Clinical characteristics and treatment patterns of patients with episodic cluster headache: results from the United States, United Kingdom and Germany

Abstract

Objective

To describe clinical characteristics and regional treatment patterns of episodic cluster headache (CH).

Methods

A point-in-time survey of physicians and their patients with CH was conducted in the United States, United Kingdom and Germany in 2017.

Results

Overall, 1012 patients with episodic CH were analyzed. Demographic and clinical findings were generally consistent across regions. Most patients were men (66.6%) and the mean age was 40.9 years. The greatest proportion of patients (38.3%) had ≤1 attack per day. The mean number of attacks per day (APD) was 2.4 and mean number of cluster periods per year was 2.6; the mean cluster period duration was 30.8 days. Most patients (69.3%) did not report a specific or predicable time when cluster periods occurred. Acute treatment was prescribed for 47.6% of patients, 10.3% of patients received preventive treatment, and 37.9% of patients received combined acute and preventive treatment; 4.2% of patients were not receiving treatment. Frequently prescribed acute treatments were sumatriptan, oxygen, and zolmitriptan; oxygen use varied considerably across countries and was prescribed least often in the United States. Frequently prescribed preventive treatments were verapamil, topiramate, and lithium. Lack of efficacy and tolerability were the most common reasons for discontinuing preventive treatment.

Conclusions

We observed high use of acute treatments, but only half of patients used preventive treatments despite experiencing several cluster periods per year with multiple cluster APD. Further studies about the need for and benefits of preventive treatment for episodic CH are warranted.

PLAIN LANGUAGE SUMMARY

People with cluster headache (CH) experience headache attacks of excruciating stabbing pain, usually on one side of the head around the eye. These headache attacks typically last between 15 min and 3 h, and come in clusters (or bouts) occurring up to several times a day for a few weeks or months at a time. This greatly impacts a patient’s quality of life.

We surveyed doctors and their patients across the United States, the United Kingdom and Germany, looking at symptoms that occurred during CH attacks, how long the headache attacks lasted, how often the patient had them, and what medicines were being given.

Our results showed that patients with CH suffered from clusters (bouts) of headache attacks several times a year. Nearly, a third of patients had a wrong diagnosis before being diagnosed with CH. Patients experienced stress, agitation, restlessness, difficulty relaxing and depression during a headache attack, especially those who had more CH attacks each day.

Although many patients were taking medication, only half of patients were prescribed medicines to prevent their headache attack from starting. Side effects and the medicines not working were the most common reasons patients stopped taking medicine to prevent their headache attacks. The differences seen in medicines prescribed between countries suggest differences in guidance, or in doctors’ awareness of current medication guidelines. Further studies about the need for and benefits of medicines to prevent CH attacks are needed.

Keywords:

• Episodic cluster headache
• treatment patterns
• acute treatment
• preventive treatment
• efficacy
• tolerability

Introduction

Cluster headache (CH) is a primary headache disorder that belongs to a group of headache disorders known as trigeminal autonomic cephalalgias^1,2. Patients with CH experience severe unilateral or periorbital pain lasting 15–180 min, which occur with a frequency ranging from once every other day to eight times a day². Pain is accompanied with symptoms such as conjunctival injection, lacrimation, eyelid edema, nasal congestion, rhinorrhea, forehead and facial perspiration, miosis, or ptosis². Patients with CH may also experience a sense of restlessness or agitation².

CH is classified as episodic (attacks occur in periods of seven days to a year with remission periods ≥3 months) or chronic (attacks occur ≥1 year without remission period or remission periods <3 months)². According to the International Classification of Headache Disorders (ICHD)-III diagnostic guidelines, patients with episodic CH typically have one to two active cluster periods per year²; cluster periods last on average approximately 8 weeks³. Episodic CH is approximately six times more common than chronic CH⁴.

CH has a lifetime prevalence of just over one in 1000 and the 1-year prevalence of CH may be as high as 53 per 100,000 adults⁴. The usual age of CH onset is between 20 and 40 years². It has been previously reported that for episodic CH, men are affected approximately four times more often than women⁴.

CH has a significant impact on both society and patients; for patients, CH is considered one of the most intensely painful human conditions⁵. A Danish study that evaluated the socioeconomic impact of CH revealed reduced participation in social activities and increased absenteeism and use of healthcare resources among patients with CH compared with a population-based survey of subjects with non-CH types⁶. Studies using the 36-Item Short Form Health Survey revealed considerable quality-of-life impairment in patients with CH^7,8; one study revealed that patients with CH had significantly more bodily pain and lower social functioning than patients with migraine⁷. Additional studies undertaken in Sweden indicate that patients with CH have a greater degree of morbidity and sickness absence than the normal population, with female patients reporting a higher level of disease burden^9–11.

The mainstay of CH treatment is pharmacotherapy¹². Currently available pharmacotherapies consist of acute treatments (taken at the onset of each individual attack to abort the attack) or preventive treatments (for prevention of attacks and reduction of attack frequency)¹³. Subcutaneous sumatriptan, intranasal zolmitriptan, and oxygen are established as effective acute therapies for CH¹³. Verapamil is largely considered the first-line preventive treatment across many countries¹⁴. However, according to the American Headache Society, only suboccipital steroid injections have been established as effective for preventive treatment; lithium, verapamil, and nasal civamide are also preventive treatment options but have less supporting evidence¹³. Guidelines from the European Federation of Neurological Societies recommend oxygen and subcutaneous or nasal sumatriptan for acute treatment, and verapamil and steroids for preventive treatment¹⁵. Until recently, the available preventive therapies were not specifically developed for the treatment of CH and were used off-label¹. For example, the recommended dose of verapamil for treating CH is above the recommended dose in other indications¹⁴. Galcanezumab-gnlm is a humanized monoclonal antibody that binds to and inhibits calcitonin gene-related peptide, a neuropeptide associated with CH attacks¹⁶, and was approved in 2019 by the United States Food and Drug Administration as a treatment for episodic CH¹⁷. Other non-pharmacologic treatment options (either newly available or under investigation) include occipital nerve, vagal nerve, and sphenopalatine ganglion stimulation¹. A retrospective analysis of insurance claims of 7589 patients with CH from the United States (USA) revealed that only one third of patients were prescribed recognized CH treatments¹⁸.

Real-world data on clinical characteristics of patients with CH (including symptoms, pain severity, and attack duration and frequency) and on treatment practices across regions are limited. We therefore performed a point-in-time survey of physicians and their consulting patients with CH in the USA, United Kingdom (UK), and Germany. Our aim was to gain an improved understanding of the real-world clinical features and treatment patterns of patients with CH to facilitate addressing the unmet needs of this population.

Methods

Study background

Data were extracted from the Adelphi CH Disease Specific Programme (DSP) conducted in the USA, UK, and Germany from June to December 2017. DSPs are large, multinational surveys conducted in clinical practice that describe current disease management, disease-burden impact, and associated treatment patterns (both clinical and physician-perceived). DSPs are point-in-time surveys of physicians and their patients presenting in real-world clinical practice. The DSP methodology has been published and validated^19–21.

Participating physicians and patients

Participating physicians were primary care physicians (PCPs) or neurologists who consulted ≥4 patients with CH in a typical month, and were actively involved in the treatment and management of patients with CH. Physicians were identified and recruited through networks of field-based interviewers, using public and internal databases, and their eligibility was confirmed based on criteria described above. Physicians were compensated for participating in the DSP according to fair market research rates consistent with the time involved.

Participating patients were aged ≥18 years and had a diagnosis of CH. For this study, participating patients were considered to have episodic CH (attacks occurring in periods lasting from 7 days to 1 year separated by pain-free periods lasting ≥1 month) or chronic CH (attacks occur ≥1 year without remission periods or remission periods <1 month) based on ICHD-3 beta guidelines, which were the most recent guidelines at the time this study was conducted²².

Data collection

Participating physicians completed a patient record form (PRF) for the next 10 consecutively consulting patients with a CH diagnosis. Completion of the PRFs was undertaken through consultation of existing patient clinical records, as well as the judgment and diagnostic skills of the physician. This represents the evidence physicians had when making clinical decisions, consistent with those made by physicians in routine clinical practice. PRFs contain questions on patient demographics, diagnosis, management, clinical status, concomitant conditions, current treatment, and treatment history. Concomitant conditions may have been diagnosed by any physician (i.e. not necessarily only those participating in the survey) and were selected from a pre-coded list. For patients currently in an active cluster period, physicians provided their current treatment regimen; for patients not in an active cluster period, the patient’s most recent treatment regimen during their most recent cluster period was recorded. All previously known treatments up until the most recent treatment were also provided. Physicians were provided with a precoded list of the most commonly prescribed acute and preventive treatments to select from for ease of completion, but there was also space for physicians to write treatments not included in the precoded list.

For assessment of when patients took their preventive treatment in relation to cluster periods, patients were categorized as follows: only during a cluster period; before and during a cluster period; during and for a specified time after a cluster period; during and continuous use after a cluster period; before, during and for a specified time after a cluster period; and before, during, and continuous use after a cluster period.

Each patient for whom the physician completed a PRF was invited to complete a voluntary Patient Self-Completion (PSC) questionnaire. PSCs contain questions on demographics and current condition. As not all patients were in an active cluster period at the point of data collection, patient-reported symptoms, feelings, and pain severity during attacks were based on what the patient recalled experiencing in a typical attack. PSCs were completed by the patient independently of the physician immediately after consultation and were returned in a sealed envelope to ensure confidentiality. Patients with episodic CH were segmented into those experiencing ≤1, >1 to ≤2, >2 to ≤3, >3, and an unknown number of attacks per day (APD), on average, and were analyzed descriptively.

Missing data were not imputed; therefore, the base of patients for analysis may vary from variable to variable and were reported separately for each analysis.

Ethics

Using a check box, physicians and patients provided informed consent to participate in the survey. Data were collected in such a way that patients and physicians could not be identified directly. DSP data collection was undertaken in line with European Pharmaceutical Market Research Association Code of Conduct²³ and as such it did not require ethics committee approval. Each survey was performed in full accordance with relevant legislation at the time of data collection, including the Health Information Technology for Economic and Clinical Health Act²⁴ and the Health Insurance Portability and Accountability Act²⁵. Adelphi Real World routinely establishes a Master Services Agreement with each fieldwork partner, which stipulates agencies must follow local guidelines in addition to adhering to the European Pharmaceutical Market Research Association Code of Conduct²³. Ethical approval was granted by the Freiburger Ethik Kommission International, which is an official independent body that reviewed the DSP methodology and questionnaires offering international coverage.

Statistical analysis

Analysis was conducted in Stata 16.1 (StataCorp, 2019, Stata Statistical Software: Release 16, StataCorp LLC, College Station, TX). Descriptive statistics are presented as numerical (count, mean, and standard deviation (SD)) or categorical (count and percentage of patients falling into each response category). Comparisons were made between APD categories (≤1, >1 to ≤2, >2 to ≤3, >3, unknown number) using one-way analysis of variance for continuous variables, χ² tests (two-tailed) for unordered categorical variables and Kruskal–Wallis tests (two-tailed) for ordered categorical variables. p Values <.05 were considered statistically significant.

No statistical power calculation was conducted prior to the study. As the data were collected before the study objectives were determined, the sample size was based on available data.

Results

A total of 142 physicians (41 PCPs, 101 neurologists) completed PRFs for 1344 patients, of which 763 provided PSCs (patient participation rate 56.8%). Of the 101 neurologists, 54 were headache specialists. Due to sample size, data from a total of 1012 patients with episodic CH were analyzed in this study; 605 of these patients provided PSCs (patient participation rate 59.8%). APDs in the total sample and by country are shown in Table 1. Across all countries, most patients were men (66.6%, 674/1012) and the mean age was 40.9 years (Table 2). Patients with >3 APD were older than those with ≤3 APD (Table 2 and Supplementary Table 1). Mean body mass index (BMI) of all patients was 25.9 kg/m²; BMI was comparable across countries and APD groups. Anxiety (16.7%, 169/1012), hypertension (14.2%, 144/1012), depression (12.0%, 121/1012), insomnia (6.5%, 66/1012), and asthma (4.5%, 46/1012) were the five most common concomitant conditions. Overall, depression was reported more often in patients with >3 APD than those with ≤1 APD (19.9%, 40/201 vs. 10.6%, 41/388) (Table 2). This result was most pronounced in the UK patient population (>3 APD, 29.4%, 20/68 vs. ≤1 APD, 6.3%, 8/126) (Supplementary Table 1). Across all countries, the proportion of current smokers was the highest in those with >3 APD (Table 2). Contrasting results were observed across countries; in Germany those with >3 APD were most often current smokers, while in the UK those with >3 APD were less often current smokers than those in other APD categories (Supplementary Table 1).

Table 1. Patient sample sizes.

	All countries (n = 1012)	USA (n = 375)	Germany (n = 309)	UK (n = 328)
≤1 attack per day	388 (38.3)	156 (41.6)	106 (34.3)	126 (38.4)
>1 to ≤2 attacks per day	175 (17.3)	74 (19.7)	51 (16.5)	50 (15.2)
>2 to ≤3 attacks per day	112 (11.1)	45 (12.0)	38 (12.3)	29 (8.8)
>3 attacks per day	201 (19.9)	51 (13.6)	82 (26.5)	68 (20.7)
Unknown number of attacks per day	136 (13.4)	49 (13.1)	32 (10.4)	55 (16.8)

All values are n (%) unless otherwise indicated. Based on physician report forms.

Table 2. Patient demographic and general clinical characteristics.

	All patients	≤1 APD	>1 to ≤2 APD	>2 to ≤3 APD	>3 APD	Don’t know APD	p Value
Age
n	1012	388	175	112	201	136
Mean (SD)	40.9 (11.7)	39.9 (11.9)	39.8 (11.9)	40.1 (10.9)	43.3 (11.4)	42.2 (11.6)	.004
Median	40	38	38	39	44	40
IQR	31.5, 49	30, 48.5	31, 49	31, 47.5	35, 51	33.5, 50
Sex
n	1012	388	175	112	201	136
Male	674 (66.6)	250 (64.4)	123 (70.3)	84 (75.0)	138 (68.7)	79 (58.1)	.036
BMI, kg/m^2
n	1012	388	175	112	201	136
Mean (SD)	25.9 (3.9)	25.7 (4.1)	25.5 (3.7)	25.7 (3.6)	26.3 (3.7)	26.7 (4.2)
Median	25.4	25.2	25.2	25.1	25.8	25.9
IQR	23.4, 27.8	23.3, 27.7	23.1, 27.2	23.1, 27.2	23.7, 28.6	23.5, 29.7
Top 5 concomitant conditions
n	1012	388	175	112	201	136
Anxiety	169 (16.7)	75 (19.3)	26 (14.9)	13 (11.6)	36 (17.9)	19 (14.0)	.250
Hypertension	144 (14.2)	49 (12.6)	17 (9.7)	16 (14.3)	39 (19.4)	23 (16.9)	.062
Depression	121 (12.0)	41 (10.6)	21 (12.0)	8 (7.1)	40 (19.9)	11 (8.1)	.002
Insomnia	66 (6.5)	22 (5.7)	13 (7.4)	8 (7.1)	18 (9.0)	5 (3.7)	.339
Asthma	46 (4.5)	14 (3.6)	8 (4.6)	4 (3.6)	17 (8.5)	3 (2.2)	.043
Smoking status
n	1,012	388	175	112	201	136
Current smoker	275 (27.2)	106 (27.3)	43 (24.6)	32 (28.6)	66 (32.8)	28 (20.6)	.009
Ex-smoker	218 (21.5)	78 (20.1)	45 (25.7)	29 (25.9)	36 (17.9)	30 (22.1)
Never smoked	472 (46.6)	193 (49.7)	80 (45.7)	49 (43.8)	86 (42.8)	64 (47.1)
Don’t know	47 (4.6)	11 (2.8)	7 (4.0)	2 (1.8)	13 (6.5)	14 (10.3)

Abbreviations. APD, Attacks per day; IQR, Interquartile range; SD, Standard deviation.

All values are n (%) unless otherwise indicated. Based on physician report forms.

The mean (SD) number of APD across all countries was 2.4 (2.4) (Table 3). The greatest proportion of patients in the USA (30.9%, 116/375) and UK (37.2%, 122/328) had an average CH attack duration of >30 min but ≤1 h; in Germany, approximately, 30% (of patients had a CH attack duration of >15 min but ≤30 min (91/309) or >30 min but ≤1 h (88/309) (Supplementary Table 2). The mean (SD) number of cluster periods per year across all countries was 2.6 (2.0) (Table 3); the greatest number was observed in Germany (3.2 [1.6]) and the lowest in the UK (1.9 [1.3]) (Supplementary Table 2). No statistically significant association between APD and cluster period frequency per year was observed (Table 3). Of the 1012 patients with episodic CH, almost a quarter of patients (24.0%, 243/1012) were in an active cluster period at the point of data collection; 67.4% (682/1012) were in remission and 8.6% (87/1012) were experiencing symptoms of an impending cluster period (Table 3). Across all countries, the mean (SD) cluster period was 30.8 (34.9) days (Table 3). The longest mean (SD) cluster period was observed in the UK (42.7 [45.2] days) and the shortest in Germany (22.5 [16.7] days) (Supplementary Table 2).

Table 3. Cluster headache-specific clinical characteristics.

	All patients	≤1 APD	>1 to ≤2 APD	>2 to ≤3 APD	>3 APD	Don’t know APD	p Value
APD
n	876	388	175	112	201	0
Mean (SD)	2.4 (2.4)	0.7 (0.3)	1.9 (0.2)	3.0 (0.1)	5.9 (2.4)		<.001
Median	2	0.7	2	3	5
IQR	0.9, 3	0.4, 1	2, 2	3, 3	4, 7
CP frequency, per year
n	737	319	138	83	154	43
Mean (SD)	2.6 (2.0)	2.6 (2.3)	2.4 (1.7)	2.5 (1.3)	2.8 (1.8)	2.7 (2.2)	.528
Median	2	2	2	2	2	2
IQR	1, 3	1, 3	1, 3	2, 3	1, 4	1, 3
CP, days
n	787	331	141	94	178	43
Mean (SD)	30.8 (34.9)	31.6 (38.3)	33.7 (31.0)	28.3 (29.8)	28.2 (31.9)	31.0 (40.8)	.618
Median	21	21	21	21	21	14
IQR	10, 35	10, 35	14, 42	8, 42	7, 30.4	7, 42
Duration of average CH attack
n	1012	388	175	112	201	136
≤15 min	42 (4.2)	7 (1.8)	6 (3.4)	6 (5.4)	21 (10.4)	2 (1.5)	<.001
>15 min but ≤30 min	207 (20.5)	68 (17.5)	35 (20)	31 (27.7)	60 (29.9)	13 (9.6)
>30 min but ≤1 h	326 (32.2)	142 (36.6)	59 (33.7)	29 (25.9)	65 (32.3)	31 (22.8)
>1 h but ≤2 h	229 (22.6)	98 (25.3)	47 (26.9)	28 (25)	27 (13.4)	29 (21.3)
>2 h but ≤3 h	91 (9)	38 (9.8)	8 (4.6)	11 (9.8)	20 (10)	14 (10.3)
>3 h	28 (2.8)	13 (3.4)	4 (2.3)	1 (0.9)	3 (1.5)	7 (5.1)
Length varies/unpredictable	68 (6.7)	20 (5.2)	13 (7.4)	6 (5.4)	5 (2.5)	24 (17.6)
Don’t know	21 (2.1)	2 (0.5)	3 (1.7)	0 (0)	0 (0)	16 (11.8)
Current CH status
n	1012	388	175	112	201	136
In remission	682 (67.4)	280 (72.2)	116 (66.3)	70 (62.5)	126 (62.7)	90 (66.2)	.163
Impending	87 (8.6)	32 (8.2)	11 (6.3)	14 (12.5)	17 (8.5)	13 (9.6)
Active	243 (24.0)	76 (19.6)	48 (27.4)	28 (25.0)	58 (28.9)	33 (24.3)

Abbreviations. APD, Attacks per day; CP, Cluster period; IQR, Interquartile range; SD, Standard deviation.

All values are n (%) unless otherwise indicated. Data presented derived from physician-reported patient record forms.

A total of 29.2% (218/745) of all patients were misdiagnosed at least once prior to a confirmed CH diagnosis. The five most common misdiagnoses were migraine (70.2%, 153/218), tension headache (21.6%, 47/218), trigeminal neuralgia (8.7%, 19/218), headache (3.2%, 7/218), and sinusitis (2.8%, 6/218) (Table 4). Considerable variation between countries was observed with misdiagnoses of tension headache, trigeminal neuralgia, and headache (Supplementary Table 2).

Table 4. Diagnosis of cluster headaches.

	All patients	≤1 APD	>1 to ≤2 APD	>2 to ≤3 APD	>3 APD	Don’t know APD	p Value
n	745	312	140	87	139	67
Diagnosed first time	527 (70.7)	211 (67.6)	105 (75.0)	60 (69.0)	94 (67.6)	57 (85.1)	.005
1 misdiagnosis	152 (20.4)	82 (26.3)	21 (15.0)	19 (21.8)	23 (16.5)	7 (10.4)
2 misdiagnoses	54 (7.2)	16 (5.1)	13 (9.3)	7 (8.0)	15 (10.8)	3 (4.5)
3 misdiagnoses	7 (0.9)	2 (0.6)	0 (0)	1 (1.1)	4 (2.9)	0 (0)
≥4 misdiagnoses	5 (0.7)	1 (0.3)	1 (0.7)	0 (0)	3 (2.2)	0 (0)
Top 5 misdiagnoses^a
n	218	101	35	27	45	10
Migraine	146 (67.0)	69 (68.3)	25 (71.4)	17 (63.0)	28 (62.2)	7 (70.0)	.895
Tension headache	47 (21.6)	25 (24.8)	4 (11.4)	4 (14.8)	12 (26.7)	2 (20.0)	.384
Trigeminal neuralgia	19 (8.7)	5 (5.0)	2 (5.7)	3 (11.1)	9 (20.0)	0 (0)	.032
Headache	7 (3.2)	1 (1.0)	0 (0)	2 (7.4)	4 (8.9)	0 (0)	.054
Migraine with no aura	6 (2.8)	3 (3.0)	0 (0)	1 (3.7)	2 (4.4)	0 (0)	.761
Sinusitis	6 (2.8)	1 (1.0)	1 (2.9)	2 (7.4)	2 (4.4)	0 (0)	.390
Other migraine types	1 (.05)	0 (0)	0 (0)	0 (0)	1 (2.2)	0 (0)	.425

Abbreviation. APD, Attacks per day.

All values are n (%) unless otherwise indicated. Data presented derived from physician-reported patient record forms.

^a For sake of completeness, other migraine types are included here.

The five most common current symptoms reported by physicians during a typical cluster attack across all countries were unilateral pain (69.2%, 700/1012), orbital or supraorbital pain (46.4%, 470/1012), lacrimation (45.8%, 463/1012), conjunctival injection (42.6%, 431/1012), and temporal pain (41.0%, 415/1012) (Table 5). Considerable variation between countries was observed for physician-reported symptoms of unilateral pain, orbital or supraorbital pain, temporal pain, and conjunctival injection (Supplementary Table 2). The five most common feelings or reactions reported by all patients during a typical attack were stress (45.5%, 265/582), anxiety (44.7%, 260/582) or agitation (44.7%, 260/582), pacing (41.2%, 240/582), and difficulty relaxing (41.1%, 239/582). The proportion of patients that reported stress, agitation, restlessness, and difficulty relaxing was the highest in those with >3 APD. Physicians reported that 74.1% (737/955) of patients experienced severe or very severe pain during a typical CH attack; 67.2% (387/576) of patients self-reported severe or very severe pain during a typical CH attack (Table 5).

Table 5. Symptoms of cluster headaches.

	All patients	≤1 APD	>1 to ≤2 APD	>2 to ≤3 APD	>3 APD	Don’t know APD	p Value
Top 5 current symptoms experienced during attack in typical cluster period
n	1012	388	175	112	201	136
Unilateral pain	700 (69.2)	282 (72.7)	129 (73.7)	77 (68.8)	146 (72.6)	66 (48.5)	<.001
Orbital/supraorbital pain	470 (46.4)	173 (44.6)	79 (45.1)	47 (42.0)	115 (57.2)	56 (41.2)	.014
Lacrimation	463 (45.8)	182 (46.9)	84 (48.0)	54 (48.2)	100 (49.8)	43 (31.6)	.011
Conjunctival injection	431 (42.6)	162 (41.8)	86 (49.1)	56 (50.0)	95 (47.3)	32 (23.5)	<.001
Temporal pain	415 (41.0)	145 (37.4)	77 (44.0)	52 (46.4)	89 (44.3)	52 (38.2)	.244
Top 5 most common feelings during attack
n	582	244	81	66	120	71
Stress	265 (45.5)	113 (46.3)	28 (34.6)	31 (47.0)	69 (57.5)	24 (33.8)	.005
Anxiety	260 (44.7)	106 (43.4)	38 (46.9)	32 (48.5)	54 (45.0)	30 (42.3)	.928
Agitation	260 (44.7)	111 (45.5)	27 (33.3)	27 (40.9)	66 (55.0)	29 (40.8)	.036
Pacing	240 (41.2)	101 (41.4)	35 (43.2)	31 (47.0)	53 (44.2)	20 (28.2)	.168
Difficulty relaxing	239 (41.1)	104 (42.6)	25 (30.9)	25 (37.9)	71 (59.2)	14 (19.7)	<.001
Restlessness	230 (39.5)	99 (40.6)	27 (33.3)	23 (34.8)	68 (56.7)	13 (18.3)	<.001
CH pain severity, physician-reported
n	995	388	174	112	200	121
Mild	22 (2.2)	6 (1.5)	7 (4.0)	1 (0.9)	4 (2.0)	4 (3.3)	<.001
Moderate	236 (23.7)	120 (30.9)	32 (18.4)	17 (15.2)	29 (14.5)	38 (31.4)
Severe	437 (43.9)	165 (42.5)	74 (42.5)	59 (52.7)	87 (43.5)	52 (43.0)
Very severe	300 (30.2)	97 (25.0)	61 (35.1)	35 (31.3)	80 (40.0)	27 (22.3)
CH pain severity, patient-reported
n	576	242	81	65	122	66
Mild	20 (3.5)	8 (3.3)	4 (4.9)	0 (0)	4 (3.3)	4 (6.1)	<.001
Moderate	169 (29.3)	98 (40.5)	23 (28.4)	11 (16.9)	19 (15.6)	18 (27.3)
Severe	264 (45.8)	108 (44.6)	36 (44.4)	37 (56.9)	52 (42.6)	31 (47.0)
Very severe	123 (21.4)	28 (11.6)	18 (22.2)	17 (26.2)	47 (38.5)	13 (19.7)

Abbreviations. APD, Attacks per day; CP, Cluster period.

All values are n (%) unless otherwise indicated. Data presented derived from physician-reported patient record forms, except for patient-reported pain severity.

Across all countries, most patients (69.3%, 391/564) reported that there was no specific or predicable time when cluster periods occurred (Table 6). The most common cluster period triggers reported by all patients were no known triggers (38.9% 225/579), stress (30.7%, 178/579), changes in weather (24.5%, 142/579), lack of sleep or exhaustion (21.4%, 124/579), change in seasons (17.1%, 99/579) and alcohol (15.4%, 89/579). The most common attack triggers reported by all patients during a cluster period were stress (37.1%, 208/561), no known triggers (36.2%, 203/561), lack of sleep or exhaustion (26.0%, 146/561), changes in weather (17.8%, 100/561), time of day (14.4%, 81/561) and alcohol (14.3%, 80/561) (Table 6).

Table 6. Triggers for and within cluster periods.

	All countries	USA	Germany	UK
Typical time of year for cluster period	n = 564	n = 216	n = 227	n = 121
No specific or predictable time of year	391 (69.3)	148 (68.5)	158 (69.6)	85 (70.3)
Top CP triggers	n = 579	n = 223	n = 233	n = 123
No known triggers	225 (38.9)	62 (27.8)	115 (49.4)	48 (39.0)
Stress	178 (30.7)	77 (34.5)	48 (20.6)	53 (43.1)
Changes in weather	142 (24.5)	77 (34.5)	50 (21.5)	15 (12.2)
Lack of sleep/exhaustion	124 (21.4)	54 (24.2)	25 (10.7)	45 (36.6)
Change in seasons	99 (17.1)	38 (17.0)	47 (20.2)	14 (11.4)
Alcohol	89 (15.4)	41 (18.4)	30 (12.9)	18 (14.6)
Top triggers during a CP	n = 561	n = 215	n = 223	n = 123
Stress	208 (37.1)	97 (45.1)	58 (26.0)	53 (43.1)
No known triggers	203 (36.2)	53 (24.7)	107 (48.0)	43 (35.0)
Lack of sleep/exhaustion	146 (26.0)	50 (23.3)	49 (22.0)	47 (38.2)
Change in weather	100 (17.8)	53 (24.7)	35 (15.7)	12 (9.8)
Time of day	81 (14.4)	19 (8.8)	44 (19.7)	18 (14.6)
Alcohol	80 (14.3)	37 (17.2)	25 (11.2)	18 (14.6)

Abbreviation. CP, Cluster period.

All values are n (%) unless otherwise indicated. Based on patient self-completed questionnaires.

During their current or most recent cluster period, most patients were prescribed acute treatment only (overall, 47.6%, 482/1012; USA, 42.9%, 161/375; UK, 48.2%, 158/328; Germany, 52.8%, 163/309), followed by acute and preventive treatment (overall, 37.9%, 384/1012; USA, 42.4%, 159/375; UK, 36.6%, 120/328; Germany, 34.0%, 105/309) and preventive treatment only (overall, 10.3%, 104/1012; USA, 12.3%, 46/375; UK, 7.9%, 26/328; Germany, 10.4%, 32/309) (Table 7 and Supplementary Table 3). A minority of patients were not prescribed any treatment (overall, 4.2%, 42/1012; USA, 2.4%, 9/375; UK, 7.3%, 24/378; Germany, 2.9%, 9/309). Differences in prescribing patterns between PCPs and neurologists were observed; when compared to neurologists, PCPs prescribed acute treatment only more often (62.1%, 180/290 vs. 41.8%, 302/722) and acute and preventive treatment less often (26.6%, 77/290 vs. 42.5%, 307/722) (Supplementary Table 4). When stratified by duration of cluster period (≤4 weeks or >4 weeks), patients with a cluster period ≤4 weeks were more often prescribed acute treatment only (48.1%, 245/509 vs. 37.1%, 103/278); this was most pronounced in the UK (50.8%, 66/130 vs. 31.3%, 35/112) (Supplementary Table 5).

Table 7. Treatment patterns.

	All patients	≤1 APD	>1 to ≤2 APD	>2 to ≤3 APD	>3 APD	Don’t know APD	p Value
Current/most recently prescribed treatments
n	1012	388	175	112	201	136
No treatment	42 (4.2)	16 (4.1)	7 (4.0)	2 (1.8)	6 (3.0)	11 (8.1)	.003
Acute only	482 (47.6)	174 (44.8)	65 (37.1)	54 (48.2)	105 (52.2)	84 (61.8)
Preventive only	104 (10.3)	39 (10.1)	22 (12.6)	10 (8.9)	19 (9.5)	14 (10.3)
Acute and preventive	384 (37.9)	159 (41.0)	81 (46.3)	46 (41.1)	71 (35.3)	27 (19.9)
Current/most recently prescribed acute treatments
n	1012	388	175	112	201	136
Sumatriptan	590 (58.3)	239 (61.6)	101 (57.7)	66 (58.9)	111 (55.2)	73 (53.7)	.440
Oxygen	319 (31.5)	117 (30.2)	54 (30.9)	35 (31.3)	82 (40.8)	31 (22.8)	.010
Zolmitriptan	119 (11.8)	45 (11.6)	16 (9.1)	14 (12.5)	30 (14.9)	14 (10.3)	.486
Sumatriptan formulation
n	589	238	101	66	111	73	<.001
Oral	236 (40.1)	108 (45.4)	27 (26.7)	19 (28.8)	42 (37.8)	40 (54.8)
Intranasal	241 (40.9)	81 (34.0)	62 (61.4)	34 (51.5)	48 (43.2)	16 (21.9)
Injection	110 (18.7)	48 (20.2)	12 (11.9)	13 (19.7)	20 (18.0)	17 (23.3)
Don’t know	2 (0.3)	1 (0.4)	0 (0)	0 (0)	1 (0.9)	0 (0)
Current/most recently prescribed preventive treatments
n	1012	388	175	112	201	136
Verapamil	298 (29.4)	132 (34.0)	65 (37.1)	37 (33.0)	47 (23.4)	17 (12.5)	<.001
Topiramate	107 (10.6)	37 (9.5)	23 (13.1)	12 (10.7)	23 (11.4)	12 (8.8)	.687
Lithium	39 (3.9)	19 (4.9)	5 (2.9)	2 (1.8)	9 (4.5)	4 (2.9)	.490

All values are n (%) unless otherwise indicated. Based on physician report forms.

The most frequently prescribed acute treatments across all countries were sumatriptan (58.3%, 590/1012), oxygen (31.5%, 319/1012), and zolmitriptan (11.8%, 119/1012) (Table 7). Compared to the other countries, sumatriptan was most often prescribed in the USA (62.4%, 234/375) (Supplementary Table 3). Across all countries, there was a similar proportion of sumatriptan oral (40.1%, 236/589) and injection (40.9%, 241/589) formulations prescribed; the remaining 18.7% (110/589) was for the intranasal formulation (Table 7). Oxygen (45.0%, 139/309) and zolmitriptan (17.5%, 54/309) were most often prescribed in Germany compared to the other countries; considerable variation was observed between countries in oxygen prescribing rates (USA, 18.9%, 71/375; UK, 33.2%, 109/328) (Supplementary Table 3). Across all countries, the proportion of patients prescribed oxygen increased as APD increased (≤1 APD, 30.2%, 117/388 vs. >3 APD, 40.8%, 82/201) (Table 7).

The most frequently prescribed preventive treatments across all countries were verapamil (29.4% 298/1012), topiramate (10.6%, 107/1012), and lithium (3.9%, 39/1012) (Table 7). Corticosteroid use was low (0.8%, 9/1012, data not shown). Compared to the other countries, verapamil (34.3%, 106/309) was most often prescribed in Germany, topiramate in the USA (17.9%, 67/375), and lithium in Germany (4.9%, 15/309); considerable variation was observed between countries in topiramate prescribing rates. In the UK, there was an inverse relationship between APD and proportion of patients prescribed verapamil (≤1 APD, 37.3%, 47/126 vs. >3 APD, 17.6%, 12/68) (Supplementary Table 3). Of patients ever prescribed preventive treatment, 15.0% (75/501) had discontinued a preventive treatment at some point. Lack of efficacy (60.0%, 45/75) and tolerability (45.3%, 34/75) were the most common reasons for discontinuing any preventive treatment at any point during the patient’s treatment history (Supplementary Table 6).

Of the patients who took preventive treatments, across all countries most patients (50.7%, 176/347) took preventive treatment before, during, and continuously (no planned treatment holidays between cluster periods) after a cluster period; this was followed by patients who took preventive treatment during and continuously after a cluster period (32.3%, 112/347). A low proportion of patients took preventive treatment only during a cluster period (5.2%, 18/347) or before and during cluster period (0.3%, 1/347) (Figure 1).

Figure 1. Use of preventive treatment in relation to cluster periods.

Discussion

This analysis revealed that patients with episodic CH suffer from multiple cluster periods per year. Despite availability of evidence-based guidelines recommending the use of preventive treatment for CH, only approximately half of patients in this study were prescribed a preventive treatment during a cluster period.

This study provides important insights on the burden of CH in relation to the number of APD experienced. The proportion of patients that experienced stress, agitation, restlessness, and difficulty relaxing during an attack was the highest in those with >3 APD. In addition, the greatest proportion of patients with comorbid depression was observed in those with >3 APD.

Some of the clinical features reported by patients in this study may contrast with previous findings. Most patients in this study did not report any clear seasonality to cluster periods; cluster periods have previously been described as usually occurring in spring or autumn²⁶. The ICHD states that episodic cluster periods usually last between 2 weeks and 3 months²; while within such range, physicians in this study reported cluster periods of relatively brief duration (average 30.8 days). A Danish survey of patients with CH revealed a mean cluster period of 8.3 weeks, although that study also included patients with chronic CH³, who generally have cluster periods of longer duration. The shorter duration observed in this study may reflect a more heterogeneous population seeking care across a more diverse range of practice settings and physician specialties than previous reports.

A meta-analysis published in 2008 of population-based studies revealed men are affected by episodic CH approximately four times more often than women⁴. In our study, 66.6% of patients in our sample were men. Previous studies have indicated that there are differences between males and females with respect to the symptomatic burden of CH. While prevalence is greater in males, females appear to be more severely affected^27,28. Our observations suggest that clinicians should suspect CH in patients who present with appropriate symptoms and history regardless of sex.

Given that a number of physician specialties can be involved in the diagnosis and treatment of CH, it is important to understand differences in diagnosis and treatment between physician types. Nearly, a third of patients in this study had been misdiagnosed at least once prior to receiving a confirmed CH diagnosis. Similarly, high rates of misdiagnosis in CH have been reported in the literature, with one Danish study finding that 49% of their CH patient sample had been initially misdiagnosed²⁹. Similar to our findings, Frederiksen et al. reported that migraine and tension headache were the most common misdiagnoses received by these patients.

The considerable proportion of patients (40%) prescribed oral sumatriptan as acute treatment was surprising given that subcutaneous (injection) sumatriptan has the highest level of evidence for efficacy according to current evidence-based guidelines¹³. Given the speed of onset, injectable or nasal formulations are preferred in an acute attack; oral routes of administration are not optimal, as indicated in a Cochrane systematic review³⁰. Given the evidence in support of subcutaneous sumatriptan, it is possible that physician lack of awareness or factors related to patient preference, such as fear of needles, price or insufficient supply of injections, may explain the use of the oral formulation in our study.

Despite its effectiveness and tolerability, oxygen was prescribed the least often in the USA as acute treatment among the countries in this study. Some studies have investigated oxygen use for CH in the USA and suggested barriers to use. The United States Cluster Headache Survey revealed that although 93% of respondents were aware of oxygen as a CH therapy, 34% never attempted oxygen therapy³¹. A total of 44% of respondents had to suggest oxygen to their physicians for a prescription and 12% of physicians refused to prescribe oxygen³¹. About half of respondents did not receive proper training or had to find the oxygen source themselves³¹. The authors of that study concluded that physicians should be better educated on oxygen use in CH³¹. A more recent study on the economics of oxygen use in CH in the USA revealed that oxygen was not prohibitively expensive for patients or insurance providers; however, many insurers did not reimburse oxygen costs in CH³².

In this study, we observed that preventive treatment was only used in half of patients. Of those ever prescribed preventive treatment, 15% had discontinued a preventive treatment at some point, with lack of efficacy and tolerability the main reasons for discontinuation. This suggests that the driver for the lack of preventive treatment use may possibly be patients not being started on treatment, rather than of treatment failure. The variation observed in treatment patterns and the underuse of preventive treatments suggests regional differences in local guidelines or in physician awareness of current evidence-based guidelines, which could potentially vary depending on physician specialty and background. Other factors such as lack of adherence or low compliance may have also driven this variation, but these factors are difficult to investigate in self-reported surveys. The most commonly prescribed preventive treatments were verapamil, topiramate, and lithium. While verapamil and lithium across many countries are considered first- and second-choice preventive treatments, respectively, these treatments are supported by limited evidence in CH and may require regular monitoring due to side effects or toxicity^13,33. In the UK, there was an inverse relationship between APD and the proportion of patients prescribed verapamil. As this observation reflected the current or most recent treatment regimen, it is possible that patients had tried verapamil in the past but discontinued it due to lack of efficacy, poor tolerability or both. Likewise, the point-in-time nature of the data collection may explain the low rate observed for therapies that are typically used for a transitional period, such as corticosteroids. Although lack of efficacy followed by tolerability issues were the most common reasons for discontinuation of preventive treatments, most patients who took preventive medication did so before, during, and continuously after a cluster period. Taken together, these observations highlight the need for other preventive options, and for further research to understand the reasons for and expectations around preventive treatment initiation, continuation, and switching.

Limitations

This study had some limitations. We did not include a true random sample of physicians or patients. Physician willingness to complete the survey influenced participation and, although respondent physicians were requested to collect data on a series of consecutive patients to avoid selection bias, this was contingent upon the integrity of the participating physician rather than formalized source verification procedures. It is possible that patients treated by the same physician may be more similar than those treated by different physicians; this was not accounted for in the statistical tests performed. Although the point-in-time study design prevents any conclusions about causal relationships, it is possible to identify significant associations. The data were not powered specifically for this analysis and adjustments were not made for multiple testing. Therefore, any statistically significant differences observed may not indicate important or otherwise clinically relevant differences. Due to this study design, it is possible that patients could be categorized as taking preventive treatment before, during, and continuously after a cluster period or during (but not before) and continuously after a cluster period. Although going forward these two patient groups would essentially follow the same preventive treatment plan (i.e. take preventive treatment continuously), at the time of data capture the first patient group had taken preventive treatment before their cluster period, whereas the second patient group had not. Participating patients may have been more likely to visit their physician due to more frequent cluster periods or more severe disease, and therefore may not have been representative of all patients with episodic CH. Physicians were asked to indicate from a precoded list which symptoms patients experience when suffering a typical cluster attack. It is possible that the reporting of unilateral pain is an underestimation if the physician prioritized the specific location of the pain (i.e. orbital/supraorbital, temporal, etc.) above whether it was one sided or not. The concept of pain is subjective, and patient perceptions of pain can change over time. Nearly, all patients were on some form of treatment (either acute and/or preventive); therefore, they may have answered with respect to their level of pain when medicated and achieving some form of pain relief. Data on reasons for preventive treatment discontinuation were derived from low patient bases and should be interpreted with caution. When asked to report their triggers, patients selected from a list of responses that may have excluded other potential triggers, such as nitroglycerin. Recall bias, a common limitation of surveys, may have affected patient and physician responses to the questionnaires. However, data for these analyses were collected at the time of each patient’s consultation and this was expected to reduce the likelihood of recall bias. It is worth noting that a substantially smaller proportion of patients in the current study were in an active cluster period (24%); patients and their physicians were instructed to recall a typical cluster period or attack when providing responses. The degree that recall bias may have impacted these findings is unknown. Given that data collection occurred several years before publication, any recent changes in the treatment landscape of episodic CH and patient and physician awareness of the disease are not reflected in the results presented here.

Conclusions

Real-world data from patients with CH and their treating physicians revealed that patients with episodic CH had a considerable clinical burden marked with attacks of severe or very severe pain in cluster periods multiple times each year. Misdiagnosis was common in this population, thus, highlighting the need for additional awareness of this disease among physicians. Preventive and acute treatment were prescribed less often than acute treatment only, and both the preventive and acute treatments prescribed varied by geographic region and were not consistent with available evidence-based guidelines. Our results suggest that improving physician awareness of both episodic CH itself and the available treatments, both acute and preventive, will improve patient care. Preventive treatments that are both effective and well-tolerated are an unmet need in this population and are greatly underutilized.

Transparency

Declaration of funding

Data collection was undertaken by Adelphi Real World, as part of an independent survey entitled The Adelphi Cluster Headache Disease Specific Programme (DSP). The analysis described here used data from the Adelphi Cluster Headache DSP. The DSP is a wholly owned Adelphi product; Eli Lilly and Company were one of the multiple subscribers to the DSP. All authors had full access to all data and had final responsibility for the decision to submit for publication. Neither honoraria nor payments were made for authorship.

Declaration of financial/other relationships

Sarah Cotton: employee of Adelphi Real World, Bollington, UK; Jeffrey S. Andrews: employee of Takeda, Pharmaceuticals, Cambridge, MA. Former employee and shareholder of Eli Lilly and Company, Indianapolis, IN, USA; Russell M. Nichols: employee and shareholder of Eli Lilly and Company, Indianapolis, IN, USA; James Jackson: employee of Adelphi Real World, Bollington, UK; Antje Tockhorn-Heidenreich: employee and shareholder of Eli Lilly and Company, Indianapolis, IN, USA; Gary Milligan: employee of Adelphi Real World, Bollington, UK; James M. Martinez: employee and shareholder of Eli Lilly and Company, Indianapolis, IN, USA.

Author contributions

All authors were involved in (1) conception or design, or analysis and interpretation of data; (2) drafting and revising the article; (3) providing intellectual content of critical importance to the work described; and (4) final approval of the version to be published, and therefore meet the criteria for authorship in accordance with the International Committee of Medical Journal Editors (ICMJE) guidelines. In addition, all named authors take responsibility for the integrity of the work as a whole, and have given their approval for this version to be published.

Acknowledgements

Medical writing assistance under the guidance of the authors was provided by Derek Ho, on behalf of Adelphi Real World and Eli Lilly and Company in accordance with Good Publication Practice (GPP3) guidelines³⁴.

Data availability statement

All data, i.e. methodology, materials, data and data analysis, that support the findings of this survey are the intellectual property of Adelphi Real World. As such, the questionnaires used in this study are not publicly available. All relevant data are presented in full, and descriptions of the methods, the tools used, questions asked and how answers were derived are included within the methodology and results sections of the manuscript, where appropriate. All requests for access should be addressed directly to Sarah Cotton at [email protected].