Migraine-like headaches: a systematic review of secondary etiologies, clinical features and therapeutic patterns (1977–2024)

Abstract

Introduction

Secondary headaches presenting similar clinical findings to migraine are known as migraine-like headaches. Currently, there is no broad consensus regarding their prevalence, characteristics, or underlying causes. Existing evidence is based primarily on descriptive studies and a single systematic review, with vascular, traumatic, and epileptic seizures as the most frequent causes. The aim of this systematic review was to describe and analyze secondary etiologies behind migraine-like headaches, including their epidemiological, clinical, and therapeutic characteristics.

Methodology

A database search of studies between 1977 and 2024 was conducted in February 2025 on EBSCO, SCOPUS, PubMed, and Web of Science using the terms “migraine-like” or “migraine mimic”. Included articles were human observational studies and case reports that documented a migraine-like headache with a diagnosed underlying cause. Articles lacking explicit causality, as well as review and experimental studies were excluded. Data on demographics, clinical features, etiology, imaging, and treatment were extracted and analyzed. Etiologies were categorized into vascular, neoplastic, traumatic, epileptic, autoimmune, infectious and others. Sensitivity analysis (SA) was performed using low-bias concern studies.

Results

Of 745 studies retrieved, 220 met inclusion criteria, totaling 4,422 patients (mean age 33.7 years, 63.4% female). Most presented severe (68.8%), unilateral (66.9%), gradual onset (56.6%) headaches with aura (52.5% and 47.8% in SA), primarily localized to frontal (29.3%) and temporal (22.8%) regions. Stroke (6.4% and 7.8% in SA), epilepsy (7.3%), and traumatic brain injury (5%) were the most frequent causes, while NSAIDs and triptans were the most common treatments (30.5 and 18.2%). After SA the associations between younger age and parietal pain (p = 0.023) and between imaging modality and etiology (p < 0.001) remained significant, where magnetic resonance imaging was predominantly used for autoimmune diseases and post-traumatic headaches frequently lacked imaging. Sudden onset and post-traumatic headaches were identified as frequent red flags.

Conclusion

The present study findings underscore the importance of recognizing red flags in migraine-like headaches and early identification of underlying causes. Severe, sudden onset headaches alongside systemic symptoms or trauma history should prompt thorough evaluation. Future research should aim to refine the definition of migraine-like headache and to improve diagnostic precision and treatment timing in these patients.

Supplementary Information

The online version contains supplementary material available at 10.1186/s10194-025-02151-8.

Background

Secondary headaches that portray clinical migraine characteristics are known as migraine-like headaches (also referred to as symptomatic migraine or migraine mimic) [1–3]. These patients are typically treated as primary migraine initially, only to be diagnosed with an underlying cause later. Therefore, careful clinical evaluation and imaging studies are crucial in these cases [4].

The prevalence of secondary headaches is variable and influenced by population characteristics. While studies in emergency settings report a 2–7% prevalence amongst headache patients, the World Health Organization estimated that 18% of headaches are due to secondary causes [5, 6]. In terms of migraine-like headache specifically, there is no broad consensus regarding prevalence, clinical characteristics, and underlying pathologies, as evidence is mainly based on case reports and small observational studies with extremely diverse reporting qualities. The most frequently reported secondary causes of migraine-like headache include arteriovenous malformations (AVMs), traumatic brain injury (TBI), stroke, epilepsy, cerebral aneurysms, intracranial hemorrhages, arachnoid cysts and intracranial neoplasms [2, 3, 7–20].

As per our current knowledge, this area of research is supported by a solitary systematic review published in 2011, in which a working criteria for migraine-like headache was designed based on the International Classification of Headache Disorders, 3rd edition (ICHD-3) criteria for primary migraine and secondary headaches [21]. Only studies which strictly adhered to the proposed definition of “symptomatic migraine” or “possible symptomatic migraine” were included. Therefore, only 11 cases fulfilled definite criteria, and 39 patients were identified as “possible”. All definite cases reported aura characteristics in their patients and identified five AVMs, two cases of carotid artery stenosis, one case of carotid artery dissection, one case of carotid artery aneurysm, one case of parietal metastasis and one case of meningioma as underlying pathologies This strict approach ensured evidence of causality between the headache and the secondary cause. Yet, ICHD-3 criteria also encourage the attribution of both migraine-like headache and primary migraine diagnosis to patients previously known with primary migraine that later presented causal evidence of another disorder [1].

Accurate identification of individuals with migraine-like headache has broad therapeutic and prognostic implications. Unlike primary migraine, secondary headaches require addressing the underlying cause, which is potentially curable in many cases with the aid of surgery and specific therapies [2, 3, 18]. The main clinical challenge lies in distinguishing primary migraine from secondary migraine-like headaches. For this purpose, the identification of “red flag” characteristics and the use of neuroimaging studies, particularly magnetic resonance imaging (MRI), are of utmost importance [4, 5, 22, 23].

Although many observational studies and case reports pertaining to migraine-like headaches have been put forward, there is still no broad consensus regarding its prevalence, clinical characteristics, causes and treatment. A more accurate clinical definition and therapeutic proposal for migraine-like headaches based on review studies is needed. Therefore, the purpose of this systematic review is to describe underlying causes, epidemiological, clinical, and therapeutic characteristics of individuals with migraine-like headache and to investigate statistically significant findings.

Methodology

In accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyzes (PRISMA) guidelines, this systematic review was registered in PROSPERO under the number CRD420251054061.

Search strategy and eligibility criteria

The keywords used were "migraine-like" OR "migraine mimic" in title or abstract and the databases searched were EBSCO, SCOPUS, PubMed, and Web of Science. Inclusion criteria: 1) observational studies and case reports 2) reporting human patients diagnosed with migraine-like headache with an identified underlying cause 3) published articles from January 1 st, 1977 to December 31 st, 2024, 4) written in English or Spanish. The articles excluded were those reporting individuals without migraine-like headache and individuals without explicitly mentioned causality between headache and etiology, as well as narrative reviews, systematic reviews, meta-analyses, letters to the editor, clinical trials, and experimental studies. Filters were applied to restrict results according to the inclusion criteria. The complete search strings for each database are provided in supplementary material, Table 1.

Study selection and data extraction

The study selection process is depicted in the PRISMA flowchart (Fig. 1). Duplicate studies were removed, and the remaining studies were screened by two authors by reviewing title and abstract according to the inclusion and exclusion criteria and discrepancies were solved by consensus of both parties. Due to the highly variable degree of headache description, studies were considered for inclusion whenever a causal relationship between a recognized migraine-like headache and a diagnosed underlying cause was explicitly stated, even if headache characteristics were not broadly detailed. Studies excluded were due to study subjects not being humans, not meeting criteria for migraine-like headache (ICHD-3 primary migraine and secondary headaches) whenever headache characteristics were stated, not observational or case reports, not in English or Spanish [1]. A team of six researchers performed data extraction. Extracted data were centered on the number of patients included, number of individuals with migraine-like headache, comorbidities reported, headache characteristics (severity, duration, onset, location, laterality and presence of aura), as well as type of brain imaging performed, whether the patients required cranial surgery, etiology of migraine-like headache and number of deceased patients during study. The characteristics of the study and the study population such as author, year of publication, country, mean age of patients, sex distribution and type of study, was also extracted. The data were tabulated, coded, and imported into a datasheet for analysis. For subsequent statistical analysis, etiologies were grouped into seven distinct categories: Vascular, Neoplasms, Autoimmune, Infectious, Epilepsy, Post-traumatic and Others.

Fig. 1.

PRISMA flowchart for study selection

Quality and evaluation

Two independent reviewers performed a quality analysis using the Joanna Briggs Institute (JBI) Cohort/Case–Control Tools for observational studies and the Case Reports/Series Critical Appraisal Checklist. Each tool includes items addressing study selection, reporting, validity, and outcome assessment. Items were rated as “Yes,” “No,” “Unclear,” or “Not applicable.” To enable sensitivity analysis, we categorized studies into levels of bias concern according to the proportion of checklist items fulfilled: low bias concern when ≥ 70% of items were met, moderate bias concern when 50–69% were met, and high bias concern when < 50% were met. Discrepancies were resolved by consensus. No studies were excluded based on this analysis, but rather a sensitivity analysis was performed and reported alongside full analysis. Meta-analyses could not be performed due to the heterogeneity of extracted data and reporting methods.

Statistical analysis

Statistical analysis was performed by two researchers, using SPSS ver. 26.0 and RStudio ver. 2025.05.1 + 513. Missing data were managed by complete-case analysis; only observations with available data for the variables of interest were included in the analyses. No imputation methods were applied. Shapiro–Wilk test was used to identify normally distributed variables. Quantitative variables were expressed through mean and standard deviation for normally distributed variables and median and IQR for non-normally distributed variables. Likewise, qualitative variables were represented as frequency and percentage. In terms of correlation, quantitative variables were assessed through Pearson’s correlation for normally distributed variables and Spearman’s Rank correlation for non-normally distributed variables. Associations between categorical variables were analyzed through chi-square tests, likelihood ratio tests, Fisher’s tests, and Cramér’s V when appropriate. On the other hand, differences were compared using ANOVA, Welch’s ANOVA, Kruskal–Wallis H and Mann–Whitney U tests and post-hoc analysis included Tukey’s HSD, Bonferroni Correction and Dunn’s test whenever appropriate. Finally, sensitivity analysis was performed using the low-bias concern subset for confirmation of full-analysis descriptive and inferential results.

Results

The database search was performed on February 12th, 2025, and retrieved 745 studies (Fig. 1). Following screening based on title and abstract, 624 studies were reviewed in full text. After exclusion, 220 studies were included, comprising a total of 4,422 individuals with migraine-like headache with a mean age of 33.7 years (SD 16.3). 137 articles were identified as case reports (62.3%) and 83 as observational studies (37.7%). In these articles, both median sample size and median number of individuals with migraine-like headache was 2 (IQR 1–56, IQR 1–10 respectively) (Table 1). Across the 220 included studies, comorbidity data were explicitly reported in 190 (86.4%), imaging findings in 191 (86.8%), and treatment or medication use in 105 (47.7%). Clinical features such as headache onset were reported in 136 studies (61.8%), intensity in 141 (64.1%), aura in 171 (77.7%), location in 92 (41.8%), and laterality in 124 (56.4%). Whether patients required invasive treatment such as surgical interventions was described in 193 studies (87.7%). A full summary of the 220 included studies is provided in supplementary material, Table 2.

Table 1.

Overview of included studies

Study type	N of studies (%)	N of individuals with migraine-like headache (%)	Bias concern n (%)	Main data reported
Case reports	137 (62.3%)	287 (6.5%)	Low: 82 (59.9%) Moderate: 28 (20.4%) High: 27 (19.7%)	Mostly clinical features, rare etiologies
Observational studies	83 (37.7%)	4,135 (93.5%)	Low: 34 (41.0%) Moderate: 31 (37.3%) High: 17 (20.5%)	Comorbidities, imaging, treatments and vascular etiologies
Total	220 (100%)	4,422 (100%)	-	-

Additionally, most studies achieved low bias concern scores (52.7%, n = 116), 60 were marked as moderate concern (27.3%), and only 44 studies were identified as high concern (20%) (Fig. 2).

Fig. 2.

Proportion of bias concern levels among studies. Bias concern level: 1-low concern, 2-moderate concern, 3-high concern

In terms of sex distribution between studies, most patients were female, with a median percentage of migraine-like headache affected females of 63.4% (IQR 0–100%). 1,198 patients described presence or absence of previous history of migraine with a median percentage of 57.7% having previous primary migraine (IQR 0–100%). Out of 3,720 patients with explicitly reported comorbidities, only a few patients reported a previous history of hypertension (11.6%, n = 432) and type 2 diabetes (6.8%, n = 253), while the majority had neither (74.1%, n = 2,755). Only eight studies reported deaths occurring during the study period, including eight cases in a study regarding epileptic encephalopathy and nine cases identified in an AVMs study.

Clinical headache characteristics had a tendency towards severe, unilateral pain localized to frontal, occipital, or temporal regions. Within 1,116 patients reporting number of migraine days, most reported low migraine days, with a median of 4 days (IQR 1–10 days). The number of patients with mentioned presence or absence of aura was 3,693, of which around half of patients reported presence of aura characteristics (52.5%, n = 1,937). Similarly, explicit description of headache onset was available for 2,308 patients, with a sizable proportion corresponding to gradual onset (56.6%, n = 1,306) (Table 2). 2,924 patients explicitly reported imaging details, with most patients undergoing MRI (76.9%, n = 2,249) and only 10.0% (n = 292) of patients did not have any imaging performed. With regards to treatment, 1,779 cases specified therapeutic agents, of which NSAIDs (30.5%, n = 542), triptans (18.2%, n = 323) and antiseizure medication (13.3%, n = 236) were reported as the most frequent agents (Table 3). A total of 3906 cases explicitly reported whether the patient required further interventional therapies such as surgery. Of these, cranial surgery was performed in 16.1% (n = 629) of patients [24–54]. Most common causes of surgery included cerebral aneurysm clipping (11.3%, n = 71) and pituitary adenoma resection (10%, n = 63). Other causes reported in less than 10 patients included arteriovenous malformation resection, moyamoya disease, stroke management, epilepsy management, astrocytoma resection, IIH shunt, Chiari malformation decompression, arachnoid cyst resection and other intracranial neoplasms and cerebral vascular disorders.

Table 2.

Clinical headache characteristics among the studied population

Characteristic	n/N of patients with reported characteristic	%
Sudden onset	1002/2308	43.4
Gradual onset	1306/2308	56.6
Mild intensity	47/2256	2.1
Moderate intensity	657/2256	29.1
Severe intensity	1552/2256	68.8
Frontal headache	263/897	29.3
Occipital headache	195/897	21.7
Temporal headache	205/897	22.8
Parietal headache	39/897	4.3
Hemicranial headache	146/897	16.3
Retro-orbital headache	49/897	5.5
Unilateral headache	1845/2758	66.9
Bilateral headache	913/2758	33.1
Presence of aura	1937/3693	52.5

Table 3.

Most reported therapeutic agents

Drug	n/1779	%
NSAIDs	542	30.5
Triptans	323	18.2
Antiseizure	236	13.3
Aspirin	221	12.4
Antidepressants	85	4.8
Steroids	68	3.8
Beta blockers	68	3.8
Migraine specific combination	68	3.8
Other	168	9.4

NSAIDs Nonsteroidal Anti-Inflammatory Drugs

Underlying causes corresponding to all 4,422 patients were listed according to frequency, with the most frequent individual migraine-like headache etiologies reported as stroke (6.4%, n = 283), epilepsy (5.9%, n = 261), TBI (5%, n = 221) and Mitochondrial Encephalopathy, Lactic Acidosis, and Stroke-like Episodes (MELAS) (4.1%, n = 181). Nonetheless, individual causes marked as others (25%, n = 1106) and other cerebrovascular disorders (9.5%, n = 420) accounted for a substantial proportion of patients. (Supplementary Material, Table 3) Amongst the “other,” “other cerebrovascular disorders” and “other intracranial neoplasms” causes, Schimke immuno-osseous dysplasia and Susac syndrome (both 1.4%, n = 62) were the most frequent. After grouping, most causes persisted under the “others” group (31.9%, n = 1,411), followed closely by vascular (29.1%, n = 1,287), autoimmune and infectious (9.5%, n = 420 both) (Table 4). Within the 1,406 patients grouped as “Others”, the most frequent causes included others as individual cause (50%, n = 703), MELAS syndrome (11.4%, n = 160), Stroke-like Migraine Attacks after Radiation Therapy (SMART) syndrome (10%, n = 141) and Idiopathic Intracranial Hypertension (IIH) (7.1%, n = 100).

Table 4.

Etiology group behind migraine-like headache

Etiology group	n/4422	%	References
Others	1411	31.9	[11, 46–51, 54–116]
Vascular	1287	29.1	[7, 9, 15, 18, 24–27, 29, 31–33, 117–165]
Autoimmune	420	9.5	[166–185]
Infectious	420	9.5	[17, 186–206]
Neoplastic	340	7.7	[12, 34–45, 207–209]
Epilepsy	323	7.3	[53, 210–223]
TBI	221	5	[14, 224–234]

TBI Traumatic Brain Injury

Imaging modality and etiology

Imaging of choice was shown to be associated with etiological group. A likelihood ratio test result revealed a statistically significant association of small to moderate strength (Likelihood Ratio = 49.369, df = 18, p < 0.001, Cramer’s V = 0.286) between these variables. In this context, autoimmune etiologies were particularly evaluated with MRI and none with computed tomography (CT) when compared to other causes, while TBI showed a tendency towards a lack of imaging evaluation (Fig. 3).

Fig. 3.

Imaging technique by etiology group

Mean migraine days

Studies including a larger proportion of migraine-like patients, but not necessarily larger study sample size, reported higher mean migraine days (r = 0.344, N = 69, p = 0,004). Subsequently, significant differences were found between mean migraine days, etiology group (Kruskal–Wallis H [6] = 20.189, p = 0.003) and imaging type (Kruskal–Wallis H [3] = 16.933, p = 0.001). Post-hoc analysis revealed that neoplastic causes reported fewer mean migraine days when compared to TBI cases (adjusted p = 0.001). Likewise, patients in which both CT and MRI were used at evaluation reported fewer migraine days than studies with lack of imaging, CT, and MRI cases individually. (adjusted p = 0.003, p = 0.018, p = 0.013 respectively) (Fig. 4).

Fig. 4.

Associations pertaining to mean migraine days. A Correlation between mean migraine days and number of individuals with migraine-like headache included. B Distribution of mean migraine days by etiology group. C Distribution of mean migraine days by type of imaging study

Mean age of patients

Previous hypertension/diabetes history (ANOVA: F(3,165) = 4.49, p = 0.005), headache intensity (ANOVA F(2,129) = 3.607, p = 0.030) and headache location (ANOVA: F(5,82) = 1.515, p = 0.194; Welch’s F(5,20.44) = 3.909, p = 0.012) showed significant differences when compared to mean age of patients. Post-hoc analysis revealed that patients with severe intensity (mean = 29.62 ± 16.48) were significantly younger than those with mild (mean = 41.30 ± 3.32, p = 0.008) and moderate (mean = 37.69 ± 17.76, p = 0.047) intensities. In parallel, parietal headache location (mean = 20.25 years) represented patients significantly younger than temporal (mean = 37.62 years, p = 0.018) and hemicranial (mean = 36.40 years, p = 0.035) locations (Fig. 5).

Fig. 5.

Associations between mean age and clinical findings. A Distribution of mean age by previous history of diabetes/hypertension, B Distribution of mean age by pain intensity. C Distribution of mean age by headache location

Sensitivity analysis

Sensitivity analysis included 116 low-bias concern studies and a total of 2,095 patients, which showed a reduction in the proportion of case reports (n = 34, 29.3%). While aura presence proportion did not change significantly, 1,747 low-bias concern cases which mentioned aura presence or absence, reported a slightly lower prevalence of aura patients than full analysis (n = 834, 47.8%). On the other hand, out of 415 studies reporting previous migraine history, the median percentage of patients with previous primary migraine increased in the low-bias concern group (35.7%, IQR 0–100) and a small decrease in neoplasm group etiology patients was also noted (n = 109/2095, 5.2%). In terms of individual etiologies, out of 2095 patient, low-bias concern cases showed slight increase in stroke (n = 163, 7.8%) and SMART syndrome (n = 90, 4.3%). Interestingly, some relevant changes were revealed regarding the most frequent treatment agents, where 1,003 cases mentioned therapeutic agents used in the low-bias concern group: NSAIDs (n = 264, 26.3%), triptans (n = 212, 21.1%), aspirin (n = 158, 15.8%) and antiseizure medication (n = 123, 12.3%). All other descriptive data did not show substantial change (Fig. 6).

Fig. 6.

Comparison of clinical and study characteristics: Full dataset vs. Low-bias subset. TBI: Traumatic Brain Injury

While the association between etiology group and mean migraine days showed a similar trend as in the full dataset, it no longer reached statistical significance (p = 0.076).

The association between mean age and previous hypertension/diabetes history did not reach statistical significance (p = 0.069), although the trend suggested older patients in groups with these comorbidities. Similarly, the association between mean age and headache intensity was not significant after sensitivity analysis (p = 0.104). However, the association between mean age and headache location remained significant (p = 0.023). Although no individual post-hoc comparisons survived multiple testing corrections, the pattern of younger age in parietal cases compared to temporal and hemicranial locations was directionally consistent with the original findings. All other inferential analysis did not show substantial change (Fig. 7).

Fig. 7.

Comparison of statistical significance: Full dataset vs. Low-bias subset. HTN: Hypertension, DM: Diabetes

Discussion

The current study is focused on describing and analyzing the epidemiological, etiological, clinical, and therapeutic characteristics of individuals with migraine-like headache and their relevant statistically significant findings. For this purpose, the amount of screened and included studies was drastically larger when compared to the previous systematic review on the subject. This difference is primarily due to the implementation of strict working criteria for definite and probable migraine-like cases in the previous review [21]. On the other hand, the present study aimed at understanding the full breadth of reports linking migraine-like headaches with secondary causes and therefore considered studies for inclusion whenever a causal relationship between a recognized migraine-like headache and a diagnosed underlying cause was explicitly stated, even if headache characteristics were not broadly described. While this method can introduce diagnostic bias, possibly including cases not adhering to strict migraine-like headache criteria, it broadens identification of headache characteristics and red flags for subsequent consideration in future diagnostic criteria.

Patient characteristics

Patients identified were most frequently young, female without previous history of diabetes or hypertension. Although full analysis showed a substantial proportion of patients with previous migraine history, sensitivity analysis signaled towards a more conservative presentation (57.7% vs. 35.7%). This finding highlights the diagnostic challenge and delay, common in individuals with migraine-like headache [2–4, 72]. While it may be argued that a relatively high proportion of primary migraine patients could potentially be hiding an underlying cause, the development of a secondary cause after the initial diagnosis of migraine could also be argued [3, 21]. While previous observational studies have identified positive associations between hypertension and migraine, diabetes has not been independently associated and may even confer lower risk of migraine [235, 236]. A low prevalence of hypertension and diabetes among the studied population might represent a lack of direct involvement with migraine-like headache pathophysiology [237, 238]. Additionally, the almost negligible death rate could be linked to the lack of long-term follow-up, common in case reports and some observational studies [239, 240].

Headache characteristics

Highly relevant headache characteristics revealed that most patients had severe unilateral episodic migraine-like pain localized to frontal, temporal, or occipital areas. In addition, the relatively high prevalence of aura in individuals with migraine-like headache included in this review compared to primary migraine descriptions (47.8%−52.5% vs 30–38%), suggests relevant cortical spreading mechanisms related to the most frequent underlying causes. [1, 241, 242] Similarly, an even higher proportion of sudden onset headache among individuals with migraine-like headache could strengthen red flag criteria evidence. [243–245] While the intensity of pain confirms the disabling consequences that underlying pathologies originate, low mean migraine days show an underreported pattern possibly related to the episodic nature of the main causes described [1, 14].

Etiology characteristics

When it comes to etiological entities, reported causes were highly heterogeneous, with a considerable proportion of pathologies being represented by a single or a couple of reports. Nevertheless, frequent causes can be clearly identified, with individual vascular, epileptic, and traumatic causes reflecting high prevalence, as highlighted by sensitivity analysis. Additionally, a relatively high proportion of MELAS and SMART syndrome reports highlight their known association with migraine-like headaches [61, 71]. Etiological grouping enables a more comprehensive understanding of identified causes. A strong predominance of vascular etiologies is consistent with involvement of vascular phenomena in migraine-like headache pathogenesis, which in turn signals towards an overlap with primary migraine physiology [21, 33, 38, 57, 136, 246]. Although the epileptic, traumatic and neoplastic relationship with migraine-like headache was relatively known, autoimmune and infectious causes appeared as frequent unexpected etiologies that require further analysis through pathophysiology [8, 17, 174, 193]. These findings contrast with the previous systematic review when it comes to individual etiologies, highlighting stroke instead of AVMs as the most frequent association. On the other hand, the current and previous studies detected vascular pathologies as responsible for a substantial proportion of migraine-like headaches, which strengthens this notion [21].

Imaging and treatment

Recent MRI dominance at evaluation stage seems consistent with current neurological practice, as MRI is commonly preferred after red flag detection in headache [2–5, 247]. Interestingly, headache specific therapies were not the most frequent medication prescribed, showing a lack of targeted treatment possibly related to the episodic pain seen in most patients. Subsequently, triptans and antiseizure medication were the main migraine-specific agents used when confronted with individuals with migraine-like headache, which may be similarly attributed to the episodic and aura characteristics that prompted clinicians towards a symptomatic approach [17, 18]. The particularly low incidence of cranial surgery confirms the notion that non-surgical treatment remains the norm for migraine-like headaches [2, 5, 21].

Associations after sensitivity analysis

Sensitivity analysis confirmed statistical significance for relevant associations. Although mean age was initially associated with previous history of hypertension and diabetes, headache location and intensity, sensitivity analysis did not find comparable associations except for younger patients’ tendency towards parietal headache. Nevertheless, post-hoc analysis revealed that, despite parietal location being a key driver towards this difference, it could not explain the entirety of the overall age difference. Similarly, higher mean migraine days prompted more aggressive imaging evaluation using both CT and MRI, which reflects clinician behavior when confronted with severe migraine-like headaches [23].

Choice of imaging study appeared to be related to etiological grouping, possibly reflecting clinician’s reasoning and preference for MRI when suspecting an autoimmune disease, which would reveal lesions in detail [21, 23]. Conversely, the lack of imaging studies among traumatic causes hints towards a substantial proportion of mild TBI, while the use of CT and MRI could represent severe TBI cases [14, 248].

Pathophysiological parallels

Primary migraine pathophysiology is currently understood as a process including a 1) premonitory phase, characterized by nonspecific symptoms due to hypothalamic activation, 2) Aura, whose process is attributed to cortical spread depression (CSD), 3) Headache, commonly associated with trigeminocervical complex activation, CGRP and substance P pain transmission and 4) Postdrome phase related to locus coeruleus activation [246, 249, 250].

Aura and vascular causes among individuals with migraine-like headache could reflect a strong association with CSD mechanisms. To support this relationship, CSD phenomena have been previously observed in stroke, TBI and epileptic disorders, which conform to the most frequent etiological groups in our study [27, 246, 251, 252]. Additionally, this trend may represent the importance of endothelial dysfunction and vasoactive substance secretion associated with neurovascular events and their subsequent trigeminovascular complex activation, which could in turn precipitate migraine [253, 254]. In addition, CSD has been proposed to worsen ischemic damage in the context of stroke, suggesting a positive feedback cycle [29]. Migraine and epilepsy have been previously observed to share paroxysmal characteristics and CSD production through cortical excitability changes is a possibility. These pathways may be represented by the high epilepsy prevalence found in this study [255, 256]. In addition, the considerable proportion of autoimmune and infectious diseases found may be explained by migraine occurrence through systemic inflammatory responses or neuroinflammation, which in turn enhances trigeminal nociceptive pain transmission and hypothalamic involvement [257, 258].

The lack of significant associations, following sensitivity analysis, between etiology group and demographic or clinical characteristics may signal that migraine-like headache causative etiologies follow similar pathophysiological pathways as primary migraine, and therefore do not significantly change patient characteristics [21, 246]. Nevertheless, associations between mean migraine days and etiology could yield statistical significance in more robust controlled observational studies. Likewise, the predominance of unilateral, severe pain could be influenced by a similar trigeminocervical complex activation and subsequent central and peripheral sensitization that results in truly migraine-like pain [253, 256].

Diagnostic impact

The frequent association between individuals with migraine-like headache and previous primary migraine diagnosis highlights the importance of careful evaluation and red flag identification. Several associations related to red flags were identified in these studies. Infectious causes being the second most frequent etiology group alongside autoimmune diseases, signals towards possible systemic symptoms being present in individuals with migraine-like headache [5, 259]. This could also be strengthened by the many reported cases of COVID-19 as an underlying cause [187, 192, 193]. Similarly, around half of patients acknowledged a sudden onset of headache as an important red flag, which adds value to its future recognition as a prognostic feature [5, 23]. Most importantly, for patients with post-traumatic headaches, who represent a sizable proportion of cases, increased consideration of neuroimaging could be proposed. Additionally, identified cases of migraine-like headache due to drug dependence reiterate their inclusion as a proposed red flag [23, 260]. Although not identified as a red flag, a higher proportion of aura patients may represent a need for higher suspicion index in these patients [5, 23]. It is worth noting that cranial surgery was reported in 16% of patients, regardless of etiological group, reflecting the resource intensive therapeutic strategy used in patients with structural pathologies [2, 21, 23].

Limitations and future perspectives

The current review included a large number of studies, extensive geographical and temporal distribution and therefore strived towards characterization of the “full clinical picture” regarding migraine-like headache. Nonetheless, reliance on case reports and small observational studies inherently included publication bias. Etiology grouping was encouraged to diminish this bias, which produced a high proportion of patients included in the “others” etiological group. Incomplete data and lack of a clear migraine-like definition were important challenges when confronted with highly variable reporting techniques and could indicate that weak tendencies were not robust, therefore solid statistical and sensitivity analysis was paramount.

An urgent need for widely recognized migraine-like headache definition based on etiological frequency and clinical characteristics is recognized. For this purpose, future prospective studies are needed to propose clinical algorithms beyond red flag identification for patients at risk of secondary causes. Furthermore, cost–benefit analysis of early MRI use in high-risk migraines should be encouraged, as well as an international registry of migraine-like cases.

Conclusion

This systematic review consolidates the epidemiological, clinical, and therapeutic characteristics of individuals with migraine-like headaches and their underlying etiologies reported in the literature. Although clinically similar to primary migraine, a substantial proportion of these cases were associated with identifiable secondary causes such as vascular, autoimmune, and infectious disorders. The frequent presence of aura, severe unilateral pain, and sudden onset, along with systemic symptoms, a history of trauma or drug-dependency, emerged as key red flags that should attract attention towards thorough clinical evaluation and early neuroimaging.

Clinical features observed in the study cohort and the predominance of vascular causes found may reflect pathophysiological mechanisms similar to primary migraine, such as CSD, neuroinflammation and trigeminovascular activation. These similarities may account for the frequent use of triptans and antiseizure agents, which are typically employed in the management of primary headache disorders.

The considerable variability in reported etiologies, combined with the predominance of case reports and observational data, reflects both a diagnostic challenge and a research gap. An ongoing need to refine the definition of migraine-like headache, with particular emphasis on highlighting red flags to distinguish it from primary migraine and other secondary diagnoses, is recognized. Similarly, prospective studies should aim to develop robust evidence-based algorithms that integrate clinical features, high-risk characteristics and imaging strategies for diagnosis and management of patients at risk for secondary headache etiologies.

Abbreviations

SA

Sensitivity analysis

MRI

Magnetic resonance imaging

NSAIDs

Nonsteroidal anti-inflammatory drugs

AVMs

Arteriovenous malformations

TBI

Traumatic brain injury

MELAS

Mitochondrial encephalopathy, lactic acidosis, and stroke-like episodes

SMART

Stroke-like migraine attacks after radiation therapy

CSF

Cerebrospinal fluid

IIH

Idiopathic intracranial hypertension

MS

Multiple sclerosis

CADASIL

Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy

CVT

Cerebral venous thrombosis

HANDL

Headache with associated neurological deficits and lymphocytosis

COVID-19

Coronavirus disease 2019

AHS

Alpers-huttenlocher syndrome

MOH

Medication overuse headache

SLE

Systemic lupus erythematosus

Post-EVT

Post-endovascular treatment

ICHD-3

International classification of headache disorders, 3rd edition

PRISMA

Preferred reporting items for systematic reviews and meta-analyses

CSD

Cortical spread depression

IQR

Interquartile range

CT

Computed tomography

SD

Standard deviation

JBI

Joanna Briggs Institute

Authors’ contributions

EGGR was a major contributor to the overall project supervision and administration, research conceptualization, data curation, formal analysis, validation, visualization, writing and reviewing of the manuscript. ANCG was a major contributor to research conceptualization, data curation and statistical analysis, as well as manuscript writing. MFCV was a major contributor to research conceptualization, data curation and manuscript writing. JHR contributed to the data curation process and manuscript writing. AEJR contributed to the data curation process and project administration. JMFM contributed to the data curation process. ERL was a substantial contributor to manuscript reviewing. CADG was a substantial contributor to manuscript reviewing. JAGO was a major contributor to research conceptualization, project supervision and manuscript reviewing. All authors read and approved the final manuscript.

Funding

This research received no external funding.

Data availability

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.