BackgroundWe demonstrated in the randomised controlled ICON study that 48-week treatment of medically intractable chronic cluster headache (MICCH) with occipital nerve stimulation (ONS) is safe and... Show moreBackgroundWe demonstrated in the randomised controlled ICON study that 48-week treatment of medically intractable chronic cluster headache (MICCH) with occipital nerve stimulation (ONS) is safe and effective. In L-ICON we prospectively evaluate its long-term effectiveness and safety.MethodsICON participants were enrolled in L-ICON immediately after completing ICON. Therefore, earlier ICON participants could be followed longer than later ones. L-ICON inclusion was stopped after the last ICON participant was enrolled in L-ICON and followed for ≥2 years by completing six-monthly questionnaires on attack frequency, side effects, subjective improvement and whether they would recommend ONS to others. Primary outcome was the change in mean weekly attack frequency 2 years after completion of the ICON study compared to baseline. Missing values for log-transformed attack-frequency were imputed for up to 5 years of follow-up. Descriptive analyses are presented as (pooled) geometric or arithmetic means and 95% confidence intervals.FindingsOf 103 eligible participants, 88 (85%) gave informed consent and 73 (83%) were followed for ≥2 year, 61 (69%) ≥ 3 year, 33 (38%) ≥ 5 years and 3 (3%) ≥ 8.5 years. Mean (±SD) follow-up was 4.2 ± 2.2 years for a total of 370 person years (84% of potentially 442 years). The pooled geometric mean (95% CI) weekly attack frequency remained considerably lower after one (4.2; 2.8–6.3), two (5.1; 3.5–7.6) and five years (4.1; 3.0–5.5) compared to baseline (16.2; 14.4–18.3). Of the 49/88 (56%) ICON ≥50% responders, 35/49 (71%) retained this response and 15/39 (38%) ICON non-responders still became a ≥50% responder for at least half the follow-up period. Most participants (69/88; 78% [0.68–0.86]) reported a subjective improvement from baseline at last follow-up and 70/88 (81% [0.70–0.87]) would recommend ONS to others. Hardware-related surgery was required in 44/88 (50%) participants in 112/122 (92%) events (0.35 person-year−1 [0.28–0.41]). We didn't find predictive factors for effectiveness. Show less
