A tie assembled from two opposite movements

In A-CLOSE the combined endpoint came out at 5.0 per cent against 5.1 per cent, the ischaemic component at 3.7 per cent against 1.6 per cent and the bleeding component at 1.8 per cent against 4.1 per cent. Inside that same total sits all-cause mortality at 1.3 per cent against 0.4 per cent. Because the three figures are gathered under one endpoint, the equality in the headline puts two separate risks the patient faces on one scale.[1]

An endpoint that adds death, myocardial infarction, stent thrombosis, stroke and bleeding into a single total can present two components moving in opposite directions as a tie. A second reading is available: the two components may genuinely balance for the average patient, and the total may reflect that balance correctly. Which reading holds depends on whether the components cost the patient the same, and the principal investigator himself describes the result as a clear trade-off.[1]

A small difference and a difference nobody could see

In TARGET-CTCA the rate after three years was 7.1 per cent in the scan arm and 7.3 per cent in the standard-care arm; the trial took 3,170 people across 14 emergency departments, and entry required a troponin level above 5 nanograms per litre with a heart attack already excluded. A defined patient group, close to four years of follow-up and an outcome measure fixed in advance stand together here.[2]

The trial's published protocol shows how that difference was looked for: the primary outcome measure is defined as myocardial infarction or cardiac death, the study was sized with 90 per cent power to detect a 40 per cent relative risk reduction, and 97 events were anticipated in the standard-care arm. The 7.1 per cent against 7.3 per cent that came out means a difference of that size was looked for and not found.[2]

In EMAIL-HF initiation within six months was 18.6 per cent against 8.2 per cent, while hospitalisation or death from any cause came out at 13.0 per cent against 14.0 per cent and did not reach significance. The trial lead says that only a proportion of patients responded to the invitation, that the difference in drug use between the groups therefore stayed limited, and that a much larger study would have been needed to detect the difference in clinical outcomes reliably.[3]

In both trials the gap between groups stayed small. The distinction lies in how each gap was looked for: one measured an outcome fixed in advance by following 3,170 people for three years, while in the other the clinical outcome stayed a secondary endpoint and the trial lead noted that a much larger study would have been needed to see it. Two figures of similar size therefore carry different information.[2], [3]

The line to read before the headline

The first line to read in a trial is which events were gathered into a single endpoint. The A-CLOSE total counts ischaemic events and bleeding together; the TARGET-CTCA measure covers heart attack and cardiac death and returns 7.1 per cent in the scan arm against 7.3 per cent in the standard-care arm. Two sentences built the same way do not imply the same composition underneath.[1], [2]