Introduction
Over 69 million people suffer from traumatic brain injury (TBI) worldwide every year, with 90% of these classified as mild traumatic brain injury (mTBI).1–3 The terms ‘concussion’ and ‘mild traumatic brain injury’ are considered synonymous.4–6 Concussions are most commonly caused by motor vehicle accidents, followed by falls and sports-related injuries.7,8 Although most people recover from concussion, approximately 30% develop persistent symptoms over 3 months and 5% to 15% develop symptoms over 12 months.7,9–11
Numerous symptoms result from concussion, including headache, dizziness, changes in sleep, poor concentration, depression, anxiety and cognitive impairments.6,10,12 Post-traumatic headache (PTH) is defined by the onset of headache within 7 days following trauma or injury to the head and/or neck and is further characterized as either acute (the first 3 months from headache onset) or persistent (beyond 3 months) as per the International Classification of Headache Disorders (ICHD).13 Persistent PTH is among the most prevalent sequelae of concussion, affecting up to 89% of patients. It is one of the longest lasting post-concussion symptoms, contributes substantially to morbidity and may be associated with slower neurocognitive recovery.14 The complexity of PTHs can also be related to other concussion sequelae such as cervical or oculomotor dysfunction and anxiety leading to bruxism.15–19
The headache profile of individuals with persistent PTH most often resembles a chronic migraine phenotype or a combined episodic migraine and tension-type headache phenotype according to the ICHD; however, the underlying pathophysiology is not the same, and trials including those of injectables which successfully treat migraine have been shown to be less effective in PTH.20 Heterogeneity in the population, the causes of concussion and the associated symptoms may also contribute to the difficulty of showing the effectiveness of medication.12,20–22 Prophylactic treatment for headaches aims to reduce the frequency or severity of headaches over time; however, there is insufficient evidence to support its widespread use. Consequently, the International Headache Society’s guidelines are based largely on expert consensus rather than robust clinical data.13,15,22–25 First-line oral prophylactic medications that could be used for persistent PTH include gabapentin, amitriptyline, nortriptyline, topiramate, candesartan, verapamil, sodium valproate, propranolol, atenolol and others.23,26 Success rates for the different medications vary. One observational study with 100 military personnel who received either amitriptyline or nortriptyline, topiramate, propranolol or valproate found that 48% of patients experienced a 50% or more decrease in the frequency of headaches, and this was significant for patients treated with topiramate.24 A paediatric retrospective chart review on PTH found that amitriptyline yielded a treatment success rate of 68%.27 A systematic review of migraine prophylaxis revealed that topiramate, propranolol, metoprolol and amitriptyline are strongly recommended with high-quality evidence for migraine prophylaxis, while nadolol, gabapentin and candesartan are strongly recommended with moderate-quality evidence for migraine prophylaxis.23 Amitriptyline was effective in the treatment of PTH in patients who sustained concussions through various mechanism, including motor vehicle accidents, falls and other forms of trauma.22,28,29
The optimal management of persistent PTH is uniquely challenging.30 Regulatory bodies including the Food and Drug Administration (FDA), European Medicines Agency, Health Canada or the Therapeutic Goods Administration (TGA) have no approved drugs for the treatment of PTH, with all medication considered ‘off-label’.31–34 No large randomized controlled clinical trials (RCTs) have shown effectiveness of oral prophylactic treatments for PTH.12,35 Other symptoms including neck pain, oculomotor and visual changes that contribute to headache are also challenging to treat and may impact the efficacy of these medications.36 There is an urgent need to evaluate outcomes in current clinical practice to best provide safe and effective care for patients suffering from persistent PTH.37
As current guidelines for pharmacological treatment of PTH are based on expert consensus rather than evidence-based studies, we aimed to evaluate and compare the effectiveness of commonly prescribed medications for persistent PTH prophylaxis using patient-reported outcomes. We also assessed the impact of various demographic and injury-related factors on PTH treatment outcomes to help inform the development of future evidence-based clinical guidelines.
Methods
Study design and participants
This study is a retrospective chart review of patients referred to the Canadian Concussion Centre (CCC) at Toronto Western Hospital, a tertiary care centre. Most patients were seen for at least two visits by one physician between 2016 and 2021 for persisting symptoms of concussion but in four patients, data about previous management of PTH were obtained from the patient. This study was approved by the University Health Network Research Ethics Board number 23-5091.
Patients were included in this study if they (1) were diagnosed with at least one concussion,23 (2) had persistent PTH due to concussion (note: 90% of patients met the 3-month criterion at intake; the remainder met this criterion at a follow-up), (3) were prescribed and took at least one of the headache prevention medications which included amitriptyline, candesartan, gabapentin or topiramate and (4) reported whether their headaches improved after taking their first headache prevention medication. Patients were excluded if they did not present with PTH, were not prescribed one of the four selected headache prevention medications, were prescribed a medication but had no follow-up appointment, didn’t take their prescribed medication, did not report whether their headaches improved, had an unclear medical file with respect to headache medication, had a positive computed tomography (CT) scan or magnetic resonance imaging (MRI) scan of their brain (e.g. haemorrhage, haematoma), had a neurodegenerative disease or had other medical complications affecting treatment (e.g. long COVID).
Study variables
The following information was collected from a retrospective chart review of all eligible participants:
Demographic data: age, sex, history of headache/migraine, total number of concussions (1 or >1, including index concussion)
Concussion information: mechanism of index concussion (motor vehicle collision [MVC] or other)
Headache prevention medication data: headache prevention medication data were only collected for six medications of interest: amitriptyline, nortriptyline, gabapentin, candesartan, atenolol and topiramate. These are all first-line medications used in primary headache prevention and are commonly prescribed by the study physician (MCT) for persistent PTH following concussion. For the analysis, amitriptyline and nortriptyline were considered one medication group as they belong to the same medication family, and the same for candesartan and atenolol as they are both antihypertensives.
For each medication a participant took, headache improvement was recorded as binary yes/no. This was the primary outcome of the study. A ‘yes’ was recorded if there was any indication of headache improvement, regardless of extent. For each patient, we recorded whether any of the headache prevention medications they took were effective, or if none of the medications they took showed effectiveness, regardless of the number of medications each patient tried.
Medications taken prior to being seen at the CCC were only recorded if they were one of the medications of interest listed earlier.
Descriptive statistical methods were employed to describe the study sample and the pattern of prescribed medications. Categorical variables were expressed as counts and percentages, and group comparisons were performed using the Chi-squared test. Binominal logistic regression was used to calculate an odds ratio with 95% confidence interval with gabapentin as the reference. Continuous variables were summarized using the mean with standard deviation, and group comparisons were performed using the t-test. To explore the associations between headache improvement and the two specific medications, as well as the relationships between headache improvement following treatment with any of the four first-line medications and demographic and injury-related variables, univariable and multivariable logistic regression analyses were performed. A subgroup analysis, stratified by mechanism of injury, was additionally conducted to investigate the association between medication usage and headache improvement for patients who sustained concussion in either a MVC or a non-MVC incident. Multivariable models were adjusted for potential confounders selected based on both statistical and clinical relevance. *p-*Values were deemed statistically significant at p < 0.05. All statistical analyses were performed using R, version 4.2.2.
Results
Participants
A total of 418 charts were reviewed. Of these, 318 patients were excluded (Table 1), leaving 100 patients for inclusion in the final analysis. The demographic and injury-related characteristics for all 100 participants are presented in Table 2. Our retrospective dataset included 69 (69.0%) females and 31 (31.0%) males. The average age was 42.6 years (standard deviation [SD] = 14.2). MVCs were the most common mechanism of injury (56.0%) compared to all other mechanisms combined (44.0%). More than half of the patients (55.0%) had a history of more than one concussion (including the index concussion), whereas the remaining 45.0% had sustained only one concussion. Additionally, 26.0% of patients reported a prior history of headache or migraine.
Response to initial medication prescribed
Gabapentin and amitriptyline/nortriptyline were the most frequently prescribed first-line medications for headache prevention (45.0% and 44.0%, respectively), while topiramate and candesartan/atenolol were much less frequently prescribed as a first medication (6.0% and 5.0%, respectively) (Table 3). Overall headache improvement after the first medication was seen in 64.0% of participants. Headache improvement was most common in patients who took gabapentin as their first medication (73.3%), followed by topiramate (66.7%), amitriptyline/nortriptyline (56.8%) and candesartan/atenolol (40.0%).
Headache improvement – Amitriptyline/nortriptyline versus gabapentin
Of the 100 participants, 89 took either gabapentin (n = 45) or amitriptyline (n = 44) as their first medication and were therefore included in our primary analysis. Significantly more patients in the amitriptyline/nortriptyline group had a history of more than one concussion (including the index concussion) compared to the gabapentin group (68.2% vs 37.8%, respectively, p = 0.008) (Table 4). There were no other significant differences between the two groups for any other demographic or injury-related variables.
Univariable logistic modelling suggested that gabapentin doubled the odds of headache improvement compared to amitriptyline/nortriptyline, although this was not significant (odds ratio [OR] = 2.09, 95% confidence interval [CI] 0.86–5.09, p = 0.105) (Table 5). With respect to the demographic and injury-related covariates, patients with a history of headache/migraine showed a trend toward being less likely to experience PTH improvement (OR = 0.43, 95% CI 0.16–1.41, p = 0.089). Although sustaining a concussion in an MVC was also associated with a decreased odds of PTH improvement, this was not statistically significant (OR = 0.48, 95% CI 0.19–1.87, p = 0.111). We found no association with PTH improvement for age, sex and number of concussions (Table 5).
First medication, mechanism of injury and history of headache/migraine were entered into a multivariable logistic regression model. Age, sex and concussion history were not included as they showed no association with headache improvement in the univariable analysis. After adjustment, results remained similar to the univariable model (Table 5).
Subgroup analyses: Headache improvement – Amitriptyline/nortriptyline versus gabapentin, split by mechanism of injury
In a subgroup analysis of patients who sustained concussion in an MVC (n = 50), there were no significant differences between patients who took amitriptyline/nortriptyline (n = 24) versus gabapentin (n = 26) as their first medication with respect to age, sex and history of headache/migraine. However, significantly more patients in the amitriptyline/nortriptyline group had more than one concussion compared to the gabapentin group (62.5% vs 23.1%, respectively, p = 0.011) (Supplementary Table 1). Univariable logistic modelling revealed a significant association between medication and PTH improvement, such that the odds of PTH improvement were 3.8 times higher for patients who sustained concussion in an MVC and took gabapentin as their first headache prevention medication compared to those who sustained concussion in an MVC and took amitriptyline/nortriptyline as their first medication (OR = 3.8, 95% CI 1.16–12.50, p = 0.028). In this MVC subgroup, the odds of PTH improvement were reduced for patients who had a history of headache/migraine (OR = 0.19, 95% CI 0.04–0.83, p = 0.027). After adjusting for history of headache/migraine in a multivariable model, the odds of PTH improvement remained significantly higher for patients who took gabapentin as their first medication in this subgroup (OR = 4.05, 95% CI 1.13–14.40, p = 0.030) (Table 6).
With respect to patients who sustained concussion in any incident other than an MVC (n = 39), there were no significant differences between patients who took amitriptyline/nortriptyline (n = 20) versus gabapentin (n = 19) as their first medication with respect to sex, total number of concussions and history of headache/migraine. The gabapentin group was, however, on average older than the amitriptyline/nortriptyline group, and this difference approached significance (35.35 SD = 11.1 vs 44.84 SD = 18.53, respectively, p = 0.057) (Supplementary Table 2). In a univariable model, we found no association between medication and headache improvement for this subgroup (results not shown).
Profile differences between patients with PTH who improved compared to those who did not
In our secondary analysis, we examined potential factors that could be associated with headache improvement among patients taking any one of the four first-line headache prevention medications and demonstrating improvement. Considering headache improvement overall, whether on first or subsequent therapy, 79 out of 100 showed improvement, while 15 out of 100 did not. Six patients were excluded from this analysis due to missing data.
All demographic and injury-related variables were assessed using univariable logistic regression models. None of age, sex, mechanism of injury and history of headache/migraine were associated with the outcome, each with ORs close to 1 and p-values >0.5. However, a history of more than one concussion (including the index concussion) decreased the odds of PTH improvement on any of the four first-line medications of interest (OR = 0.27, 95% CI 0.07–1.03, p = 0.055) (Table 7). In the multivariable model, there was a significant association between total number of concussions and the outcome, suggesting that independent of all other variables, a history of more than one concussion decreases the odds of headache improvement after taking any of the four first-line medications evaluated (OR = 0.23, 95% CI 0.06–0.93, p = 0.039) (Table 7).
Discussion
Our results showed that headache improvement occurred in 64% of patients after the first medication and 79% after all medications tried. Improvement was most common in patients whose first medication was gabapentin (73.3%), followed by topiramate (66.7%), amitriptyline/nortriptyline (56.8%) and candesartan/atenolol (40.0%). The odds of headache improvement in patients who sustained concussion in an MVC were 3.8 times higher for those who took gabapentin as their first headache prevention medication compared to those who took amitriptyline/nortriptyline (Table 6). This was statistically significant (p = 0.03), and although the 95% CI was wide [1.16–12.50], it did not cross 1. Importantly, those in the MVC group had reduced odds of headache improvement when there was a history of a previous headache disorder (OR = 0.19, 95% CI 0.04–0.83, p = 0.027). However, the odds of headache improvement remained significantly higher for patients who took gabapentin as their first medication in the MVC group after adjusting for history of headache/migraine in a multivariable model. Notably, more patients on amitriptyline/nortriptyline had a history of multiple concussions, which could have impacted the effectiveness of this drug, but importantly there was no association between total number of concussions and headache improvement in the univariable model for improvement after first medication taken.
PTH is a debilitating post-concussive symptom. The strongly disabling impact of headache is often associated with somatic symptoms such as photophobia, phonophobia and nausea in many patients and also accompanied by symptoms of anxiety, depression and cognitive issues.21,38 Studies have shown that patients with chronic PTH after head injury were significantly less likely to return to work than those without, by a difference of 21%.22 It is important to note that 42% of patients suffering from PTH also met criteria for medication overuse headache,22 which could be prevented with prophylactic therapies that effectively reduce headaches.22–24,26
Recent meta-analyses of PTH have highlighted the lack of evidence-based studies on pharmacological treatment and the need for rigorous methodology and RCTs with placebo-controlled designs for PTH.37 Emphasizing that PTH often occurs alongside other post-concussive symptoms including neck pain such as whiplash and psychiatric symptoms, it becomes unclear whether approaches derived solely from the headache literature are applicable, given that most patients in that context do not present with the same array of symptoms.11
Different factors were evaluated with respect to PTH treatment outcomes. The mechanism of injury evaluation revealed that patients who sustained concussion in an MVC had higher odds of headache improvement after taking gabapentin compared to amitriptyline. This may be due to the different constellation of symptoms that can ensue after an MVC, such as anxiety and post-traumatic stress disorder (PTSD), which may impact headache.11,39
Another headache classification according to the ICHD is persistent headache attributed to whiplash, which may also be relevant for the patients included in the study.13 The diagnostic criteria for this secondary headache condition specify that the symptoms have commenced within 7 days after the incident, persist more than 3 months and have pain associated with whiplash. Neck-related symptoms may increase the risk of persistent symptoms by 2.58 to 6.38.40 There are no large trials that look at anticonvulsants (such as gabapentin and topiramate) and antidepressants (such as amitriptyline or nortriptyline) for the treatment of whiplash.41 As such, consensus guidelines often recommend against the use of these drugs, and the impact of these medications may have limited benefit for whiplash symptoms.36,42–44 This type of headache may, however, be important to consider in patients who did not respond to treatment.
Interestingly, gabapentin has been used off-label as an anxiolytic, which may be partially responsible for the improved outcome in PTH in MVC.39,45–47 Some studies have stressed that a secondary diagnosis of PTSD plays a critical role in persistent PTH, and since PTSD is common in MVC, it is an important consideration when choosing PTH medications.39,48 It is worth noting that almost 30% of individuals with persistent PTH also have PTSD.48 Conversely, in patients who sustained their injury by any other mechanism (e.g. sports, falls, striking object), we found no difference in the odds of headache improvement when comparing patients who took gabapentin versus amitriptyline as their first medication. Also, our secondary analysis comparing profiles of patients who had headache improvement found that a history of more than one concussion decreased the odds of PTH improvement by 77%. Some studies have noted that a history of concussion(s) is a risk factor for sustaining a subsequent concussion and for developing persistent symptoms.46,47
Our study has some limitations. Firstly, this was a retrospective study, and so improvement was binarized as opposed to having a spectrum of improvement, which likely represents clinical reality best. As well, over half of our patients had more than one concussion, so some may have had PTH and/or other ongoing symptoms from previous concussions. This may have impacted PTH improvement. Also, patients with chronic PTH may have more than one type of headache, in particular persistent headache associated with whiplash,39 impacting response to treatment, which was not included in the data.
Additionally, some patients may not have reported headache prevention medication(s) tried before their visit to the clinic. Including patients at both initial referral and follow-up introduces temporal heterogeneity, as treatment approaches may differ across stages. However, because only 10% of patients were enrolled at follow-up, this is unlikely to have had a substantial impact on the results.
An additional limitation is that medication dose, side effects, adherence and time to response were not captured, which may limit interpretation of treatment effectiveness. Moreover, medications such as atenolol and candesartan were combined due to limited numbers; however, as they each have a different mechanism of action, one could have been more effective than the other. Furthermore, some patients had longer follow-up than others, which could also impact results. As a single-centre, single-physician study, the findings are subject to potential selection and practice-related bias. Consequently, the effectiveness of the second-line or lesser prescribed prophylactic medications may not be fully or accurately represented. There was no validated outcome measure used to quantify improvement; instead, outcomes were recorded as a binary yes/no. This limits the ability to assess the magnitude of treatment effects and compare differences between medications.
Of the 418 charts reviewed, 318 patients were excluded. The most common reasons were absence of a prescribed medication, unclear documentation of headache treatment, lack of headache symptoms or missing data on treatment response, accounting for 73% of exclusions. This likely represents the setting of a specialized concussion clinic, where headache is not universally present, unlike in dedicated headache clinics. Additionally, excluding patients without follow-up or with missing outcome data introduces attrition bias and may lead to an overestimation of the treatment effects. The relatively small sample size (n = 100) limits the ability to perform robust multivariable and subgroup analyses and increases the risk of overinterpreting the findings.
Conclusion
There are currently no clinical studies or evidence-based guidelines for PTH management.21,30 However, our results suggest that patients may experience improvement following initiation of pharmacological treatment, supporting the broader treatability of this condition. In particular, gabapentin may be associated with greater benefit, especially in patients who sustained their injury in a motor vehicle accident. These observations are limited by the small sample size (n = 100) and underscore the need for larger prospective trials in PTH. Future trials should include stratification by clinical and injury-related factors, such as prior concussion history, baseline headache or migraine disorder, age, sex, mechanism of injury and concomitant symptoms, to better define treatment response across subgroups.
Author contributions
Olivia FT Scott: Design, methodology, data curation, investigation, result preparation. Goldin Joghataie: Manuscript preparation, writing, review and editing of discussion, conclusions, abstract, reference management. Artee Srivastava: Writing of introduction. Asma Mushtaque: Investigation. Ella Huszti and Qixuan Li: Visualization, software, data curation and statistical analysis. Dina Safarini: Investigation. Timothy Butson: Manuscript preparation, writing, review, editing of discussion. Charles H Tator: Conceptualization, supervision. Carmela Tartaglia: Supervision, principal investigator, conceptualization.
Conflict of Interest Statement
The authors have no competing interest to disclose.
Funding statement
University Health Network, Canadian Concussion Centre