Post Stroke Use of Selective Serotonin Reuptake Inhibitors and Clinical Outcome Among Patients With Ischemic Stroke
A Nationwide Propensity Score–matched Follow-up Study
Background and Purpose—Selective serotonin reuptake inhibitors (SSRIs) are widely prescribed after stroke. We aimed to investigate whether potential antiplatelet or vasospastic effects have important clinical implications.
Methods—Using data from Danish medical registries, we did a nationwide follow-up study among ischemic stroke patients between 2003 and 2009. We identified 5833 SSRI users, and propensity score matched these patients with nonusers in a 1:1 ratio, followed by Cox regression analysis to compute hazard ratios (HRs) of acute myocardial infarction, recurrent stroke, major bleeding, and death.
Results—Median follow-up time (from 30 days after discharge to death/end of follow-up) was 1159 days. In total, 2.9% had myocardial infarction, 8.1% recurrent ischemic stroke, 20.2% major bleeding, 1.4% intracranial bleeding, and 34.4% died during follow-up. SSRI users had a lower risk of the combined outcome of myocardial infarction or recurrent ischemic stroke (adjusted HR, 0.77; confidence interval [CI], 0.62–0.96). However, the SSRI users also experienced a higher risk of overall major bleeding (adjusted HR, 1.33; CI, 1.14–1.55) and a nonsignificantly higher risk of intracranial bleedings (adjusted HR, 1.14; CI, 0.62–2.12). Mortality increased in SSRI users (adjusted HR, 1.13; CI, 1.00–1.28) and death caused by bleeding increased (adjusted HR, 1.89; CI, 0.97–3.66) as compared with death by other causes (adjusted HR, 1.11; CI; 0.98–1.26).
Conclusions—SSRI use after ischemic stroke was associated with a lower risk of new cardiovascular events and also with an increased bleeding risk. There was an increased mortality among SSRI users, which may be related to the increased bleeding risk.
Antidepressants, in particular selective serotonin reuptake inhibitors (SSRIs), are widely used for treating post stroke depression (PSD) and pathological crying. PSD is common, affecting ≈40% of stroke patients within the first year, whereas pathological crying is seen in ≈15% of the patients within the first year post stroke. Approximately 80% of the PSD cases are diagnosed within the first 6 months post stroke.1–3 The use of SSRI for treating PSD results in significantly less depressive symptoms after 3 weeks, whereas the treatment of pathological crying has an immediate and long-term effect.4,5 The therapy is, in general, considered to be well tolerated regardless of high age, comorbidities, comedications, or potential drug interactions. Recent studies also suggest that SSRI treatment enhances cognitive recovery6 and motor recovery when used in the subacute phase post stroke.7,8
However, epidemiological studies have shown an increased gastrointestinal bleeding risk in SSRI users, in general.9,10 The increased bleeding risk may be explained by the SSRI depletion of platelet serotonin, resulting in inhibition of aggregation, although the exact mechanism is unknown.11 In line with a proved bleeding risk it has been suggested that SSRI treatment in depressed patients with acute myocardial infarction (MI) may reduce the risk of subsequent cardiovascular morbidity and mortality,12 possibly also a result from inhibition of platelet aggregation, although other explanations are possible.12,13 In stroke patients the situation is less clear because serotonin also regulates cerebral vasomotor tonus and may cause vasoconstriction14,15 thereby counteracting a mild antithrombotic effect. An increased risk of stroke in association with serotonin transporter inhibitor treatment has thus been suggested,16 whereas post stroke SSRI treatment has been associated with longer survival.17
The effects of post stroke SSRI use regarding the possible clinical consequences of the antiplatelet or vasospastic effects remain uncertain, and population-based data reflecting routine clinical practice among unselected patients are sparse. We, therefore, examined the association between post stroke SSRI use and clinical outcomes, including hospitalization with acute MI, recurrent stroke, bleeding, and mortality in ischemic stroke patients in a nationwide, propensity score–matched follow-up study.
The study was conducted within the entire Danish population (≈5.5 million). The Danish National Health Service provides tax-supported health care to all Danish residents, including free access to general practitioners and hospitals and refunds a variable proportion of the prescription medication costs. The Danish Civil Registration System keeps electronic records on sex, date of birth, change of address, date of emigration, and changes in vital status since 1968.18 The records carry a unique 10-digit civil registration number, assigned to every Danish citizen and used in all Danish registries, enabling unambiguous individual-level linkage between them.
The study was approved by the Danish Data Protection Agency. According to Danish law, ethical approval is not needed for registry-based studies.
We identified all patients (≥18 years) with ischemic stroke discharged from Danish hospitals between January 13, 2003 and December 31, 2009 (n=64 118). Patients (592) who were not Danish residents for at least 1 year before admission or did not have a valid social security number and 16 909 patients registered as recurrent stroke were excluded. 3125 patients died, and 1038 who had MI, recurrent stroke, or hemorrhage within 30 days after discharge were excluded. Finally, 6279 patients who redeemed a prescription for antidepressants within 1 year before admission were excluded. Of the remaining (n=36 175), 5837 were SSRI users. 5833 could be matched in a 1:1 ratio using propensity score matching (Figure 1). The patients were identified in the Danish Stroke Registry. The registry is a nationwide initiative to monitor and improve the quality of early stroke care. Participation in the project is mandatory for all hospital departments in Denmark treating patients with acute stroke.19 All patients admitted to Danish hospitals with acute stroke according to the WHO criteria (ie, rapidly developed clinical signs of focal or global disturbance of cerebral function, lasting >24 hours or until death, with no apparent nonvascular cause) are eligible for inclusion in the Danish Stroke Registry. All patients have brain computed tomography or magnetic resonance imaging scans in the acute stroke phase. Patients with transient ischemic attack (symptoms lasting <24 hours), subdural hematoma, epidural or subarachnoid hemorrhage, retinal infarct, and infarct caused by trauma, infection, surgery, or an intracerebral malignant process are not included.
Data on all SSRI prescriptions filled before and after hospital discharge were obtained on each ischemic stroke patient by linkage with the Danish Medicines Agency’s Medical Register. The register contains data from 1995 onwards on all prescription drugs dispensed at all Danish pharmacies, including patients’ civil registration numbers, type of drug according to the Anatomical Therapeutic Chemical Classification System, and date of dispensing the drug. In Denmark, SSRIs are available by prescription only. In our study, antidepressants are categorized into 3 groups according to their action on serotonin and norepinephrine reuptake mechanisms.20 SSRIs, non-SSRIs, and other antidepressants (antidepressants with an inhibitory effect on both serotonin and norepinephrine reuptake, and antidepressants with either selective norepinephrine reuptake inhibition or no effect on any reuptake; Appendix I in online-only Data Supplement).
Patients were followed from the date of discharge to the date of the individual endpoints, emigration, or end of study period, whichever came first. Median follow-up time (from 30 days after discharge to death/end of follow-up) was 1159 days with a range from 1 to 2875 days and an interquartile range of 582 to 1839 days. Patients who died within the first 30 days after stroke discharge were excluded because patients with a presumed poor prognosis at the time of discharge were less likely to be prescribed long-term treatment with SSRIs and other prescription drugs. Seven endpoints were assessed: hospitalization with MI, recurrent ischemic stroke, major bleeding, intracranial bleeding, all-cause mortality, mortality caused by bleeding, and mortality by other cause. Major bleeding included intracranial bleeding and other bleedings (gastrointestinal bleeding and other bleedings requiring blood transfusion). Information on the vital status of the patients during follow-up was obtained by linkage with the Danish Civil Registration System. Data on hospitalizations with MI or major bleeding were obtained from The Danish National Patient Register that holds information on all discharges from Danish nonpsychiatric hospitals since 1977, including date and diagnosis of discharge, all encoded according to the International Classification of Diseases (8th revision until 1993 and 10th revision thereafter)21 (Appendix I in online-only Data Supplement). Data on recurrent ischemic stroke were obtained from the Danish Stroke Registry. Data on use of transfusions were obtained from the Danish Transfusion Database, which is a national clinical registry monitoring the use of blood components. The database, which holds data from 1999 onwards and has complete nationwide coverage from 2005 and onward, retains data on all types of blood transfusions administered during the study period, including data on dates, types and number of blood components administered to the patient.
At the time of hospital admission, data were collected on the following characteristics: sex, living arrangements (living with partner, family or friend, and living alone), type of residence (own home, nursing home, or other form of institution), Scandinavian Stroke Scale Score,22 history of stroke and MI, previous and current atrial fibrillation, hypertension, diabetes mellitus or intermittent claudication, former coronary revascularization, smoking habits (smoker, ex-smoker, and never), and alcohol intake (≤14/21, >14/21 drinks per week for women and men, respectively) (Appendix I in online-only Data Supplement). The categories of alcohol intake reflected recommendations in the study period on maximum alcohol intake issued by the National Board of Health in Denmark.
We also computed the Charlson comorbidity index score23 for each patient based on all discharge diagnoses recorded before the stroke hospitalization. Data on previous hospitalizations were obtained from The National Registry of Patients that contains data on all discharges from all nonpsychiatric hospitals in Denmark since 1977. The Charlson comorbidity index covers 19 major disease categories, including congestive heart failure, renal disease, and cancer, weighted according to their prognostic impact on patient survival. The index has previously been adapted for use with hospital discharge registry data and has been reported to be useful also among patients with stroke. We defined 3 levels of comorbidity for each patient, based on their complete hospital discharge history, as follows: 0 comorbidities (low), 1 to 2 comorbidities (moderate) and >2 comorbidities (high). Scandinavian Stroke Scale was used to assess admission stroke severity. The scale is a validated and widely used neurological stroke scale in Scandinavia with a total score that ranges from 0 (no function) to 58 (intact neurological function).
Information on drug use before and after the admission with stroke was also obtained, including use of non-SSRIs, other antidepressants, blood pressure–lowering drugs (including angiotensin-converting enzyme inhibitors, β-blockers, calcium antagonists, diuretics, and angiotensin II receptor antagonists), platelet inhibitors (low-dose aspirin, dipyridamole, and clopidogrel), Vitamin K antagonist, statins, and hormone replacement therapy (Appendix I in online-only Data Supplement). All drugs are available on prescription only, except for low-dose aspirin; however, low-dose aspirin is generally prescribed by physicians, because chronic users and pensioners receive 50% reimbursement through the national health insurance program. Finally, information on the quality of in-hospital care was obtained from the Danish Stroke Registry and included data on early admission to a specialized stroke unit, early administration of antiplatelet or anticoagulant therapy, early examination with computed tomography or magnetic resonance imaging, early assessment by a physiotherapist and an occupational therapist, and assessment of nutritional risk. We computed the percentage of relevant processes of care received for each patient as a measure of the quality of in-hospital stroke care (0%–25%, 26%–50%, 51%–75%, 76%–99%, and 100%, respectively).
Because post stroke SSRI use was not randomly assigned in the study population, we used propensity score matching (greedy method) to reduce the risk of bias owing to confounding.24,25 Each SSRI user (defined as having redeemed at least 1 prescription within 180 days after hospital discharge) was matched to 1 nonuser with similar propensity score. The propensity score was based on the following covariates: sex, age, atrial fibrillation, hypertension, former revascularization, former use of SSRI, non-SSRI, other antidepressants, blood pressure–lowering drugs, platelet inhibitors, Vitamin K antagonist, statins, hormone replacement therapy, Scandinavian Stroke Scale Charlson comorbidity index, living arrangements, housing, alcohol intake, smoking, and quality of in-hospital stroke care. An absolute standardized difference <10% and a variance ratio between 0.8 and 1.25 were considered to support the assumption of balance between the groups26,27 (Figures 2 and 3; • for the entire study population [36 175 patients] and ° for the 11 666 propensity score–matched patients).
The matching was followed by Cox regression analysis, which enabled us to adjust for potential residual confounding. In the Cox regression analyses, we adjusted for use of other drugs during follow-up, including non-SSRIs, other antidepressants, blood pressure–lowering drugs, platelet inhibitors, Vitamin K antagonist, and statins. All drugs were included as time-dependent variables with an assumed average prescription length of 90 days.
All data analyses were performed using SAS 9.2, figures were generated using R (x64 version 2.12.1).
Table 1 displays patient characteristics of the included 5833 SSRI users and 5833-matched SSRI nonusers. The imbalances in the covariates were largely removed by the propensity score matching (Figures 2 and 3). During the follow-up study period (median follow-up time from discharge plus 30 days to death/end of follow-up was 1159 days), we recorded, in the entire study population, 337 MIs (2.9%), 940 recurrent strokes (8.1%), 2357 major bleedings (20.2%) of which 167 were intracranial (1.4%) and 4010 deaths (34.4%). Table 2 shows crude and adjusted risk ratios for adverse outcomes among the 2 groups. Patients who were treated with SSRI had a lower risk of MI and recurrent stroke with unadjusted hazard ratios (HRs) of 0.67 (95% confidence interval [CI], 0.44–1.02) and 0.86 (95% CI, 0.67–1.09), respectively, although it did not reach statistical significance. Adjustment for use of other drugs during follow-up, including other antidepressant drugs, left the adjusted HRs virtually unchanged (ie, 0.65 [95% CI, 0.42–1.00] and 0.87 [95% CI, 0.68–1.11]). For the combined outcome (MI or recurrent stroke) the adjusted HR was 0.77 (95% CI, 0.62–0.96). SSRI treatment was associated with an increased bleeding risk because SSRI users had an adjusted HR of major bleeding of 1.33 (95% CI, 1.14–1.55) and an adjusted HR of intracranial bleeding of 1.14 (95% CI, 0.62–2.12), the latter not reaching statistical significance. SSRI treatment was also associated with increased mortality with an adjusted HR of 1.13 (95% CI, 1.00–1.28). SSRI users had an adjusted HR of death caused by bleeding of 1.89 (95% CI, 0.97–3.66) and an adjusted HR of death by other causes of 1.11 (95% CI, 0.98–1.26). In comparison, we observed that the adjusted HRs of death in the multivariable analyses for use of tri- and tetracyclic antidepressants were 1.88 (95% CI, 1.17–3.03) and 1.87 (95% CI, 1.46–2.40), respectively.
SSRI treatment post stroke was in this large propensity score–matched population-based follow-up study associated with a lower risk of MI and recurrent ischemic stroke and also associated with a higher risk of major bleedings and death.
Our findings among ischemic stroke patients are in line with a former study among MI patients, showing a significant lower risk of new MI among users of SSRIs.12 However, stroke pathophysiology may differ from the pathophysiology of ischemic heart disease, and cerebral vasospasms caused by SSRI treatment could, at least in theory, counteract a potential antithrombotic effect in stroke patients and explain the slight difference in magnitude of protection against MI and recurrent strokes. Data on vasospasms were not available in our study population, but this theoretical risk seemed, in any case, to be outweighed by the lower risk of recurrent strokes among SSRI users. This apparently positive balance between possible beneficial and adverse effects in patients treated with SSRI is encouraging.
Intracranial bleeding always seems to be a drawback of antithrombotic treatment. It seems that SSRI treatment might have the same tendency. The absolute risk of intracranial bleeding was low, as only 1.4% in the whole population experienced an intracranial bleeding during follow-up. However, the rate of overall major bleeding was substantially higher and seemed to contribute to the increase in mortality among SSRI-treated patients.
It is well known that PSD is associated with higher mortality28, and antidepressant treatment has, to our knowledge, not been shown to eliminate this risk. Assuming that the majority of patients in our study received SSRI owing to depression, a higher mortality in this group was to be expected.
SSRI treatment has been linked with a range of physiological mechanisms that could play a role in the outcomes assessed in our study, including neuronal remodeling, serotonin receptor changes, and effect on platelet function.11,29,30 The apparent protective effect of SSRI against ischemic events was independent of concomitant aspirin treatment or other antithrombotic treatment. The fact that we found both a lower risk of ischemic events and an increase in bleeding events in this population points to an effect on platelet inhibition owing to lowering of platelet serotonin by SSRI treatment.
The strengths of our study included the prospective, population-based design with complete long-term follow-up. Furthermore, our analyses were based on a large cohort, with detailed information on all individuals that reduced the risk of confounding and chance findings. In Denmark, stroke treatment, in general, is well organized in the acute setting, where approximately >90% of all stroke patients are examined according to key indicators as described in the Danish Stroke Registry.31
In this study, we wished to study the effects of SSRI treatment initiated after stroke. To remove potential effects of prestroke SSRI use on the study outcomes, prestroke antidepressant users were excluded. As ≈80% of the PSD cases are diagnosed within the first 6 months post stroke, the 180-day limit for SSRI use identification was chosen. This was done to reduce the possibility that the treatment was initiated on other indications.
We used prescription data as an indicator of drug use, but we had no information on actual patient compliance (whether the patient used all of the prescribed drugs). This lack of information could potentially introduce a misclassification into the study. However, the data are likely to provide a good reflection of actual drug use since we only included data on prescriptions that had been filled and the patients’ medical expenses were only partly reimbursed, making them more likely to have taken the prescribed drugs.
Although we had detailed information about the patients and used a well-balanced propensity score–matched design, we cannot exclude the possibility that our results remain influenced by confounding factors because of the observational nature of the study design. The uncontrolled confounding by indication owing to the underlying depression among the majority of the SSRI users is a particular concern in this context and it seems likely that this may have contributed to the increased mortality associated with SSRI use and it may have led us to underestimate the protective vascular effect. On the contrary, misdiagnosis may also have occurred resulting in untreated depressed patients in the control group, which could have drawn the comparisons in a favorable direction for the SSRI-treated group. To overcome some of the weaknesses in the present study, further research in the form of a prospective, randomized controlled trial is needed. We believe that such a trial should be designed to study the effect of SSRI in stroke patients regardless of depressive symptoms because nondepressed patients could potentially benefit from the treatment.
In conclusion, SSRI use among patients with ischemic stroke was associated with a lower risk of new ischemic events, an increased overall bleeding risk, and a nonsignificantly increased risk of intracranial bleeding. SSRI users were found to have an increased mortality, which may reflect a combination of uncontrolled confounding by indication owing to the underlying depression and an increased bleeding risk. Our data suggest that SSRI use may have important implications for the clinical outcomes of patients with ischemic stroke; however, additional efforts are warranted to ensure optimal patient selection and monitoring of the treatment.
Source of Funding
The study was funded by a grant from the Tryg Foundation (TrygFonden, Denmark). There has been no interference with the study design, data collection, analysis, or publication from the funding source.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.112.674242/-/DC1.
- Received August 17, 2012.
- Revision received October 29, 2012.
- Accepted November 19, 2012.
- © 2013 American Heart Association, Inc.
- Andersen G
- Hackett ML,
- Yang M,
- Anderson CS,
- Horrocks JA,
- House A
- Hackett ML,
- Anderson CS,
- House AO,
- Halteh C
- Ducros A,
- Boukobza M,
- Porcher R,
- Sarov M,
- Valade D,
- Bousser MG
- Bak S,
- Tsiropoulos I,
- Kjaersgaard JO,
- Andersen M,
- Mellerup E,
- Hallas J,
- et al
- Mainz J,
- Krog BR,
- Bjørnshave B,
- Bartels P
- 22.↵Scandinavian Stroke Study Group. Multicenter trial of hemodilution in ischemic stroke--background and study protocol. Stroke.1985; 16:885–890.
- 31.↵Kompetencecenter for Epidemiologi og Biostatistik (Nord). Dansk Apopleksiregister: Årsrapport 2011 [Danish Stroke Registry: Annual Report 2011]. Endelig version ed. Aarhus: Kompetencecenter for Klinisk Kvalitet og Sundhedsinformatik (Vest); 2012.