Observer bias affects the results of cardiac studies with the computerized nuclear probe

To assess the reliability of a nonimaging computerized nuclear probe in the evaluation of global left ventricular (LV) function, the authors compared LV ejection fraction (LVEF) and LV filling rate (LVFR) calculated from gamma camera-derived and probe-derived time activity curves. 40 consecutive patients with a variety of cardiac diseases were assigned to Group 1 (N=21) or Group 2 (N=19). In Group 1, where the operator knew results from camera studies, LVEF and LVFR by probe correlated well with camera data (R=0.96 and R=0.93, respectively, both rho<0.01), whereas in Group II where camera results were not yet known, the correlation of probe and camera data was less satisfactory (R=0.76, rho<0.05 for LVEF, and R=0.23, NS, for LVFR). This difference was significant (rho<0.01). Thus, prior knowledge of LV function introduces a bias which improves the concordance of nuclear probe and gamma camera studies, and the reliability of LVEF and LVFR from the nuclear probe may be reduced without such prior information