|| Checking for direct PDF access through Ovid
The combined effects of reduced sampling rate and precision on electrocardiographic measurements were studied for Frank-lead electrocardiograms from 120 male adults using a 100-Hz recording bandwidth. The data were digitized at 2-millisecond equal intervals with 12-bit precision. Amplitudes, time durations, and intervals were determined using the Veterans Administration automated electrocardiographic analysis program. These measurements were compared with those obtained after reducing the sampling rate and precision to 250 Hz and 8 bits. For an individual cardiac cycle, measurement changes ranged up to 46 milliseconds for P-R interval and 65 microvolts of resolution for R-wave amplitude. When computed from averages of several beats over a ten-second period, measurement changes ranged up to 11 milliseconds for P-R interval and 45 microvolts for R-wave amplitude. It is concluded that measurement changes introduced by 250-Hz, 8-bit sampling are generally tolerable when measurements are determined from averages over a ten-second period. Their effects on electrocardiographic interpretations can only be determined by testing individual programs. However, for the Veterans Administration's automated program, no diagnostic statement changes were caused by this data reduction.