26
46 adults on peripheral VA ECMO imaged by TTE at 100%, 120% and 50% of target blood flow on mean day 1 and mean day 4. LV global longitudinal strain improved from −6.1% (IQR −8.9 to −4.0) at 120% flow to −8.8% (−11.5 to −6.0) at 50% flow, p under 0.001. Percentage change in GLS between 50% and 120% flow: 72.7% (26.8–100.0) acute phase versus 22.5% (14.9–43.8) delayed phase, p under 0.001; LVEF 43.0% versus 22.8%, p = 0.012; LVOT VTI 45.8% versus 24.2%, p = 0.001. 24 of 46 (52.2%) weaned successfully. The three standard echocardiographic criteria as a model gave AUROC 0.86 (0.73–0.98) for weanability and 0.68 (0.51–0.85) for eventual decannulation; the SAVE score gave 0.58 and 0.54. Both models well calibrated (Hosmer–Lemeshow p over 0.05).
Why it matters: the direct measurement behind Chapter 26's organising claim that an echocardiographic number on ECMO is a measurement at a flow, and that the swing across flows is the informative quantity. Also the book's clearest example of one model discriminating well for one endpoint and poorly for another in the same patients.
Read as indexed full-text passages. DOI not independently resolved.