Response.

To the Editor:

We note with interest the letter from Trisolini et al in response to our recent report of nasopharyngeal-lung gradient in severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) among critically ill patients. They draw attention to the apparently contrasting findings of Geri et al, who found that BAL in patients who are not ventilated with hypoxemic respiratory failure with negative nasopharyngeal polymerase chain reaction (PCR) for SARS-CoV-2 identified only two additional cases of coronavirus disease 2019 (COVID-19). In our report, we noted that 33% of patients with positive deep lung samples (BAL or endotracheal aspirate) had negative nasopharyngeal PCR. Our finding of both false-negative nasopharyngeal swabs and higher viral load in the lungs is consistent with other reports. Wang et al noted a significantly higher positivity rate for BAL (93%) compared with a nasal swab (63%), findings which were replicated in a recent metanalysis of seven diagnostic studies that included those of Wang et al.

We believe the key to understanding these apparently divergent results lies in the differences in the populations that were examined. Our study and those analyzed by Bwire et al included only patients with PCR-confirmed COVID-19 and examined viral detection at different sites, whereas Geri et al investigated patients with undifferentiated respiratory failure. Our study included patients who were admitted at the peak of the first wave of the pandemic in the United Kingdom, and the ICUs that were involved were largely or completely occupied by patients with COVID-19, whereas the prevalence in centers in the study by Geri et al was 21%. Furthermore, all patients had sufficiently severe respiratory failure to merit admission to ICU, and all but one were receiving ventilatory support. Finally, 40% of the patients in the study of Geri et al had no evidence of viral pneumonitis on CT scans. Overall, it appears the divergent results arise from differing pretest probabilities of infection. With a relatively low pretest probability, it is perhaps unsurprising that Geri et al detected only an additional 2.5% cases by bronchoscopy. We do not believe this invalidates the use of deep lung samples to investigate undifferentiated severe respiratory failure, especially as we enter the influenza season in the northern hemisphere. Sampling of the distal lungs can aid the identification of both SARS-CoV-2 and other viral or bacterial pathogens, although the relative roles of endotracheal aspirate and BAL in this setting remain to be determined. When the implications of any study are being interpreted, it is vital to consider the population sampled and to be wary of applying findings to populations that were not well represented in the study under consideration.