This systematic review was carried out in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines.²³ A systematic search was performed focusing on PTH phenotype following mTBI in military personnel. Two reviewers (H.S.L. and M.S.) from the University of Birmingham independently screened the abstracts retrieved from the above-mentioned searches using Rayyan software.²⁴ Any discrepancies were resolved through a third senior author (J.M.). The names of authors and journals were not masked from the reviewers. Following abstract review based on the inclusion criteria, selected articles underwent full-text review by two reviewers (H.S.L. and M.S.) working independently. Additionally, the reference lists of these selected articles were searched to identify additional articles that potentially met the inclusion criteria and were subject to the same review process.

The search criteria for this systematic review included peer-reviewed and published studies meeting the following criteria: (1) military population active personnel; (2) participants aged 18 years or older; (3) participants diagnosed with mTBI or concussion; (4) participants experiencing subsequent PTH; (5) studies published from inception until March 1, 2023; and (5) studies written in the English language. Searches were conducted in PubMed, Web of Science, Cochrane, and Clinicaltrials.gov. A complete list of the search terms for each database is provided in Supplemental Table S1.

Reviews, commentaries, editorials, conference proceedings, case reports, and case series (defined as studies involving fewer than four participants) were excluded. Studies in which PTH was not defined as an outcome or did not include PTH sub-phenotypes were also excluded.

Data extraction was independently performed by the two reviewers (H.S.L. and M.S.) using a standardized form (Supplemental Table S2). Any discrepancies were discussed and resolved between the two investigators, with a senior researcher (J.M.) serving as moderator. Mixed cohort studies were also included in the analysis.^25,26

Each selected study was independently assessed by two investigators (H.S.L. and M.S.) to determine the strength of evidence. The level of evidence was classified and scored based on a modified version of the Newcastle-Ottawa scale.²⁷ A quality assessment rating of “high” (corresponding to ≥7 scores), “medium” (score 6), or “low” (≤5 scores) was assigned to each study.

The military search initially yielded 719 articles (Figure 1). Following the screening of abstracts, 13 articles met the inclusion criteria and underwent a full-text review. Upon completion of full-text review, eight military studies^{25,26,28–33} satisfied the final criteria. The methodological risk of bias for each study is reported in Supplemental Table S3; The Newcastle-Ottawa scores ranged from five to seven (mean 6.1).²⁷

An overview of the studies included is displayed in Supplemental Table S4. Half of the military studies had a prospective design,^25,26,31,33 while the other half were retrospective.^28–30,32 Two studies did not define their population as “military” but included a large proportion of blast injury, 43% to 44%,^25,26 and therefore given that blast injuries predominantly occur among military personnel, they have been included in the analysis. Six studies focused on soldiers in active duty, with three studies specifying that their cohorts had recently returned from deployment.^14,29,33 The duration of these deployments were typically one year, and two studies evaluated participants within seven days of injury.^31,32 The two mixed cohort studies observed those with persistent PTH, with a mean average time of assessment post-injury at 10.6 years.^25,26

In terms of sex distribution, female participants accounted for 0% to 10% of the exclusive military cohorts, with 32 to 35% reported in the mixed cohort studies.^25,26 Blast injuries were a unique aspect of military personnel, comprising 31% to 92% of the reported injuries.

Although most studies phenotype headache using the ICHD, the definition of mTBI, however, varied across the studies, with some employing the criteria set by the VA/DoD,³⁴ Centers for Disease Control and Prevention (CDC),³⁵ or no specific criteria. Regarding family history of primary headaches, it was reported in 28% to 32% of the studies,^30,32,36 with six studies including participants with previous concussions, which affected 18% to 79% of the cohorts.^{25,26,29,30,32,33} Three studies excluded participants with pre-existing headaches.^25,26,31

The headache phenotypes are shown in Table 1.

In three studies, participants were allowed to report multiple headache types,^29,30,32 with up to 30% of participants reporting multiple headache types.²⁹ Migraine-like was the most common headache sub-phenotype in seven studies, ranging from 33% to 92% of participants.^14,29–32 However, one study reported lower numbers, with 33% of participants suffering from migraine-like and 64% classified as “other headaches.”³³ Different studies had varying approaches to categorizing migraine-like and probable migraine-like, with some treating them as distinct phenotypes,^{14,25,29,31,32} while others combined them,^26,30 or did not provide information on probable migraine-like at all.³³ Five of the included studies further separated “non-migrainous” headaches into other primary headache phenotypes.^{25,26,29,30,32} Tension-type headache (TTH)-like was the second most common headache sub-phenotype in four studies.^26,29,30,32 There were 13% to 36% of participants with trigeminal autonomic cephalalgias (TACs) in two studies (same cohort), with blast injuries and prior concussions being contributing factors.^30,32 A study comparing mTBI cohorts with healthy controls found that those with PTH had predominantly migrainous features, and specific symptoms strongly associated with PTH included visual or sensory aura, allodynia, and daily or continuous headaches.³³ The headache assessments in the eight studies involved a variety of professionals, including neurologist/headache specialist,^{25,26,29,30,32} and healthcare professionals in an emergency department.³¹ Self-administered questionnaires were also used in some studies.^14,33

Additional headache characteristics were reported in Table 2. In the seven military studies that reported monthly headache days (MHDs), the mean range was 9.9 to 26.7 MHD.^{14,25,26,29,30,32,33} Most headaches lasted more than four hours and had a moderate to severe severity.^{14,29,30,32,33}

We were unable to perform a meta-analysis of these results because of substantial heterogeneity highlighted above, and instead provided a narrative description of the findings. This review has highlighted that characterizing PTH in deployed military personnel faces several challenges. When looking at the variability in Newcastle-Ottawa scores, two studies received a score of five (high risk of bias),^25,26 which raises concerns regarding its methodological limitations and potential biases. Some of the included studies lacked documentation of participants’ baseline characteristics regarding pre-existing headache disorders, which are known to be associated with the development of PTH.^14,33 Notably, three studies explicitly excluded patients with pre-existing headaches before their head injury.^25,26,31 Since PTH can also be defined as the worsening of pre-existing headaches, excluding this sub-cohort of patients could introduce selection bias. This, in turn, can alter study outcomes, such as the sub-phenotype of PTH. Conversely, a review article argues that including those with pre-existing headaches is unlikely to represent the prevalence rates of the general population, thus limiting generalizability.⁵⁴ This discrepancy underscores the importance of careful study design to balance the accurate representation of PTH characteristics and the generalizability of findings to the broader population. Additionally, there is the risk of recall bias because of assessments being conducted after their return from operations, along with possible cognitive disturbances associated with wider post-concussion syndrome. This bias extends not only to the timing of headache onset but also to the accurate identification of the injury as a mTBI. It is important to note that the military participants were usually assessed within a year of injury, whereas the mixed cohorts included individuals up to 11 years post injury. Limited participation of females in military studies poses a potential limitation in generalizing the findings to the broader population. It is important to acknowledge there may be variability in blast injury severity within the category of mTBI. The use of different professionals for documenting headache phenotypes introduces variability in expertise, training, and consistency, highlighting the need for standardized criteria and protocols to minimize variability and potential bias in the classification of headache phenotypes. In addition, there is some symptom overlap with migraine and TTH, with photophobia, phonophobia and kinesiophobia also being experienced by those with TTH, and this can lead to discrepancy in results.⁵⁵ The exclusion of probable migraine-like as a phenotype in certain studies raises the possibility that some headaches categorized as “unclassified” may actually meet the criteria for this phenotype, leading to under-reporting of this phenotype. The 2 studies that reported higher numbers of TAC were from the same cohort at different time points and so may not be representative of the wider military cohort.^30,32 While using the term “military” as a population there are many different roles within a military population that may be more susceptible to PTH and mTBI, such as those involved in close combat roles versus those involved in shaping operational environments.

In future research, we would recommend appropriately recording baseline characteristics, including the presence of any pre-existing headache, family history of headache, and any previous concussion history. It would also be useful to investigate when the headaches started and observe any potential changes in phenotype over time i.e., prospective contemporaneous longitudinal studies. The study populations analyzed in this systematic review display heterogeneity, underscoring the necessity for additional research features, more stringent criteria, and comprehensive recording of baseline characteristics.