# Mask trials reading notes ## Scope and approach This table extracts the ten randomized trials in the supplied full-text source pack only. Each row keeps the trial's randomized unit, assigned intervention, primary outcome, and primary analysis population together. DOI, publication year, title/source identity, and evidence locator are retained. “Unreported” means the requested item was not stated clearly enough in the supplied section used for extraction; no value was inferred from outside material. ## Trial identities 1. 2020 — *Facemask against viral respiratory infections among Hajj pilgrims: A challenging cluster-randomized trial* — DOI 10.1371/journal.pone.0240287. 2. 2022 — *Impact of community masking on COVID-19: A cluster-randomized trial in Bangladesh* — DOI 10.1126/science.abi9069. 3. 2016 — *Cluster randomised controlled trial to examine medical mask use as source control for people with respiratory illness* — DOI 10.1136/bmjopen-2016-012330. 4. 2015 — *A cluster randomised trial of cloth masks compared with medical masks in healthcare workers* — DOI 10.1136/bmjopen-2014-006577. 5. 2010 — *Surgical mask to prevent influenza transmission in households: a cluster randomized trial* — DOI 10.1371/journal.pone.0013998. 6. 2012 — *Facemasks, hand hygiene, and influenza among young adults: a randomized intervention trial* — DOI 10.1371/journal.pone.0029744. 7. 2012 — *The role of facemasks and hand hygiene in the prevention of influenza transmission in households* — DOI 10.1186/1471-2334-12-26. 8. 2011 — *A cluster randomized clinical trial comparing fit-tested and non-fit-tested N95 respirators to medical masks to prevent respiratory virus infection in health care workers* — DOI 10.1111/j.1750-2659.2011.00198.x. 9. 2009 — *Face mask use and control of respiratory virus transmission in households* — DOI 10.3201/eid1502.081167. 10. 2008 — *Preliminary findings of a randomized trial of non-pharmaceutical interventions to prevent influenza transmission in households* — DOI 10.1371/journal.pone.0002101. ## What can be compared Within a trial, randomized arms can be compared using the stated primary analysis and cluster-aware uncertainty. The Hajj trial compares an offer/instruction strategy with no mask supplied. The Bangladesh trial estimates a village-level mask-promotion package and reports prespecified pooled and mask-type estimates. Source-control household trials assign masks to infected index cases, whereas wearer-protection household trials assign masks to contacts; these address different mechanisms. Healthcare-worker trials compare occupational use during shifts. Cloth masks, medical/surgical masks, P2/N95 respirators, and multimodal mask-plus-hand-hygiene packages must remain distinct. ## What cannot be compared directly Outcome definitions differ: clinical respiratory illness, ILI, laboratory-confirmed influenza, any respiratory virus, and symptomatic SARS-CoV-2 seroprevalence are not interchangeable. Follow-up spans days to weeks, settings range from households and dormitories to hospitals, villages, and pilgrimage tents, and randomization occurs at different cluster levels. Event-rate differences, risk/rate ratios, odds ratios, and prevalence ratios are on different scales. Adjusted and unadjusted estimates should not be placed on a common numerical scale without the original model and denominator structure. The Bangladesh intervention combines mask distribution, communication, reminders, and role modelling, so it is not a pure material-efficacy test. The cloth-mask hospital trial used reusable cloth masks under a specific occupational protocol and should not be generalized to all cloth masks. The respirator trial's convenience no-mask group was not randomized and should not be treated as a randomized comparator. ## Adherence, crossover, and secondary analyses Adherence was often incomplete and sometimes differed sharply by arm. Control participants in the Hajj and Beijing source-control trials used masks, diluting assignment contrasts. Several papers report stronger effects among adherent participants, with early implementation, or in per-protocol/as-treated analyses. Those analyses may be informative about implementation but are vulnerable to post-randomization selection and are not substitutes for the randomized intention-to-treat result. The CSV therefore labels them as limitations/context rather than primary estimates. ## Missing or imprecise evidence Full-table review corrected the abstract-limited entries. Aiello 2012 Table 3 gives cumulative ITT ILI estimates for both intervention arms (unadjusted), while the Results paragraph/Table 4 gives adjusted cumulative laboratory-confirmed influenza estimates; these are labeled separately and are not mixed. MacIntyre 2011 randomized hospitals, not wards, and Table 2 supplies hospital-cluster-adjusted ITT odds ratios for fit-tested N95, non-fit-tested N95, and pooled N95 versus medical masks. MacIntyre 2009 Table 4 supplies the primary randomized ITT risk ratios and confidence intervals by household and by individual for all masks, surgical masks, and P2 respirators. The significant adherence analysis remains secondary. The 2016 source-control study reports 597 household members (302 mask arm and 295 control arm) in addition to 245 index cases. Cowling 2008 was also resolved from the full results. Of 198 randomized households, 128 were followed with home visits (370 contacts); 122 households had laboratory-confirmed influenza in the index case and contributed 350 contacts to crude secondary-attack analyses. Five contacts lacked age, so adjusted Table 3 included 345 contacts: 202 control, 60 face mask, and 83 hand hygiene. The adjusted laboratory-confirmed influenza estimates were OR 1.16 (95% CI 0.31–4.34) for face masks versus control and OR 1.07 (0.29–4.00) for hand hygiene versus control. These analyzed subsets are narrower than the randomized population and are explicitly labeled. No pooling, cross-trial effect synthesis, or clinical recommendation is provided.