A critical appraisal of a case-control study on healthcare workers

Sir,

We read the article with interest by Chatterjee et al^l, published recently and want to congratulate the authors for conducting this study in such a short span of time during this pandemic. We would like to contribute a few points.

In this study, controls are taken in less number than cases. During a case-control study, the desired ratio between cases and controls should be at least 1:1 and preferably may be 1:2 or even 1:32. This study was most appropriate to have 1:2 or 1:3 ratio as only a small proportion of healthcare workers (HCWs) were affected. Matching is usually done for age and sex between cases and controls3. Only symptomatic persons were considered in this study for cases and controls. Asymptomatic cases were not considered which constitute 40-45 per cent of the infective cases3. Hence, selection bias must have been introduced vitiating results. The proportion of non-responders is extremely high ranging from 32 to 40 per cent. The number of participants who completed interview was far less than the minimum required number. It is likely that the history of consuming hydroxychloroquine (HCQ) may be different among them. It is usual experience that non-respondents are either from uppermost or lowermost quartile4. High non-response rate certainly vitiates the results. The time period of enrolment and data collection constitutes approximately seven weeks. Chatterjee et al1 have not given precise information about details of consumption of HCQ. Initial guidelines were to take HCQ for seven weeks. There may be some HCWs who consumed HCQ for more than six weeks but stopped after that. The effect will be washed off. It was also not clear if the consumption history was at the time of telephonic interview or at the time of collection of sample. It could have been mentioned clearly. When adjusted odds ratio was calculated, there was no need to give crude odds ratio in the first few Tables. Further, the risk-benefit ratio of HCQ administration needs to be closely examined and this could have been easily done. It has major side effects such as retinal disorders, prolongation of QT interval and haemolytic anaemia5.

These are the few observations and comments from my side.