Errors in manual measurement of QT intervals.

OBJECTIVE: To quantify the errors associated with manual measurement of QT intervals and to determine the source of the errors. DESIGN: A randomised study of QT measurement by four cardiologists of electrocardiograms plotted on paper in presentations with different noise levels, paper speeds, amplifier gains, and with and without a second QRST complex to indicate the RR interval. SUBJECTS: Four electrocardiograph leads (I, aVR, V1, V5) recorded in eight healthy people relaxing in a semirecumbent position. MAIN OUTCOME MEASURES: Manual measurement of QT interval in 512 electrocardiograms (eight subjects x four leads x eight presentations x two repeats) by each of four cardiologists. RESULTS: QT intervals measured were significantly longer with greater amplifier gain: by 8 ms for a doubling of gain (p < 0.005), equivalent to a doubling of T wave height. QT intervals measured were significantly longer at slower paper speeds: by 11 ms when paper speed was reduced from 100 to 50 mm/s (p < 0.001) and by 16 ms when speed was further reduced from 50 to 25 mm/s (p < 0.001). Neither the presence of noise nor the presence of a second QRST complex altered the mean QT measurements. There were consistent differences in the measurements between cardiologists, amounting to a maximum mean difference of 20 ms. CONCLUSIONS: Manual measurement of QT interval is significantly affected by the paper speed used to plot the electrocardiogram and by electrocardiogram gain, and hence also T wave amplitude. Manual QT measurement also differed consistently with different cardiologists.