key: cord-0904235-xdxhsdyz
authors: Zhao, B.; Wei, Y.; Sun, W.; Qin, C.; Zhou, X.; Wang, Z.; Li, T.; Cao, H.; Wang, W.; Wang, Y.
title: Distinguish Coronavirus Disease 2019 Patients in General Surgery Emergency by CIAAD Scale: Development and Validation of a Prediction Model Based on 822 Cases in China
date: 2020-04-23
journal: nan
DOI: 10.1101/2020.04.18.20071019
sha: dc80e8daaa793db15b1c9fa8b8dc7482a3425dd7
doc_id: 904235
cord_uid: xdxhsdyz

IMPORTANCE In the epidemic, surgeons cannot distinguish infectious acute abdomen patients suspected COVID-19 quickly and effectively. OBJECTIVE To develop and validate a predication model, presented as nomogram and scale, to distinguish infectious acute abdomen patients suspected coronavirus disease 2019 (COVID-19). DESIGN Diagnostic model based on retrospective case series. SETTING Two hospitals in Wuhan and Beijing, China. PTRTICIPANTS 584 patients admitted to hospital with laboratory confirmed SARS-CoV-2 from 2 Jan 2020 to15 Feb 2020 and 238 infectious acute abdomen patients receiving emergency operation from 28 Feb 2019 to 3 Apr 2020. METHODS LASSO regression and multivariable logistic regression analysis were conducted to develop the prediction model in training cohort. The performance of the nomogram was evaluated by calibration curves, receiver operating characteristic (ROC) curves, decision curve analysis (DCA) and clinical impact curves in training and validation cohort. A simplified screening scale and managing algorithm was generated according to the nomogram. RESULTS Six potential COVID-19 prediction variables were selected and the variable abdominal pain was excluded for overmuch weight. The five potential predictors, including fever, chest computed tomography (CT), leukocytes (white blood cells, WBC), C-reactive protein (CRP) and procalcitonin (PCT), were all independent predictors in multivariable logistic regression analysis (p[≤]0.001) and the nomogram, named COVID-19 Infectious Acute Abdomen Distinguishment (CIAAD) nomogram, was generated. The CIAAD nomogram showed good discrimination and calibration (C-index of 0.981 (95% CI, 0.963 to 0.999) and AUC of 0.970 (95% CI, 0.961 to 0.982)), which was validated in the validation cohort (C-index of 0.966 (95% CI, 0.960 to 0.972) and AUC of 0.966 (95% CI, 0.957 to 0.975)). Decision curve analysis revealed that the CIAAD nomogram was clinically useful. The nomogram was further simplified into the CIAAD scale. CONCLUSIONS We established an easy and effective screening model and scale for surgeons in emergency department to distinguish COVID-19 patients from infectious acute abdomen patients. The algorithm based on CIAAD scale will help surgeons manage infectious acute abdomen patients suspected COVID-19 more efficiently.

Since the outbreak of coronavirus disease 2019 in Wuhan, China, which has been characterized as a pandemic by the World Health Organization on Mar 11, 2020, the cunning virus has drastically spread all over the world 1, 2 . Millions of people have been infected, resulting in tens of thousands died 3 . The still ongoing pandemic is not only a huge threat to the public physical health but also an acid test for the medical system regardless of developed counties or developing countries 4 . In addition to prevention, quick and accurate diagnosis of COVID-19 is one of the most important tasks currently.

The medical management of other diseases has been critically disturbed, especially for some with fever, the typical symptom of COVID-19 5 . There are numerous high-risk people in close contact with the confirmed patients. As we all know, interdicting transmission is the most effective way to control the epidemic of COVID-19. Under current situation for surgeons, infectious acute abdomen is still one of the most common surgery emergencies, the patients of which often have fever, diarrhea and other atypical symptom, and also have similar change of blood routine and other biochemistry items with COVID-19 for infectious or chemical peritonitis 6 . Hence, when surgeons managing patients of infectious acute abdomen, the typical symptoms and blood test indicators of COVID-19 are easy to be covered up. Currently, diagnosis of COVID-19 mainly depends on severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) nucleicacid detection 7 . However, the approach has defects of time-consuming and false-negative results 8 , which go against the urgency of emergency operation and the aim to avoid cross infection in the hospital, and therefore, there is a pressing need of an easier and more feasible method to distinguish the genuine COVID-19 patients among the infectious acute abdomen patients with mimical symptoms.

Based on the clinical data of 822 patients, 584 of COVID-19 and 238 of infectious acute abdomen, we compared the demographic, clinical, imaging and laboratory characteristics to obtain significant predictors of COVID-19.

Further, a prediction model to distinguish the two diseases was generated based on machine learning and presented in form of nomogram, which has a good discrimination performance in both training and validation cohort.

Ultimately, we offered a practical screening scale, named CIAAD scale, and an algorithm, including precaution advice for surgeons, in infectious acute abdomen patients encounters.

Ethical approval was obtained from the Ethics Committees of Peking Union Medical College Hospital and the Central Hospital of Wuhan for this retrospective stud. We brought 584 COVID-19 patients adopted into the Central Hospital of Wuhan between Jan 2, 2020 and Feb 15, 2020 into our . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . https://doi.org/10.1101/2020.04.18.20071019 doi: medRxiv preprint study, the diagnostic criteria of COVID-19 were positive RT-PCR results for SARS-CoV-2 or highly homologous viral gene sequencing results with SARS-CoV-2 of respiratory or blood samples 7 . Since the routine medical order of other diseases in Wuhan were severely disturbed by the epidemic, the clinical data of infectious acute abdomen patients were collected from 283 patients receiving emergency operation in Peking Union Medical College Hospital between Feb 28, 2019 and Apr 3, 2020. The inclusion criteria were: (1) fever; or (2) abnormal blood routine results or other infection indicators; or (3) signs of pneumonia. The patients with infectious acute abdomen adopted after Jan 20, 2020 were all tested for SARS-CoV-2 and none of them was positive.

Demographic, clinical, laboratory, treatment and outcome data of the COVID-19 and infectious acute abdomen patients were extracted from electronic medical system of the Central Hospital of Wuhan and Peking Union Medical College Hospital respectively.

Fever was defined as axillary temperature of at least 37.3℃. The chest CT scores were graded retrospectively by two radiologists back to back. Each lung is divided into the upper, middle and lower parts and the scoring criterion is ＜1/3 lung infected, 0 point; 1/3-2/3 lung infected, 1 point; ＞2/3 lung . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 23, 2020. . https://doi.org/10.1101/2020.04.18.20071019 doi: medRxiv preprint

The primary cohort of the whole 822 patients was divided into a training cohort and a validation cohort randomly by ratio of 2:1. The least absolute shrinkage and selection operator (LASSO) method, one of the most effective ones in regularized regression with great advantage when managing multicolinearity data, was used to select the most useful predictive variables for COVID-19 in the training cohort 9 .

We conduct multivariate logistic regression involving the potential predictors to further verify their predictive efficacy and build a nomogram to distinguish COVID-19 patients from infectious acute abdomen patients on the basis of multivariable logistic analysis in the training cohort. We named the nomogram as COVID-19 and Infectious Acute Abdomen Distinguishment (CIAAD) nomogram. Calibration curves were plotted to assess the calibration of the CIAAD nomogram and C-index was measured to quantify its discrimination performance. Then, CIAAD nomogram the training cohort was applied to the patients in the validation cohort and the calibration curve and C-index were derived on the basis of the regression analysis. The distinguishment capacity of CIAAD nomogram in both training and validation cohorts were also accessed by calculating the area under the receiver operating characteristic curve (AUC).

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 23, 2020. Decision curve analysis and clinical impact curves was conducted to evaluate the clinical practicability of CIAAD nomogram by quantifying the net benefits at different threshold probabilities in both training and validation dataset.

The corresponding score of each item in CIAAD nomogram was divided by 25 and rounded to obtain a simplified score. The simplified score are verified to have the same efficacy comparing the original nomogram. We subdivide the risk of COVID-19 into low (＜0.3), moderate (0.3-0.7) and high (＞0.7) risk.

The CIAAD Scale was formed on basis of the simplified scoring criterion and risk classification.

Categorical variables were expressed as numbers and percentage.

Continuous variables were expressed as medians with interquartile ranges.

Chi-square test and Mann-Whitney U-test were used to evaluate categorical and continuous data respectively. Statistical analysis was conducted with R software (version 3.6.1; http:// www.Rproject.org) and SPSS statistical software package (version 25.0). P＜0.05 was considered statistically significant.

A total of 822 patients, 584 COVID-19 patients without infectious acute . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 23, 2020. . https://doi.org/10.1101/2020.04.18.20071019 doi: medRxiv preprint abdomen and 238 infectious acute abdomen patients without COVID-19, were included in this study (Table 1) . Nearly 16% of the COVID-19 patients were severe or critical ( Figure 1 ). The disease spectrum of the infectious acute abdomen patients was principally acute appendicitis (60.5%), perforation 

All collected 40 variables were reduced to 6 potential predictors (abdominal pain, fever, chest CT, CRP, PCT and WBC) with nonzero coefficients in LASSO regression on the basis of 547 patients in the training cohort ( Figure 2 ).

It should be noted that 99.2% of the infectious acute abdomen patients had the symptom of abdominal pain, the proportion of which in COVID-19 patients was merely 0.2%, making it very likely to have overmuch weight in the model and go against the distinguish efficacy (Table 1) . Concerned above, abdominal . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 23, 2020. . https://doi.org/10.1101/2020.04.18.20071019 doi: medRxiv preprint pain was excluded from the potential predictors. AUC of the left five variables, fever, chest CT, CRP, PCT and WBC, was yielded respectively in the training cohort and validation cohort ( Figure S1 and S2).

To simplify the model, the concrete values of CRP, PCT and WBC were transformed into categorical variables (CRP and PCT: 1 for normal, 2 for high; WBC: 1 for low, 2 for normal and 3 for high) and the symptom fever was also defined as 0 for "no" and 1 for "yes". Multivariate logistic regression analysis was performed among the five variables and all of these potential predictors have great value in distinguishing COVID-19 from infectious acute abdomen (Table 2) . A risk score formula was preliminarily built to predict COVID-19 probability as follows: Logit (P = COVID-19) = 10.104 + 1.915×fever + 1.753×chest CT + (-2.508)×CRP + (-0.8)×PCT + (-1.836)×WBC. The CIAAD nomogram was generated on basis of the above result ( Figure 3A ).

The calibration curve of CIAAD nomogram for the risk of COVID-19 demonstrated good agreement between prediction and reality in the training cohort ( Figure 3B ). The C-index for the prediction nomogram was 0.981 (95% CI, 0.963 to 0.999) for the training cohort. Good calibration was also observed in the validation cohort with a C-index of 0.966 (95% CI, 0.960 to 0.972) ( Figure 3C ). The ROC analysis for the training and validation cohort yielded AUC values of 0.970 (95% CI, 0.961 to 0.982) and 0.966 (95% CI, 0.957 to . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . https://doi.org/10.1101/2020.04.18.20071019 doi: medRxiv preprint 0.975), which implied the prediction performance was favorable ( Figure 3D and 2E).

Decision curve analysis and clinical impact curves was conducted for the CIAAD nomogram in both training and validation cohort (Figure 4 ), demonstrating high clinical net benefit nearly over the entire threshold probability.

With the purpose of making our prediction model more concise and practical in surgery emergency, we simplified the scoring criterion of the CIAAD nomogram and create a brand new scale, named CIAAD scale ( Figure 5 ). In CIAAD scale, the lowest and highest score of this scale is 3.5 and 14.5. The item with a higher corresponding score is more common in COVID-19 patients, such as fever, abnormal chest CT, normal level of CRP and WBC. If the total score of a patient is less than 5, he/she gets low risk (＜30%) of confirmed COVID-19, and the risk rises to more than 70% as the total score reaches 7.

With the global outbreak of COVID-19, the latest sum of infected patients has exceeded 1.8 million, and humankind will face the threat all through the near future 2 . Vast medical resource has been put into rolling back the virus resulting in the treatment for many other diseases postponed. However, for surgeons confronted with patients of infectious acute abdomen urgent for emergency . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. 

Current reports of the COVID-19 cohorts show that respiratory symptoms, such as fever, cough and dyspnea, are the main clinical manifestations 10 .

Nevertheless, the digestive symptoms like diarrhea, nausea, vomiting and . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . https://doi.org/10.1101/2020.04.18.20071019 doi: medRxiv preprint COVID-19 cases from Wuhan showed that the positive rate for chest CT was 88%, comparing the 59% of RT-PCR 8 . Several prediction models based on the integration of demographic, clinical, imaging and laboratory variables have been developed to evaluate the disease risk or prognosis 17 . Unfortunately, the target population of published diagnostic models was patients presenting at fever clinic or ordinary patients suspected COVID-19 17 . It is not suitable for surgeons to borrow these models directly to screen infectious acute abdomen patients and our CIAAD nomogram and scale make up the pity.

The set of our model has the superiority of strong pertinence. The recommended user of CIAAD scale is surgeon in the emergency department and the recommended assessed population is infectious acute abdomen patients suspected COVID-19. To this end, we collected firsthand and high-quality data of COVID-19 patients and enrolled infectious acute abdomen patients strictly. In addition, by virtue of LASSO regression analysis, five quantifiable indicators were successfully selected. Though many variables like diabetes, cough and D-dimer varied considerably between COVID-19 and infectious acute abdomen patients, they were ruled out by LASSO regression analysis as overmuch weight or causing the prediction model cumbersome.

The selected indictors were all included in previous prediction models, which verified the prediction capacity of these variables from the side 17 .

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. Whereas, there is some inadequacy in our study as well. Firstly, the disturbance for routine medical work by the epidemic resulted in the appropriate lack of patients with both COVID-19 and acute abdomen. As the number of emergency operations of acute abdomen decreased sharply in Wuhan, the data of acute abdomen patients was from Peking Union Medical College Hospital, a renowned and leader hospital in China.

An algorithm helpful for allowing both a focused workup and expeditious therapy was given, including necessary prevention advice for medical staff with the guidelines published by World Health Organization for reference 18 ( Figure 6 ). What needs to be pointed out is that standard precautions are needed for all patients. For an infectious acute abdomen patient suspected COVID-19, the first step is to evaluate his/her surgical status and screen the patient by CIAAD scale. The urgency degree of surgical status determines whether the medical staff wait for the results of nucleicacid detection or take precautions according to our algorithm. Level of precautions adopted should be instructed by the risk degree of COVID-19 from CIAAD scale.

As the scale is harmless and the has net benefit nearly over the entire threshold probability according to the decision analysis curves, strong recommendation of our CIAAD scale and the algorithm was made to all surgeons in countries severely affected by the epidemic. With the wide use in lager population, efficacy of CIAAD scale will be further tested in a prospective . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . https://doi.org/10.1101/2020.04.18.20071019 doi: medRxiv preprint Transparency: The lead authors and manuscript's guarantor affirm that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained.

Dissemination to participants and related patient and public communities: No patients or members of the public involved in the design, or conduct, or reporting, or dissemination plans of this study.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. https://apps.who.int/iris/rest/bitstreams/1272420/retrieve. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . Binomial Deviance 23 20 18 18 18 11 8 3 2 1 1 1 1 1 1 1 1 1 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 23, 2020. . is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 23, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 23, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 23, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 23, 2020. . https://doi.org/10.1101/2020.04.18.20071019 doi: medRxiv preprint

A novel coronavirus genome identified in a cluster of pneumonia cases -Wuhan

Director-General's opening remarks at the media briefing on COVID-19

Coronavirus disease (COVID-19) outbreak situation

The novel coronavirus originating in Wuhan, China: challenges for global health governance

Clinical Characteristics of Coronavirus Disease 2019 in China

Sabiston Textbook of Surgery: The Biological Basis of Modern Surgical Practice

National Health Commission of the People's Republic of China. Chinese management guideline for COVID-19 (version 7.0)