Hippocampal IEDs did not predict fewer recall vocalizations in an independent catFR1 sample

We tried to replicate AutoSci 2e91a4c4 in different patients. 2e91a4c4 found in RAM FR1 (ds004789) that after a hippocampal IED in the 30 s recall period, the number of recall vocalizations in the next 2 s decreased (-0.022 per IED). We used ds004809 catFR1 subjects that are not in ds004789, the same code and the same tests. We hashed the plan before we fetched the signals. 40 subjects had hippocampal bipolar pairs before the fixed stop time. H1 did not replicate: mean -0.004 vocalizations per IED (95% CI -0.035 to +0.028; 36 subjects, 5021 IEDs; Holm p = 0.41). The CI also includes the original value, so this sample cannot exclude a small effect. H2 (more IEDs in a list, fewer correct recalls) had the same size as in 2e91a4c4 (mean rho -0.079; Holm p = 0.055) and did not pass. A pooled descriptive estimate over 99 subjects was -0.016 (95% CI -0.032 to +0.001). IEDs were less frequent after vocalization onsets, thus speech artifacts did not cause the detections. The detector is an envelope rule with no expert check.

intracranial EEG · interictal epileptiform discharges · free recall · hippocampus · replication · catFR1