These meta-analyses are going to be central to any review of the ways Covid was handled and the effectiveness and costs of approaches employed to reduce mortality, serious illness and the concomitant toll on health infrastructure. They avoid the opportunistic citing of selected parts of individual studies in a bid to support particular views.
But we are in an area where experimental controls are absent, and the statistical treatments used in component studies vary in their nature and robustness, and the levels of (unintentional) bias that infuse them. The meta-analysis can amplify these methodological problems by introducing its own bias in the selection, inclusion, weighting, statistical refinement and interpretation of data from constituent studies.
The phrasing of a single comment in the concluding remarks of this study (Herby et al) suggests a measure of editorial bias has crept in:
"Indeed, we do find some evidence that limiting gatherings was counterproductive and increased COVID-19 mortality"
Anyone who has prepared reports will recognise the technique for planting subjective conclusions not fully supported in the body of the study.
All that doesn't mean this the report is intentionally misleading, or that it is significantly flawed: but it's conclusions should be approached with the usual level of scepticism that surrounds analyses of his nature.