Authors' response.

We thank the authors of the letter for reading our article1 with interest and emphasize the importance of appropriate use of personal protective equipment (PPE) but would guard against interpreting the importance of any one component over the others based on our study findings. Though there is a relative paucity of high-quality evidence on the role of PPE in averting infections, a recent Cochrane review has found that PPE made of more breathable materials may not be associated with higher infections, gowns provide better protection than aprons, spoken instructions provide fewer doffing errors and various ensembles of PPE sets do not have significant differences in infection events2. Since our study was not designed or statistically powered to examine relative protective effects afforded by various PPE components, we recommend that the associations be interpreted with caution and standard guidelines for PPE use be followed3.

Our study1 was undertaken to inform public health responses during the COVID-19 outbreak in the country. While we acknowledge the shortfall in reaching the calculated sample size, the response rate in our study has been higher than those reported in the literature from India and abroad4567. We also adopted several strategies to reduce the non-response rates, such as training of interviewers, multiple call attempts, targeted call times and establishing credentials and significance of the research topic at the beginning of the interview8910. We did not intend to match the cases and controls for gender and other demographic factors to avoid overmatching. As we selected them from the eligible pool (1073 SARS-CoV-2-infected and 20329 non-infected HCWs) in a random manner, any baseline differences that were captured in the cases and controls could be reflective of the existing differences in demographic variables in the databases forming the pools. Further, we chose a parsimonious model and avoided individual consideration of masks and gloves in the final multivariate model as the use of these items had a conceivable chance of being correlated. We agree that the sizes of some of the diagnostic subgroups in the multivariate model were small.

We would like to highlight that the authors of the letter were rightly alarmed by the lack of mask usage in HCWs, but some of them were HCWs in low-risk settings, such as administrative staff in healthcare setting or security personnel. Moreover, some of the responses could be timed to the earlier phase of the pandemic in the country when the use and availability of PPE was patchy and perceived risks between members within a particular occupational group also varied.