key: cord-0894138-0xhc816b
authors: Zeng, Zihang; Ma, Yiming; Zeng, Huihui; Huang, Peng; Liu, Wenlong; Jiang, Mingyan; Xiang, Xudong; Deng, Dingding; Liao, Xin; Chen, Ping; Chen, Yan
title: Simple nomogram based on initial laboratory data for predicting the probability of ICU transfer of COVID‐19 patients: Multicenter retrospective study
date: 2020-06-30
journal: J Med Virol
DOI: 10.1002/jmv.26244
sha: 790b3e146728c9edd5fbdfd33f52a7cf50a98e8f
doc_id: 894138
cord_uid: 0xhc816b

This retrospective, multicenter study investigated risk factors associated with intensive care unit (ICU) admission and transfer in 461 adult patients with confirmed coronavirus disease 2019 (COVID‐19) hospitalized from January 22 to March 14, 2020 in Hunan Province, China. Outcomes of ICU and non‐ICU patients were compared, and a simple nomogram for predicting the probability of ICU transfer after hospital admission was developed based on initial laboratory data using a Cox proportional hazards regression model. Differences in laboratory indices were observed between patients admitted to the ICU and those who were not admitted. Several independent predictors of ICU transfer in COVID‐19 patients were identified including older age (≥65 years) (hazard ratio [HR]=4.02), hypertension (HR=2.65), neutrophil count (HR=1.11), procalcitonin level (HR=3.67), prothrombin time (HR=1.28), and d‐dimer level (HR=1.25). Lymphocyte count and albumin level were negatively associated with mortality (HR=0.08 and 0.86, respectively). The developed model provides a means for identifying, at hospital admission, the subset of patients with COVID‐19 who are at high risk of progression and would require transfer to the ICU within 3 and 7 days after hospitalization. This method of early patient triage allows more effective allocation of limited medical resources. This article is protected by copyright. All rights reserved.

In December 2019, an outbreak of a new disease that was eventually named coronavirus disease 2019 (COVID- 19) was reported in Wuhan (Hubei Province, China). 1 The spread of COVID-19 was designated as a pandemic by the World Health Organization (WHO) on March 11, 2020. As of April 3, 2020, there were 960,000 confirmed cases of COVID-19 in 204 countries, with 50,000 deaths. The causative Accepted Article agent of COVID-19 was officially named severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) on February 11, 2020 by the WHO. 2 COVID-19 can progress rapidly in critically ill patients; a meta-analysis of 50,466 patients with COVID-19 showed that 18.1% were severe cases. 3 Without timely and appropriate care, the outcome of patients can be very poor. 4 It is therefore essential to identify at an early stage those patients who will require admission to the intensive care unit (ICU). However, ICUs in hospitals around the world are over capacity because of the scale of this pandemic. 4 ,5 A simple model for predicting COVID-19 progression can guide clinical decision-making with regard to critical-care capacity and resource allocation. Although numerous differences in laboratory indices have been reported between patients with and those without ICU care, 6, 7 there are no models for predicting the risk of progression and need for ICU admission based on these data.

To address this issue, this study investigated the laboratory indices and factors associated with ICU admission and transfer in COVID-19 patients in Hunan Province, China, and used this information to develop a nomogram for predicting the probability of ICU transfer after hospital admission. This article is protected by copyright. All rights reserved.

The ethics committee of Xiangya Hospital (Changsha, China) approved the study protocol (2020-010). Written, informed consent was obtained from participants or their families for retrospectively collected data. This article is protected by copyright. All rights reserved.

Patients were divided into 2 groups based on ICU transfer status-namely, the ICU care (ICUC) group (n=55) and non-(N)ICUC group (n=406 patients). Patients were followed up for 30 days after hospital admission, until the end of the observation period (March 29, 2020), or until referral to the ICU. Time-to-event was defined as the time from hospitalization to ICU admission.

The Cox proportional hazard model was used to generate nomograms for predicting the risk of ICU transfer using rms (www.rms.com). A score based on regression coefficients was assigned to factors that would be convenient for clinical decision-making. The discrimination and predictive abilities of the nomogram were assessed with Harrell's concordance index (C-index), where a larger index reflected a more accurate prediction of prognosis. To validate the nomogram, calibration curves plotted based on nomogram-predicted and actual probabilities of ICU transfer were analyzed.

Descriptive data are presented as interquartile range (IQR) for continuous variables and as numbers (%) for categorical variables. Differences in the distribution of patient characteristics between the ICUC and NICUC groups are presented with This article is protected by copyright. All rights reserved.

95% confidence intervals (CIs). Continuous variables were analyzed with the Mann-Whitney U test, and categorical variables were compared between subgroups with the χ² test or Fisher's exact test. The Cox proportional hazards model was used to determine hazard ratio (HR) and 95% CI between individual factors and ICU transfer status. Statistical analyses were performed using SPSS v25.0 (IBM, Armonk, NY, USA) or R v3.6.3 (R Foundation for Statistical Computing, Vienna, Austria).

A total of 461 hospitalized patients with confirmed COVID-19 were included in this study ( Table 1 ). The median age of patients was 45 years (IQR, 34.5-57 years), and 239 (51.48%) were male. The most common comorbidity was hypertension (n=84 

A total of 55 (11.93%) patients were admitted to the ICU; these patients (ICUC n=406). Differences in laboratory findings between the 2 groups are summarized in Table 3 . White blood cell (WBC) and neutrophil counts were higher in the ICUC group than in the NICUC group (P<0.001). Lymphocyte and platelet counts were predictors of ICU transfer. These factors were used to develop a nomogram for predicting the probability of ICU transfer after hospital admission (Fig. 1) . Each point This article is protected by copyright. All rights reserved.

of an independent variable was determined according to the intersection of the vertical line drawn from the variable to the point axis. The total risk score was then calculated by adding each variable point; the probability of ICU admission was obtained from the total point axis. The C-index of the nomogram was 0.848. Calibration curves indicated that the probability predicted using the nomogram showed good concordance with real-world data (Fig. 2) .

In this study, blood biochemistry, coagulation function, and infection-related biomarkers were examined in 461 adult patients with laboratory-confirmed COVID-19, and several risk factors associated with ICU admission of COVID-19 patients using the Cox proportional hazard model were identified. Based on these factors, a simple nomogram was developed for predicting the probability of ICU transfer 3 and 7 days after hospitalization. This is the first model for predicting the probability of ICU transfer in patients with COVID-19.

The clinical presentation of COVID-19 varies from asymptomatic to mild upper respiratory tract symptoms, severe viral pneumonia with respiratory failure, and death; 9,10 20.3% of patients with COVID-19 require ICU admission. 11 Predicting which patients are at risk of COVID-19 progression and thus require ICU transfer at the time of hospital admission could improve patient outcome by ensuring that those Accepted Article who are most likely to become critically ill can receive appropriate care early on, thereby alleviating pressure on ICU capacity.

In terms of laboratory findings, high levels of ESR and CRP as well as hypoalbuminemia and lymphopenia were observed in over half of the patients, which is consistent with other reports. 12 There were also numerous differences between the ICUC and NICUC groups. The ICU patients had increased WBC count, neutrophilia, lymphopenia, thrombocytopenia, and elevated indices of liver damage (AST and LDH), renal dysfunction (urea), infection (CRP, PCT, and ESR), and coagulation function (PT and D-dimer), which is in agreement with previous studies. 6, 7, 13 The uniand multivariate Cox regression models showed that lymphopenia was a significant predictor of ICU transfer; lymphopenia has been shown to be a feature of severe COVID-19. 14,15 These data suggest that severe cases of SARS-CoV-2 infection are likely associated with bacterial infection; immune deficiency; activation of coagulation; and impaired myocardial, hepatic, and kidney functions.

Immune dysregulation was found to be associated with the critical illness caused by SARS-CoV-2 infection. Elevated infection-related indices and poor coagulation functions were also found to be risk factors for ICU transfer, as previously reported. 13, 16 CRP is widely used as a biochemical indicator for inflammation because it reflects the acute severe systemic inflammatory response caused by viral infections; it has been suggested that the cytokine storm is involved in This article is protected by copyright. All rights reserved.

severe disease. 17 Interestingly, a higher PCT level was associated with a higher probability of ICU transfer in the present study. As demonstrated in a recent meta-analysis, 18 PCT level may predict evolution to a more severe form of disease.

Moreover, moderately or markedly increased D-dimer levels suggest the activation of coagulation in patients who are later admitted to the ICU. Platelet-to-lymphocyte ratio was shown to be associated with prognosis in patients with COVID-19; 19 here it was shown that thrombocytopenia was related to COVID-19 severity.

A nomogram was established to guide clinical decision-making with regard to critical care resource allocation. Five variables were included in the model including lymphocyte and platelet counts and AST, LDH, and CRP levels. The C-index of the nomogram was 0.848, suggesting that the model was effective in identifying patients at risk for ICU transfer. In addition, the calibration curves indicated that the probability of ICU transfer predicted using the nomogram matched well with real-world data.

This study had 2 major limitations. Firstly, because of its retrospective design, not all laboratory tests were performed in all patients, which reduced the sample size for constructing the nomogram. Secondly, prospective studies were not carried out to validate the model; this will be done in the future in a larger cohort of COVID-19

patients.

This article is protected by copyright. All rights reserved.

In conclusion, although it requires verification and validation in a larger number of patients, the predictive model developed in this study can aid physicians in identifying early on (ie, at the time of admission) patients who are at risk of COVID-19 progression and therefore require transfer to the ICU, so that medical resources can be more effectively allocated and patient outcomes improved. ALT, alanine aminotransferase; APTT, activated partial thromboplastin time; AST, aspartate aminotransferase; CK-MB, creatine kinase muscle-brain isoform; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; ICU, intensive care unit; IQR, interquartile range; LDH, lactate dehydrogenase; PT, prothrombin time. This article is protected by copyright. All rights reserved.

A novel coronavirus from patients with pneumonia in China

Coronaviridae Study Group of the International Committee on Taxonomy of Viruses. The species severe acute respiratory syndrome-related coronavirus: Classifying 2019-nCoV and naming it SARS-CoV-2

Clinical characteristics of 50466 hospitalized patients with 2019-nCoV infection

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

Critical care utilization for the COVID-19 outbreak in Lombardy, Italy: Early experience and forecast during an emergency response

Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan, China

Clinical Management of Severe Acute Respiratory Infection When Novel Coronavirus (nCoV) Infection is Suspected: Interim Guidance

(ncov)-infection-is-suspected

Zhong NS; China Medical Treatment Expert Group for Covid-19. Clinical characteristics of coronavirus disease 2019 in China

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: A descriptive study

LANCOVID-19). Clinical, laboratory and imaging features of COVID-19: A systematic review and meta-analysis

Novel coronavirus infection (COVID-19) in humans: A scoping review and meta-analysis

Clinical characteristics of refractory COVID-19 pneumonia in Wuhan, China

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

Analysis of factors associated with disease outcomes in hospitalized patients with 2019 novel coronavirus disease

Abnormal coagulation parameters are associated with poor prognosis in patients with novel coronavirus pneumonia

Risk factors associated with acute respiratory distress syndrome and death in patients with coronavirus disease

Procalcitonin in patients with severe coronavirus disease 2019 (COVID-19): A meta-analysis

This article is protected by copyright. All rights reserved. Accepted Article Lymphocyte count ×10 9

Creatine kinase, U/l 66

Lactate dehydrogenase, U/l 172

Infection-related indices CRP, mg/l 5

Lymphocyte count ×10 9

Creatine kinase

Lactate dehydrogenase

This article is protected by copyright. All rights reserved

The authors thank all study participants. This work was supported by Emergency

Hospital (no. 502701002).

The authors declare that they have no competing interests.

The study was designed by Z Zeng, Y Ma and Y Chen. P Huang, W Liu, M Jiang, X Xiang, D Deng, X Liao and P Chen were responsible for data collection. Z Zeng, Y Ma, H Zeng were responsible for the data analysis. Z Zeng drafted the first draft and all other authors provided guidance on revision of the manuscript. All authors read and approved the final manuscript. This article is protected by copyright. All rights reserved.

 Figure 1 . Nomogram predicting the probability of ICU transfer. All 5 prognostic factors must be available for this model to be used. AST, aspartate aminotransferase; CRP, C-reactive protein; LDH, lactate dehydrogenase; PLT, platelet. Urea ALT, alanine aminotransferase; APTT, activated partial thromboplastin time; AST, aspartate aminotransferase; CI, confidence interval; CK-MB, creatine kinase muscle-brain isoform; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; HR, hazard ratio; ICU, intensive care unit; LDH, lactate dehydrogenase; PT, prothrombin time.