No Blinding and No Protocol

Studies with unverifiable bias

Here are examples of comparative studies which made no mention of a protocol or of researcher blinding: Doung-ngern et al., Galow et al., Liu et al., Hong et al., Rebmann et al., and Sugimura et al.

Liu is the only prospective study, the rest are retrospective. One is a case control the others are comparative cohorts. As retrospective studies, they start from the outcome (Covid ascertained in a variety of usually ill-defined ways and then look at exposure (wearing or not wearing a mask). They then assess an association between the two.

If the analyst looking at the data is not blinded to who was wearing a mask and who was not, and there is no protocol to direct the analyses, the play of bias is almost certain, especially in a politically charged situation like mask policy.

These studies have a lot of other problems, which we have annotated in the next pages below each UKHSA extraction strip. They make their conclusions unreliable.

Doung-Ngern et al. is a retrospective case-control study of asymptomatic contacts of confirmed covid-19 cases, including 211 cases of covid-19 and 839 controls in Thailand. Telephone calls were used to obtain the details. Case and control definitions are rudimentary; however, the authors report phone questions carried out over the period April 30–May 27, 2020, relating to contacts in the month of March 2020.

The result showed that self-reported always wearing a mask all of the time during contact with a case compared to not wearing led to a lower risk of testing positive (adjusted OR 0.23; 95% CI 0.09–0.60). 29 cases reported wearing a mask all the time  (13.8%) versus 198 controls (24.1%).

It’s important to note that the OR should not be interpreted as though it is a relative risk.