Pre-clinical studies provide clear and consistent evidence that a variety of centrally acting drugs affecting specific neurotransmitters can either facilitate or interfere with functional recovery after brain injury. Although at least some clinical trials suggest similar effects in humans, results have been inconsistent. The impact of important factors such as drug dose, duration, and intensity of physiotherapy, and timing between injury and treatment are difficult to translate from preclinical studies. Issues related to variability in stroke severity, location of injury, and comorbid conditions further complicate trial design and could obscure a true treatment effect. Because of these and other issues, the design of efficacy trials assessing putative neuro-restorative interventions is not trivial. Although a proven pharmacological approach resulting in a clinically meaningful improvement in post-stroke recovery remains elusive, it is reasonable to avoid medications that may have harmful effects in patients who have had a stroke. It is also important to control for these possible harmful effects in future clinical trials assessing the outcomes of stroke patients after the acute period.
Preclinical studies show that a variety of systemically administered drugs can affect post-stroke functional recovery; however, data from prospective clinical trials are inconsistent. There are likely many reasons for these failures of translational medicine, yet there is reason to believe that the identification of effective pharmacological approaches for improving functional outcomes after stroke may be possible (Cramer, 2011). Novel strategies include stem cell transplantation, hormonal therapy, and the administration of exogenous growth factors; statins hold promise. This discussion focuses on the potential effects of medications acting on central neurotransmitters on the recovery process.
Amphetamine is among the most extensively studied, systemically administered drugs with the potential to modulate the recovery process after stroke and traumatic brain injury. Experiments indicating that the administration of amphetamine beneficially affects functional outcome date to at least the 1940s (Maling and Acheson, 1946). Nearly 40 years later, the drug’s therapeutic potential was highlighted by a series of experiments by Feeney and coworkers (Feeney et al., 1981, 1982). Their work showed that when combined with task-relevant training, a single dose of d-amphetamine given 24 hours following unilateral sensorimotor cortex ablation in the rat increased the rate of locomotor recovery, an observation confirmed in other laboratories (Dunbar et al., 1989; Goldstein and Davis, 1990b; Goldstein, 1995; Goldstein and Bullman, 1999). Additional studies found that post-injury administration of amphetamine not only accelerated the rate of recovery, but also could promote the recovery of otherwise permanent neurological deficits, extending the data from rodents to other species. For example, in cats, amphetamine given after unilateral or bilateral frontal motor cortex ablations leads to a restoration of both locomotor ability (Meyer et al., 1963; Hovda and Feeney, 1984; Sutton et al., 1989) and tactile placing (Feeney and Hovda, 1983). After occipital lobe injury, amphetamine combined with visual experience leads to the recovery of stereoscopic vision (Feeney and Hovda, 1985; Hovda et al., 1989) in addition to tactile placing (Hovda et al., 1987) and to the recovery of sensory function after ischaemic injury to the barrel cortex in rats (Hurwitz, et al., 1989, 1990; Dietrich et al., 1990). Improved recovery of motor function with amphetamine administration was also found after middle cerebral artery distribution infarction (Stroemer et al., 1994), traumatic brain injury (Prasad et al., 1995), and skilled reaching after both cortical ischaemia (Adkins & Jones, 2005) and ablation injury (Ramic et al., 2006). In a primate model, d-amphetamine combined with training led to a long-term improvement in motor task performance after cortical infarction in squirrel monkeys (Barbay et al., 2006). Although largely positive, there are reports of other studies that failed to find a benefit of amphetamine, possibly due to differences in experimental conditions (Schmanke et al., 1996; Dose et al., 1997; Brown et al., 2004; Alaverdashvili et al., 2007).
Amphetamine may affect the release and activity of several neurotransmitters including dopamine, serotonin, and norepinephrine (Fuxe and Ungerstedt, 1970), but several lines of evidence suggest its action on recovery is mediated through epinephrine. Intraventricular infusion of norepinephrine in rats facilitates motor recovery, mimicking the amphetamine effect (Boyeson and Feeney, 1990). Similar infusion of dopamine is ineffective if its conversion to norepinephrine is blocked by the administration of a dopamine-beta-hydroxylase inhibitor (Boyeson and Feeney, 1990). Selective depletion of central norepinephrine with (N-[(2-chloroethyl)]-N-ethyl-2-bromobenzylamine; DSP-4) impairs motor recovery after a later injury to the cerebral cortex (Goldstein et al., 1991; Boyeson et al., 1992a). Selective injury to the locus coeruleus, the primary source of central noradrenergic projection fibres, also has a detrimental effect on motor recovery after a subsequent injury to the rat sensorimotor cortex, although the effect varies among experiments (Weaver et al., 1988; Boyeson et al., 1992b; Boyeson et al., 1993; Goldstein, 1997). When performed at least 2 weeks before a unilateral sensorimotor cortex ablation, bilateral, contralateral, and ipsilateral locus coeruleus lesions block locomotor recovery (Goldstein, 1997).
Because each locus coeruleus has widespread projections, selective lesions of the ipsilateral and contralateral dorsal noradrenergic bundle were used to determine whether norepinephrine levels in the damaged or undamaged hemisphere correlated with recovery. Locomotor recovery was impaired by contralateral but not ipsilateral dorsal noradrenergic bundle lesions and the overall rate of recovery correlated with norepinephrine content in the contralateral, but not ipsilateral, cerebral hemisphere (Goldstein and Bullman, 1997). The observation is consistent with functional magnetic resonance imaging (MRI) studies in humans that find changes in the cerebral hemisphere contralateral to a stroke that correlate with motor recovery (Cramer et al., 1997; Bütefisch et al., 2005).
The neurobiological processes underlying amphetamine (i.e. norepinephrine) modulated recovery remain uncertain. One theme that arises from laboratory experiments is that the effect of d-amphetamine on recovery depends on concomitant training. In the original experiments by Feeney and colleagues (1982), the amphetamine effect was blocked if the rats were not allowed to walk rather than being trained in conjunction with drug administration. There was no drug effect if rats were handled rather than trained, suggesting that specific locomotor experience was required (Goldstein and Davis, 1990c). Motor recovery in cats was not enhanced with d-amphetamine administration in the absence of training, and there was no reinstatement of depth perception in cats after visual cortex ablation if they were kept in the dark after being given d-amphetamine (Hovda and Feeney, 1984; Feeney and Hovda, 1985).
d-Amphetamine given in addition to training leads to neuronal changes in both the ipsilateral and contralateral cerebral cortex (Stroemer et al., 1995). The combination of post-injury training and amphetamine administration is associated with an increase in projection fibres from the contralateral, non-injured cerebral cortex to the pontine motor nuclei (Ramic et al., 2006). Therefore, amphetamine may induce long-term changes in neuronal structure that could, in part, underlie its impact on recovery.
d-Amphetamine’s effect on locomotor recovery after sensorimotor cortex injury in the rat is dose dependent (Goldstein, 1990). There is increasing benefit as the dose is increased to 2.6 mg/kg base weight, but decreasing benefit as the dose is increased further. In cats, motor recovery after cortex injury is facilitated when a series of amphetamine/training sessions are carried out every 4 days for 2 weeks as compared with a single session (Hovda and Feeney, 1984). Amphetamine improves binocular depth perception in visually decorticated cats when given beginning 10 days after injury, but there is no benefit if treatment is delayed for 3 months (Feeney and Hovda, 1985).
Several principles arise from these experimental studies. Some systemically administered drugs such as d-amphetamine may modulate post-brain injury recovery. The effect depends on concomitant behavioural experience/training. Timing and dosing are critical. For d-amphetamine, at least part of the drug effect appears to be exerted in the cerebral hemisphere contralateral to the injury.
|Transmitter or drug class/drug||Action||Recovery effect|
|Trazodone||5-HT reuptake blocker||–||+|
|Fluoxetine||5-HT reuptake blocker||Neutral||+/Neutral|
|Desipramine||NE reuptake blocker||+||…|
|Amitriptyline||Mixed 5-HT and NE reuptake blocker||–/Neutral||…|
* The effects of selected drugs on recovery in laboratory animal models and in preliminary clinical studies in humans recovering from stroke. GABA indicates γ-aminobutyric acid; AR, adrenergic receptor; 5-HT, serotonin; NE, norepinephrine; +, a beneficial effect; ?, the drug has been insufficiently tested in humans to reach preliminary conclusions about its effect on recovery; –, a detrimental effect; symbols in parentheses, drugs that were included in the list of potentially harmful agents in 2 retrospective studies but have not been examined separately; and ellipses, lack of data.
Given the hypothesis that d-amphetamine acts through norepinephrine, other centrally acting noradrenergic drugs would be anticipated to affect post-brain injury recovery. Motor recovery is accelerated by the norepinephrine precursor, L-threo-3,4-dihydroxyphenylserine (L-DOPS) (Kikuchi et al., 1999, 2000). Effects on recovery similar to amphetamine also occur after administration of phenylpropanolamine (Chen et al., 1986), phentermine (Hovda et al., 1983), and, depending on dosing regimen, methylphenidate (Kline et al., 1994). Further illustrating the complex interaction of dose and experience, the effect of methylphenidate depends on the number and timing of treatment sessions (Kline et al., 1994). Single doses of the α2-adrenergic receptor antagonists idazoxan and yohimbine, which increase synaptic norepinephrine release, facilitate motor recovery when given to rats after unilateral sensorimotor cortex injury (Goldstein et al., 1989; Feeney and Westerberg, 1990; Sutton and Feeney, 1992).
Antihypertensives crossing the blood–brain barrier that act on noradrenergic receptors have the potential to affect post-brain injury recovery. The α2-adrenergic receptor agonist clonidine (Goldstein and Davis, 1990a) and the α1-adrenergic receptor antagonists phenoxybenzamine (Feeney and Westerberg, 1990) and prazosin (Feeney and Westerberg, 1990; Sutton and Feeney, 1992) interfere with locomotor recovery after cortex injury. Deficits also transiently re-emerge in animals that recovered motor function when given either clonidine or prazosin (Sutton and Feeney, 1992). In contrast to drugs active at α-adrenergic receptors, the β-adrenergic receptor antagonist propranolol has no effect on locomotor recovery (Feeney and Westerberg, 1990).
Feeney and colleagues (1982) reported not only that d-amphetamine coupled with training facilitated locomotor recovery in rats, but also that even a single dose of haloperidol was harmful. Haloperidol administration also blocked d-amphetamine-facilitated recovery of stereoscopic vision in visually decorticated cats (Hovda et al., 1989). Butyrophenones such as fluanisone and droperidol reinstate neurological deficits in rats that recovered motor function after cortex injury (Van Hasselt, 1973). Although these observations raise the possibility that dopamine, in addition to norepinephrine, might modulate post-brain injury recovery, as noted above, intraventricular administration of norepinephrine improves recovery similar to d-amphetamine, but intraventicular dopamine is neutral (Boyeson and Feeney, 1990). Because haloperidol and other major tranquillizers also have effects at noradrenergic receptors (Peroutka et al., 1977; Cohen and Lipinski, 1986), it is hypothesized that their impact on recovery is noradrenergically mediated. The relative dose-dependent detrimental effects of haloperidol and clozapine on post-sensorimotor cortex injury in rats is correlated with their relative potencies at noradrenergic receptors (Goldstein and Bullman, 2002).
Antidepressants affect the reuptake and metabolism of a variety of central neurotransmitters, including norepinephrine. Serotonin, a target of several antidepressants, modulates the release of norepinephrine through activation of 5-HT3 and possibly 5-HT1C receptors in rat hippocampal neurones (Blandina et al., 1991). In addition, various antidepressants, including fluoxetine, induce neurogenesis in the hippocampus, an effect thought to underlie their delayed impact on depression (Santarelli et al., 2003). The administration of a single dose of trazodone, however, transiently slows motor recovery in rats with cortical injury and reinstates the hemiparesis in recovered animals (Boyeson and Harmon, 1993). In contrast, desimpramine facilitates, whereas fluoxetine and amitriptyline had no effect on, motor recovery in experimental animal studies (Boyeson and Harmon, 1993; Boyeson et al., 1994). Another study found the administration of fluoxetine after traumatic brain injury in rats had no impact on memory, balance, or gait (Wilson and Hamm, 2002), and chronic fluoxetine negatively affected memory and had no effect on electrophysiological measures after dentate gyrus injury in rats (Keith et al., 2007). Thus, preclinical data supporting a direct effect of antidepressants on post-stroke recovery are inconsistent.
Diazepam, an indirect γ-amino butyric acid (GABA) agonist, has a permanent and severe detrimental impact on recovery of sensory function when given after anteromedial neocortex damage in the rat, an action associated with neuroanatomical changes in the thalamus and substantia nigra (Schallert et al., 1986; Jones and Schallert, 1992). Co-administration of a benzodiazepine antagonist blocks this detrimental effect (Hernandez et al., 1989). Because anxiolytics that do not act through the GABA/benzodiazepine receptor are neutral with respect to recovery, the harmful actions of benzodiazepines seem to be due to a specific, receptor-mediation action (Schallert et al., 1992).
The negative effects of diazepam and other benzodiazepines on post brain injury recovery raise concern for similar actions of other anticonvulsants. Phenobarbital delays recovery from somatosensory deficits after unilateral injury to the cerebral cortex in rats (Hernandez and Holling, 1994) and phenytoin administration worsens sensorimotor deficits after cortical injury (Brailowsky et al., 1986). In contrast, chronic administration of carbamazepine does not affect recovery (Schallert et al., 1992).
Diaschisis, remote metabolic depression in brain structures anatomically and functionally linked, but not adjacent to the area of primary injury, can be demonstrated in both animal models (Jaspers et al., 1990; Feeney, 1991; Theodore et al., 1992) and humans (Lenzi et al., 1982; Martin and Raichle, 1983; Fiorelli et al., 1991; Tanaka et al., 1992). Because d-amphetamine’s effect on recovery is evident within hours in some animal models, it was hypothesized that drugs that prolong or worsen diaschisis would be detrimental, whereas those that reverse diaschisis would be beneficial (Feeney, 1991). Resolution of crossed cerebellar diaschisis in humans, however, is not associated with recovery after hemispheric stroke-related hemiparesis (Infeld et al., 1995).
Long-term potentiation (LTP) refers to changes in synaptic efficiency following specific types of neurotransmitter exposure and is considered a physiological mechanism for learning and memory (Bliss and Collingridge, 1993). Because the initial effects of d-amphetamine on post-brain injury recovery occur within hours, and because of the need for concomitant training, induction of LTP is an attractive potential mechanism underlying the effects of at least certain classes of drugs on recovery. LTP is mediated through the N-methyl-d-aspartate (NMDA) receptor complex and leads to activation of both pre- and postsynaptic mechanisms, resulting in a long-lasting effect on synaptic strength. In addition to the hippocampus, LTP (and its correlate, long-term depression) occurs in a variety of brain regions, including the motor (Iriki et al., 1989; Keller et al., 1990) and visual cortex (Artola and Singer, 1989; Aroniadou and Teyler, 1991). In addition to amphetamine (Dunwiddie et al., 1982; Delanoy et al., 1983; Gold et al., 1984), neurotransmitters including norepinephrine (Dunwiddie et al., 1982; Gold et al., 1984; Stanton and Sarvey, 1985; Burgard et al., 1989; Dahl and Sarvey, 1989; Bröcher et al., 1992), serotonin (Kulla and Manahan-Vaughan, 2002), dopamine (Manahan-Vaughan and Kulla, 2003), GABA (Wigstrom and Gustafsson, 1985; Satoh et al., 1986; Olpe and Karlsson, 1990), and acetylcholine (Ito et al., 1988; Williams and Johnston, 1988; Burgard and Sarvey, 1990; Bröcher et al., 1992) can affect the induction of LTP. Consistent with this possible mechanism, the impact on recovery of a variety of drugs acting on central neurotransmitters can be predicted based on their effect on LTP (Goldstein, 1998, 2000, 2006). The effects of norepinephrine on synaptic function and plasticity, however, are complex and can vary in different brain regions (Marzo et al., 2009).
Disability and Dependence
There have been several clinical studies assessing the effects of amphetamine on post-stroke recovery in humans (Table 25.2). The studies were small and have important differences in methodologies. An initial ‘proof-of-concept’ study randomized eight subjects with stable post-stroke motor deficits to receive either 10 mg of d-amphetamine or placebo coupled with physical therapy within 10 days of ischaemic stroke (Crisostomo et al., 1988). As assessed 24 hours later, the d-amphetamine-treated group had a significant improvement in motor performance compared with the placebo-treated group. The study, however, involved only a small number of highly selected patients, and the clinical significance of the effect was uncertain.
|Study||N||Stroke-treatment interval||d-amphetamine dose / treatment frequency||Drug-therapy session interval (duration)||Outcome assessment|
|Crisostomo et al., 1988||8||<10 days||10 mg, one session||<3 hour (45 min)||1 day|
|Reding et al., 1995||21||>1 month||10 mg daily for 14 days, then 5 mg daily for 3 days||Same day (? Duration)||1 month|
|Walker-Batson et al., 1995||10||16–30 days||10 mg every 4 days for 10 sessions||“Peak of drug action” (? Duration)||1 week and 1 year|
|Sonde et al., 2001||39||5–10 days||10 mg twice weekly*||1 hour (30 min)||3 months|
|Martinsson et al., 2003||30||<96 hours||5 or 10 mg once or twice daily for 5 days||Same day (15 min vs 30–45 min)||3 months and 1 year|
|Treig et al., 2003||24||<6 weeks||10 mg every 4 days for 10 sessions||1 hour (45 min)||90 days and 1 year|
|Gladstone et al., 2006||71||5–10 days||10 mg twice weekly for 10 sessions||90 min (1 hour)||6 weeks and 3 months|
** Duration of physiotherapy varied (both groups received d-amphetamine)
A second double-blind, placebo-controlled trial included 5 d-amphetamine-treated and 5 placebo-treated patients with treatment given in conjunction with physical therapy once every 4 days for 10 sessions beginning 15 to 30 days after stroke (Walker-Batson et al., 2001). Amphetamine-treated patients had significantly greater improvements in motor scores compared with placebo-treated patients with a consolidation of the benefit after treatment was completed.
Other studies using different or similar trial designs have failed to confirm these observations. One randomized 24 subjects to receive 10 mg of amphetamine daily for 14 days, followed by 5 mg for 3 days or placebo (Reding et al., 1995). The subjects were enrolled more than 1 month after stroke and there was not a tight coupling between drug administration and physical therapy. Compared with the previous studies, there was a longer delay between stroke and treatment, and physical therapy was not temporally linked to drug exposure. Other negative studies evaluated dl– rather than d-amphetamine (Sonde et al., 2001), or used different dosing regimes or intervals between treatment sessions and treatment durations, or only included subjects with severe deficits (Martinsson et al., 2003). Two additional trials using a treatment and dosing regimen similar to the single positive study with longer-term follow-up also failed to find benefit (Treig et al., 2003; Gladstone et al., 2006).
Overall, the results of clinical trials evaluating the effect of d-amphetamine combined with physical therapy on post-stroke motor recovery have been disappointing and have not been consistent with the available extensive preclinical data. A meta-analysis including data from 3 small trials noted a trend towards more deaths at the end of follow-up among subjects randomized to amphetamine (Figure 25.1; OR 2.8, 95% confidence interval [CI]: 0.9–8.6), which may have been due to baseline imbalances between the groups (Martinsson et al., 2007). Including data from 9 trials (n = 114 amphetamine, n = 112 controls), the same meta-analysis found no overall effect on motor recovery (Figure 25.2) or activities of daily living (Figure 25.3; 4 trials, n = 58 amphetamine, n = 55 controls). As reflected above, it was concluded that ‘too few patients have been studied to draw any definite conclusions about the effects of amphetamine treatment on recovery from stroke’.
Figure 25.2 Meta-analysis of effects of post-stroke treatment with amphetamines on motor recovery.
Methylphenidate is used as a psychostimulant in apathetic patients to improve their participation in physiotherapy (Kaplitz, 1975; Johnson et al., 1992). The preclinical data reviewed above suggested benefit in improving post-brain injury motor recovery, but the dosing and treatment schedule is critical (Kline et al., 1994). Two clinical studies did not find any treatment-associated improvement in motor recovery with methylphenidate, although there were cardiovascular side effects (Larsson et al., 1988; Grade et al., 1998).
An uncontrolled study of L-DOPS combined with physiotherapy was conducted in a group of subjects with chronic, stroke-related motor deficits (Nishinoet al., 2001). The subjects’ average Fugl-Meyer motor score improved by 4.4 points (p < 0.001), and 10-minute walk time was shortened by 16% (p < 0.001) after 28 days of drug administration. Because the study was not controlled and physiotherapy alone can improve motor function even in the setting of established deficits (Wade et al., 1992; Green et al., 2002), the clinical significance of the observation is not certain.
Post-stroke depression is common, and antidepressants of various classes are used not only to treat the attendant psychiatric symptoms but also to improve the patient’s participation in rehabilitative interventions (El Husseini et al., 2012). The effects of two norepinephrine reuptake blockers (i.e. nortriptyline [Lipsey et al., 1984], maprotiline [Dam et al., 1996]) on post-stroke disability were neutral when given chronically. Dosing, however, may be critical. Brain norepinephrine content is reduced after chronic, but not acute, administration of desipramine (Roffler-Tarlov et al., 1973). Trazadone is an antidepressant that blocks α1-adrenergic receptors and interferes with motor recovery after cortex injury in rats (Boyeson and Harmon, 1993). A small clinical trial, however, found that chronic administration of trazadone to depressed stroke patients receiving physical therapy led to improvements in their activities of daily living (Reding et al., 1986). The discrepancy could be due to different pharmacological effects of one-time versus chronic administration on norepinephrine levels, related to its action on serotonin levels, or to its antidepressant effects.
As noted above, there are scant preclinical data showing an effect of serotonergic drugs on post-brain injury recovery. Despite this, a small clinical study suggested that the selective serotonin reuptake inhibitor (SSRI) fluoxetine might facilitate post-stroke recovery (Dam et al., 1996). This was followed by the Fluoxetine for Motor Recovery after Acute Ischemic Stroke (FLAME) trial that randomly assigned 118 subjects with stroke-related moderate to severe hemiparesis to fluoxetine 20 mg daily or placebo for 3 months beginning 5 to 10 days after symptom onset (Chollet et al., 2011). Improvement at 90 days was greater in the fluoxetine (adjusted mean 34.0 points [95% CI: 29.7–38.4]) than in the placebo group (24.3 points [19.9–28.7]; p = 0.003). There was also benefit as assessed by the modified Rankin score (mRS) at 90 days with 26% of those treated with fluoxetine compared with 9% of those who received placebo being independent (mRS 0–2, p = 0.015). The effect was independent of depression. The trial, however, included only a small number of subjects; whether the benefit is sustained over time is not known.
A meta-analysis of the effects of SSRIs on post-stroke disability included data from 13 trials of fluoxetine, 1 trial of sertraline, 3 trials of citalopram, and 5 trials of paroxetine (n = 1310) (Mead et al., 2013). Treatment with an SSRI was superior to controls (standard mean difference 0.92, 95% CI: 0.62–1.23; number needed to treat = 3) with no evidence of differences among the antidepressants (Figure 25.4). Many of the trials were small and there was significant heterogeneity among the studies. It was concluded that SSRIs might be associated with improved post-stroke recovery, but that much of the evidence is of poor quality, and larger, high-quality studies are needed.
The TALOS study (The Efficacy of Citalopram Treatment in Acute Stroke) was a Danish placebo-controlled, randomized, double-blind study in which 642 nondepressed patients with recent (<7 days) first-ever ischaemic stroke were randomized to either citalopram (n = 319) or placebo (n = 323) for 6 months as add-on to standard medical care. Improvement in functional recovery on the mRS from 1 to 6 months occurred in 160 (50%) patients on citalopram and 136 (42%) on placebo (odds ratio, 1.27; 95% CI: 0.92–1.74; p = 0.057). When dropouts before 31 days were excluded (n = 90), the analysis population showed an odds ratio of 1.37 (95% CI: 0.97–1.91; p = 0.07). It was concluded that early citalopram treatment did not improve functional recovery in nondepressed ischaemic stroke patients within the first 6 months, although a borderline, statistically significant effect was observed in the analysis population. The risk of cardiovascular events was similar between treatment groups, and citalopram treatment was well tolerated (Kraglund et al., 2018).
The UK-based Fluoxetine Or Control Under Supervision (FOCUS) trial randomized 3127 patients with a clinical diagnosis of recent (2–15 days previously) stroke to fluoxetine 20 mg once daily (n = 1564) or matching placebo (n = 1563) for 6 months. For the primary outcome measure, the mRs at 6 months, which was available in 1553 (99.3%) patients in each treatment group, random allocation to fluoxetine did not alter the distribution across mRs categories at 6 months compared with placebo (common OR, adjusted for minimization variables, 0.951; 95% CI: 0.839–1.079; p = 0.439).
Among secondary outcomes, allocation to fluoxetine did reduce the frequency of new depression by 6 months compared with placebo (13.0% fluoxetine vs 16.9% placebo, absolute difference 3.78%, 95% CI of difference: 1.26–6.30%; p = 0.0033),
However, allocation to fluoxetine also increased the frequency of bone fractures (2.9% fluoxetine vs 1.5% placebo, absolute difference 1.41%, 95% CI of diff: 0.38–2.43; p = 0.0070).
There were no significant differences in other events or outcomes at 6 or 12 months.
These results suggest that fluoxetine 20 mg given daily for 6 months after acute stroke does not improve functional outcome, but it likely reduces the occurrence of depression (although a secondary outcome, this is a well-recognized effect of fluoxetine), and also likely increases bone fractures (noted in some observational studies, but again, a secondary outcome) (FOCUS Trial Collaboration, 2019).