Abstract
Background and objective
There are five headache disorders composing the trigeminal autonomic cephalalgias (cluster headache, paroxysmal hemicrania, short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing (SUNCT), short-lasting unilateral neuralgiform headache attacks with cranial autonomic symptoms (SUNA), and hemicrania continua). Little is known about these disorders in the pediatric population. The objectives of this study are to report the full age ranges of pediatric trigeminal autonomic cephalalgias and to determine if pediatric-onset trigeminal autonomic cephalalgias display similar signs and symptoms as adult onset.
Methods
Search criteria in Medline Ovid, Embase, PsycINFO, and Cochrane Library were created by a librarian. The remainder of the steps were independently performed by two neurologists using PRISMA guidelines. Inclusion criteria for titles and abstracts were articles discussing cases of trigeminal autonomic cephalalgias with age of onset 18 or younger, as well as any epidemiological report on trigeminal autonomic cephalalgias (as age of onset data was often found in the results section but not in the title or abstract). Data extracted included age of onset, sex, and International Classification of Headache Disorders criteria for trigeminal autonomic cephalalgias (including pain location, duration, frequency, autonomic features, restlessness) and some migraine criteria (photophobia, phonophobia, and nausea). Studies that did not meet full criteria for trigeminal autonomic cephalalgias were examined separately as “atypical trigeminal autonomic cephalalgias”; secondary headaches were excluded from this category.
Results
In all, 1788 studies were searched, 86 met inclusion criteria, and most (56) examined cluster headache. In cluster headache, onset occurred at every pediatric age (range 1–18 years) with a full range of associated features. Autonomic and restlessness features were less common in pediatric patients, while migrainous features (nausea, photophobia, and phonophobia) were found at similar rates. The sex ratio of pediatric-onset cluster headache (1.8, 79 male and 43 female) may be lower than that of adult-onset cluster headache. Data for other trigeminal autonomic cephalalgias, while more limited, displayed most of the full range of official criteria. The data for atypical trigeminal autonomic cephalalgias were also limited, but the most common deviations from the official criteria were abnormal frequencies and locations of attacks.
Conclusions
Trigeminal autonomic cephalalgias can start early in life and have similar features to adult-onset trigeminal autonomic cephalalgias. Specifically, pediatric-onset cluster headache patients display the full range of each criterion for cluster headache (except maximum frequency of six instead of eight attacks per day). However, cranial autonomic features and restlessness occur at a lower rate in pediatrics. Additional information is needed for the other trigeminal autonomic cephalalgias. As for expanding the ICHD-3 criteria for pediatric-onset trigeminal autonomic cephalalgias, we have only preliminary data from atypical cases, which suggests that the frequency and location of attacks sometimes extend beyond the official criteria.
Keywords
Introduction
The trigeminal autonomic cephalalgias (TACs) are a group of five primary headache disorders: Cluster headache, paroxysmal hemicrania, short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing (SUNCT), short-lasting unilateral neuralgiform headache attacks with cranial autonomic symptoms (SUNA), and hemicrania continua (1). They share features of unilateral location and ipsilateral facial autonomic features such as conjunctival injection, nasal congestion, and ptosis. However, they differ in duration, frequency, and treatments (Table 1).
Comparison of the trigeminal autonomic cephalalgias. The official criteria from the International Classification of Headache Disorders 3rd edition (ICHD-3) (1) are shown. The primary differences are duration (Criterion B), frequency (Criterion B for hemicrania continua and criterion D for all others), and first-line treatments.
*SUNCT and SUNA differ in only one characteristic: SUNCT has both conjunctival injection and lacrimation ipsilateral to the pain, while SUNA has one or neither of these features.
**General population data are reported as data in pediatrics are unknown.
***First line treatments are based on adults as first line treatments in pediatrics are unclear.
SUNCT: short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing; SUNA: short-lasting unilateral neuralgiform headache attacks with cranial autonomic symptoms.
Reports from clinical reviews and expert opinions often note that the TACs typically begin in adulthood and are exceptionally rare in pre-teens (2–7). However, eight epidemiologic studies (of 200–808 patients each) report an age range that includes pre-teens (8–15). In cluster headache, the most common and best studied of the TACs, there is typically a delay in diagnosis of several years, and one of the contributing factors is lower age at onset (10,16). A possible reason for delay in diagnosis at a younger age is atypical headache features (migraine, for example, can have a shorter duration in pediatric patients (1)). Another possible reason is a lack of awareness of cluster headache in younger patients. The objectives of this systematic review and meta-analysis are to report the full age ranges of pediatric TACs and to determine if pediatric-onset TACs display similar signs and symptoms to adult-onset TACs.
Methods
Eligibility criteria: Inclusion criteria were established a priori and included all study types that reported i) any trigeminal autonomic cephalalgia and ii) any study that mentioned either a) onset of age 18 years or younger, or b) any epidemiological report (as we found in our initial research that epidemiology studies often mentioned age of onset data in the results section). We included studies that mentioned an average age of onset ± standard deviation as long as the lower end of the standard deviation was age 18 or younger. Exclusion criteria were non-English articles, letters to the editor and similar articles that may have less strict peer review, and articles before 1988. We chose 1988 as the starting date because the first edition of the International Classification of Headache Disorders (ICHD) was published in 1988 (17). For SUNCT and hemicrania continua, which are not listed in the first edition of the ICHD, we allowed the original descriptions by Sjaastad and colleagues (18,19). For SUNCT and SUNA, we also allowed the criteria listed in the appendix of the second edition of the ICHD (20), as they were not listed in the main text.
Some articles did not meet full ICHD criteria such as probable TACs and secondary TACs. These were analyzed separately from the rest of the articles. In this group, we excluded secondary TACs and considered the remainder to be “atypical TACs”. Analysis of the atypical TACs is listed at the end of the results section.
Information sources and search: An experienced librarian (ES) developed searches for Medline Ovid, Embase, PsycINFO, and Cochrane Library. The study was registered with PROSPERO (registration number CRD42020165256), and we followed Guidelines for the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) (21). In general, search terms focused on i) trigeminal autonomic cephalalgias and ii) age of onset and age terms such as child, youth, adolescent, elementary school, middle school, and high school (Medline Ovid search terms are available in Supplemental Table 1). The main search was conducted in Medline Ovid using MeSH Headings as well as equivalent keywords and phrases. These terms were then tested for relevancy and the main search was finalized in Medline Ovid on 2/20/2020. These search terms were then translated into Embase, PsycINFO, and Cochrane Library on 2/20/2020.
Study selection: Screening was performed independently by two neurologists (AG and MJB). In the screening step, we examined titles and abstracts in Rayyan (22) and applied our inclusion and exclusion criteria. In the eligibility step, we examined full-length articles for inclusion and exclusion criteria. At the end of each stage, disagreements were settled by discussion between the two neurologists. In all cases, a consensus was reached.
Data collection process, data items: Two authors (AG and MJB) independently extracted data elements from each article. Data extracted included the author names, article title, journal name, publication year, country of authors, country of patients, source of patients, type of TAC, sex, age at the time of interview, and age of onset. Detailed data for associated features was also extracted including attack duration, attack frequency, headache location, conjunctival injection, lacrimation, nasal congestion, rhinorrhea, eyelid edema, forehead/facial sweating, miosis, and ptosis, as well as restlessness/agitation, nausea, photophobia, phonophobia, and a history of primary or second-hand tobacco use. For quality assessment, the two authors independently assessed the method of diagnosis and excluded all that did not meet ICHD criteria. For additional quality assessment, we proposed in PROSPERO to use the Newcastle–Ottawa appraisal checklist (23) but found that it was not applicable in its current form as it evaluates cohort studies; instead we used the NIH Quality Assessment Tool for Case Series Studies (24) for all studies involving two or more patients (Supplemental Table 2). We rated all studies of two or more patients as Fair or Good and thus included all of these in the meta-analysis.
Study measures and synthesis of results: Summary measures were age of onset as well as sex and associated features (attack duration, attack frequency, headache location, conjunctival injection, lacrimation, nasal congestion, rhinorrhea, eyelid edema, forehead/facial sweating, miosis, and ptosis, restlessness/agitation, nausea, photophobia, phonophobia). Studies that reported only age range or average ± standard deviation were not included in the meta-analysis because of the lack of detailed data. Associated features were included in the meta-analysis only for case series of five or more patients because there was substantial missing data for the case reports and smaller case series.
Statistical measures: To examine sex and age, Fisher’s exact test was calculated in Stata version 14.2 (StataCorp, College Station, TX). For missing data, one study (25) was not included in the statistical calculation because sex was not provided: Ages of onset were 17 and 18 years.
Results
Our search identified 1780 records prior to de-duplication, with eight additional studies added from our general review of the literature (Figure 1). Our dataset included patients from 24 countries and five continents (excluding Africa and Antarctica). Our search seemed reasonably inclusive as it captured 46 of the 47 studies identified in six prior narrative reviews (26–31) and the one missing study (32) met an exclusion criterion (published before 1988). The remaining 46 identified studies included 28 cluster headache studies (of these 14 (8,9,29,31,33–42) were included in the final dataset based on our inclusion criteria), 12 paroxysmal hemicrania studies (of these five (43–47) were included in the final dataset), five SUNCT/SUNA studies (of these four (48–51) were included in the final dataset), and one hemicrania continua study (52) (which was included in the final dataset).
Flow diagram for identification, screening, eligibility, and inclusion. When assessing for eligibility, duplicates include not only duplicate articles, but also use of the same dataset in different articles.
Before we began our systematic review, we noted that some types of articles, particularly epidemiological studies, mentioned age of onset data in the results but not the title or abstract. As we screened articles only by title and abstract, we deliberately reviewed the full text of a substantial proportion of articles (n = 380, or 30% of the 1252 records identified after de-duplication) because we wanted to search the full text of epidemiological studies. Ultimately, we identified 86 studies for our systematic review, including four (42,53–55) of the eight studies added from our general review of the literature. Most focused on a specific TAC, though SUNCT and SUNA were often grouped together. Cluster headache was the most common diagnosis (n = 56, or 65% of the final 86 articles).
Age of onset data was reported in several forms, and each is discussed as a separate part: i) individual patient data of patients with a full ICHD diagnosis, with sufficient information available for a meta-analysis; ii) grouped patient data with a full ICHD diagnosis that lists age range or an average ± standard deviation where the standard deviation includes age ≤ 18, and iii) individual patient data without a full ICHD diagnosis (“atypical TACs”).
Part 1: Studies with individual patient data and full ICHD-3 criteria (n = 40). Pediatric age of onset data is presented in Table 2.
Age of onset and sex data for pediatric-onset cluster headache. Additional information is listed in Supplemental Table 3, including and associated features for the studies listed here.
*11 patients were described in this study and three were excluded as they did not meet International Classification of Headache Disorders (ICHD) criteria.
**Cluster headache criteria state a headache duration of 30–120 min, while paroxysmal hemicrania state a duration of 2–30 min and frequency more than five times per day, so some of these patients’ headaches meet full ICHD criteria while others do not.
***Headaches occurred less than once per day, such as 3–6 days per week (0.5–0.9) or every other day (0.5).
****Though the frequency suggests the headaches were not daily, the authors report that the child fulfilled all diagnostic criteria.
UK: United Kingdom; USA: United States of America; NR: not reported.Bolded rows are summary.
Cluster headache: Cluster headache had an onset as young as 1 year old with many pre-adolescent patients. Pediatric-onset cluster headache patients (n = 124) had pain in all typical locations, with a full range of durations (15–180 minutes), but a lower frequency (between every other day and six per day, official criteria is between every other day and eight per day).
We also investigated the frequency of positive and negative cranial autonomic features in pediatric-onset cluster headache. However, to avoid bias, we restricted our analysis to studies of five or more patients (29,37,38,40,56) because for small studies of 1–4 patients, it was typical to list only positive findings. All the autonomic and restlessness features were seen for pediatric-onset cluster headache (Supplemental Table 3). Cluster headache was most commonly associated with lacrimation in 81% (68/84), followed by restlessness in 63% (24/38), conjunctival injection in 61% (51/84), nasal congestion in 57% (48/84), ptosis in 52% (44/84), rhinorrhea in 45% (38/84), facial sweating in 16% (6/38), miosis in 14% (10/73), and eyelid edema in 3% (1/38). Cluster headache patients often displayed migrainous symptoms, and pediatric onset patients most commonly display photophobia in 58% (49/84), nausea in 37% (27/73), and phonophobia in 29% (24/84).
The sex ratio in pediatric-onset cluster headache patients was 1.8 (79 male, 43 female). There were no significant differences between males and females for the proportion of pre-adolescents versus adolescents (p = 0.374, see Supplemental Table 4). Both male and female patients displayed a wide range of ages (Supplemental Figure 1). However, the data are too limited to comment on a particular age at which patients are at risk: The largest study (29), making up 44% (38/87) of the patients, only examined patients under 14 years old and thus is a potential confounder for the decrease in pediatric onset for ages 14–18.
We also attempted to extract data on tobacco use or second-hand smoke in patients with pediatric onset, as tobacco use is associated with cluster headache and may be a risk factor (57,58). Unfortunately only two articles reported tobacco history (29,40). In one study, six of nine patients had a parent that was a smoker (and two patients could not be contacted) (40), and in the other study 20 of 38 patients were smokers as adults (29).
Paroxysmal hemicrania: Pediatric onset was found for 11 patients across six studies, with a range of onset between 2–14 years old. For official criteria, patients displayed the full range of locations, durations, and frequencies. At least one patient had every cranial autonomic and restlessness symptom except rhinorrhea. The sex ratio was 4:7 for paroxysmal hemicrania (four male, seven female).
SUNCT: Pediatric onset was found for six patients across six studies, with a range of onset between 2–17 years old. For official criteria, patients displayed the full range of locations with the exception of the mandibular division of the trigeminal nerve, with a nearly full duration of 5–600 seconds (official duration is 1–600 seconds), and a slightly increased frequency of 3–200 attacks per day (official frequency is at least one per day). By definition, all patients had conjunctival injection and lacrimation, and at least one patient had every other cranial autonomic symptom with the exception of facial sweating and miosis. The sex ratio was 1:1 for SUNCT (three male, three female).
SUNA: Pediatric onset was found for three patients across three studies, with a range of onset between 1–18 years old. For official criteria, location was restricted to orbital and temporal (not supraorbital and not the maxillary or mandibular divisions of the trigeminal nerve), duration restricted to 20–300 sec (official duration is 1–600 sec), and frequency restricted to 10–30 per day (official frequency is at least one per day). At least one patient had every cranial autonomic symptom except eyelid edema, facial sweating, miosis, and ptosis. The sex ratio was 1:2 for SUNA (one male, two female).
Hemicrania continua: Pediatric onset was found in two patients across two studies, ages 8 and 12. For official criteria, by definition both patients had unilateral and constant pain, and at least one patient had every cranial autonomic symptom and restlessness except nasal congestion, eyelid edema, facial sweating, and miosis. Hemicrania continua patients are known to have migrainous symptoms but the published data are limited: Migrainous data were not reported in either case with the exception of nausea in one case. The sex ratio was 1:1 for hemicrania continua (one male, one female).
Part 2: Studies with full ICHD-3 criteria that list age range or an average ± standard deviation (n = 35). Thirty-five studies reported either an age range that included pediatric onset, or an average and standard deviation where the lower end of the standard deviation included 18 years or younger (Supplemental Table 5). We consider the age range data more accurate, as the average and standard deviation may not actually include pediatric-onset, but for completeness we chose to include all the studies. For cluster headache, the lower end of the age range varied from 3–18 years old across 21 studies. For paroxysmal hemicrania, the lower end of the age range was five in one study while for SUNCT and SUNA, the lower end of the age range varied from 13–16 years of age across two studies. Finally, in hemicrania continua, the lower end of the age range varied between 6–10 years of age across two studies.
Part 3. Atypical TACs (studies of individual patient data that do not meet full ICHD-3 criteria, n = 11). We searched for atypical TACs to identify possible differences in features between pediatric and adult TACs. Our reasoning was that another primary headache, migraine, differs in its duration between pediatrics (2–72 h) and adults (4–72 h) (1). We identified 11 studies of atypical TACs that were not due to secondary causes. Atypical studies were found for all TACs except SUNA (Supplemental Table 6). The most common finding in atypical TACs was an abnormal frequency of attacks (n = 6), followed by an abnormal location of pain (n = 5), lack of an indomethacin trial (n = 3), lack of autonomic/restlessness features (n = 2), lower intensity of pain (n = 1), and failure of an indomethacin trial (n = 1).
Discussion
The current literature on pediatric onset, based on 86 studies identified in our systematic search, suggests that TACs can start early in life, with the youngest documented cases at 1 year old for cluster headache and SUNA, 2 years old for paroxysmal hemicrania and SUNCT, and 6 years old for hemicrania continua. In cluster headache, where there is more published data, onset has been documented at every pediatric age. In the other TACs, where there are few published reports, the data do not suggest a particular pediatric age for which patients are at risk. In our dataset, we excluded studies that only reported age of onset above 18 years old. Thus, we cannot comment on the prevalence of pediatric-onset TACs, only on the existence of pediatric-onset TACs. Previous large epidemiologic studies (1000 patients or more) have been performed on cluster headache, and they differ on the prevalence of pediatric onset. In an American study, 35% of 1134 patients had onset 20 years or younger (59). In a Dutch study, the average age of onset was 32 ± 14 years in 1163 patients (54), but in an international study the average age of onset was 27 ± 13 years in 1604 patients (60). Additional large studies or systematic reviews are needed.
The pediatric-onset data is similar to the adult data in several respects. In cluster headache, all the features of the official ICHD-3 criteria were present in at least one patient with the exception of a frequency of 1–6 attacks per day instead of 1–8 attacks per day. Data for the other TACs met most but not all ICHD-3 criteria, but our search found very few studies that examined these disorders. Overall, these data suggest that the full range of each ICHD-3 criterion is appropriate for pediatric-onset cluster headache patients, and additional information is needed for the other TACs. As for expanding the ICHD-3 criteria for pediatric onset TACs to include a wider range of durations, frequencies, or other features, we have only preliminary data from a review of 11 articles on atypical TACs. Our preliminary data suggests that the frequency and location of attacks sometimes extend beyond the official criteria in pediatric-onset TACs.
The pediatric-onset data also differs from the general literature on trigeminal autonomic cephalalgias in several respects. We will focus on cluster headache, because the reason for the differences in the other TACs may simply be limited data. First, cranial autonomic symptoms and restlessness appear to be less common in pediatric-onset cluster headache. In pediatric-onset cluster headache the most common associated symptoms were lacrimation (81%) and restlessness (63%). In the general cluster headache population (using studies of >1000 participants), these features are more common: Lacrimation in 91% and restlessness in 76–99% (54,59,60). Interestingly, migrainous symptoms may have a similar or higher rate: Pediatric-onset cluster headache is associated with nausea in 37%, photophobia in 58%, and phonophobia in 29%. In large studies of migrainous features of cluster headache, nausea is found in 18–50%, photophobia in 12–56%, and phonophobia in 5–43% (8,59,61–63). Additional studies are needed given the limited amount of information for pediatric-onset cluster headache. However, cluster headache might be unrecognized or misdiagnosed as migraine in pediatric patients because of less cranial autonomic features and restlessness, but similar rates of migrainous features. Furthermore, migraine patients are known to have cranial autonomic features (64), thus cranial autonomic and migrainous features are poor distinguishing characteristics between cluster headache and migraine.
Cluster headache patients who have delays in their diagnosis are often misdiagnosed with migraine (16). Shared autonomic and migrainous features, however, are not the only consideration. Unlike adults, where there is no overlap in attack duration between cluster headache (between 15 min and 3 h) and migraine (4–72 h), in pediatric patients a migraine attack can be as short as 2 h (1). We recommend to clinicians that attack duration is still quite useful diagnostically when less than 2 h (more consistent with cluster headache) or more than 3 h (more consistent with migraine). However, when the attack duration is 2–3 h, a detailed assessment of other features is critical to distinguish cluster headache from migraine. Those other features include attacks per day (1–8 per day for cluster headache, one per day for migraine), restlessness (present for cluster headache, often absent for migraine where motion aggravates the headache), and the location of pain (strictly unilateral for cluster headache, unilateral or bilateral for migraine) (7).
Also in cluster headache, the sex ratio of 1.8 in pediatric-onset patients is much lower than the reported 4.3 in the general population (65). It should be noted that the true sex ratio is unclear in cluster headache: The sex ratio has been decreasing over the years, with large studies showing ratios of 3.7 (913:250) (54), 2.6 (816:318) (59), 2.2 (1104:497) (60), and 1.3 (4356:3233) (66). However, if the sex ratio is indeed lower in patients with pediatric onset, it suggests sex-specific genetic, epigenetic, or environmental triggers that lead to an increased prevalence of cluster headache in males. One possibility is hormonal triggers that occur during puberty, similar to migraine, where the sex ratio of migraine increases in females after puberty possibly because of estrogen (67–69). Secondary causes of cluster headache are known to be caused by pituitary abnormalities, especially prolactinomas (70–72). Unfortunately, we have insufficient numbers in this dataset to determine if the incidence of cluster headache onset differs after puberty. Larger studies are needed to determine if there is a difference in sex ratio between pre-pubertal pediatric onset, post-pubertal pediatric onset, and adult-onset cluster headache.
This study has several limitations. First, we assumed that articles that made the diagnosis of TACs using the ICHD criteria in fact were valid diagnoses, even if all the specific criteria were not mentioned in the article. Second, the meta-analysis for all TACs, in particular the non-cluster headache TACs, must be interpreted with caution given the small number of studies. We chose to proceed with a meta-analysis as data in general for pediatric-onset TACs are limited, and meta-analysis is useful in summarizing the existing literature. Third, the evaluation of probable TACs is likely an incomplete evaluation of atypical and undifferentiated TAC-like headaches in pediatrics as our search criteria may not have been mentioned in the title or abstract, especially if the headaches resemble migraines. Additionally, it is reasonable to think that the full range of atypical or undifferentiated TAC-like headaches are not reported in the literature and require a methodology other than a systematic review (such as an observational study of atypical TACs). Finally, we may have excluded articles during the screening step because the age of onset data was available in the main text but not in the abstract or title. We attempted to review the full text of many epidemiological studies that did not mention age of onset in the abstract or title, but a more comprehensive search of entire articles may identify additional studies.
In summary, the current literature strongly suggests that trigeminal autonomic cephalalgias can start very early in life. Pediatric-onset cluster headache has similar features to adult-onset cluster headache but may be confused with migraine because of lower rates of cranial autonomic features and a similar rate of prototypical migrainous features.
Clinical implications
All five trigeminal autonomic cephalalgias can start early in life, including in pre-teens. The data is strongest for cluster headache. Pediatric-onset cluster headache has similar features to adult-onset cluster headache but may be confused with migraine because of lower rates of cranial autonomic features and a similar rate of prototypical migrainous features. Pediatric cluster headache can be distinguished from pediatric migraine by attack duration when this is less than 2 h (more consistent with cluster headache) or more than 3 h (more consistent with migraine). When the attack duration is 2–3 h, pediatric cluster headache can be distinguished by the number of attacks per day (more than one), presence of restlessness, and the strictly unilateral (i.e. never bilateral) pain.
Footnotes
Acknowledgements
We thank the Texas Medical Center Library for assistance with the systematic review search.
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: Dr. Burish receives funding from the Will Erwin Headache Research Foundation.
