Reconciling aerosol generation data with actual rates of infection data

Two recent papers in Anaesthesia studied aerosol generation with anaesthesia (tracheal intubation and extubation) and appear to present opposing conclusions. Dhillon et al. [1] unambiguously conclude that “intubation and extubation are aerosol generating procedures” while Brown et al. do not “support the designation of elective tracheal intubation as an aerosol-generating procedure” [2]. No doubt the authors will need to ‘fight this one out’ amicably, and perhaps there are nuances that allow both of these statements to be at once partially true (see https://theanaesthesia.blog). Nevertheless, the ability to generate aerosol is just one side of the equation. The data (especially that of Brown et al. [2]) need to be reconciled with the best data we have of rates of infection amongst anaesthetists engaged in aerosol-generating procedures, which is that from the intubateCOVID study [3, 4]. Regardless of how much aerosol is generated, the reality is that an anaesthetist engaged in an aerosol-generating procedure in a COVID-19-positive patient has a 1 in 670 chance of contracting COVID-19 (positive test; or 1 in 240 chance of being ill with symptoms) [4]. This is despite the wearing of appropriate levels of personal protective equipment, and is a risk that accumulates binomially with continued work. This is really quite high and is entirely compatible with the results of Dhillon et al. [1]. In contrast, the results of Brown et al. [2] are compatible with these intubateCOVID data only if at least one of the following is true: COVID-19 is so infectious that even very few and/or very small particles will transmit disease; the sample size in Brown et al.’s study was too low to reflect clinical conditions of transmissibility [2]; and anaesthetists in the intubateCOVID study did not, in the main, contract infection from the aerosol-generating procedure(which was only a coincidental correlate) but from some other contact, either with patients in other contexts or with community contact. Binomial modelling of the intubateCOVID dataset allowed predictions of real infection rates among anaesthetists to be made; these departmental-level predictions matched actual rates at the time of publication in July 2020 [4]. Now, in November 2020 we can present updated data from one large Trust (department size 240) which shows near exact correspondence with the binomial modelling predictions (Fig. 1). In other words, if the intubateCOVID data are true and predicated on the assumption that aerosol-generating procedures are the source of infection, then binomial modelling of these is able to predict infection rates that match precisely real rates in anaesthetists over the entire course of the pandemic [4]. This, to some extent, validates the original assumption. This is consistent with the data of Dhillon et al. [1], but less so with the contention of Brown et al. that aerosol-generating procedures in COVID-positive patients are safe [2].