Prognostic implications of aspiration pneumonia in patients with community acquired pneumonia: A systematic review with meta-analysis

Aspiration pneumonia is thought to be associated with a poor outcome in patients with community acquired pneumonia (CAP). However, there has been no systematic review regarding the impact of aspiration pneumonia on the outcomes in patients with CAP. This review was conducted using the MOOSE guidelines: Patients: patients defined CAP. Exposure: aspiration pneumonia defined as pneumonia in patients who have aspiration risk. Comparison: confirmed pneumonia in patients who were not considered to be at high risk for oral aspiration. Outcomes: mortality, hospital readmission or recurrent pneumonia. Three investigators independently identified published cohort studies from PubMed, CENTRAL database, and EMBASE. Nineteen studies were included for this systematic review. Aspiration pneumonia increased in-hospital mortality (relative risk, 3.62; 95% CI, 2.65–4.96; P < 0.001, seven studies) and 30-day mortality (3.57; 2.18–5.86; P < 0.001, five studies). In contrast, aspiration pneumonia was associated with decreased ICU mortality (relative risk, 0.40; 95% CI, 0.26–0.60; P < 0.00001, four studies). Although there are insufficient data to perform a meta-analysis on long-term mortality, recurrent pneumonia, and hospital readmission, the few reported studies suggest that aspiration pneumonia is also associated with these poor outcomes. In conclusion, aspiration pneumonia was associated with both higher in-hospital and 30-day mortality in patients with CAP outside ICU settings.

Search criteria. We identified published cohort studies written in English from the PubMed database using the search terms: "community acquired pneumonia [All Fields] OR healthcare associated pneumonia [All Fields] AND aspiration [All Fields]", from the Cochrane Central Register of Controlled Trials (CENTRAL) database using the search terms: "community acquired pneumonia AND aspiration" and "healthcare associated pneumonia AND aspiration", and from the EMBASE using the search terms: "community acquired pneumonia AND aspiration" or "healthcare associated pneumonia AND aspiration" (accessed on August 31, 2016). Studies published only in abstract form were excluded because the methods and results could not be fully assessed. Full texts of articles were further evaluated by three investigators (KK, TA and JK). Data extraction. We extracted the following information from included studies: study design, sample size, inclusion and exclusion criteria, subject demographics, prevalence rate of aspiration pneumonia, type of outcome, type of statistical analysis and other significant predictive factors for each outcome.
Assessing risk of bias. The risk of bias in the included studies was assessed according to the recommendations outlined in the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0. and MOOSE guidelines for the following items: selection, performance, detection, attrition, and publication bias 24 . Each study included in this review was assessed for quality as good, moderate, or poor based on biases using the modified Hayden's criteria 30 . Disagreements among the investigators were resolved by review of the assessments to reach consensus.

Data analysis.
We conducted meta-analysis by the outcomes of studies as follows: in-hospital mortality, and 30-day mortality. ICU mortality, long-term mortality, readmission and recurrence, and treatment failure. Outcomes were pooled using Mantel-Haenszel risk ratios, and the precision of the estimates was expressed as the 95% confidence interval (CI). Statistical heterogeneity was assessed using the Higgins I 2 tests. A random-effects model was used when significant heterogeneity was found. Publication bias was assessed by examination of funnel plots 31 . Statistical significance was defined by a P value < 0.05 for all analyses. The meta-analysis was performed with the Review Manager ver. 5.3 software program (The Nordic Cochrane Centre, The Cochrane Collaboration).
In-hospital mortality. Seven studies (69,129 subjects) evaluated in-hospital mortality outside of the ICU 57,59,63,64,[69][70][71] . Mortality in patients with aspiration was significantly higher than that in non-aspiration pneumonia (Table 1). Meta-analysis indicated that aspiration risk increased in-hospital mortality (relative risk, 3.62; 95% CI, 2.65-4.96; P < 0.00001; I 2 = 86%) as shown in Fig. 2. Most of the included studies individually assessed the impact of aspiration pneumonia using multivariate analysis ( Table 1). Three of these studies found that aspiration pneumonia independently increased in-hospital mortality after multivariate adjustments. Hayashi et al.   65 . In studies that showed that aspiration pneumonia was not associated with 30-day mortality in multivariate analysis 58 ICU mortality. Five studies including 1,644 subjects analyzed ICU mortality 67,[72][73][74] . These studies were all published from the same research group in France. Paradoxically, the mortality in subjects with aspiration pneumonia was significantly lower than that in those with non-aspiration pneumonia in two of these studies (Table 3), and aspiration pneumonia was associated with decreased ICU mortality (relative risk, 0.40; 95% CI, 0.26-0.60; P < 0.00001) (Fig. 4). One of these studies showed that aspiration risk was a better prognostic factor for ICU survival compared to non-aspiration pneumonia in the multivariate analysis 72 . This study also identified other prognostic factors for ICU mortality including ineffective antimicrobial therapy (prognostic score resulted from validation cohort, + 1.5), immunosuppression (+ 1.38), and higher organ system failure score (OSFS) (+ 0.64) 75 .

90-day mortality and one year mortality.
One study evaluated 90-day mortality 68 , and two assessed one year mortality after discharge 58,69 . Aspiration pneumonia increased one year mortality (40.4% in aspiration pneumonia vs 22.1% in non-aspiration pneumonia, p < 0.001) 69 . The other studies showed that aspiration pneumonia was significantly associated with 90-day mortality (HR 3.09; 95% CI 1.90-5.03) 68 (Table 4). Two studies showed that aspiration risk was associated with increased readmission rate after multivariate analysis, and 2 studies identified that aspiration risk was associated with a higher rate of recurrent pneumonia 58 Treatment failure. One research group analyzed the association between aspiration risk and treatment failure; defined as fever for more than 3 days with clinical deterioration necessitating a change in initial empiric antibiotic therapy, the occurrence of a severe side effect, or death occurring after at least 48 h of antibiotic treatment 26 . The rate of treatment failure was higher in subjects with aspiration risk than that in non-aspiration pneumonia ( Publication bias. There appeared to be funnel plot asymmetry for in-hospital mortality (Fig. 5) suggesting the possibility of publication bias. Due to small number of studies included in each meta-analysis, Sterne's test 31 was not appropriate for detecting funnel plot asymmetry.

Discussion
This systematic review suggests that aspiration risk is associated with greater in-hospital and 30-day mortality in subjects with CAP except, perhaps, in the ICU setting. Although there are insufficient data to perform a meta-analysis on long-term mortality, recurrent pneumonia, and hospital readmission, the few reported studies suggest that aspiration pneumonia is also associated with these outcomes. Although aspiration pneumonia was significantly associated with in-hospital mortality when all studies were combined, multivariate analysis in individual studies suggested greater variability (Table 1). For example, Hayashi et al. reported that aspiration pneumonia did not remain as a significant risk factor for in-hospital mortality after multivariate analysis 63 , but that CURB-65 and ECOG PS increased in-hospital mortality in the same analysis. The CURB-65 includes confusion as a scoring item and the ECOG PS includes weakness as a severity of disability so these scoring systems may, in part, encompass aspiration risk factors. The Fujiki study reported the highest mortality of aspiration pneumonia at 59.1% with an especially strong association with in-hospital mortality (OR 49.9; 95% CI 6.23-398.94) 59 . Differences in study size, study population, and in the definition of what constitutes aspiration pneumonia may account for these markedly divergent results.
The risk ratio for 30-day mortality (3.57; 95% CI, 2.18-5.86) was similar to that for in-hospital mortality (4.67, 95% CI, 2.59-8.41) most likely because these outcomes are similar. Fernandez et al. focused on patients aged over 80 years and in these subjects, gram-negative pneumonia was the main risk factors for 30-day mortality 60 . HCAP patients, many of whom are elderly, may be at greater risk of death from aspiration pneumonia, but only one such study has been reported. In that study, aspiration pneumonia in subjects with HCAP was   associated with an increased 30-day mortality compared to those with pneumonia but no aspiration (14.8% vs 4.3%, p = 0.025) 32 .
Paradoxically, in the studies that analyzed mortality in the ICU, aspiration risk was associated with a lower risk of death 67,[72][73][74] . It was hypothesized that the younger study population, the use of intravascular volume repletion, the earlier appreciation and treatment of pneumonia in the ICU and close monitoring might improve prognosis 74 . It is also possible that patients with pneumonia who are admitted to the ICU are, on the whole, sicker and with comorbidities that might minimize any influence of aspiration on outcomes.
Aspiration risk was associated with all cause readmission, perhaps because many of these subjects were debilitated and had comorbidities 58,66 . In subjects with weakness or neurologic dysfunction (e.g. stroke), dysphagia and aspiration often continues even with careful medical management. In general, recurrent pneumonia within 3 to 5 years of an episode of CAP occurs in 9 to 12% of subjects with a median time to recurrence of 123 to 317 days and mortality ranging from 4 to 10% 76 . While some risk factors for aspiration pneumonia result from the natural course of aging 77 , a multidisciplinary approach might reduce the risk of aspiration pneumonia. El Solh et al. reviewed ways to prevent aspiration pneumonia and concluded that few data were available to guide an evidence-based approach to the prevention using drugs such as angiotensin-converting enzyme inhibitors or capsaicin 78 . The evidence relating to non-pharmacologic approaches; swallowing rehabilitation, thickening feeds, oral hygiene, gastroesophageal reflux management, and a head-up position are also limited, but the combination approach using pharmacologic and non-pharmacologic methods may be of value in high risk patients 79,80 .
To summarize our results, aspiration risk is associated with in-hospital and 30 day mortality outside the ICU, with long term mortality, all cause readmission, and recurrent pneumonia. Most of the studies demonstrated that aspiration pneumonia is an independent risk factor for these outcomes. There are several limitations to interpreting these data. All studies were observational and most are retrospective. This has the potential to introduce selection, measurement, and possibly publication bias. Due to the small number of publications for each outcome, we could not conduct Stem's test or meta-regression analysis for confirm funnel plot asymmetry as a measure of publication bias, however, visual examination of the funnel plots suggested the possibility of bias for in-hospital   Table 4. All cause readmission and recurrent pneumonia. HR; hazard ratio, ND; not described, NS; not significant, OR; odds ratio. ***good, **moderate or *poor quality assessed based on biases using the modified Hayden's criteria. # Multiple outcome study. mortality but not for other outcomes. Although we have provided specific ad hoc definitions for pneumonia and for aspiration risk; these are still somewhat subjective making case ascertainment challenging. If uniform criteria to establish aspiration risk can be developed and accepted for future studies, this will enable well controlled and appropriately powered studies to determine if interventions that can decrease aspiration risk, will also affect morbidity and mortality in this population.