Errors in gated equilibrium radionuclide measurements of resting left ventricular ejection and filling.

An assessment was made of the reproducibility (between study differences) interobserver variability and intraobserver variability of 7 radionuclide measurements describing both resting left ventricular ejection (ejection fraction--EF, average ejection rate--AVER, peak ejection rate--PER, time to peak ejection rate--TPER) and filling (average filling rate--AVFR, peak filling rate--PFR, time to peak filling rate--TPFR). Gated blood pool studies were performed one week apart in 42 patients and between study correlation coefficients for these measurements ranged from r = 0.58 (TPER) to r = 0.99 (PFR) but there were spontaneous changes in measurements of up to 82% (AVER). Interobserver variability was determined in 44 studies. Correlation coefficients for measurements obtained by two independent observers in 44 studies ranged from r = 0.87 (AVER) to r = 0.96 (TPFR) but spontaneous changes of up to 52% occurred (AVFR). Intraobserver variability was determined in 53 studies and correlation coefficients ranged from r = 0.95 (AVER) to r = 0.99 (AVFR, PFR). The maximum percentage difference between observations was 46% (AVER). Use of correlation coefficients alone to assess the reproducibility of these measurements fails to highlight the marked spontaneous changes which occur in left ventricular function. Changes in radionuclide indices describing left ventricular function in response to a therapeutic intervention must be interpreted in the light of these findings.