Effects of recording speed on precision of time-based polycardiographic measurements. Optimal paper speeds for measuring points and intervals.

Optimal paper speeds have not been established for all time-based measurements of the cardiac cycle by appropriately designed observer performance studies. In 10 subjects (5 normals and 5 cardiac patients) carotid pulse, phonocardiogram, and electrocardiogram were recorded on magnetic tape for measurement of all fiducial points for systolic time intervals, the systolic time intervals themselves, the pulse transmission time, cycle length (RR), qR time, and R-to-point versus q-to-point measurements at recording speeds of 25, 50, 75, 100, and 200 mm/s. Tracings were coded numerically and randomised. Three observers measured all points and calculated intervals in a sequence determined by individual tables of random numbers. Left ventricular ejection time was the only calculation that could be made at 25 mm/s statistically equally well as at all other speeds. The smallest numerical observer differences occurred uniformly at 100 mm/s paper speed when all recording speeds were considered. However, after excluding the 25 mm/s speed there were no significant differences among point measurements. Measurements of points from R (rather than q) reduced observer variability. We conclude that for point measurements, for systolic time intervals, pulse transmission time, and RR interval, recording speed between 50 and 200 mm/s showed no statistical differences, though smallest numerical differences occurred at 100 mm/s. For LVET, 25 mm/s was satisfactory.