The effect of corticosteroids on mortality in hospitalized patients with COVID-19

In a recent study published on medRxiv* prepress server, researchers evaluated the impact of modern causal inference methods in estimating the effect of corticosteroids on coronavirus disease-related mortality 2019 (COVID-19).

Observational research has provided great opportunities to understand causal effects in the absence of randomized trials. However, the accuracy of observational research depends on various analytical and design choices.

About the studio

In this study, the researchers described the benefits of using modern causal inference methods compared to standard research approaches in estimating the effect of corticosteroids on coronavirus 2 (SARS-CoV-2) infections due to severe acute respiratory syndrome.

The approach proposed by the team, which used modern causal inference, was the first approach. This method used a target study framework and a plan that outlined causal inference. The plan used helped to design the study without time alignment and similar design biases and mitigated the bias of the incorrect model specification by estimating optimal machine learning. The team also described one randomized study that could hypothetically analyze the impact of corticosteroids on patients with severe COVID-19.

The population of the hypothetical study included adults admitted for COVID-19 to either New York-Presbyterian Hospital (NYPH) / Weill Cornell, Lower Manhattan Hospital or NYPH Queens. Infections were confirmed by performing reverse transcriptase-polymerase chain reaction (RT-PCR) assays on nasopharyngeal swabs taken from patients. The team randomized patients on the first day of admission to the hospital to receive either standard care with corticosteroids or standard care without any corticosteroid treatment.

The dose of corticosteroid administered was 0.5 mg per kg body weight of methylprednisolone corticosteroid, which was equivalent for 24 hours to six days. The corticosteroids included in the study were hydrocortisone, prednisolone, prednisone, methylprednisolone and dexamethasone. The team defined severe hypoxic criteria such as initiation of a high-flow ventilation mask, nasal cannulae, invasive or non-invasive mechanical ventilation, or an oxygen saturation level of less than 93%.

The primary endpoint of the study included mortality reported within 28 days of randomization. Of interest was the variation in mortality over 28 days between the two randomized groups.

The target study emulation used data from patients who met the inclusion criteria between March 3 and May 15, 2020. The team obtained demographic, intubation, comorbidity, death, and hospital discharge information from electronic medical records. Participants were followed for 28 days after hospitalization until they were discharged or transferred to another hospital. The team evaluated the impact of a hypothetical dynamic treatment plan in which each patient was treated with corticosteroids for six days if and when the patient had severe hypoxia.

Result

The study cohort included a total of 3298 patients with a median age of 65 years with almost 60% men and a mean body mass index (BMI) of 27. Approximately 31% of patients had diabetes mellitus, 14% had coronary artery disease, 54% had hypertension and 4.8% had kidney disease. A total of 1690 patients met the criteria for randomization for severe hypoxia, while 423 patients reported a history of corticosteroids during follow-up. In addition, 699 patients succumbed within 28 days of randomization.

The team also noted that 20% of patients who completed the study duration received no corticosteroid treatment, while 30% of patients died after corticosteroid treatment. The estimated mortality rate was 32% in subjects who did not receive corticosteroids, while 23% of patients who received corticosteroids for six days had severe hypoxia. This indicates a 9.6% reduction in mortality between the corticosteroid groups and the hypothetical non-corticosteroid groups.

In the target emulation study, among 1,690 patients who met hypoxia, 72 patients were treated with corticosteroids within one day of diagnosis of hypoxia, while 191 hypoxic patients received corticosteroids within five days. Among them, 18 and 451 patients succumbed to hypoxia within one and five days without any corticosteroid treatment.

Conclusion

Overall, the results of the study showed that the current standard used for clinical research cannot sufficiently restore an appropriate treatment approach, unlike the modern method of causal inference. The researchers believe that the current study has shown the importance of incorporating such innovative methods into future research proposals.

* Important notice

medRxiv publishes preliminary scientific reports that are not reviewed and therefore should not be considered conclusive, leading clinical practice / health-related behavior, nor should they be treated as validated information.