Headache after pediatric traumatic brain injury: a comparison between a post-acute sample of children and adolescents and general population

Abstract

Background

Headache is one of the most common post-concussion symptoms following pediatric traumatic brain injury (TBI). To better understand its impact on young individuals, this study aims to investigate the prevalence of headache in a German-speaking post-acute pediatric TBI sample and compare it with the general population. In addition, factors associated with the development of pediatric post-TBI headache are investigated to improve the understanding of this condition.

Methods

A post-acute sample (3 months up to 10 years post-injury) comprising N = 463 children and adolescents aged 8 to 17 years from the TBI sample and N = 463 individuals from the general population matched for gender, age, and health status were included in the study. The Postconcussion Symptom Inventory (PCSI) item assessing headache was used as the outcome variable. Logistic regression was used to examine the association between the risk of developing headache and sociodemographic and health-related factors.

Results

Slightly less than half of the participants reported the presence of headache (TBI sample: 46%; matched controls: 44%). Compared with matched controls, the odds of headache in the TBI sample were not significantly different (OR = 1.09, 95% CI 0.85 to 1.4, p = 0.49). The association between PCSI symptoms was generally stronger in adolescents than in children and in the matched controls than in the TBI sample. In the TBI sample, the probability of reporting headache increased with age.

Conclusions

The results of this study suggest that the prevalence of headache in the post-acute phase of pediatric TBI is not significantly different from that in the matched non-TBI population, indicating good recovery from injury. However, due to its high prevalence, follow-up screening for this common TBI symptom, especially in adolescents, may be helpful to prevent further chronification.

Trial registration

The study is retrospectively registered in German Clinical Trials Register and in International Clinical Trials Registry Platform (ID DRKS00032854).

Introduction

According to the latest Global Burden of Disease study [1], headache disorders remain one of the most common health problems worldwide. At the same time, headache is one of the leading symptoms after traumatic brain injury (TBI) in children and adolescents [2] and adults [3]. As stated in the International Classification of Headache Disorders (ICHD), post-traumatic headache (PTH) occurs within one week of a trauma or injury to the head and/or neck and is classified as secondary headache, with a further distinction between acute (i.e., lasting no longer than three months) and chronic (i.e., lasting longer than three months) [4]. PTH affects up to 90% of children and adolescents after TBI [2, 5, 6], with a prevalence of non-specific chronic headache reaching 39% and chronic PTH according to the ICHD definition reaching approximately 8% [7].

According to multiple reviews [2, 8, 9], the most common risk factors for developing short- and long-term headaches following pediatric TBI are female gender, adolescent age, recurrent and/or more severe TBI, history of headache/migraines (also in family members), preinjury mental health problems, and an increased number of other TBI-related symptoms. In particular, adolescents appear to be more likely to experience headache after TBI than younger children and adults [5, 8]. Headache can negatively affect the ability to concentrate and participate in school [10], interfere with social relationships [11], or limit daily activities and sports [6]. Therefore, understanding headache, its etiology, and associated factors after pediatric TBI is critical for more targeted post-injury treatment.

Headaches are also common in the general population. According to a systematic review of population-based studies [12], the prevalence of headache in children and adolescents reaches more than 50%. Consideration of the prevalence of headache in the general population may therefore be helpful in evaluating the clinical impact of PTH.

Despite the large number of studies examining headache in children and adolescents after TBI, direct comparisons with the general population are lacking [e.g., 7]. Therefore, the present study aims to investigate the prevalence of headache in a German-speaking post-acute sample (3 months to 10 years) after pediatric TBI, to compare the prevalence of headache in a post-acute TBI sample with the general population, and to investigate its association with other symptoms and sociodemographic and clinical factors. The investigation of the known factors associated with the development of headache after injury will further contribute to the understanding of post-TBI headache and to the research on this topic.

Methods

Study participants

The TBI sample was recruited in the Quality of Life after Brain Injury in Children and Adolescents project [13, 14], which collected data from children (8–12 years) and adolescents (13–17 years) after TBI in Germany and Austria from January 2019 to February 2023 in two study phases. In the pilot phase, the first questionnaire was developed to measure TBI-specific health-related quality of life (HRQoL); the second phase aimed at final validation of the questionnaire. Potential participants were selected from the databases of 12 medical centers, and families were contacted via mail if children and adolescents after TBI met the following criteria: age 8 to 17 years, TBI diagnosed 3 months to 10 years prior to enrollment (ICD-10 [15] diagnosis S06.*), including information on its severity (classified as mild, moderate, or severe, based on either the Glasgow Coma Scale score [16] or clinical rating), and ability to understand and answer questions in German. Participants were excluded if they had pre-injury epilepsy, very severe polytrauma (as rated by the investigator to avoid confounding of TBI and polytrauma with effects on HRQoL), severe premorbid mental illness, or life-threatening medical conditions. Based on power analysis, 150 participants from each age group were deemed necessary to meet the original project goal. The recruitment procedure and data collection is described in detail elsewhere [13, 14]. For the present study, sample selection was based on the availability of the self-report post-TBI version of the Postconcussion Symptom Inventory (PCSI) [17], administered in the pilot and final study phases of the project to assess symptom burden after injury. In total, N = 477 children and adolescents were included in the analyses.

The general population sample was recruited online from March 2022 to April 2022 through the databases of two survey research agencies: Dynata (London, UK) and bilendi (former: respondi; London, UK). Details of the recruitment procedure can be found elsewhere [18]. In total, N = 1997 children and adolescents participated in the study. To provide meaningful comparisons, participants in the general population sample were matched to those in the TBI sample based on complete cases for sociodemographic and health-related characteristics (see Statistical analyses). This resulted in N = 463 matched controls. Figure 1 provides further details on sample flow.

Fig. 1

Sample flowchart

Materials and measures

Information on headaches

The PCSI is a system of age-appropriate self-report instruments designed to assess post-concussion symptoms [17], suggested by the Common Data Elements recommendations [19] for use after pediatric TBI as a supplemental outcome measure. Symptoms (PCSI-SR8: 17 for 8–12-year-olds and PCSI-SR13: 21 for 13–18-year-olds) are rated on a Guttmann scale for the past two days. The post-TBI version used in this study aims to assess symptom burden for the time after injury. Matched controls completed an adapted version of the respective PCSI, in which any references to TBI was removed [18, 20].

We used the item “Headache” from the respective German versions of the PCSI [18, 21]. Since different PCSI versions have different response scales (i.e., PCSI-SR8: 0–2 and PCSI-SR13: 0–7), all responses were summarized in a manner comparable to the original study by Sady et al. [17] to allow for direct comparisons. This resulted in three categories (0: ‘no headache’, 1: ‘a little headache’, 2: ‘a lot of headache’) for descriptive statistics and two categories (0: ‘no headache’, 1: ‘headache present’) for statistical analyses.

To additionally examine the influence of symptoms other than headache (e.g., dizziness, sensitivity to light and noise, difficulty concentrating and remembering, drowsiness, fatigue, and others) on probability of developing headache after injury, the sum of the corresponding items rated as at least a little problem (PCSI-SR8) or at least a moderate problem (PCSI-SR13) was calculated.

Sociodemographic, premorbid, and injury-related information

Comparisons with matched controls included the influence of belonging to the TBI vs. control sample, while controlling for gender, age, and the presence of chronic health conditions (and in a separate step, the number of symptoms other than headache) through matching. The presence of chronic conditions in the TBI sample was determined by the investigator (Are there any chronic conditions? [yes/no]). A separate text box provided an opportunity to add the diagnosis. Information on the health status of the controls was reported by the parents across nine health areas: central nervous system disease, alcohol and/or psychotropic drug abuse, active or uncontrolled systemic disease, psychiatric disorder, severe sensory disorder, use of psychotropic or other drugs, mental retardation or other neurobehavioral disorder, problems before, during, and after childbirth, and other. Participants were classified as having at least one chronic condition if at least one area was endorsed.

The following sociodemographic, premorbid, and injury-related factors were used to examine their influence on the probability of developing headache (groups with expected adverse outcome in italics): gender (male vs. female), age (8–12 years vs. 13–17 years), repeated TBI (no vs. yes), more severe TBI (mild vs. moderate/severe), premorbid mental health problems (no vs. yes), while controlling for time since injury (in years).

Statistical analyses

First, descriptive analyses were performed to determine the prevalence of headache in the TBI sample and in the matched controls. Then, two separate regression models were fitted.

Model I examined differences in the odds of developing headache between the TBI sample and matched controls. Matching occurred using nearest neighbor approach based on propensity scores estimated using logistic regression analyses adjusted for age, gender (male vs. female; diverse participants had to be skipped due to low case numbers), and presence of chronic health conditions (no vs. yes). Goodness of matching was assessed using the propensity score variance ratio (a value of 1 indicates a perfect match) and visualized using the Love plot [22, 23]. Marginal effects reported the exposure of headache in the TBI sample compared with matched controls using odds ratio (OR) with 95% confidence intervals (CI). To account for other symptoms and their potential impact on the probability of developing headache, while attempting to avoid predicting outcome by other potentially related outcomes, we reran sensitivity analyses controlling for the number of symptoms other than headache in the matching.

The association between headache and other reported symptoms was examined separately in children and adolescents using Spearman correlation coefficients.

Model II used binary logistic regression to examine the influence of sociodemographic (age, gender), premorbid (health status prior TBI, previous TBI experience), and injury-related characteristics (TBI severity, presence of lesions, and time since injury) on the development of headache in the post-acute phase after TBI. This analysis was limited to the TBI patients.

Information on migraine before TBI could not be included in the model due to low prevalence (n < 30). To better understand the effect of a one-year increase in age on the occurrence of headache, we included a linear and a quadratic age term in the model. The goodness of the model fit was assessed by Nagelkerke’s R², which represents the relative improvement in predictive accuracy between the null model with no predictors and the defined model (range 0–1, with values closer to 1 indicating greater improvement). We provided information on the area under the receiver operating curve (AUC-ROC) characteristic using trapezoidal rule [24] to show the ability to distinguish those without headache from those with headache. The AUC-ROC values > 0.70 indicate at least an acceptable differentiation [25]. Finally, in a sensitivity analysis, we applied elastic net regularization [26] to the regression model to select those sociodemographic and health-related characteristics that meaningfully improved the model prediction. Data was randomly split 1:1 into training data and test data. Standard cross-validation was used to select the best tuning parameters for regularization in the training data. The predictive power of the regularized model was then assessed in the test data.

All analyses were performed using R version 4.4.0 [27] using the packages ‘MatchIt’ [28] for sample matching, ‘marginaleffects’ [29] for calculation of marginal effects, ‘caret’ [30] for regression training, and ‘corrplot’ [31] for visualization. The significance level was 5% two-tailed.

Results

Sample characteristics

For Model I, the sample consisted of N = 463 post-TBI children and adolescents (56% male) and N = 463 matched controls (Table 1). Balance of the matched data was achieved with a propensity score variance ratio of 1 and an absolute standardized mean difference of close to 0 for all variables (Supplementary Figures S1 and S2). A total of 81% of the participants had no chronic health conditions. Slightly less than half of the participants reported the presence of headache (TBI sample: 46%; matched controls: 44%), and 50% reported two symptoms other than headache in both samples.

Table 1 Sample characteristics of the matched TBI and control samples (Model I)

For Model II, the sample comprised N = 465 children and adolescents post-TBI (56% male) (Table 2). The majority had no pre-existing chronic health conditions (81%). Most participants experienced a single TBI (85%), which occurred on average 4.82 (SD = 2.80) years ago. TBI severity was predominantly mild (80%) with no lesions detected on neuroimaging (76%). Somewhat less than half (46%) reported post-TBI headache.

Table 2 Sample characteristics of the TBI sample (Model II)

Prevalence of headache and other symptoms

Figure 2 provides an overview of the prevalence of headache in the TBI and matched controls. Overall, 46% of children and adolescents in the TBI sample and 44% in the matched control sample reported the presence of headache, with a slightly higher representation of post-TBI adolescents in this category. The distribution of symptom severity showed some variation between the two samples, but did not exceed a relative difference of 5%. Among other symptoms, irritability and difficulty concentrating were the most common in both age groups and both samples, ranging from 23 to 49%. Prevalence of symptoms other than headache is presented in Supplemental Table S1.

Fig. 2

Distribution of headache ratings in children and adolescents after TBI (N = 463) and matched controls (N = 463)

Model I – Risk of developing headache in the TBI sample vs. matched controls

Analysis of marginal effects revealed a small, non-significant increase in the odds of headache in the TBI sample compared to matched controls (OR = 1.09, 95% CI 0.85 to 1.4, p = 0.49). Including the number of symptoms other than headache in the matching did not significantly affect the results, again suggesting no significant difference (OR = 1.18, 95% CI 0.92 to 1.50, p = 0.19).

Association between the symptoms

The association between PCSI symptoms was generally stronger in adolescents than in children and in the matched controls than in the TBI sample. While Spearman correlations between the headache and other symptoms ranged from 0.12 (“Thinking more slowly”) to 0.38 (“Nausea”) in the children after TBI aged 8 to 12 years and from 0.28 (“Irritability”) to 0.43 (“Blurred vision” and “Feelings slowed down”) in the matched controls (Fig. 3, Panel A), the correlations in the adolescent samples varied from 0.26 (“Visual problems”) to 0.52 (“Fatigue”) after TBI and 0.44 (multiple symptoms) to 0.62 (“Nausea”) in the controls (Fig. 3, Panel B).

Fig. 3

Correlations between symptoms in children aged 8–12 years (Panel A) and in adolescents aged 13–17 years (Panel B). TBI samples are shown in the lower part of the correlation matrix (A1 and B1, respectively), matched controls are shown in the upper part of the correlation matrix (A2 and B2, respectively). Deeper colors represent higher correlations, the main diagonal represents correlations of the same symptoms in both samples

Model II – Risk of developing headache in the TBI sample

Logistic regression analysis revealed a significant effect of age on the probability of developing headache after TBI (Table 3). Specifically, the linear age term (OR = 2.00 per year, CI 1.01 to 3.99, p = 0.048) was significant. The Nagelkerke’s R² = 0.08 indicated a slight increase between the null model without predictors and the model estimated. The AUC = 0.64 suggested a poor ability of the model to distinguish those without headache from those with headache.

Table 3 Results of logistic regression analysis (Model II)

In line with this, only the predictors age (OR = 1.26 per year), sex (OR = 0.99 for male) and repeat TBI (OR = 1.07 if repeated) ended up in the reduced model by elastic net regularization. In the cross-validation, the accuracy of the regularized model was 61%, significantly above the no information rate of 54% (p = 0.036).

Discussion

The purpose of the present study was to examine the prevalence of headache after pediatric TBI and to investigate factors associated with the probability of developing headache in a post-acute TBI sample, also in comparison with matched controls from the general German population, which has not been done to date.

When controlling for age, sex, and the presence of chronic health conditions, the differences in the relative frequency of reported headaches between children and adolescents from the post-TBI and general populations do not exceed 5%. Overall, the results indicate no significant increase in the odds of headache in children and adolescents in the post-acute TBI phase compared with matched controls. However, there is an age effect, suggesting that the probability of reporting headache increases with age in children and adolescents.

An increasing tendency for children and adolescents to suffer from headaches has been reported in recent years [32]. Our findings on the prevalence of headache in the matched control sample of 44% are largely consistent with previous findings [32, 33]. Given the high prevalence of headache in (relatively) healthy young people, it is particularly important to understand the impact of a traumatic event such as TBI on the development and persistence of headache, also in terms of differential diagnosis. Since headache has a negative impact on HRQoL [34], particularly in terms of physical functioning and social and psychological well-being [35], it is important to treat it to allow children and adolescents to lead a normal life.

The prevalence of headache in the TBI sample reaches 46%, which exceeds the previously reported frequency of non-specific chronic headache after TBI [7]. Given that the vast majority of the sample sustained mild TBI without detectable neuroimaging abnormalities, this finding deserves special attention. On the one hand, this relatively high prevalence may be due to the way headache was defined in the present study. The post-TBI version of the PCSI requires an assessment of symptom intensity compared with the pre-injury period, which may be affected by recall bias and is not sufficient to make a clinical diagnosis. On the other hand, it may be an indication of headache onset independent of the traumatic event. Post-concussion symptoms in general have repeatedly been found to be TBI-independent, following non-brain injuries or occurring in otherwise healthy populations [36]. In light of these considerations, the high frequency of self-reported headache in the post-acute phase of TBI suggests the need for further attention, special clinical implications (e.g., thorough medical history review, appropriate diagnosis, and timely treatment), and long-term follow-up of affected children and adolescents.

Headache intensity correlates with that of other symptoms, and vice versa, especially in adolescents. However, it is not evident whether other symptoms cause the development of headache or whether headache contributes to the development of more/additional symptoms. Some recent reviews have reported the presence of a greater number of other symptoms as critical to the development of headache both after TBI [3] and in the general pediatric population [37]. Therefore, screening for multiple symptoms such as fatigue, sleep disturbance, difficulty concentrating, clumsiness, etc. in addition to headache would provide a more accurate symptom profile after pediatric TBI and guide further intervention strategies. By comparing the scores of children and adolescents after TBI with those of the general population, conclusions could be drawn more easily. Such reference values have recently been provided for two age-adjusted German post versions of the PCSI [18, 20] and are recommended for use in both clinical and research settings.

According to some studies, different types of headaches in children and adolescents tend to peak around 13 to 15 years of age [38,39,40]. Headaches can be a common symptom of a changing and maturing body, caused by hormonal changes, growth stimulation, but also by other reasons such as physical or mental health problems [37]. In this case, post-TBI headaches may also be exacerbated. For this reason, appropriate developmental considerations and a detailed differential diagnosis should be made [41].

In contrast to previous studies [2, 3, 8, 9], other factors such as gender, repeat TBI, time since injury, TBI severity, presence of lesions on neuroimaging and chronic health complaints had no significant association with post-TBI headache. However, both gender and repeat TBI remained in the reduced model by elastic net regulation, indicating importance of both factors.

Our findings are indicative at least in the post-acute phase of TBI, but are not generalizable to the acute TBI setting. In conclusion, we recommend paying special attention to the multiple symptoms reported and to the age of the patients, complemented by a thorough diagnosis and comparison with children and adolescents from the non-TBI population, to prevent symptom burden, also in the post-acute TBI phase.

Limitations

The present study is the first to examine the occurrence of headache in children and adolescents after TBI in comparison with the general population. However, some limitations should be noted. First, the information on headaches was based on a single self-report item with a different type of response scale for children and adolescents, which may have overestimated the prevalence. Therefore, we were unable to distinguish between different types and pattern of headache (e.g., migraine or cluster headache [42]). Further research focusing on different types of headaches may contribute to a better understanding of headache incidence, also in comparison with non-TBI populations. Second, the study sample is a post-acute sample with a relatively long time since injury. Although time since injury does not appear to significantly influence outcome, further studies in acute post-injury populations are indicated, especially given that we cannot distinguish whether the occurrence of headache is directly related to TBI or caused by other conditions or recall bias. Third, longitudinal studies at multiple time points would clarify the course of headache after TBI. In addition, a more consistent assessment of the presence of chronic health conditions in both samples would allow for more accurate comparisons of the occurrence of headache with or without TBI. Finally, some factors known to influence the development of headache, such as migraine (including family history), could not be included in the model due to small case numbers or unavailability of data. In addition, the interaction terms could not be considered because of the group size of some variables included in the model. Overall, further studies are recommended to improve the understanding of this common post-TBI symptom.

Conclusion

The results of the present study suggest that the probability of occurrence of headache in the post-acute phase of pediatric TBI is not different from that in the general population. Young patients of adolescent age are however at higher risk, which may interfere with common symptoms of puberty. With this information in mind, adolescents should be targeted and followed up appropriately for timely treatment and prevention of chronicity.