Discussion

We observed no signal to indicate that outpatients prescribed with (and presumably exposed to) fluvoxamine around the time of COVID-19 diagnosis were at a reduced risk of subsequent hospitalization or death compared to their non-prescribed (non-exposed, control) peers. By definition, both subject subsets had to have ICD-10 code entries documenting history of a particular spectrum of psychiatric difficulties, and were then matched exactly on a range of psychiatric and other (co)morbidities, demographic and epidemiological characteristics. In both analyses (Study 1, Study 2), exposure to the treatment of interest (fluvoxamine) was always positively identifed, and in Study 2 it also included explicit exclusion of exposure to paroxetine. Control status was always defined by explicit exclusion of exposure to fluvoxamine, and in Study 2 it also included a positive identification of exposure to paroxetine. Apart from these definitions, both exposed and control subjects in both analyses could have been exposed (prescribed with) other psychiatric treatments including other antidepressants/anxiolytics, e.g., as a part of augmentation strategies and/or to treat residual symptoms [26-28]. Although we did not explicitly match patient subsets regarding these “other treatments”, it is reasonable to consider that potential imbalances in this respect were minor, if any, given that patients were exactly matched on a wide range of psychiatric diagnoses. Moreover, and as elaborated in the Introduction section, it is highly unlikely that any of the “other treatments” exerted any clinically relevant anti-COVID-19 effect. Furthermore, even the estimates corrected for a strong hypothetical confounding bias arising from a large imbalance between fluvoxamine-exposed and control subjects in prevalence of a highly effective anti-COVID-19 treatment did not indicate any relevant benefit of exposure to fluvoxamine. In this respect, and having in mind all the (elaborated) limitations for such extrapolations, the present data are more in line with RCTs not supporting a benefit of early fluvoxamine therapy in COVID-19 outpatients [3-6] than with RCTs suggesting a benefit [1,2].
Since based on administrative data, present work has several (inherent) limitations common to studies of this type. First, “(non)exposure” is implied based on prescription (non)issuance within a certain time-frame, however compliance and actual doses taken remained unknown. Next, information about presence and severity of symptoms at COVID-19 diagnosis was missing. To minimize the impact of this drawback, we restricted the analysis to patients diagnosed exclusively in outpatient settings, hence likely (at this point) suffering only milder symptoms (if any). Further, some inaccurracies in identification of exposures, comorbidities and outcomes were probable. We believe, however, that if present, such inaccurracies did not relevantly bias present observations: i) we used data managed by dedicated professional institutions; ii) data on key variables such as age, sex, vaccination status, date of COVID-19 test/test result or diagnosis were missing or erroneously entered in only 0.38% of the identified COVID-19 diagnoses (and these patients were excluded) indicating that if present, chance errors were sporadic; iii) it does not seem resonable to think that occurrence of chance errors is “prejudiced” in respect to (non)-issuance of fluvoxamine (or any other) prescriptions; iv) we left a period of a minimum one year + 2 months (from January 1 2019 to the first COVID-19 case in February 2020) to precede the index COVID-19 diagnosis not to miss entries of relevant comorbidities and issued prescriptions and other health care services into the Central Health Information System; v) raw incidence of all outcomes was within the expectations. Incidence of 30-day all-cause hospitalization was closely similar in all cohorts that included patients suffering psychiatric difficulties (around 12.0%), and these patients were also closely similar regarding age and comorbidities. Incidence was twice lower in patients free of such difficulties (Cohort C in Study 1), who were also younger and less comorbid. In Study 1 (larger), overall incidence of 6.9% (across all three cohorts) at the average age of 46.5 years is in agreement with expected 4.3% to 8.5% hospitalizations among people aged 40-49 years who test positive for COVID-19 [29]. Although one could consider all hospitalizations that occur within a month since the COVID-19 diagnosis as “COVID-19-related”, we defined a separate outcome where COVID-19 was the lead or at least one of the discharge diagnoses (the latter part of the definition implying that COVID-19 could have triggerred/worsened some underlying condition). It seems reasonable to assume that these were the “more severe” COVID-19 patients. Again, all cohorts including patients with psychiatric difficulties were similar in this respect (around 3.3%) and incidence was (expectedly) much lower (0.94%) in the younger and considerably less comorbid patients free of psychiatric difficulties (Cohort C in Study 1). The overall crude incidence across all cohorts in Study 1 of 1.5% is within the range of the reported expected incidence of severe/critical disease in 30-50-year olds who tested positive for COVID-19 (1.2-2.5%) [30]. The relationship between cohorts in Study 1 regarding (COVID-19-related) mortality was similar to that regarding other two outcomes, and the overall incidence (across all cohorts) of 2.5% is in line with the ratio of cumulative COVID-19-confirmed deaths and COVID-19 confirmed cases in Croatia up to October 31, 2021 [31]. Finally, due to exact matching on a number of covariates, matches between cohorts were found mainly among less comorbid subjects resulting in relatively low incidence of COVID-19 related death outcomes in matched sets and consequent imprecise estimates. However, all comparisons (A vs. B, or A or B vs. C in Study 1, and Fluvoxamine vs. Paroxetine in Study 2) were numerically closely similar indicating consistency of findings. Overall, it appears safe to conclude that we were able to resonably accurrately capture exposures, comorbidities, cotreatments and outcomes, and to adequatly control confounding. Under such circumstances, we observed no estimate that would go “in favor” of the fact of being prescribed fluvoxamine around the time of COVID-19 diagnosis.
In conclusion, the present population-based matched cohort studies strongly suggest that outpatients prescribed with fluvoxamine around the time of COVID-19 diagnosis are not at a reduced risk of subsequent hospitalizations or death compared to their peers suffering similar psychiatric difficulties but not prescribed with fluvoxamine, or prescribed paroxetine, or as compared to their peers free of psychiatric difficulties and respective treatments. Considering the specifics of the setting, extrapolation of the present data to the general question of efficacy of fluvoxamine in early COVID-19 can only be an approximation under several strong assumptions; in this context, present observations agree with trial data that failed to demonstrate a practically relevant benefit of fluvoxamine in this setting.