Sweden switched whole regions to a different heart-attack drug, and events did not move

Two drugs, prasugrel and ticagrelor, sit in the same slot of treatment after a heart attack: a P2Y12 inhibitor added to aspirin after a stent. Which one should be preferred has been uncertain for years [s1]. SWITCH-SWEDEHEART, published in the New England Journal of Medicine on 29 August and presented at ESC Congress 2026, tried to settle it by randomising hospitals rather than patients — and the answer it returns is as much about that design as about the drugs.

A trial of a policy, not a prescription

SWITCH-SWEDEHEART was a registry-based, open-label, stepped-wedge, cluster-randomised trial in Sweden involving adults who had undergone percutaneous coronary intervention for an acute coronary syndrome [s1]. Seven regions, grouped into three clusters, switched from ticagrelor to prasugrel as the default P2Y12 inhibitor in a randomised sequence across four nine-month periods between 2021 and 2024 [s1].

Nobody consented to a pill. A region changed which drug it reached for first, and every eligible patient treated there during that window became part of the comparison, with outcomes read out of national registries [s1]. The primary end point was a composite of death from any cause, fatal or non-fatal myocardial infarction, or fatal or non-fatal stroke through one year [s1]. The intention-to-treat analysis used a mixed model adjusted for cluster and calendar time to produce what the investigators call a policy-level estimate [s1].

The New England Journal published an accompanying editorial specifically on this class of design, titled "The Emerging Role of Treatment Policy Implementation Trials in Acute Coronary Syndrome" [s2] — a signal that the method, not only the finding, is what the field is being asked to absorb.

Who was enrolled

A total of 17,095 patients were included: 9,444 treated under the default ticagrelor policy and 7,651 under the default prasugrel policy [s1].

The population is the point. Mean age was 69.8 years, 36.8% were at least 75 years old, and 27.8% were female [s1]. Presentations split 39.4% ST-segment elevation myocardial infarction, 43.7% non-STEMI and 16.8% unstable angina [s1]. That is closer to who actually arrives at a Swedish catheter laboratory than the filtered population of a conventional randomised trial.

What happened

At one year, a primary end-point event had occurred in 11.8% of patients under the ticagrelor policy and 11.1% under the prasugrel policy, an adjusted odds ratio for prasugrel versus ticagrelor of 0.90 (95% CI 0.77 to 1.06) [s1].

The interval crosses one. On ischaemic events, this is a null result, and the authors state it as such: the risk of ischaemic events was not lower with a default prasugrel policy than with a default ticagrelor policy [s1].

Major bleeding occurred in 4.4% under ticagrelor and 4.2% under prasugrel, an adjusted odds ratio of 0.80 (95% CI 0.64 to 0.99) [s1]. That interval sits just below one. It is a secondary comparison in a trial whose primary end point was null, which is exactly the situation where a nominally significant result should be read as hypothesis-generating rather than as an established difference.

What a policy trial can and cannot tell you

The strength of the design is generalisability. Because the unit of randomisation was a region and the outcomes came from national registries, the estimate describes what happens when a health system changes its default — including the patients who would have been excluded from a conventional trial, the ones who switch drugs mid-course, and the ones who stop taking either.

The corresponding weakness is that it dilutes. A policy-level estimate is not a drug-level estimate. If adherence to the default was incomplete in either direction — and in a system-wide switch it will not have been perfect — the comparison of prasugrel against ticagrelor in the individual patients who actually took each drug would be expected to look different from the comparison of one default against the other.

The trial was also open-label, ran across four periods between 2021 and 2024, and adjusted for calendar time precisely because background care changes over three years [s1]. Stepped-wedge designs are vulnerable to secular trends in a way parallel-group trials are not; the adjustment is the standard remedy, not a guarantee.

The trial was funded by Region Västra Götaland and registered as NCT05183178 [s1].

What to watch

Two questions follow. First, whether the bleeding difference — an adjusted odds ratio of 0.80 with an upper bound of 0.99 — is reproduced anywhere else, because on its own it is not enough to move a guideline [s1]. Second, whether other health systems adopt this design. A registry-based stepped-wedge trial answers a question no individually randomised trial can answer, which is what happens when a whole service changes its habit, and the editorial's framing suggests the journal expects more of them [s2].

This article describes trial results. It is not advice about any medicine, and nothing here should be used to start, stop or change treatment.

Sources

• Prasugrel versus Ticagrelor in Acute Coronary Syndromes, New England Journal of Medicine, 29 August 2026
• The Emerging Role of Treatment Policy Implementation Trials in Acute Coronary Syndrome, New England Journal of Medicine, 29 August 2026