Predictors of the chest CT score in COVID-19 patients: a cross-sectional study

The study protocol and consent notes were reviewed and approved by the ethics board of our institute (approval code: TUMS.VCR.REC.1399.138). This study is a cross-sectional, observational, single-center study conducted between April and August 2020 at a tertiary medical center. All referred patients to the COVID-19 clinic of Imam Khomeini Hospital, university hospital, were evaluated for eligibility of participation in the study. Patients with mild, moderate, or severe clinical symptoms suggestive of COVID-19 who had positive RT-PCR (226, 47.2%) confirming COVID-19 or suggestive chest CT scan (252, 52.8%) were included after signing a written informed consent form. Patients with history of previous lung diseases excluded from study. Four physicians did the patient examination and data registration, and five medical residents followed up with the patients through telephone calls. Both hospitalized (after discharge) and outpatient participants were followed weekly through telephone calls until symptoms resolved. Treatment regimens were based on the latest version of the national protocol of COVID-19.Hydroxychloroquine was used as the primary therapeutic option in the outpatient setting; hydroxychloroquine, lopinavir/ritonavir (Kaletra), atazanavir (nonboosted with ritonavir or cobicistat) were administered for hospitalized patients, and in case of severe hypoxemia (not attaining SpO2 of > 88% with reservoir mask), corticosteroid was given.

Following clinical data were collected in the specific forms for each patient, including: (a) demographics information: (age and sex); (b) vital signs: temperature (T-Celsius), oxygen saturation (SpO₂), respiratory rate (RR per minute), and pulse rate (PR per minute); (c) symptoms: myalgia, generalized weakness, fever, chills, headache, chest pain, dyspnea, sore throat, cough, sputum, loss of appetite, loss of taste, anosmia; (d) comorbidities: hypertension (HTN), diabetes mellitus (DM); (e) laboratory data: white blood cell (WBC-cell/mm3), lymphocyte count (cell/mm3), platelet(cell/mm3), C-reactive protein (CRP-mg/L), erythrocyte sedimentation rate (ESR-mm/hr), and lactate dehydrogenase (units/L); (f) admission status: inpatient or outpatient; (g) Intensive care unit (ICU) admission; (h) death; (i) radiologic findings (will be mentioned in following section); (j) length of hospitalization (day) and recovery time (day). The recovery time was defined as the patient's subjective statement indicating no symptoms other than his/her baseline.

All chest CT scans were performed on either lightspeed 64-detector CT (GE Healthcare) or the 16-slice (Siemens SOMATOM Emotion) MDCT scanner with patients in the supine position at full inspiration breath-hold. The leading scanning parameters were as follows: 120 kVp tube voltage; 50–150 mAs tube current; 0.75 s tube rotation time; 0.5–0.75 s gantry rotation time; 2–3-mm section thickness; and 0.6–2 mm beam collimation.

The visual chest CT interpretation was performed using a single, experienced (10 years) thoracic radiologist. The radiologist was blinded to clinical and laboratory data of participants and reviewed on both lung and mediastinal window settings. The presence of CT features including (a) predominant pattern of lesions: ground-glass opacification (GGO), consolidation, or mixed GGO and consolidation; (b) dominant distribution of lesions: central, peripheral, diffuse, or peri- broncho vascular; (c) shape of lesions: round, elongated, wedged, or confluent; (d) additional findings: crazy paving pattern, reverse-halo sign, interlobular septal thickening, linear opacities combined, air bronchogram sign, tree in bud, adjacent pleural thickening, pleural effusion, pericardial effusion, lymphadenopathy, and pulmonary emphysema, was reported as defined in Fleischner Society Glossary of terms for Thoracic Imaging [17].

To quantify the extension of lung lesions, a scoring system as follows was used: each of the five lobes of lungs visually scored from 0 to 5 (0, no involvement; 1, < 5% involvement; 2, 5–25% involvement; 3, 26–49% involvement; 4, 50–75% involvement; 5, > 75% involvement). Then, the total chest CT score was calculated by the sum of each lob's scores, ranging from 0 to 25 [18].

The statistical analysis was performed using SPSS version 16 (SPSS Inc. Chicago, IL). Continuous variables were presented as mean (standard deviation), and categorical variables as frequency and percentages. Variables were tested for normality using Kolmogorov–Smirnov test. Normally distributed continuous variables were analyzed using the independent sample t-test; otherwise, the Mann–Whitney U test was used. A Chi-square test was employed for nominal variables. P values less than 0.05 were considered statistically significant.

A total of 478 participants were recruited in the current study with convenient sampling. Of those, 353 (73.6%) participants were admitted to the hospital, and among hospitalized participants, 42 (8.7%) were admitted to ICU. The patient's mean age was 53.92 (15.4), and 267 (55.8%) was male. The most common complaints of patients were myalgia (60.4%), fever (59.4%), dyspnea (57.9%), chills (49.5%), and generalized weakness (45.3%). Demographics, initial signs, symptoms, and laboratory data of all participants are listed (Table 1).

The most common patterns of pulmonary involvement were GGO (58.5%) and consolidation 99 (20.7%) with peri-broncho-vascular (33.7%), peripheral (33%), and diffuse (32%) distribution, and the most common shape of lesions were confluent (47.2%). Detailed chest CT scan findings are presented (Table 2).

Our results showed that the total chest CT score was significantly higher in patients with anosmia (mean of 12.46 ± 7.73 vs. 8.73 ± 8.33; P = 0.01), and had a significant positive association with RR with a cut point of 25 (mean of 8.37 ± 7.94 vs. 12.37 ± 9.23; P = 0.001), CRP with a cut point of 90 (mean of 7.82 ± 7.89 vs. 10.83 ± 8.67; P = 0.002), WBC with a cut point of 10,000 (mean of 8.70 ± 8.33 vs. 12.02 ± 8.17; P = 0.009) and negative association with SpO2 with a cut point of 93 (mean of 9.75 ± 8.54 vs. 7.76 ± 7.58; P = 0.04). Chest CT score was also associated with a higher risk of ICU admission (mean of 11.10 ± 9 vs. 7.71 ± 7.88; P = 0.003), longer hospital stay (mean of 9.08 ± 8.23 vs. 12 ± 9.21; P = 0.037) and recovery period (mean of 7.28 ± 8.02 vs. 9.29 ± 8.09; P = 0.009) (Table 3).

Further analysis showed consolidation on chest CT scan had a significant association with initial lower SpO2 (P = 0.001), higher risk of ICU admission (P = 0.005), extended hospitalization (P = 0.003), and longer recovery time (P = 0.008) (Table 4).

In line with previous studies, our results showed that higher initial WBC and CRP are associated with more severe cases of COVID-19 patients, as WBC > 1000 cell/mm3 and CRP > 90 mg/L were associated with higher chest CT scores. Similar to our findings, Salvatore et al. reported hospitalized and critical COVID-19 patients had higher CRP, leukocyte count, neutrophils, LDH, D-dimer, and troponin [19]. Zhang et al. also reported that chest CT score positively associated with WBC count, CRP, ESR, procalcitonin, and abnormal coagulation function [20].

Among several clinical symptoms on the first visit, just anosmia was associated with extended pulmonary involvement. A literature review did not show any previous report of extended lung involvement in patients with anosmia, and our study is the first to report this correlation. Association of anosmia and pulmonary involvement can be justified by the roll ofchemokines, as previous reports defined CXCL10 contribution in both cytokine storm of COVID-19 patients causing acute respiratory distress syndrome (ARDS) and demyelination process of the olfactory nerve causing anosmia [21].

Several clinical signs of patients on the first visit, including higher RR and lower SpO2, were associated with higher total chest CT scores. Lower SpO2 was also associated with consolidation opacities as the most severe lesion of COVID-19 in chest CT scan. In line with our findings, Kunwei Li et al. also showed that patients with RR ≥ 30 times/min, and SpO2 ≤ 93% as categorized in severe type patients, had significantly higher total chest CT scores than common type patients [22]. Another study also confirmed that severe/critical patients with respiratory rate ≥ 30 times/min and SpO2 of 93% or less in a resting state had higher total CT scores than ordinary COVID-19 patients [23]. Aalinezhad [24] and Osman[25] et al. also reported higher chest CT score is inversely associated with O2 saturation. Furthermore, a multicenter cohort study demonstrated that consolidation in upper lungs on the initial chest CT of COVID-19 patients was associated with increased odds of adverse endpoints, including SpO2 < 93% and partial arterial pressure of oxygen less than 60 mm Hg on room air [26].

We found that ICU admission, more extended hospitalization, and recovery period were associated with higher total CT scores and consolidation opacities in chest CT of these patients. Similarly, a previous study showed a higher pulmonary score ≥ of 8 accompanied by age ≥ 53, and SpO₂ ≤ 91 predicts ICU admission and mortality [27]. Deepak Nagra et al. also reported a higher lung opacification score is a reliable predictor of ICU admission for COVID-19 patients [28]. Another study on baseline chest CT scans and clinical and laboratory data of 72 patients admitted with COVID-19 pneumonia showed lung severity score >  was associated with a significantly lower recovery rate and discharge and extended hospitalization in patients admitted for COVID-19 pneumonia [29]. Similar to our findings, Ahlstrand et al. previously reported that chest CT score at hospital admission correlates closely with hospital length of stay and ICU admission [30]. CT severity score combined with age and history of at least one underlying disease had 79.7% sensitivity and 65.5% specificity in predicting the adverse outcomes in a previous study [31].

Initial assessment of patients in the clinic determines the treatment plan, so we assessed lung involvement based on initial signs, symptoms, and laboratory tests. Taken together, we found that patients with anosmia, higher respiratory rate, WBC count, or CRP, and lower SpO2 had extended pulmonary involvement of COVID-19 pneumonia, which can cause adverse outcomes, including more extended hospitalization and recovery period. Hence, those patients should be prioritized for greater attention and intensive care.

As recovery time was defined as a subjective statement of patients, this could have resulted in a conclusion error; inviting the patients to the clinic for evaluation could have been more precise to confirm recovery. In addition, in this study, we assessed just the presence of symptoms, not their severity; we recommend further studies assessing the severity of symptoms and their correlation with lung involvement or adverse outcomes.