared with MSSA VAP, does not necessarily prove a causative relationship between methicillin resistance and excess lethality. It has previously been demonstrated that, compared with patients with MSSA bacteraemia, those with MRSA bacteraemia are more likely to be older, have a terminal disease and/or a more debilitated physical condition, require prolonged hospitalisation and mechanical ventilation, and receive inappropriate empirical treatment; it is interesting that, after adjustment for these confounding factors, methicillin resistance was not independently associated with excess mortality 30, 31. Thus, it may be difficult to characterise MRSA as a more virulent pathogen than MSSA in terms of mortality, since its presence is associated with increased age 14, 19, 21, severity of illness 25 and inadequate treatment 6, all of which are independently associated with a worse prognosis.
In an attempt to explore the influence of numerous potential confounders on mortality, relevant information was collected from all of the selected studies. As depicted in table 2⇑, in six 14, 18, 19, 21–23 of the eligible reports, patients with MRSA and MSSA differed in terms of age. However, in three 14, 19, 21 of these six studies 14, 18, 19, 21–23, this difference was resolved by performing a multivariable analysis. Furthermore, the evaluation of underlying diseases, a factor potentially influencing mortality, was performed in only six 14, 18–20, 22, 23 of the eligible reports. In addition, the assessment of severity of illness on admission and adequacy of empirical treatment was performed in only five 14, 18, 19, 20, 22 and three 14, 19, 20 of the eligible studies, respectively. Conversely, duration of mechanical ventilation and length of stay before VAP onset were available for three 14, 22, 23 and four reports 19, 20, 22, 24; they were more prolonged in the MRSA than the MSSA groups. Parameters potentially influencing mortality other than the above, such as the presence of outbreak and polymicrobial status of VAP were assessed in four 14, 20, 22, 23 and five studies 14, 18, 19, 22, 23, respectively. These discrepancies in potential confounding factors presumably reflect substantial differences in patient population among eligible reports and cause difficulties in estimating the impact of methicillin resistance on mortality.
The present authors recognise that the current contribution has limitations. For example, the findings of the present systematic review may be influenced by the study that enrolled the largest number of patients with S. aureus VAP 21. However, the main findings did not change substantially after exclusion of the above report. In addition, unfortunately, no data regarding the number of patients with early- or late-onset VAP in the MRSA and MSSA groups were provided by the great majority of the eligible reports. Finally, the heterogeneity among the included reports with regard to the factors for which there was adjustment, as well as the fact that, in half of the studies, there was absolutely no adjustment for confounding factors inevitably limit the value of the present evaluation of the available literature. However, in an attempt to investigate the influence of several confounding factors on mortality, relevant information was carefully gathered from all of the eligible reports (table 2⇑); this probably permits the disclosure of possible associations between these factors and mortality.
In conclusion, despite the above mentioned limitations, the present authors believe that the current contribution is of clinical and health economic importance. The limited available and analysed evidence seems to suggest that methicillin resistance may be associated with excess mortality among persons who acquire Staphylococcus aureus ventilator-associated pneumonia, a finding that may differentiate the efforts to control methicillin-resistant Staphylococcus aureus. However, even more limited evidence exists suggesting that this association is not retained after adjustment for clinically significant potential confounding factors, and, thereby, that this association may not be causal but probably due to confounders, such as the adequacy of the empirical treatment and severity of illness.
Statement of interest
None declared.
- Received July 2, 2007.
- Accepted October 17, 2007.
- © ERS Journals Ltd
References
