Design
This was a descriptive, observational, cross-sectional study, including patients with non-traumatic neck pain and a control group of asymptomatic participants based on a previously described study protocol [31].
This study was designed and the findings are reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) [32].
The study was approved by the Research Ethics Committee of CEU San Pablo University (495/21/39). Participants provided informed written consent before being enrolled into the study and they were able to withdraw their consent at any time during the study, in compliance with the WHO standards and the Declaration of Helsinki [33].
Sample and selection
The sample was composed of a group of patients with non-traumatic subacute and chronic neck pain and another group of asymptomatic subjects. We recruited a non-probabilistic convenience sample via flyers, online forms through social networks, e-mail, or direct verbal communication at San Pablo-CEU University, the CEU San Pablo University clinic, as well as in private physiotherapy clinics in the Community of Madrid.
Patients with neck pain were eligible to be included in the study if they were 18–65 years old and fulfill the following selection criteria: (a) neck pain of at least 1 month of evolution, (b) neck pain from non-specific mechanical origin associated or not with primary headache (e.g. migraine or tension-type headache), shoulder pain, or upper limb pain. Patients with neck pain were excluded if they presented any of the following criteria: (a) complex regional syndrome, (b) previous surgeries in the neck and/or head region, (c) vestibular alterations, (d) otogenic or idiopathic vertigo/dizziness, (e) presence of tumors in the craniocervical region, (f) previous fracture in the head or neck region, (g) osseous deformities in the thoracic, cervical, or cranial region.
Asymptomatic subjects should not have any pain in the cervical region during the last year and no previous treatment for neck pain. The exclusion criteria were the same as described for patients with neck pain.
Once deemed eligible, subjects were asked to read and sign the informed consent prior to participation and then were invited to participate in the study, consisting of one session in which all variables described below were assessed.
Instrumentation and measures
Prior to sensorimotor control testing, subject demographic characteristics such as age, gender, weight, height, and dominant side were recorded. In addition, the following pain or descriptive variables were collected for all participants: (a) mean intensity of pain during the last month (visual analog scale), (b) neck disability measured using the Spanish version of the Neck Disability Index (NDI) [34], (c) duration of pain since pain started (months). Participants were considered asymptomatic in case these three variables were recorded with a score of 0. Then, these variables were included in the regression statistical analysis described below as continuous variables.
Primary outcomes: sensorimotor control assessment
Sensorimotor control kinematic variables were recorded by means of small (4 cm × 4 cm × 8 cm), light (< 200 g) Inertial Measurement Unit (IMU) sensors (Werium Solutions©, Madrid, Spain), which integrate a 3D accelerometer, a gyroscope, and a magnetometer. This inertial sensor technology has previously shown good to excellent intra-rater and inter-rater reliability in the measurement of general cervical ROM (ICC = 0.93) [35] and craniocervical flexion (ICC > 0.80) [36].
An independent assessor administered testing procedures, blinded to subject group status (neck pain or asymptomatic). Another researcher recorded the demographic variables and was aware of the subject’s pain status. As reported by similar previous research [37, 38] performance and detection biases during this type of testing procedure are less likely, since these procedures use automatic computerized data collection and processing.
The kinematics of the craniocervical flexion movement in supine position were evaluated for each subject enrolled in the study. Participants received instructions on testing procedures (see below). During the test, the assessor was able to monitor on real time the values of ROM displayed on a computer screen (Fig. 1).
Before starting with the test, participants were asked to sit naturally in a standard chair with the feet well supported on the floor and the neck and head in a neutral comfortable position with their hands resting on their thighs. Patients performed three repetitions of flexion/extension, lateral inclination, and rotations (to both sides), serving as a warm-up exercise before the measurement of craniocervical flexion.
Then, the assessor asked the subjects whether they had any questions before starting the test. Then subjects were asked to perform three consecutive active movements of the craniocervical flexion movement in supine position as described in the section below.
Secondary outcomes
All participants completed a questionnaire related to fear of movement, which is also described below. The Spanish version of the Tampa Scale for Kinesiophobia (TSK11) [42] was used to assess fear of movement and injury. This self-reporting questionnaire includes 11 items that are rated on a four-point scale, where 4 represents “strongly agree” with the statement and 1 represents “strongly disagree”. Scores range from 11 to 44, where higher scores indicate higher fear of movement. This Spanish version showed good reliability and validity, with an internal consistency of α = 0.79.
Craniocervical flexion in supine position
One wireless wearable sensor was adhered to the center of the forehead (defined as the place where the lines that bisect the forehead longitudinally and horizontally cross in supine position) before starting the test (Fig. 2a).
Placement of the sensor with these landmarks has shown to be a reliable method for measuring craniocervical flexion ROM in previous research [36]. Participants were placed in a relaxed supine position with the forearms resting on the abdomen, the knees flexed, and the neck in a neutral comfortable position. The assessor visually assessed that the cranio-cervical spine was in a mid-position in which the subjects’ chin and forehead were horizontal and an imaginary line, which extended from the tragus of the ear to bisect the neck longitudinally, was parallel to the plinth[39]. Subjects were reminded to stay and memorize the starting neutral position of the head (Fig. 2a) and return to this position as accurately as possible after each of the three repetitions. Once the participant was in this position, the sensor was set and calibrated to the starting position. At this moment, patients were asked to perform three repetitions of the full-range craniocervical flexion movement (Fig. 2b), consisting of an anterior rotation of the head in a nodding action, feeling the back of their heads sliding up on the table. During the test, the assessor provided standardized verbal cues to guide the process with a correct technique if necessary.
Additionally, the movement was repeated in case the assessor detected signs of compensation, such as lower cervical flexion, neck retraction, or overuse of the superficial flexor muscles.
This procedure was verbally explained to participants through the following standardized instructions to ensure same information was explained to all subjects: “Please lie on your back with your knees bent and your feet resting on the table. The starting position will be relaxed with the gaze directed vertically towards the ceiling. Then, please perform an anterior rotation of the head in a controlled manner without moving your neck, such that the head rotates slightly, reaching as far as possible. The posterior side of the head will slide smoothly on the table during the movement and the head should not separate from the table or push into the table during the movement. You must perform this movement three times, returning to the starting position after each one and holding the position of maximum flexion for three seconds in each repetition”.
Sensorimotor control data processing
A software application computed and exported the complete ROM of the participants, expressed as angles from the calibrating starting position in the three axes, sampled every 20 ms. In addition, data smoothing filters were applied to avoid peaks and variations not corresponding to the trend of each data set. The analysis of these data allowed for the calculation of the following variables during three repetitions of craniocervical flexion:
-
Active craniocervical flexion ROM expressed as the maximal angular displacement (°) achieved in any of the three repetitions
-
Active craniocervical extension ROM when trying to return to the neutral position from the craniocervical flexion motion, also expressed as the maximal angular displacement (°). This variable represents poorer accuracy when trying to achieve the neutral position. It is different from the variable “Head repositioning accuracy”, since this excessive extension motion was frequently observed immediately before the patient tried to further change the position to come to the point in which they were supposed to have achieved the real neutral position between repetitions.
-
Peak velocity in craniocervical flexion and extension independently, expressed as the maximal angular velocity (°/s): calculated as the discrete derivative of angular orientation applying a standard smooth filtering algorithm.
-
Smoothness of motion expressed as maximal movement jerk peak (º/s3): calculated as the third discrete derivative of the angular orientation (change in acceleration) [40, 41].
-
Head repositioning accuracy expressed as angular displacement (°): calculated as the absolute repositioning error considering the maximal difference between the neutral starting position (set up at the beginning of each movement) and the positions reached when the patient tried to come back again to the neutral after returning from flexion. The maximal difference observed during all the three repetitions was used for the analysis.
-
Peak conjunct motion (°): calculated as the angular displacement occurring in a different anatomical plane to the one that is being tested (movement in frontal or transverse plane during sagittal plane flexion–extension).
Data analysis plan
All data were analyzed using the Statistical Package for Social Sciences (SPSS) software version 24.0 (SPSS Inc, 233 S WackerDr, 11th Fl, Chicago, IL 60606) and R Ver. 4.1.3 program. (R Foundation for Statistical Computing, Institute for Statistics and Mathematics, Welthandelsplatz 1, 1020 Vienna, Austria). The level of significance was established at p < 0.05. The distribution of the quantitative variables was tested with the Kolmogorov–Smirnov test with Lillierfors correction, which showed the absence of normality. Quantitative variables are shown as mean ± standard deviation and categorical variables with absolute and relative values (%).
Nine multivariate linear regression models were built between each of the sensorimotor control dependent variables (i.e., active maximal ROM in flexion and extension, peak velocity in flexion and extension, peak jerk in flexion and extension, peak conjunct motion in lateral and rotation movements, and head repositioning accuracy) and the secondary variables of age, previous month visual analog scale (VAS), TSK11 score, pain duration, and NDI score. The VAS score was included in the model instead of the group categorical variable (pain or asymptomatic) in order to maintain the power inherent in a continuous variable, which could be lost in a discrete categorical variable with few levels defined [43]. All patients with neck pain reported some degree of pain in the last month.
The assumption of linearity between all dependent and independent variables was tested by visual inspection of the correlation graph and a value of the effective degrees of freedom (EDF) close to 1. When this assumption was not met, a generalized additive model (GAM) was applied with the double penalty method in the selection of variables, modelling the variables with the parametric or smoothed model based on fulfilling the assumption. In all models, compliance with concurrency assumptions was tested for the smoothed terms, eliminating those with a value greater than 0.8, and multicollinearity in the parametric terms, eliminating those with a variance inflation factor (VIF) greater than 2. The distribution of residuals and adjusted values around the null value and adequacy of the number of basic functions with a non-significant K index were also checked.
In the case of the peak velocity variable, a linear model was applied when the assumptions of linearity, homoscedasticity, normality of the residuals, and absence of autocorrelation were fulfilled, while in the maximum lateral flexion a weighted least square (WLS) model was applied to be able to handle the non-normality of the residuals.
Sample size
The sample size of this study was determined using G*Power, Version 3.1.9.2 (Franz Faul et al. University at Kiel, Germany), considering the results from a pilot study with 20 subjects: 10 asymptomatic and 10 subjects with neck pain. Sample size was calculated using a linear multiple regression (fixed model), with 0.95 power (1- beta error probability) and an alpha level of 0.05 [44]. Considering craniocervical flexion ROM as a dependent variable in a model with five independent variables, a total sample size of 160 subjects was estimated considering a partial R2 = 0.076 and an effect size of 0.082. Considering the probability of technical errors related to the automatic record of data from inertial sensors, an additional 20% of patients was estimated (n = 192).