Perceived fear and exercise difficulty in patients with migraine and their association with psychosocial factors: a cross-sectional study

Study design

An observational cross-sectional study was conducted from March 2022 to September 2024 to evaluate the perceived fear and difficulty reported by patients with migraine while observing various exercises and the association of these scores with psychological variables. We used non-probabilistic convenience and snowball sampling, employing social networks to spread the study information. The study was performed according to the Strengthening the Reporting of Observational Studies in Epidemiology declaration (von Elm et al., 2008). It followed the principles of the Declaration of Helsinki and obtained the approval of the Ethics Committee of the University La Salle (CSEULS-PI-005/2022) on the 3^rd of February 2022. All participants gave written informed consent to participate.

Participants

Patients with episodic or chronic migraines, with or without visual or motor aura, were included. They could have comorbidities commonly associated with migraine, such as tensional headache, irritable bowel syndrome, or fibromyalgia.

Patients were included if they (1) were 18 to 65 years of age; (2) had physician-diagnosed migraine; (3) had their migraine diagnosis corroborated by the Penn Online Evaluation of Migraine (POEM) questionnaire; and (4) were fluent in Spanish. Patients were excluded if they: (1) had a history of cervical, facial, or cranioencephalic trauma that could have started the headache; or (2) were pregnant. Severe comorbidities, such as undergoing cancer treatment, were also considered for exclusion.

Procedure

First, patients received an e-mail containing links to a video call and the questionnaires, which were adapted online with Cognito Forms (https://www.cognitoforms.com/), a free online form builder. Upon opening the first link and reading the informed consent, they gave their consent by clicking the “yes” button. Next, patients confirmed the inclusion criteria and migraine diagnosis by completing the POEM questionnaire. The evaluator always supervised this process during the video call to guide patients through the test and to address any questions that arose. After confirming the inclusion criteria and migraine diagnosis, participants proceeded to the next part of the evaluation.

The second part consisted of various sociodemographic and anthropometric questions (age, height, weight, academic level, and work status) and included inquiries about other health conditions, such as additional headache diagnoses (tensional headache, cluster headache, hemicrania continua, or new persistent daily headache) or comorbidities (cardiovascular diseases, type 2 diabetes mellitus, rheumatoid arthritis, or fibromyalgia). Following these questions, participants completed the Spanish versions of the short form of the International Physical Activity Questionnaire (IPAQ-SF), the shortened version of the TAMPA Scale of Kinesiophobia (TSK-11), the Pain Catastrophizing Scale (PCS), the Fear-Avoidance Beliefs Questionnaire (FAB-Q), the Headache Impact Test (HIT-6), the Neck Disability Index (NDI), the Chronic Pain Self-Efficacy Scale (CPSES), and the Self-Efficacy Questionnaire to Regulate Exercise (SEQRE).

Lastly, the third part involved assessments of nine different exercises (walking, running, jumping, neck rotation and flexion/extension, spine flexion and extension, squatting, and shoulder press) to evaluate the perceived fear of initiating or aggravating a migraine episode and the difficulty of performing the exercise on a scale from 1 to 10. Given the differences in perception between interictal and migraine phases, patients were advised to consider both moments and weigh their final scores accordingly. The selection of these exercises was based on their common inclusion as aerobic (walking, running, and jumping), mobility (neck rotation, flexion, and extension, and spine flexion), and strength exercises (spine extension, squatting, and shoulder press) in conventional exercise and rehabilitation programs. The evaluator showed videos of these exercises to the participants to facilitate the scoring process (available in Material S1). In each video, a female demonstrator briefly performed each exercise to ensure clarity on what was being described. After viewing the videos, participants rated their perceived fear and difficulty for each exercise.

Outcomes

The POEM questionnaire is a validated instrument for confirming migraine diagnosis, with specificities of 84%, 93%, and 90% for migraine without aura, migraine with aura, and overall migraine diagnosis, respectively (Kaiser et al., 2019).

The Spanish version of the IPAQ-SF was used to evaluate each participant’s physical activity level. This questionnaire is valid and reliable, showing moderate concordance between both long and short forms (r = 0.67; 95% CI [0.64–0.7]) and reliability (r = 0.76; 95% CI [0.73–0.77]) (Toloza & Gómez-Conesa, 2007). It includes seven items that evaluate the days per week and minutes per day of vigorous, moderate, and low physical activity (walking), as well as the minutes per day spent sitting. The physical activity level is then calculated in metabolic equivalents for vigorous, moderate, and low intensities, and for total physical activity.

Kinesiophobia, a construct that assesses fear of pain and movement, was evaluated with the Spanish version of the TSK-11, which is composed of two subscales: fear of physical activity and fear of injury. This is a valid and reliable instrument (Cronbach’s α = 0.79; intraclass correlation coefficient (ICC) > 0.7) (Gómez-Pérez, López-Martínez & Ruiz-Párraga, 2011).

Catastrophism was assessed with the Spanish version of the PCS. This scale contains three domains: rumination, magnification, and hopelessness. It is a valid and reliable instrument (Cronbach’s α = 0.79; ICC = 0.84) (Campayo et al., 2008). The authors have permission to use this instrument from the copyright holders.

The FAB-Q was employed to evaluate fear-avoidance beliefs and behaviors related to physical activity and work. It is a valid and reliable instrument (Cronbach’s α = 0.93; ICC = 0.97) (Kovacs et al., 2006). However, items 3 and 11 were eliminated because they evaluate the perception that physical activity/work might harm the back. A reliability analysis was conducted after recruitment to ensure high internal consistency, even if these items were removed.

Headache-related disability was assessed using the HIT-6 questionnaire, which has a high internal consistency (Cronbach’s α = 0.87) (Martin et al., 2004). It consists of six questions related to pain severity, limited daily activities, frequency of wishing to lie down, tiredness to do work and daily activities, frequency of feeling irritated, and limited concentration.

Neck pain is a common comorbidity among patients with migraine (Ashina et al., 2015). Therefore, the Spanish version of the NDI questionnaire was included to evaluate the disability generated by neck pain. This version is a valid and reliable instrument (Cronbach’s α = 0.89; ICC = 0.98) (Kovacs et al., 2008; Ortega, Martínez & Ruiz, 2010). The authors have permission to use this instrument from the copyright holders.

The self-efficacy construct was assessed with the Spanish version of the CPSES. This questionnaire contains three domains: self-efficacy for pain management, physical functioning, and coping with symptoms. The validity of this instrument is good (Cronbach’s α = 0.91), and it is reliable (ICC = 0.75) (Martín-Aragón et al., 1999). The authors have permission to use this instrument from the copyright holders.

Lastly, the short form of the SEQRE was used to evaluate the perceived self-efficacy to maintain a daily exercise routine. The Spanish version shows good internal consistency (Cronbach’s α = 0.84) (de los Fuentes Vega & González Lomelí, 2020). It contains seven different contexts that could be interpreted as barriers to exercise adherence, and each item is scored from 0 (‘I cannot do it’) to 100 points (‘I am confident I can do it’).

Sample size calculation

A sample size calculation was made based on a multiple linear regression evaluating the association between the analyzed variables and the perceived fear of jumping. The sample size required in this study for the results of this regression analysis was calculated using an alpha level of 0.05 and a 90% power. We calculated the effect size (f² = 0.19) based on the adjusted coefficient of determination (R² = 0.16). These parameters and results were introduced into the G*Power 3.1 software (University of Düsseldorf, Düsseldorf, Germany) (Faul et al., 2007). A total sample size of 109 participants was estimated based on these parameters.

Statistical analysis

All statistical analyses were performed with R software (version 4.4.1; R Core Team, 2024) in the RStudio environment (RStudio Team, 2023) version 2023.06.0+421. Descriptive analyses were conducted to summarize the demographic and clinical characteristics of participants, including age, height, weight, body mass index (BMI), sex, academic level, employment status, and migraine characteristics. Mean, standard deviation, medians, minimum, maximum, and quartiles 1 and 3 were reported for continuous variables. Categorical variables were summarized using frequencies and percentages.

The primary aim was to assess differences in perceived fear and difficulty across nine exercises. Given the nature of the data, which consisted of bounded ordinal variables, robust statistical methods were applied, using the WRS and WRS2 packages (Mair & Wilcox, 2019; Wilcox, 2021). We used a robust one-way analysis of variance (ANOVA) for medians (bd1way function), with 2,000 bootstrap resamples, given the non-normality of the data. Multiple comparisons of medians between exercises were conducted using the percentile bootstrap method (dmedpb function). Results are presented as median difference values along with their 95% percentile bootstrap confidence interval and interquartile ranges.

For the secondary aim, beta regression models were employed to examine the relationships between the total scores of the various questionnaires (IPAQ-SF, TSK-11, PCS, FAB-Q, HIT-6, NDI, CPSES, and SEQRE) and the perceived fear and difficulty during each exercise. Only those exercises with the highest scores for fear and difficulty were selected for the beta regression analyses. These scores were standardized to values between 0 and 1 and were analyzed with the betareg package. Specifically, the distribution of these standardized values was examined to ensure they were bounded between 0 and 1 (Ferrari & Cribari-Neto, 2004). Transformations were applied to ensure that values which reached exactly 0 or 1 were shifted slightly (e.g., 0.0001 and 0.9999) to fit within the bounds required for beta regression.

Before running the beta regressions, the assumptions of the model were evaluated. Model diagnostics such as residual vs fitted, Cook’s distance, leverage, and Q-Q plot measures were evaluated to check for violations of assumptions.

To interpret significant findings from the beta regression models, predicted values and their 95% prediction intervals were calculated and plotted for each questionnaire score. The relationships between variables were visualized using smooth curves.

We also included two Spearman correlation analyses to evaluate the strength of the association between the perceived fear in different exercises and the questionnaires’ scores, and between the perceived difficulty in various exercises and the questionnaires’ scores (Schober, Boer & Schwarte, 2018).

The internal consistency of the FAB-Q questionnaire was analyzed with a reliability analysis. Both Cronbach’s α (0.87) and McDonald’s ω (0.88) indicated high internal consistency (results are available in Material S2).

Descriptive statistics

A total of 110 patients with migraine participated in the study, with 88% identifying as female. The average age was 36 years (SD ± 12), and 88% had a college-level education. Most (63%) participants were employed, and 25% were students. The cohort’s mean BMI was 25 (SD ± 4). Regarding migraine diagnosis, 53% had chronic migraine, 47% had episodic migraine, 30% had migraine without aura, 64% had visual aura symptoms, and 36% had motor aura symptoms. Table 1 shows the descriptive analysis of the participants. Histograms and pairwise correlations are provided in Material S3.

Primary analysis: perceived fear and difficulty across exercises

The one-way robust ANOVA for median values revealed significant differences in perceived fear across the nine exercises (test = 13.56; p < 0.001). Jumping was perceived as the most fear-inducing exercise, with a median value of 5. The differences in perceived fear between the exercises are presented in Fig. 1A.

Similarly, there were significant differences in perceived difficulty among the exercises (test = 8.89; p < 0.001). Spine extension was rated as the most difficult, with a median score of 4. The differences in perceived difficulty between the exercises are presented in Fig. 1B.

Secondary analysis: regression models for perceived fear and difficulty

Beta regression models were employed to assess the relationship between perceived fear of jumping and the various questionnaire scores, and between perceived difficulty of spine extension and the questionnaire scores. The assumptions of the beta regression models were examined. Although Cook’s distance showed no significant influence of each score in the various models, some participants appeared to influence the models’ estimation based on the leverage plot. However, there was no justification for their elimination. Some deviations from the half-normal plot of residuals were also presented. The assumptions of the beta regression models are included in Material S4.

Jumping fear

The IPAQ-SF was negatively associated with jumping fear (β = −0.00016, 95% CI [−0.00029 to −0.00003]; p = 0.013), whereas the FAB-Q was positively associated (β = 0.03; 95% CI [0.0094–0.0499]; p = 0.004). Both associations are shown in Fig. 2.

Spine extension difficulty

Both the IPAQ-SF (β = −0.00015; 95% CI [−0.00026 to −0.00003]; p = 0.014) and the CPSES (β = −0.0126; 95% CI [−0.0208 to −0.0044]; p = 0.003) scores were negatively associated with spine extension difficulty, whereas the FAB-Q was positively associated (β = 0.028; 95% CI [0.0091–0.0462]; p = 0.0035). These associations are presented in Fig. 3.

Correlation analyses

The results of the fear correlation analysis are summarized in Table 2. The strongest association was between the IPAQ-SF and perceived fear of running, which showed a moderately significant correlation (ρ = −0.425, p < 0.001). The results of the difficulty correlation analysis are shown in Table 3. No moderate or strong correlations were found in this analysis.

Clinical implications and future directions

Understanding the levels of physical activity, kinesiophobia, catastrophizing, fear-avoidance beliefs, headache-related disability, and self-efficacy in patients with migraine can be especially beneficial during the initial evaluation and treatment planning, particularly when considering therapeutic exercise interventions within a biobehavioral framework. Knowing these patients’ physical condition and exercise perceptions could help healthcare professionals tailor exercise modalities and prescription parameters to each patient’s abilities and perceived sense of safety, promoting a more personalized intervention.

Moreover, considering that various psychological and functional variables, such as self-efficacy and physical activity levels, have been associated with a worse migraine condition, and that increasing the amount of physical activity may directly impact symptoms, studying the relationship between these exercise perceptions and migraine symptoms in prospective and experimental studies could help clarify whether these perceptions influence migraine frequency, intensity, and duration. Perhaps, decreasing the anticipatory anxiety response triggered by exposure to these exercises could reduce migraine-related symptoms.

Furthermore, the findings of this study provide a foundation for designing graded exposure programs for exercise. Given that certain exercises were perceived as more fear-inducing or challenging, these results can help establish a progressive exposure hierarchy. Initiating treatment with exercises perceived as less intense, such as walking, and gradually progressing to more challenging exercises could reduce initial fear and improve adherence. This approach allows for a gradual reduction in psychological barriers, facilitating patient adaptation to exercise levels that could ultimately benefit their condition.

Implementing a biobehavioral intervention that integrates therapeutic education and exercise could also help address these psychological factors, enhancing exercise adherence and improving its overall effectiveness. Therapeutic education could be instrumental in modifying misconceptions and strengthening patients’ confidence in their ability to engage in physical activities without fearing migraine exacerbation.

Limitations

This study presents several limitations that should be considered when interpreting its results and guiding future research.

Firstly, the lack of randomization in the order of exercise presentation is a significant limitation. The assessment of one exercise could have influenced the perception of fear or difficulty in subsequent exercises, particularly those perceived as more intense or challenging. Randomizing the order of presentation would have minimized this carryover effect and enhanced the validity of the findings, allowing for a more accurate analysis of each exercise’s perception independently.

Additionally, modifying the FAB-Q questionnaire limited its original scope. However, the internal consistency analysis of the adapted version demonstrated good reliability, supporting its applicability in this population. Nonetheless, the adaptation could have affected comparability with previous studies that used the full version. In addition, although the IPAQ-SF questionnaire is a valid tool to evaluate the level of physical activity, its accuracy in estimating the total METs per week is limited by participants’ interpretation regarding each intensity category and their ability to recall the time they spent doing exercise. However, due to the cross-sectional design and online evaluation, using more precise instruments to evaluate the amount of physical activity, such as accelerometers, was not feasible.

Another important limitation is the use of a unidimensional scale to evaluate perceived difficulty and fear. This approach could have restricted measurement sensitivity, with potential “floor” or “ceiling” effects that limit the ability to capture more nuanced variations in patients’ perceptions. Employing a multidimensional scale in future studies could provide a more detailed understanding of the factors contributing to perceptions of difficulty and fear across various exercises.

The sample selection also presents a limitation, because it was primarily conducted through online and social media channels. This could have introduced selection bias, potentially excluding patients who might exhibit higher levels of disability and lower exercise tolerance due to a more advanced severity of their condition. Therefore, findings might not be fully generalizable to the broader population of migraine patients in clinical settings, highlighting the need for future studies to include more representative samples across different contexts. In fact, the disproportionate number of females compared to males impeded any stratified analysis or comparison based on the sex category. This could be considered a limitation raised by the non-probabilistic convenience sampling strategy. However, updated evidence shows that women have 3.25 times more risk of developing migraine than men, and even in countries like Italy the female-to-male ratio is 4.9:1 (Allais et al., 2020). Considering these data, the present study is almost in alignment with the proportion of women and men affected by migraine.

Lastly, its cross-sectional design limits the ability to establish causal relationships between variables. This design constrains the ability to determine whether physical activity levels, kinesiophobia, and other psychological and clinical factors are causes or consequences of perceived fear and difficulty during exercises. Longitudinal studies are needed to explore the directionality of these associations and provide a foundation for interventions aimed at modifying long-term risk factors. Moreover, non-evaluated factors, such as sleep quality, migraine frequency, intensity, duration, and medication use, could have influenced the model, explaining the high variability in some analyzed outcomes. However, controlling these confounders would need a more complex design. Future studies should prospectively evaluate the effect of these variables on exercise perception.