Implementation of Recommendations on the Use of Corticosteroids in Severe COVID-19

Key Points Question How did the percentage of patients with severe COVID-19 who received treatment with corticosteroids vary over time (January 2020-September 2022) and across World Health Organization geographic regions? Findings In this cohort study of 434 851 patients with severe COVID-19, corticosteroid use increased over time but unequally across geographic regions. The statistical associations between increased corticosteroid use and publication of the Randomised Evaluation of COVID-19 Therapy trial and World Health Organization guidelines were significant only in Europe, where most of the trial recruitment occurred. Meaning The findings of this study suggest that a lack of research representativeness may hinder guideline implementation.


Introduction
The COVID-19 pandemic was undeniably a global crisis.In every country, large numbers of individuals required hospitalization following SARS-CoV-2 infection, and many died. 1 The response of the international research community was also unprecedented.A surge of studies emerged, many making use of novel designs, rapidly followed by evidence syntheses and, eventually, by clinical practice guidelines created under the auspices of the World Health Organization (WHO).To date, those guidelines include strong recommendations in favor of a limited number of therapeutic interventions for severe (ie, corticosteroids, interleukin 6 receptor blockers, and baricitinib) and nonsevere (nirmatrelvir-ritonavir) COVID-19. 24][5][6][7] However, guideline implementation may vary regardless of the underlying methodological rigor according to intrinsic (ie, related to the guideline) or extrinsic (ie, related to the context in which guidelines are implemented) barriers and facilitators. 8,9Given the global scale of the COVID-19 pandemic and the scope of clinical practice guidelines, questions related to the implementation of recommendations by region are particularly relevant. 102][13][14] The main objective of this cohort study was to compare variations over time and across WHO regions of corticosteroid use for severe COVID-19.The first secondary objective was to evaluate the association between publication of the RECOVERY (Randomised Evaluation of COVID-19 Therapy) trial 15 and of the WHO guidelines 2 and the temporal trends in corticosteroid use by region.The second secondary objective was to describe the geographical distribution of the recruitment in clinical trials that informed the WHO guidelines.

Study Design and Participants
This prospective observational cohort study is nested within the ISARIC-WHO Clinical Characterisation Protocol for Severe Emerging Infections. 16,17Information on informed patient consent, the case report forms, and the study protocol are available on the ISARIC website. 18As of September 2, 2022, investigators from 63 countries used the data collection instruments, which were adapted for a range of resource settings, to prospectively collect data using the ISARIC case report form built using the Research Electronic Data Capture (version 8.11.11,Vanderbilt University) hosted by the University of Oxford (Oxford, England).In addition, some collaborating investigators collected data using locally available systems and submitted data to ISARIC for centralized mapping.
All investigators retained full rights to their data.The WHO Ethics Review Committee approved the ISARIC-WHO Clinical Characterisation Protocol.In addition, local ethics approval was obtained for each participating country and site according to local requirements.This report adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for observational studies.
The inclusion criteria were laboratory confirmation of SARS-CoV-2 infection (as defined by the WHO 2,19 ) and admission to the hospital for acute illness due to COVID-19.Patients were excluded if

JAMA Network Open | Equity, Diversity, and Inclusion
Implementing Recommendations on the Use of Corticosteroids in Severe COVID-19 they did not have a country code or admission date.We collected information on the following variables defined in the ISARIC data dictionary: age, sex, use of corticosteroids, use of oxygen therapy, admission to the intensive care unit, a selection of comorbidities (hypertension, diabetes, HIV infection, asthma, chronic pulmonary disease, obesity, and tuberculosis), country, country's income level (lower, lower-middle, and upper-middle vs high income), admission date, and site.

Operational Definitions of Severe or Critical COVID-19 and of Corticosteroid Use
Cases were categorized as severe or critical COVID-19 if they met any of the following criteria: receipt of oxygen therapy (such as noninvasive positive pressure ventilation, high-flow oxygen nasal cannula, noninvasive ventilation, extracorporeal membrane oxygenation, artificial respiration, oxygen, prone body position, insertion of a tracheostomy tube, removal of an endotracheal tube, or intubation) or admission to an intensive care unit during hospitalization.Corticosteroid use was defined by the administration of any systemic (ie, intravenous or oral) corticosteroid agent at any point during the patients' COVID-19 hospitalization episodes.The ISARIC database was reviewed for the following terms: steroids, corticosteroids, hydrocortisone, dexamethasone, methylprednisolone, prednisolone, or prednisone.Patients who were already treated with corticosteroids for chronic conditions before their hospitalization were not considered as having received corticosteroids for treatment of COVID-19.No restrictions were imposed based on the corticosteroid dose or duration.

Statistical Analysis
The host repository at the University of Oxford curated data from contributing sites, which used a variety of local data systems, into the Study Data Tabulation Model standards, version 1.7 (Clinical Data Interchange Standards Consortium).The requested data fields for the data extraction period from January 31, 2020, to September 2, 2022, were sent to a secure server of the Study Design and Biostatistics Center at the Université de Sherbrooke, Quebec, Canada.Because the ISARIC collaboration process accommodates sites with a wide range of data collection capabilities, the proportion of complete case reports available for analysis varied substantially across sites and global regions.Our models were based on complete cases (ie, we did not impute missing data).Analyses were conducted using R, version 4.0.3(R Project for Statistical Computing).To account for multiple testing, we applied Bonferroni corrections to all statistical tests.We reported continuous variables as means (SDs) or medians (IQRs), as appropriate, and categorical variables as counts (percentages).
To compare variations in corticosteroid use for severe COVID-19 across WHO regions (objective 1), we estimated the percentages and corresponding confidence intervals (CIs) of patients with severe COVID-19 who received corticosteroids overall and within WHO geographic regions.Two-sided Wald-type CIs were calculated, with Bonferroni correction applied to maintain a global joint CI of 95%.A χ 2 test at a significance level of α = 5% was used to test for differences in the percentages of patients who received corticosteroids.
To evaluate the association between publication of the RECOVERY trial 15 and of the WHO To describe the geographic distribution of the recruitment in clinical trials that informed the WHO guidelines (objective 3), we reviewed the published reports of meta-analyses and clinical trials that informed the guidelines.2][23][24][25][26][27] Information regarding the countries in which participants were recruited appears in the appendices of the meta-analysis as well as the primary clinical trials. 2,20,21e concurrent recommendation against corticosteroid use in nonsevere COVID-19 was informed entirely by the nonsevere subgroup of the RECOVERY trial, 15 which enrolled participants in the United Kingdom exclusively.We created heat maps using R packages ggplot2, rgeos, rworldmap, and maps to report the distribution of clinical trial participants alongside the global distribution of ISARIC cases between January 31, 2020, and September 2, 2022.

Results
The total ISARIC cohort comprised 823 771 patients hospitalized for COVID-19 (mean [SD] age, 55.6 [21.4] years; 50.6% female and 49.4% male).Of these, severity and corticosteroid use could be ascertained for 784 101 patients from 54 countries (eTable 1 in Supplement 1).The age and sex of 39 670 patients who were excluded from this analysis due to missing data regarding COVID-19 severity and corticosteroid use are provided in eTable 2 in Supplement 1.

Variations in Corticosteroid Use
Among 434 851 patients with confirmed severe or critical COVID-19 for whom receipt of corticosteroids could be ascertained, 174 307 (40.1%) received corticosteroids during the study period.In all regions, corticosteroid use for treatment of severe cases increased over time (Table 3; Figure 1).Globally, the number of severe or critical patients in the ISARIC data set who received corticosteroids was 16  Corticosteroid use increased abruptly and significantly in Europe both in June 2020 (timeinterruption coefficient for the uptake, 1.0 [Bonferroni-corrected 95% CI, 0.9-1.2])and September 2020 (time-interruption coefficient for uptake, 1.9 [95% Bonferroni-corrected CI, 1.7-2.0]).The increase was not associated with those dates in the other regions (ie, all other Bonferroni-corrected 95% CIs for the associated time-interruption coefficients contained the value 0 and were, therefore, not statistically significant).

JAMA Network Open | Equity, Diversity, and Inclusion
Implementing Recommendations on the Use of Corticosteroids in Severe COVID-19

Geographic Distribution of Trial Participants vs ISARIC Cases
The WHO recommendation to administer corticosteroids to patients with severe and critical COVID-19 hinged on data from 8 clinical trials [21][22][23][24][25][26][27] that enrolled a total of 7184 participants (including 1 trial 28 that remained unpublished as of publication of the present study).Of the trial participants included, 91.6% were recruited from the United Kingdom.Figure 2 illustrates the geographic distribution of patients hospitalized for COVID-19 who were included in the ISARIC database and who were recruited in clinical trials of corticosteroids.

Discussion
The results of this cohort study including 434 851 patients hospitalized for severe COVID-19 across 54 countries from all regions of the globe suggest that while corticosteroid use increased in all regions, important geographic variations persisted until the end of the study period, September 2022.
Moreover, the timing of changes observed in corticosteroid use coincided with publication of the  RECOVERY trial 15 and of the WHO guidelines 2 for corticosteroids, most evidently in Europe, which is also where the clinical trials recruited a majority of the research participants.
Although many other factors influence guideline implementation, these results underscore the need to address the issue of global clinical research representativeness.Overwhelmingly, clinical trial data are derived mainly from a limited number of high-income countries, even when these countries bear a relatively small portion of the global burden of a disease (eg, sepsis 29 ).In comparison with many other therapies, corticosteroids are widely regarded as safe, available, and inexpensive.In this context, it is noteworthy that this study found that uptake of the WHO recommendation was more modest and slower in regions that were underrepresented in the clinical trials that informed the guidelines.While it is unclear to what extent a lack of trust in the regional applicability of the underlying trial evidence explains the heterogeneous guideline implementation, it is widely accepted that diversity and representativeness in research is paramount in "building trust, promoting fairness, and generating biomedical knowledge." 30Ultimately, it may prove an uphill battle to argue for global implementation of guidelines

guidelines 2
and the temporal trends in corticosteroid use by region (objective 2), we used timevarying curve estimates and defined 3 study periods: from the beginning of the COVID-19 pandemic (January 31, 2020) to June 1, 2020 (publication of the RECOVERY trial results for corticosteroids 15 ); between June 1 and September 1, 2020 (publication of the WHO guidelines for use of corticosteroids in the treatment of COVID-19 2 ); and from September 2, 2020, to September 2, 2022 (end of the study period).We looked for abrupt changes in trends of corticosteroid use over time and across regions using a time-interrupted logistic regression model.The model incorporated 3 variables: corticosteroid use status, time (grouped by month and year), and WHO region (Europe, Africa, Americas, South-East Asia, Eastern Mediterranean, and Western Pacific).Natural cubic spline additive components were used to model smooth time variations that could be attributed to regionspecific effects, and time-interruption parameters together with their interactions with the WHO regions variable were added to allow for potential trend interruptions after June 1 and September 1, 2020.To compare region-specific time-interruption parameters, we computed 2-sided CIs for the JAMA Network Open | Equity, Diversity, and Inclusion Implementing Recommendations on the Use of Corticosteroids in Severe COVID-19 model coefficients at a global level of 95% using the Wald method and Bonferroni correction.A likelihood ratio test based on the χ 2 statistic was used at a significance level of α = 5% to determine the significance of any time-interruption parameter.We also compared the percentages of patients who received corticosteroids before June 1 and after September 1, 2020, overall and by WHO geographic region based on CIs.The time-interrupted logistic regression model was based on data collected between March 1, 2020, and May 31, 2022, despite the available data ranging from January 31, 2020, to September 2, 2022, because of insufficient data points in some regions before March 1, 2020, and after May 31, 2022.

Figure 2 .
Figure 2. Global Distribution of Severe COVID-19 Cases in the International Severe Acute Respiratory and Emerging Infections Consortium (ISARIC) Database and in the Clinical Trials That Informed World Health Organization Guidelines for Corticosteroid Use in COVID-19 ISARIC severe and critical COVID-19 cases with available information on receipt of corticosteroids A

Table 1 .
General Characteristics of the ISARIC Cohort of Patients With Severe COVID-19 by World Health Organization Geographic Region Abbreviations: ICU, intensive care unit; ISARIC, International Severe Acute Respiratory and Emerging Infections Consortium.

Table 2 .
Characteristics of Participating Sites by World Health Organization Geographic Region Patients from South Africa were treated across 631 hospitals.Although information was obtained regarding the treatments and outcomes of individual patients, the distribution of patients across South African sites, as well as site-level characteristics, are missing.Accordingly, 631 sites from South Africa were removed from these proportions. a

Table 3 .
Receipt of Corticosteroids Among Patients in ISARIC Data Set With Severe or Critical COVID-19 Disease, by Time Period and World Health Organization Geographic Region Figure 1.Corticosteroid Use Among Severe Cases of COVID-19 in the International Severe Acute Respiratory and Emerging Infections Consortium (ISARIC) Cohort

JAMA Network Open | Equity, Diversity, and Inclusion Implementing
Recommendations on the Use of Corticosteroids in Severe COVID-19 Countries Represented in the ISARIC Cohort Study Conducted Between January 31st 2020 and September 2nd 2022 eTable 2. Age and Sex of Cases With Missing Data for Illness Severity and/or Corticosteroid Use (Excluded From This Analysis)