Search strategy
In this study, the published articles were collected from databases such as Science direct, Pub med, Web of Science, Scopus, Cochrane and Google scholar between January 2019 to April 2021.The search strategy was performed using English keywords as well as combination of important, and sensitive words. Search was done using the keywords "Tosilizumab", "Clinical trial", "COVID-19", "RR", "SARS-COV-2", "Risk Ratio", "Steroids", "Corticosteroids", "Methylprednisolone" by adding AND, OR in the title and abstract. In addition, the reference of articles was examined for the increasing of the search sensitivity. Finally, all references entered into reference management software (EndNote).
Inclusion criteria
In this research, the search strategy tool (PICO) aimed to determine the impact of patient (P), intervention (I), comparison (C), and outcome (O). "P" signifies COVID-19 patients in studies; "I" means intervention with Tosilizumab and steroids. "C" includes three groups of patients: 1- COVID-19 patients who treated with Tosilizumab and steroids. 2- COVID-19 patients who did receive Tosilizumab alone 2: COVID-19 patients who did receive standard of care treatment (control). "O" means the effectiveness of the Tosilizumab and steroids in case group.
All clinical trial studies were included in this meta-analysis for evaluating the efficacy of Tosilizumab and steroids in treatment of COVID-19.
Exclusion criteria
1- Those studies that scored less than 5 based on the quality assessment checklist. 2- Studies those patients treated with other antivirals 5. Case reports or case series studies. 5- Studies in non-English language.
Selection studies
In this study, the full text or summary of articles, documents and reports were screened and extracted. Firstly, the irrelevant articles were removed from this study. Then, by reviewing and studying, the case reports and review studies were omitted. Finally, according to inclusion and exclusion criteria, data were extracted from full-text articles.
Extracting the data
Screening and data extraction were performed by two independent reviewers based on article title, first author’s name, year of study, journal name, place of study, type of study, total number of patients in TCZ +/ CS group, number of recovered patients in TCZ +/ CS group, number of deaths in TCZ +/ CS group, total number of patients in control group, number of recovered patients in control group, and number of deaths in control group were entered into the excel.
Evaluation of quality
In this study, Newcastle-Ottawa scale (NOS) checklist (12) was used for the evaluation of the quality of articles. In this checklist, 9 questions were presented that cover the type of study, sample size, study objectives, study population, sampling method, data analysis method, presentation of findings in an appropriate way and presentation of results based on objectives. Each question was assigned a score. If a study scored at least 5 points, it was included in this study.
Analysis
Stata ver. 14 package (Stata Corp, College Station, TX, USA) was used to analyze the obtained data. Data were weighted and combined based on the inverse variance. Cochran (Q) and I2 tests were used to determine heterogeneity index between studies. Based on the heterogeneity results, a random effect model was used to estimate the risk of death in different types of treatment groups. Publication bias was estimated through the Egger's test and the impact of each study on the overall estimate was assessed by sensitivity analysis.