Early mortality in patients with community-acquired pneumonia: causes and risk factors

Community-acquired pneumonia (CAP) continues to be a major health problem worldwide 1–3. Despite more accurate aetiological diagnosis, effective antibiotic therapy and advances in supportive care, the morbidity and mortality rates associated with this infection remain high. Recent studies report complications in 15–50% of hospitalised patients and overall mortality rates that range from ∼10% for patients treated in a hospital setting to >30% for patients treated in an intensive care unit 4–7. According to these studies, most deaths occurring within 30 days of presentation appear to be pneumonia related. A substantial number of identifiable risk factors may influence mortality, and some of the factors associated with mortality within the first days may differ from those associated with mortality occurring later.

In this setting, information regarding the causes and factors related to mortality within the first 48 h of pneumonia are scarce and there is no clear consensus on whether this very early mortality is modifiable by medical intervention 8–10.

The aim of the present study was to determine the frequency, causes and factors associated with early mortality in a large prospective cohort of hospitalised patients with CAP.

MATERIALS AND METHODS

RESULTS

A total of 2,457 hospitalised patients with CAP were included in the present study. Out of these, 57 (2.3%) were early deaths (≤48 h), 131 (5.4%) were late deaths and 2,269 (92.3%) were survivors. Overall mortality (<30 days) was 7.7% (188 patients). Demographic characteristics and the main clinical features of patient groups are compared in table 1⇓. Early deaths were older and more often classified as having a high-risk pneumonia, compared with the remaining patients. Altered mental status, renal failure, tachycardia, increased respiratory rate, high fever, multilobar infiltrates, respiratory failure and shock were more frequently found at baseline in early deaths. Late deaths had greater comorbidity than early deaths, mainly chronic heart disease and chronic renal disease. Conversely, shock at admission and bacteraemic pneumonia was more frequent in early deaths.

Causes of early mortality are shown in table 2⇓. Acute respiratory failure secondary to pneumonia and multiorgan failure associated with septic shock were the most frequent.

As shown in table 3⇓, S. pneumoniae was the most frequently identified pathogen; there were no differences between groups. Bacteraemic pneumococcal pneumonia was significantly more frequent in early and late deaths. Out of 324 S. pneumoniae strains isolated from the 718 patients with pneumococcal pneumonia, no significant differences were found in the rates of antibiotic resistance between groups (early deaths versus late deaths versus survivors versus all (late deaths and survivors)): penicillin (Minimum Inhibitory Concentration (MIC) ≥4 μg·mL⁻¹; 20.0% versus 8.7% (p = 0.738), versus 8.1% (p = 0.448), and versus 8.3% (p = 0.521)), cefotaxime/ceftriaxone (MIC ≥2 μg·mL⁻¹; 0% versus 0% (p = 1), versus 6.3% (p = 0.563), and versus 6.4% (p = 0.586), erythromycin (MIC ≥1 μg·mL⁻¹; 37.5% versus 15.8% (p = 0.464), versus 13.7% (p = 0.058), and versus 13.4 (p = 0.088)), and ciprofloxacin (MIC ≥4 μg·mL⁻¹; 0% versus 0% (p = 1), versus 1.9% (p = 1), and versus 1.75% (p = 1)). Pneumonia due to Gram-negative bacilli was significantly more frequent in patients who died (early and late deaths), especially in the case of Pseudomonas aeruginosa pneumonia. No significant differences were observed regarding the frequency of P. aeruginosa between nursing home residents and the remaining patients (1.9 versus 0.8%; p = 0.361).

As shown in table 4⇓, most patients were initially treated with a single antimicrobial agent. Concordance of antibiotic therapy could be determined in 38 out of 57 early deaths, 85 out of 131 late deaths, and 1,223 out of 2,269 survivors. Overall, early deaths received discordant antibiotic therapy more frequently than all others. Discordant empirical antibiotic therapy in early deaths was mainly due to lack of coverage against P. aeruginosa infection (five out of six patients; 83%). All these five patients were aged >70 yrs and were classified into the Pneumonia Severity Index (PSI) risk class V; three of them had bacteraemia and septic shock and two had chronic obstructive pulmonary disease (COPD), but none had bronchiectasis or were receiving chronic steroid therapy.

Table 5⇓ shows factors associated with early mortality by multivariate analyses. After adjustment, factors associated with early death (≤48 h) were increased age, altered mental status, multilobar pneumonia, shock at admission, pneumococcal bacteraemia and discordant empiric antibiotic therapy. No significant differences were found in these factors associated with death when analysing 5-day instead of 2-day mortality, by univariate and multivariate analyses. Among all 188 patients who died, shock at admission was independently associated with early deaths (odds ratio (OR) 2.683, 95% confidence interval (CI) 1.014–7.097), whereas chronic heart disease was associated with late deaths (OR 0.382, 95% CI 0.153–0.956).

DISCUSSION

It is widely recognised that the evolution of patients with CAP within the first 48 h is crucial 1, 15, 16. In fact, once clinical stability is achieved, substantial clinical deterioration owing to pneumonia is rare 17. In a previous study, the current authors analysed the causes and factors associated with early failure in hospitalised patients with CAP 11. In the definition of early failure as “lack of response or worsening of clinical or radiological status at 48–72 h requiring changes in antibiotic therapy or invasive procedures,” patients who had died within the first 48 h of admission were specifically excluded. The current prospective study offers a comprehensive evaluation of this group of patients in order to establish the causes of, and risk factors for, early mortality in CAP.

In the present series of patients admitted according to predefined criteria 18, in which the severe immunosuppressed population is excluded, the early mortality rate was 2.3%; that is, one third of the total patients with CAP who died during hospital admission. This figure, though relatively low in the current authors’ view, is difficult to compare with others obtained in previous series, owing to differences in definitions and study populations.

Overall, although differences in the frequency and types of underlying diseases were not observed among early deaths compared with all the others, the demographic and clinical characteristics of early deaths define a group with more severe pneumonia, as shown by the fact that 91% of cases were classified in the PSI high severity risk classes.

The most frequent causes of early death were respiratory failure and shock/multiorgan failure. Indeed, a large proportion of them had bacteraemia and presented with septic shock and/or respiratory failure at admission. The vast majority of deaths were pneumonia related, in the setting of an unbalanced inflammatory response.

Both bacteraemic pneumococcal pneumonia and P. aeruginosa pneumonia were significantly more frequent in early deaths. In the case of S. pneumoniae, no relationship between mortality (early and late deaths) and drug resistance could be demonstrated. This observation is in agreement with most previous studies of bacteraemic pneumococcal pneumonia which have not shown differences in mortality between those with susceptible and those with nonsusceptible pneumococci when controlling for age, underlying disease, severity of illness on presentation, and appropriate treatment 19, 20. In fact, all patients who died early and had pneumococcal infection were given concordant antibiotic therapy from admission. This evidence reinforces the classical concept that early deaths are less dependent on antibiotic efficacy than on other factors, including inadequate host response 21. Recent studies suggest that modulation of the immune system could improve the outcomes of patients with severe pneumonia 22, 23. However, further studies are warranted to evaluate the relationship among excessive host response and early deaths. Importantly, randomised clinical trials addressing the potential role of steroids as adjunctive therapy in severe CAP are needed.

Conversely, it has been shown previously that polysaccharide pneumococcal vaccination may prevent invasive pneumococcal disease in adults, and improve outcomes 24, 25. In the present study, <10% of patients who died had received pneumococcal vaccination, despite the fact that >75% had underlying disease. These findings concur with other studies showing that current vaccination rates among target persons remain low 26, 27. The current authors believe that a wider use of the pneumococcal polysaccharide vaccine may help to prevent bacteraemic pneumococcal pneumonia and, conceivably, lowering the rates of early deaths in CAP.

However, the present data suggest a possible relationship between early deaths and discordant therapy in cases of P. aeruginosa pneumonia, since five out of six patients with early deaths while receiving discordant therapy had this diagnosis. None of these patients had a previous diagnosis of bronchiectasis, nor had they received corticosteroid therapy; thus, they did not present with the major risk factors for P. aeruginosa pneumonia indicated in the current guidelines for the management of CAP 1. Nevertheless, it should be borne in mind that all these patients had severe pneumonia with high risk of death, in spite of appropriate antibiotic therapy.

The clinical risk factors for early mortality identified by multivariate analysis, such as increased age, altered mental status, multilobar pneumonia and shock, have also been recognised in previous studies as factors associated with overall mortality and with mortality occurring within the initial five days after hospital admission 4–7, 28. Such factors would be expected to influence early evolution if present at admission. In addition, the present study identified chronic heart disease as a factor associated with late mortality. Apart from age (the major driver in the PSI score) the other factors would also be expected to influence early evolution if present at admission. Additionally, the present study identified discordant therapy as an independent risk factor for early mortality. Nevertheless, this finding is supported almost exclusively by the aforementioned cases of patients with P. aeruginosa pneumonia, and the issue requires further study. The question of whether the use of appropriate antibiotics has a clear impact on survival in the first hours after admission remains unanswered, and should be addressed by future research.

In conclusion, current early mortality is relatively low, representing around one third of deaths in patients with community-acquired pneumonia who died during hospital admission. It occurs mainly in elderly patients or patients presenting with altered mental status and septic shock. Pneumococcal bacteraemia and discordant antibiotic therapy, mainly due to lack of coverage against Pseudomonas aeruginosa, are also significant risk factors. The major causes of early death are pneumonia related, such as respiratory failure and shock in the setting of an inadequate host response.

Support statement

This study was supported by research grant REIPI RD06/0008 from the Ministry of Health and Consumption, Institute of Health Carlos III, Spanish Network for the Research in Infectious Diseases; and by the Institute of Biomedical Investigation of Bellvitge (C. Garcia-Vidal). The funding sources had no role: in the study design; in the collection, analysis or interpretation of the data; or in the decision to submit the manuscript for publication. Only the authors had full access to the data files for the study.

Statement of interest

None declared.