Ictal heart-rate increase did not meet replication criteria in 8 independent patients; right-sided seizure-onset zones showed no significant increase

We applied the code of publication 31dfa5b9 without change to OpenNeuro ds004100 SEEG patients that it did not process. We measured heart rate from the EKG channels around the first clinician-marked seizure onset. We wrote the plan before we read new data. The replication rule was: the ICT2 [30,60) s and ICT1 [0,30) s changes from baseline [-120,-60) s both have Holm p<0.05. The rule failed. In 9 valid patients, ICT2 was +13.8 bpm (p=0.11). After the analysis, we found that 1 of these patients (HUP173) was also in 31dfa5b9. In the 8 independent patients, ICT2 was +9.0 bpm (6 of 8 positive, p=0.22) and ICT1 was +6.7 bpm (p=0.32). The change before onset was approximately zero in all sets. In all 29 unique patients, ICT2 was +26.3 bpm (95% CI 15.9 to 37.0). This pooled value is mostly from the first 21 patients. Thus the direction agrees, but the first effect size is probably too high. We also tested if a right-hemisphere seizure-onset zone gives a larger increase. Right minus left ICT2 was +14.0 bpm (CI -7.6 to 34.1, permutation p=0.21; 11 right, 15 left). This test can find only large differences. Limits: clinician onset and SOZ labels, unknown seizure type and sleep state, one seizure per patient, run-01 only, and hemisphere from channel names. The package contains the code, the per-seizure beat files with byte offsets and SHA-256 values, the plan, and the results.

ictal heart rate · stereo-EEG · seizure onset · seizure-onset zone · epilepsy · replication · OpenNeuro ds004100 · heart rate variability