Study design
This cross-sectional study employed purposive sampling, adhering to STROBE guidelines [20]. Participants received comprehensive information and provided voluntary informed consent. Ethical principles following the Declaration of Helsinki were upheld [21], with approval from the Centro Superior de Estudios Universitarios La Salle ethics committee (CSEULS-PI-034/2019).
Participants
The participants included in the study were required to meet the proposed inclusion criteria. Male and female adult patients aged between 18 and 65 years were recruited and were required to have a good command of the Spanish language. Participants in the CM group were recruited from a clinic specializing in treating patients with temporomandibular disorders, headaches, and craniofacial pain (Madrid, Spain). The patients had to have a previous medical diagnosis and meet the CM criteria of the International Classification for Headache Disorders [22], which are as follows: a) headache frequency ≥ 15 days per month; b) migraine symptom frequency ≥ 8 days; c) chronicity ≥ 3 months; and d) a history of migraine starting before the age of 50 years. The following cases were excluded: a) previous cervical and cranial trauma; b) infectious or tumor diseases; and c) recent surgical procedures (in the previous 12 months).
The control group consisted of asymptomatic individuals with no history of head and neck pain for at least one year and who did not require any medical treatment or physiotherapy. Participants in this group were intentionally age-matched with those in the CM group to achieve similar groups. The control group was recruited through social networking from a university population.
Procedure
After giving their consent to participate in the study, all participants were given a set of questionnaires, which included a socio-demographic assessment and were asked to complete a series of self-reports: Head Impact Test (HIT-6) [23], International Physical Activity Questionnaire (IPAQ) [24], Tampa scale of Kinesiophobia (TSK-11) [25], Pain Catastrophism Scale (PCS) [26], chronic pain self-efficacy scale (CPSS) [27], Pain Behaviors Questionnaire (PBQ) [28].
After the participants had completed the self-report measures, the following physical measures were assessed: cervical range of motion, lumbar range of motion, cervical flexor muscle endurance, maximal isometric contraction in cervical flexion and extension, maximal lumbar isometric contraction and handgrip dynamometry. The assessor was a physical therapist blinded to the participants’ condition. All patients’ assessments were done during the interictal periods.
Physical measures
Cervical range of motion
Cervical range of motion (CROM) was measured with a cervical range-of-motion device referred to as a CROM (Performance Attainment Associates, Lindstrom, MN) [29], which consists of three independent inclinometers, one for each plane of motion, attached to a plastic frame similar to a pair of glasses. The cervical ranges of motion measured were 1) flexion-extension, 3) right-left lateral flexion, 5) right-left rotation. The CROM has proven to be a valid and reliable tool for measuring the range of motion of the cervical region [30].
Lumbar range of movement
Lumbar flexion range of motion was assessed with a digital inclinometer based on the iHandy mobile application. To perform the measurement, the assessor holds the mobile device over the participants’ sacrum and applies light pressure while the participants perform a lumbar flexion movement. This measurement has been shown to have good intra-rater and inter-rater reliability, with an intra-class correlation coefficient ≥ 0.86 [31].
Endurance of the cervical musculature
The endurance of the cervical musculature was measured with the deep neck flexor endurance test, which has good reliability [32]. Participants were placed in the supine position. The examiner raised the participants’ head 2.5 cm above the couch and instructed the participants to hold this position for as long as possible. The examiner then let go of the participants’ head, leaving it suspended, held in place only by the participants’ muscle exertion.
Endurance of the lumbar musculature
Lumbar extensor muscle strength was assessed using the Ito test. Participants lay prone with a 10-cm pillow beneath their lower abdomen to reduce lumbar lordosis. With arms parallel to the body axis, they raised their upper body to a 15° angle, maintaining a neutral cervical spine position, and both feet on the couch. The test continued until fatigue, with termination upon a >10° decrease in trunk angle. Two brief practice attempts (5 seconds) ensured correct execution. The Ito test is a valid and reliable measure of lumbar extensor muscle strength [33].
Cervical strength
Maximal isometric contraction (MIC) of cervical flexion and extension was assessed using a calibrated handheld digital dynamometer (MicroFET 2 dynamometer, Hoggan Health Industries, Salt Lake City, UT). The dynamometer, with a cushioned pad, was placed on the area to be assessed. For flexion MIC, participants were supine, with the pad on their forehead, performing maximal craniocervical flexion. For extension MIC, participants were prone, with the pad on the occipital area, resisting the assessor's opposing force. Each movement was tested three times for 5 seconds with a 60-second rest, showing good reliability [34].
Lumbar strength
The extension MIC of the lumbar region was measured using a foot dynamometer (Takei TM 5420, Takei Scientific Instruments CO., Niigata City, Japan). This device has been validated and can be used to determine leg and back strength in held positions, provided that the measurement protocol is standardized (r, 0.91; P < 0.001) [35]. For the measurement, the participants let their arms hang down to hold the dynamometer’s bar with both hands with the palms facing the body. The dynamometer chain was then adjusted so that the knees were flexed to approximately 110°. The evaluator took 3 measurements; the mean was used in the data analysis.
Handgrip strength
Isometric handgrip strength was measured using a JAMAR hydraulic handgrip dynamometer (Sammons Preston, Rolyon, Bolingbrook, IL), following the procedure recommended by Roberts et al. Participants sat upright with feet flat on the floor, elbows flexed at 90°, and wrists and forearms in a neutral position [36]. Grip strength was recorded thrice on the dominant hand with a 30-second interval between measurements. This test demonstrates excellent reliability across various populations and conditions [37–39].
Psychological and disability measures
Headache-related disability
Disability was assessed using the Spanish HIT-6, comprising 6 items to measure headache-related disability in CM patients [23]. The questionnaire exhibits acceptable psychometric properties and validation for CM patients [40]. Scores range from 36 to 78 points, categorized into four severity levels: little or no impact (36–49), some impact (50–55), substantial impact (56–59), and severe impact (60–78).
Pain behaviors
The PBQ assesses pain-related behaviors, initially validated in headache patients [41, 42]. The Spanish version, validated in migraine and tension headache patients, exhibits strong psychometric properties, comprising 19 items across six factors: avoidance behaviors (5 items), active non-verbal complaint (4 items), passive non-verbal complaint (3 items), verbal complaint (3 items), rest (2 items), and medication (2 items) [28].
Fear of movement
We assessed fear of movement using the Spanish TSK-11, with good psychometric properties (Cronbach’s α, 0.81) [25]. It has 2 subscales: one for fear of physical activity and another for fear of harm. Each of 11 items was scored 1–4 (1 = “strongly disagree”, 2 = “disagree”, 3 = “agree”, 4 = “strongly agree”), yielding scores from 11 to 44.
Physical activity level
The IPAQ assessed participants' physical activity, categorizing them into three levels (high, moderate, low/sedentary) and estimating activity in METs. IPAQ's psychometric properties are accepted; it has a reliability of about 0.77 (95% CI 0.67–0.84) [43].
Pain intensity
Self-reported pain intensity was assessed using the numerical pain scale (NPS) (0–10/10). A score of 0 indicates “no pain”, while a score of 10 indicates “maximum possible pain intensity” [44].
Sample size
The sample size was estimated with G*Power 3.1.7 (G*Power of the University of Düsseldorf, Germany) [45]. A pilot study was conducted with 16 patients with CM and 16 asymptomatic participants to determine differences using Student’s t-test and effect size to compare cervical muscle endurance variables and the handgrip dynamometry pressure. The study employed an alpha error level of 0.05, a statistical power of 80% (1-B error) and an effect size d (0.68 and 0.81). The estimated total sample size was 56 for the cervical muscle endurance variable and 40 for the handgrip dynamometry variable; ultimately, the larger sample size (28 patients with CM and 28 asymptomatic participants) was chosen. An additional 5% of the sample was included to allow for possible withdrawals that may occur during the physical evaluation. For these comparisons, the sample was finally 60 participants.
The sample calculation required for the multiple regression analysis was performed taking into account the use of 8 predictor variables, an alpha error level of 0.05, a statistical power of 99% (1-B error), an R2 of 0.2 and an effect size f2 of 0.25, resulting in a total sample estimate of 65 patients with CM.
Statistical analysis
All analyses were performed using SPSS statistical software, version 27.0 (SPSS Inc., Chicago, IL). Statistical analyses were performed at a 95% confidence level; P-values < 0.05 were considered statistically significant. To compare descriptive statistics, physical variables and self-reports scores between the CM group and asymptomatic participants 𝑡-test for independent samples was used and the chi-square test was used to compare categorical variables. Effect sizes (Cohen’s 𝑑) were calculated for the outcome variables. According to Cohen’s method, the effect size was classified as small (0.20–0.49), moderate (0.50–0.79) or large (≥0.8) [46].
An analysis of covariance (ANCOVA) was used, it included “physical activity level” as a covariate for between-group comparisons of physical measures and kinesiophobia. For this analysis the effect size was estimated with partial eta squared (ηp2).
The relationship between headache-related disability and physical and psychological variables in the CM group was examined using Pearson's correlation coefficients. A Pearson correlation coefficient > 0.60 indicated a strong correlation, a coefficient between 0.30 and 0.60 indicated a moderate correlation, and a coefficient < 0.30 indicated a low or very low correlation [47].
A stepwise multiple linear regression analysis was used to estimate the strength of the association between headache-related disability (criterion variable) and psychological, behavioral, and physical variables (predictor variables). Only variables that obtained moderate correlations in the correlation analysis were included in the regression model. We assessed multicollinearity in the models using the Variance Inflation Factor (VIF). A VIF near 1 implies minimal multicollinearity; values between 1 and 5 suggest moderate correlation among predictors; and a VIF over 10 indicates significant multicollinearity.
In the development of our multiple linear regression model, we have performed comprehensive diagnostic analyses to verify its robustness and adherence to key assumptions.
The model's evaluation began with an investigation into the homogeneity of variance, which is crucial for the reliability of the regression estimates. A graphical approach was employed, plotting residuals against fitted values to visually inspect the data. This plot revealed a random dispersion of residuals with no apparent patterns or funnels, leading us to confirm that the variance of the residuals is consistent across all levels of the independent variables, thereby satisfying the homogeneity criterion.
Another critical assumption, the independence of observations, was rigorously tested using the Durbin-Watson statistic. This test is instrumental in detecting any autocorrelation in the residuals that could compromise the integrity of the regression analysis. A value of the statistic proximate to 2.0 was indicative of the absence of autocorrelation, thereby upholding the model's assumption of independent observations.
Moreover, the normality of the residuals' distribution was scrutinized using Q-Q plots. These plots provided a visual assessment by comparing the distribution of the residuals to a perfectly normal distribution. The alignment of the residuals along a straight line on the Q-Q plots suggested that the distribution of the residuals aligns well with the assumption of normality.
Lastly, the linearity assumption was substantiated by examining scatter plots of observed versus predicted values and normal P-P plots of standardized residuals. These assessments disclosed a clear linear trajectory, thereby reinforcing our confidence in the linearity of the relationship modeled.