Our study enrolled 1087 patients with COVID-19 who were registry from a multi-center of Sichuan and Wuhan provinces where the outbreak risk levels are different. In primary study, based on patents’ demographic and clinical characteristics obtained on first admission, we established and validated a nomogram predicting the risk for admission to ICU through the LASSO and logistic regression analysis. The independently statistically significant factors including in the prediction model were age, respiratory rate, systolic blood pressure, smoking status, fever and chronic kidney disease. The validation of the model demonstrated its great performance using different statistical methods. As those factors could be obtained easily when first on admission, our nomogram is a convenient and valuable clinical warning tool for predicting ICU admission of COVID-19, especially in emergency department and even in community health center.
Most COVID-19 cases have mild disease with good prognosis, but some patients may develop severe respiratory distress syndrome and poor prognosis[16]. To mitigate the burden on the healthcare system and provide the best care for patients, it is necessary to effectively predict the prognosis of the disease[17]. A predictive model that combines multiple variables or features to estimate the risk of an infected person's poor outcomes can help healthcare staff classify patients' severity when allocating limited medical resources[18]. Previous studies have reported prediction models for diagnosing and prognosis of COVID-19, and for detecting the risk of being admitted to hospital for COVID-19. Chen et al constructed a diagnosis prediction models with 10 clinical factors based on 136 participants[19]. Wang et al enrolled 296 in-hospital COVID-19 patients and developed a clinical model for predicting the mortality of in-hospital COVID-19 patients[17]. Dong et al developed a scoring model to predict the progression risk with COVID-19 pneumonia base on 209 patients[20]. However, those proposed models are poorly reported and at high risk of bias, raising concern that their predictions could be unreliable when applied in daily practice for diagnosing. In a recent study, a risk score was reported to estimate the risk of critical illness of patients with COVID-19 based on 10 variables[21]. Although the study has modest sample size and satisfying performance, this scoring system was complicated with some laboratory examination data which cannot be obtained before admission or as soon as possible on admission. It is necessary to develop and validate a convenient prediction model for healthcare staff or emergency staff to use quickly and easily. Based on multi-center study from different cities and different severity of outbreak in Wuhan and Sichuan province, we constructed a warning model for predicting the risk of ICU admission. In our model, the independently statistically significant factors were age, respiratory rate, systolic blood pressure, smoking status, fever and comorbidity with chronic kidney disease, which could be obtained simply, practically, reliability, and fast. This prediction model could be used in prehospital care or emergency department, allowing medical staff to intervene at an early stage and determine their treatment location and the type of intervention. Our model showed a good discriminative ability and potential clinical benefit using different statistical methods. This prediction model is more practical to evaluate COVID-19 patients than other scoring tools and greatly improve the applicability and robustness of prediction models in routine care.
It identified that comorbidities played a key role in the prognosis of COVID-19. Cardiovascular system disease, especially hypertension, was reported to be one of the most important independent risk factors[22]. In this study, we observed the patients with presence of kidney disease more likely to be admitted to the ICU and it was an independent risk factor for ICU admission of COVID-19, suggesting that the patients with comorbidity of kidney disease on admission might have a high risk of deterioration[23, 24]. Previous study showed that kidney injury was associated with an increased risk of death in patients with influenza A virus subtype H1N1 and SARS[25–27]. Multiple organ involvement including the liver, gastrointestinal tract, and kidney have been reported during the course of SARS in 2003 and very recently in patients with COVID-19[28]. In our hypothesis, such patients might have a proinflammatory state with functional defects in innate and adaptive immune cell populations and were known to have a higher risk for upper respiratory tract infection and pneumonia. The 2019-nCoV itself may also cause kidney injury through multiple mechanisms: the 2019-nCoV may uses angiotensin converting enzyme 2 (ACE2) as a cell entry receptor and exert direct cytopathic effects on kidney tissue. It has been reoprted ACE2 expression in urinary organs (kidney) was nearly 100-fold higher than in respiratory organs (lung)[28]. Viral antigens or virus-induced specific immune effect mechanisms (specific T-cell lymphocytes or antibodies) deposits of immune complexes may damage the kidneys[29]. Early detection and treatment of renal abnormalities, including assess volume status and renal transplantation pressure, avoidance of nephrotoxic drugs and adequate hemodynamic support may help to improve the vital prognosis of COVID-19.
In most prognosis prediction models published, elder age, comorbidities, lactate dehydrogenase, lymphocyte and C-reactive protein were reported as risk factors for poor prognosis[20]. Some other indicators such as heart rate, breath rate, oxygen saturation, procalcitonin, procalcitonin, direct bilirubin, albumin, D-dimer levels, activated partial thromboplastin time, glomerular filtration rate, chest radiography (CXR) abnormality have controversial conclusions[30, 31]. Our study also demonstrated that the COVID-19 infected patients with elder age, especially greater than 65 years, would have worse prognosis than younger patients. In our study, fever (63.0% in training cohort, 66.0% in validation cohort), cough (59.8% in training cohort, 65.1% in validation cohort) and fatigue (37.1% in training cohort, 38.3% in validation cohort) were the most common symptoms. But of all the symptoms, only fever is an independent risk factor for prognosis, which is different from other studies. The reasons for the inconsistent reports of these models may be related to the risk of bias caused by the sample size and geographical differences of each model.
Our study has some limitations. First, the design was retrospective. Second, some cases had incomplete data on symptoms, laboratory tests, and imaging examinations, given the variation in the structure of electronic databases across different participating hospitals and an urgent data extraction schedule. Third, Sample size limit, future studies with larger sample sizes are warranted to validate our results. Fourth, severe patients were older than non-severe patients, and this difference in age may be a confounding factor. Fifth, although the study is multicenter, the results cannot be generalized to other populations. Sixth, we did not collect treatment-related data which may be critical to patient's outcome. However, all patients received treatment in accordance with the guidelines issued by the National health commission of China.