Intra- and inter-individual attack frequency variability of chronic cluster headache

Abstract text (incl. figure legends and references)

Question: Little is known regarding the AF variability of CCH, hampering power and sample size calculations, and consensus on the most optimal duration of pre-trial baseline observation periods.

Methods: We used detailed data from the 12-week baseline period of the randomized controlled occipital nerve stimulation ICON trial in patients with medically intractable CCH. Participants were post hoc divided into four mean daily AF groups: <2; >2-3; >3-4; >4. We analyzed the following four variables for the total and four AF groups: (i) weekly vs. instantaneous recording of the AF; (ii) intra-individual AF variability by using (a) the mean absolute deviation from the mean and (b) the coefficient of variation; (iii) seasonal variability of the AF ; (iv) the smallest number of weeks to obtain a reliable estimate of the baseline AF over the entire 12-week period.

Results: Weekly median (14.4 [8.2 – 24.0]) and instantaneous (14.2 [8.0 – 24.5]) AF recordings were similar (p=0.20; Bland-Altman plot). The median weekly AF over all 12 weeks was 15.3 (range 4.2-140). Absolute AF variation was lower in the lowest AF group in comparison to the other AF groups (p<0.001). Relative AF variability decreased with increasing AF (p=0.010). During spring AF was higher compared to the other seasons (p=0.001). We tabulated the weekly AF estimation accuracies compared to, and the associated deviations from the 12-week gold standard for different lengths of the observation period.

Conclusion: Weekly retrospective recording of the AF is as good as instantaneous recording and more convenient. Participants with ≥3 daily attacks show less AF variability than those with <3 daily attacks. Mean AF is highest in spring. The data suggest that an optimal balance between feasibility and an accuracy of 90% with a deviation of no more than 20% is achieved at an observation period of 7 weeks.