Interventions for treatment of COVID-19: Second edition of a living systematic review with meta-analyses and trial sequential analyses (The LIVING Project)
Bibliographic record
Abstract
BACKGROUND: COVID-19 is a rapidly spreading disease that has caused extensive burden to individuals, families, countries, and the world. Effective treatments of COVID-19 are urgently needed. This is the second edition of a living systematic review of randomized clinical trials assessing the effects of all treatment interventions for participants in all age groups with COVID-19. METHODS AND FINDINGS: We planned to conduct aggregate data meta-analyses, trial sequential analyses, network meta-analysis, and individual patient data meta-analyses. Our systematic review was based on PRISMA and Cochrane guidelines, and our eight-step procedure for better validation of clinical significance of meta-analysis results. We performed both fixed-effect and random-effects meta-analyses. Primary outcomes were all-cause mortality and serious adverse events. Secondary outcomes were admission to intensive care, mechanical ventilation, renal replacement therapy, quality of life, and non-serious adverse events. According to the number of outcome comparisons, we adjusted our threshold for significance to p = 0.033. We used GRADE to assess the certainty of evidence. We searched relevant databases and websites for published and unpublished trials until November 2, 2020. Two reviewers independently extracted data and assessed trial methodology. We included 82 randomized clinical trials enrolling a total of 40,249 participants. 81 out of 82 trials were at overall high risk of bias. Meta-analyses showed no evidence of a difference between corticosteroids versus control on all-cause mortality (risk ratio [RR] 0.89; 95% confidence interval [CI] 0.79 to 1.00; p = 0.05; I2 = 23.1%; eight trials; very low certainty), on serious adverse events (RR 0.89; 95% CI 0.80 to 0.99; p = 0.04; I2 = 39.1%; eight trials; very low certainty), and on mechanical ventilation (RR 0.86; 95% CI 0.55 to 1.33; p = 0.49; I2 = 55.3%; two trials; very low certainty). The fixed-effect meta-analyses showed indications of beneficial effects. Trial sequential analyses showed that the required information size for all three analyses was not reached. Meta-analysis (RR 0.93; 95% CI 0.82 to 1.07; p = 0.31; I2 = 0%; four trials; moderate certainty) and trial sequential analysis (boundary for futility crossed) showed that we could reject that remdesivir versus control reduced the risk of death by 20%. Meta-analysis (RR 0.82; 95% CI 0.68 to 1.00; p = 0.05; I2 = 38.9%; four trials; very low certainty) and trial sequential analysis (required information size not reached) showed no evidence of difference between remdesivir versus control on serious adverse events. Fixed-effect meta-analysis showed indications of a beneficial effect of remdesivir on serious adverse events. Meta-analysis (RR 0.40; 95% CI 0.19 to 0.87; p = 0.02; I2 = 0%; two trials; very low certainty) showed evidence of a beneficial effect of intravenous immunoglobulin versus control on all-cause mortality, but trial sequential analysis (required information size not reached) showed that the result was severely underpowered to confirm or reject realistic intervention effects. Meta-analysis (RR 0.63; 95% CI 0.35 to 1.14; p = 0.12; I2 = 77.4%; five trials; very low certainty) and trial sequential analysis (required information size not reached) showed no evidence of a difference between tocilizumab versus control on serious adverse events. Fixed-effect meta-analysis showed indications of a beneficial effect of tocilizumab on serious adverse events. Meta-analysis (RR 0.70; 95% CI 0.51 to 0.96; p = 0.02; I2 = 0%; three trials; very low certainty) showed evidence of a beneficial effect of tocilizumab versus control on mechanical ventilation, but trial sequential analysis (required information size not reached) showed that the result was severely underpowered to confirm of reject realistic intervention effects. Meta-analysis (RR 0.32; 95% CI 0.15 to 0.69; p < 0.00; I2 = 0%; two trials; very low certainty) showed evidence of a beneficial effect of bromhexine versus standard care on non-serious adverse events, but trial sequential analysis (required information size not reached) showed that the result was severely underpowered to confirm or reject realistic intervention effects. Meta-analyses and trial sequential analyses (boundary for futility crossed) showed that we could reject that hydroxychloroquine versus control reduced the risk of death and serious adverse events by 20%. Meta-analyses and trial sequential analyses (boundary for futility crossed) showed that we could reject that lopinavir-ritonavir versus control reduced the risk of death, serious adverse events, and mechanical ventilation by 20%. All remaining outcome comparisons showed that we did not have enough information to confirm or reject realistic intervention effects. Nine single trials showed statistically significant results on our outcomes, but were underpowered to confirm or reject realistic intervention effects. Due to lack of data, it was not relevant to perform network meta-analysis or possible to perform individual patient data meta-analyses. CONCLUSIONS: No evidence-based treatment for COVID-19 currently exists. Very low certainty evidence indicates that corticosteroids might reduce the risk of death, serious adverse events, and mechanical ventilation; that remdesivir might reduce the risk of serious adverse events; that intravenous immunoglobin might reduce the risk of death and serious adverse events; that tocilizumab might reduce the risk of serious adverse events and mechanical ventilation; and that bromhexine might reduce the risk of non-serious adverse events. More trials with low risks of bias and random errors are urgently needed. This review will continuously inform best practice in treatment and clinical research of COVID-19. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42020178787.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.133 | 0.273 |
| Meta-epidemiology (narrow) | 0.005 | 0.004 |
| Meta-epidemiology (broad) | 0.022 | 0.038 |
| Bibliometrics | 0.015 | 0.012 |
| Science and technology studies | 0.002 | 0.002 |
| Scholarly communication | 0.007 | 0.005 |
| Open science | 0.005 | 0.008 |
| Research integrity | 0.004 | 0.004 |
| Insufficient payload (model declined to judge) | 0.021 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".