Chest radiographic findings in high-risk Covid-19 pneumonia patients

Introduction: The severity of Covid-19 pneumonia has shown a positive association with co-existing risk factors. However, the exact nature of lung involvement in high-risk Covid-19 patients is yet to be resolved. Therefore, we evaluated chest X-ray (CXR) features, temporal progression, and the factors associated with CXR severity in high-risk patients. Methods: Chest X-rays (n=289) of Covid-19 infected high-risk adults (n=228) treated at the Base Hospital Homagama were evaluated to record CXR features, their temporal progression, CXR severity score and the patient outcomes. Results: The evaluated patients (48.2% men) were in mean age of 59 (SD 15) years. The most frequent CXR features were patchy ground-glass (GG) opacities (49%) and patchy consolidations (CON) (42%); They showed bilateral (100%) involvement, superoinferior gradient (100%) and diffuse (27%), peripheral (18%) or perihilar (10%) distribution. CON was the predominant opacity among the non-survivors and GG 2 among the survivors (χ =14.73; p =0.001). Right lung predominant (28%) asymmetrical lung involvement was more frequent than bilateral symmetrical (16%) or left lung predominance (7%). Progression to fatal disease was significantly higher when the lung involvement is asymmetrical: right predominance: ODDs: 0.502; p =0.023; left predominance: ODDs:0.268; p =0.002. The CXRs were frequently normal in early (66%) 2 and progressive (56%; χ =36.64; p <0.001) stages than in peak or resolving stages. The predictors of CXR severity included age (β : 0.140; 95% CI : 0.041 - 0.233; p =0.004), male gender (β : 4.140; 95% CI : 1.452 - 6.481; p =0.003), and disease day (β : 0.622; 95% CI : 0.301 - 0.942; p <0.001). Conclusion: This study decoded the CXR features of Covid-19 pneumonia in a high-risk cohort while describing their associations.


Introduction
Covid-19 disease is caused by the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) and it principally affects the respiratory epithelium.Immunocompetent hosts mostly experience a mild infection confined to the upper respiratory tract (1).However, a minority, with or without other risk factors, can progress into acute respiratory distress syndrome (ARDS) that needs intensive care (ICU) management.Newly mutated viral strains such as the Delta variant are shown to have much higher DOI: https://doi.org/10.4038/gmj.v28i4.8172virulence with frequent lower respiratory tract involvement (2,3).
The diagnosis of severe Covid-19 disease is done with clinical, biochemical & radiological investigations.Clinical features of critical lung involvement include increased work of breath, low peripheral oxygen saturation and increasing oxygen requirements.For clinically suspected severe disease, chest radiographs (CXR) and computed tomographic scans of the chest (CT) are employed to detect the extent of lung involvement (4,5).
Despite the high sensitivity of CT in detecting lung pathology, on-site CXRs are helpful when CT facilities are unavailable or with practical difficulties for critically ill patient mobilization (5,6).Among other features, bilateral-symmetrical lower zone CXR opacities have been described as the most common pattern in Covid-19 pneumonia.The CXR severity score has been recognized as a reliable risk predictor (6)(7)(8)(9)(10)(11)(12).
A strong correlation has been described between the severity of Covid-19 pneumonia and co-existing risk factors (2,9,10).Since most available studies focused on lung involvement in otherwise healthy adults, the lung involvement patterns in highrisk patients are yet to understand.Geographical or ethnic factors have also shown an association with the severity of Covid-19 lung disease.Therefore, more population-based studies are warranted to recognize the true nature of this novel infection (13).Precisely, the knowledge gap on lung involvement patterns of high-risk Covid infected patients has to be bridged.
Since the first report of Covid-19 infection in Sri Lanka, diagnosed patients were managed in health care facilities as in-patients.The study centre functioned as a dedicated national multidisciplinary treatment centre catering for the entire country admitting high-risk, symptomatic Covid patients.Therefore, the study centre provides an excellent opportunity to study a cohort of high risk symptomatic Covid-19 patients.
This study aimed to describe the CXR features and their temporal sequence of high-risk Covid infected patients and to identify the factors associated with CXR severity.The CXRs were been obtained using the same portable X-ray unit as anteroposterior (AP) or supine projection.These CXRs were evaluated by two experienced radiologists (experience more than seven years) under the same image viewing conditions.The CXRs were reported on consensus agreement.Based on the clinical and radiographic features, the CXRs were categorized as normal/ abnormal or changes of Covid-19 lung disease/ non-Covid lung disease.The patients (n = 73) who had indeterminate CXR features, poor image quality or were suggestive of associated comorbidity/ preexisting lung disease were excluded from the study.

Chest radiographs (CXR) of
The CXR opacities were categorized as follows using the Fleischner Society glossary (14): consolidation (patchy/ lobar or segmental), groundglass opacities (GGO), reticulation/ interstitial thickening, atelectasis, pulmonary nodules, pleural effusion, hilar lymphadenopathy, lung cavitation, reverse halo sign, or pneumothorax.The distribution of CXR opacities was categorized: 1. peripheral, peri-hilar predominance or diffuse; 2. right, left, bilateral lung involvement; 3. upper, mid and lower zone involvement.The peripheral and peri-hilar demarcation was made by drawing an imaginary line midway between the lateral lung edge and hilum -the area medial to the line was considered peri-hilar.
The CXR severity was defined using a validated severity score (15).The lung was divided into three zones by drawing two horizontal lines -just below the aortic arch and right pulmonary vein.A score was allocated for each zone considering the area involved (score 0 = no involvement; 1 ≤ 25%; 2 = 25 -50%; 3 = 50 -75%; 4 ≥ 75% area involvement) and the predominant density pattern (score 1 = ground glass/ reticular opacity; score 2 = consolidation).The severity score for each lung zone (upper, middle, and lower zones) was calculated by multiplying the scores obtained for the area involved and density.The total score of each patient is ranged from 0 to 48.A CXR severity grade was assigned using the total severity scores of the patient: mild -severity score 1 to 16; moderate -17 to 32; severe -33 to 48.
The temporal progression of CXR features was assessed by considering the date of diagnosis as day 0. Disease stages were categorized as stage 1 or early stage (0 to 4 days); stage 2 or progressive stage (5 to 8 days); stage 3 or peak stage (9 to 13 days); and stage 4 or resolving stage (≥ 14 days) (16).

Statistical analysis
Once confirming the normalcy of the data set, parametric analysis was done.Continuous variables were expressed as means and standard deviations and categorical variables as percentages.The groups 2 were compared with one-way ANOVA, T-test, χ analysis and linear regression analysis.A p-value less than 0.05 was considered statistically significant.

Patient characteristics
The 228 Covid-19 infected patients (48.2% men) included in the study were in the mean age (SD) of 59 (15) years, and a range of 18 to 87 years.Table 01 describes the baseline characteristics of the study cohort.Of them, 89.5% have discharged uncomplicated, while 17.1% had a fatal outcome.In this high-risk cohort, most patients (85%) have had at least one coexisting disease, which raised to 92.3% among the non-survivors.Figure 1 shows the CXR features of Covid-19 pneumonia identified in the study cohort.Patchy ground-glass opacities (49%; GG) and patchy consolidations (42%; CON) were the most frequent CXR findings of Covid-19 pneumonia (Table 2; Figure 1).Importantly, lobar or segmental consolidations were not present in the study cohort.Lung nodules, masses, hilar lymph nodes, pneumothorax and pneumomediastinum were not present in this study cohort.
Compared to survivors, CON and GG were more 2 frequent among non-survivors (χ = 33.08;p<0.001).Importantly, patchy CON was the predominant opacity pattern among non-survivors, whereas the 2 GG was among the survivors (χ = 4.73; p=0.001).Some features, such as reticular shadows, atelectatic bands and pleural effusions, have not shown an association with disease severity (Table 2).
The progress into the fatal disease was significantly higher in asymmetrical lung involvement: right predominance: ODDs 0.502; p=0.023; left predominance: ODDs 0.268; p=0.002.
There was a supero-inferior gradient in the severity of lung involvement; the lower lung zones were more severely affected (with high severity scores) than the upper zones (right lung T = 10.7;left lung T =11.3; p<0.001).This supero-inferior gradient in lung involvement was more obvious among the non-survivors (T = 6.886; p<0.001).The area of lung involvement has also been associated with disease severity.In 90% of the study cohort and among 90% of survivors, the area involved was less than 75%, it was seen only in 30% of 2 non-survivors (χ = 50.552;p<0.001;Table 03).
Similarly, the proportion of patients who were classified as having a severe radiographic grade was significantly higher in the non-survived group 2 (15%; survived group: 3%; χ = 33.806;p<0.001).3a).With the disease progression, the lungs were infiltrated with consolidation and ground-glass opacities (Figure 3a; consolidations: ** <0.001 ‡ Right lung ** <0.001 ‡ Left lung    The main CXR features of Covid-19 pneumonia detected in this study include consolidations and ground-glass opacities, which showed a superoinferior gradient.Additionally, a temporal sequence of lung involvement has been established with the peak in the second week of illness.Also, the age of the patient and male gender were found as independent predictors of chest X-ray severity. Many previous studies have recognized the factors associated with severe Covid-19 infection using risk un-stratified cohorts; associated comorbidity, male gender and ethnicity were among them (10,12).Considering the possible risk for progression of severe pneumonia into chronic lung injury, predicting and identifying the disease severity is valuable to anticipate subsequent chronic lung injury.Many previous studies have shown an association between risk factors and severe pneumonia; however, the small sample size was their limitation.Associated co-morbidities such as diabetes, hypertension were the recognized risk factors for severe Covid-19 disease in the acute stage (17).During the acute phase of illness, the consolidated lung volume (p=0.031) and proportion of lung involvement (p=0.019)among the Covid-19 infected diabetics (n=15) were significantly higher than that of non-diabetics (n=47) (17).A retrospective study done in Wuhan, China (n=41) reported a complication rate of 32% among the patients with associated comorbidity (18).
Similarly, we found a fatal outcome in 17.1% and radiographically severe disease in 50.9% in a cohort (n=228) with single or multiple risk factors.Thus, by studying a larger high-risk cohort, we confirm that the severity of lung infection is higher in the presence of risk factors.Notably, the limited experiences on post-Covid follow up imaging raises the possibility of chronic lung disease in the high-risk patients and patients with severe disease (19,20).Therefore, particularly for highrisk cohorts, imaging follow-ups may be necessary to detect Covid associated chronic lung disease early.
The Covid-19 pneumonia CXR features among the risk un-stratified cohorts included patchy consolidations, ground glass and reticular opacities (6,8,10,21,22).Patchy consolidations favoured Covid-19 pneumonia, while lobar or segmental consolidations ruled out Covid lung disease (23).Interestingly, the CXR opacities in Covid-19 pneumonia have shown a peripheral and lower lobe predominant distribution.Pleural effusion has been recognized as a rare feature (6,8,10,21,22,24).Cardiomegaly was described among the Covid infected patients without delineating a direct relationship to the coronavirus infection (6).The CXR findings described in our high-risk cohort have also followed a similar pattern.Even though pulmonary nodules have occasionally been reported in Covid patients, none from the current study cohort have had pulmonary nodules (10,21,24).The temporal progression of CXR features detected in this study was in agreement with previous studies done for risk un-stratified cohorts.
The CXR features and severity scores have changed over time and peaked in the second week of infection (9).All in all, the CXR features described in this study for a high-risk group were not different from the radiographic features of risk unstratified populations.
The CXR severity score has been correlated well with the patient outcome; the severity score has been increased with disease severity and fatality (7,10,24).Irrespective of patients' risk status, there was a uniformity in finding lower lobe predominant lung involvement in Covid pneumonia (6,7,21,24) with a higher severity score in the lower zones.A similar pattern has been observed in this high-risk cohort as well.Therefore, severity score appears to be a reliable tool for risk stratification in high-risk patients as well.
In risk unstratified groups, most Covid infected patients had either a normal CXR or a CXR with mild severity (7).Anyhow, as expected, we have not observed a hike in CXR severity grading in this high-risk cohort.Since previous CXR based studies have also included inward patients, though not mentioned clearly, their samples may have included at least a proportion of high-risk patients.These concealed overlaps in the study populations may have created comparable findings in high-risk and low-risk populations.Thus, comparing both high and low-risk patients using an adequate sample would be helpful to identify the actual burden of lung involvement in high-risk populations.
The lung involvement pattern in Covid-19 pneumonia has been described as bilateral symmetrical (6,7,21,24).However, recent studies have described a predominant right lung involvement pattern (7,25).We also noticed a right-predominant, asymmetrical lung involvement pattern more frequently than the bilaterally symmetrical pattern.Though it has been stated that the patients with right lung predominant disease were at a higher risk of hospitalization and a fatal outcome (OR = 2.662; p=0.0252), our findings did not agree with it (25).
The strengths of this study design include evaluating a large sample exclusively from a high-risk cohort and the objective assessment of the CXR features with consensus agreement of experienced observers.However, a case-control study design would have further strengthened the study design.

Conclusions
In conclusion, a typical chest X-ray appearance and a distinct temporal sequence described for high-risk Covid-19 pneumonia patients correspond with reported data from the risk un-stratified population.
Covid-19 infected adult (age >18 years) patients (n = 228) were isolated who were at high risk based on clinical criteria treated at the Base Hospital Homagama st from 1st of December 2019 to 1 of February 2020.The Covid-19 diagnosis was made either by positive RT-PCR or rapid antigen test to detect SARS-CoV-2 antigen.The patient details -sociodemographic data; details of current presentation; previous co-morbidities; risk factors; outcome; the date of diagnosis -were retrospectively obtained from the records.The ethical clearance for the study was granted by the Ethics Review Committee of Sri Lanka Medical Association (Protocol No: ERC/21-001).

Figure 1 :
Figure 1: Radiological features of Covid-19 pneumonia identified in the study cohort

3a:
Spatial distribution of opacities in the lungs 3c: Distribution of opacities in the lung zones 3b: Pattern of distribution of opacities in the lungs 3d: Distribution of severity score in the lungs 3e: Distribution of radiological severity according to the disease stage Original article 125 Galle Medical Journal, Vol.28: No. 4, December 2023

Figure 3 :
Figure 3: The distribution of chest X-ray features at various disease stages of Covid-19 pneumonia a. Spatial distribution of opacities in the lungs b.Pattern of distribution of opacities in the lungs c.Distribution of opacities in the lung zones d.Distribution of severity score in the lungs e. Distribution of radiological severity according to the disease stage

Table 1 :
Baseline characteristics of the study population Galle Medical Journal, Vol 28: No. 4, December 2023 Chest x-ray features Excluding 73 patients in whom the CXR features were either non-conclusive, features of pre-existing lung disease or associated comorbidity, 289 CXRs of 228 patients have evaluated to describe CXR features.Generally the chest imaging was done 7 days after the diagnosis (range of 1 to 21 days).