This case-crossover study investigates the association between the use of triptans and risk of stroke and myocardial infarction in patients with ischemic risk factors.
Key Points
Question
Is the use of triptans associated with ischemic events to an extent that warrants current contraindication in patients with ischemic risk factors?
Findings
In this case-crossover study including 429 612 individuals, triptan use was associated with a higher risk of ischemic events, but the risk for the individual user was very low. Patients with ischemic outcomes generally had a profile of high cardiovascular risk.
Meaning
Results suggest that avoiding triptan treatment due to concern of triptan-associated ischemic events is rational in patients with a high-risk cardiovascular profile.
Abstract
Importance
Triptans are contraindicated in patients with ischemic heart disease or previous myocardial infarction, and caution is advised when prescribing these drugs to patients with vascular risk factors. However, controlled observational studies have either shown no association or an apparent lower risk, possibly owing to a channeling of triptans to individuals at low risk of cardiovascular outcomes, and it remains unclear whether avoiding triptan treatment for these patients is meaningful.
Objective
To establish whether an association between triptans and ischemic events could be demonstrated using a self-controlled design because this type of design is robust to the previously mentioned type of confounding.
Design, Setting, and Participants
All people in nationwide Danish registries who were initiating triptans and all the ischemic events that they experienced were identified. A case-crossover design was used to estimate odds ratios (OR) for associations between first-ever triptan use and ischemic outcomes, comparing triptan exposure in the 2-week period up to the event with four 2-week reference periods. Data were obtained for the period January 1995 to August 2022. Included from the population of Denmark were individuals redeeming a prescription for any triptan and experiencing at least 1 of 3 predefined ischemic outcomes. No one was excluded.
Exposure
Initiation of any triptan.
Main Outcomes and Measures
Acute myocardial infarction, ischemic stroke, or nonspecified stroke.
Results
Identified were a total of 429 612 individuals (median [IQR] age, 38 [28-48] years; 325 687 female [75.8%]) who redeemed a first prescription for a triptan in the study period. Of these patients, 11 (0.003%) had a myocardial infarction with the first triptan prescription in either a focal or referent window (odds ratio [OR], 3.3; 95% CI, 1.0-10.9), 18 (0.004%) had ischemic stroke (OR, 3.2; 95% CI, 1.3-8.1), and 35 (0.008%) had ischemic/nonspecified stroke (OR, 3.0; 95% CI, 1.5-5.9). Case patients had a median age of approximately 60 years and had a high-risk cardiovascular profile.
Conclusions and Relevance
Results of this case-crossover study suggest that triptan initiation was associated with higher risk of ischemic stroke and myocardial infarction. For the individual patient with low background cardiovascular risk, the risk of an ischemic event after triptan initiation was very low.
Introduction
Triptans are selective serotonin 1B/1D/1F agonists that are among the first-line treatment options for acute migraine attacks. The mechanism of action of triptans in the treatment of migraine pain is not yet clarified.1 Triptans act on serotonin receptors located on multiple sites including peripheral trigeminal sensory nerve endings, on neurons in the trigeminal cervical complex, rostral brainstem, and thalamus as well as on receptors located on intracranial, extracranial, and systemic blood vessels.2 In vitro and in vivo studies have shown vasoconstrictor effects in coronary arteries,3,4,5,6 and myocardial infarctions have been reported after triptan use.7,8,9,10,11
Due to concerns of vasoconstriction leading to ischemic vascular events, triptans are contraindicated in patients with ischemic heart disease or previous myocardial infarction, and caution is advised when prescribing the drugs to patients with vascular risk factors.12,13,14 Although triptans seem to have minimal or no vasoconstrictor effects on cerebral arteries in patients with migraine in vivo,15 triptans are also contraindicated for patients with previous stroke or transient ischemic attack due to a concern of a higher risk of triptan-induced cerebral ischemia in individuals at risk.12,13,14 Observational studies have suggested that patients with migraine with aura have a higher risk of stroke, and it has been debated if triptans are an independent risk factor or mediator for this outcome.16,17
Large case-control studies and cohort studies have not found a higher risk of cardiovascular events in patients with migraine who are treated with triptans.18 This may be explained by the cautions against using triptans in patients at high cardiovascular risk,19,20 such that those who take vs those who do not take triptans have different baseline risks for CV events. If there is incomplete adjustment and residual confounding by these baseline characteristics, a spurious null finding or negative association can emerge. A self-controlled design using patients exposed to triptans as their own controls could avoid this type of confounding because it is immune to confounders that are stable over time, such as genetic disposition, atherosclerosis, smoking behavior, and overweight. We, therefore, conducted this study using a case-crossover design based exclusively on patients who used triptans and experienced a stroke or myocardial infarction.
Methods
Data Sources and Study Population
This case-crossover study was approved by the institutional review board of the University of Southern Denmark. Ethics committee approval is not required for pure register-based studies in Denmark. Owing to the use of deidentified patient data, informed participant consent was not required for this type of study. Data were obtained from Danish nationwide health registers and were linked between registers using the unique civil registration number held by all Danish citizens. Data on prescription drug use were obtained from the Danish National Prescription Register (prescription register) in which redeemed prescriptions were identified according to the Anatomical Therapeutic Chemical (ATC) classification. The indication for the drug is not available in the prescription register. Basic demographic data, medical history including outcome events, and other hospital and outpatient diagnosis codes were obtained from the Danish National Patient Register (patient register). Diagnoses were identified according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) system. We obtained data from the predefined study period January 1995 to August 2022. The analysis was conducted within the setting of the Danish National Health Data Board (Sundhedsdatastyrelsen), a governmental institution that collects health data for administrative, statistical, and research purposes. According to institutional regulation on data confidentiality, researchers are not allowed to report cell counts below 5. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
The prescription register has covered all prescriptions for all residents of Denmark since 1995, ie, from before triptans became widely used. The diagnosis register has covered all hospitalizations for Danish residents since 1977 and all other secondary care contacts since 1994.
All individuals who (1) filled a prescription for a triptan and (2) had 1 of 3 outcomes, ie, myocardial infarction, ischemic stroke, or ischemic/nonspecified stroke, during the study period were eligible for inclusion in this study. As a feature of the case-crossover design, individuals who did not redeem their first prescription of a triptan in either the focal window or one of the reference windows (explanation available subsequently under design considerations) did not contribute to the analyses and were excluded. Persons younger than 18 years at the time of an outcome of interest were also excluded, as were patients for whom data were not available for a full year before redemption of a first triptan prescription. The latter criterion was implemented to allow us to accurately chart comorbidity and coprescribed medication. Data on race and ethnicity were not available.
As an aid to interpretation, we compared baseline characteristics for the individuals contributing to the most extensive analysis (14-day window using 2 years of quarantine) to a random sample of triptan initiators matched according to sex and birth year in a 1:70 ratio. In addition, we charted the same baseline characteristics for all triptan initiators (Table 1).
Table 1. Characteristics of Patients Initiating a Triptan in the Study Period, Patients With an Outcome Contributing to the Case-Crossover Analysis (Cases), and a Sample of Triptan Initiators Age and Sex Matched to the Cases.
| Characteristic | All patients who initiated triptans in the study period (N = 429 612)a | Patients included in the case-crossover analysis (n = 59)b | Random sample of triptan initiators matched to cases on age and sex (n = 4130)c |
|---|---|---|---|
| Sex, No. (%) | |||
| Male | 103 925 (24.2) | 23 (39.7) | 1610 (39.0) |
| Female | 325 687 (75.8) | 35 (60.3) | 2520 (61) |
| Age, median (IQR), y | 38 (28-48) | 59 (44-68) | 59 (44-69) |
| Age, No. (%) | |||
| <60 y | 399 290 (92.9) | 30 (51.7) | 2100 (50.8) |
| ≥60 y | 30 322 (7.1) | 28 (48.3) | 2030 (49.2) |
| Migraine diagnosis in hospital registry | 20 242 (4.7) | 5 (8.6) | 174 (4.2) |
| Migraine with aura | 6453 (1.5) | <5d | 64 (1.5) |
| Migraine without aura | 7024 (1.6) | NA | 52 (1.3) |
| Hypertension | 44 640 (10.4) | 29 (50.0) | 1151 (27.9) |
| Diabetes | 10 106 (2.4) | 8 (13.8) | 217 (5.3) |
| Atrial fibrillation | 3088 (0.7) | <5d | 175 (4.2) |
| Previous acute coronary syndrome | 2226 (0.5) | 12 (20.7) | 87 (2.1) |
| Previous ischemic stroke | 1196 (0.3) | 18 (31.0) | 57 (1.4) |
| Lipid-lowering therapy | 15 466 (3.6) | 22 (37.9) | 561 (13.6) |
| Anticoagulation therapy | 1622 (0.4) | <5d | 71 (1.8) |
| Antiplatelet therapy | 5494 (1.3) | 19 (32.8) | 294 (7.1) |
Abbreviation: NA, not applicable.
Patients who redeemed their first-ever prescription of a triptan in the study period.
Individuals contributing to the most extensive case-crossover analysis, ie, patients with a redemption of a triptan following a 2-year prescription-free period and included in the 14-day exposure period analysis.
Matched to cases included in the case-crossover analysis at a ratio of 1:70. Due to sparse numbers, patients 90 years and older were categorized as 90+ years and matched accordingly.
According to patient data protection regulations, we are not allowed to report counts below 5.
Exposure, Outcomes, and Covariates
Exposure was defined as the patient’s first-ever redemption of a prescription for any triptan (ATC codes N02CC01-N02CC07) in any formulation. We only included individuals who redeemed a first prescription after the first full year of the study period to maximize the likelihood that this was truly their first-ever prescription. As exposure assessment in this study is based on prescriptions only, we assumed that a patient was exposed to a triptan for a 14-day period beginning on the date a triptan prescription was redeemed. We used this 14-day time window of exposure in the main analysis (Figure) and ran sensitivity analyses using alternative time windows.
Figure. Schematic Representation of the Case-Crossover Sampling Method With 4 Case Examples.
Case 1 has an exposure in one of the reference windows and is not exposed during the focal window, while the opposite is true of case 2. Case 3 has no exposures in either the focal window or reference windows and does not contribute to the analysis. Like case 1, case 4 has an exposure in a reference window and contributes to the analysis in the same way as case 1 with regard to exposure status.
The outcomes of interest (ICD-10 codes in parentheses) were acute myocardial infarction (I21), ischemic stroke (I63), and ischemic or unspecified stroke (I63, I64). We required that an eligible diagnosis (ie, the date of a hospital contact coded with the diagnosis) should have at least 6 months’ distance to any previous hospital contact with that diagnosis in the patient register. This ensured that an admission was not included as an outcome when it was, in fact, a late manifestation of a previous outcome. To characterize triptan users according to risk profile, comorbid cardiovascular risk factors were defined according to their diagnoses and use of prescription drugs (eTable 1 in Supplement 1).
Design Considerations
The study was designed as a case-crossover study in which the past experience of each case patient serves as the patient’s own control.21 The exposure attributes of a focal window immediately before the outcome were compared—within the same individual—with exposure attributes in one or more reference windows before the focal window (Figure). Thereby, confounders that were stable over the individual participant’s study period (84 days or less) were eliminated by design.
This ensures that confounders such as genetic disposition, atherosclerosis, smoking behavior, and overweight cancel out. Such near-optimal control of confounding is difficult to achieve in, eg, a conventional case-control design. The case-crossover design is particularly useful for intermittent exposure with immediate, transient effect on risk and abrupt outcomes22 as would be the case for a possible association between triptan use and cardiovascular events.
In accordance with the assumed exposure duration, we set the length of the focal window (preceding and including the day of an outcome of interest) and reference windows in the main analysis to 14 days. Four consecutive reference windows of 14 days were applied, separated from the focal window by a 14-day washout window. A patient was considered exposed in a specific time window if that window contained the patient’s first-ever triptan prescription. The use of 4 reference windows was motivated by a better statistical precision achieved.23
Statistical Analysis
The exposure in the focal windows and in the referent windows were used to estimate odds ratios (ORs) with 95% CIs according to the Mantel-Haenszel OR.24 We performed 5 sensitivity analyses. In one sensitivity analysis, the focal and reference windows were 7 days instead of 14 days. In another set of sensitivity analyses we did not require that the triptan prescription should be the first ever but also included triptan prescriptions that were the first observed after 2 or 3 years with no triptan prescriptions. Thereby, 5 separate sensitivity analyses (14 day windows, first prescription in 2 years; 14 day windows, first prescription in 3 years; 7-day windows, first-ever prescription; 7 day windows, first prescription in 2 years; 7 day windows, first prescription in 3 years) were carried out apart from the main analysis. Subgroup-analyses were performed according to sex, age (< or ≥60 years), and selected comorbid risk factors. Data on race or ethnicity were not available. All analyses were performed in Stata, version 18.0 (StataCorp).
Results
Study Population Inclusion and Characteristics
Individuals who redeemed a first-ever prescription for a triptan in the study period comprised 429 612 persons (median [IQR] age, 38 [28-48] years; 103 925 male [25.2%]; 325 687 female [75.8%]). Of the 429 612 individuals with a first-ever prescription for a triptan, 11 (0.003%) redeemed this first-ever prescription within one of the focal or referent windows preceding an acute myocardial infarction, 18 (0.004%) had ischemic stroke, and 35 (0.008%) had ischemic/nonspecified stroke. These individuals were included in the main analysis.
In the sensitivity analyses, the number of included individuals varied due to variations in definitions of exposure. Baseline characteristics are presented in Table 1 and Table 2. Compared with triptan users, individuals with outcomes had more cardiovascular risk factors (hypertension, 29 of 59 [50.0%] vs 44 640 of 429 612 [10.4%]) and were older (aged ≥60 years, 28 of 59 [48/3%] vs 30 322 of 429 612 [7.1%]), and the differences were sustained when comparing with a large group of patients matched to the patients with outcomes according to sex and birth year.
Table 2. Characteristics of Patients With a Redemption of a Triptan After a 2-Year Prescription-Free Period and Included in 14-Day Exposure Period Analysis.
| Characteristic | Cases per outcome of interest (%) | ||
|---|---|---|---|
| Myocardial infarction (n = 13) | Ischemic stroke (n = 25) | Ischemic or unspecified stroke (n = 46) | |
| Sex, No. (%) | |||
| Male | 7 (53.8) | 10 (40.0) | 17 (37.0) |
| Female | 6 (46.2) | 15 (60) | 29 (63) |
| Age, median (IQR), y | 63 (46-71) | 61 (52-70) | 61 (48-71) |
| Age, No. (%) | |||
| <60 y | 5 (38.5) | 11 (44.0) | 22 (47.8) |
| ≥60 y | 8 (61.5) | 14 (56.0) | 24 (52.2) |
| Migraine diagnosis in hospital registry | <5a | <5a | 6 (13.0) |
| Migraine with aura | 0 | <5a | <5a |
| Migraine without aura | 0 | 0 | 0 |
| Hypertension | 10 (76.9) | 17 (68.0) | 24 (52.2) |
| Diabetes | <5a | 6 (24.0) | 8 (17.4) |
| Atrial fibrillation | <5a | 0 | <5a |
| Previous acute coronary syndrome | 9 (69.2) | <5a | 7 (15.2) |
| Previous ischemic stroke | 0 | 15 (60.0) | 19 (41.3) |
| Lipid-lowering therapy | 8 (61.5) | 13 (52.0) | 20 (43.5) |
| Anticoagulation therapy | <5a | <5a | 0 |
| Antiplatelet therapy | <5a | 13 (52.0) | 23 (50.0) |
According to patient data protection regulations, we are not allowed to report counts below 5.
For patients included in the analysis with the highest number of patients (14 day windows, first prescription in 2 years), median age was approximately 60 years and was similar between outcome groups (median [IQR] age: myocardial infarction, 63 [46-71] years; ischemic stroke, 61 [52-70] years; ischemic or unspecified stroke, 61 [48-71] years). Approximately two-thirds of patients were female (50 of 84 [59.5%]). For the 3 outcome groups of myocardial infarction, ischemic stroke, and any stroke, 76.9% (10 of 13), 68.0% (17 of 25), and 52.2% (24 of 46), respectively, had hypertension. For a history of lipid-lowering treatment, the corresponding numbers in the 3 outcome groups were 61.5% (8 of 13), 52.0% (13 of 25), and 43.5% (20 of 46). Prevalence of other comorbidities varied between groups.
Less than 5 patients in the myocardial infarction and ischemic stroke outcome groups had a registered migraine diagnosis in the patient register, whereas 13% of patients (6 of 46) with any stroke had an existing migraine diagnosis.
Case-Crossover Analysis
Results of the case-crossover analyses are presented in Table 3. Eleven patients contributed to the analysis for the outcome myocardial infarction. These patients were more likely to have been exposed to a triptan in the outcome window compared with the reference windows (OR, 3.3; 95% CI, 1.0-10.9). In subgroup analyses, triptan use was associated with higher risk of myocardial infarction in patients with previous acute coronary syndrome (OR, 5.3; 95% CI, 1.2-23.8), hypertension (OR, 4.0; 95% CI, 1.0-16.0), and in patients younger than 60 years (OR, 6.0; 95% CI, 1.0-35.9).
Table 3. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios, Between First-Ever Prescription of a Triptan and 14-Day Risk of Case-Defining Outcome.
| Characteristic | Myocardial infarction | Ischemic stroke | Ischemic or unspecified stroke | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No. of events | Exposed in focal window, No. (%) | Exposed in reference window, No. (%) | OR (95% CI) | No. of events | Exposed in focal window, No. (%) | Exposed in reference window, No. (%) | OR (95% CI) | No. of events | Exposed in focal window, No. (%) | Exposed in reference window, No. (%) | OR (95% CI) | |
| Overall | 11 | 5 (45.5) | 6 (13.6) | 3.3 (1.0 - 10.9) | 18 | 8 (44.4) | 10 (13.9) | 3.20 (1.3-8.1) | 35 | 15 (42.9) | 20 (14.3) | 3.00 (1.5-5.9) |
| Sex | ||||||||||||
| Male | 6 | <5a | <5a | 4.0 (0.8-19.8) | 9 | 6 (66.7) | <5a | 8.0 (2.0-32.0) | 15 | 10 (66.7) | 5 (8.3) | 8.0 (2.7-23.4) |
| Female | 5 | <5a | <5a | 2.7 (0.5-16.0) | 9 | <5a | 7 (19.4) | 1.1 (0.2-5.5) | 20 | 5 (25.0) | 15 (18.8) | 1.3 (0.5-3.7) |
| Age group | ||||||||||||
| <60 y | 5 | <5a | <5a | 6.0 (1.0-35.9) | 7 | <5a | <5a | 5.3 (1.2-23.8) | 16 | 7 (43.8) | 9 (14.1) | 3.1 (1.2-8.4) |
| ≥60 y | 6 | <5a | <5a | 2.0 (0.4-10.9) | 11 | <5a | 7 (15.9) | 2.3 (0.7-7.8) | 19 | 8 (42.1) | 11 (14.5) | 2.9 (1.2-7.2) |
| Comorbidity | ||||||||||||
| Hypertension | 8 | <5a | <5a | 4.0 (1.0-16.0) | 12 | 6 (50.0) | 6 (12.5) | 4.0 (1.3-12.4) | 17 | 9 (52.9) | 8 (11.8) | 4.5 (1.7-11.7) |
| Diabetes | <5a | <5a | <5a | NA | <5a | <5a | <5a | NA | 5 | <5a | <5a | 16.0 (1.8-143.2) |
| Atrial fibrillation | <5a | <5a | <5a | NA | <5a | <5a | <5a | NA | <5a | <5a | <5a | NA |
| Previous acute coronary syndrome | 7 | <5a | <5a | 5.3 (1.2-23.8) | <5a | <5a | <5a | NA | <5a | <5a | <5a | NA |
| Previous ischemic stroke | <5a | <5a | <5a | NA | 8 | <5a | 5 (15.6) | 2.4 (0.6-10.0) | 13 | 5 (38.5) | 8 (15.4) | 2.5 (0.8-7.6) |
| Lipid-lowering therapy | 6 | <5a | <5a | 4.0 (0.8-19.8) | 7 | <5a | <5a | 5.3 (1.2 - 23.8) | 12 | 5 (41.7) | 7 (14.6) | 2.9 (0.9-9.0) |
| Anticoagulation therapy | <5a | <5a | <5a | NA | <5a | <5a | <5a | NA | <5a | <5a | <5a | NA |
| Antiplatelet therapy | <5a | <5a | <5a | NA | 6 | <5a | <5a | 2.0 (0.4-10.9) | 16 | 7 (43.8) | 9 (14.1) | 3.1 (1.2-8.4) |
Abbreviation: NA, not applicable.
According to patient data protection regulations, we are not allowed to report counts below 5.
Eighteen patients contributed to the analysis for the outcome ischemic stroke. For these patients, the OR was 3.2 (95% CI, 1.3-8.1). Subgroup analyses identified higher risks in patients with hypertension (OR, 4.0; 95% CI, 1.3-12.4) and patients younger than 60 years (OR, 5.3; 95% CI, 1.2-23.8) at the time of their stroke. The risk was especially higher in male patients (OR, 8.0; 95% CI, 2.0-32.0). A higher risk was also found in patients with previous ischemic stroke although the CI included unity (OR, 2.4; 95% CI, 0.6-10.0).
Thirty-five patients contributed to the analysis for the outcome ischemic or unspecified stroke. These were 3 times as likely to be exposed to a triptan in the focal window leading up to their stroke, compared with reference windows (OR, 3.0; 95% CI, 1.5-5.9). The risk was highest in men (OR, 8.0; 95% CI, 2.7-23.4) and in patients with diabetes (OR, 16.0; 95% CI, 1.8-143.2), and the risk was similarly high for patients younger than 60 years and those 60 years and older (<60 years: OR, 3.1; 95% CI, 1.2-8.4; ≥60 years: OR, 2.9; 95% CI, 1.2-7.2). All subgroup analyses involved very low numbers of patients for all outcomes (for the most part, <5 and no more than 15).
Sensitivity Analyses
The most restrictive analysis (7-day exposure periods and first-ever prescription) found higher risks for all 3 outcomes with triptan exposure, with higher estimates of ORs and wider CIs, as these were based on low numbers of included individuals (eTable 1 in Supplement 1). Patients with acute myocardial infarction were more likely to have been exposed to a triptan in the outcome window compared with the reference windows (OR, 5.3; 95% CI, 1.2-23.8). For patients experiencing ischemic stroke, OR was 4.0 (95% CI, 1.2-13.8), and for patients with any stroke, OR was 5.0 (95% CI, 2.0-12.7).
The analyses applying a 2- or 3-year prescription-free period instead of a first-ever prescription all found higher risks with triptan exposure of similar statistical strength. For acute myocardial infarction, however, the CI included the null. Subgroup analyses involved very low numbers of individuals for all outcomes in the sensitivity analyses (eTables 2-6 in Supplement 1).
Discussion
In this case-crossover study of patients using triptans, we identified higher risks of myocardial infarction and ischemic stroke associated with the period after redemption of a prescription for a triptan. The association with myocardial infarction was strongest in patients younger than 60 years and in those with previous acute coronary syndrome, whereas the association with ischemic stroke was strongest in men and in patients younger than 60 years. Alternative definitions of exposure and exposure periods applied in sensitivity analyses did not substantially alter the association, although not all sensitivity analyses reached statistical significance. It should be emphasized that these are relative risk measures and that the high values for young individuals should be seen in context with a low baseline risk in these individuals.
Our finding of a higher risk of ischemic outcomes with triptan use is at odds with other pharmacoepidemiologic studies finding no change in risk.18,20,25,26 We interpret this as a matter of residual confounding in these other studies by triptans being channeled to low-risk individuals, thereby obscuring the increased risk associated with triptan use. Our study is self-controlled and thereby robust to such confounding, and it has produced an estimate that is consistent with the pharmacologic understanding of triptans as vasoconstrictors. It is also noteworthy that the patients included as cases in our study generally appeared to have a high-risk cardiovascular profile, also higher than an age- and sex-matched group of other triptan initiators.
The very low number of outcomes included in the analyses is a limitation but also an important result in and of itself. Our data allowed us to identify all relevant outcomes up to 84 days after initiation for more than 400 000 triptan users. The low number of outcomes thus testifies to the low absolute risk among initiators. Only 20 exposed cases occurred with either myocardial infarction or stroke in the main analysis. Assuming an OR of 3.00, the attributable proportion among exposed is 66% (= [OR − 1] / OR). Thereby, taking our results at face value, only 13 outcomes can be considered attributable to triptans, or 1 in 30 000 new users. Thus, although we did find higher risks of ischemic stroke associated with triptan use, the risk for the individual patient was very low. It should be emphasized, though, that this risk pertains to an average initiator of triptans, most of whom are young women. Nearly all our patients with outcomes had cardiovascular risk factors, and the majority were older than 60 years. For the corresponding subpopulation of triptan initiators, the risk is likely of a higher magnitude than 1 in 30 000.
Strengths and Limitations
This study has several strengths. The case-crossover design allowed us to perform analyses that were robust to several unmeasured confounders that were stable over each user’s reference and focal windows (eg, smoking, overweight, degree of atherosclerosis) and also eliminated confounding by contraindication. The use of nationwide health registers with complete coverage and follow-up for all citizens ensured practically complete eligibility for inclusion inside the population of Denmark, thereby minimizing the risk of selection bias. We were thus able to estimate the risk for the individual triptan initiators. By using these registers, we had the opportunity to investigate associations in a large-scale, real-life setting.
Our study also has several limitations. Most importantly, the applied definition of exposure is based on prescription redemption. We cannot know if or when the patient actually ingested the medication. The defined 14-day exposure period serves as a rough proxy of the timing of exposure. We may assume that triptans are bought when the patient needs treatment for a migraine attack, at least for the first-ever prescription. We note that the sensitivity analysis using a 7-day exposure window (as opposed to the 14-day time window in the main analysis) returned higher ORs than the main analysis, which supports a conservative misclassification of the risk estimates even when using a 14-day window.
Our data were from people who were primarily of European ancestry and may not be generalizable to other populations. As mentioned, the low number of individuals contributing to the main analyses is a limitation, warranting careful interpretation of our findings, particularly for the subanalyses.
Finally, we need to consider the possibility of protopathic bias, ie, that the triptans were prescribed for something that was really a first manifestation of the outcome. Cerebrovascular events may themselves present primarily as migrainelike headache, and misdiagnosis of transient ischemic attack and minor stroke as migraine in the emergency department setting is not rare.27 A recent study reported that patients discharged to home from the emergency department with a headache diagnosis have a 3- to 4-fold higher risk of stroke in the following month compared with patients presenting with kidney colic or acute back pain.28 This higher risk was most pronounced within the first 10 days following discharge. We find it less likely that this protopathic bias also applies to myocardial infarction as an outcome, but we acknowledge that it cannot be dismissed entirely because cardiac ischemia may occasionally manifest as migrainelike headache.29 The very low proportion of patients with a triptan prescription who had a registered migraine diagnosis (4.7%) could be interpreted as a sign of protopathic bias, but we believe this to be explained by limitations in our data sources, as only diagnoses registered in hospital settings were available, and most patients with migraine in Denmark are managed exclusively in the primary sector.
Conclusions
Results of this case-crossover study suggest that although use of triptans may be associated with a short-term higher risk of stroke or myocardial infarction, this amounted to a very small number of patients initiating triptans. However, the results of this study support the current US Food and Drug Administration recommendation that triptans should not be prescribed to patients with a history of coronary artery disease, transient ischemic attack, or stroke. Triptans are generally considered safe in patients with low cardiovascular risk, and our findings do not raise concern about triptan use in this population group.
eTable 1. Definitions of Risk Factors According to Diagnoses and Use of Prescription Drugs
eTable 2. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios, Between First-Ever Prescription of a Triptan and 7-Day Risk of Case-Defining Outcome
eTable 3. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios—14-Day Exposure Period Analysis on Patients With a 2-Year Prescription-Free Period
eTable 4. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios—14-Day Exposure Period Analysis on Patients With a 3-Year Prescription-Free Period
eTable 5. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios—7-Day Exposure Period Analysis on Patients With a 2-Year Prescription-Free Period
eTable 6. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios—7-Day Exposure Period Analysis on Patients With a 3-Year Prescription-Free Period
Data Sharing Statement.
References
- 1.Tfelt-Hansen P. Pharmacological strategies to treat attacks of episodic migraine in adults. Expert Opin Pharmacother. 2021;22(3):305-316. doi: 10.1080/14656566.2020.1828347 [DOI] [PubMed] [Google Scholar]
- 2.Rubio-Beltrán E, Labastida-Ramírez A, Villalón CM, MaassenVanDenBrink A. Is selective 5-HT1F receptor agonism an entity apart from that of the triptans in antimigraine therapy? Pharmacol Ther. 2018;186:88-97. doi: 10.1016/j.pharmthera.2018.01.005 [DOI] [PubMed] [Google Scholar]
- 3.Barra S, Lanero S, Madrid A, et al. Sumatriptan therapy for headache and acute myocardial infarction. Expert Opin Pharmacother. 2010;11(16):2727-2737. doi: 10.1517/14656566.2010.522567 [DOI] [PubMed] [Google Scholar]
- 4.Welch KM, Mathew NT, Stone P, Rosamond W, Saiers J, Gutterman D. Tolerability of sumatriptan: clinical trials and postmarketing experience. Cephalalgia. 2000;20(8):687-695. doi: 10.1046/j.1468-2982.2000.00116.x [DOI] [PubMed] [Google Scholar]
- 5.Macintyre PD, Bhargava B, Hogg KJ, Gemmill JD, Hillis WS. The effect of IV sumatriptan, a selective 5-HT1-receptor agonist, on central haemodynamics and the coronary circulation. Br J Clin Pharmacol. 1992;34(6):541-546. [PMC free article] [PubMed] [Google Scholar]
- 6.MacIntyre PD, Bhargava B, Hogg KJ, Gemmill JD, Hillis WS. Effect of subcutaneous sumatriptan, a selective 5HT1 agonist, on the systemic, pulmonary, and coronary circulation. Circulation. 1993;87(2):401-405. doi: 10.1161/01.CIR.87.2.401 [DOI] [PubMed] [Google Scholar]
- 7.Jensen C, Riddle M. ST-elevation myocardial infarction after sumitriptan ingestion in patient with normal coronary arteries. West J Emerg Med. 2015;16(5):781-783. doi: 10.5811/westjem.2015.6.25920 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chalaupka FD. Acute myocardial infarction with sumatriptan: a case report and review of the literature. Headache. 2009;49(5):762-764. doi: 10.1111/j.1526-4610.2009.01409.x [DOI] [PubMed] [Google Scholar]
- 9.Okonkwo K, Ojha U. Vasospasm induced myocardial ischaemia secondary to sumatriptan use. BMJ Case Rep. 2020;13(8):e235387. doi: 10.1136/bcr-2020-235387 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Weir RAP, Oldroyd KG, Walters MR. Myocardial infarction after oral sumatriptan use in a woman with normal coronary arteries. Am J Med. 2007;120(11):e7-e8. doi: 10.1016/j.amjmed.2006.03.047 [DOI] [PubMed] [Google Scholar]
- 11.Erbilen E, Ozhan H, Akdemir R, Yazici M. A case of myocardial infarction with sumatriptan use. Pediatr Cardiol. 2005;26(4):464-466. doi: 10.1007/s00246-004-0768-4 [DOI] [PubMed] [Google Scholar]
- 12.Dodick DW. Migraine. Lancet. 2018;391(10127):1315-1330. doi: 10.1016/S0140-6736(18)30478-1 [DOI] [PubMed] [Google Scholar]
- 13.Danish Medicines Agency . Sumatriptan accord. Accessed February 22, 2023. https://www.produktresume.dk/AppBuilder/search?q=%22Sumatriptan+Accord%2C+filmovertrukne+tabletter+50+mg+og+100+mg.docx%22
- 14.US Food and Drug Administration . Sumatriptan succinate product information. Accessed March 27, 2023. https://www.accessdata.fda.gov/drugsatfda_docs/label/2013/020132s028,020626s025lbl.pdf
- 15.Amin FM, Asghar MS, Hougaard A, et al. Magnetic resonance angiography of intracranial and extracranial arteries in patients with spontaneous migraine without aura: a cross-sectional study. Lancet Neurol. 2013;12(5):454-461. doi: 10.1016/S1474-4422(13)70067-X [DOI] [PubMed] [Google Scholar]
- 16.Becker C, Brobert GP, Almqvist PM, Johansson S, Jick SS, Meier CR. Migraine and the risk of stroke, TIA, or death in the UK (CME). Headache. 2007;47(10):1374-1384. doi: 10.1111/j.1526-4610.2007.00937.x [DOI] [PubMed] [Google Scholar]
- 17.Albieri V, Olsen TS, Andersen KK. Risk of stroke in migraineurs using triptans. associations with age, sex, stroke severity and subtype. EBioMedicine. 2016;6:199-205. doi: 10.1016/j.ebiom.2016.02.039 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Hall GC, Brown MM, Mo J, MacRae KD. Triptans in migraine: the risks of stroke, cardiovascular disease, and death in practice. Neurology. 2004;62(4):563-568. doi: 10.1212/01.WNL.0000110312.36809.7F [DOI] [PubMed] [Google Scholar]
- 19.Bigal ME, Golden W, Buse D, Chen YT, Lipton RB. Triptan use as a function of cardiovascular risk. A population-based study. Headache. 2010;50(2):256-263. doi: 10.1111/j.1526-4610.2009.01595.x [DOI] [PubMed] [Google Scholar]
- 20.Velentgas P, Cole JA, Mo J, Sikes CR, Walker AM. Severe vascular events in migraine patients. Headache. 2004;44(7):642-651. doi: 10.1111/j.1526-4610.2004.04122.x [DOI] [PubMed] [Google Scholar]
- 21.Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol. 1991;133(2):144-153. doi: 10.1093/oxfordjournals.aje.a115853 [DOI] [PubMed] [Google Scholar]
- 22.Cadarette SM, Maclure M, Delaney JAC, et al. Control yourself: ISPE-endorsed guidance in the application of self-controlled study designs in pharmacoepidemiology. Pharmacoepidemiol Drug Saf. 2021;30(6):671-684. doi: 10.1002/pds.5227 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Mittleman MA, Maclure M, Robins JM. Control sampling strategies for case-crossover studies: an assessment of relative efficiency. Am J Epidemiol. 1995;142(1):91-98. doi: 10.1093/oxfordjournals.aje.a117550 [DOI] [PubMed] [Google Scholar]
- 24.Kubota K, Kelly TL, Sato T, Pratt N, Roughead E, Yamaguchi T. A novel weighting method to remove bias from within-subject exposure dependency in case-crossover studies. BMC Med Res Methodol. 2021;21(1):214. doi: 10.1186/s12874-021-01408-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ghanshani S, Chen C, Lin B, Duan L, Shen YA, Lee MS. Risk of acute myocardial infarction, heart failure, and death in migraine patients treated with triptans. Headache. 2020;60(10):2166-2175. doi: 10.1111/head.13959 [DOI] [PubMed] [Google Scholar]
- 26.Roberto G, Raschi E, Piccinni C, et al. Adverse cardiovascular events associated with triptans and ergotamines for treatment of migraine: systematic review of observational studies. Cephalalgia. 2015;35(2):118-131. doi: 10.1177/0333102414550416 [DOI] [PubMed] [Google Scholar]
- 27.Tarnutzer AA, Lee SH, Robinson KA, Wang Z, Edlow JA, Newman-Toker DE. ED misdiagnosis of cerebrovascular events in the era of modern neuroimaging: a meta-analysis. Neurology. 2017;88(15):1468-1477. doi: 10.1212/WNL.0000000000003814 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Liberman AL, Zhang C, Lipton RB, et al. Short-term stroke risk after emergency department treat-and-release headache visit. Headache. 2022;62(9):1198-1206. doi: 10.1111/head.14387 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Wei JH, Wang HF. Cardiac cephalalgia: case reports and review. Cephalalgia. 2008;28(8):892-896. doi: 10.1111/j.1468-2982.2008.01590.x [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Definitions of Risk Factors According to Diagnoses and Use of Prescription Drugs
eTable 2. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios, Between First-Ever Prescription of a Triptan and 7-Day Risk of Case-Defining Outcome
eTable 3. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios—14-Day Exposure Period Analysis on Patients With a 2-Year Prescription-Free Period
eTable 4. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios—14-Day Exposure Period Analysis on Patients With a 3-Year Prescription-Free Period
eTable 5. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios—7-Day Exposure Period Analysis on Patients With a 2-Year Prescription-Free Period
eTable 6. Frequencies of Events and Exposure and Case-Crossover Estimates for Odds Ratios—7-Day Exposure Period Analysis on Patients With a 3-Year Prescription-Free Period
Data Sharing Statement.