Angiogram strategy (with a stent, balloon or bypass if needed) vs medicines alone after a heart attack, for people aged 75 and over

“Our final sample size was 1518, as opposed to the planned 1668”

Kunadian et al., 2024, Discussion. The authors add that fewer primary-outcome events occurred than expected, that the Covid-19 pandemic hampered recruitment of frail patients, and that about 1 in 5 screened patients joined the trial. The trial was open-label, and the authors state that subgroup confidence intervals were not adjusted for multiplicity.

“The early 1-year benefit of the invasive approach was not maintained at 5 years”

This critical appraisal also lists the trial's weaknesses. It was smaller than planned. Only 22% of eligible patients were enrolled. Bypass surgery was used in only 3.3% of patients, so the results may not apply to people who need it. Multivessel PCI was done in only 29.9%, so most patients were incompletely revascularised, which the author says may have affected the findings. The appraisal concludes that the result can be argued both ways, and that treatment should be individualised.

Narayan, 2025, Indian J Thorac Cardiovasc Surg (read in full, PDF supplied). The trial paper itself reports that the proportional-hazards assumption was violated, and that the gain in average event-free time over 5 years was 29 days (95% CI −40 to 98).

Heart attacks that were not fatal were fewer with the angiogram strategy (11.7% vs 15.0%), while deaths from heart causes were numerically more frequent (15.8% vs 14.2%, a difference that could be chance). Combining them gives a neutral overall result. Crossover also narrows the comparison: 24.2% of the medicines-only group later had an angiogram, and 9.7% of the angiogram group never had one. The letters also raised the crossover from medicines to angiogram; the other direction is our reading of the paper's Tables 2 and 3.

The paper's abstract says procedural complications occurred in fewer than 1% of patients. Its Supplementary Table S16 shows that each listed complication type was below 1%, except bleeding caused by the procedure, which affected 13 of 680 people who had an angiogram (1.9%). The table does not report how many people had at least one complication. This is our reading of the appendix, not a published critique.

Three letters to NEJM raised concerns. Reinstadler, Metzler and Reindl note that 3,251 eligible patients (47%) were not enrolled because of clinical decision, and 65% of those with data (2,046 of 3,129) received invasive treatment, which they say

“introduces a substantial selection bias and may result in misleading conclusions”

They add that 1,704 eligible patients (24%) declined, compared with 7% in the FIRE trial, and that a 5-day delay likely penalised the invasive strategy. Nanna, Kochar and Damluji point out that 24% of the conservative group crossed over to angiography, more than half of them were revascularised, and this may dilute any treatment effect; they still see the trial as supporting a conservative strategy for appropriately selected older patients. Komamura, Murohara and Iwase ask what guided clinicians' decisions for patients who were not randomised. In reply, Kunadian says eligible patients who were not randomised had similar characteristics, that long-term outcomes were not collected for them, and that the revascularisation rate resembles other trials (49% in After Eighty, about 60% in RITA-3). The reply adds that guidelines downgraded very early angiography after a meta-analysis found no survival benefit, and that the trial allowed 7 days because of the clinical complexity of older patients.

N Engl J Med, 2025, correspondence and author reply (read in full, PDF supplied). Our check: 3,251 and 1,704 are 47% and 24% of all 6,977 eligible patients (60% and 31% of those not recruited, Supplementary Table S1); the 65% matches Supplementary Table S2. The reply says the oldest patient was 102, while the Supplementary Appendix says 103. The reply's statement that 5.6% of the conservative group had angiography within 7 days could not be matched in the paper or appendix; in the paper, 5.6% is the share of the invasive group with a later angiogram.

In this exploratory analysis of the 469 patients who were frail on the Fried criteria, the primary outcome occurred in 37.7% with the angiogram strategy and 29.4% with medicines only, a difference that could be chance. When frailty was treated as a continuous score, the authors found

“a potential signal for increased risk of harm”

Rubino et al., 2026, JAMA Network Open (read in full via Scite). The authors caution that the frail subgroup may be too small to detect modest differences.

A 2025 meta-analysis of randomised trials in patients aged 70 or older with acute coronary syndrome, including SENIOR-RITA, found no difference in death from any cause, fewer repeat heart attacks and repeat procedures, and more major bleeding with an invasive strategy. Its supplement gives absolute numbers, shown in the fact box above for people with the non-ST-elevation type. Our check: the later-procedure rows of that table (eTable 7) are out by a factor of ten, so the fact box takes 14 in 100 from its forest plot counts (208 of 1,508) and 6 in 100 from its combined result. Most of its MOSCA-FRAIL counts appear to be readmission episodes, not people, including 8 and 8 for later procedures. A published correction amended the P value for major bleeding, which only just excluded no difference. See the Reddy 2025 fact box for details. Its authors conclude that the competing risks of an early invasive strategy

“should be weighed in shared therapeutic decision-making”

Reddy et al., 2025, JAMA Internal Medicine (full text and Supplement 1 read, PDFs supplied).

In a EuroIntervention debate, both sides point to timing. The median time was 3 days from randomisation to angiogram and 5 days from admission, while the debate cites a recommended angiogram within 24 hours. The authors arguing against routine angiograms call this delay

“which is very late and could have impacted the results”

The authors arguing for angiograms (Biscaglia and Campo) add more points. Only about half of patients had a stent or bypass, and they put complete revascularisation at about 30%. Only 22% of eligible patients were randomised (1,518 of 6,977), and 55% of screened patients who were not randomised received invasive treatment, suggesting clinicians may have kept likely beneficiaries out of the trial. They argue that fewer heart attacks and repeat procedures matter as much as survival at this age. The authors arguing against (Cuisset and Cayla) conclude that a routine invasive strategy is probably excessive for many older patients and should be individualised by cognitive function, coexisting conditions and presentation.

Biscaglia, Campo, Cayla and Cuisset, 2024, EuroIntervention (read in full, PDF supplied). Our check: the trial paper reports multivessel PCI in 29.9% of those who had PCI, not a complete-revascularisation rate. The debate's figure that 65% of patients excluded by clinician judgement received invasive treatment is confirmed by Supplementary Table S2 (2,046 of those with data).

An editorial in the European Heart Journal says SENIOR-RITA's results should be read in the context of a

“relatively low revascularization rate in the invasive arm (49.9%)”

It credits the trial for enrolling about one third frail patients, and warns against caution so excessive that frail patients are denied treatment that might help them.

Oliva and Mehran, 2025, European Heart Journal (read in full, PDF supplied). The editorial's main subject is a different, observational study (Roman et al.); its comments on SENIOR-RITA are two sentences.

The NEJM editorial published with the trial argues that the lack of a survival benefit may reflect selection bias. Only 1 in 5 eligible patients was enrolled, and 65% of the eligible patients excluded on the treating physician's judgement (more than 2,000 people) had an invasive procedure, suggesting

“a bias toward an invasive-treatment strategy in those patients”

It adds that earlier studies found a survival benefit in patients with a GRACE risk score above 140, whereas the median in SENIOR-RITA was 135. Angiograms came 5 days after admission. Only 50% of patients were revascularised, although all had a type 1 heart attack. Only 11.5% had pressure-wire (physiology) testing. The editorial also stresses that the trial showed fewer repeat heart attacks and very few complications, and asks whether survival is the right goal at this age rather than quality of life and independence.

Biscaglia, 2024, N Engl J Med (read in full, PDF supplied). Biscaglia also wrote the "for" side of the EuroIntervention debate above. Our check: the trial paper reports a median of 5 days from admission to angiogram; the editorial says mean. The paper's 105 multivessel PCIs out of 351 (29.9%) are out of all patients who had PCI, not, as the editorial puts it, out of patients with multivessel disease. The Supplementary Appendix confirms the median GRACE score of 135 (invasive group; 136 conservative, Table S5), physiology testing in 78 of 680 angiograms (11.5%, Table S9) and the 65% figure (Table S2). Multivessel disease was present in 380 of the 680 patients who had an angiogram (Table S9).

A Bayesian meta-analysis combining SENIOR-RITA with an individual-patient meta-analysis of 6 smaller trials found an 86% probability that the invasive strategy reduced the combined outcome of death or heart attack. It found almost certain benefit for heart attacks and urgent revascularisation, and no difference for stroke. For death from any cause, the authors found

“weak evidence that an invasive strategy may be detrimental”

with a 91% probability favouring the conservative approach. They note that no quality-of-life data were available to weigh against these outcomes.

Siontis and Efthimiou, 2025, EuroIntervention (read in full, PDF supplied).

These could not be read, so any criticism in them is missing here: a letter in Revista Española de Cardiología ("Frailty and invasive management of NSTEACS: lights and shadows"), Supplementary Appendix 1 of Siontis and Efthimiou 2025, and Supplement 2 and the tables of Rubino et al. 2026. The authors of Buendía Santiago 2025 and the details of McDowell 2016, which we did not hold as files, could not be confirmed independently.