Prior Dual Antiplatelet Therapy and Mortality After Intracerebral Hemorrhage: A Danish Nationwide Cohort Study

Clinical Question (PICO)

Among adults aged 50 years or older with a first-ever intracerebral hemorrhage (ICH) who were receiving antiplatelet therapy at admission (P), is current dual antiplatelet therapy (DAPT) associated with different mortality than single antiplatelet therapy (SAPT) (I/C), and does the association vary by concurrent oral anticoagulation, age, sex, or duration of antiplatelet use (prespecified subgroup analyses)? The outcomes were cumulative all-cause mortality by day 1, day 30, and day 180, with admission neurological severity as a secondary outcome. This was an observational registry study, not a treatment trial.

Bottom Line

In this Danish nationwide cohort, 180-day mortality was 52.5% among 265 patients receiving DAPT and 45.6% among 5,535 receiving SAPT. After stabilized inverse probability of treatment weighting, DAPT was associated with higher mortality at day 1 (weighted RR 1.43, 95% CI 1.13–1.82; P=.003), day 30 (1.29, 1.13–1.47; P<.001), and day 180 (1.14, 1.01–1.29; P=.03). The excess was concentrated in the first day, consistent with—but not proving—a connection to early hematoma expansion.

Design

Trial type: Nationwide retrospective observational cohort using linked Danish health and prescription registries; no randomization or blinding.
N: 5,800 patients with ICH who were current antiplatelet users at admission.
- DAPT: 265 (4.6%).
- SAPT: 5,535 (95.4%).
Randomization: None; stabilized inverse probability of treatment weighting was used to construct a covariate-balanced pseudo-population.
Setting: Denmark’s nationwide stroke, patient, prescription, and civil-registration systems; Denmark has a population of approximately 6 million and tax-funded access to acute stroke care.
Enrollment: January 1, 2005, through August 31, 2022.
Follow-up: From index admission to death or a maximum of 180 days; mortality landmarks were day 1, day 30, and day 180.
Analysis: Ten multiply imputed data sets; generalized linear binomial models for risk ratios and absolute risk differences; stabilized IPTW for confounding control; bootstrap sensitivity analyses; Kaplan–Meier estimates for cumulative mortality. Admission Scandinavian Stroke Scale (SSS) was excluded from the primary propensity model because it was treated as a possible mediator and was examined separately.
Primary outcome: Cumulative all-cause mortality by day 1, day 30, and day 180 after ICH admission.

Population

Inclusion Criteria

Age 50 years or older.
First-ever ICH recorded in the Danish Stroke Registry.
Admission from January 2005 through August 2022.
Current antiplatelet medication covering the index date.

Exclusion Criteria

Previous ICH.
Younger than 50 years, an age restriction selected to enrich for spontaneous rather than secondary ICH.
No current antiplatelet exposure at admission.

Baseline Characteristics (before weighting)

Mean age was 75.7 years in the DAPT group and 77.3 years in the SAPT group; 61.9% and 53.1%, respectively, were men.
Mean admission SSS was 27.6 with DAPT and 28.2 with SAPT; lower SSS indicates more severe neurological deficit.
DAPT users had more ischemic heart disease (64.9% vs 28.2%), congestive heart failure (21.5% vs 9.1%), chronic kidney disease (9.4% vs 3.7%), and prior ischemic stroke or TIA (50.6% vs 43.2%).
Among SAPT users, 3,362 (60.7%) were taking aspirin and 1,208 (21.8%) clopidogrel. Among DAPT users, 222 (83.8%) were taking aspirin plus clopidogrel.

Interventions

DAPT exposure: Current prescriptions for two antiplatelet agents at the time of ICH admission. Antiplatelets included aspirin, clopidogrel, ticagrelor, prasugrel, and dipyridamole; the dominant combination was aspirin plus clopidogrel.
SAPT comparator: Current prescription for one antiplatelet agent at ICH admission. Aspirin alone and clopidogrel alone were also used as restricted comparator groups in sensitivity analyses.
This was not an intervention study: the article evaluated medication exposure already present when ICH occurred and did not assign reversal, surgery, transfusion, or other acute treatments.

Outcomes

Primary Outcome (cumulative all-cause mortality after ICH):

Day 1: Crude mortality was 24.2% with DAPT versus 15.3% with SAPT; weighted RR 1.43 (95% CI 1.13–1.82; P=.003).
Day 30: Crude mortality was 47.2% versus 37.0%; weighted RR 1.29 (95% CI 1.13–1.47; P<.001).
Day 180: Crude mortality was 52.5% versus 45.6%; weighted RR 1.14 (95% CI 1.01–1.29; P=.03). When uncertainty from propensity-score estimation was incorporated, evidence for a difference was retained at day 1 and day 30 but weakened at day 180.
Incident mortality analyses suggested that the separation was concentrated in day 0–1 (crude RR 1.58, 95% CI 1.27–1.98); there was no comparable evidence for the later intervals.

Secondary Outcomes:

After weighting, DAPT was associated with a 15% higher likelihood of severe neurological deficits on admission (SSS 0–25; weighted RR 1.15, 95% CI 1.00–1.32). The median SSS was otherwise similar: 30.0 with DAPT versus 29.5 with SAPT.
In prespecified subgroup analyses, the day-1 association among patients without concurrent oral anticoagulant use was weighted RR 1.40 (95% CI 1.07–1.84); among those with concurrent anticoagulant use it was 0.74 (0.30–1.79), with no evidence of heterogeneity (P=.17).
For day-180 mortality, concurrent anticoagulant use modified the association: weighted RR 1.58 (95% CI 1.30–1.93) with anticoagulant use versus 1.08 (0.94–1.24) without it; P for heterogeneity=.002. The article reports similar direction across age and sex strata. Duration of current antiplatelet use showed evidence of heterogeneity for day-1 mortality (P=.001), but the small DAPT subgroup makes this exploratory in practical interpretation.
Compared with aspirin alone, DAPT was associated with weighted RR 1.60 (95% CI 1.27–2.03) at day 1, 1.43 (1.26–1.63) at day 30, and 1.23 (1.09–1.38) at day 180. Compared with clopidogrel alone, the corresponding estimates were 1.23 (0.95–1.61), 1.16 (0.99–1.35), and 1.10 (0.96–1.26), without clear evidence of excess risk.
Reclassifying aspirin plus dipyridamole as DAPT yielded weighted RRs of 1.20 (95% CI 1.03–1.39) at day 1, 1.17 (1.08–1.27) at day 30, and 1.13 (1.05–1.21) at day 180.

Adverse Events / Safety:

Mortality was the principal safety outcome. This study included patients after ICH had already occurred, so it does not estimate the risk of causing ICH, symptomatic ICH, or bleeding from initiating DAPT.
Hematoma volume, hematoma expansion, intraventricular extension, reversal treatment, functional outcome, and cause-specific mortality were not directly measured in the primary registry analysis. The proposed mechanism—more aggressive early hematoma dynamics—therefore remains a hypothesis.

Figures

Study flow chart for the Danish ICH cohort — *Figure 1.* Study flow chart. APT indicates antiplatelet therapy; DAPT, dual antiplatelet therapy; DNPR, Danish National Prescription Registry; DSR, Danish Stroke Registry; ICH, intracerebral hemorrhage; and SAPT, single antiplatelet therapy.

Cumulative 180-day mortality after ICH by antiplatelet exposure — *Figure 2.* Crude cumulative all-cause mortality after intracerebral hemorrhage during 180 days of follow-up. DAPT indicates dual antiplatelet therapy; ND, not disclosed because numbers below five were redacted under data-protection regulations; and SAPT, single antiplatelet therapy.

Criticisms

This is an association study, not evidence that DAPT caused the excess mortality. DAPT users had markedly more ischemic heart disease, heart failure, kidney disease, prior ischemic events, and statin use before weighting; key drivers of why DAPT was prescribed may remain unmeasured.
The exposed group contained only 265 patients, and fewer than 5% of antiplatelet-treated ICH patients were on DAPT. Estimates in the anticoagulant, duration, aspirin, and clopidogrel strata are therefore vulnerable to imprecision and sparse-data behavior.
Admission SSS was treated as a mediator rather than included in the primary propensity model. That is defensible if DAPT contributes to early neurological injury, but it also means the headline estimate is not a comparison fully adjusted for baseline neurological severity; adding SSS weakened the estimates while leaving day-1 and day-30 evidence.
Prescription coverage does not establish adherence, the actual dose, in-hospital continuation, or acute reversal treatment. Over-the-counter low-dose aspirin was not captured, although the authors estimate that approximately 90% of low-dose aspirin use in Denmark was prescribed.
The registry analysis did not directly measure hematoma volume or expansion, intraventricular extension, ICH location, functional outcome, or cause of death. The early mortality pattern supports a hematoma-expansion hypothesis but cannot test it.
The cohort was restricted to Danish residents aged 50 years or older receiving antiplatelet therapy, limiting transportability to younger patients, other health systems, and patients with ICH who were not taking antiplatelets.

Funding

No specific funding was obtained for this article. This was not a registered interventional trial and no NCT registration was reported. The paper discloses author relationships and external grants, including grants to investigators from Novo Nordisk, Leo Pharma, Roche, Pfizer, Astellas, the National Institutes of Health/National Institute of Neurological Disorders and Stroke, and the National Institute on Aging; these were reported as outside the submitted work where specified.

The paper

Authors. Christian Ovesen, Luis Alberto García Rodríguez, Jesper Hallas, Camilla Storm, Stine Munk Hald, Sören Möller, Magdy Selim, and David Gaist.
Title. Prior Dual Antiplatelet Therapy and Mortality After Intracerebral Hemorrhage: A Danish Nationwide Cohort Study.
Journal. Journal of the American Heart Association.
Year. 2025.
DOI. 10.1161/JAHA.125.043268
PMCID. PMC12684552

Deep Dive — click to expand

What this is

This paper asks whether an ICH that occurs while a patient is receiving two antiplatelet agents carries a worse prognosis than one occurring during single-agent therapy. The headline is an association between DAPT and higher mortality, but the more operational finding is temporal: after adjustment, the signal is strongest on day 1 (weighted RR 1.43) and attenuates by day 180 (1.14), while the cohort has no direct hematoma-volume or expansion data. It is best read as a risk-stratification and mechanism-generating study—not as a reason to withhold indicated DAPT before an ICH occurs, and not as proof that a particular reversal strategy improves survival.

1. Shadow Audit

The paper’s framing can make the day-180 result feel like the main discovery, but the day-180 difference is largely inherited from the first 24 hours. Crude mortality was 24.2% versus 15.3% by day 1, then 47.2% versus 37.0% by day 30, and 52.5% versus 45.6% by day 180. The incident-mortality analysis found evidence concentrated in day 0–1, with no comparable evidence in days 2–30 or 31–180. That pattern says “something about the initial hemorrhage or initial care differs” more strongly than it says “DAPT causes a persistent six-month hazard.”

The second concealed point is that this is not a clean medication comparison. Before weighting, ischemic heart disease was present in 64.9% of DAPT users versus 28.2% of SAPT users, and heart failure in 21.5% versus 9.1%. Those covariates were better balanced after IPTW, but the clinical reasons for choosing DAPT—recent coronary or cerebrovascular events, stents, treatment intensity, and unmeasured disease burden—are not fully encoded in prescriptions and diagnosis codes. The association is credible enough to matter at the bedside, but not clean enough to label causal.

2. Inversion Engine

To erase the headline day-1 signal, the unmeasured confounding would need to account for a weighted RR of 1.43, with a 95% CI of 1.13–1.82. In other words, after the measured covariates are balanced, a remaining difference in disease severity, hematoma anatomy, prehospital delay, or initial treatment would need to produce roughly a 43% relative mortality excess—and at least about a 13% excess at the lower confidence boundary—to explain the point estimate entirely.

The inversion is not implausible. The study did not measure initial hematoma volume, location, intraventricular extension, or hematoma expansion, each of which can dominate ICH prognosis. Nor did it capture actual antiplatelet adherence, dose, acute reversal, or the urgency of the indication for DAPT. Including admission SSS in the propensity model weakened the DAPT association, though day-1 and day-30 evidence remained. A future prospective cohort with serial CT, exact last-dose timing, platelet-function or drug-exposure data, and standardized reversal would change the causal reading more than another administrative adjustment.

3. Second-Order Catalyst

The first protocol to change should be the ICH arrival bundle for patients taking DAPT—not the upstream secondary-prevention protocol that prescribed DAPT. On the first scan and medication reconciliation, the team should treat DAPT exposure as a high-priority variable for rapid neurocritical-care triage, repeat imaging decisions, blood-pressure control, and early discussion with hematology or neurosurgery about reversal and expansion risk. This is a workflow change that can happen immediately because the signal is front-loaded in the first day, not a claim that this cohort proves which reversal product to use.

The subgroup most likely to move first is the patient with DAPT plus concurrent oral anticoagulation: day-180 weighted RR was 1.58 (95% CI 1.30–1.93) in that subgroup, although the day-1 estimate was imprecise (0.74, 0.30–1.79). That discordance is a warning against a simplistic “more drugs equals a fixed acute RR” rule. It supports a medication-stack alert and expedited source-control evaluation, while leaving the actual treatment decision to imaging, timing, indication, and guideline-based reversal pathways.

4. Asymmetric Leverage

The exposed group is small—265 of 5,800 antiplatelet-treated ICH patients—but the possible leverage is large because early death is common and early hematoma dynamics are time-sensitive. A modest improvement in identifying and stabilizing the highest-risk DAPT patients could affect a clinically important number of lives across a large ICH denominator, even if the relative effect is smaller than the point estimate suggests.

The reverse asymmetry matters just as much: a small, selected subgroup can generate a large-looking relative estimate. The day-1 crude RR was 1.58, while the weighted estimate was 1.43; the day-180 weighted estimate was 1.14. The practical payoff is therefore not to count 43% as a transportable treatment effect, but to use DAPT as a trigger for fast, high-fidelity assessment while awaiting evidence that links exposure to modifiable hematoma biology.

5. Paradigm Destroyer

The paper kills the reflex that “antiplatelet exposure is one homogeneous category” after ICH. DAPT is uncommon among antiplatelet-treated ICH patients, but its associated mortality signal is concentrated before the first day is over; waiting for the six-month outcome obscures where the clinical problem is occurring.

Protocol update for tomorrow morning: “At ICH arrival, document the exact antiplatelet and anticoagulant stack, last known doses, and indication before the first handoff. If DAPT is present, prioritize immediate neurocritical-care review and a documented plan for repeat imaging and evidence-based reversal; do not infer causality or delay indicated care from this observational association.”

MVP — Minimum Viable Proof

The minimum statement that would change bedside practice is: among comparable patients with first-ever ICH, exact DAPT exposure independently predicts serial-CT hematoma expansion or early death, and a standardized DAPT-specific management pathway reduces that outcome without unacceptable ischemic or thrombotic harm. This paper supplies the early mortality association—weighted RR 1.43 by day 1—but not the serial-imaging mediator or the intervention effect.

Best Combination

Combine this cohort with the prior literature on antithrombotic-associated ICH in three layers. First, use the present study to identify the timing signal and to separate DAPT from the broader SAPT category. Second, use imaging cohorts and meta-analyses of antiplatelet-associated ICH to test whether DAPT tracks with larger baseline hematoma, more expansion, or worse functional outcome. Third, use randomized evidence on acute platelet-directed or other reversal strategies to decide what to do—not this observational study alone. The strongest synthesis would preserve the paper’s useful distinction between risk after ICH and the separate question of whether DAPT was appropriate before ICH.

Overvalue Warning

Do not convert a weighted RR of 1.43 into “DAPT causes 43% more deaths.” IPTW reduces measured confounding; it cannot balance hematoma anatomy, exact drug exposure, adherence, indication intensity, or acute treatment that was not recorded.
Do not treat the day-180 result as evidence of a durable pharmacologic effect or use it to compare aspirin with clopidogrel definitively. The day-1 excess drives much of the later separation, and the clopidogrel-restricted estimates were imprecise and explicitly hypothesis generating.

Educational use notice. The Clotbust Review is dedicated to medical education. Figures from published articles are embedded for teaching purposes, with attribution to the original publication. We do not claim copyright over any embedded figure; all rights remain with the original publisher and authors under the article's published license. If you are a rights holder and would like a figure removed, please contact the editorial team. See our editorial notice for the full policy.