Post-stroke infection: A systematic review and meta-analysis

Infection is a common complication in the acute phase after stroke. Reported infection rates after stroke vary considerably, ranging 5-65%. Differences in patient populations, study design and definition of infection may account for these large variations in post stroke infection rates [1]. However, a reliable pooled estimate of the infection rate in patients with stroke is lacking. Pneumonia is the most common post-stroke infection and been associated with a relative risk of 3.0 for mortality in a study including 14293 patients with stroke [2]. Consequently, new treatment strategies, i.e. preventive antibiotics, are currently under investigation [3]. In this systematic review and meta-analysis we calculated the pooled post-stroke infection rates, identified study and population characteristics associated with infection, and estimated the impact of pneumonia on outcome after stroke.

This meta-analysis shows that infections commonly complicate the acute phase after stroke. The pooled overall infection rate was 30% and pneumonia and urinary tract infections occurred each in 10% of patients. Previous studies showed a wide range of post-stroke infection rates, from 5%- 65% for infections, 1%-33% for pneumonia, and 2%-27% for urinary tract infection [1, 20]. This meta-analysis included a large number of patients and therefore provides a reliable estimate of infection after stroke.

We identified several study and population characteristics that were associated with infection rate. Given the prevalence of pneumonia and urinary tract infection in general wards in Dutch hospitals - 1.1% and 1.7% respectively - our findings confirm that infection rate in patients in the acute phase of stroke is high [21]. This increased vulnerability of patients in the acute phase of stroke for infections can be attributed to different factors.

First, infection was associated with study characteristics. Studies aimed on infection and those with consecutive enrolment were associated with higher infection rate. Possibly, these studies benefit of more rigorous detection of infection. Rate of infection was also higher in studies with a longer observation time. In a prospective observational study, most post stroke infections occurred within three days of hospital admission [1]. Nevertheless, 25% of infections occurred after these three days. Also, most urinary tract infections occur after 48 hours, i.e. most UTI's are hospital acquired, and therefore it seems logical that a longer observation time yielded a higher urinary tract infection rate in our study [17]. The absence of an association between observation time and pneumonia is not surprising. Pneumonia is mostly diagnosed within the first days following a stroke, both in studies performed on general wards as well as ICU's [1, 13, 22‐24].

Microbiologic data of patients with post-stroke pneumonia shows a pattern of mostly early onset nosocomial pneumonia, or a community acquired aspiration syndrome. Staphylococcus aureus and gram-negative bacteria such as Klebsiella pneumoniae, Pseudomonas aeruginosa, Escherichia coli or Enterobacter spp. were commonly identified; also Streptococcus species are occasionally found. Gram-negative bacteria and Staphylococcus aureus are known to cause pneumonia by aspiration of endogenous material from the colonized oropharynx [25, 26]. These pathogens are often seen in nosocomial infections [27]. Streptococcus species is still the most detected pathogen in community acquired pneumonia [28]. In stroke patients, it could be a cause of 'community acquired aspiration pneumonia', with aspiration occurring at the stroke ictus [29]. Often, no causative organism is detected in post-stroke pneumonia. Yield of cultures is usually not very high in patients with pneumonia; collection of material could be even more difficult in stroke patients due to neurologic deficit or lowered level of consciousness. Also, some cases of suspected pneumonia could in fact be a non-infectious aspiration pneumonitis, or the infection could be caused by anaerobic bacteria that require special culturing techniques. However, the role of anaerobic bacteria in the development of pneumonia is unclear [29, 30].

Second, infection rate was associated with the patients' clinical condition. Studies including patients with a higher stroke severity or lower levels of consciousness showed higher infection rates, in particular for pneumonia. This effect corresponds with previous studies that often report both characteristics as risk factors for pneumonia [14, 31, 32]. Risk for aspiration is increased in these patients due to the absence of protective reflexes, and this risk is related to the degree of consciousness impairment [33]. Most stroke related pneumonias are believed to result from dysphagia and subsequent aspiration of oropharyngeal material or gastric content. A systematic review by Martino et al showed that dysphagia occurs in 37- 78% of stroke patients and increases the risk for pneumonia 3-fold and 11-fold in patients with confirmed aspiration [34]. However, up to half of patients with post stroke pneumonia do not aspirate, which implies that also other mechanisms are involved, e.g., stroke-induced immunodepression which is discussed below [2].

We also found higher infection rates in ICU studies. Patients admitted to an ICU generally suffer from more severe strokes and also the frequency of invasive procedures is higher [15, 24, 32, 35]. The use of invasive procedures - i.e. urinary catheterization or mechanical ventilation in ICU patients - increase infection risk by facilitating the entry of a pathogen [17, 35].

We identified a higher age and female sex as risk factors for urinary tract infection. Both of these characteristics have previously been reported as risk factors for urinary tract infection [16, 17]. Studies frequently report urinary catheterization as an important risk factor; we could not investigate this association because this characteristic was mostly not reported [17]. Advanced age as risk factor for post-stroke infection has been reported previously [16].

In addition to above mentioned characteristics, acute stroke may lead to stroke-induced immunodepression, a systemic anti-inflammatory response that is related to susceptibility to infection [20, 36, 37]. This anti-inflammatory response was found in different clinical studies in acute stroke patients, and includes an excessive counter-inflammatory cytokine responses and impairments in cell-mediated immunity [38]. Certain features of this response - i.e., reduced lymphocyte count, delay in the recovery of T-lymphocyte loss - were more pronounced in patients developing post stroke infections [37, 39]. These results suggest that immunological changes could facilitate infection in acute stroke.

Outcome is affected by post-stroke infections, as shown in our review. Pneumonia and urinary tract infection both increase the risk for unfavorable outcome and pneumonia is associated with mortality with an OR of 3.62. Infections could affect outcome in several ways. Firstly, they lead to immobilization, general frailty and a delay in rehabilitation due to prolonged hospital stay [1, 40]. More importantly, immunological effects of infections could worsen outcome. Evidence from experimental studies suggests that infection promotes antigen presentation and autoimmunity against the brain [41]. However, the evidence on this topic is scarce and the exact pathofysiology remains to be investigated.

Our systematic review has limitations. First, results are limited by publication bias. Data was derived from randomized controlled trials, cohort studies and stroke registries; infection rate could differ in hospitals without stroke research or complication registries. Second, included studies were heterogeneous in definition of infection, which was based on clinical grounds and in some cases not described. A standardized definition for infection - as described by the Centers for Disease Control and Prevention [42] - is preferred since stricter criteria could permit identification of fewer infections. Next, not all relevant characteristics - for example use of antibiotics, differences in primary stroke care or use of a urinary catheter - could be evaluated, due to lack of data in included studies. Some of these characteristics have previously been described as a risk factor for infection and could have confounded our results. Thirdly, for some characteristics data was lacking in many studies. Surprisingly, no association was found between age or dysphagia and pneumonia rate, often described risk factors for pneumonia [24, 31, 32, 35, 43, 44]. Dysphagia was not reported in all studies, which was a limiting factor in the analysis. Lack of these significant associations could be due to few studies reporting adequate information on all these characteristics and therefore the analyses have low explanatory power. Finally, many of the characteristics act on patient level, but studies report aggregate data (i.e., mean age of study population) also reducing explanatory power. Due to these limitations - most importantly the heterogeneity in definition of infections - results of this review need to be interpreted with caution.

Our review shows the potential of strategies aiming to prevent infections in patients with stroke. Some of these strategies - i.e., prevention of aspiration and reduction of urinary catheterisation - are incorporated in stroke unit care, which reduces the risk of death after stroke through the prevention of infections in particular [45]. Infections can also be prevented by use of preventive antibiotic therapy, as shown in a recent meta-analysis. This meta-analysis did not establish a reduction in mortality, however, included studies were small and heterogeneous and functional outcome was not evaluated [12]. Also, limited data was reported on the effect on antibiotic resistance. Currently, the use of preventive antibiotics in stroke, and the effect of this therapy on antibiotic resistance, is investigated in a large randomized controlled trial with functional endpoint [3]. This trial will be able to establish whether preventive antibiotic treatment is an effective strategy to prevent infection and its adverse effect on outcome in patients in the acute phase of stroke.

This work was supported by the Netherlands Organisation for Health Research and Development (ZonMW): 171002302 and the Netherlands Heart Foundation (Hartstichting): 2009B095. DvdB is supported by grants from the Netherlands Organization for Health Research and Development (ZonMw; NWOVeni grant 2006 [916.76.023], NWO-Vidi grant 2010 [016.116.358]) and the Academic Medical Center (AMC Fellowship 2008).