CO-RADS is an alternative COVID-19 pneumonia classification system to the Radiology Society of North America (RNSA) classification [5, 9]. In one study, CO-RADS was even reported to have better diagnostic performance the system recommended by RSNA .
In our study, a significant difference was found between the CO-RADS groups in terms of age. This significance resulted from the difference between the CO-RADS groups 1 and 5 and groups 4 and 5. Patients in CO-RADS 5 represented the oldest group. We can state that age is an important factor in the lung involvement of the disease. In a meta-analysis study, it was reported that clinical and laboratory findings showed the course of severe disease, especially in elderly patients . In another study, it was determined that the disease progressed differently in different age groups . This was mostly attributed to the weakness of immunity .
In the current study, the significant difference between the CO-RADS groups in terms of ICU transfer resulted from the comparison of the groups 1 and 3. Twenty percent of the patients in the CO-RADS group 3 were transferred to ICU. From this, we can conclude that despite their low risk of COVID-19, it is important to follow up patients in the CO-RADS group. This group should perhaps be regarded as the unstable group. Currently, there are not sufficient data on this subject in the literature, but more studies will likely be carried out in the future.
When we examined the relationship between CO-RADS and oxygen saturation, we determined that saturation was lower in cases with more lung involvement, as expected. There are new treatment options available in the literature in patients with hypoxia, and one of them is tocilizumab . Although tocilizumab is an effective treatment for hypoxia, it is a risky agent in terms of its side-effect profile. In particular, the risk of hepatotoxicity has been previously reported [14, 15]. Therefore, the CO-RADS classification can be used as a parameter in determining treatment indication together with saturation values. The clinician may consider the CO-RADS classification in making a treatment decision.
Considering the relationship between CO-RADS and creatinine, significant differences were observed between the CO-RADS groups 1 and 5 and groups 4 and 5. Creatinine may be an important marker in lung involvement, which is known to be associated with kidney functions. In one study, creatinine values were found to be significantly higher in critically ill patients who were hospitalized , which is consistent with our study.
In relation to LDH, we observed significant differences between the CO-RADS group 3 and 5, groups 4 and 5, and groups 1 and 5. The highest LDH value was in the CO-RADS 5 category, and therefore LDH can distinguish this group from all the other categories. LDH can be considered as an important marker in parenchymal involvement, especially in the presence of bilateral and diffuse involvement. In particular, LDH values above 200 can support CO-RADS 5. Since LDH is a parameter associated with tissue damage, it can be stated that tissue damage is more intense in cases classified as CO-RADS 5. We consider that LDH can be used to predict the long-term outcomes of patients in this group.
When the relationship between CO-RADS and D-dimer was analyzed, there were significant differences between the CO-RADS groups 3 and 5 and groups 1 and 5. The D-dimer level was the highest in the CO-RADS group 5. From this, we can conclude that one of the causes of diffuse lung involvement is the tendency to clot. In a meta-analysis study, lymphopenia, thrombocytopenia and high CRP, LDH and D-dimer values were found to be associated with advanced disease . However, to our knowledge, there is no other that has compared the laboratory tests of patients according to their CO-RADS groups. In this respect, our study makes a contribution to the literature.
The hs-cTnT level was the highest in the CO-RADS group 5 compared to the remaining groups. Although our study did not evaluate the myocardial involvement of COVID-19, this result suggests that patients in the CO-RADS 5 group are at higher risk of cardiac involvement.
The highest CRP value was also observed in the CO-RADS group 5. Previous studies showed that elevated CRP levels were generally correlated with the severity of the disease at the time of detection and lung lesions [18, 19]. Therefore, our CRP results are in agreement with the literature.
In our study, we also grouped the parenchymal involvement of the patients in terms of ground glass, mixed and consolidation patterns, since we consider that the CT pattern should have a place in the scoring system. For this purpose, we obtained the structured total CT score value in the regression analysis performed with the total CT scoring  and the scores obtained from the pattern groups. In our study, when the total CT score and the structured total CT score values were compared according to the CO-RADS groups 3, 4 and 5, significant differences were found between the groups 3 and 5 in terms of the structured total CT score and between the groups 3 and 5 and groups 4 and 5 in terms of the total CT score value. In a previous study, scoring was performed according to < 25% lung involvement, mild; 25–50% involvement, moderate; and > 50% involvement, advanced . In our study, scoring was undertaken separately for each lung lobe, and therefore we consider this to be an innovative feature of our study since we obtained a structured total CT lung score by taking into account the pattern groups.
In another study using the scoring system utilized in our study, each lobe was evaluated separately, and then the total score of the whole lung was found to be positively correlated with RT-PCR positivity and male gender . In our study, no significant difference was found in terms of gender.
When the relationship of creatinine with the pattern groups was examined, creatinine was found to significantly differ between the normal lung parenchyma (pattern 0) and the ground glass pattern (pattern 1) groups. In addition to showing the extent of the disease, creatinine is also an effective parameter in showing the intensity of lung involvement. When the pattern groups were compared according to the total CT score values, no significant difference was observed. In a similar study conducted with 165 patients to evaluate lung involvement pattern groups, the CT patterns were divided into groups 0–4, representing normal findings, bronchopneumonia, organizing pneumonia, progressive organizing pneumonia, and diffuse alveolar damage . In our study, the pattern groups ranged from 0 to 3, and the CT pattern was categorized according to the density of the lesions in the lung. We consider that our own classification is more applicable and practical in daily clinical practice.
In a previous study, mortality was found to be significantly higher in patients with a higher total CT score value at the time of initial diagnosis . In another study, the total CT score value was found to be associated with requirement of hospitalization, requirement of intensive care unit, and one-month mortality . In our study, when the structured total CT score and the total CT score values were compared in terms of ICU transfer, survival, oxygen saturation and laboratory values, they were observed to be correlated with each other. The comparison of many parameters revealed that the structured total CT scoring system provided similar results to the total CT scoring system. Structured scoring may even be regarded a more accurate scoring system because it also includes the lung involvement pattern.
One of the limitations of our study is that we only included RT-PCR-positive patients. The alternative diagnosis group followed up in the CO-RADS category 2 was excluded. In addition, since we did not encounter pattern 3 in any of our patients, we did not include this pattern in our evaluations.