Introduction
Fluvoxamine, a selective serotonin re-uptake inhibitor (SSRI) has
attracted much attention in efforts on drug repurposing for COVID-19
disease. Two randomized controlled trials (RCTs) [1,2] indicated that it
reduced the risk of disease deterioration when started early upon
diagnosis, but other RCTs indicated no benefit [3-6]. Fluvoxamine came
into focus based on a reasonably sound pharmacodynamic rationale [7],
but also owing to early non-randomized observations (reviewed in [8,9]):
fluvoxamine was either proactively offerred and patients opted to take
it or not, or “standard” pharmacoepidemiological studies (on
administrative data) were performed. Regarding the former, it should be
noted that besides other potential limitations (like sample size,
confouding), studies of this type are burdened by uncorrectable
selection bias. Regarding the latter, it should be appreciated that in
the setting of fluvoxamine for COVID-19 outpatients, such studies of
interventions face obstacles beyond their standard limitations [10].
Since fluvoxamine has no approved use in infectious/inflammatory
conditions, people treated with fluvoxamine during early COVID-19 are
treated for some underlying psychiatric disorder, and mental disorders
might contribute to poorer COVID-19 outcomes (reviewed in e.g. [11,12]).
Hence, when using such data to evaluate benefits/harms of fluvoxamine,
one needs to separate the effects of two co-exposures (fluvoxamine,
underlying psychiatric condition): fluvoxamine exposed and non-exposed
subjects should come from the same population (i.e., people suffering
psychiatric difficulties). This limits the exposed vs. non-exposed
comparison to only a subset of people with COVID-19 and data are
informative about the treatment in broader terms (i.e., are
generlizable) only if its effect is not conditional on the
presence/absence of psychiatric difficulties. Furthermore, if
participants in such studies are identified in databases of hospitalized
patients or patients seeking help for COVID-19, estimates about the
“fluvoxamine effect” are likely to be biased: by selective inclusion one
conditions on a factor (a non-mild disease form) on a path between the
cause (COVID-19 infection) and the outcome (hospitalization/death) and
this is likely to generate spurious associations between the
intervention of interest (fluvoxamine) and the outcome [13]. On the
other hand, population-based studies embracing COVID-19 outpatients at
the time of the diagnosis are likely to be devoid of such biases, even
if including only COVID-19 positive patients, since there is
considerable evidence that psychiatric disorders (including mood
disorders and their treatments) do not affect one’s susceptibility to
SarsCov2 infection [14]. An important step is such an effort is
definition of “fluvoxamine non-exposed” subjects. People suffering
conditions requiring antidepressant/anxiolytic treatment not treated
with fluvoxamine are likely exposed to other treatments. Biological
(mechanistic) rationale to support their use in COVID-19 has been argued
for a range of antidepressants/anxiolytics not only for SSRIs [8,9],
with, seemingly, an emphasis on fluvoxamine and fluoxetine [15]. One
early observational study based on administrative data on COVID-19
patients identfied based on the fact of emergency department vistis of
hospitalizations suggested that those prescribed fluoxetine (n=470) had
by 28% lower mortality than propensity score-matched controls not
prescribed with any SSRI [16]. However, unlike for fluvoxamine, there
are no RCTs pertainting to fluoxetine or any other SSRI or non-SSRI
antidepressant/anxiolytic to indicate clinical benefit (reviewed in
[8,9]; we also could not identify any data by searching PubMed and OVID
Medline, Scopus, Web of Science, ClinicalTrials.gov and the
International Clinical Trials Registry Platform; September 25, 2022).
Therefore, “no exposure to fluvoxamine” seems to be an adequate
definition of a control condition vs. “exposure to fluvoxamine”.In
an attempt to estimate whether in adults diagnosed with COVID-19 in
outpatient settings the fact of being prescribed with (and presumably
exposed to) fluvoxamine at the time around the COVID-19 diagnosis
affected the probability of subsequent hospitalization or death, we
conducted two population-based matched cohort studies comparing
fluvoxamine-exposed outpatients to their non-exposed peers. Although the
study (primarily) pertains to people who at the time of COVID-19
diagnosis suffered psychiatric conditions for which fluvoxamine was one
of the possible treatments, it may still contribute to the overall
knowledge on utility of fluvoxamine in early COVID-19 treatment by
providing complementary information in the sense of being in agreement
or not with the estimates generated in RCTs.