Predictive Value of Chest CT Score in Assessing Disease Severity and Short-term Mortality in COVID-19 Pneumonia at a Tertiary Care Centre in Northern India: A Prospective Observational Study

1. Junior Resident, Department of Radiodiagnosis and Imaging, All India Institute of Medical Sciences, Rishikesh, Uttarakhand, India. 2. Senior Resident, Department of Radiodiagnosis and Imaging, All India Institute of Medical Sciences, Rishikesh, Uttarakhand, India. 3. Associate Professor, Department of Radiodiagnosis and Imaging, All India Institute of Medical Sciences, Rishikesh, Uttarakhand, India. 4. Senior Resident, Department of Community and Family Medicine, All India Institute of Medical Sciences, Bhopal, Madhya Pradesh, India. 5. Junior Resident, Department of Radiodiagnosis and Imaging, All India Institute of Medical Sciences, Rishikesh, Uttarakhand, India. 6. Junior Resident, Department of General Medicine, All India Institute of Medical Sciences, Rishikesh, Uttarakhand, India. 7. Additional Professor, Department of General Medicine, All India Institute of Medical Sciences, Rishikesh, Uttarakhand, India. 8. Assistant Professor, Department of General Medicine, All Ind

Abstract

Introduction: Severe Acute Respiratory Syndrome-Coronavirus-2 (SARS-CoV-2) infection, also known as Coronavirus Disease-2019 (COVID-19) is the global pandemic, first described in Wuhan city of China in December 2019. Its diagnosis depends upon real time Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR). On chest Computerised Tomography (CT), it is almost similar to other viral pneumonia with extensive parenchymal involvement. Semiquantitative scores depicting this extensiveness of involvement could correlate with disease severity, laboratory parameters, mortality like Intensive Care Unit (ICU) admission, requirements of ventilatory support and longer hospital stay.

Aim: To define the role of chest CT score in determining disease severity, predicting poor prognosis and mortality of COVID-19 pneumonia in short-term follow-up.

Materials and Methods: This prospective study enrolled 547 admitted real time RT-PCR positive patients for COVID-19 at All India Institute of Medical Sciences, Rishikesh, India from 15^th April 2021 to 31^st May 2021. All patients were assigned semi-quantitative CT scores based on the extent of lung parenchymal involvement of 20 lung regions in chest CT. Finally, 205 patients were enrolled for the final analysis. Clinical severity was matched with chest CT scoring and laboratory findings. Survival curves along with univariate and multivariate analysis was applied to define the role of CT scoring in predicting short-term prognosis.

Results: Total 205 subjects were included in the study, of which the chest CT score showed a significant association with clinical severities (p-value <0.001). CT score was correlating significantly with increased serum C-Reactive Protein (CRP) (p-value=0.001) and D-dimer (p-value=0.01), and decreased lymphocyte count (p-value <0.001). A CT score ≥31 was found to be associated with an increased risk of mortality in both univariate and multivariate analysis {Odd Ratio (OR)=276.8; 95% Confidence Interval (CI)=45.21-1695.43; p-value <0.001}.

Conclusion: Chest CT score can be imaging measure of disease severity and predict a higher probability of mortality in score ≥31. It can also predict other defined variables of short-term prognosis. So, it has an advantage in speedy diagnostic workflow of symptomatic cases, timely referral of patients to higher centre, and better management of critical care resources.

Coronavirus disease-19, Computed tomography severity score, Ground glass opacity

The COVID-19 infection has resulted in 190,860,860 confirmed cases and 4,101,414 deaths as of 21^st July 2020 (1). The concern regarding high infectivity, morbidity and mortality with COVID-19 infection has resulted in worldwide lockdown to contain the spread of disease (2). Currently, different parts of the world are preparing for the third and subsequent wave of infection including community transmission (1). Fever, fatigue, cough and dyspnoea are the common clinically presenting complaints and quite similar to other respiratory virus infections, more specifically other coronavirus infections like Middle East Respiratory Syndrome (MERS) Coronavirus and SARS-CoV-2 (3),(4).

On non contrast chest CT imaging, it resembles viral pneumonia as symmetrical involvement of lung parenchyma with Ground Glass Opacity (GGO), with or without associated consolidation, predominately in peripheral and posterior distribution (5),(6). Authors hypothesise that extent of lung parenchymal involvement, depicting as chest CT score could correlate with the clinical severity of COVID-19 infection and predicts disease outcome.

There is growing evidence of insensitivity of single RT-PCR testing for diagnosis of COVID-19 infection (7),(8). The sensitivity of RT-PCR testing depends upon sample collection technique and test’s technical characteristics. Inadequate sampling is also an add-on to this cumbersome test when a quick diagnosis is sought. In symptomatic patients with the first RT-PCR negative test, chest CT imaging can be a supplement (9). The relatively lower sensitivity of a single RT-PCR testing with a long turnaround time insinuates that a large bulk of COVID-19 patients could not be isolated quickly to contain the disease. Management of these patients based on the clinical severity.

The primary objective of the present research was to examine the role of CT score in assessing the disease severity and predicting short-term mortality. Secondary objective was to bring out a cut-off of CT score beyond which they had experienced a higher mortality in the present study population. Thirdly, its role in predicting other variables of prognosis like ICU admission, ventilatory support, and long hospital stay was also investigated.

Material and Methods

This prospective observational study was conducted in the Department of Radiodiagnosis and Imaging and Department of General Medicine at All India Institute of Medical Sciences, Rishikesh, India, from 15^th April 2021 to 31^st May 2021 after getting ethical clearance from Institutional Ethics Committee (letter No- AIIMS/IEC/20/441, Reg No: ECR/736/Inst/UK/2015/RR-18), following the principles of the Declaration of Helsinki.

Authors initially enrolled all patients who were admitted to the Emergency Department of the institute with clinical suspicion of COVID-19 infection during the study period (n=547). The criteria for clinical suspicion of COVID-19 infections were based upon guidelines laid by World Health Organisation (WHO) (10). Informed consent was obtained from all patients before enrolment into the study.

Inclusion criteria: Patient admitted in the institute with clinical suspicion of COVID-19 infection.

Exclusion criteria: Two serial RT-PCR reports coming out to be negative with a gap of one day within them (As per our institutional protocol to say someone negative for COVID-19 infection), patients with primary and metastatic lung neoplasms, active pulmonary tuberculosis, any acute medical and surgical conditions that had independent risk of mortality were excluded from the study. Previously diagnosed with Interstitial Lung Disease (ILD) cases or patients with contraindication of CT scan, patients who refused for consent or had poor chest CT imaging due to excessive motion artifacts etc., were also excluded from the study.

Finally, 205 patients were enrolled for the study. During this study, all patients were managed with standard guidelines for clinical management and they were not enrolled in any other studies (Table/Fig 1).

Clinical Workflow and Disease Staging

Detailed present and past clinical history, and vital parameters such as respiratory rate, pulse rate, oxygen saturation was maintained in predefined clinical sheets. Chest CT and laboratory investigation like complete haemogram, Arterial Blood Gas (ABG), serum level of acute phase reactants like C-Reactive Protein (CRP), D-dimer, Procalcitonin (PCT) and Lactate Dehydrogenase (LDH) was carried out routinely within one day of hospital admission. Whenever a first RT-PCR came negative, a repeat RT-PCR was done after a gap of one day.

Disease severity classification was done using the Chinese Centers for Disease Control and Prevention (CDC) guidelines as mild, severe and critical (11). To know the evolution of disease in chest CT, the course of the disease was divided into early (<7 days) and late (≥7 days) phase based on the days of symptoms (12). All patients were followed for clinical progression throughout their hospital stay. Hospital stay was divided as short (<20 days) and long (≥20 days) (13).

In all stages blinding was maintained, neither the clinician nor the radiologist knew about each other’s findings to prevent selection bias, only conventional CT reports were provided.

CT Protocol

Maintaining appropriate infection prevention and control measures, image acquisition was done using a single source Multidetector Computed Tomography (MDCT) scanner Ingenuity core 64 slice (Philips, Netherlands), in a supine position during a single inspiratory breath-hold, from the apex of the lung to the costophrenic angle. The scanning parameters were KVp=120; mAs=40; rotation time-0.5 second; pitch-1.0; section thickness-5 mm; intersection space-5 mm. Images were reconstructed at 1 mm slice thickness in all three planes and viewed in the mediastinal (C=60, W=400 and Matrix=512) and lung (C=-600, W=1600 and Matrix=768) windows.

Image analysis: Three radiologists with 5-10 years of experience in chest radiological reporting reviewed all the provided reconstructed images independently and had been completely blinded to the clinical and laboratory findings. In case of any discrepancies in the interpretation, the final result was reached by blinded voting among them. The standard radiological terms were used as described in the standard glossary for thoracic imaging reported by the Fleischner Society (14). CT scoring was done as proposed by Yang R et al., which was an adaptation of previously clinically and laboratory parameters correlating scoring technique to describe lung involvement in patients of SARS (15),(16). The 18 anatomical segments of both lobes of the lung were divided into twenty lung regions, each lung region was scored as 0 (no involvement) 1 (<50% lung involvement), and 2 (>50% lung involvement) (Table/Fig 2).

Statistical Analysis

Statistical analysis was applied using Statistical Package for the Social Sciences (SPSS) software version 23.0. For a single and multiple comparisons, Mann-Whitney and Kruskal-Wallis tests were performed respectively. The association with CT Severity Score (CT SS) was done using the 2-tailed Chi-square test or Fisher’s-exact test. The Receiver Operating Characteristic (ROC) curve was drawn to determine the optimal cut-off point for CT score as an all-cause mortality. Pearson Chi-square, continuity correction, likelihood ratio, Fisher’s-exact test and linear by linear association tests were applied to define association of CT score with variables. Kaplan-Meier test which was used to evaluate the relationship between CT score and all-cause mortality, which was compared with the logrank test. Cox proportional hazards regression modelling was performed to determine the Hazard Ratio (HR) for CT score as an all-cause mortality predictor.

Results

The mean turn around time was 14.3±2 hours for RT-PCR and 22±10 minutes for chest CT. Another 30 minutes were required for sanitation of the CT machine before it was ready for the next patient. Out of 205 cases, 71.7% were males and 28.3% were females. Maximum cases were seen in the age range of >45 to 65 years. Diabetes mellitus was the most common co-morbidity, seen in 40.5% of patients; followed by hypertension (37.1%). A 20% of patients had both diabetes mellitus and hypertension. In clinical presentation; fever was seen in 77.1% of patients; followed by cough (63.4%) and shortness of breath (53.6%) (Table/Fig 3).

Chest CT finding regarding imaging features, complications, lobar involvement, disease localisation are presented in (Table/Fig 4). The most common chest CT finding (Table/Fig 5) was Ground Glass Opacities (GGO) seen in 156 patients (76.1%), followed by parenchymal consolidation (n=143; 69.7%) and crazy paving pattern (n=101; 49.3%).

Auxillary findings (Table/Fig 6) like fibrosis (n=98; 47.8%), subpleural lines (n=82; 40%), mediastinal lymphadenopathy (n=46; 22.4%), and reversed halo sign (n=3; 1.5%) was also seen. Significant lower lobe involvement was seen with right sided preference.

On the anterio-posterior dimension, there is more involvement of the posterior location with peripheral predominance (Table/Fig 7). Nine patients showed normal chest CT.

Authors also noticed various complications like pleural effusion, systemic venous thrombosis and spontaneous pneumothorax (Table/Fig 8). CT score in early versus late phase disease (Table/Fig 9) (12). GGO was more seen in the early phase whereas consolidation, crazy paving, fibrosis, subpleural lines and mediastinal lymphadenopathy were more seen in the late phase with a significant p-value (p-value <0.05).

CT score versus clinical severity: The chest CT score showed significant association with clinical severity; higher the score, more the severity clinically. The mean CT scores in mild, severe, and critical groups of patients were 13.73±8.51, 23.49±3.75 and 32.42±4.01, respectively (p-value <0.0001). It is described in (Table/Fig 10), (Table/Fig 11).

Clinical severity in age, sex and co-morbidities adjusted cohort: The milder form of the disease was more common in the younger population; however, there were no statistically significant patterns in the cases of severe and critical patients. There was no sex predilection while comparing different clinical severity groups. More severe forms of disease, predominantly critical forms were seen in patients having any known co-morbidities. CT score has no significant association with age, sex and co-morbidities (Table/Fig 12), (Table/Fig 13), (Table/Fig 14).

Short term prognosis in age, sex and co-morbidities adjusted cohort: Although parameters of short-term prognosis were more prevalent in the older age group; however, statistical significance was seen in ICU admission, ventilator support and death. On crosstab statistics, a significant association of CT score was seen with ICU admission, ventilator support and death. However, no significant association with long hospital stays was seen (Table/Fig 12), (Table/Fig 13), (Table/Fig 14).

A statistically significant higher number of ICU admission and deaths were noticed in patients with any known co-morbidities in comparison to previously healthy individuals. In case of diabetes and hypertension, the same prognostic parameters were seen at lower CT score than previously healthy individuals. On crosstab statistics, CT score had no significant association with co-morbidities.

Kaplan-Meier survival curves and univariate and multivariate analyses: Out of the 205 cases in the present study cohort, 46 patients (22.4%) died during hospital stay out of which 42 had known co-morbidities. Diabetes mellitus was reported in 27 (58.7%) of 46 deaths, followed by hypertension in 23 (50%) patients and other co-morbidities in 7/46 (15.2%). Four (6.1%) patients who had no known co-morbidities also died during hospital stay. In the present study death was only seen in critical group patients (Table/Fig 14).

As per Kaplan-Meier analysis and ROC curve, the mortality risk was significantly higher with the increase in CT Score, using an estimated cut-off of ≥30.5 {logrank p-value <0.0001; HR-46.30 (CI:14.35-149.34); p-value <0.001} on a follow-up period of 30 days. Area under ROC Curve (Table/Fig 15) was reported to be 0.97 (p-value <0.0001). Bivariate analysis showed a significant association of CT SS with phase of disease (p-value <0.001), clinical severity (p-value <0.001), lymphocytopenia (p-value <0.001), raised CRP level (p-value <0.001), raised LDH level (p-value <0.001), raised D-dimer level (p-value <0.001), raised PCT level (p-value <0.001) and mortality (p-value <0.001).

Univariate analysis demonstrated a higher risk of mortality with an increase in age, associated co-morbidities, higher CT SS, and raised CRP and D-dimer levels. Multivariate analysis applied on significant statistical variables proven by univariate analysis confirmed the role of CT score as an independent predictor of death (OR=276.8; 95% CI=45.21-1695.43; p-value <0.001) together with co-morbidities (OR=19.36; 95% CI=3.50-107.09; p-value=0.001) and raised D-dimer (OR=25.02; 95% CI=1.90–328.43; p-value=0.014). The Nagelkerke R Square is estimated at 0.820 indicating that 82.0% of the variance in mortality can be predicted from the linear combination of high CT severity score, presence of co-morbidity and high D-dimer levels supposed to be the predictors of mortality.

Discussion

The most common imaging findings of COVID-19, in the present study were bilateral GGOs with or without consolidation, with a predominant peripheral, lower lobe and posterior anatomic distribution. It is quite consistent with previous studies (12),(15),(17). The reason for the more common imaging occurrence of GGO in the early phase of disease can be attributed to the acute phase alveolar injury leading to air space oedema, bronchiolar fibrin depositions and interstitial thickening (18). As the disease progress, there is activation of humoral as well as cell-mediated immune system by virus specific B and T-cells; causing intense production of proinflammatory markers leading to uncontrolled autoimmune reaction. A combination of alveolar oedema, bacterial superinfection, and interstitial inflammatory changes are seen in the late phase, which may explain the higher prevalence of consolidations and crazy-paving pattern in the late phase (19).

Raised leukocyte count, decreased lymphocyte count, raised serum CRP, LDH, PCT and D-dimer levels were commonly observed in COVID-19 patients. These correlated strongly with higher CT scores. Raised serum CRP and D-dimer levels may be explained as a result of the pronounced inflammatory activation and disseminated coagulopathy, characteristics of severe disease (20),(21). Raised serum PCT can be seen in the setting of secondary bacterial infection, suggesting a bad prognosis (22). CT score, co-morbidities and serum level of D-dimer has a significant predictor of mortality on multivariate analysis. Age and serum level of CRP shows significant risk of mortality on univariate analysis; however, no significant association was defined on multivariate analysis.

A Kaplan-Meier survival analysis was performed between CT score and days of hospital stay, to confirm chest CT findings’ prognostic significance for an observational period of 30 days. By using this method, the present study was able to demonstrate that a cut-off CT score of 30.5 is predictive of mortality. But a score of 30.5 is not possible in this study scoring system, so authors took it as ≥31. Similar observations were previously published by Colombi D et al., and Francone M et al., (23),(24). CT severity score more than 31 out of 40 was associated with poor prognosis in the present study population which is comparable to observation made by Francone M et al., (24). They have found score equal or more than 18 out of 25 had poor outcome.

There was a significantly higher mortality in a population of more than 60 years of age in comparison to younger ones. The univariate analysis also proved increased risk of mortality with an increase in age. No significant gender preponderance was seen with the severe form of disease and bad prognosis.

A chest CT score, the objective value of the radiologist’s observation, depicting the extent of lung involvement can correlate well with disease severity and active phase inflammatory marker findings. It can also predict the outcome or clinical course of the disease, as it represents the disease burden. This prospective study explored all the clinical utility of chest CT score in predicting disease severity and short-term prognosis of the disease. Apart from this, a CT score can come in handy in the management of patients in a few more ways.

Due to the short turnaround time of chest CT in comparison to RT-PCR, it can be very useful in the early isolation of patients to contain the disease in the hospital. There is a statistically significant difference between the turnaround time of chest CT and RT-PCR. In 29 patients with first RT-PCR negative, authors had to repeat RT-PCR for diagnosis. This further increases the time for diagnosis. Sometimes sampling is also poor; causing more delay in diagnosis. Although, the diagnostic role of CT remains controversial and a hot debate topic in the current pandemic situation. A group of researchers believe chest CT has higher sensitivity (25),(26) in comparison to RT-PCR while others believe vice-versa (27). Several authors and radiological fraternities do not believe in the use of CT as a first-line investigation due to radiation exposures (28). In the present study, CT also did not appear much sensitive; nine patients were RT-PCR positive despite their chest CT imaging were normal. Due to less turnaround time and significant association with laboratory parameters, CT score can be helpful speedy diagnostic workflow of symptomatic patients. In COVID-19 infection, there is more extensive involvement of lung parenchyma in comparison to other viral pneumonia (5),(6),(29). A higher CT score has a more probability of the COVID-19 infection. So, authors recommend the use of chest CT routinely in all symptomatic severe patients.

Limitation(s)

The present study had some limitations as recall bias regarding the previous diagnosis of ILD or any co-morbidities and authors experienced a higher number of critical patients and mortality in this study group in comparison to our national data as we were a referral center. The survival analysis study lacks longer follow-up data. Due to the smaller sample size, authors could not define the role of individual co-morbidity in disease mortality. Authors suggest a future prospective study with a larger sample size, co-morbidity adjusted individual cohort and a study to look predictive value of CT score for delayed complications.

Conclusion

Chest CT findings are quite evident and correlating well with the novel acute phase reactants. So, CT score can be used as an imaging tool to predict the future course of disease and plan management accordingly in areas lacking with the modern laboratories. In a developing country, this can guide timely referral of patients to higher centers with better intensive care facilities. Whereas in the developed country, it can help hospital administration for better preparedness for critical events and management of hospital resources. Due to its objectiveness, it can make communication easier between the caregivers and the caregiver, and the patient’s caretaker. The higher chest CT score with higher probability of COVID-19 infection can be helpful in containing the disease by early isolation.

DOI and Others

DOI: 10.7860/JCDR/2022/51808.16168

Date of Submission: Aug 08, 2021
Date of Peer Review: Nov 22, 2021
Date of Acceptance: Jan 05, 2022
Date of Publishing: Mar 01, 2022

AUTHOR DECLARATION:
• Financial or Other Competing Interests: None
• Was Ethics Committee Approval obtained for this study? Yes
• Was informed consent obtained from the subjects involved in the study? Yes
• For any images presented appropriate consent has been obtained from the subjects. Yes

PLAGIARISM CHECKING METHODS:
• Plagiarism X-checker: Aug 12, 2021
• Manual Googling: Jan 04, 2022
• iThenticate Software: Jan 11, 2022 (9%)

ETYMOLOGY: Author Origin