If you're building anything that flags a change in a patient's walking speed, there's a number you have to put somewhere in the code: how big a change counts.
For usual gait speed, the 2024 Age and Ageing review I cited in my latest article reports a minimal important change anywhere from 0.03 to 0.12 meters per second. That's a fourfold range.
Plug in 0.05 m/s and see what you get. Against the low end of that range it's a real change. Against the high end it's well inside the noise. A patient whose speed moved by 0.05 may have improved, or may have just had a good morning.
So whatever threshold ends up in the config is really a choice about which study you trust, whether or not anybody wrote that down. I'd write it down, next to the value, so the next person who touches the rule knows where it came from.
The evidence underneath is thin, too. The review found only three articles on responsiveness at all. It rated usual speed sufficient and fast speed insufficient, both on low-quality evidence.
My own read is that lots of readings over time could narrow that gap compared with a single before-and-after. That's a guess about where research could go. Nobody has shown it.
Responsiveness is one of five questions I walk through in this piece on what separates a movement measurement from a real biomarker. Reliability, validity, reference ranges and whether anyone acts on the number are the other four.
Disclosure: I'm VP of Operations at BioMech Health, which works in movement measurement.
Top comments (0)