Background
Worldwide, several severity scoring systems have been used to guide decisions on the site of care and to assess the prognosis of community-acquired pneumonia (CAP). Examples of such scoring systems are the Pneumonia Severity Index (PSI) [
1], severe pneumonia criteria by the Infectious Diseases Society of America (IDSA)/American Thoracic Society (ATS) (IDSA/ATS severe pneumonia criteria) [
2], CURB-65 by the British Thoracic Society [
3] and A-DROP by the Japanese Respiratory Society [
4]. In the 2005 IDSA/ATS guidelines [
5], healthcare-associated pneumonia (HCAP) was defined as being associated with a greater risk of antimicrobial-resistant infections and worse prognosis than CAP.
Although these pneumonia severity scoring indices incorporate a variety of prognostic factors for CAP, including patient characteristics, such as age, sex, comorbidities, vital signs and laboratory findings, the prognostic factors are distinct from severity scoring indices. All these pneumonia severity scores mentioned above were established before the definition of HCAP was proposed. Although many reports have investigated the prognostic factors and utility of these scoring systems for assessing the severity of CAP, only few have evaluated them in terms of CAP defined by the latest criteria, which excludes HCAP. In this study, we aimed to investigate the prognostic factors of CAP defined by the latest criteria, including not only patient characteristics, vital signs and laboratory findings, but also initial antibiotic therapy as prognostic factors; we also evaluated whether pre-existing pneumonia severity scoring systems are useful for predicting prognosis in CAP excluding HCAP.
Methods
Study design and setting
This study retrospectively analyzed hospitalized CAP patients who were enrolled in a prospective, observational, cohort study at Kurashiki Central Hospital between April 2007 and February 2016. CAP was diagnosed in accordance with the IDSA/ATS guidelines as [
2]: presence of at least one of the clinical symptoms of cough, sputum, fever, dyspnea, and pleuritic chest pain, plus at least more than one finding of coarse crackles on auscultation or elevated inflammatory biomarkers, in addition to a new infiltrate on chest radiography. We enrolled consecutively hospitalized patients diagnosed with pneumonia in this cohort. Exclusion criteria were: age <15 years, acquired immune deficiency syndrome, hospital-acquired pneumonia, and HCAP [
5]. This study was performed as part of a clinical study for pneumonia (UMIN000004353) after October 2010 and was approved by the institutional review board of Kurashiki Central Hospital (approval number 2235). All patients gave their informed consent to participate in this study.
In all the patients, severity of pneumonia was assessed on admission with the use of PSI [
1], IDSA/ATS severe pneumonia criteria [
2], CURB-65 score [confusion, urea >7 mmol/L, respiratory rate ≥30 breaths per minute, low blood pressure (systolic <90 mmHg or diastolic ≤60 mmHg), and age ≥65 y] [
3] and A-DROP score [age ≥70 years in men or age ≥75 years in women, blood urea nitrogen ≥21 mg · dL
−1 or dehydration, oxyhemoglobin saturation measured by pulse oximetry ≤90% or partial pressure of oxygen in arterial blood ≤60 mmHg, confusion, or systolic blood pressure ≤90 mmHg] [
4].
All patients received antimicrobial agents at the discretion of the physician in charge and in accordance with the recommendations of the CAP guidelines of the Japanese Respiratory Society [
4]. Blood tests and chest X-ray images were examined to assess the effectiveness of the antimicrobials. Basically, patients were treated in the intensive care unit (ICU) if they needed mechanical ventilatory support and/or vasopressor drugs.
Microbiological investigations
As far as possible, we tried to obtain samples of sputum and blood for cultures at the time of admission, and blood for measuring serum antibodies to detect the causative pathogens of CAP. A bacterial cause was identified if the following criteria were met: (1) positive sputum culture of more than 1+ on a qualitative test or 105 on a quantitative test, with reference to the sputum Gram stain; (2) positive blood culture (excluding contaminated normal skin flora); (3) positive pleural fluid culture; (4) positive urinary antigen test for Streptococcus pneumoniae and Legionella pneumophila; (5) seroconversion or four-fold increase in antibodies for Mycoplasma pneumoniae and Chlamydophila pneumoniae; and (6) ≥1:320 on a single antibody test for M. pneumoniae PA antibody (FUJIREBIO, Tokyo, Japan) or a cut-off index of ≥2.0 on a C. pneumoniae IgM antibody test using the Hitazyme® assay (Hitachi Chemical, Tokyo, Japan).
Prognostic variables
In this study, we assessed age, sex, smoking status, comorbidities, vital signs and laboratory findings that influenced prognosis. Comorbidities included chronic heart disease, chronic obstructive pulmonary disease (COPD), diabetes mellitus, cerebrovascular disease, malignant disease, chronic kidney disease, and chronic liver disease. We diagnosed COPD using the Global Initiative for chronic obstructive lung disease (GOLD) definition [
6], and patients who were already diagnosed and treated as COPD in other hospitals and had emphysema on chest tomography were also included. We defined malignant disease as one that was active at the time of admission or was diagnosed within one year of admission.
We did not evaluate disturbances in consciousness as a prognostic variable in this study, because it is included as a factor in severity scoring systems and because we did not assess consciousness separately in all patients. Also, we used a partial pressure of arterial oxygen/fraction of inspired oxygen (PaO2/FiO2) ratio of ≤250 mmHg as a surrogate indicator of oxygenation, since we did not perform arterial blood gas analysis in all patients. We estimated PaO2 from oxyhemoglobin saturation measured by pulse oximetry.
Previous reports have shown that CAP patients with bacteremia [
7] or multilobar pneumonia [
8] have a worse prognosis; therefore, we included these variables in the analysis.
For the treatment of CAP, some systematic reviews and meta-analyses of observational cohort studies have shown that combination therapy of macrolides and β-lactams (BLs) improve prognosis [
9,
10]; however, this was opposed by two randomized controlled studies [
11,
12]. In this study, we investigated the efficacy of macrolide combination therapy, fluoroquinolone (FQN) combination therapy, and BL monotherapy in terms of 30-day mortality.
The primary study outcome assessed was 30-day mortality and secondary outcome was direct admission to the ICU at the time of hospitalization. We also evaluated the usefulness of existing pneumonia severity scores, such as PSI, IDSA/ATS severe pneumonia criteria, CURB-65 and A-DROP, for predicting 30-day mortality in CAP excluding HCAP. We designated patients who were discharged within 30 days of admission and those who did not follow-up after discharge as survivors.
Statistical analysis
Continuous variables were expressed as mean and standard deviation (SD), whereas categorical variables were expressed as frequency (percentage), as appropriate. Continuous variables were analyzed by Student’s t-test and categorical variables were assessed with the chi-square test. Univariate analysis was performed for predicting 30-day mortality. Multivariate analysis using stepwise logistic regression analysis was conducted for all variables that were found to have a P value of ≤0.1 on univariate analysis. For antimicrobial therapy, azithromycin (AZM) was used in almost all patients who received macrolide combination therapy, except in 4 patients who were given clarithromycin (n = 3) and erythromycin (n = 1). Therefore, we excluded these 4 patients and analyzed macrolide combination therapy as AZM combination therapy. To assess the usefulness of existing pneumonia severity scores for predicting the prognosis of CAP, we performed Receiver Operating Characteristic (ROC) curve analysis with application of the Bonferroni correction for two-way comparisons of the AUCs of pneumonia severity scores. All statistical tests were two-tailed, and we considered a P value <0.05 as significant. Analyses were performed using R (version 3.0.3, Vienna, Austria).
Discussion
In this study, we showed that increasing age, presence of COPD and malignancy as comorbidities, low body temperature, tachypnea, PaO2/FiO2 ≤ 250, low Alb levels, high BUN levels and the need for mechanical ventilatory support were predictors of a poor prognosis in CAP patients. We also found that AZM combination therapy with BLs was a predictor of good prognosis, and that the existing pneumonia severity indices have a good predictive ability for the prognosis of CAP. The results of this study are important because the study population comprised patients who were diagnosed as CAP based on the new criteria, which excludes those with HCAP.
All four pneumonia severity scores, including PSI [
1], IDSA/ATS severe pneumonia criteria [
2], CURB-65 [
3] and A-DROP [
4] were previously shown to be useful for predicting prognosis in patients diagnosed with CAP based on the old criteria, which includes HCAP patients. Although some previous reports have investigated the prognostic factors and usefulness of existing pneumonia severity scores in predicting the prognosis of CAP [
13,
14], our report is valuable because patient vital signs and antibiotic therapy, especially macrolide combination therapy, were not evaluated in previous studies.
Age is included as a prognostic factor in the PSI [
1], CURB-65 [
3] and A-DROP scoring systems [
4]. Our study showed that increasing age is also an important prognostic factor for CAP defined according to the latest criteria, which exclude HCAP. As for the comorbidities, Restrepo et al reported that CAP patients with COPD showed significantly higher 30-day mortality (HR 1.32; 95% CI 1.01–1.74) and 90-day mortality (HR 1.34; 95% CI 1.02–1.76) than those without COPD [
15]. Molinos et al also showed that COPD was an independent poor prognostic factor in CAP, after adjusting for age (OR 2.62; 95% CI 1.08–6.39) [
16]. In this study, we showed similar results. Therefore, our study underscores the importance of pneumococcal vaccination in the prevention of pneumonia in COPD patients, as recommended by the Global Initiative for Chronic Obstructive Lung Disease for stable COPD [
6]. In fact, Maruyama et al [
17] reported in their study that the 23-valent pneumococcal vaccine significantly reduced pneumococcal pneumonia by 63.8% (95% CI 32.1–80.7;
P = 0.0015) and all-cause pneumonia by 44.8% (95% CI 22.4–60.8;
P = 0.0006). Malignancy is also a known prognostic factor for CAP, with a score of 30 points in PSI [
1]. Tashiro and colleagues reported that the presence of malignancy is a poor prognostic factor for CAP in patients aged 18–64 years [
18]. We also showed that malignancy is a prognostic factor for CAP in patients aged over 15 years.
All four pneumonia severity scores include vital signs as prognostic factors, although they are different from pneumonia severity indices. Respiratory status is included in all pneumonia severity scores and respiratory rate is adopted in CURB-65 [
3], while respiratory failure is adopted in PSI [
1], IDSA/ATS severe pneumonia criteria [
2] and A-DROP [
4]. A PaO
2/FiO
2 ratio of ≤250 is one of the prognostic factors in IDSA/ATS severe pneumonia criteria. Our study showed that both tachypnea and PaO
2/FiO
2 ≤ 250 are poor prognostic factors. Hypothermia is adopted as a prognostic factor in PSI (body temperature less than 35 °C or more than 40 °C, 15 points) and IDSA/ATS severe pneumonia criteria (body temperature less than 36 °C); our study showed similar results. However, since the cut-off values of these factors for predicting prognosis are unknown, further studies are needed to determine these.
Regarding laboratory findings, our data indicated that Alb and BUN were poor prognostic factors of CAP. BUN is included in all four pneumonia severity scores, while Alb is not included. Previous reports showed that low levels of Alb are a poor prognostic factor in CAP [
19,
20] and in both CAP and HCAP [
21]. Our study also supported these findings, which suggests that, in future, Alb should be included as a prognostic factor in existing pneumonia severity indices.
Recently, some systematic reviews and meta-analysis indicated that compared with BL monotherapy, macrolide combination therapy reduced CAP mortality rate [
9,
10]. However, the studies included in these reviews were all observational in design. On the other hand, two randomized controlled trials [
11,
12] did not demonstrate a reduced mortality rate with macrolide combination therapy. Therefore, the efficacy of macrolide combination therapy in reducing mortality in CAP is controversial. Previous reports that assessed the usefulness of macrolide therapy in patients with CAP, including some cases of HCAP, included erythromycin, clarithromycin, and AZM. In this study, we showed that AZM combination therapy with BLs reduced mortality rate in CAP patients, excluding HCAP patients.
Macrolides, including AZM, have anti-inflammatory properties and immunomodulating effects, such as regulation of neutrophil chemotaxis, decreased pro-inflammatory cytokine production, and regulation of adhesion molecule expression [
22,
23]. AZM demonstrated anti-inflammatory and antivirulent characteristics in mouse and human studies on
P. aeruginosa [
24,
25]. An experimental study on pneumococcal pneumonia in mice showed that AZM combination therapy with ampicillin was effective in downregulating lung inflammation and accelerating bacterial clearance [
26]. These effects of AZM, in addition to its antibacterial properties, may have brought about the reduction of mortality in CAP.
In systematic reviews and meta-analyses, the AUC of summary ROC curves for predicting 30-day mortality with PSI and CURB-65 in CAP were reported as 0.81 and 0.80, respectively [
27]. Shindo et al. [
28] reported that A-DROP was as useful in assessing the severity of CAP as CURB-65, and their AUCs for predicting 30-day mortality were 0.846 and 0.835, respectively. Compared to previous reports, the results of our study indicated mild low AUCs for all pneumonia severity scores.
However, since the AUC was about 0.75 for each scoring system, we believe that the existing pneumonia severity scores are useful for predicting the prognosis of CAP defined by recent criteria.
This study has certain limitations. First, since it was performed at a single center, the applicability of our results to other areas or countries is uncertain. Regardless, the present study analyzed over 1800 CAP patients, which is a large number of patients. Second, our study had some missing data. We did not include disturbance of consciousness as a prognostic variable because we could not assess it in all patients as a separate item. Although arterial blood gas analysis was not performed in 513 patients from among the survivors and 12 patients from among non-survivors, oxyhemoglobin saturation, as measured by pulse oximetry, was ≥90% in almost all these patients. Hence, we estimated that the PaO2/FiO2 ratio in these patients was at least >250. Analysis of disturbance of consciousness and PaO2/FiO2 ratio as separate prognostic factors might have revealed different results. Finally, although AZM combination therapy was associated with a good prognosis in CAP, the best formulation for a particular population is unclear. In this study, 91.6% of patients received AZM in the oral form; therefore, oral AZM may be sufficient, at least for its anti-inflammatory effect. However, it is important to note that use of AZM combination therapy in all CAP patients may increase antimicrobial resistance and cost. Therefore, the CAP population that would benefit from AZM combination therapy should be determined in future randomized controlled trials.
Acknowledgements
The authors would like to thank all their colleagues who recruited and treated the CAP patients.