Analysis in matched sets
Before matching, 1016 Group A patients moderately differed from 95984 Group B patients in a range of covariates (listed in Table 2) (Appendix, Table A5), and 749 of the former could be exactly matched to 31336 of the latter (Appendix,Table A5). All but one Group A patients and 86.2% Group B patients were prescribed with at least one drug from the ATC groups N05 (psycholeptics), N06 (psychoanaleptics) and N07 (other nervous system drugs) at any time between January 1 2019 and the date of the index COVID-19 diagnosis. Next, 866/1016 Group A patients could be exactly matched to 222792/275804 Group C patients (Appendix , Table A6), and 82323/95984 Group B patients could be exactly matched to 268778/275805 Group C patients (Appendix, Table A7).
Incidence of COVID-19-related hospitalizations and of COVID-19-related mortality was <1.5% in all matched sets, while all-cause hospitalizations were considerably more common (Figure 3). There was no indication that outpatients prescribed fluvoxmine around the time of COVID-19 diagnosis (Group A) were at a reduced risk of any of the outcomes as compared to their peers burdened with similar psychiatric difficulties but not prescribed fluvoxamine over the critical period of time (Group B), or as compared to COVID-19 outpatients free of psychiatric difficulties and related treatments (Group C) (Figure 3): all relative risk estimates, frequentist and Bayesian with different priors, were around 1.0 or somewhat higher than 1.0. Comparisons between matched Group B and Group C patients were based on considerably larger number of subjects than other comparisons (Figure 3), hence estimates were much more precise (narrower confidence intervals), but in terms of the point-estimates, A vs. C and B vs. C differences were closely similar.