Increased suicidality in patients with cluster headache

Abstract

Objective

To investigate suicidality related to cluster headache and factors associated with increased suicidality in cluster headache patients.

Methods

In this multicenter study, 193 cluster headache patients were recruited between September 2016 and August 2018. Patients were asked about their suicidality during and between attacks, specifically about passive suicidal ideation, active suicidal ideation, suicide plan, and suicide attempt. Univariable and multivariable logistic regression analyses were performed to evaluate the factors associated with high ictal suicidality (sum of positive response ≥ 2). Patients were followed up when they were in the between-bouts period.

Results

A total of 175 cluster headache patients in the in-bout period were included in this study. Passive suicidal ideation, active suicidal ideation, suicidal planning, and suicidal attempt were reported by 111 (64.2%), 62 (35.8%), 10 (5.8%), and four (2.3%) patients during attacks; seven (4.0%), six (3.5%), five (2.9%) and two (1.2%) patients interictally; and none (0%), one (1.9%), one (1.9%), and none (0%) among patients in the between-bouts period. Factors associated with high ictal suicidality were longer disease duration, the Headache Impact Test score, and the Patient Health Question-9 score (multivariable OR = 1.90 per 10-year increase in disease duration, 95% CI = 1.18–3.05, p = 0.008; multivariable OR = 3.19 per 10-point increase in HIT-6, 95% CI = 1.73–5.87, p < 0.001; multivariable OR = 2.11 per 10-point increase in PHQ-9, 95% CI = 1.13–3.95, p = 0.020, respectively).

Conclusions

Cluster headache attack carries a high suicidality compared to the interictal or between-bouts state. An intensive treatment to reduce cluster headache burden may be helpful to alleviate suicide risk in cluster headache patients.

Keywords

Cluster headache suicide disease burden comorbidity

Introduction

Cluster headache (CH) is characterized by strictly unilateral, severe headaches, which accompany cranial autonomic symptoms and typical recurrence patterns (1). CH is often considered as one of the most painful disorders (2). Functional impairments from the burden of CH seriously impact the patients’ lives, most of whom are in the most productive age for social and family life (3).

Due to the excruciating attacks and highly-frequent recurrence during the bout, CH has also been referred to as the “suicide headache” (4). However, a limited number of studies investigating suicidality in CH patients exist to date (5,6). One prospective study, using a web-based survey, examined the suicidality of CH patients (6). Based on this study, the rate of suicidal attempt was considered similar to that of general population (7). However, this study result has not been replicated by studies recruiting CH patients with a verified diagnosis. Furthermore, suicidality might change according to phases of CH (i.e. during attack, within-bout interictal, and between-bouts periods). In addition, questions regarding the factors influencing suicide risk are yet to be answered.

To comprehensively investigate the suicidality related to CH and the factors associated with increased suicidality, we conducted a prospective multicenter study of CH. Patients with CH within their active periods were assessed regarding their suicidality during and outside the CH attacks and were followed up during their between-bouts periods. Both the risk of suicidality during attacks and the clinical factors associated with active suicidality were analyzed.

Methods

Study design

This study was designed as a part of the Korean Cluster Headache Registry Study (KCHR), which is a prospective, longitudinal, multicenter study that enrolled consecutive patients with CH from 15 hospitals (13 university hospitals, including eight tertiary and five secondary referral hospitals; and two general hospitals) in Korea. The suicidality study was conceived as a major axis of the KCHR and implemented prior to the first patient recruitment.

The study protocol was reviewed and approved by the institutional review board (IRB) of each participating hospital and written informed consent was obtained from every patient prior to the recruitment. Patient recruitment was conducted following the IRB approval in each hospital, mostly between September 2016 and February 2017. Only patients who visited the hospital and were diagnosed with CH by investigators were recruited at the clinic for the KCHR study. The diagnosis of CH was performed by each investigator based on the criteria from the International Classification of Headache Disorders (ICHD), 3rd edition, beta version. Patients with episodic, chronic, and probable CHs were included in this study (8).

Details about the KCHR

The KCHR comprises several sub-studies, including (but not restricted to) clinical manifestations, quality of life, suicidality, treatment response, validation of ICHD diagnostic criteria, and longitudinal follow-up for prognosis. All the participants were seen and interviewed by investigators, who determined the type, onset, clinical course, characteristics of CH, and treatment response. Comorbid depression and anxiety were assessed using the Korean version of the Patient Health Questionnaire-9 (PHQ-9) and the General Anxiety Disorder-7 item scale (GAD-7) (9,10). Patients were followed up during their between-bouts periods.

Assessment of suicidality

Suicidality was assessed through the four structured questions according to three (i.e. ictal, interictal, and between-bout) clinical situations. For ictal suicidality assessment, we asked patients if they experienced a passive suicidal ideation (“Have you thought that it was better to die while you experienced CH attacks in the active period?”), active suicidal ideation (“Have you thought of killing yourself while you experienced CH attacks in the active period?”), suicidal planning (“Have you planned suicide while you experienced CH attacks in the active period?”), and suicidal attempt (“Have you attempted suicide while you experienced CH attacks in the active period?”). Interictal suicidality was assessed by using the same questions about those experiences when they were between attacks but within the active cluster period (“Have you… when you were between attacks but in the active period?). The ictal and interictal suicidality scores were defined as the sum of the positive responses to the four questions. The between-bout suicidality was followed up in patients who visited the clinic during their between-bouts periods using same questions: “Have you… during the remission period?”.

Validation of questionnaires

We validated the suicidality questionnaire with the Korean version of the Columbia suicide severity rating scale (C-SSRS) for screening during bouts in 26 CH patients (11,12). Comparing the sum of positive response from five questions regarding suicidal ideation (wish to be dead; non-specific active suicidal thoughts; active suicidal ideation with any methods without intent to act; active suicidal ideation with some intent to act, without specific plan; active suicidal ideation with specific plan and intent) and actual suicidal attempt, the ictal suicidality score showed a good agreement (intra-class correlation coefficient = 0.854, 95% CI = 0.673–0.934, p < 0.001).

Statistical analysis

Categorical data were presented as numbers (%), and continuous variables were summarized as means (SD). The intra-subject odds ratio for suicidality during the attack compared to the interictal and between-bouts periods was calculated using the generalized estimating equation (GEE). To identify whether the ictal suicidal scores predicted interictal suicidal risk, the receiver operating characteristic (ROC) curve was analyzed. The area under the curve (AUC) and the best cutoff score were identified using the Youden’s index. The cutoff score of the ictal suicidality score was used as the surrogate outcome variable of regression analysis, due to the occurrence rate of the hard outcome; that is, the interictal suicidal attempt, was low. A logistic regression analysis was performed to assess the factors associated with high suicidality. Clinical factors were tested as explanatory variables and were selected in the multivariable logistic regression analysis using the forward stepwise method. Interictal suicidality scores were analyzed using the same method.

All the statistical analysis was performed using the Stata 15.0 software (StataCorp 2017. Stata Statistical Software: Release 15; StataCorp LLC, College Station, TX, USA). A p-value < 0.05 was considered statistically significant.

Results

Patients

Of the 193 patients enrolled in the study period, 18 were in the between-bout period and excluded from this study. A total of 175 in their active period were included in this investigation. Both the demographics and characteristics of included patients are summarized in Table 1. The mean age was 38.4 (SD 11.11, range 18–79) years at enrollment and the mean age at onset was 29.9 (SD 12.91, range 9 – 78) years. The male to female ratio was 5.7:1.

Table 1.

Demographics and characteristics of study participants.

	Patients within the active period (n = 175)
Demographic data
Age, y	38.4 (11.11)
Male sex	149 (85.1%)
BMI	24.1 (3.19)
Smoking
Never	68 (38.9%)
Ex-smoker	34 (19.4%)
Current smoker	73 (41.7%)
Characteristics of cluster headache
Age at onset, y	29.9 (12.91)
Disease duration, y	8.5 (8.41)
Total lifetime bout, n	8.8 (13.72)
ICHD-3b
Definite	153 (87.4%)
Probable	22 (12.6%)
Subtype
Episodic	154 (88.0%)
Chronic	8 (4.6%)
Undetermined^a	13 (7.4%)
Circadian rhythmicity	86 (49.1%)
Seasonal rhythmicity^b	71 (51.5%)
Characteristics of the current bout
Side
Left	78 (44.6%)
Right	88 (50.3%)
Both (alternating or spreading)	9 (5.1%)
Index bout
First onset	36 (20.6%)
Recurrent	139 (79.4%)
Attack frequency per day	2.3 (1.94)
Attack duration, min	99.3 (71.13)
Attack intensity, NRS	8.9 (1.28)
Length of active period, week	10.8 (35.49)
HIT-6 score	68.3 (7.98)
PHQ-9 score	7.4 (6.29)
GAD-7 score	7.6 (5.71)

BMI: body mass index; ICHD-3b: International Classification of Headache Disorders 3rd edition beta version; NRS: numeric rating scale; HIT-6: headache impact test-6; GAD7: general anxiety disorder-7; PHQ9: patient health questionnaire-9.

In the case of first-onset CH with active periods that last without 1 month of attack-free periods, but with an observation period of less than 1 year.

Thirty-six patients with first-onset bout and one who could not recall the season of the previous bout (which was 13 years before the current bout) were excluded.

Data are presented as means (SD) or numbers (%).

Ictal vs. interictal suicidality during active periods

Ictal passive and active suicidal ideation were reported in 111 (64.2%) and 62 (35.8%) patients, respectively. A total of 10 (5.8%) patients had a suicidal plan, whereas four (2.3%) patients attempted suicide during the CH attack. In contrast, only seven (4.0%) and six (3.5%) patients reported passive and active suicidal ideation, respectively, interictally. Additionally, interictal suicidal planning and attempt were reported by five (2.9%) and two (1.2%) patients, respectively (Figure 1). There was no difference in suicidality between patients with definite CH and those with probable CH (definite vs. probable CH: 64.0% vs. 65.2%, p = 0.910 for passive suicidal ideation; 36.7% vs. 30.4%, p = 0.562 for active suicidal ideation; 6.0% vs. 4.4%, p > 0.999 for suicidal planning; and 2.7% vs. 0%, p > 0.999 for suicidal attempt).

Cluster attacks increased the odds of passive suicidal ideation (OR = 42.46, 95% CI = 19.61–91.90, p < 0.001), active suicidal ideation (OR = 15.55, 95% CI = 7.02–34.41, p < 0.001), suicidal planning (OR = 2.06, 95% CI = 0.87–4.88, p = 0.100), and suicidal attempt (OR = 2.02, 95% CI = 0.5–8.26, p = 0.326) when compared to the interictal state, although the latter two were not statistically significant due to low event occurrence.

Ictal suicidality score and interictal suicidality

Using the ROC analysis, the ictal suicidality score was predictable of interictal passive suicidal ideation (AUC = 0.857, 95% CI = 0.769–0.945, p = 0.001), active suicidal ideation (AUC = 0.864, 95% CI = 0.769–0.959, p = 0.002), and suicidal planning (AUC = 0.875, 95% CI = 0.772–0.978, p = 0.004). Although not statistically significant, the ictal suicidality score yielded a high AUC to predict the interictal suicidal attempt (AUC = 0.895, 95% CI = 0.728–1.000, p = 0.055). In all four analyses, a cutoff of ictal suicidality score ≥ 2 provided the best discriminating accuracy (sensitivity and specificity: Respectively, 100% and 66.3% for interictal passive suicidal ideation; 100% and 65.9% for interictal active suicidal ideation; 100% and 65.5% for interictal suicidal planning; and 100% and 64.3% for interictal active suicidal ideation). None of the 62 patients without any (passive or active) ictal suicidal ideation planned or attempted suicide either ictally or interictally.

Factors associated with ictal suicidality

Factors associated with an ictal suicidal score ≥ 2 in the univariable analysis were longer disease duration, higher attack intensity, higher HIT score, GAD-7 score, and PHQ-9 score (Table 2). The multivariable analysis showed disease duration, HIT-6, and PHQ-9 were independently associated with a high ictal suicidality score (OR = 1.90 per 10-year increase in disease duration, 95% CI = 1.18–3.05, p = 0.008; OR = 3.19 per 10-point increase in HIT-6, 95% CI = 1.73–5.87, p < 0.001; OR = 2.11 per 10-point increase in PHQ-9, 95% CI = 1.13–3.95, p = 0.020, respectively).

Table 2.

Factors associated with a high (≥2) ictal suicidality score.

	Univariable			Multivariable
	OR	95% CI	p	OR	95% CI	p
Age	1.00	0.98–1.03	0.794
Male sex	1.35	0.55–3.30	0.517
Smoking
Never	(ref)
Ex-smoker	1.10	0.45–2.68	0.834
Current	1.80	0.89–3.61	0.101
Age at onset	0.97	0.95–1.00	0.050
Disease duration, per 10-year increase	1.91	1.29–2.83	0.001	1.90	1.18–3.05	0.008
Total number of lifetime bouts, per 5 increase	1.07	0.96–1.20	0.212
Chronic subtype	1.86	0.45–7.76	0.394
Probable CH	0.79	0.30–2.06	0.632
Current bout
Length of active period	1.01	1.00–1.02	0.133
Attack frequency	1.02	0.87–1.19	0.837
Attack duration	1.00	1.00–1.01	0.093
Attack intensity	1.55	1.14–2.11	0.005
HIT-6 score, per 10-point increase	3.42	2.01–5.84	<0.001	3.19	1.73–5.87	<0.001
GAD-7 score, per 10-point increase	3.16	1.75–5.71	<0.001
PHQ-9 score, per 10-point increase	3.37	1.93–5.91	<0.001	2.11	1.13–3.95	0.015

CH: cluster headache, HIT-6: headache impact test-6; GAD7: general anxiety disorder-7; PHQ9: patient health questionnaire-9.

Factors associated with interictal suicidality

Table 3 shows that younger age at onset, longer disease duration, total number of lifetime bouts, and higher GAD-7 and PHQ-9 scores are significantly associated with interictal suicidality in the univariable analysis. Among these, both the total number of lifetime bouts and the PHQ-9 scores remained significant in the multivariable analysis.

Table 3.

Factors associated with a high (≥2) interictal suicidality score.

	Univariable			Multivariable
	OR	95% CI	p	OR	95% CI	p
Age	0.98	0.91–1.05	0.541
Male sex	1.25	0.15–10.60	0.838
Smoking
Never	(ref)
Ex-smoker	1.97	0.12–32.50	0.636
Current	5.82	0.68–49.69	0.107
Age at onset	0.91	0.83–1.00	0.040
Disease duration, per 10-year increase	2.13	1.11–4.12	0.024
Total number of lifetime bouts, per 5 increase	1.20	1.05–1.38	0.009	1.20	1.01–1.42	0.034
Chronic subtype	1.00	NA*	NA*
Probable CH	0.98	0.11–8.37	0.985
Current bout
Length of active period	1.00	0.97–1.03	0.811
Attack frequency	0.72	0.40–1.30	0.274
Attack duration	1.01	1.00–1.01	0.106
Attack intensity	2.38	0.84–6.75	0.103
HIT-6 score, per 10-point increase	3.29	0.89–12.18	0.075
GAD7 score, per 10-point increase	3.97	1.19–13.28	0.025
PHQ9 score, per 10-point increase	4.62	1.73–12.33	0.002	3.09	1.13–8.44	0.028

No patients with chronic CH had interictal suicidal scores ≥ 2.

CH: cluster headache, HIT-6: headache impact test-6; GAD7: general anxiety disorder-7; PHQ9: patient health questionnaire-9.

In-bout vs. between-bouts suicidality

Of the 175 patients included in this study, 54 were followed up during their between-bouts periods (median 81 days [IQR 34–121] after the baseline visit). The in-bout (ictal and interictal) and between-bouts suicidality of these 54 patients are shown in Figure 2. Suicidality was overall very low outside the bout; specifically, none of the patients reported having passive suicidal ideation or suicidal attempt, whereas only one (1.9%) patient had both active suicidal ideation and suicidal planning.

When compared to the between-bouts period, the CH attacks were associated with significantly higher odds of active suicidal ideation (OR = 37.32, 95% CI = 3.22–433.01, p = 0.004) and increased suicidal planning, although not statistically significant (OR = 3.20, 95% CI = 0.62–16.42, p = 0.163). In contrast, given that passive suicidal ideation and suicidal attempt in the between-bouts period were not reported, the corresponding odds ratios could not be calculated.

Discussion

In this prospective multicenter study, we found a significant increase in passive and active suicidal ideation, suicidal planning, and suicidal attempt during attacks in patients with CH. Furthermore, increased ictal suicidality was suggested to predict interictal suicidality. Interictal suicidality was overall low, although still higher than between-bouts suicidality. In addition to depressive symptoms, longer disease duration and greater headache-related impact were associated with suicidality in CH patients, highlighting the importance of proper management of CH to reduce its burden.

The effect of CH on actual suicidal risk was believed to be low based on a previous study’s finding that the overall risk of suicidal attempt was not higher in CH patients when compared to the general population (6). However, our within-subject comparison demonstrated a significant increase in suicidality during attacks when compared to the interictal and between-bouts periods. In addition, both suicidal planning and attempt showed a twofold increase during attacks. Such an increase in ictal suicidality predicted interictal suicidality. Given that greater headache-related impact was independently associated with ictal suicidality, an intensive treatment to reduce the headache-related impact might be beneficial to prevent suicide in CH patients.

Our study showed the long-term disease burden (i.e. longer disease duration and number of lifetime bouts) was associated with increased suicidality. Specifically, longer disease duration and the number of lifetime bouts were associated with increased ictal and interictal suicidality, respectively. This association was not confounded by chronic CH, probably due to the small number of chronic CH patients in our data, similarly to other Asian studies (13 –16). Rather, our data showed episodic CH with long-term disease burden might be as significant as chronic CH. In addition to reducing the short-term disease burden; that is, headache-related impact and headache intensity, a disease-modifying treatment to prevent further bouts is warranted to decrease suicidality in CH patients.

Our prospective evaluation showed that the between-bouts suicidality further decreased compared to interictal suicidality. Given that the lifetime prevalence of suicidal ideation, planning, and attempt was reported to be 15.2%, 3.3%, and 3.2% in Korean adults and 12.9%, 2.2%, and 2.5% in men, respectively, our CH patients had lower suicidality compared to the general population when they were in the between-bouts periods (17). As psychiatric disorders or chronic pain disorders are not considered as risk factors for CH, our study results suggest that patients remain mentally healthy when the bouts are effectively ended. As patients had relatively increased suicidality during bouts even when they were attack-free (interictally), a strategy to shorten the length of bout is warranted.

In our study, depressive symptoms measured by the PHQ-9 are associated with both ictal and interictal suicidality. Recently, a national claims data-based study showed an association between depression and CH (18). Given that depression is not considered as a risk factor for CH, the increased odds of depression could be a result, rather than a cause, of CH. More attention to depressive symptoms is warranted as they are associated with increased suicidality during and between attacks. For this reason, a further study on depressive symptoms in CH patients is currently ongoing in the KCHR.

Interestingly, suicidality did not differ between patients with definite versus probable CHs. This prompts clinicians for an increased identification and intensive treatment strategy for probable CH, as it carries a suicidality rate similar to that of definite CH. Therefore, the burden of probable CH should not be underestimated. In fact, probable CH might be either a milder (e.g. lower attack frequencies) or more severe (e.g. longer attack durations) end of CH spectrum. It should also be noticed that Asian patients may have more atypical features; that is, a lower mean number of attacks per day and less agitation/restlessness as an accompanying symptom, compared to study results from Europe, which may result in an increased diagnosis of probable CH (13 –15,19 –21).

The strengths of our study are as follows. To our knowledge, this is the first prospective study of detailed suicidality in different phases of CH. Although we collected data based on patients’ retrospective recall, we asked patients only about their active period when they were in active bouts, whereas suicidality of the recall period was followed up during the remission period. The latter allowed patients to focus on their recent remission period and to be less influenced by positive events (i.e. pain experience in CH), which affect the memory more powerfully than do negative events. In addition, an accurate diagnosis of CH was ensured by experienced neurologists who interviewed all the patients. Furthermore, a comprehensive evaluation was performed to fully elucidate our study questions. Our study also has limitations. First, the prevalence of chronic CH was low in our patients. Second, not all patients underwent follow-up visits in the between-bouts period. In fact, it was extremely hard to achieve an optimal follow-up rate in this investigator-initiated study, especially when patients were asymptomatic (i.e. in the between-bouts period). Improvements in the follow-up procedures were made for another sub-study on CH recurrence and its clinical course, for which the follow-up was conducted via telephone calls and e-mails. Third, although we used our best efforts to minimize recall bias and sampling error, as mentioned above, data obtained from recall can be less accurate than prospective recording. Fourth, the questions regarding suicidality were not quantitative: Frequencies of suicidal ideation, plan, or attempt were not taken into consideration. Lastly, we did not collect data on personality disorders, insomnia, or substance abuse or addiction, all of which can influence the suicidality in patients with chronic pain (22).

In conclusion, CH is associated with an increased suicidality during attacks and in within-bout periods. Furthermore, short- and long-term disease burden from CH as well as depressive symptoms contribute to the ictal and interictal suicidality in CH patients. Early diagnosis and effective treatment aimed at both reducing the headache-related burden and preventing further bouts are warranted.

Clinical implications

Cluster headache is associated with an increased suicidality during attacks and within the active cluster period.

Increased suicidality during CH attack is associated with increased suicidality when patients do not have headaches.

Short- and long-term disease burden from CH as well as depressive symptoms contribute to the ictal and interictal suicidality in CH patients.

Footnotes

Acknowledgements

We thank Ms Miran Jung and Ms Jinju Lee for aiding in data management. Sook-Young Woo helped with the statistical analyses used in this study.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by a grant from the Korean Neurological Association (KNA-16-MI-09).

References

Headache Classification Committee of the International Headache Society (IHS). The International Classification of Headache Disorders, 3rd edition. Cephalalgia 2018; 38: 1–211.

Nesbitt

Goadsby

. Cluster headache. BMJ 2012; 344: e2407.

Rossi

Little

De La Torre

, et al. If you want to understand what it really means to live with cluster headache, imagine… fostering empathy through European patients’ own stories of their experiences. Funct Neurol 2018; 33: 57–59.

Horton

. Histaminic cephalgia. Lancet 1952; 72: 92–98.

Jurgens

Gaul

Lindwurm

, et al. Impairment in episodic and chronic cluster headache. Cephalalgia 2011; 31: 671–682.

Rozen

Fishman

. Cluster headache in the United States of America: Demographics, clinical characteristics, triggers, suicidality, and personal burden. Headache 2012; 52: 99–113.

Markley

Buse

. Cluster headache: Myths and the evidence. Curr Pain Headache Rep 2006; 10: 137–141.

Headache Classification Committee of the International Headache Society (IHS). The International Classification of Headache Disorders, 3rd edition (beta version). Cephalalgia 2013; 33: 629–808.

Seo

Park

. Validation of the Patient Health Questionnaire-9 (PHQ-9) and PHQ-2 in patients with migraine. J Headache Pain 2015; 16: 65.

10.

Seo

Park

. Validation of the Generalized Anxiety Disorder-7 (GAD-7) and GAD-2 in patients with migraine. J Headache Pain 2015; 16: 97.

11.

Posner

Brown

Stanley

, et al. The Columbia-Suicide Severity Rating Scale: Initial validity and internal consistency findings from three multisite studies with adolescents and adults. Am J Psychiatry 2011; 168: 1266–1277.

12.

Pai

Woo

J-M

Son

, et al. The reliability and validity of the Korean version of Columbia-Suicide Severity Rating Scale in alcohol dependent patients. Journal of Korean Neuropsychiatric Association 2015; 54: 222–227.

13.

Lin

Wang

Fuh

, et al. Cluster headache in the Taiwanese – a clinic-based study. Cephalalgia 2004; 24: 631–638.

14.

Dong

Dai

, et al. Clinical profile of cluster headaches in China – a clinic-based study. J Headache Pain 2013; 14: 27.

15.

Moon

Park

Lee

, et al. Clinical features of cluster headache patients in Korea. J Korean Med Sci 2017; 32: 502–506.

16.

Imai

Yagi

Kuroda

, et al. Clinical profile of cluster headaches in Japan: Low prevalence of chronic cluster headache, and uncoupling of sense and behaviour of restlessness. Cephalalgia 2011; 31: 628–633.

17.

Jeon

Lee

, et al. Lifetime prevalence and correlates of suicidal ideation, plan, and single and multiple attempts in a Korean nationwide study. J Nerv Ment Dis 2010; 198: 643–646.

18.

Liang

Chen

Fuh

, et al. Cluster headache is associated with an increased risk of depression: A nationwide population-based cohort study. Cephalalgia 2013; 33: 182–189.

19.

Bahra

May

Goadsby

. Cluster headache: A prospective clinical study with diagnostic implications. Neurology 2002; 58: 354–361.

20.

Schurks

Kurth

de Jesus

, et al. Cluster headache: Clinical presentation, lifestyle features, and medical treatment. Headache 2006; 46: 1246–1254.

21.

Gaul

Christmann

Schroder

, et al. Differences in clinical characteristics and frequency of accompanying migraine features in episodic and chronic cluster headache. Cephalalgia 2012; 32: 571–577.

22.

Tang

Crane

. Suicidality in chronic pain: A review of the prevalence, risk factors and psychological links. Psychol Med 2006; 36: 575–586.