I think the question of mask wearing is one of: are multiple studies with a weak or neutral signal the evidence of a strong signal? Or is this the hallmark of pseudoscience instead?
I'll let you judge the studies for yourself if you're interested.
Aggarwal et al., 2020: Using results from 9 non-healthcare RCTs, found that mask use, both with hand hygiene (P=.714) and without (P=.226), was not associated with lower rates of ILI infection in community settings.
Brainard et al., 2020: Did not report any statistically significant results when analyzing RCT data. Reported that mask use was not associated with statistically significant reductions in ILIs when used by a well person (11.2% (116/1032) vs. 12.1% (127/1046), P=.68), when used as source control by an ill person in a home setting (5.6% (25/450) vs. 6.2% (28/453), P=.87), or when used by all parties in a home with a sick individual (11.0% (79/715) vs. 12.0% (107/890), P=.43).
Chaabna et. al, 2020: Reported a significant protective effect of medical facemask use when evaluated in conjunction with other interventions (e.g. handwashing) (6.8% (273/4029) vs. 9.8% (458/4677), 95% CI 0.54–0.81).
Chu et. al, 2020: Using data from six observational studies on SARS-CoV-1, reported a statistically significant reduction in infections associated with face masks (adjusted OR: 0.33) compared to no mask controls.
Jefferson et al,, 2020: Analyzing 15 RCTs, found no reductions in ILIs (RR 0.93, 95% CI 0.83-1.05) or influenzas (RR 0.84, 95% CI 0.61-1.17) for masks in the general population or healthcare workers (RR 0.37, 95% CI 0.05-2.50).
Liang et al., 2020: Using data from both observational and RCT studies, the authors reported a significant protective effect on lab-confirmed respiratory viral infection (5.9% (307/5217) vs. 12.1% (419/3469), P<.00001). Using RCT-only data, between-group differences declined (5.4% (44/816) vs. 7.8% (77/989)).
Ollila et al., 2020: Analyzing data from 5 RCTs, reported strong and statistically significant results in favor of face mask efficacy at maximum follow up (7.8% (297/3793) vs. 18.4% (704/3830); RR: 0.608). For 2 of the 5 papers studied the authors utilized data from face mask + other intervention arms instead of available data from face mask-only arm, which if used would lead to considerably different odds ratio of 14.3% (542/3793) and 16.4% (629/3830).
Perski et al., 2020: Authors considered 10 observational studies and 11 RCTs (only one of which found a reduction in self-reported ILIs in participants wearing face masks) and, using a Bayesian analysis, reported a “moderate likelihood of a small effect for the wearing of face masks” in reducing self-reported ILI (cumulative posterior odds=3.61), but determined that evidence was equivocal as to clinically- and laboratory-confirmed infections (cumulative posterior odds of 1.07 and 1.22, respectively).
Wang et al., 2020: Using 15 non-healthcare studies (10 observational and 5 RCTs), authors reported a slightly decreased pooled odds ratio (OR: 0.96, 95% CI 0.8–1.15) but the results were not statistically significant.
Xiao et al., 2020: Incorporating data from 10 RCTs in non-healthcare settings, reported no statistically significant effect for the use of masks on laboratory-confirmed influenza (2.3% (29/1276) vs. 3.3% (51/1567), P=.25).
Li et al., 2021: Using data from 6 COVID-19 case-control studies––5 in healthcare settings––to report a significantly-reduced risk of infection (11.4% (82/718) vs. 20.0% (202/1008); OR: 0.38). In the only non-HCW study considered the results were non-significant (12.8% (29/227) vs. 16.9% (102/602); OR: 0.72, 95% CI: 0.46–1.12).
Tabatabaeizadeh et al., 2020: Authors used data from 4 observational COVID-19 studies to conclude that mask-wearing is correlated with statistically significant risk ratio decrease of 0.12. However, 70.8% (n=5442) of the study’s total participants (n=7688) came from a single paper where participants used N95 respirators, not facemasks.
Coclite et al., 2021: Authors used data from 3 RCTs and 10 observational papers to conduct two separate meta -analyses. Concluded that neither RCT data (11.7% (187/1598)vs. 11.2% (272/2419); RR: 0.97, P=0.85) nor any of the observational data (cross-sectional: 20.2% (1302/6438) vs. 17.2% (1714/9975); RR: 0.90, 95% CI: 0.74–1.10) (case-control: 19.9% (138/694) vs. 40.5% (327/807); RR: 0.59, 95% CI: 0.34–1.03) (prospective: 20.5% (88/429) vs. 58.4% (310/531); RR: 0.55, 95% CI: 0.11–2.75)) were statistically significant.
Abullahi et al., 2020: Considering data from 2 RCTs and 3 observational studies in the SARS-CoV-1 and influenza contexts, authors failed to find a statistically significant benefit of face mask use (18.7% (142/758) vs. 33.1% (480/1451); RR: 0.78, P=0.52).
Nanda et al., 2020: Incorporating data from 7 RCTs (all previously discussed) evaluating ILI transmission, found no significant difference in infection between mask and no-mask groups (2.8% (37/1301) vs. 3.6% (57/1592); RR: 1.00, P=0.93).
