key: cord-0696982-n6c8cd9g
authors: Menéndez, Rosario; Méndez, Raúl; González-Jiménez, Paula; Zalacain, Rafael; Ruiz, Luis A.; Serrano, Leyre; España, Pedro P.; Uranga, Ane; Cillóniz, Catia; Pérez-de-Llano, Luis; Golpe, Rafael; Torres, Antoni
title: Early recognition of low-risk SARS-CoV-2 pneumonia: A model validated with initial data and IDSA/ATS minor criteria
date: 2022-05-21
journal: Chest
DOI: 10.1016/j.chest.2022.05.013
sha: 7159d1d02276be21109172d4116e9fd387219828
doc_id: 696982
cord_uid: n6c8cd9g

Background A shortage of beds in the intensive care unit (ICU) and conventional ward during the COVID-19 pandemic led to a collapse of healthcare resources. Research Question. Can admission data and minor criteria by the Infectious Diseases Society of America (IDSA) and American Thoracic Society (ATS) help identify patients with low-risk SARS-CoV-2 pneumonia? Study Design and Methods This multicenter cohort study included 1274 patients in a derivation cohort and 830 (first wave) and 754 (second wave) in two validation cohorts. A multinomial regression analysis was performed on the derivation cohort to compare the following patients: those admitted to the ward (assessed as low-risk); those admitted to the ICU directly; those transferred to the ICU after general ward admission; and those who died. A regression analysis identified independent factors for low-risk pneumonia. The model was subsequently validated. Results In the derivation cohort, similarities existed among those either directly admitted or transferred to the ICU and those who died. These patients could, therefore, be merged into one group. We identified five independently-associated factors as being predictors of low risk (not dying and/or requiring ICU admission) (odds ratios, with 95% confidence intervals): SpO2/FiO2 > 450 (0.233; CI 0.149-0.364); < 3 IDSA/ATS minor criteria (0.231; 0.146-0.365); lymphocyte count > 723 cells/mL (0.539; 0.360-0.806); urea < 40 mg/dL (0.651; 0.426-0.996); and C-reactive protein < 60 mg/L (0.454; 0.285-0.724). The areas under the curve were 0.802 (0.769-0.835) in the derivation cohort, and 0.779 (0.742-0.816) and 0.801 (0.757-0.845) for the validation cohorts (first and second waves, respectively). Interpretation. Initial biochemical findings and the application of <3 IDSA/ATS minor criteria make early identification of low-risk SARS-CoV-2 pneumonia (approximately 80% of hospitalized patients) feasible. This scenario could facilitate and streamline healthcare resource allocation for patients with COVID-19.

7 such variables in the emergency room (ER), and there are no specific ICU admission criteria for COVID- 19 . In the latter case, many clinicians use scores to predict mortality and disease progression. [12] [13] [14] [15] [16] [17] However, the ability to triage patients per ward allocation using initial data from the ER could prove significant, facilitating early identification of low-risk COVID-19 pneumonia that will either not require a later ICU transfer and result in death.

In other words, successive pandemic waves could continue witnessing healthcare institutions exceed their capacity. Early identification may address the issue of a fair use of healthcare resources and promote the safe and prompt redirection of patients deemed suitable to wards, home hospitalization, field hospitals, or medicalized hotels requiring less complex resources. 5 We hypothesized that a combination between clinical and biochemical analysis data obtained at the ER and the minor criteria for ICU admission set forth by the Infectious Diseases Society of America (IDSA)/American Thoracic Society (ATS) could help differentiate between low-risk and severe COVID-19 pneumonia (leading to death or requiring ICU admission). 18 We also hypothesized that either patients transferred from the ward to the ICU or those who die may share some similarities regarding disease severity as those directly admitted to the ICU due to an underestimation of their conditions.

The primary aim of our study was to identify factors for low-risk SARS-CoV-2 pneumonia that will not require ICU admission (directly or transferred from the ward) or lead to death during hospitalization, with the use of clinical and biochemical data obtained in the ER and the IDSA/ATS minor criteria for ICU admission in community acquired pneumonia (CAP). Additionally, we would validate the model in patients J o u r n a l P r e -p r o o f 8 from the second COVID-19 wave. We then had two secondary aims. First, we aimed to compare the initial characteristics of patients directly admitted to the ICU and those either transferred to the ICU from the ward or dying from COVID-19 during hospitalization. Second, we aimed to validate our model in patients from the second COVID-19 wave.

The TRIPOD Statement for the reporting of studies developing, validating, or updating prediction models, whether for diagnostic or prognostic purposes, was followed (see the Supplemental File for the checklist). 19 This was a multi-site study of derivation and validation cohorts across several Spanish hospitals that involved patients from the first and second COVID-19 waves. We included patients with clinical symptoms and a microbiologically confirmed diagnosis of SARS-CoV-2 by reverse transcription polymerase chain reaction (RT-PCR) testing performed on nasopharyngeal swabs. However, we excluded those patients transferred from other medical facilities or nursing homes.

Polytechnic Hospital (2020-122-1) approved the study for the derivation cohort and the Ethics Committee of Galicia (Cod. 2020/239) approved the study for the validation cohort. The need for written informed consent was waived due to the noninterventional study design.

J o u r n a l P r e -p r o o f 9

The following data were collected: demographics, smoking or alcohol habits, number of days from symptom onset, comorbidities (hypertension, cardiovascular disease, obesity, liver or renal disease, and chronic respiratory disease such as chronic obstructive pulmonary disease [COPD] or asthma) and immunosuppression due to malignancy, transplantations or any immunosuppressive therapy. Initial blood levels were recorded for biochemistry, creatinine, aspartate transaminase (AST), lactate dehydrogenase (LDH), D-dimer (DD), C-reactive protein (CRP) and white blood cell (WBC) counts (including absolute neutrophil, lymphocyte and platelet counts).

Oxygen saturation measurements of room air and/or blood gas analysis results were recorded, and peripheral blood oxygen saturation/fraction of inspired oxygen (SpO2/FiO2) ratios were calculated. 20 Initial chest X-ray findings were recorded as unilobar or multilobar (≥ 2 lobes) infiltrates. Initial assessments of disease severity were conducted by calculating the CURB-65 score 21 were diagnosed between August 1 st and November 2020, during the second COVID-19 wave. Both cohorts were divided into two subsets: (a) patients admitted to a ward who were discharged alive and did not require an ICU transfer and (b) patients who died or required ICU admission.

The statistical analysis was performed using IBM SPSS version 25.0 (IBM Corp., Armonk, NY, USA), with a p-value < 0.05 being considered statistically significant.

Qualitative variables were compared with the χ 2 test while quantitative variables were compared with the ANOVA or Kruskal-Wallis test. One-way ANOVA or Kruskal-Wallis tests were used for comparisons of more than two groups. Age was stratified into three groups: < 50, 50-70 and > 70 years.

Blood results were expressed as medians with the interquartile range. For the multivariate analyses, these results were stratified using the following thresholds obtained from the cohort medians and/or by considering previous data: DD > 1000 ng/mL or urea > 40 mg/dL (medians in our cohort); CRP ≥ 60 mg/L (as reported in previous studies and the median in the subset of patients admitted to a ward); SpO2/FiO2 ≤ 450 (median in the subset of patients admitted to a ward); and an absolute lymphocyte count < 724 cells/mL (very close to the median in the subset of J o u r n a l P r e -p r o o f patients admitted to a ward; this cut-off has also been validated by our group as an independent risk factor for death in pneumonia). 22 Radiographic pulmonary infiltrates were grouped as either unilobar or multilobar. Finally, we dichotomized both the IDSA/ATS minor criteria (< 3 or ≥ 3) and the number of days of symptoms (< 7 or ≥ 7) days. 23 The following steps were taken.

First, we performed a multinomial stepwise logistic regression analysis on the derivation cohort to compare the three patient subsets (ward, ICU direct, and ICU transfer plus death) and estimate significantly independent variables. The subset of patients admitted to a ward was the reference group. The model included independent variables found to be either significant in univariate analyses or deemed clinically relevant: hypertension, cardiovascular diseases, age, the SpO2/FiO2 ratio, urea concentrations, CRP levels, lymphocyte counts, DD levels, IDSA/ATS minor criteria, and number of days of symptoms. To avoid overfitting, we performed a stepby-step variable selection (conditional method) to detect collinearity.

Second, we completed a binary logistic regression analysis to predict low-risk (ward admitted) versus high-risk (ICU admission directly from the ER or ward) SARS-CoV-2 pneumonia and/or death (by merging these two subsets). We adjusted for independent variables found to be significant in any of the logits of the prior multinomial analysis. We then assessed model calibration using the Hosmer-Lemeshow test (the distance between the observed and expected values). Also, we calculated the area under the receiver operating characteristic curve (AUROC). Table 1 ). There were 122 (14.7%) deaths in the first validation cohort and 54 (7.1%) in the second validation cohort.

Univariate analysis. Table 1 shows the demographic characteristics, comorbidities and initial biochemical results of the three patient subsets. One hundred twenty-eight patients were initially admitted to the ICU, while 139 were transferred to the ICU after being admitted to a ward. For the analysis, patients who died were included in the subset of those requiring an eventual transfer to the ICU.

The ICU group had a higher proportion of comorbidities-mainly, cardiovascular diseases and cardiovascular risk factors. The laboratory analysis showed lower lymphocyte and platelet counts in the ICU direct group when compared to the ward and ICU transfer plus death groups.

Multivariate analyses. Figure 1 shows results for the two logit models obtained in the multinomial regression analysis, compared with the ward admission reference group. Model 1 compared the ICU direct and ward groups, while Model 2 compared J o u r n a l P r e -p r o o f 13 the ICU transfer plus deaths and ward groups. The independent variables were set as age, sex, cardiovascular disease, arterial hypertension, time since symptom onset, IDSA/ATS minor criteria, the SpO2/FiO2 ratio, urea concentration, CRP level, lymphocyte count and DD level. Three independent factors were found to have a similar odds ratio (OR) for three predictive factors in both cohorts-namely, SpO2/FiO2 ≤ 450; the presence of <3 IDSA/ATS minor criteria; and CRP < 60 mg/L-with the lymphocyte count only appearing relevant in Model 2.

After confirming similar predictors and ORs in the two comparator groups (ICU direct versus ICU transfer or death), the whole cohort was stratified into two subsets: patients admitted to a ward and discharged alive (considered as low-risk) versus any ICU admission and/or death ( Table 2 ). The regression logistic analysis identified five independent variables (Table 4) 

First validation cohort. Table 2 details the demographic characteristics, comorbidities and initial biochemical results of the two validation cohort groups.

When compared to the derivation cohort, comorbidities were similar; however, patients had fewer multilobar infiltrates and lower CRP levels. Table 3 details the IDSA/ATS minor criteria. Table 4 shows that the independent variables behaved quite like those of the derivation cohort, with comparable ORs. The AUC was 0.779 (95% CI 0.742-0.816; p < 0.001) (Figure 2) , with a sensitivity and specificity of 84.2% and 53.6%, respectively, and a PPV and NPV of 84.9% and 52.3%, respectively.

Second validation cohort. The second validation cohort presented similar demographics, comorbidities, analytical parameters and IDSA/ATS minor criteria (Tables 2 and 3 ). The same independent variables were identified as in the other two cohorts, with the exception of urea, which did not enter the model ( Table 4 ). The AUC was 0.801 (0.757-0.845; p<0.001), with a sensitivity and specificity of 89.5% and 50%, respectively, and a PPV and NPV of 92% and 42.2%, respectively.

In summary, around 80% of patients hospitalized with COVID-19 had low-risk pneumonia (discharged alive and did not require ICU admission) after being admitted to a conventional ward. We identified five factors during initial ER evaluation that would predict no ICU requirement or death (i.e., low-risk pneumonia; AUC, 0.802):

IDSA/ATS minor criteria, SpO2/FiO2 ratio, CRP, lymphocyte count and urea. These factors were validated in both a different multicenter cohort (AUC, 0.779) and patients from the second COVID-19 wave (AUC, 0.801). Given that the scale of the pandemic has led to shortages of hospital beds, it is crucial to have simple criteria to improve the safe triage of both mild and severe episodes of pneumonia and ensure better, appropriate allocation of resources.

In our study, 71.8% of the patients in the derivation cohort (75.3% and 86.5%

in the first and second validation cohorts, respectively) were admitted to a conventional ward and remained there until discharged alive. Only 10.1% required direct ICU admission, while 18.1% either died or were later transferred to the ICU. The derivation and validation cohorts presented similar ages and comorbidity features to those previously reported. 3 Mortality rates were 12.2%, 14.7% and 7.1% in the derivation, and first and second validation cohorts, respectively, indicating lower mortality during the second wave.

Using a multinomial regression, we compared the two subsets of severely ill patients (ICU direct admissions and later ward transfers to the ICU or deaths) in relation to patients in the ward. Both groups of severely ill patients exhibited similarities when compared to ward patients: more males, more cardiovascular diseases, lower SpO2/FiO2 ratios, lower lymphocyte counts, and higher urea, lactate dehydrogenase, DD and CRP levels. They also both presented ≥3 IDSA/ATS minor criteria (34.4 and 30.3%) in the ER. Indeed, the independent risk was identical for three factors, allowing the two subsets of severely ill patients to be merged for later analyses and validation. The findings regarding the IDSA/ATS minor criteria are interesting, as patients initially admitted to a ward with ≥ 3 criteria faced a higher risk of a later transfer to the ICU or death. This could suggest that disease severity of these patients could be insufficiently recognized. In CAP, as validated elsewhere, the presence of ≥ 3 IDSA/ATS minor criteria indicates a requirement for ICU admission in those patients who do not need mechanical ventilation or vasopressor treatment. 23, 24 To our knowledge, however, the IDSA/ATS minor criteria have not been evaluated for identifying severe episodes of COVID-19. Additionally, we independently assessed two parameters that provide similar biological information to those used in this study (the PaO2/FiO2 ratio compared to the SpO2/FiO2 ratio and blood urea nitrogen compared to urea)-which are more widely used in ERs-and applied thresholds obtained from our cohort and frequently used in literature.

In SARS-CoV-2 disease, low-risk pneumonia has required the greatest use of hospital resources and bed occupancy per day. It is, therefore, vital that patients with such cases are quickly differentiated from those with more severe cases in ERs. This is can be accomplished promptly using five independent predictive factors adjusted for age, hypertension, comorbidities and other biochemical findings. The model obtained in this study has a good discriminating ability to identify these patients. AUC values were 0.802 and 0.779 in the derivation and validation cohorts, respectively, and the model showed similar sensitivity, specificity and predictive values. Although urea was not entered as an independent variable for patients from the second COVID-19 wave, the model was also validated in this group, with an AUC of 0.801, proving its robustness.

Four biochemical variables independently predicted low-risk pneumonia: lymphocyte count ≥ 724 cells/mL; urea < 40 mg/dL; SpO2/FiO2 ratio > 450; and CRP < 60 mg/L. Interestingly, urea was only an independent variable during the first wave. This may be due to the fact that patients admitted to a ward during the second wave J o u r n a l P r e -p r o o f 17 tended to be younger and present lower urea levels compared with the first wave.

The presence of lymphopenia has been reported in severe cases. Yang et al. 25 reported that up to 85% of severely ill patients presented lymphopenia, which has since been considered a signature of severe disease 26 and immunological misfiring. 27 Huang et al. 2 and Liu et al. 28 reported the importance of initial lymphocyte counts and their evolution during the course of infection. In our study, we selected a threshold (< 724 cells/mL) that has been associated with higher mortality in CAP. 22, 29 The reasons for lymphopenia are not clear, although a direct toxic action against lymphocytes resulting in their apoptosis or necrosis is possible. Indeed, reductions in lymph nodes have been noted in some autopsies. Another important aspect is the possible endothelial dysfunction triggered by SARS-CoV-2. 26 Low initial CRP levels (< 60 mg/L) independently predicted the lack of requirement for ICU admission or progression to death. Higher CRP levels, with thresholds ranging from 40 to 100 mg/L, have been associated with poor prognosis. Castro et al. highlighted that CRP levels as a laboratory result could estimate mortality. 30 Although this study has several strengths, such as the inclusion of three multicenter cohorts and double validation in different disease waves, important limitations must be considered. First, some variables were missing, and there were potential differences in ICU strategies among hospitals. However, we did exclude patients from nursing homes where therapeutic effort could have been limited. 31 Second, a biochemical analysis was performed only at admission and did not include dynamic monitoring. 28 The current study was performed when the population was not vaccinated. Similarly, we did not include a subset of non-admitted patients.

18 A combination of parameters-including host response (e.g., lymphocyte count and CRP levels); lung function (e.g., the SpO2/FiO2 ratio); and < 3 IDSA/ATS minor criteria-provides a feasible tool for decision-making processes in the ER as it relates to evaluating disease severity for safe triage and resource allocation. Similarities with some initial analytical results and IDSA/ATS criteria existed in patients admitted directly to the ICU and those who either were transferred to the ICU from the ward or died during ward hospitalization. Early identification of patients with low-risk SARS-CoV-2 pneumonia who will not require ICU admission and or progress to death could help with resource allocation during periods of hospital bed shortages.

Hospital (2020-122-1) approved this study.

All authors have accepted the publication of the manuscript.

The datasets used and/or analyzed in the current study are available from the corresponding author upon reasonable request.

The authors declare no conflicts of interest.

This study was supported by:

-Instituto de Salud Carlos III (ISCIII) through Project [COV20/00385], co-funded by the European Regional Development Fund/European Social Fund "Investing in Your 

The Funding source does not have any role in design, data obtention, analyses, interpretation of results or drafting. All authors had full access to all of the data in the study and had the final responsibility of making the decision to submit for publication. 

We would like to thank the Integrate Research Program (PII) of Respiratory Infections of SEPAR. We also would like to extend our thanks to Laura Descalzo for her collaboration in the statistical analysis, as well as to Ana Latorre and Luz Mimbiela for their help. Finally, we will always remember those who are no longer among us due to this pandemic. J o u r n a l P r e -p r o o f 1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0 SpO2/FiO2, peripheral blood oxygen saturation/fraction of inspired oxygen.

Leukocyte count, cells/mL (IQR)

Lymphocyte count, cells/mL (IQR) 1273 980

Data are summarized as n (%) or median (interquartile range), as appropriate

Pathophysiology, Transmission, Diagnosis, and Treatment of Coronavirus Disease 2019 (COVID-19): A Review

Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China

Presenting Characteristics, Comorbidities, and Outcomes Among 5700 Patients Hospitalized With COVID-19 in the

Risk Factors Associated With Acute Respiratory Distress Syndrome and Death in Patients With Coronavirus Disease

Fair Allocation of Scarce Medical Resources in the Time of Covid-19

Experiences and Perspectives on Resource Limitation and Patient Care During the COVID

Longitudinal analyses reveal immunological misfiring in severe COVID-19

A dynamic COVID-19 immune signature includes associations with poor prognosis

Deep immune profiling of COVID-19 patients reveals distinct immunotypes with therapeutic implications

COVID-19: consider cytokine storm syndromes and immunosuppression

Viral RNA load in plasma is associated with critical illness and a dysregulated host response in COVID-19

Risk stratification of patients admitted to hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: development and validation of the 4C Mortality Score

Development and Validation of a Clinical Risk Score to Predict the Occurrence of Critical Illness in Hospitalized Patients With COVID-19

Risk Factors of Severe Disease and Efficacy of Treatment in Patients Infected with COVID-19: A Systematic Review, Meta-Analysis and Meta-Regression Analysis

Prediction for Progression Risk in Patients with COVID-19 Pneumonia: the CALL Score

Development and validation of the ISARIC 4C Deterioration model for adults hospitalised with COVID-19: a prospective cohort study

Systematic evaluation and external validation of 22 prognostic models among hospitalised adults with COVID-19: An observational cohort study

IDSA/ATS consensus guidelines on the management of community-acquired pneumonia

Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): the TRIPOD statement

Comparison of the SpO2/FIO2 ratio and the PaO 2/FIO2 ratio in patients with acute lung injury or ARDS. Chest [Internet]

Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study

Lymphopenic Community Acquired Pneumonia (L-CAP), an Immunological Phenotype Associated with Higher Risk of Mortality

Validation of the infectious diseases society of America/American Thoratic Society Minor Criteria for Intensive Care Unit Admission in Community-Acquired Pneumonia Patients Without Major Criteria or Contraindications to Intensive Care Unit Care

Validation and clinical implications of the IDSA/ATS minor criteria for severe community-acquired pneumonia

Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study

Lymphopenic community acquired pneumonia as signature of severe COVID-19 infection

Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19

Longitudinal characteristics of lymphocyte responses and cytokine profiles in the peripheral blood of SARS-CoV-2 infected patients

Using IL-2R/lymphocyte for predicting the clinical progression of patients with COVID-19

Laboratory findings associated with severe illness and mortality among hospitalized individuals with coronavirus disease 2019 in Eastern Massachusetts

Practices in triage and transfer of critically ill patients: A qualitative systematic review of selection criteria

Demographics Age, years (IQR) 156

Chronic heart disease

Chronic renal disease

Radiological findings Multilobar infiltrates

Data are summarized as n (%) or median (interquartile range), as appropriate. Abbreviations: ICU, intensive care unit