TrialsgrambetaRCTs are beautiful

Adedinsewo AI-ECG (peripartum cardiomyopathy)

2024

Artificial Intelligence Guided Screening for Cardiomyopathies in an Obstetric Population: a Pragmatic Randomized Clinical Trial

In 1,232 pregnant and postpartum women in Nigeria, AI-guided screening with a digital stethoscope and AI-ECG improved identification of left ventricular systolic dysfunction compared with usual care.

Result

Efficacy: Identification of left ventricular systolic dysfunction during the study periodOR 2.12 (95% CI 1.05-4.27) · P = 0.032

Below 1 favours usual care, above 1 favours AI-guided screening with a digital stethoscope and AI-ECG.

Event rates
Absolute difference
2.1 points
NNT
48

Absolute difference and NNT are derived from the event rates.

Also reported: 12-lead AI-ECG: 3.4% (20/587) vs 2.0% (OR 1.75, 95% CI 0.85–3.62; P=0.125). 1,195 of 1,232 randomised completed the baseline visit; no serious adverse events.

Original publication
Subtopic
Heart failure & cardiomyopathy
Design
Open-label pragmatic RCT, Nigeria (6 hospitals)
Relevance
Important: one of several pillars.
Source
Adedinsewo DA, et al. Nat Med. 2024;30(10):2897-2906.doi 10.1038/s41591-024-03243-9

Discussion & critique

The first randomised trial of AI screening in pregnancy, run in the country with the highest reported incidence of peripartum cardiomyopathy and a rare AI trial in a low-resource setting. Only the digital-stethoscope model reached significance; the 12-lead AI-ECG did not. The endpoint is detection rather than maternal outcome, the trial was open-label, and the intervention arm had a baseline confirmatory echocardiogram that the control arm did not, which favours detection in that arm.