The number in the protocol arrived
ASPIRED analysed 2,233 adults seen in the emergency departments of 45 UK hospitals after unexplained fainting. Over one year, self-reported fainting averaged 1.37 episodes with a 14-day wearable ECG and 1.58 with standard care; the incidence rate ratio was 0.89, the 95 per cent confidence interval 0.68 to 1.18 and p=0.43. That interval covers both a reduction and an increase.[1]
The way to read that line runs through the trial's protocol, published in BMJ Open in 2023. The protocol declared the primary endpoint to be the number of self-reported fainting episodes at one year, and wrote down a target sample of 2,234 participants, 90 per cent power, a two-sided alpha of 5 per cent and an expected effect of an incidence rate ratio of 0.85. The trial landed one person short of that target and did not find the effect it was looking for. This is an answer taken on the design's own terms.[1]
Which column the headline comes from
The headline the trial was announced under rests on its secondary results. In the monitored arm, detection of clinically significant rhythm disorders rose from 9 per cent to 22 per cent, median time to diagnosis fell from 55 days to 22 days, pacemaker implantation went from 4.6 per cent to 6.8 per cent and anti-arrhythmic drug initiation from 7.3 per cent to 10.8 per cent. One-year all-cause mortality was 2.9 per cent with standard care and 1.5 per cent with monitoring.[1]
In ASPIRED mortality was a secondary endpoint and the protocol listed twelve secondary objectives. A single difference emerging from that many comparisons does not carry the same weight on its own; when the primary endpoint does not move, secondary results generate hypotheses. Matthew Reed presented the survival difference as an observation for further research to explain. The other reading is available too: catching a rhythm disorder earlier may genuinely have led to earlier treatment and fewer deaths.[1]
Two trials whose events never arrived
Two more trials in the Hot Line 11 session at the same congress found no difference, and in both the events arrived in smaller numbers than expected. EVAOLD compared stress imaging-guided selective invasive management with routine invasive management in patients aged 80 and over; after 587 of a planned 1,756 patients had been randomised it was stopped at the first interim analysis for futility, because the conditional power to show noninferiority was low. Death, non-fatal myocardial infarction or non-fatal stroke at one year occurred in 24.1 per cent of the imaging-guided arm and 20.7 per cent of the routine arm (hazard ratio 1.22; 95 per cent confidence interval 0.86 to 1.72; p=0.27). CorCal Outcomes randomised 5,772 people; over 4.2 years major cardiovascular events stayed at 2.7 per cent in both arms and the noninferiority criterion was not met (hazard ratio 0.99; 95 per cent confidence interval 0.71 to 1.38; p for noninferiority 0.045).[2], [3]
None of the three separated on its primary endpoint. What creates the difference between them is how many events stand behind that result. Because ASPIRED reached the target of 2,234 patients and the 90 per cent power it declared in its protocol, it narrowed the hypothesis that monitoring cuts fainting; EVAOLD ended at 587 patients and CorCal at an event rate of 2.7 per cent, so their nulls leave the question open. Joseph B Muhlestein's own conclusion runs the same way: he treated the data as hypothesis-generating input for designing a better-powered trial. The signal to watch is where mortality sits among the prespecified secondary endpoints in ASPIRED's full report.[1], [2], [3]