Self-Reported Efficacy of Treatments in Cluster Headache: a Systematic Review of Survey Studies

Cluster headache (CH) is a severe trigeminal autonomic cephalalgia characterised by recurrent, unilateral attacks that can occur several times per day and in cluster cycles. The episodic form (ECH) is more common, while chronic CH (CCH) is defined by symptom persistence without remission for over a year. Conventional acute treatments recommended by guidelines include high-flow oxygen and subcutaneous or intranasal triptans; verapamil and lithium are commonly used preventives, with several other agents and neuromodulatory approaches used for refractory cases. At the same time, a substantial minority of patients report using complementary or unconventional pharmacological treatments, many of which lack formal clinical trial evidence. Retrospective patient surveys have been used to capture self-reported treatment effects, but their methodological heterogeneity has limited attempts to synthesise consistent patterns across studies. Santeri and colleagues set out to systematically review retrospective survey studies that quantify self-reported efficacy of two or more treatments for CH, published in English from 2000 to 2020. Their primary aims were to identify treatments consistently reported as most effective across surveys, assess consistency between surveys, and compare survey findings with randomized controlled trial (RCT) evidence. A secondary aim was to evaluate survey design features that could introduce bias into the reported outcomes.

The researchers conducted a PubMed search in May 2020 and again in December 2020 using search terms combining "cluster headache" with survey-related terms (questionnaire, poll, retrospective, interview, etc.). The review was restricted to English-language, retrospective survey studies published between 2000 and 2020 that quantified and compared self-reported efficacy for two or more treatments. The search returned 994 articles; after screening against the inclusion criteria, nine articles were judged eligible for full-text assessment and inclusion. Data extraction was performed independently by two authors (SSR and SD), who read the included surveys repeatedly to identify potential sources of bias and to extract variables for synthesis. Extracted items included survey instrument and administration mode, recruitment channels, methods for verifying CH diagnosis, demographic features (age, sex ratio, ECH/CCH ratio, ethnicity), and the self-reported abortive and prophylactic efficacy of conventional and unconventional treatments. One study (Schindler et al.) supplied additional numerical data to the reviewers that had been only graphically presented in the original article. Because the included surveys used varying efficacy scales (binary, four- or five-level, or three-level), the investigators harmonised outcomes for synthesis by converting available results to a binary effective/not effective format where feasible. One study (de Coo et al.) reported effects on duration and frequency and was excluded from the primary combined efficacy synthesis due to incompatibility. In another study with a three-level scale, the "partially effective" category could not be reliably dichotomised and was therefore presented separately and omitted from the hierarchical clustering to avoid artifacts. For the pooled exploratory analysis, the authors created two combined tables—one for abortive (acute) treatments and one for preventive (prophylactic) treatments. They applied hierarchical clustering using Spearman rank correlations with average linkage to the binary-coded efficacy frequencies across eight surveys, visualising the results as a heatmap and dendrogram using the Heatmapper tool implemented in R. This approach grouped treatments by similarity in self-reported efficacy profiles across surveys.

Nine survey studies met the inclusion criteria, collectively reporting data from 5,419 respondents. Most surveys were web-based (7/9) and used structured questionnaires with closed response options; all used convenience sampling. Participant characteristics reported across studies included mean/median ages mostly between 40 and 50 years, male-to-female ratios ranging from 2.03 to 3.82, predominance of respondents of European descent, and in five surveys a majority (53.8–77%) of current or former smokers. Four of nine surveys verified respondents' CH diagnoses clinically or from records; the remainder relied on self-report. The proportion of CCH patients among respondents varied, often higher than in general CH populations. Across the synthesised surveys (eight studies used for the main binary analyses), there were marked and generally consistent differences in self-reported efficacy between treatments. For abortive (acute) treatment, high self-reported efficacy clustered for oxygen and subcutaneous triptan injections; oxygen and psilocybin appeared in the same clade with triptans in the hierarchical clustering, though oral triptans showed lower reported effectiveness than injectable formulations. For prophylactic effects, serotonergic substances—LSD and psilocybin mushrooms—along with ergoline alkaloids and methysergide clustered with conventional preventive agents such as verapamil and corticosteroids in the higher-efficacy clade. By contrast, melatonin, valproic acid, gabapentin, amitriptyline and propranolol formed a clade characterised by consistent low self-reported efficacy. Psilocybin mushrooms were reported not only as prophylactic but also as having abortive effects comparable to oxygen and injectable triptan in some surveys. Respondents frequently reported durable reductions or even cessation of attacks after single doses of LSD or psilocybin; preventive responses were also reported with small, reportedly non-psychoactive doses. Adverse effects for these serotonergic substances were reported as uncommon and mostly minor in the surveyed samples. The investigators note that the absolute number of respondents reporting use of psychedelics was small relative to users of conventional treatments. When comparing ECH and CCH subgroups, statistically significant better abortive responses were reported for oxygen and triptans in ECH versus CCH in the four surveys that assessed this.

Santeri and colleagues interpret the concordance of results across multiple retrospective surveys as evidence that such surveys, despite inherent limitations, can yield useful and externally consistent information about patient-perceived treatment effects in CH. The observed rank order of self-reported efficacy—oxygen and triptans among the most effective abortive treatments and verapamil and corticosteroids among conventional preventives—aligns broadly with clinical trial data and clinical practice patterns. Injectable sumatriptan was consistently reported as more effective than oral or intranasal triptans, mirroring RCT findings. A notable and novel finding highlighted by the authors is the consistently high self-reported prophylactic efficacy of 5‑HT2A agonists, specifically psilocybin mushrooms and LSD. These serotonergic agents clustered with highly efficacious treatments and showed the highest reported prophylactic response frequencies (around 75% in pooled survey frequencies). The authors note mechanistic plausibility—central 5‑HT2A receptor activity and expression in hypothalamic regions implicated in CH—and cite a small exploratory controlled study that found prolonged migraine reduction after a low psilocybin dose. Nevertheless, they emphasise that the absolute numbers of respondents using these substances were small relative to conventional treatments. The paper acknowledges multiple sources of bias across the included surveys: retrospective design, convenience sampling, heavy reliance on patient support groups for recruitment, incomplete verification of diagnoses in most studies, potential duplicate respondents across surveys, and heterogeneous outcome scales complicating pooled inference. These limitations affect representativeness (for example, a higher proportion of CCH patients in surveys) and the comparability of effect measures. The authors also stress methodological gaps that surveys cannot address, such as dose–response relationships, effects of combined treatments, and direct head-to-head comparisons. Despite these limitations, the investigators argue that combining diverse survey datasets can strengthen confidence in observed patterns when results are consistent across settings and methods. They recommend that future survey research employ random sampling, diagnostic verification, standardised efficacy and adverse-effect measures, direct comparative questions, and prospective designs. Clinically, the authors suggest that awareness of widespread self-treatment with unconventional agents should prompt clinicians to discuss potential harms, benefits and drug interactions with patients. Finally, they conclude that the consistent self-reported prophylactic effects of psilocybin and LSD strengthen the rationale for prospective randomized clinical trials of 5‑HT2A agonists in CH.

In this first systematic review of retrospective survey studies on CH treatment efficacy, the authors conclude that surveys capture consistent patient-reported differences in treatment effectiveness that generally align with clinical evidence. Most notably, psilocybin mushrooms and LSD were repeatedly reported as effective prophylactic agents across surveys, a finding that the authors view as sufficient rationale to prioritise formal clinical trials of 5‑HT2A agonists in cluster headache. They recommend that future research address the methodological shortcomings identified in survey studies to better inform clinical and policy decisions.