Impact of Prestroke Selective Serotonin Reuptake Inhibitor Treatment on Stroke Severity and Mortality
Background and Purpose—Selective serotonin reuptake inhibitors (SSRIs) have been associated with an increased risk of bleeding but also a possible neuroprotective effect in stroke. We aimed to examine the implications of prestroke SSRI use in hemorrhagic and ischemic stroke.
Methods—We conducted a registry-based propensity score–matched follow-up study among first-ever patients with hemorrhage and ischemic stroke in Denmark (2003–2012). Multiple conditional logistic regression was used to compute adjusted odds ratios of severe stroke and death within 30 days.
Results—Among 1252 hemorrhagic strokes (626 prestroke SSRI users and 626 propensity score–matched nonusers), prestroke SSRI use was associated with an increased risk of the strokes being severe (adjusted propensity score–matched odds ratios, 1.41; confidence interval, 1.08–1.84) and an increased risk of death within 30 days (adjusted propensity score–matched odds ratios, 1.60; confidence interval, 1.17–2.18). Among 8956 patients with ischemic stroke (4478 prestroke SSRI users and 4478 propensity score–matched nonusers), prestroke SSRI use was not associated with the risk of severe stroke or death within 30 days.
Conclusions—Prestroke SSRI use is associated with increased stroke severity and mortality in patients with hemorrhagic stroke. Although prestroke depression in itself may increase stroke severity and mortality, this was not found in SSRI users with ischemic stroke.
Selective serotonin reuptake inhibitors (SSRIs) are associated with an increased risk of bleeding, including intracranial and intracerebral hemorrhages1,2 and upper gastro intestinal bleedings.3 A neuroprotective effect in acute ischemic stroke, independent of the antidepressive effect, has also been suggested.4 The effect of prestroke SSRI use on stroke outcome has to date only been scarcely examined.5 We examined the possible implications of prestroke SSRI use in patients with first-ever hemorrhagic and ischemic stroke, including stroke severity and 30-day mortality.
We did a registry-based, nationwide follow-up study in Denmark between 2003 and 2012. We have previously described the relevant Danish registries in detail.2 For additional information, see online-only Data Supplement. The study was approved by the Danish Data Protection Agency. According to Danish law, ethical approval is not needed for registry-based studies.
All first-ever patients with hemorrhage and ischemic stroke ≥18 years discharged between January 2003 and November 2012 were identified from the Danish Stroke Registry. A total of 626 patients with hemorrhagic stroke and 4478 patients with ischemic stroke were identified as prestroke SSRI users and were propensity score matched with nonusers in a 1:1 ratio (see Figure I in the online-only Data Supplement for flowchart). Patients were defined as prestroke SSRI users if ≥1 prescription of SSRI was filled within 90 days before stroke onset (see online-only Data Supplement for prescription codes).
Two outcomes were assessed: severe stroke and 30-day mortality. Information on stroke severity was obtained from the Danish Stroke Registry as assessed by the Scandinavian Stroke Scale score at admission.6 We defined severe stroke as a Scandinavian Stroke Scale score <30 points at the time of admission.7 Additional sensitivity analyses were run with a score <15 points defining severe stroke. Information on 30-days mortality after admission for stroke was obtained by linkage with the Danish Civil Registration System.
We obtained information on a range of covariates with known or suspected association with the use of SSRI or the outcomes. The information obtained at the time of admission included sex, age, housing (eg, own home or nursing home), living arrangements (eg, living alone or living with someone), educational level, income, employment status, alcohol intake, smoking habits, body mass index, previous or current atrial fibrillation, hypertension, type I and II diabetes mellitus, previous history of dementia or myocardial infarction, prestroke use of blood pressure–lowering drugs, platelet inhibitors, vitamin K antagonists, other antidepressants (not SSRIs), statins, and hormonal replacement therapy. Finally, we computed the Charlson comorbidity index score for each patient based on all discharge diagnoses recorded within the last 10 years before stroke hospitalization.8 Because 3 of the disease categories included in the Charlson comorbidity index score (dementia, myocardial infarction, and diabetes mellitus) were included as independent variables in our analyses, these were excluded from the Charlson comorbidity index score. Information on additional covariates was obtained during admission: The quality of in-hospital stroke care was assessed as the proportion of relevant guideline recommended processes of care received by the individual patient (0%–25%, 26%–50%, 51%–75%, 76%–99%, and 100%). The processes of care included early admission to a stroke unit, early initiation of antiplatelet or anticoagulant therapy, early computed tomographic/MRI scan, early assessment by a physiotherapist or occupational therapist, and early assessment of nutritional risk. Finally, information on use of thrombolysis and thrombectomy was included (see Table I in online-only Data Supplement for further description of covariates).
SSRI treatment was not randomly assigned in the study population, so to reduce the risk of bias because of confounding, propensity score matching was used. Using logistic regression, a propensity score was calculated for each patient based on the covariates from the time of admission.9,10 Each prestroke SSRI user was propensity score matched with a nonuser with the same stroke type. We matched in a 1:1 ratio without replacement using a caliper width of 0.2.11 An absolute standardized difference <10% was considered to support the assumption of balance between the groups12 (see Figures II and III in the online-only Data Supplement).
The risk of severe stroke and 30-day mortality was compared among the propensity score–matched prestroke SSRI users and nonusers using multiple conditional12 logistic regression adjusting for the covariates gathered at the time of admission (as listed above). In addition, in the analyses on 30-day mortality, we also adjusted for quality of in-hospital stroke care and for use of thrombolysis and thrombectomy (ischemic strokes). All data analyses were performed using STATA 12.0 (StataCorp, College Station, TX) and SAS 9.3 (SAS Institute Inc, Cary, NC).
For patient characteristics, please see Table 1 (Table I in the online-only Data Supplement for a full descriptive table). In the hemorrhagic group, a total of 574 patients (45.8%) had severe stroke and 461 patients (36.8%) died within 30 days after stroke onset. In the ischemic group, a total of 1679 patients (18.7%) had severe stroke and 958 patients (10.7%) died within 30 days after stroke onset. Table 2 shows the crude and adjusted propensity score–matched odds ratios (OR) for the 2 outcomes among the SSRI users and nonusers. Among patients with hemorrhagic stroke, prestroke SSRI use was associated with an increased risk of the strokes being severe (adjusted propensity score–matched OR, 1.41; confidence interval, 1.08–1.84) and with an increased risk of death within 30 days (adjusted propensity score–matched OR, 1.60; confidence interval, 1.17–2.18). In contrast, in patients with ischemic stroke, prestroke SSRI use was not associated with a higher risk of severe stroke (adjusted propensity score–matched OR, 1.07; confidence interval, 0.97–1.18) or death within 30 days (adjusted propensity score–matched OR, 1.15; confidence interval, 0.99–1.34). Additional analyses run with a score of <15 points defining severe stroke on the Scandinavian Stroke Scale Score left our results virtually unchanged (data not shown).
Presumably the majority of the SSRI users in this study received treatment because of depression, which in itself increases the risk of cardiovascular disease and mortality.13 Given the observational nature of our study, this would lead to a phenomenon often referred to as confounding by indication. Thus, we would expect an increased risk of severe stroke and death in the SSRI-treated patients regardless of stroke type. Interestingly, we found a marked difference between hemorrhagic and ischemic strokes with an increased risk of severe stroke and increased mortality only among SSRI-treated patients with hemorrhagic stroke but not among ischemic strokes. The definition of prestroke SSRI use leaves the possibility that some of the patients defined as SSRI users were not actual users. However, the standard prescription length in Denmark is ≈90 days, and most patients with ≥1 filled prescription will be covered by SSRI for the whole 3-month period. The data provide a good reflection of compliance because we only included data on prescriptions that had been filled, and the patients’ medical expenses were only partially reimbursed, making them more likely to have taken the prescribed drugs. There is no reason to suspect any systematic differences in treatment indication, length of treatment period, severity of a potential depression, or compliance between the 2 stroke types. Any misclassification could, however, bias our risk estimates toward the null hypothesis.
SSRI treatment may possibly potentiate hematoma growth because of platelet inhibition and thus lead to more severe symptoms and increased mortality among the hemorrhagic strokes. However, a possible neuroprotective effect of preadmission SSRI treatment may be present in ischemic stroke because we observed no increased risk of adverse outcomes, despite the indication for SSRI use most likely being an underlying depression.
Prestroke SSRI use may increase stroke severity and mortality in patients with hemorrhagic stroke. In our study, prestroke SSRI use was not associated with an adverse outcome among patients with ischemic stroke. Further studies are warranted to explore the possible hemorrhagic and neuroprotective effects of SSRI treatment in acute stroke.
Sources of Funding
This study was funded by The Tryg Foundation, The Danish Council for Independent Research, and The Aase and Ejner Danielsen Foundation. There has been no interference with the study design, data collection, and analysis or publication from the funding sources.
Presented in part at the International Stroke Conference, San Diego, CA, February 12–14, 2014.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.005302/-/DC1.
- Received February 25, 2014.
- Revision received April 29, 2014.
- Accepted April 30, 2014.
- © 2014 American Heart Association, Inc.
- Mortensen JK,
- Larsson H,
- Johnsen SP,
- Andersen G
- Ried LD,
- Jia H,
- Feng H,
- Cameon R,
- Wang X,
- Tueth M,
- et al
- 6.↵Scandinavian Stroke Study Group. Multicenter trial of hemodilution in ischemic stroke--background and study protocol. Stroke. 1985;16:885–890.
- Johnsen SP,
- Svendsen ML,
- Hansen ML,
- Brandes A,
- Mehnert F,
- Husted SE