This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mHealth and uHealth, is properly cited. The complete bibliographic information, a link to the original publication on https://mhealth.jmir.org/, as well as this copyright and license information must be included.
Concomitant psychological and cognitive impairments modulate nociceptive processing and contribute to chronic low back pain (CLBP) maintenance, poorly correlated with radiological findings. Clinical practice guidelines recommend self-management and multidisciplinary educational and exercise-based interventions. However, these recommendations are based on self-reported measurements, which lack evidence of related electrophysiological changes. Furthermore, current mobile health (mHealth) tools for self-management are of low quality and scarce evidence. Thus, it is necessary to increase knowledge on mHealth and electrophysiological changes elicited by current evidence-based interventions.
The aim of this study is to investigate changes elicited by a self-managed educational and exercise-based 4-week mHealth intervention (
A 2-arm parallel nonrandomized clinical trial was conducted at the University of the Balearic Islands (Palma, Spain). A total of 50 patients with nonspecific CLBP were assigned to a self-managed group (23/50, 46%; mean age 45.00, SD 9.13 years; 10/23, 43% men) or a face-to-face group (27/50, 54%; mean age 48.63, SD 7.54 years; 7/27, 26% men). The primary outcomes were electroencephalographic activity (at rest and during a modified version of the Eriksen flanker task) and heart rate variability (at rest), PPTs, and pressure pain intensity ratings. The secondary outcomes were pain, disability, psychological functioning (mood, anxiety, kinesiophobia, pain catastrophizing, and fear-avoidance beliefs), and cognitive performance (percentage of hits and reaction times).
After the intervention, frequency analysis of electroencephalographic resting-state data showed increased beta-2 (16-23 Hz; 0.0020 vs 0.0024;
Both intervention modalities increased mainly beta activity at rest and improved psychological functioning. Given the limitations of our study, conclusions must be drawn carefully and further research will be needed. Nevertheless, to the best of our knowledge, this is the first study reporting electroencephalographic changes in patients with CLBP after an mHealth intervention.
ClinicalTrials.gov NCT04576611; https://clinicaltrials.gov/ct2/show/NCT04576611
Low back pain (LBP) is a highly experienced symptom in the general population and the main cause of disability in industrialized countries [
Evidence-based clinical guidelines consider physical exercise a key component among the nonpharmacological interventions for patients with LBP, and education has traditionally been used as an integral part of the multidisciplinary treatment, with its importance highlighted in recent decades [
Regarding physical exercise, a systematic review showed that stretching and strengthening exercises delivered with supervision may improve pain and function, respectively, in patients with CLBP [
Therefore, current interventions are inadequate because they are often based on a biomedical model, sidelining the well-documented impairments in central nociceptive processing mechanisms [
Therefore, it is necessary to clarify the usefulness of these physiological measures in patients with CLBP and the relationship of these measures to concomitant psychological (eg, pain beliefs, catastrophizing, and depression) and cognitive (eg, processing speed, memory, and executive function) alterations that may contribute to the mechanisms of central sensitization [
The goal of this study is to investigate whether a self-managed program based on education and exercise using a mobile app (
This 2-arm parallel design nonrandomized clinical trial was submitted to ClinicalTrials.gov (NCT04576611). This study is also reported according to the CONSORT-EHEALTH (Consolidated Standards of Reporting Trials of Electronic and Mobile Health Applications and Online Telehealth) statement [
A total of 59 patients with nonspecific CLBP initially participated in this study. First, participants were contacted through email or telephone using a database from a previous study [
The inclusion criteria were as follows: participants aged 18-59 years with nonspecific CLBP lasting for >12 weeks, of which they have experienced at least three episodes of LBP (lasting for >1 week) [
The sample size was calculated using GRANMO-IMIM [
After compliance to the treatment sessions was checked through the BackFit app, of the 59 participants, we excluded 6 (10%) from the analysis for having undergone fewer than 7 sessions, 1 (2%) because intensity of use was <10 minutes per session in more than one session, 1 (2%) because data were lost (server error), and 1 (2%) for a nonreported previous traffic injury; the remaining 50 (85%) participants were nonrandomly distributed (ie, considering their preferences to promote treatment adherence) into two groups of a 4-week educational and exercise program (total of 8 sessions of approximately 50 minutes’ duration;
CONSORT (Consolidated Standards of Reporting Trials) flow diagram of the progress of enrollment, intervention allocation, and data analysis.
All participants had to perform the same intervention protocol twice a week for 4 consecutive weeks, completing up to 8 sessions. Each session consisted of the following: (1) viewing a pain education video <4 minutes in duration [
Screenshots from the BackFit app showing examples of the intervention protocol (ie, session, pain rating scale, educational video, and exercise).
Sociodemographic and clinical data (using a semistructured interview, height and weight measuring scales, and a digital tensiometer [OMRON M3; OMRON Healthcare]), as well as clinical pain intensity ratings (using a digital slider integrated into the BackFit app) were collected.
All outcome measures, whether primary or secondary, were collected before and after the intervention.
Electroencephalographic signals were continuously recorded for 5 minutes in the eyes-open resting state and during the performance of a cognitive task in an acoustically attenuated room using a QuickAmp amplifier (Brain Products GmbH) at 1000 Hz sampling rate from 29 silver or silver chloride scalp electrodes placed according to the 10-20 System of Electrode Placement. Active electrodes were recorded against an average reference. A ground electrode was located at the AFz position. An electro-oculogram channel was obtained by placing an electrode above the left eye and another below the same eye. An electrocardiogram (ECG) channel was also obtained by placing an electrode at both wrists. Electrode impedances were kept below 10 k
During electroencephalography (EEG) data preprocessing performed with BrainVision Analyzer software (version 1.05; Brain Products GmbH), signals were segmented in epochs of 1000 ms (for resting-state data) or in epochs of 600 ms (−100 to 500 ms, relative to the stimulus onset for cognitive task data) and digitally filtered (high-pass filter at 0.10 Hz, low-pass filter at 30 Hz, and notch filter at 50 Hz). We corrected eye movement artifacts using the Gratton and Coles algorithm [
Regarding EEG resting-state data, frequency power densities at delta (2-4 Hz), theta (4-8 Hz), alpha (8-12 Hz), beta-1 (12-16 Hz), beta-2 (16-23 Hz), and beta-3 (23-30 Hz) were computed by using the fast Fourier transformation obtained from each artifact-free EEG epoch. A source localization of the frequency bands was also performed by using low-resolution electromagnetic tomography analysis [
Regarding the ECG data, resting-state raw signals were offline filtered (bandpass filter 0.5-30 Hz) and hand corrected for artifacts such as missed, erroneous, or ectopic beats by using QRSTool software [
Regarding EEG registration during the cognitive task data, a nonparametric cluster-based permutation test (CBPT), which allows for testing group differences in high-dimensional neural data while it deals with the multiple-comparison problem [
To assess PPTs, we used a digital algometer (FPIX 50; Wagner Instruments) at an individual unilateral low back location (spinal erector muscle, 2 cm from the spine at the most painful point) and at the forefinger (control) 3 consecutive times in counterbalanced order (maximum pressure of 5 kg/cm2). Subjective pressure pain intensity ratings were measured using a visual analog scale (0-10). The average of 3 measurements of both variables was used for the statistical analysis. Algometry was always conducted by the same researcher (OVR).
Handedness, physical disability, mood, anxiety, fear of movement, pain catastrophizing, and fear-avoidance beliefs were self-assessed on paper using the Spanish versions of the Edinburgh Handedness Inventory [
A modified computerized version of the Eriksen flanker task [
To investigate the effects of the intervention and the group differences, 2-way analyses of variance with repeated measures were performed using
We also calculated pre–post differences in each group and ran a bivariate Pearson correlation analysis only among the variables that showed significant differences in the previous analysis.
All significant results are presented with the original df, the
In all, 10 and 13 outlier values (>3 times the IQR) were excluded from the self-reported (ODI, POMS, PCS, and TSK-11) data analysis and accuracy data analysis, respectively.
This study was conducted according to the Declaration of Helsinki and approved by the research ethics committee of the Balearic Islands (IB 3186/16 PI).
As shown in
Sociodemographic, clinical, and self-reported data of participants (N=50).
Characteristics | Before the intervention | After the intervention | |||||||
|
Face-to-face group (n=27) | Self-managed group (n=23) | Face-to-face group (n=27) | Self-managed group (n=23) |
|
||||
Sex (male), n (%) | 7 (26) | 10 (43) | N/Aa | N/A | .19b | ||||
Age (years), mean (SD) | 48.63 (7.54) | 45.00 (9.13) | N/A | N/A | .13c | ||||
BMI, mean (SD) | 0.43 (0.09) | 0.41 (0.07) | N/A | N/A | .62c | ||||
WHtRd, mean (SD) | 0.55 (0.08) | 0.53 (0.06) | N/A | N/A | .43c | ||||
WHRe, mean (SD) | 1.12 (0.12) | 1.14 (0.11) | N/A | N/A | .60c | ||||
Pain duration (years), mean (SD) | 11.81 (7.47) | 8.06 (8.74) | N/A | N/A | .16c | ||||
EHIf (10-50), mean (SD) | 18.05 (5.06) | 18.22 (3.83) | N/A | N/A | .91c | ||||
Systolic BPg, mean (SD) | 112.67 (12.58) | 115.30 (14.79) | N/A | N/A | .12c | ||||
Diastolic BP, mean (SD) | 77.61 (8.62) | 76.16 (9.74) | N/A | N/A | .92c | ||||
Pain intensity (0-10), mean (SD) | 2.87 (2.27) | 3.57 (2.50) | 2.67 (2.36) | 3.83 (2.20) | .93 | ||||
ODIh (0-100, %), mean (SD) | 6.15 (5.35) | 6.01 (3.92) | 7.85 (6.22) | 7.15 (5.66) | .01i | ||||
|
|||||||||
|
Tension or anxiety (0-36) | 9.96 (7.47) | 8.83 (5.94) | 7.46 (3.67) | 8.00 (5.89) | .06 | |||
|
Anger or hostility (0-48) | 11.46 (8.48) | 9.11 (7.32) | 7.96 (4.66) | 8.78 (5.29) | .10 | |||
|
Vigor or activity (0-32) | 15.87 (4.66) | 14.28 (5.13) | 16.04 (4.75) | 16.33 (4.52) | .09 | |||
|
Fatigue or inertia (0-28) | 9.38 (7.93) | 10.15 (7.14) | 8.03 (5.91) | 8.94 (5.71) | .12 | |||
|
Depression or dejection (0-60) | 9.26 (11.10) | 6.22 (6.65) | 5.12 (5.95) | 5.17 (5.68) | .01i | |||
|
Confusion or bewilderment (0-28) | 6.28 (5.18) | 5.00 (4.51) | 5.04 (4.39) | 5.06 (3.57) | .23 | |||
|
|||||||||
|
State (0-30) | 15.51 (8.76) | 13.75 (7.29) | 15.49 (8.07) | 14.96 (10.12) | .53 | |||
|
Trait (0-30) | 19.68 (8.02) | 20.15 (8.83) | N/A | N/A | .84c | |||
|
23.14 (3.61) | 22.68 (3.59) | 20.41 (3.45) | 21.32 (3.50) | .002i | ||||
|
Activity avoidance (7-28) | 15.09 (2.27) | 13.89 (2.31) | 12.67 (2.21) | 13.05 (2.39) | <.001i | |||
|
Harm (4-16) | 8.06 (1.69) | 8.79 (1.84) | 7.74 (1.68) | 8.26 (1.52) | .21 | |||
|
12.21 (8.96) | 15.35 (8.53) | 12.29 (10.33) | 11.05 (6.74) | .19 | ||||
|
Rumination (0-16) | 3.75 (3.77) | 4.57 (3.20) | 4.29 (3.69) | 4.05 (3.39) | .98 | |||
|
Helplessness (0-24) | 5.83 (4.36) | 6.92 (4.14) | 4.92 (4.60) | 4.55 (3.10) | .02i | |||
|
Magnification (0-18) | 2.62 (1.64) | 3.55 (2.33) | 3.08 (2.60) | 2.45 (1.64) | .39 | |||
|
34.75 (24.08) | 35.25 (21.33) | 31.96 (19.72) | 26.25 (14.54) | .03i | ||||
|
Avoidance of physical activity (0-24) | 12.04 (4.93) | 12.10 (6.36) | 11.71 (6.52) | 6.85 (4.44) | .01i | |||
|
Avoidance of work (0-42) | 15.79 (12.25) | 16.90 (11.72) | 13.96 (10.77) | 13.65 (9.25) | .10 |
aN/A: not applicable.
bChi-square test.
cBoth groups were comparable in terms of gender, age, anthropometrics (BMI, waist-to-height ratio, and waist-to-hip ratio), systolic and diastolic blood pressure, pain duration, handedness, and anxiety trait.
dWHtR: waist-to-height ratio.
eWHR: waist-to-hip ratio.
fEHI: Edinburgh Handedness Inventory.
gBP: blood pressure.
hODI: Oswestry Disability Index.
iBoth groups showed decreased depression, kinesiophobia (and activity avoidance), helplessness, and fear-avoidance beliefs (and avoidance of physical activity), as well as increased disability after the intervention. No significant differences between the groups were found in any of these data.
jPOMS: Profile of Mood States.
kSTAI: State–Trait Anxiety Inventory.
lTSK-11: Tampa Scale for Kinesiophobia.
mPCS: Pain Catastrophizing Scale.
nFABQ: Fear-Avoidance Beliefs Questionnaire.
Regarding the frequency power density of the EEG resting-state data analysis, no differences between the groups were found at delta, theta, alpha, beta-1, beta-2, or beta-3 (results not shown). We only found main effect of
Differences between before and after the intervention on statistical maps of source analyses in all participants are displayed in
Regarding ECG data, no differences between the groups or sessions were found in HR, SDNN, or RMSSD. No differences between the groups or sessions were found in VLF, LF, or HF (
Summary of significant resultsa from whole-brain standardized low-resolution electromagnetic tomography analysis comparisons between before the intervention and after the intervention for delta, alpha, beta-2, and beta-3 frequency bands in all participants.
Lobe and region | BAb | Xc | Yc | Zc | ||
|
||||||
|
|
|||||
|
|
Cuneus | 30 | –5 | –70 | 5 |
|
|
Cuneus | 18 | 0 | –75 | 10 |
|
|
Middle occipital gyrus | 18 | 25 | –90 | 15 |
|
|
|||||
|
|
Precuneus | 7 | 0 | –60 | 55 |
|
||||||
|
|
|||||
|
|
Postcentral gyrus | 2 | –25 | –40 | 70 |
|
|
Postcentral gyrus | 3 | –20 | –40 | 70 |
|
|
Postcentral gyrus | 5 | –25 | –45 | 70 |
|
|
Postcentral gyrus | 7 | 5 | –65 | 65 |
|
||||||
|
|
|||||
|
|
Anterior cingulate | 32 | 0 | 35 | 20 |
|
|
|||||
|
|
Medial frontal gyrus | 10 | –5 | 50 | 15 |
|
|
Medial frontal gyrus | 9 | 5 | 50 | 20 |
|
||||||
|
|
|||||
|
|
Anterior cingulate | 32 | 0 | 20 | 35 |
|
|
Anterior cingulate | 24 | 0 | 30 | 25 |
|
|
|||||
|
|
Medial frontal gyrus | 8 | 0 | 20 | 50 |
|
|
Medial frontal gyrus | 6 | –5 | 15 | 50 |
aSignificant (
bBA: Brodmann area.
cMontreal Neurological Institute and Hospital coordinates.
Standardized low-resolution electromagnetic tomography analysis (sLORETA) results for 3 orthogonal brain slices (horizontal, sagittal, and coronal) of delta, alpha, beta-2, and beta-3 frequency bands in all participants. Yellow-red voxels represent increased (
The CBPT revealed no differences between the groups or sessions in the EEG response to the congruent or incongruent conditions (results not shown).
No significant differences between the groups or sessions were found in PPTs either in pressure pain intensity ratings at the spinal erector muscle or the forefinger (
Both groups showed decreased depression, kinesiophobia (and activity avoidance), helplessness, and fear-avoidance beliefs (and physical activity avoidance), as well as increased disability after the intervention (
The Eriksen flanker task involved a low level of difficulty (mean overall hit rate 98.93%, SD 0.14%), and no significant main effects in the percentage of hits among the groups, sessions, conditions, or interaction effects were found (results not shown). No significant main effects in RTs among the groups, sessions, or interaction effects were found (results not shown). We only found an expected significant main effect of
We only computed a bivariate Pearson correlation analysis of the pre–post differences in psychological outcomes (ODI, POMS depression and dejection scale, TSK-11 total, TSK-11 activity avoidance scale, PCS helplessness scale, FABQ total, and FABQ avoidance of physical activity scale) and EEG resting-state data (delta at the cuneus [BA30], alpha at the postcentral gyrus [BA2], and beta-2 and beta-3 at the ACC [BA32]) in all participants. After applying multiple comparison corrections, no significant correlations were found among these variables.
Both groups showed an increase in beta-2 and beta-3 in EEG resting-state data after the intervention. Source localization data analysis also showed a significant higher current density of beta-2 and beta-3 mainly located at the ACC after the intervention, as well as a higher current density of alpha mainly located at the postcentral gyrus and a significant lower current density of delta frequency located at the cuneus and precuneus. Several studies demonstrate that alpha and beta oscillations are related to feedback (top-down) brain signaling or contextual (ie, cognitive, emotional, or motivational) processing of pain [
Nevertheless, no significant differences between the groups or sessions in HRV resting-state measures were found. Previous research stated that self-reported pain and RMSSD were inversely associated in healthy individuals but not in chronic pain, concluding that this vagal tone measure is disturbed [
In addition, the self-managed intervention was as effective as the face-to-face intervention in improving depression, kinesiophobia (plus activity avoidance), helplessness, and fear-avoidance beliefs (plus physical activity avoidance). However, both modalities failed to reduce pain and disability and increase PPTs. In this regard, a previous study found improvements not only in health-related quality of life (mental and physical well-being), kinesiophobia, and hypervigilance, but also in pain sensitivity and disability in patients with CLBP after a 12-week intervention combining pain neuroscience education and cognition-targeted motor control training [
Regarding the modality of the intervention, a recent meta-analysis concluded that mHealth-based self-managed programs revealed better immediate effects on pain and disability than web-health–based programs, with better immediate effects on pain but not on disability for programs with durations of ≤8 weeks [
Furthermore, no significant differences between the groups or sessions in performance in terms of EEG activity during the Eriksen flanker task were found. Regarding cognitive performance, we found expected slower RTs in the incongruent trials, which confirmed the validity of this task to measure interference control, with greater cognitive resources needed to process stimuli in the incongruent condition. As the mean overall hit rate of this task was 98.93% (SD 0.14%), perhaps it was too effortless and not sensitive enough to observe changes produced by our educational and exercise-based intervention. Although current evidence backs cognitive improvements after aerobic exercise, we focused on a nonaerobic exercise–based intervention (including muscle strength exercises, motor control, relaxation, flexibility, and self-massage) to add novel evidence. In this regard, a previous study showed that a single session of aerobic exercise had no effect either on RTs or on brain activation in the Eriksen flanker task, but an explorative analysis revealed that RTs improved in both conditions after high-intensity exercise [
Although the results are novel and interesting, there are several limitations in the design of this study that should be considered. The main limitation was not having a passive control group to compare both interventions. As mentioned previously, clinical practice guidelines recommend accommodating patient preferences in the design of such interventions. Therefore, to promote treatment adherence, participants were not randomly distributed; as a result, a risk of selection bias must be assumed. Because of the nature of the intervention, blinding of both researchers and participants was practically unattainable; however, this is also a bias that could compromise the internal validity of the study. Because of the exclusion of the data of 15% (9/59) of the participants, the study did not reach the planned sample size to achieve an adequate statistical power. Finally, we did not control for the use of caffeine before data collection and we did not restrict the use of medications, but there were no differences between the groups (
Both intervention modalities (face-to-face group and self-managed group with the BackFit app) were equally effective at increasing beta activity at rest and located at the ACC, as well as at improving psychological functioning among patients with nonspecific CLBP. However, these results should be interpreted carefully because of the aforementioned limitations, which could compromise both internal and external validity of our study. The baseline pain and disability scores of our participants were clearly lower than those reported in previous studies; thus, they cannot be a representative sample of the population being studied. These limitations notwithstanding, to the best of our knowledge, this is the first study reporting brain changes in patients with CLBP after an mHealth intervention. Double-blinded randomized controlled studies with larger sample sizes are needed to increase the evidence for the efficacy of mHealth interventions in clinical practice for CLBP care. Furthermore, there is still conflicting evidence regarding the most adequate parameters for exercise prescription in chronic pain management, which must be considered in the design of novel exercise-based programs.
Nonsignificant results regarding electrocardiography data and pain sensitivity.
Percentage of causes of onset of pain, diagnosis, and surgery or invasive treatment received, as well as medication use.
anterior cingulate cortex
Brodmann area
cluster-based permutation test
chronic low back pain
Consolidated Standards of Reporting Trials of Electronic and Mobile Health Applications and Online Telehealth
electrocardiogram
electroencephalography
Fear-Avoidance Beliefs Questionnaire
high frequency
heart rate
heart rate variability
low back pain
low frequency
mobile health
Oswestry Disability Index
Pain Catastrophizing Scale
Profile of Mood States
pressure pain threshold
root mean square of the successive differences
reaction time
SD of the normal-to-normal (R-R) intervals
Tampa Scale for Kinesiophobia
very low frequency
This research was funded by grants from the Spanish Ministry of Science and Innovation (PSI2015-66295-R, PSI2017-88388-C4-1-R–AEI/FEDER, and PID2019-110096GB-I00–AEI/10.13039/501100011033) and the Office for Transfer of Research Results of the University of the Balearic Islands. The authors acknowledge Dr Francisco García-Ginard and Dr José A Morales from Sant Joan de Déu Hospital (Palma, Balearic Islands) for their support during patient recruitment. The authors also acknowledge Pauline Bécognée, Maria Morro, Alicia Florit, and Miguel Ruiz de la Torre for their assistance during data acquisition.
All the authors have read and approved the article and the procedures. CS, NGD, JSF, OVR, and JCP discussed the original design of the experiment and acquired the data. CS and JLT analyzed the data and drafted the original manuscript. All the authors revised the manuscript.
None declared.