Discussion
In the present study, we explored the impact of antecedent SSRI use on functional status at discharge after AIS. Based on the multiple clinical studies suggesting a benefit for poststroke initiation of an SSRI,8 10–12 23–25 our hypothesis was that preSSRI use would portend favourable outcomes. Surprisingly, despite no differences in admission NIHSS score, treatment with intravenous tPA at our facility, endovascular thrombolysis or length of stay, patients taking an SSRI or other antidepressant prior to an AIS were less likely to be sent home or ambulate unassisted on discharge. These findings would suggest that preSSRI/antidepressant use is associated with worse poststroke outcomes, and are in contrast to several studies showing a benefit of poststroke SSRI initiation on motor recovery.8 12 24 26 Understanding the mechanism for these observations could prove informative in both advancing our understanding of SSRI-mediated effects on AIS and stroke recovery and targeting rehabilitation efforts in patients with depression.
Our analysis and that of the placebo-controlled study of fluoxetine for motor recovery after AIS (FLAME) have several important differences in terms of study design and patient demographics that merit consideration.8 First, in the FLAME trial, subjects were excluded if they had a diagnosis of depression or were taking antidepressant medications in the month prior to their admission.8 In contrast, all of our subjects were on SSRIs or other antidepressants prior to their admission and 6.4% had a diagnosis of depression. Additionally, there was a significant difference in stroke severity between the FLAME trial and our study. In the FLAME trial, the trial design selected for patients with moderate to severe stroke deficits, which resulted in 96%–100% of subjects having a baseline modified Rankin Scale score of 4 or 5.8 Our study, on the other hand, had much less severe stroke deficits, illustrated by the median NIHSS score of 4–5 among the groups. The differences in study design and patient demographics of our study and the FLAME trial certainly limit comparisons; however, an outstanding question is how preSSRI use modulates the decreased odds of discharge to home or ambulatory status at discharge.
There are several plausible explanations for our observed results. In our study, patients in the preSSRI or preAD group were more likely to have experienced a prior stroke or TIA and, despite no difference in admission NIHSS score, present with weakness as the initial exam finding. A parsimonious explanation for our findings would be that the preSSRI/preAD groups are biased towards worse motor outcomes given their pre-existing infarct burden and initial presentation with weakness. In addition, the increased rates of pneumonia in the preSSRI/preAD groups could impact functional status at discharge given the multifactorial inputs that influence ambulatory status. Arguing against these hypotheses, however, there was no difference in the length of stay or admission NIHSS scores between groups.
It is likely that the apparent role of preSSRI or antidepressant use on poststroke recovery is influenced by factors extending from the molecular to psychosocial levels. In animal studies of ischaemic stroke, SSRIs have been demonstrated to profoundly improve neurobehavioural outcomes.13 14 The beneficial effects of SSRIs in animals include reduced infarct volume,16 upregulated hippocampal neurogenesis15 and improved cerebral blood flow autoregulation.17 These possible mechanisms of action, at least in part, seem to be mediated by SSRI-induced upregulation of BDNF.18 19 27 At the neurobehavioural level, poststroke depression is exceedingly common with an estimated prevalence of 9%–34%28 29 and certainly has the potential to impact participation in acute, intensive rehabilitation efforts. Even prior to AIS, a large systematic meta-analysis showed an association between depression and increased risk of stroke.30 Whether preSSRI/antidepressant use is a marker of less favourable outcomes or worse motor deficits warrants further studies.
To better understand how preSSRI use could influence poststroke functional status, future studies should assess the influence on infarct volume and location, perfusion/diffusion mismatch and changes in BDNF levels in AIS. It has already been reported that in the acute poststroke period, there is a correlation between BDNF levels and functional status at 90 days.31 For this reason, it is conceivable that chronic BDNF upregulation from SSRIs could impact the acute/compensatory response of the normal brain to ischaemia.
There are several limitations to this study. First, the assessment of functional outcomes as ambulatory status at discharge and disposition, while both informative markers of functional status, shows each to be influenced by multiple factors that complicate speculation on the underlying cause. Future studies focused on more specific metrics, such as stroke location/subtype, infarct volume, changes in depression severity poststroke and more specific assays of motor weakness (eg, NIHSS motor score), would therefore be useful. Second, the preSSRI/preAD groups had significantly more prior strokes than the control group. Whether this discrepancy is reflective of the change in clinical practice because of FLAME and other studies8 23 26 or the incidence of poststroke depression, there is potential for bias. In the acute stroke period, prior compensated infarct burden could predispose to recrudesce and thereby influence functional status at discharge. Despite this observation, there was no significant difference between groups in ambulatory status on admission. Another limitation is the discordance between patients with a diagnosis of depression and preSSRI use in our study. This observation is likely a reflection of patients taking SSRIs for diseases other than depression and/or the relatively poor sensitivity of automated approaches for extracting depression diagnoses from the medical record.32 An additional consideration is whether the results of our single-centre analysis are generalisable to the broader population. In comparing our patient population with the nationwide GWTG-Stroke registry, there are several notable differences between our preSSRI population and the nationwide database, including age (68 vs 74 years), white race (87% vs 75.1%) and discharge to rehabilitation facility (46.1% vs 15.7%), which represent important considerations.22 Lastly, in this study and data registry, we lack information on the dose and duration of SSRI or antidepressant use, both of which would be informative in interpreting our observed results.
The importance of further investigations into this area is highlighted by the mixed results of clinical studies investigating SSRI use in humans with ischaemic or haemorrhagic stroke. An observational study of preSSRI use showed increased stroke severity and 30-day mortality in patients with haemorrhagic stroke but no association in patients with AIS.20 A retrospective study on preadmission SSRI or serotonin and norepinephrine reuptake inhibitor (SNRI) use showed higher inhospital mortality in patients admitted to the intensive care unit setting.33 A small, retrospective comparison of prestroke versus poststroke SSRI treatment suggested pretreatment was associated with favourable functional outcomes and ΔNIHSS score at discharge,9 although this interpretation was limited by sample size and intergroup differences in initial stroke severity and percentage receiving intravenous thrombolysis.