key: cord-0884451-au3pupnw
authors: Goyal, A.; saigal, s.; Joshi, A.; Niwariya, Y.; Brahmam, D.; Khurana, A.; Singh, P.; Sharma, J.; Karna, S. T.; Khadanga, S.; Mitra, A.
title: Whether Early Pulse Steroid dose is associated with lower mortality in COVID-19 critically ill Patients- A retrospective Chart Review
date: 2020-08-22
journal: nan
DOI: 10.1101/2020.08.19.20171868
sha: 5f04955a323648d7df51fe86a221d74ed85d95cb
doc_id: 884451
cord_uid: au3pupnw

Abstract : Introduction: Steroids have theoretically seems useful in critically ill patients of COVID-19 . However the time of starting steroid and dose remains a matter of concern due to still emerging evidences and wide-ranging concerns of benefits and harms. We did a retrospective record analysis in an apex teaching hospital ICU setting to explore this concern. Methodology: 45 adults age more than 18 years with nasopharyngeal swab PCR-confirmed SARS-CoV-2 infection with ARDS admitted to ICU in between 20th March,2020 to 15th July 2020 were included in chart review. We did a bivariate analysis of age ,comorbidity, infections, severity of disease , timing /dose(appropriate) steroid and presence of infection on survival. In the next step we performed a Bayesian Exact regression to understand the adjusted effect of early appropriate steroid on survival in the presence of age and infections as probable confounder. Results: Bivariate analysis showed the statistically significant effect of age <60 years and steroid dose (early and classified by disease severity) had a favourable effect on outcome. Further Early Pulse Steroid (EPS ) amongst the more severe subgroup was found to be significantly associated with better survival. Conclusion: High dose steroids (more than 500 mg MPS) if given early in the course of disease to COVID19 critically ill patients (P/F<150) can significantly reduce mortality and are not associated with increased infections.

Since the start of COVID19 pandemic, multiple repurposed drugs like antivirals [Hydroxycholoroquine, Remdesivir] , anti-inflammatory (steroids, Tocilizumab or anakinra), anticoagulants and fibrinolytics have been used with variable success.1,2 As the pathophysiological pathways around COVID-19 are still advancing in the light of new acquired evidences, one may witness the differential standardized protocols adopted in different settings and gradually evolving protocol in the same setting longitudinally.

The hallmark of COVID-19 associated severe disease is increasing oxygen requirement and increasing C-reactive protein (CRP) levels, which is hypothesized to result from the cytokine storm syndrome (CSS).3 This CSS probably acts by two ways: First, it leads to pneumonia and ultimately adult respiratory disease syndrome (ARDS) and second it leads to overreactive coagulation pathway, leading to microvascular thrombosis in the pulmonary vasculature. This dual brunt is the main reason for elevated morbidity and mortality.

At the start of the pandemic, Tocilizumab, an IL6 receptor antagonists had shown benefit in early case series.4 But it was not available for most of our patients at our centre because of prohibitive cost ($600-1200 per patient). Pulse dose steroids were used to treat Haemophagocytic lymphohistocytosis (HLH) long before other IL6 antagonists were invented.5 COVID19 associated CSS is strikingly similar to HLH in some aspects. 6 Keeping this scientific pathophysiological plausibility in mind, we decided to give high doses of steroids to patients of COVID19 ARDS. With the context of strong pathophysiological linking of potential steroid therapy amongst COVID19 patients and absence of any standardised treatment protocol at that time, motivated us to initiate steroid therapy amongst COVID19 ARDS patients. As this was a relatively naïve paradigm, standardisation of the dose could not be achieved despite of a qualitative consensus on steroid therapy.

Early administration of glucocorticoid in course of ARDS was associated with better survival in subgroup analysis of patients with SARS/MERS.7 Recently published RECOVERY trial has shown positive effect of low dose glucocorticoids (6 mg dexamethasone OD for 10 days) on survival in treating COVID-19 ARDS patients on respiratory support. 8 But there has been no consensus of usefulness of high dose steroids in literature in COVID19, since pulse steroids is theoretically associated with secondary infections.

We did retrospective chart analysis for patients admitted with COVID19 related ARDS in our ICU, to see whether steroids are associated with reduced mortality and whether it was associated with increased risk of infections.

Consecutive adults ≥18 years with nasopharyngeal swab PCR-confirmed SARS-CoV-2 infection with ARDS admitted ( in ICU) in between 20th March,2020 to 15th July 2020 were included in chart review. This review was done in a single tertiary care teaching hospital in central India. The records were reviewed longitudinally from admission to outcome ( either till death or discharge from hospital.

We included only those patients who were given steroids in our ICU during the study period. All our patients in this analysis were being given varying doses of steroids. Time when patient had mMRC grade IV (breathlessness at rest) was noted according to patient history. The time difference between onset of severe breathlessness (mMRC grade IV) and first dose of steroid administration (as confirmed by the nursing chart) was counted. Dose and timing of steroid administration was decided by the consultant in-charge posted in ICU.

As we gained experience in management, we felt that high dose steroids (>500 Methylprednisolone) for 2-3 days followed by low dose (30-60 mg MPS) was better in management of patients with high oxygen requirement on admission (>10LPM). For patients with lower oxygen requirement than 10 LPM, low dose steroids (30-60 mg MPS) were sufficient to control cytokine storm. For comparative analysis, we grouped patients into two categories: Appropriate and Inappropriate. We defined appropriate steroids as either: a) If FiO2 requirement to maintain spO2 >92% was more than 0.5 and he/she was given 500 mg of Methylprednisolone (MPS) or more as first dose of steroid within 24 hours of severe breathlessness. This group is henceforth . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 22, 2020. . called as Early Pulse Steroids (EPS). b) If FiO2 requirement to maintain spO2 >92% was less than 0.5 and he/she was given 30-60 mg of MPS or more as first dose of steroid within 24 hours of severe breathlessness. This group is henceforth called as Early Low Steroids (ELS).

Patients were defined in the inappropriate steroids group (or other group) if:

I. If first dose of MPS (in any dose) was started after 24 hours of severe breathlessness (late steroids).

II. If patient's oxygen requirement (FiO2) was more than 0.5 to maintain spO2 >92% was and he/she was given less than 500 mg of MPS as first dose of steroids.

Disease severity was decided on the basis of P/F ratio on the day of admission. All patients admitted in our ICU had P/F < 300. For analysis purpose, we categorize these patients into two groups: P/F <150 (critically ill) and P/F in the range of 150-300 (mild to moderate ARDS). Requirement of NIV or invasive mechanical ventilation (IMV) or High Flow Nasal Canula (HFNC) was noted.

Process of data extraction-Data was extracted from the case records and treatment files of patients admitted in ICU. Data was retrieved by resident and consultant posted in ICU and was cross checked by another consultant. Data collection included timepoints like symptom onset, onset of shortness of breath, presumptive onset of severe breathlessness (mMRC grade IV), timing of steroid administration with respect to onset of severe breathlessness, date of hospitalization, date of outcome (discharge or death). Median value of SpO2 was taken for analysis for a 24-hour period. CRP, P/F ratio, FiO2 requirement and absolute lymphocyte counts were recorded and trend was analysed. Death was assessed during hospitalization and patients were followed until discharge from hospital. In our hospital, COVID19 ARDS patients are discharged only after negative RTPCR report. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 22, 2020. .

All patients in our ICU had hypoxemia and were started on steroids along with empiric antibiotics (for gram negative coverage to prevent infections. Blood and endotracheal aspirate Culture Sensitivity were sent whenever patient had fever or TLC increased more than 14000/l after steroid administration. Bacterial infection was diagnosed if patient had any of the culture reports positive. The retrospective chart analysis was cross checked by two senior consultants. History of known comorbidities like Hypertension, Diabetes Mellitus (DM), Coronary Artery Disease (CAD), Chronic Kidney Disease(CKD), hypothyroidism was noted from history sheet.

Statistical Analysis All statistical analysis was performed using statistical programming language R.9

The data was prepared for analysis using standard data cleaning and data wrangling procedures using base R and the 'tidyverse' universe of packages. Exploratory analysis was done by cross-tabulations and bi-variate analysis. Data visualisation was done using grouped bar charts, box-plots and density ridgeline plots using functions from the 'ggplot2' package. A multivariate exact logistic regression was performed to estimate the conditional distribution of statistic for interest parameters namely steroid category and infection status conditional to the survival status of the patient. This Bayesian method uses the Markov Chain Monte Carlo(MCMC) method to derive the exact conditional inference for binomial data modelled by logistic regression. We used this method due to smaller sample size and presence of zero observations in a category formed by the outcome and categorical predictors. We used 'elrm'v1.2 package in R software which is available in open domain. All code and data necessary for reproducing the analyses and additional results are provided in the supplementary appendix.(Annexure-1) Data was first transformed in a collapsed data set which consisted of the respected frequencies (trials; for 4 combinations of steroid and infection categories) and frequency of survival (success) in that combination. The Markov chain was created with 22000 iterations and 2000 burn in samples with final chain size as 20000. Three separate models were created with interest parameters as steroid category, infection category separately and in combination of both. In order to obtain the full conditional inference in the combination model we updated the length of the chain to 75000 (total iteration 80000 with 5000 burn-in samples). The model diagnostic was estimated by trace plot and histogram to check convergence of the Markov chain. The presence of serial correlation which may confound the results was further investigated by Autocorrelation Function (ACF) plot.

Ethics issues-All admitted patients and their relatives are explained about disease condition , therapeutic options and prognosis routinely at our institute. This retrospective chart review study is part of study titled "Predictors of severity in COVID 19 infected patients: Observational Chart review". Institutional Human Ethics Committee (IHEC) approved the protocol vide LOP Number IHEC-LOP/2020/IM0281 , dated 15/07/2020.

A total of 45 study participants were included in the study. The mean age of the population was 56.8 ± 15.9 years. The most baseline characteristics of the study population are provided in Table 1 . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 22, 2020. The results of the cross tabulation of outcome with the other variables are given in Table 2 . It was found that age over 60 years and steroid therapy were found to be significant in preliminary bivariate analysis while gender, presence of comorbidity, category of ARDS and state of infection were found to be statistically non-significant. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The relationship between severity (PFR status), infection status, age and diabetes status with assumed appropriate steroid dose and outcome is presented in a composite plot in Figure 1 . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 22, 2020. . Figure-2 Composite plot of timing and dose of steroid pulse therapy faceted by disease severity as measured in PFR .The x-axis represents the category of steroid therapy (which is subdivided into 3 subcategories) while the y-axis represents the number of study participants in each category. The colour of the bars represent the outcome of the study participants. Figure 2 is the extension of figure 1a where the survival status is further sub grouped by the three dose categories and stratified by the disease severity as measured by P/F ratio. The plot clearly indicate the benefit of early high dose (defined as >500/1000 mg early pulse steroid ) in more severe disease(PFR<150).

In the next step the effect of early steroid on survival was quantified with choosing age >60 years and presence of infection as confounder. The Bayesian exact regression was performed to estimate the conditional distribution of the sufficient statistics of the parameter of interest (steroid category) on survival in the presence of nuisance parameter( adjusted by the effect of age and infection) . The equation the result of which is shown in table no.3. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 22, 2020. . This shows that regression parameter value is away from 0 (Ho: γ=0 and H1: : γ≠0)thus there seems to be an independent effect of appropriate dose of steroid after adjusting with age and infections. The trace plot was used to diagnosed whether the Markov chain was converged ( Figure 3 ) which is not in this case, thus estimates seems robust methodologically. 

In this retrospective chart analysis study, we found that administration of early pulse dose steroids (>500 mg MPS) within 24 hours of start of severe breathlessness was associated with lower mortality in COVID19 ARDS (P/F <150). Also, we didn't find higher incidence of infections among those who . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 22, 2020. . https://doi.org/10.1101/2020.08.19.20171868 doi: medRxiv preprint received high dose steroid and it was not associated with increased mortality due to secondary infections.

According to current understanding, COVID-19 infection timeline has probably following phases:

1. Initial phase of viral replication: Often there are only mild symptoms.

Adaptive immunity is stimulated to respond to this replication. Low dose steroid (Dexamethasone 6 mg OD or methylprednisolone 30 mg OD equivalent) administration during the early inflammatory phase (i.e. Phase 2) has been shown to be beneficial by blunting inflammatory cascade.8 In RECOVERY trial benefit was shown in patients on respiratory support (either NIV/HFNC/Oxygen supplementation/ IMV). But in the preliminary report of RECOVERY trial, subgroup analysis on the basis of P/F ratio or oxygen requirement was not done. We found low dose steroids (40-120 mg MPS) to be useful in patients with P/F >150, but patients with marked hyperinflammation reaction (characterized by P/F <150 on ABG), early low-dose steroids if given were less effective to suppress Cytokine storm, as shown in Figure 2 . High doses of steroids given early in the phase of hyperinflammation (within 24 hours) can potentially blunt the inflammation cascade if oxygen requirement is high. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 22, 2020. .

In our analysis, steroids were found to be less successful in following conditions:

1. If first high dose of steroids was started after 24 hours of severe breathlessness (late steroids).

2. If low dose steroids were given in patients with initial P/F <150 even if given early (early low dose steroids)

Glucocorticoids IMV was used as last resort in our ICU; patients were intubated only when patient failed NIV or HFNC. Only one of our patients put on IMV could be extubated till the data collection.

Timing of steroids is of utmost importance in patients of COVID19 ARDS. Similar findings were seen in a study from New York ,where usage of steroids within 48 hours was associated with better survival.12 But in this series, only 7.1% patients who were given early steroids had sPO2<90% on presentation. Also the steroids dose and P/F values were not mentioned in the brief report.

Currently, we use CRP and oxygen requirement guided steroid administration in our COVID ICU practice. When patients are in early inflammatory phase (characterised by elevated CRP) and oxygen requirement upto FiO2 0.5, these patients can be successfully managed with low dose steroids (60 -120 mg MPS OD for three days then 30 mg OD for 5-7 days), provided the first dose was given early. But in patients with FiO2 > 0.5, we believe steroids should be given as early as possible in high doses (500 mg) for at least 2-3 days and then titrated according to CRP levels.

Often patients presents late in critical condition and reporting of RTPCR for COVID19 usually takes 24-36 hours in our hospital. So to avoid loss of precious time, we administer first dose of 500 mg MPS to the patient who has high pretest probability of COVID19 and oxygen requirement is high (>10 LPM). We administer second dose of MPS after confirming RTPCR report. We consider high pretest probability of patient being COVID19, if he/she fulfils all following criteria:

i We have started giving first dose of MPS to our patients, if patient fulfills all three criteria and we have found to be highly useful in our ICU setting.

Our assessment of oxygenation shows that improvements in oxygenation starts after 24 hours of steroid administration. This was similar to trend seen after tocilizumab infusion seen in a case series, where P/F ratio remains stable for 1-2 days before increasing.4 Methylprednisolone is a time tested, readily available, cost effective immunosuppressor unlike new immunosuppressor like tocilizumab, which is cost prohibitive and is available only in selected areas around world. Few RCT comparing steroids and Tocilizumab are underway and results are expected by end of 2020. 10 We used tocilizumab in only three patients due to prohibitive cost and lack of availability at our centre at the start of pandemic. It was used in two patients among appropriate steroid group (both survived) and one in late steroid group (she expired after a prolonged stay in hospital). All three patients were given high dose steroids plus tocilizumab and two of them developed superimposed fungal infections, gram positive and gram . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 22, 2020. . negative bacteraemia. So both high dose steroids and tocilizumab should not be given in a single patient as it can lead to profound immunosuppression and worse outcome.

TLC rises in a significant proportion of patients, but there was no significant difference in rates of infections among EPS and OSG. Importantly, presence of infection had no significant effect on mortality. (Figure 1B) Although, there was no difference in mortality in these two groups, we had 9 patients had gram negative, 2 had gram positive bacteraemia and two had candidemia. One HIV positive patient also developed candidemia, but he was successfully weaned off ventilator. We strongly recommend performing standard of care hygiene precautions in COVID ICU, as these patients are already immunocompromised and superimposed infections can lead to poor outcome.

Another common side effect of high dose steroids is hyperglycaemia, Most of critically ill COVID19 patients are known diabetic, and these patient present with exceptionally raised sugar (sometimes in the range of 400-600 mg/dl) at presentation. Binding of coronavirus to ACE2 receptors found on pancreatic beta cells has been postulated as cause of significant hyperglycaemia.13 Markedly raised sugar was present in significant proportion of our patients, but it was easily managed with insulin infusion in all the patients. And survival benefit of early high dose steroids was present even in diabetics.( Figure 1D ) So we believe being diabetic should not be impediment for using steroids.

Our study was a retrospective chart analysis from a single centre study with limited sample size. But we have taken the most critically ill patients and have used robust Bayesian Hypothesis. Secondly, some patients with known significantly high counts (TLC> 15000) at presentation were excluded from analysis, as these were not given steroids, so the findings should not be extrapolated to this subset of patients. Third, we did not analyse the time taken for RTPCR to become negative. Slower clearance of viral RNA was seen in patients with influenza, SARS and MERS who were treated with glucocorticoids. But significance of persistent positive RTPCR in terms of clinical outcome is unclear in patients who have improved clinically. 14 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 22, 2020. .

In conclusion, high dose steroids if given early in the course of disease to COVID19 critically ill patients (P/F<150) can significantly reduce mortality. A well designed RCT to compare efficacy of low v/s high dose steroids in patients of moderate to severe COVID19 ARDS needs to be conducted to clarify the dose requirement.

Pharmacologic Treatments for Coronavirus Disease 2019 (COVID-19): A Review

Successful use of tPA for thrombolysis in COVID related ARDS: a case series

The cytokine storm in COVID-19: An overview of the involvement of the chemokine/chemokine-receptor system

Tocilizumab treatment for Cytokine Release Syndrome in hospitalized COVID-19 patients: survival and clinical outcomes

Evidence-based diagnosis and treatment of macrophage activation syndrome in systemic juvenile idiopathic arthritis

The Role of Cytokines including Interleukin-6 in COVID-19 induced Pneumonia and Macrophage Activation Syndrome-Like Disease

Treatment of Severe Acute Respiratory Syndrome With Glucosteroids

Dexamethasone in Hospitalized Patients with Covid-19 -Preliminary Report

The R Project for Statistical Computing

Comparison of the Efficacy and Safety of Tocilizumab Versus Methylprednisolone in the Cytokine Release Syndrome of Patients With COVID-19. A Prospective Randomized Controlled Phase II Trial

Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72 314 Cases From the Chinese Center for Disease Control and Prevention

Effect of Systemic Glucocorticoids on Mortality or Mechanical Ventilation in Patients With COVID-19

New-Onset Diabetes in Covid-19

Coronavirus Disease 2019 Test Results After Clinical Recovery and Hospital Discharge Among Patients in China