← Back

EFFECTS

Effects of Fluoxetine on Outcomes at 12 Months After Acute Stroke: Results From EFFECTS, a Randomized Controlled Trial

Share Share Copied! Compare

Year of Publication: 2021

Authors: Lundström E, Isaksson E, Greilert Norin N, ..., Sunnerhagen KS; on behalf of the EFFECTS Writing Committee

Journal: Stroke

Citation: Lundström E, et al. Stroke. 2021;52:3082–3087. DOI: 10.1161/STROKEAHA.121.034705

Link: https://www.ahajournals.org/doi/full/10....EAHA.121.034705

PDF: https://www.ahajournals.org/doi/reader/1...EAHA.121.034705

Clinical Question

Does fluoxetine 20 mg daily for 6 months after acute stroke improve functional outcomes or other secondary outcomes at 12 months compared to placebo?

Study Overview

Objective

To determine whether any functional or secondary benefits of fluoxetine seen at 6 months after acute stroke persist, delay, or emerge at 12 months.

Study Summary

Fluoxetine 20 mg daily for 6 months did not improve functional outcome at 12 months after acute stroke (adjusted common OR 0.92, 95% CI 0.76–1.10).
Fluoxetine arm scored significantly worse on memory (median 89 vs 93, P=0.0021) and communication (median 93 vs 96, P=0.024) domains of the Stroke Impact Scale — authors attribute this to chance.
Antidepressant use at 12 months was similar in both groups (9.0% fluoxetine vs 8.8% placebo); the ~4% depression benefit seen at 6 months was not sustained.

Intervention

Fluoxetine 20 mg orally once daily for 6 months vs. matching placebo, with follow-up continued to 12 months. Randomization 2–15 days after acute stroke.

Patients per Arm

Fluoxetine: 750; Placebo: 750

Bottom Line

Fluoxetine 20 mg daily for 6 months after acute stroke did not improve functional outcome at 12 months (adjusted common OR 0.92, 95% CI 0.76–1.10). Patients on fluoxetine scored significantly worse on the memory and communication domains of the Stroke Impact Scale, though this is likely a chance finding given multiple comparisons. The depression benefit seen at 6 months was not sustained at 12 months (antidepressant use 9.0% vs 8.8%).

        Major Points
        EFFECTS was an investigator-led, randomized, double-blind, placebo-controlled parallel-group trial in 35 Swedish hospital stroke units enrolling 1500 patients from October 2014 to June 2019; last 12-month follow-up was July 8, 2020.
Patients were randomized 1:1 to fluoxetine 20 mg orally once daily or matching placebo for 6 months, then followed for an additional 6 months without treatment — this paper reports the predefined 12-month secondary analysis.
Stroke types: 87.4% ischemic stroke, 12.3% intracerebral hemorrhage, 0.2% non-stroke; mean age 71 years (SD 11); 38.3% women; 96.3% previously independent; median NIHSS 3.
Primary outcome (mRS distribution at 12 months): adjusted common OR 0.92 (95% CI 0.76–1.10) — no significant difference. Unadjusted common OR 0.96 (95% CI 0.80–1.15). Both favor placebo numerically but are not significant.
mRS 0–1 at 12 months: fluoxetine 53% (160+224/715) vs placebo 53% (178+201/712); mRS 6 (dead at 12 months): 34 (5%) each group.
Fluoxetine was significantly worse on memory domain of Stroke Impact Scale v3 at 12 months: median 89 (IQR 75–100) vs 93 (IQR 82–100), P=0.0021; and communication domain: median 93 (IQR 82–100) vs 96 (IQR 86–100), P=0.024 — authors attribute these to chance due to multiple comparisons. No significant differences in these domains were replicated in FOCUS or AFFINITY.
At 6 months, fluoxetine reduced new depression (antidepressant use: 4.8% fluoxetine vs 6.8% placebo); by 12 months this benefit had disappeared: 9.0% (65/725) fluoxetine vs 8.8% (64/728) placebo (diff 0.2%, 95% CI −3.1 to 3.8%). Of those on antidepressants at 6 months, 41% fluoxetine vs 59% placebo continued at 12 months.
All other secondary outcomes at 12 months were non-significant: SIS strength (P=0.88), hand ability (P=0.75), mobility (P=0.84), motor composite (P=0.88), daily activities (P=0.97), physical function composite (P=0.98), mood/emotional control (P=0.63), participation (P=0.79), recovery VAS (P=0.58), fatigue/vitality (P=0.22), MHI-5 (P=0.57), EQ-5D-5L (P=0.86).
Compliance was high: 89% (1338/1500) took trial medication for ≥150 days; median treatment duration 180 days (IQR 180–180) in both groups. 12-month mRS data available in 715 (95%) fluoxetine and 712 (95%) placebo patients. Lost to follow-up: 4.1% (62/1500).
EFFECTS (n=1500, Sweden), FOCUS (n=3124, UK), and AFFINITY (n=1260, Australia/NZ/Vietnam) all showed identical null results for fluoxetine on functional outcome. A prospective individual patient data meta-analysis of all three trials is planned.
Adverse events and safety data (fractures, hyponatremia) were only collected through 6 months; not re-assessed at 12 months except for mortality. Long-term follow-up via Swedish central registries is planned to ≥3 years.

      

Design

Study Type: Investigator-led, randomized, placebo-controlled, double-blind, parallel-group trial

Randomization: 1

Blinding: Double-blind (participant, care provider, investigator, and outcomes assessor all masked until 12-month assessment)

Randomization Method: Secure, centralized, web-based minimization algorithm; 1:1 allocation. Minimization covariates: days between stroke and randomization; probability of being alive and independent at 6 months; motor deficit; aphasia.

Enrollment Period: October 20, 2014 – June 28, 2019

Follow-up Duration: 12 months total (6 months on treatment + 6 months post-treatment observation)

Centers: 35

Countries: Sweden

Sample Size: 1500

Analysis: Intention-to-treat. Primary: ordinal logistic regression (adjusted and unadjusted common OR) for mRS. Secondary: Mann-Whitney U test (unadjusted). Software: SAS for Windows, version 9.4. Statistical analysis plan pre-published before unmasked data review.

Inclusion Criteria

Adults aged ≥18 years
Clinical diagnosis of acute stroke within the previous 2 to 15 days
Brain imaging (CT or MRI) consistent with ischemic or hemorrhagic stroke
Persisting neurological deficit at time of randomization

Exclusion Criteria

Pre-existing depression or currently taking antidepressant medication
Contraindication to fluoxetine
Unlikely to be available for follow-up during the subsequent 12 months
Another life-threatening illness making 12-month survival unlikely
Enrolled in another clinical trial of an investigational medicinal product or device
Pregnant, breast-feeding, or women of childbearing age not using contraception

Baseline Characteristics

Overall:

Mean Age (SD): 71 (11)
Female %: 38.3
Previously Independent %: 96.3
Median NIHSS: 3
Ischemic Stroke %: 87.4
Intracerebral Hemorrhage %: 12.3
Non-stroke %: 0.2

Note: Full per-arm baseline characteristics available in Data Supplement Tables I–III of the primary 2020 Lancet Neurology publication (Lundström et al. Lancet Neurol. 2020;19:661–669). Baseline was well balanced between groups per minimization algorithm.

Arms

Field	Fluoxetine	Control
Intervention	Fluoxetine 20 mg capsule orally once daily for 6 months, starting 2–15 days after acute stroke onset. Capsules were visually identical to placebo even when broken open.	Matching placebo capsule orally once daily for 6 months, starting 2–15 days after acute stroke onset. Visually identical to fluoxetine capsule even when broken open.
N Randomized	750	750
N Received Treatment	748	748
N with 12-month mRS	715	712
Duration	6 months	6 months

Outcomes

Outcome	Type	Intervention	P-value
Distribution of modified Rankin Scale (mRS) scores at 12 months (ordinal shift analysis)	Primary		Not significant
Stroke Impact Scale v3 — Strength domain \| Fluoxetine n: 662; Fluoxetine Median (IQR): 75 (50–94); Placebo n: 658; Placebo Median (IQR): 75 (50–94)	Secondary		0.88
Stroke Impact Scale v3 — Hand ability domain \| Fluoxetine n: 664; Fluoxetine Median (IQR): 81 (56–100); Placebo n: 661; Placebo Median (IQR): 81 (50–100)	Secondary		0.75
Stroke Impact Scale v3 — Mobility domain \| Fluoxetine n: 665; Fluoxetine Median (IQR): 89 (69–100); Placebo n: 662; Placebo Median (IQR): 89 (72–100)	Secondary		0.84
Stroke Impact Scale v3 — Motor composite (strength + hand + mobility) \| Fluoxetine n: 665; Fluoxetine Median (IQR): 79 (60–94); Placebo n: 661; Placebo Median (IQR): 80 (58–94)	Secondary		0.88
Stroke Impact Scale v3 — Daily Activities domain \| Fluoxetine n: 667; Fluoxetine Median (IQR): 90 (70–98); Placebo n: 662; Placebo Median (IQR): 90 (68–98)	Secondary		0.97
Stroke Impact Scale v3 — Physical function composite (strength + hand + mobility + daily activities) \| Fluoxetine n: 667; Fluoxetine Median (IQR): 76 (57–91); Placebo n: 662; Placebo Median (IQR): 77 (55–92)	Secondary		0.98
Stroke Impact Scale v3 — Memory and thinking domain \| Fluoxetine n: 666; Fluoxetine Median (IQR): 89 (75–100); Placebo n: 662; Placebo Median (IQR): 93 (82–100); Note: Significantly worse in fluoxetine group; authors consider this a chance finding	Secondary		0.0021
Stroke Impact Scale v3 — Communication and understanding domain \| Fluoxetine n: 664; Fluoxetine Median (IQR): 93 (82–100); Placebo n: 661; Placebo Median (IQR): 96 (86–100); Note: Significantly worse in fluoxetine group; authors consider this a chance finding; not replicated in FOCUS or AFFINITY	Secondary		0.024
Stroke Impact Scale v3 — Mood and emotional control domain \| Fluoxetine n: 665; Fluoxetine Median (IQR): 78 (64–89); Placebo n: 658; Placebo Median (IQR): 78 (64–89)	Secondary		0.63
Stroke Impact Scale v3 — Participation domain \| Fluoxetine n: 663; Fluoxetine Median (IQR): 69 (46–91); Placebo n: 656; Placebo Median (IQR): 71 (47–93)	Secondary		0.79
Stroke Impact Scale v3 — Recovery (VAS 0–100) \| Fluoxetine n: 666; Fluoxetine Median (IQR): 75 (50–90); Placebo n: 662; Placebo Median (IQR): 80 (50–90)	Secondary		0.58
Fatigue (SF-36 Vitality subscale, score 0–100; <50 = fatigue) \| Fluoxetine n: 662; Fluoxetine Median (IQR): 56 (44–75); Placebo n: 658; Placebo Median (IQR): 56 (44–75)	Secondary		0.22
Mental Health Inventory-5 (MHI-5, 0–100; <60 = moderate-to-poor mental health) \| Fluoxetine n: 667; Fluoxetine Median (IQR): 76 (60–88); Placebo n: 660; Placebo Median (IQR): 76 (60–88)	Secondary		0.57
Health-related quality of life (EQ-5D-5L index, UK cross-walk; 0=worst, 1=best) \| Fluoxetine n: 664; Fluoxetine Median (IQR): 0.74 (0.55–0.84); Placebo n: 658; Placebo Median (IQR): 0.74 (0.55–0.88)	Secondary		0.86
Depression proxy (open-label antidepressant use) at 12 months \| Fluoxetine: 9.0% (65/725); Placebo: 8.8% (64/728); Difference: 0.2% (95% CI −3.1 to 3.8%); Note: At 6 months, antidepressant use was 4.8% fluoxetine vs 6.8% placebo — the ~2% benefit seen at 6 months was no longer present at 12 months. Of those on antidepressants at 6 months (n=87), 60% (51/87) continued at 12 months: 41% (21/36) of fluoxetine group vs 59% (30/51) of placebo group.	Secondary		Not significant
Mortality within 12 months	Adverse	Fluoxetine: 34/750 (4.5%); Placebo: 34/750 (4.5%); Note: No difference between groups
Fractures (collected only to 6 months in primary paper)	Adverse	Fluoxetine (6 months): 2.9%; Placebo (6 months): 1.5%; P-value at 6 months: 0.047; Note: Adverse event data (fractures, hyponatremia) not collected between 6 and 12 months in this paper
Hyponatremia (collected only to 6 months in primary paper)	Adverse	Fluoxetine (6 months): 2.2%; Placebo (6 months): 0.7%; Note: Adverse event data not collected between 6 and 12 months in this paper
Lost to follow-up at 12 months	Adverse	62/1500 (4.1%)

Subgroup Analysis

No formal subgroup analyses for the 12-month outcomes are reported in this paper. Baseline covariates used in the minimization algorithm (days stroke-to-randomization, probability alive and independent at 6 months, motor deficit, aphasia) were included as co-variables in the adjusted ordinal logistic regression. No treatment-by-subgroup interactions were identified. Subgroup analyses for the primary 6-month outcome are reported in the original Lancet Neurology 2020 paper.

Criticisms

Adverse event data (fractures, hyponatremia) not collected between 6 and 12 months — the 6-month safety signals cannot be assessed for continuation or resolution after drug cessation.
Depression at 12 months defined only as antidepressant use (ATC code N06A), which is a crude proxy — likely underestimates true prevalence; no face-to-face psychiatric assessment performed at 12 months as was done at 6 months.
The statistically significant worse memory and communication scores in the fluoxetine group are not supported by FOCUS or AFFINITY and are interpreted as chance findings from multiple comparisons — no correction for multiplicity was applied to secondary outcomes.
Excluded patients with pre-existing depression or antidepressant use — limits generalizability to a population with lower baseline depression risk; pooled observational data suggest post-stroke depression rates of 31–33% at 6–12 months vs only ~7–11% in EFFECTS.
Rehabilitation intensity between 6 and 12 months was not monitored or controlled — any differential rehabilitation could confound functional outcome comparisons.
Only 12.3% hemorrhagic stroke — primarily an ischemic stroke population; results may not generalize to hemorrhagic stroke.
Single country (Sweden) with high-income healthcare structure — limits generalizability to lower-income settings where stroke burden is highest.
This paper reports secondary predefined analyses of the EFFECTS trial; the primary 6-month outcomes were published separately in Lancet Neurology 2020.

Funding

Swedish Research Council (grant 2014-07072); Swedish Heart-Lung Foundation (application 2013-0496 and 2016-0245); Swedish Brain Foundation (application FO2017-0115); Swedish Society of Medicine (ID 692921); King Gustav V and Queen Victoria's Foundation of Freemasons (2014); Swedish Stroke Association (2012 and 2013). Sponsor: Karolinska Institutet, Department of Clinical Sciences, Danderyd Hospital, 182 88 Stockholm, Sweden. Funders had no role in design, data collection, analysis, interpretation, or writing.

Based on: EFFECTS (Stroke, 2021)