Greater occipital nerve block versus intranasal sphenopalatine ganglion block in chronic migraine with failure of at least two oral preventive treatments: a prospective, non-randomised patient-preference comparative study

Background: Prospective comparative evidence for greater occipital nerve (GON) block and intranasal sphenopalatine ganglion (SPG) block is limited. This study describes the six-month course of two treatment cohorts of patients with chronic migraine in whom at least two oral preventive treatments had failed. Objective: To describe changes in migraine attack frequency, pain intensity, acute medication intake and associated symptoms after distal landmark-guided GON block or topical intranasal SPG block, and to examine whether the trajectories of the two cohorts differed. Methods: In this single-centre prospective observational study, treatment followed the patient’s acceptance or refusal of an injection; allocation was not randomised. Of 100 enrolled patients, 78 had complete baseline and month 1, 3 and 6 data (GONB group n = 45; SPG block group n = 33). Each procedure used 3 mL of 0.5% bupivacaine weekly for four weeks. Attack frequency, Numerical Rating Scale (NRS) pain intensity and triptan and NSAID intake days were analysed with linear mixed-effects models containing group, categorical time and their interaction, with Holm adjustment across the four outcomes and covariate-adjusted sensitivity models. Associated symptoms were self-reported and analysed only in patients with the symptom at baseline. Results: Both cohorts showed lower attack frequency, pain intensity and acute medication intake throughout follow-up (Friedman p < 0.001 in both groups). At month six, model-estimated attack frequency was 3.69/month (95% CI 3.40–3.98) in the GONB group and 4.24/month (3.90–4.58) in the SPG block group, and the NRS score was 4.56 (4.26–4.85) versus 5.48 (5.14–5.83). In the primary unadjusted models the group-by-time interaction was significant for attack frequency (p = 0.034) and NRS (p = 0.007) and not for triptan (p = 0.057) or NSAID days (p = 0.298); after Holm adjustment across the four outcomes only NRS remained significant (p = 0.030), and in the covariate-by-time sensitivity models no interaction remained significant after adjustment. A ≥ 50% reduction in attack frequency occurred in 40/45 (88.9%) and 28/33 (84.8%) patients (risk difference 4.0% points, 95% CI − 11.0 to 21.0). Baseline stabbing (p = 0.019) and electric shock-like pain (p = 0.009) were more frequent in the SPG block group. Reported adverse events were mild dizziness in three GONB patients and brief epistaxis in one SPG block patient. Conclusion: Both patient-preference cohorts showed within-group improvement over six months. Preference-based allocation, baseline phenotype differences, incomplete follow-up and the sensitivity of the between-group comparisons to multiplicity and covariate adjustment prevent conclusions about efficacy or superiority. Randomised, controlled studies using standardised headache-day endpoints are required.