Besides being a new and unknown disease, the increasing number of COVID-19 patients in a short time caused heavy burden and stress on healthcare providers [5]. The lack of human resources in the field of health, the shortage of hospital bed capacities and limited financial resources have made it necessary to select severe patients who may have a serious course to be provided health services first. Therefore, we developed a simple, accessible, easy to use calculation tool that makes physician’s decisions on the management of COVID-19 patients easier and provides ways of recognizing severe illness requiring ICU by using available and accessible values of patients on admission. The physician who first evaluated patient in primary care hospitals or in emergency department can decide where the patient should be followed-up as an inpatient or as an outpatient. The calculation tool also helps the physicians who are working in neighboring provinces and districts to decide whether patients will be transferred to tertiary hospitals or not.
The strongest calculation tool was constructed with two clinical (sex and oxygen saturation) and seven laboratory parameters (hemoglobin level, platelet count, GFR, AST and PCT level, ferritin and D-dimer) and it exhibited very strong distinctive power in the prediction of severe illness upon admission. It can identify a small proportion of cases whose illness will progress with a positive predictive value of 74.3% and a very high sensitivity of 92%. To our knowledge, it has the strongest discrimination properties for predicting severe cases of COVID-19 in the literature. These parameters included in ACCSES actually reflect the multi-organ functions and represent almost the whole human body including pulmonary, renal, hepatic and hematological system. Additionally, it contains inflammatory indicators. Therefore, it provides the opportunity to make an integrated decision about the patient and results in accurate estimation.
There is a number of studies investigating severity [6-12] and mortality [8,13-16] predictors in COVID-19 infection. Some of the studies reported that they developed scores or models for the prediction of disease severity [6-12]. Shi et al. defined a host risk score consisting of three parameters (age, sex, hypertension) and reported an increase in severe infection risk with the increasing score point (from 0 to 3) [12]. However, this scoring system does not determine a specific risk for each patient. Liang W et al. reported that they developed and validated a clinical risk score named as COVID-GRAM with ten parameters (chest radiographic abnormality, hemoptysis, dyspnea, age, unconsciousness, number of comorbidities, cancer histories, neutrophil-to-lymphocyte ratio, LDH, and direct bilirubin) [6]. The mean AUC was reported as 0.88 (95% CI, 0.85-0.91) in the development group and as 0.88 (95% CI, 0.84-0.93) in the validation group. Zhou et al. also reported that they developed a nomogram to predict severe patients. Seven clinical variables (body temperature at admission, oxygen saturation, cough, dyspnea, hypertension, cardiovascular disease, chronic liver disease, and chronic kidney disease on admission) were included in the nomogram. [7]. The nomogram was reported to have a good discrimination with an area under the curve of 0.862, C-index of 0.863 (95% confidence interval, 0.801–0.925). National Early Warning Systems 2 (NEWS2), an early warning systems used for prediction of cardiac arrest risk, unexpected intensive care unit admission or death, was recommended by the National Institute of Clinical Excellence (NICE) for the prediction of critical patients in COVID-19. However, Sze S et al. reported that NEWS2 score does not predict disease severity, no difference was detected between the groups in terms of NEW2 score [17].
Levy et al. established some clinical predictive tools consisting of 6 factors (blood urea nitrogen, patient age, absolute neutrophil count, red cell distribution width, oxygen saturation, and serum sodium) which resulted in an AUC of 0.86 to predict 7-day mortality. In another study developing models, clinical (age, hypertension, and coronary arterial disease) and laboratory (age, high sensitive CRP, oxygen saturation, neutrophil and lymphocyte count, D-dimer, AST and GFR) mortality-prediction models were developed. The AUC values were reported as 0.83 (95% CI 0.68-0.93) and 0.88 (95% CI 0.75-0.96) for the clinical model and laboratory model, respectively [15].
Age was reported as a predictor for severe COVID-19 infection [15,16,11,18-21]. In our study, severe cases were detected to be older than non-severe cases, and age was included in the score development process, however, it could not enter the final calculation tool. The association between comorbid diseases and the development of severe infection was indicated in previous studies [22,15,11,16,12,13] Some were the severity and mortality risk scores [15,11]. Similarly, although we found that comorbid diseases were higher in severe cases than non-severe patients, and severe patients had a higher number of comorbidities than non-severe cases, comorbidities were not identified as optimal predictors during the development process of the calculation tool.
Whether sex difference creates a predisposition to COVID-19 infection or severe form of the disease was discussed in a few studies [23-25]. It was emphasized that higher smoking rates or higher angiotensin-converting enzyme-2 (ACE2) expression in men may be a predisposing factor [23]. A current study from the United Kingdom (UK) reported that men had a higher risk for COVID-19 infection than women in a large population of 17,278,392 adults (fully adjusted hazard ratio [HR] 1.59, 1.53-1.65) [23]. In our study, male sex was found as a predictor for severe outcome and entered the final version of the score. Similar to our study, Shi et al. reported male sex as an independent factor (OR 3.68, 95% CI 1.75-7.75) for severe disease and included it in the host risk score [15].
Our study has some limitations. Firstly, although the study population is enough for the development of severity score, it was not sufficient to validate the results of study population. Maybe further studies with large number can validate ACCSES Calculation Tool in a different study population. Secondly, we can perform the study in single center. The more extensive, multicenter and large sample size studies will be better to represent the whole population.