An observational study published in JAMA Network Open looks at mortality risk among patients with COVID-19 prescribed selective serotonin reuptake inhibitor antidepressants (SSRIs).
Prof Allan Young, Director, Centre for Affective Disorders, Institute of Psychiatry, Psychology & Neuroscience (IoPPN), King’s College London, said:
“The recent “Together” platform trial showed that treatment with fluvoxamine (100 mg twice daily for 10 days) among high-risk outpatients with early diagnosed COVID-19 reduced the need for hospitalisation defined as retention in a COVID-19 emergency setting or transfer to a tertiary hospital. Although a preliminary study, this trial was quite large (approximately 750 patients each in the fluvoxamine and placebo groups) and certainly seemed to indicate benefit from treatment of this group with this old, cheap antidepressant. The obvious question arises as to whether this is a benefit unique to fluvoxamine or are all selective serotonin reuptake inhibitors (SSRIs), associated with a lower mortality risk among patients with COVID-19? The present study examined electronic health records of over 80,000 patients diagnosed with COVID-19. A reduced relative risk of mortality was found to be associated with the use of SSRIs—specifically fluoxetine and fluvoxamine—compared with patients who were not prescribed SSRIs. Considered with the results of the Together trial, these findings suggest that SSRI use (perhaps just fluoxetine and fluvoxamine) may reduce mortality among patients with COVID-19. Although further studies are needed this raises the prospect of repurposing these antidepressants as treatments for COVID-19.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“This new piece of research does, as the researchers say, support some previous evidence from smaller-scale studies that taking certain antidepressant drugs, from the SSRI family, might be associated with reduced severity of Covid-19. The researchers, very sensibly, aren’t making any claims that taking this kind of drug definitely has an effect on the severity of the disease – just that their study and others indicate that it might well be worth examining that possibility in randomized clinical trials (some of which have already begun to take place). The word ‘associated’ is important here – this particular piece of new research can’t tell us whether taking a particular SSRI can cause a reduction in Covid severity, only that, on average, people who took certain SSRIs were less likely to die after being admitted to health care facilities with severe Covid-19. And the reasons for that might not be because they took the SSRIs – there are other possibilities.
“Despite the large number of patients included in the research, about 83,000 in all, the research can’t tell us about cause and effect because it’s observational. Patients weren’t allocated by the researchers to receive SSRIs or not. Patients were included in the study if they were diagnosed with Covid-19 and had emergency or urgent care treatment, or were admitted to hospital. They were prescribed SSRIs, or not prescribed them, by their doctors in the usual way, and that was recorded in their medical records. Most of the 83,000 were not prescribed SSRIs, but something over 3,000 of them were. The medical records also recorded whether or not the patients died, and, sadly, quite a lot of them did, whether they were taking SSRIs or not. It wouldn’t make much sense just to compare what percentage of the SSRI-takers and non-SSRI-takers died – in fact, a smaller percentage of those who weren’t prescribed SSRIs died, by quite a margin, which on the face of it might look as if taking SSRIs is bad for you. That comparison isn’t helpful, however, because there are a great number of other differences between the people who were and weren’t prescribed SSRIs, other than whether they were prescribed SSRIs or not. They differed in terms of whether they were admitted as a hospital inpatient (a much greater proportion of the SSRI-takers were admitted as inpatients), in terms of other pre-existing health conditions, and in very many other ways. Any of those ways might have been the real cause of any different in the chance of dying, and not the SSRI prescriptions at all.
“To get round this limitation as best they could, the researchers used a method called propensity score matching. This is a respectable and quite commonly used statistical approach, though it has been criticised in several ways. The idea behind it is to try to emulate what would happen in a randomised clinical trial, where some patients, chosen at random, would take SSRIs and the others would not. (If that had been done, and there were a difference in mortality rates, it would probably be caused by the SSRIs.) Instead, the researchers used a statistical model to calculate what’s called a propensity score for each patient, which is a measure of how likely it is that a person with similar characteristics to that patient would actually be prescribed an SSRI. Then, a patient who was prescribed an SSRI, and another who wasn’t, but who had the same chance of being prescribed the SSRI according to the propensity score, would be matched and included in the comparison that the researchers made.
“The idea is that two matched patients should, if the statistical model is adequate, have the same chance of being prescribed the SSRI, and so comparing them should be roughly equivalent to comparing two people, one of whom had at random been allocated to take an SSRI in a randomised trial, while the other one had not. In fact these researchers used a more complicated method of propensity score matching, matching each patient who took an SSRI to more than one control, and doing the whole propensity score matching process several times, to guard against the possibility that one set of matchings might lead to a misleading result, because of statistical variability. But none of that changes the overall idea.
“Though the whole process is intended to treat the results as far as possible as if they came from a randomised clinical trial, that just can’t change the fact that they don’t come from a randomised trial, so that one still can’t deduce that the drugs cause the outcome to change. The propensity score matching gets rid of some biases, but it doesn’t deal with everything. One snag is that it’s impossible to be sure that the statistical model that produces the propensity scores has taken into account everything that might have affected the decision to prescribe and SSRI or not. In this study, the propensity scores were calculated from a wide range of factors – the patient’s age, sex, race and ethnicity, the circumstance under which they were first recorded with Covid-19 in the health records, a long list of comorbidities (other diseases or health conditions) that they might have, and possible indications for whether they might benefit from an SSRI. But are these everything that a doctor might take into account on deciding whether to prescribe? If they aren’t, there could well be biases in the matching and hence in the results.
“As well as pointing out that the study can’t give evidence of cause and effect, the researchers point out some other limitations. Some information on previous health conditions was not in the database for all patients. The timing of taking the SSRIs was uncertain, to some extent.
“Overall, then, yes, the study does give some indication that taking some SSRIs, particularly fluoxetine and fluvoxamine, are associated with a lower chance of dying during a Covid-19 episode, in these patients. But the study can’t show that it was the SSRIs that reduced the chance of dying. Also, unfortunately, the study can’t tell us much specifically about the SSRI fluvoxamine, which has been discussed quite widely as a possible treatment for Covid-19, and for which there have been some positive results from trials. That’s because only 11 patients had fluvoxamine in this study, because it was observational and was not specifically planned to look at possible effects of fluvoxamine. Fluvoxamine just wasn’t prescribed very often by the doctors caring for these patients. Data from just 11 patients isn’t really going to tell you anything in this type of study (and the researchers certainly do not claim that it tells us anything specifically about fluvoxamine). So, in all, that’s why the researchers can’t conclude any more from this study other than that more research needs to be done.”
All our previous output on this subject can be seen at this weblink:
‘Mortality Risk Among Patients With COVID-19 Prescribed Selective Serotonin Reuptake Inhibitor Antidepressants’ by Tomiko Oskotsky et al. was published in JAMA Network Open at 16:00 UK time on Monday 15 November.
Prof Allan Young: “Employed by King’s College London; Honorary Consultant SLaM (NHS UK)
Deputy Editor, BJPsych Open
Paid lectures and advisory boards for the following companies with drugs used in affective and related disorders: Astrazenaca, Eli Lilly, Lundbeck, Sunovion, Servier, Livanova, Janssen, Allegan, Bionomics, Sumitomo Dainippon Pharma, COMPASS, Sage
Consultant to Johnson & Johnson
Consultant to Livanova
Received honoraria for attending advisory boards and presenting talks at meetings organised by LivaNova. Principal Investigator in the Restore-Life VNS registry study funded by LivaNova.
Principal Investigator on ESKETINTRD3004: “An Open-label, Long-term, Safety and Efficacy Study of Intranasal Esketamine in Treatment-resistant Depression.”
Principal Investigator on “The Effects of Psilocybin on Cognitive Function in Healthy Participants”
Principal Investigator on “The Safety and Efficacy of Psilocybin in Participants with Treatment-Resistant Depression (P-TRD)”
UK Chief Investigator for Novartis MDD study MIJ821A12201
Grant funding (past and present): NIMH (USA); CIHR (Canada); NARSAD (USA); Stanley Medical Research Institute (USA); MRC (UK); Wellcome Trust (UK); Royal College of Physicians (Edin); BMA (UK); UBC-VGH Foundation (Canada); WEDC (Canada); CCS Depression Research Fund (Canada); MSFHR (Canada); NIHR (UK). Janssen (UK)
No shareholdings in pharmaceutical companies
Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee. I am also a member of the Public Data Advisory Group, which provides expert advice to the Cabinet Office on aspects of public understanding of data during the pandemic. My quote above is in my capacity as an independent professional statistician.”
None others received.