The mortality risk factor of community acquired pneumonia patients with chronic obstructive pulmonary disease: a retrospective cohort study

Background Chronic obstructive pulmonary disease (COPD) is one of the most common comorbidities in community acquired pneumonia (CAP) patients. We aimed to investigate the characteristics and mortality risk factors of COPD patients hospitalized with CAP. Methods A retrospective cohort study was conducted at Shanghai Pulmonary Hospital and Shanghai Dahua Hospital. Clinical and demographic data in patients diagnosed with CAP were collected between January 2015 and June 2016. Logistic regression analysis was performed to screen mortality risk factors of COPD patients hospitalized with CAP. Results Of the total 520 CAP patients, 230 (44.2%) patients had been diagnosed comorbid with COPD (COPD-CAP). CAP patients comorbid with COPD patients had higher rate of need for ICU admission (18.3% vs 13.1%) and need for NIMV (26.1% vs 1.4%) than without COPD (nCOPD-CAP). The PSI, CURB-65 and APACHE-II scores in COPD-CAP patients were higher than that in nCOPD-CAP patients (95 vs 79, P < 0.001; 1 vs 1, P < 0.001; 13 vs 8, P < 0.001, respectively). Logistic regression analysis indicated that aspiration, D-dimer > 2.0 μg/mL and CURB-65 ≥ 3 were risk factors associated with in-hospital mortality ((odd ratio) OR = 5.678, OR = 4.268, OR = 20.764, respectively) in COPD-CAP patients. The risk factors associated with 60-day mortality in COPD-CAP patients were comorbid with coronary heart disease, aspiration, need for NIMV (non-invasive mechanical ventilation) and CURB-65 ≥ 3 (OR = 5.206, OR = 7.921, OR = 3.974, OR = 18.002, respectively). Conclusions COPD patients hospitalized with CAP had higher rate of need for NIMV, need for ICU admission and severity scores than those without COPD. Aspiration, D-dimer > 2.0 μg/mL, comorbid with coronary heart disease, need for NIMV and CURB-65 ≥ 3 were mortality risk factors in CAP patients comorbid with COPD. Electronic supplementary material The online version of this article (10.1186/s12890-018-0587-7) contains supplementary material, which is available to authorized users.

Chronic obstructive pulmonary disease (COPD) is a disease with persistent airflow limitation and chronic inflammatory response, which progresses slowly [5]. The number of COPD cases in China increases dramatically from 32.4 million in 1990 to 54.8 million in 2013, which poses a heavy economic burden in China [6,7].
COPD is one of the most common comorbidities in CAP. In the U.S., from 2005 to 2007, the incidence of inpatients primary pneumonia among elders with COPD is 54.2/1000 person-years, which is nearly seven times higher than that in elders without COPD [8]. Fine M. J. et al. derived a prediction rule to evaluate 30-day mortality in patients with CAP, chronic pulmonary diseases were excluded as a risk factor [9]. Meanwhile, previous study demonstrated that COPD patients hospitalized with CAP had the same mortality rate and demographic features, except for age and gender, as patients without COPD [10]. On the contrary, Restrepo M. I. et al. studied CAP population and indicated that COPD patients hospitalized with CAP had higher 30-and 90-day mortality than patients without COPD [11]. Besides, Guertler C. et al. reported that COPD was one of the risk factors for long-term mortality in CAP patients during 18-month follow-up [12]. Since the influences of COPD on severity scoring and mortality in CAP are controversial, we want to make clear that whether COPD patients hospitalized with CAP have higher severity scores and higher mortality rate. Meanwhile the mortality risk factors for COPD patients hospitalized with CAP still need to be further studied.
The aim of the current study is to investigate the characteristics of CAP patients comorbid with and without COPD, and to find out the mortality risk factors in COPD patients hospitalized with CAP.

Study design and inclusion and exclusion criteria
To perform a retrospective cohort study, we recruited patients with the diagnosis of CAP who were admitted in hospital between January 2015 and June 2016.
Patients were included if they were: 1) age ≥ 18 years; 2) symptoms of an acute lower respiratory tract illness (cough and at least one other lower respiratory tract symptom); 3) new focal chest signs on examination; 4) at least one systemic feature (either a symptom complex of sweating, fevers, shivers, aches and pains and/or temperature of 38°C or more); 5) no other explanation for the illness, which is treated as CAP with antibiotics [3]. Patients were not eligible for inclusion if they had: 1) been diagnosed with asthma, bronchiectasis, interstitial lung disease or other pulmonary infectious diseases; 2) life-threatening diseases (i. e. advanced malignant tumor, severe cardiopulmonary dysfunction, renal or liver dysfunction, etc.) within the past year; 3) severe immunosuppressive such as HIV infection, using longtime high dose of immunosuppressive agents, chemotherapy or organ transplantation; 4) research-related data missing. We only used the first hospitalization information if patients were admitted more than once during our study period.

Definitions
Diagnosis of COPD was made in patients with clinical symptom (dyspnea, chronic cough or sputum production) and smoking history at least 10 years, and it was confirmed by spirometric evidence of airflow obstruction in stable stage (post-bronchodilator FEV 1 /FVC < 0.70) [5]. The classification of COPD severity was based on the Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria: GOLD stage I was defined as a forced expiratory volume in 1 s (FEV 1 ) greater than 80% predicted, GOLD stage II as FEV 1 between 50 and 80% pred., GOLD stage III as FEV 1 between 30 and 50% pred. and GOLD stage IV less than FEV 1 30% pred.. Aspiration pneumonia was applied to situations where a patient with risk factors including altered level of consciousness, neurological disorders such as stroke, presence of dysphagia, gastric disorders such as gastro-oesophageal reflux and setting up gastric tube [3]. Need for ICU admission was applied to situations when a patient met the criteria of severe pneumonia and admission to ICU was necessary [1]. Need for non-invasive mechanical ventilation (NIMV) was applied to situations when a patient met one of these criteria, diagnosed COPD with a respiratory acidosis pH 7.25-7.35 (H + 45-56 nmol/l), hypercapnic respiratory failure secondary to chest wall deformity (scoliosis, thoracoplasty) or neuromuscular diseases, cardiogenic pulmonary edema unresponsive to continuous positive airway pressure (CPAP), weaning from tracheal intubation [13]. In-hospital mortality and 60-day mortality were categorized as primary outcomes and need for ICU admission and need for NIMV were categorized as secondary outcomes.

Data collection
All data were independently collected by Dr. RXD and Dr. QHK retrospectively according to the inclusion criteria (1) age ≥ 18 years; 2) diagnosed with CAP on admission) with unified Excel table. The unified Excel table was designed by Dr. RXD and Dr. QHK, it included demographic data (gender, age, smoking status), comorbidities (diabetes, coronary heart disease, hypertension, cerebral infarction, asthma, bronchiectasis, interstitial lung diseases, pulmonary infectious diseases, tumor), physical examination (mental status, respiratory rate, pulse, blood pressure and body temperature), laboratory examination, lung function test and X-ray/chest radiographic reports. The severity scores were assessed using PSI, CURB-65 and APACHE-II scores by two researchers. The highest score within 72 h of admission was chosen and recorded. The in-hospital mortalities were assessed by medical records and 60-day mortalities were assessed by telephone survey.

Statistical analysis
Quantitative data, which were normally distributed, were analyzed with the Student's t-test. Mann-Whitney U-test was used for calculating differences between data of two groups which were not normally distributed. For qualitative data, the χ 2 test were used to assess differences. Covariates were recruited in the univariate analysis with in-hospital mortality and 60-day mortality as outcomes. Only when the covariates with a significant (P < 0.05) by univariate analysis were entered in the multivariate analysis. Multicollinearity was screened using Spearman rank correlation matrices when variables with |r| > 0.7. A backward stepwise approach in the multivariate analysis was considered to keep significant covariates with a cut-point of 0.05. The discriminatory values of PSI, CURB-65 and APACHE-II scores in relation to in-hospital mortality and 60-day mortality were analyzed by the receiver-operating characteristic (ROC) analysis. ROC curves were performed to obtain the area under the curves (AUCs), cutoff values, sensitivities, specificities. The diagnostic performance was referred as high (AUC > 0.9), moderate (AUC = 0.7-0.9), or low (AUC = 0.5-0.7). The optimal cut-off values were determined by the Youden index (maximum of sensitivity + specificity -1) (Additional file 1: Figure S1). AUCs were compared among PSI, CURB-65 and APACHE-II using the method of DeLong et al. [16]. All the data were analyzed with SPSS Version 20.0 software (SPSS, Chicago, IL) and Medcalc Version 15.2.2 software (Mariakerke, Belgium).

Patient characteristics
Among 605 CAP patients recruited in this study, 291 patients were diagnosed with COPD. 230 (44.2%) patients had been diagnosed comorbid with COPD (COPD-CAP), the recruiting procedure was described in Fig. 1. COPD patients hospitalized with CAP were significantly more likely to be older males and smokers (Table 1). Whereas CAP patients without COPD (nCOPD-CAP) showed higher rates of cough, fever and purulent expectoration. In addition, COPD-CAP patients were significantly more likely to be dyspnea and tachypnea. As for laboratory examinations, COPD-CAP patients presented more acidosis, hypoxemia and hypercapnia, and higher hemoglobin and hematocrit level than nCOPD-CAP. We also did arterial blood gas analysis of COPD-CAP patients during stable stage, and the records were collected in Additional file 2: Table S1. 24 (10.4%) COPD-CAP patients in stable stage had type 2 respiratory failure, less than on admission (52/22.6%). As a retrospective study, patients with post-bronchodilator FEV 1 /FVC ≥ 70% were recruited in nCOPD-CAP group. Therefore, nCOPD-CAP patients in this study had normal lung function. Whereas in COPD-CAP, most patients had severe airflow obstruction (mean pre FEV 1 % = 45.01, SD = 15.76). COPD-CAP patients had higher rate of need for NIMV, need for ICU admission and higher length of hospital stay than those without COPD. Although the mortality in COPD-CAP patients was higher than that in nCOPD-CAP patients, the difference in discriminate value between COPD-CAP and nCOPD-CAP was not significant.

Comparison of different primary outcomes of predictive rules
The COPD-CAP patients were more severe than nCOPD-CAP patients in PSI, CURB-65 and APACHE-II severity score systems (79 [ [9.75,17], P < 0.001) ( Table 2). As for inhospital mortality and 60-day mortality, only the constituent ratio (mild, moderate, severe) of APACHE-II score was statistically significant.

Discussion
In the present study, we showed that COPD patients hospitalized with CAP had higher rate of need for NIMV, need for ICU admission and severity scores than those without COPD. Aspiration, D-dimer > 2.0 μg/mL, comorbid with coronary heart disease, need for NIMV and CURB-65 ≥ 3 were mortality risk factors in CAP patients comorbid with COPD.
We found that COPD was a frequent comorbidity in CAP patients, less than half of the whole population (44.2%). These results were consistence with other researches in which the rates were between 16 and 45.4% [10,11,[17][18][19][20]. In the present study, COPD patients  hospitalized with CAP were male, older age and smokers in predominance, and significantly more likely to be dyspnea and tachypnea. In addition, COPD-CAP patients presented more acidosis, hypoxemia and hypercapnia than nCOPD-CAP patients. COPD-CAP patients showed high rate of need for NIMV and need for ICU admission, meanwhile with higher PSI, CURB-65 and APACHE-II scores than nCOPD-CAP patients. We deemed that COPD-CAP patients presented more severe than nCOPD-CAP patients on admission. Therefore, chronic obstructive pulmonary disease should be taken into account as one of the comorbidities in assessment of severity of community acquired pneumonia. The in-hospital mortality and 60-day mortality of COPD-CAP patients in our study was 8.3 and 12.6%. In a cohort of 262 sample size, the 30-day mortality of CAP patients with COPD was 8.4%, and there was no difference between CAP patient with and without COPD [10]. The same result was found in a cohort with 367 CAP patients [21]. Nevertheless, in the study of Restrepo M. I., COPD patients hospitalized with CAP, compared to patients without COPD, showed significantly higher 30-and 90-day mortality [11]. Three main reasons might account for the same mortality rate between COPD-CAP and nCOPD-CAP patients in our study. First, the CAP patients recruited in this study, no matter nCOPD-CAP or COPD-CAP, they were more likely to have higher mean age, more coexisting illnesses, different distribution of etiology and atypical clinical presentations. Meanwhile, the all-cause mortality, which was used in this study, was significantly related to individual conditions. Secondly, using systemic corticosteroid was found to be associated with decreased mortality of CAP-COPD patients admitted to ICU [22], which might account for same mortality between nCOPD-CAP and COPD-CAP patients. Thirdly, since CAP population was relatively small in this research and previous researches [10,21], enlarge study population would be available to find the difference of mortality between nCOPD-CAP and COPD-CAP patients.
Few studies paid attention to the mortality risk factors of COPD patients hospitalized with COPD. We found that aspiration was associated with in-hospital mortality and 60-day mortality in COPD-CAP patients. Abnormal Data given as n(%) or mean ± SD or median [interquartile range] CAP community acquired pneumonia, COPD chronic obstructive pulmonary disease, nCOPD-CAP CAP patients without COPD, COPD-CAP CAP patients with COPD, ICU intensive care unit, NIMV non-invasive mechanical ventilation, WBC white blood cell, PaO 2 partial pressure of arterial oxygen, PaCO 2 partial pressure of arterial carbon dioxide, FEV 1 forced expiratory volume in one second a Values in bold indicate P < 0.05 b GOLD I was defined as FEV 1 /FVC ratio < 70% and FEV 1 > 80% predicted; GOLD II as FEV 1 /FVC ratio < 70% and FEV 1 50%-80% predicted; GOLD III as FEV 1 /FVC ratio < 70% and FEV 1 30%-50% predicted; GOLD IV as FEV 1 /FVC ratio < 70% and FEV 1 < 30% predicted swallowing reflexes frequently occurred in COPD patients. Clayton N. A. et al. found that in COPD patients, reduced laryngopharyngeal sensitivity and impaired swallowing function might cause pharyngeal residue and subsequent inhalation of pharyngeal contents [23], which might cause bacterial colonization and recurrence of aspiration [24]. Patients at risk of aspiration pneumonia had poor short-term prognosis and the risk of recurrence of pneumonia [25][26][27], which in accordance with COPD-CAP population in present study. D-dimer > 2.0 μg/mL were associated with in-hospital mortality in COPD-CAP patients. Coagulation abnormalities were common in CAP patients without organ dysfunction [28]. The endotoxins of pathogens might cause proinflammatory states and result in activation of the coagulation cascade and inhibition of fibrinolysis [29,30], which impacted on the formation of D-dimer. Previous reports found that the level of D-dimer had predictive value for mechanical ventilation therapy and mortality risk (AUC = 0.75, AUC = 0.859) [31,32]. But D-dimer > 2.0 μg/mL could not be a risk factor for long-term mortality. Incident cardiac complications were associated with increased risk of death at 30 days in CAP patients [33]. In the present study, comorbid with coronary heart  disease was one of the risk factors of 60-day mortality in COPD-CAP patients. Previous study indicated that CAP promoted platelet activation and thrombosis, at the same time, created an imbalance between myocardial oxygen supply and demand, which might increase the cardiovascular events [34]. Whereas, in our present study, need for NIMV was associated with 60-day mortality in COPD-CAP patients. Non-invasive positive pressure ventilation was used when CAP patients in acute hypoxemic respiratory failure, which was well tolerated and was associated with a reduction of intubation rate and of stay in ICU [35]. The ICU mortality was 50% for COPD patients with CAP failed non-invasive ventilation [36]. CURB-65 ≥ 3 and PSI > 90 were found to be independent risk factors for case-fatality rate in CAP patients with COPD [20,37]. In our analysis, CURB-65 ≥ 3, other than PSI > 130 or APACHE-II ≥ 20, was identified to be a high risk factor in relation to in-hospital mortality and 60-day mortality in COPD-CAP patients.
In ROC analysis, all scoring systems in COPD-CAP inpatients had good predictive values in mortality, whereas the predictive abilities among three severity scoring systems had no discrimination. In previous study, it was confirmed that PSI score was the best indicator in elder patients with CAP [38]. Richards G. et al. reported that PSI, CURB-65 and APACHE-II performed similarly in predicting in-hospital mortality and 28-day mortality in CAP population [39]. According to the results of ROC analysis, we thought PSI, CURB-65 and APACHE-II scores performed similarly in predicting in-hospital mortality and 60-day mortality in COPD-CAP inpatients.
There were some limitations that should be acknowledged in this study. Firstly, as a retrospective observational study, we still had a potential selection bias. To tackle these problems, we chose two researchers to collect and organize data independently; patients with indefinite diagnosis were not recruited in this study. Second, this study was performed in only two centers and the sample size was relatively small. Therefore, conducting multicenter study and enlarging sample size would be more available for further research.

Conclusions
In conclusion, our study demonstrated that COPD patients hospitalized with CAP were older, easier to present respiratory failure. COPD-CAP patients had higher rate of need for NIMV, need for ICU admission and severity scores than those without COPD. Nevertheless, CAP patients comorbid with COPD had the same mortality rate as patients without COPD. Aspiration, Ddimer > 2.0 μg/mL, comorbid with coronary heart disease, need for NIMV and CURB-65 ≥ 3 were mortality risk factors in CAP patients comorbid with COPD. Availability of data and materials Data relating to this study is available from corresponding author on reasonable request.
Authors' contributions JFX made a great contribution to study design, data acquisition, analysis and interpretation, manuscript drafting and critical manuscript revision. RXD and QHK contributed equally to study design, data acquisition, analysis and interpretation, manuscript drafting and critical manuscript revision. BM contributed to study design, data acquisition, data analysis and critical manuscript revision. WX contributed to study design, data acquisition, manuscript drafting and critical manuscript revision. RJT contributed to study design, data acquisition, manuscript drafting and critical manuscript revision. XRW contributed to data acquisition, manuscript drafting and critical manuscript revision. QYK contributed to manuscript drafting and critical manuscript revision. All authors read and approved the final manuscript.

Consent for publication
Not applicable.