Performance of the CORB (Confusion, Oxygenation, Respiratory Rate, and Blood Pressure) Scale for the Prediction of Clinical Outcomes in Pneumonia

Background Community-acquired pneumonia (CAP) is a common cause of morbidity and mortality due to misdiagnosis and inappropriate treatment approaches. Objective To assess the performance of the CORB score in subjects with CAP for predicting in-hospital mortality, death within 30 days of admission, and requirement for invasive mechanical ventilation (IMV) and vasopressor support. Methods A retrospective, cohort study with diagnostic test analysis of CORB and CURB-65 scores in subjects with CAP according to ATS criteria was undertaken. An alternative CORB score was estimated by replacing SpO2 ≤90% by the SpO2/FiO2 ratio. Crude and adjusted odd ratios (AOR) were calculated for each variable. The area under the receiver operating characteristics curve (AUROC) was constructed for each score, and outcomes were analyzed. AUROCs were compared with the DeLong test, considering a p value <0,05 statistically significant. Results From 1,811 subjects who entered the analysis, 15.1% (273/1,811) died in hospital, 8.78% required IMV (159/1,811), and 9.77% (177/1,811) needed vasopressor support. CORB had an AUROC of 0,660 (95% CI: 0,623–0,697) for in-hospital mortality; an AUROC of 0,657 (95% CI: 0,621–0,692) for 30-day mortality; an AUROC of 0,637 (CI 95%: 0,589–0,685) for IMV requirement; and an AUROC of 0,635 (95% CI: 0,589–0,681) for vasopressor support. CORB performance increases when the SpO2/FiO2 ratio <300 is used as oxygenation criterion in the prediction of requirement for IMV and vasopressor support, with AUROC of 0,700 (95% CI: 0,654–0,746; p < 0.001) and AUROC of 0,702 (95% CI: 0,66–0,745; p < 0.001), respectively. CURB-65 score presents an in-hospital mortality AUROC of 0,727 (95% CI: 0,695–0,759) and 30-day mortality AUROC of 0,726 (95% CI: 0,695–0,756). Conclusions CORB score has a good performance in predicting the need for IMV and vasopressor support in CAP patients. This performance improves when the SpO2/FiO2 ratio <300 is used instead of the SpO2 ≤90% as the oxygenation parameter. CURB-65 score is superior in the prediction of mortality.


Introduction
Community-acquired pneumonia (CAP) continues to be one of the main causes of morbidity and mortality; its occurrence in adults is estimated at approximately 16 to 23 cases per 1,000 person-years, and it increases with age [1]. It is the leading cause of death from infection worldwide [2,3]. A global mortality of 10% to 14% is attributed to it, being less than 2% in healthy young people. However, mortality can increase to 11%-14% in adults who require hospitalization and may reach 25% to 50% in patients who are admitted to an intensive care unit (ICU) [4,5]. Its early recognition and treatment are essential to avoid an ICU delayed admission, which is considered an independent factor related to a long hospital stay and higher mortality rates [6]. For this reason, different scores have been created and validated to carry out an effective prognostic identification, thus providing a guide to the most appropriate site for handling and monitoring the CAP patient.
Among the most widely used scores to predict mortality are the PSI (pneumonia severity index) [5,7] and CURB-65 (confusion, urea nitrogen >7 mmol/L (19 mg/dL), respiratory rate ≥30/min, systolic blood pressure <90 mmHg or diastolic blood pressure ≤60 mmHg, and age ≥65 years) [8]. Additionally, given their superiority in predicting the need for ICU or mechanical ventilation, the criteria for pneumonia severity of the Infectious Diseases Society of America/ American oracic Society (IDSA/ATS) [5,8] and the SMART-COP score have been used [9]. However, one of the restrictions to applying these scores in different care settings is the need to perform at least one invasive procedure.
In search of a practical score that does not require the use of invasive measures in its construction, Buising et al. [10] proposed in 2007 the CORB score, which uses the information of consciousness state, oxygen saturation by pulse oximetry, respiratory rate, and blood pressure, reaching a sensitivity of 72.2% and a specificity of 70.1% for a composite outcome of mortality and requirement for invasive mechanical ventilation (IMV) and vasopressor support [10,11]. Nevertheless, performance data of CORB score are still scarce for independent prediction of these outcomes, and its performance related to scores such as the CURB-65 is not clear, which is why this study proposes to assess performance of the CORB score compared to the CURB-65 score as a predictor of IMV, vasopressor support, in-hospital mortality, and 30-day mortality.

Material and Methods
A retrospective cohort study was carried out in subjects with CAP in the third-level center Clínica Universidad de La Sabana (Chía, Cundinamarca, Colombia). Patients were treated in the emergency department, general ward, or ICUs. Data were collected from January to August 2020, from medical records dated between January 2012 and February 2020.

Selection Criteria.
Inclusion criteria included an age ≥18 years, regardless of the gender; stay in the emergency room (resuscitation, observation room, and transit room), general ward, or ICU (for at least 6 hours); CAP diagnosis according to the ATS guidelines [5,7]; and clinical records that included information for CORB and CURB-65 scores' assessment. Patients with acute decompensation of chronic diseases such as exacerbated COPD, previous congenital heart disease or decompensated heart failure, and chronic or acute interstitial lung disease were excluded. Patients whose pneumonia diagnosis changed or was ruled out during hospitalization, whose pneumonia was related to bronchial obstruction, or those who required IMV prior to taking arterial gases were also excluded.

Variables.
e requirement for IMV and vasopressor support, in-hospital death, and 30-day mortality were the outcomes; the composite outcome included the variables of in-hospital mortality, IMV requirement, and vasopressor requirement. In addition, clinical presentation, findings on physical examination, vital signs, FiO 2 and SpO 2 upon admission, comorbidities, laboratory tests results, diagnostic imaging findings (chest X-ray and/or chest CT), and arterial blood gases were considered as independent variables.
With the obtained variables, CURB-65 and CORB were calculated; the latter scores one point for the variables confusion, oxygenation by SpO 2 ≤90%, respiratory rate ≥30 breaths/minute, and blood pressure (systolic blood pressure <90 mmHg or diastolic blood pressure ≤60 mmHg), with a score of ≥2 points being considered as severe pneumonia. In addition, a CORB score was constructed in which the oxygenation assessment of SpO 2 ≤90% was replaced by a SpO 2 / FiO 2 ratio <300.
Data were obtained from admission registry and clinical records during entire hospital stay, while 30-day mortality was obtained from the national source of death. In order to reduce transcription biases, medical records were revised by at least two different reviewers, and data were verified by them when they were recorded into the database.

Sample
Size. Sample size estimation in diagnostic test [11] and data from the study by Williams et al. [12] were considered, in which an incidence of in-hospital mortality due to CAP of 12.7% and a performance for this outcome of the CORB score ≥2 points of 78% for sensitivity and 40% for specificity were reported. For a confidence level of 95% and a precision level of 7%, a minimum total of 1,060 subjects were required. Patients who did not meet the eligibility criteria were replaced until the sample size was completed.

Statistical
Analysis. Data were compiled in the electronic data capture software Research Electronic Data Capture (REDCap) [13] and later downloaded to a spreadsheet to perform the final analysis in the licensed in STATA 14 and SPSS 25 program. Qualitative variables were reported in frequencies and percentages, while quantitative variables were summarized in median and interquartile ranges if their distribution was normal in mean and standard deviation and if their distribution did not meet normality parameters. A bivariate analysis was performed comparing the quantitative variables with Student's t or Mann-Whitney U test according to their distribution, whereas the qualitative variables were compared using the Chi-square test. By using the scores obtained from CURB-65 and two CORB scores (one score with SpO 2 and another one with the SpO 2 /FiO 2 ratio as oxygenation parameters), the respective areas under the ROC curve (AUC-ROC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR+), and negative likelihood ratio (LR−) with their respective 95% confidence intervals were determined. e association strength of each variable of the studied scores regarding the proposed outcomes was estimated by calculating the odds ratio (OR) and adjusted odds ratio (AOR) using a logistic regression model. e ACORs of the different scores were compared with the DeLong test. For the estimates, a p value <0,05 was considered significant.

Ethical Considerations.
is study was approved by the research committee of the Universidad de La Sabana and the institutional ethics committee of the Clínica Universidad de La Sabana, as registered on Act Minute No 526 of January 29, 2021.
Sensitivity for mortality, IMV, and vasopressor support outcomes was higher for CURB-65. Specificity was higher for CORB with SpO 2 ≤90% for in-hospital and 30-day mortality outcomes; however, when the CORB is used with SpO 2 /FiO 2 <300 in its score, specificity improves for vasopressor support and IMV outcomes. In composite results, high sensitivity was found in the CORB ≥2 (SpO 2 /FiO 2 <300) with 91.3% and specificity in the CURB-65 > 2 with 77.9%. Table 3 and Figure 2 show the complete performance results for each of the CORB scores (with SpO 2 ≤90% and with SpO 2 /FiO 2 <300) and CURB-65 regarding in-hospital mortality, 30-day mortality, and requirement for vasopressor support and IMV outcomes and composite outcomes.

Discussion
It was found that the CORB score presents a good performance as a predictor of IMV and vasopressor support requirement, being superior to CURB-65 in estimating these outcomes. In the evaluation of in-hospital and 30-day mortality, CURB-65 shows higher performance than CORB calculated with SpO 2 ≤90% and CORB calculated with SpO 2 / FiO 2 <300. On the other hand, replacing the oxygenation parameter of SpO 2 ≤90% of the CORB score with the SpO 2 / FiO 2 <300 index turns out to be superior in the prediction of outcomes in pneumonia.
Canadian Respiratory Journal e inclusion of oxygenation parameters in the CORB score improves the performance for the prediction of IMV compared to the CURB-65. In addition, it is necessary to clarify that the CURB-65 was designed to predict only 30-day mortality, not other outcomes such as the need for IMV or vasopressors [8]. e measurement of oxygen saturation, through pulse oximetry or by calculating the SpO 2 /FiO 2 index, has been correlated with different degrees of  hypoxemia in the patients with pneumonia [14,15]. In the context of acute respiratory failure and adult respiratory distress syndrome (ARDS), PaO 2 /FiO 2 represents one of the most important physiological variables for determining the degree of lung injury [16], and it is also a severity benchmark for defining severe pneumonia [5]. In an observational study    carried out by Luna et al. which included 63 patients with ventilator-associated pneumonia, it was found that PaO 2 / FiO 2 was the factor with the greatest discrimination to predict short-term survival [17]. Regarding the particular case of pneumonia at high altitude, Martínez et al. argued that the measurement of oxygenation through D(A-a)O 2 and PaO 2 /FiO 2 of arterial gases showed the best discrimination capacity for IMV at 2,600 meters above sea level [18].
In a multivariate analysis, Buising et al. found that the urea level and age ≥65 years proposed in CURB-65 were not strongly associated variables for the composite outcome of death, IMV, or vasopressor support (OR � 1.71; 95% CI: 0.82-3,58; OR � 0.52; 95% CI: 0,23-1,16, respectively), so the authors proposed to eliminate age and replace urea by SpO 2 , which improved the association for prediction of severity by CAP (OR � 3,49; 95% CI: 1,77-6,89) [10], results that are similar to those reported in our study where oxygenation levels were more associated with the requirement for IMV and vasopressor support. Removing the variable associated with age ≥65 years can reduce the misperception that some scores may present with the mortality outcome, decreases the possibility of obtaining higher scores for patients with advanced age who are not comorbid, and reduces the proportion of low scores in younger patients with severe CAP pictures when incorporating a variable correlated with different degrees of hypoxemia. Moreover, excluding the age variable may favor the use of the CORB score in the daily evolution of the patient, since this score would be entirely composed of variables that can change rapidly.
Babu et al. analyzed if the noninvasive index SpO 2 /FiO 2 ratio could replace invasive index PaO 2 /FiO 2 in all modes of oxygen supplementation in acute hypoxemic respiratory failure; a total of 300 patients were included in this study [19]. In the result, a strong positive linear correlation was noted between PF ratio and SF ratio (r � 0,66; p < 0.001). Moreover, SF values are 285 and 323 corresponding to PF ratios of 200 and 300 with a sensitivity and specificity of 70 to 80%, respectively. Similar data were found in the study carried out by Pandharipande et al. calculating the respiratory parameter of the SOFA score with noninvasive index SpO 2 /FiO 2 and index PaO 2 /FiO 2 with a strong positive correlation [20]. Currently, SpO 2 /FiO 2 is a diagnostic tool for the early recognition of respiratory failure and the need for IVM, because the pulse oximeter is simple, inexpensive, and available for continuous monitoring of saturation.
Replacing SpO 2 <90% with SpO 2 /FiO 2 <300 ratio in CORB score significantly improves the AUC-ROC for pneumonia outcomes. Some authors suggest assessing oxygenation status through indices including FiO 2 since they have shown good prediction of mortality in ARDS [21]. e persistent decrease of PaO 2 /FiO 2 is related to significant alveolar involvement in patients with pneumonia as well as to the progression to ARDS; PaO 2 /FiO 2 values less than 150 have been associated with a worse prognosis [22]. Arterial blood gas measurement is the gold standard for oxygenation assessment; however, it is an invasive, time-consuming, and uncomfortable procedure for patients. Considering these difficulties, the use of indices from pulse oximetry may be a valid alternative; SpO 2 /FiO 2 has shown a reliable correlation with the PaO 2 /FiO 2 index [23]. A SpO 2 /FiO 2 ratio of 235-315 is equivalent to PaO 2 /FiO 2 levels of 200-300 [23]; the SpO 2 /FiO 2 <300 would be equivalent to moderate or severe hypoxemia, a parameter that is considered a minor criterion for pneumonia according to ATS/IDSA [5,7].
is study has several limitations. It is a single-center study, which could restrain its generalizability, although the good sample size supports the results. In addition, subjects older than 18 years were enrolled, which restricts its applicability in younger groups, but adult population included is considered representative. Moreover, since it is a retrospective study, the quality of the information may be affected in the completion of the medical records, but the reduction of the transcription bias could be achieved by the verification carried out by at least two study investigators in charge of corroborating the data obtained from the clinical records. Furthermore, an altitude of 2,640 m.a.s.l. at which the cohort was evaluated could be considered a limitation to the generalization of the results due to acclimatization phenomena [24,25]. However, there are no significant variations in the results compared to previous studies carried out at sea level [10]. Prospective studies are required to corroborate the performance of the CORB scale with the modification of the oxygenation parameter.

Conclusions
e CORB scale for pneumonia proves to have a good performance for the prediction of IMV and vasopressor support, with a higher performance when the SpO 2 /FiO 2 ratio <300 is used as an oxygenation parameter. Finally, when compared to the CURB-65 scale, the CORB scale is not superior in predicting mortality.

Data Availability
Data supporting the findings of this study are in medical records from the research center.

Canadian Respiratory Journal
Disclosure e research did not receive specific funding but was conducted as part of the authors' employment at the Universidad de La Sabana.