Comparison of CURB-65, PSI, and qSOFA for predicting pneumonia mortality in patients with idiopathic pulmonary fibrosis

Some patients with idiopathic pulmonary fibrosis (IPF) require hospitalization due to pneumonia. Although predictive scoring tools have been developed and validated for community-acquired pneumonia (CAP), their usefulness in IPF is unknown. The Confusion, Urea, Respiratory Rate, Blood Pressure and Age (CURB-65) score and the Pneumonia Severity Index (PSI) are validated for CAP. The quick Sequential Organ Failure Assessment (qSOFA) is also reported to be useful. The aim of this study was to investigate the ability of these tools to predict pneumonia mortality among hospitalized patients with IPF. A total of 79 patients with IPF and pneumonia were hospitalized for the first time between January 2008 and December 2017. The hospital mortality rate was 15.1%. A univariate logistic regression analysis revealed that the CURB-65 (odds ratio 4.04, 95% confidence interval 1.60–10.2, p = 0.003), PSI (4.00, 1.48–10.7, 0.006), and qSOFA (5.00, 1.44–1.72, 0.01) scores were significantly associated with hospital mortality. There was no statistically significant difference between the three receiver operating characteristic curves (0.712, 0.736, and 0.692, respectively). The CURB-65, PSI, and qSOFA are useful tools for predicting pneumonia mortality among hospitalized patients with IPF. Because of its simplicity, the qSOFA may be most suitable for early assessment.

The ROC curves for hospital mortality are shown in Fig. 1. The clinical utility of the CURB-65, PSI, and qSOFA to predict in-hospital mortality is shown in Table 5. The qSOFA had a sensitivity and specificity (98.5% and 75.0%, respectively) higher than or equal to those of the CURB-65 and PSI. The PSI had the best discriminatory value (AUC 0.736; 95% CI 0.660-0.811), followed by the CURB-65 (AUC, 0.712; 95% CI 0.620-0.801), and the qSOFA (AUC, 0.692; 95% CI 0.602-0.779). However, there were no significant differences among the three scoring systems.

Discussion
The CURB-65 and PSI were developed to predict prognosis in patients with CAP. Although the qSOFA was proposed as a simple bedside scoring tool for early identification of sepsis, it has also been reported that this prognostic tool could be used in patients with CAP who required hospitalization 11 . To the best of our knowledge, this is the first study to show that these tools predict the survival of patients with IPF with pneumonia as well. Given its comparable discriminatory power with 2 existing tools, the qSOFA seems to be the best tool for assessment in the clinical setting.  13,14 . However, pulmonary infection of patients with IPF is associated with a high mortality rate, ranging from 18 to 30% 6,7 . Our study showed that the pneumonia mortality rate of hospitalized patients with IPF was 15.1%. Hence, it is important to recognize that pneumonia is more lethal in patients with IPF than in patients without IPF. Therefore, discriminating patients who would die of pneumonia is crucial for patients with IPF and pneumonia.
As for the qSOFA, there were no statistically significant differences in ROC curves when the qSOFA ROC curve was compared with those of the CURB-65 and the PSI. However, the sensitivity of the qSOFA was higher than that of the other tools. It was reported that the sensitivity of the qSOFA ≥ 2 for mortality in patients hospitalized with CAP was 39.1-50% 11,15 . In this study, only a small number of patients had a qSOFA score of ≥ 2 points (5.0%) with extremely high hospital mortality (66.6-100%). This result might be associated with the high sensitivity of the qSOFA.
When using these tools for patients with pneumonia and IPF it is also important to take the characteristics of the three tools into account. The PSI may overestimate cancer which is unrelated to the lung such as prostate cancer. It may also overestimate the severity in elderly patients because it is heavily weighted towards age. The CURB-65 also includes age as a scoring variable, but only categorizes age as either ≧ 65 or not. The qSOFA does not include any age variable, resulting in possible underestimation in elderly patients. Table 2. Patient clinical data at the first hospitalization. Values are expressed as mean ± standard deviation or actual number. APACHE II Acute Physiology and Chronic Health Evaluation II; BUN, blood urea nitrogen; Cr creatinine; CRP C-reactive protein; CURB-65 confusion, urea, respiratory rate, blood pressure and age score; FDP fibrinogen and fibrin degradation products; GCS Glasgow coma scale; KL-6 Krebs von der Lungen-6; NHF nasal high flow; NIPPV noninvasive positive pressure ventilation; PaCO 2 partial pressure of carbon dioxide; PaO 2 /FiO 2 partial pressure of atrial oxygen / fraction of inspiratory oxygen; PSI Pneumonia Severity Index; PT INR prothrombin tome-international normalized ratio; qSOFA quick Sequential Organ Failure Assessment; SOFA sequential organ failure assessment; WBC white blood cell. a n = 74; b n = 48; c n = 62; d n = 54; e n = 77.  Table 3. Results of the univariate logistic regression analysis of hospital mortality (n = 79). APACHE II Acute Physiology and Chronic Health Evaluation II; CURB-65 confusion, urea, respiratory rate, blood pressure and age; CRP C reactive protein; DLco diffusing capacity for carbon monoxide; FDP fibrinogen and fibrin degradation products; FVC forced vital capacity; KL-6 Krebs von der Lungen-6; PaCO 2 partial pressure of carbon dioxide; PaO 2 /FiO 2 partial pressure of arterial oxygen / fraction of inspiratory oxygen; PSI Pneumonia Severity Index; qSOFA quick Sequential Organ Failure Assessment; SOFA sequential organ failure assessment; WBC white blood cell. a n = 77; b n = 52; c n = 32; d n = 48; e n = 62; f n = 54; g n = 77.   17 . However, validation of triggered AE has been not performed in a multicenter study. In this study, we made major efforts to exclude patients with triggered and suspected triggered AE after careful discussion involving several specialists. Despite our efforts, some patients may have been included.
In conclusion, three scoring tools, the CURB-65, PSI, and qSOFA can predict mortality from pneumonia in hospitalized patients with IPF. Discriminatory power was comparative among the three tools. Hence, the qSOFA would be useful in the clinical setting based on its simplicity.

Methods
Patients. From January 2008 through December 2017, we retrospectively reviewed the medical data of all patients with IPF who required admission to the Kindai University Hospital for pneumonia. IPF was diagnosed based on a recent official statement 1 . Pneumonia was defined as: (1) fever, productive cough, or abnormal white blood cell count, and (2) newly developed consolidation and/or ground-glass opacities on a chest radiograph or chest high-resolution computed tomography (HRCT). The study protocol was approved by the ethics committee of the Kindai University Faculty of Medicine (No. 31-244). Informed consent was waived, because this study was based on a retrospective analysis of case records from our university hospital. All methods were performed in accordance with the relevant guidelines and regulations (Declaration of Helsinki).  Tools for predicting pneumonia mortality. The qSOFA score was calculated according to the Sepsis-3 Task Force scoring system. This score includes systolic blood pressure ≤ 100 mmHg, respiratory rate ≥ 22 breaths/ min, and altered mental status. A total qSOFA score of ≥ 2 points indicates possible organ dysfunction 10 . The CURB-65 is a predictive tool for CAP recommended by the British Thoracic Society (BTS) 20 . The criteria include confusion status, blood urea nitrogen > 20 mg/dL, respiratory rate ≥ 30, systolic blood pressure < 90 mmHg or diastolic blood pressure ≤ 60 mmHg, and age ≥ 65 years 8 . In this study, patients who had a CURB-65 score of ≥ 3 points were classified as being at a high risk of death according to the BTS guidelines 20 . The PSI proposed in 1997 is a useful tool for predicting mortality in patients with CAP 9 . The PSI includes demographics, comorbidities, a physical examination, and laboratory and radiological findings. A PSI class of I-III was reported to represent a low risk of death 8 . In our study, patients who had a PSI class of ≥ IV were defined as being at a high risk of death.

Assessment of survival.
We evaluated the 30-day mortality and the total hospital mortality of the patients.
All deaths were confirmed by hospital chart review.
Statistical analysis. Continuous variables were expressed as means ± standard deviation (SD) and categorical variables as frequencies. Univariate and multivariate logistic regression analyses were used to identify potential risk factors for hospital mortality. The area under the receiver operating characteristic (ROC) curve (AUC) with a 95% confidence interval (CI) was used to assess discriminatory value. Z tests as described by Hanley and McNeil were used to compare pairs of ROC curves 21 . A p value of < 0.05 was considered statistically significant. The analyses were performed with Statflex ver.6 (Artech, Co., Ltd., Osaka, Japan).