Comparison of criteria to truncate gastric emptying tests

Abstract

Standardized scintigraphic gastric emptying (GE) protocols to detect gastroparesis (GP) require collecting data for 4 h. This investigation was undertaken to compare seven methods to reduce the duration of the test. This was a retrospective study of GE data collected using a standardized protocol at 0, 1, 2, 3, and 4 h for 602 patients being evaluated for GP. The reference standard was GP defined conventionally as percentage of gastric retention (GR) at 4 h (p4) of greater than 10%. For data up to 2 h the results were derived as follows: (a) confirming as being positive for GP if GR at 2 h was greater than 65%, negative for GP if GR at 2 h was less than 45%, and indeterminate otherwise; (b) by linear extrapolation; and (c) by monoexponential extrapolation. For data beyond 2 h, further evaluations were made and results were derived as follows: (a) confirming as being positive for GP if GR at 2.5 h was greater than 40%; (b) ascertainment of GR at 3 h; (c) by biphasic fit; and (d) by observation of maximum GR for normal patients at time points earlier than 4 h. Thirty percent of all patients had GP. Eighty percent were determinate by Method 1; for these patients sensitivity to detect GP was similar (P=0.11) for Methods 1-3 (69-79%). For data beyond 2 h, sensitivity of the seven methods ranged from 64 to 92%, and the sensitivity of every method was significantly lower than that of the reference standard (P<0.001). Considering that sensitivity to detect GP was significantly reduced for data collection limited to 3 h or less, it is not advisable to truncate GE studies earlier than 4 h.