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 Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on https://formative.jmir.org, as well as this copyright and license information must be included.
People with autism spectrum disorder (ASD) frequently experience high levels of anxiety. Despite this, many clinical settings do not provide specialist ASD mental health services, and demand for professional support frequently outstrips supply. Across many sectors of health, investigators have explored digital health solutions to mitigate demand and extend the reach of professional practice beyond traditional clinical settings.
This critical appraisal and pilot feasibility study examines
We assessed the feasibility of using home-based HRV biofeedback for self-management of anxiety in young people with ASD. We adopted coproduction, involving people with ASD, to facilitate development of the study design. Next, a separate pilot with 20 participants with ASD (n=16, 80% male participants and n=4, 20% female participants, aged 13-24 years; IQ>70) assessed adoption and acceptability of HRV biofeedback devices for home use over a 12-week period. Data were collected from both carers and participants through questionnaires and interviews; participants also provided single-lead electrocardiogram recordings as well as daily reports through smartphone on adoption and use of their device.
Pre-post participant questionnaires indicated a significant reduction in anxiety in children (
HRV biofeedback devices have shown promise in this pilot study. There is now a need for larger evaluation of biofeedback to determine which delivery methods achieve the greatest effect for people with ASD.
ClinicalTrials.gov NCT04955093; https://clinicaltrials.gov/ct2/show/NCT04955093
Autism is a lifelong developmental disability that affects how people communicate and interact with the world [
Approximately 1% of the British population are likely to have some form of ASD, with studies reflecting this prevalence both in children [
People with ASD can experience a range of mental health difficulties [
There is evidence that digital health interventions can aid compliance with traditional treatments and help to reduce increasing demands on health provision, while extending the reach of professional practice beyond traditional clinical settings [
Digital solutions can be tailored to the needs of people with ASD and can be associated with less stress than that reported in face-to-face interventions [
A variety of biofeedback equipment now exists for both psychophysiological stress profile assessment [
Although biofeedback has been used since the 1970s, there is now growing interest in its use in stress management [
To date, the most widespread use of biofeedback in individuals with ASD has involved the use of neurofeedback [
HRV was first recognized 60 years ago as an indicator of fetal distress [
A number of theoretical models have been proposed describing the links between HRV and health, mediated through connections between the heart and the brain [
HRV has been used in conjunction with sensor technology to develop
Portable biofeedback devices typically derive HRV using infrared photoplethysmography (PPG), which measures blood flow, usually through either the fingertip or the earlobe. Peripheral blood flow can be used to assess heart rate and to estimate HRV [
There are wide variations in study design, using different devices, training protocols, and outcome measurements. Nevertheless, systematic reviews and meta-analyses have concluded that HRV biofeedback can be an effective treatment for symptoms in a range of different populations, including both adults [
For people with ASD, biofeedback may provide specific advantages for the management of anxiety. First, it provides a technique for developing control over specific symptoms without the need for verbally based techniques designed for non-ASD populations or behavioral interventions that may be anxiety provoking for people with ASD [
Finally, people with ASD show a range of different physiological reactions compared with non-ASD peers [
Accordingly, we proceeded with an exploratory study to investigate the use of HRV biofeedback outside of clinical settings as a potential intervention to help people with ASD by reducing symptoms of anxiety.
The aim of this pilot and feasibility trial comprised two main objectives: first, to explore the use of HRV biofeedback as a suitable methodology to support people with ASD to manage anxiety outside of clinical settings and, second, to assess the risks, benefits, and challenges of using HRV biofeedback within this population.
We involved adults with ASD and professionals working in the field in the initial study development and design. We then recruited a separate group of young people with ASD in an experimental design with appropriate pre- and postintervention outcome measures.
We recorded demographic data and mental health status and used participant anxiety and depression as the primary outcome measures. Statistical analyses were carried out to assess mean group differences in reported anxiety and depression. A per-protocol analysis was used. This led to several participants being excluded after randomization because they met the exclusion criteria, and their pre-post data were therefore not included in the quantitative data analysis. The pre-post data sets were analyzed using a standard statistical package (SPSS software [version 24.0; IBM Corp]). Correlational analyses were also used to review associations between baseline measurements and HRV data.
To assess the risks, benefits, and challenges we used several methods of data collection, including daily monitoring of device use, perceived participant stress levels using a questionnaire delivered through smartphone, standardized interviews, and short debriefing reports. Participants who dropped out early or had continuing difficulties using their biofeedback device or had electrocardiogram (ECG) recording difficulties after randomization had their monitoring data included for further analysis, provided that they had consented for these data to be collected. The aim of the study was to ensure methodological robustness and feasibility, with a focus upon potential risks, problems, or difficulties as well as potential benefits of using HRV biofeedback.
The sample was drawn from a population of patients with an existing diagnosis of ASD who had attended regional health services for help with anxiety. The young people with ASD were invited to participate in experimental adoption of a portable HRV biofeedback device over a 12-week period. All participants in this study had been diagnosed with ASD in specialist health service assessments clinics using standardized measures. None had a learning disability, and all had attended mainstream education. Additional preassessment screening of participants was carried out using the
Ethics approval was granted by the regional National Health Services ethics committee (15/NI/0255; IRAS: 139122). After review by the regional National Health Services ethics committee, approval was granted under the UK governance arrangements for research ethics committees.
Potential participants were recruited using two methods: either through direct contact with their therapist or through letters sent to patients who were already discharged. It was not possible to determine the exact reasons for nonparticipation in the study because of the ethical constraints regarding contacting those who declined to participate through their therapist or those who did not respond to the invitation letters. Initially, 20 people took part: 16 (80%) male participants and 4 (20%) female participants. Their ages ranged from 13 to 22 (mean 16.2, SD 2.63) years. Detailed demographic information is presented in
CONSORT (Consolidated Standards of Reporting Trials) flow diagram summarizing the processes adopted for screening and enrollment, allocation into groups, follow-up, and final analysis. Participant numbers are provided for each stage.
To decrease the risk of group allocation bias, random assignment to treatment group or control group was carried out
Allocation was made in blocks to ensure that adequate numbers of participants were allocated to each condition. A randomized number sequence was generated through computer.
Between-group comparisons were planned to compare 6 weeks of intervention in the immediate group with 6 weeks of no intervention in the delayed group. The small sample size and additional exclusions from the study prevented these comparisons from being carried out.
HRV home trainer devices differ in terms of form, feedback mechanisms, data storage, training guidelines, and underlying software, all of which may affect user experience and effectiveness. For this study, biofeedback devices were selected based on evidence of their use in previous research. To explore whether these aspects were of any concern, participants were allocated to 1 of 2 different biofeedback devices (
Group A participants were provided with a home trainer biofeedback device that used PPG ear sensors [
We used 2 personal home trainer devices to provide biofeedback during pilot testing: StressEraser, left, and Inner Balance, right (Inner Balance image reproduced with permission of HeartMath).
Participants were invited to complete anxiety and depression questionnaires before and after the intervention. The measures included
During the intervention period, participants were asked to complete short daily reports on stress levels and device use. This enabled us to track participant stress levels and monitor perceived usefulness of the device. The progress report was devised based on initial phase 1 evaluations and served a 2-fold purpose: tracking participant stress levels and monitoring use of the device over the course of the intervention. This short report asked questions on sources of stress, levels of stress, and use of the biofeedback device.
At the end of the intervention, participants and carers independently completed a short debriefing interview, and participants completed equipment usability ratings using the
All participants were offered initial assessment and training in their own home, with their carer present. This was carried out by a chartered clinical psychologist with certified training in HRV biofeedback (HC). The training involved sending video clips and written instructions produced by the device manufacturers to participants before an agreed home visit in which a direct demonstration of device use was given. Participants were then all seen and assessed by the first author (HC) who offered 2 training sessions (lasting for 30 minutes each) of home instruction in use of the allocated biofeedback device. No other intervention regarding anxiety management was provided. The HRV biofeedback intervention was available to use daily for a period of 12 weeks. Self-tailoring of intervention intensity was allowed to gauge acceptability and record self-reported compliance. Use was encouraged by the first author (HC) at the outset and indirectly through monitoring uptake. Information on participant anxiety and depression was collected through questionnaire measures completed before and after the intervention.
Participant HRV was measured in the home through a psychophysiological stress profile assessment using a single-lead ECG recorder before and after the intervention. Once the intervention commenced, remote monitoring continued daily, using SMS text message prompting at preset intervals agreed with participants. Questions used to monitor use of the intervention were sent to participants at an agreed time in the evening, recording whether they had used the device that day and whether it had been perceived as useful. Finally, additional information on risks, perceived problems, and benefits of the intervention and equipment usability was collected in face-to-face interviews with participants and their carers at the end of the intervention. This information was also collected from those who dropped out and those whose pre-post data were excluded, provided they consented to provide these data. This was seen as crucial to capture information on any potential difficulties and ensure a representative assessment of risks and benefits. Further details on the intervention are presented in the CONSORT (Consolidated Standards of Reporting Trials) and Template for Intervention Description and Replication checklists (
All participants were nonsmokers, and none reported taking illegal drugs. Of the 20 participants, 10 (50%) reported prescribed medication. Of these 10 participants, 6 (60%) were prescribed selective serotonin reuptake inhibitor antidepressants, and 4 (40%) took stimulant medication. Sleep disturbance was reported by 75% (15/20) of the participants; 15% (3/20) reported needing to carry auto injectors with adrenaline (EpiPen). Of the 20 participants, 16 (80%) were employed or were students.
Carers were asked standardized questions regarding their main concerns about participant behavior, including whether there were any triggers for participant anxiety attacks or
Of the 20 participants, 4 (20%) were excluded after randomization because of the identification of cardiac concerns (n=3, 75%) and significant mental health concerns (n=1, 25%), whereas 1 (5%) dropped out and declined further questionnaire and physiological assessment, leaving 15 participants (n=11, 73%, male participants and n=4, 27%, female participants) for before-and-after data analysis (
Statistical analyses were carried out using SPSS software (version 24.0) to assess mean group differences in reported anxiety and depression. Parametric statistical tests were conducted using paired sample 2-tailed
Of the 15 participants, 1 (7%) adult participant did not complete the initial depression questionnaires: 14 sets of depression data were analyzed. Parametric statistical tests were also conducted using paired sample 2-tailed
Data collected from children and adults showed statistically significant reductions in mean score for anxiety after the intervention; the results for both adults and children showed no significant reduction in mean scores for depression (
Participant questionnaire data showing mean scores for anxiety and depression before and after using heart rate variability biofeedback (N=15).
Participants (aged 13-24 years) | Before the intervention, mean (SD) | After the intervention, mean (SD) | Mean difference (SD) | Coefficient, |
Cohen |
||||
|
|||||||||
|
BYI-Aa (children: n=7) | 24.43 (8.98) | 14.43 (10.97) | 10.00 (10.39) | 0.472 | 2.55 (6) | .04 | 0.99 | |
|
BAIb (adults: n=8) | 21.12 (11.2) | 15.00 (11.49) | 6.12 (4.39) | 0.925 | 3.95 (7) | .006 | 0.54 | |
|
|||||||||
|
BYI-Dc (children: n=7) | 20.43 (14.04) | 13.71 (14.16) | 6.71 (9.53) | 0.077 | 1.86 (6) | .11 | 0.48 | |
|
BDId (adultse: n=7) | 17.0 (12.74) | 13.86 (10.91) | 3.14 (4.85) | 0.928 | 1.72 (6) | .14 | 0.25 |
aBYI-A: Beck Youth Inventory, Anxiety scale.
bBAI: Beck Anxiety Inventory.
cBYI-D: Beck Youth Inventory, Depression scale.
dBDI: Beck Depression Inventory.
eAn adult participant did not complete the depression questionnaire.
Carers were asked to rate the frequency of participant behavioral outbursts or meltdowns at the initial and debriefing interviews. Carer ratings indicated a significant reduction (Wilcoxon signed-rank test) in the frequency of behavioral outbursts comparing initial interview data with debriefing interview data (
Attempts were made to record wireless ECG data to detect changes in HRV over time using a psychophysiological stress test [
Pearson bivariate correlation statistics before and after the intervention between physiological measures of heart rate and heart rate variability recorded with participant age and level of ASD (autism spectrum disorder) symptoms as measured by the Social Communication Questionnaire [
|
Age | ASD | HF-HRVa | RMSSDb | SDNNc | HRd | ||||||
|
||||||||||||
|
Before the intervention | 1 | –0.250 | –0.580e | –0.596e | –0.548e | –0.001 | |||||
|
After the intervention | —f | — | –0.389 | –0.381 | –0.522e | –0.062 | |||||
|
||||||||||||
|
Before the intervention | — | 1 | –0.221 | –0.134 | –0.184 | 0.541e | |||||
|
After the intervention | — | — | –0.557e | –0.442 | –0.504e | 0.743h | |||||
|
||||||||||||
|
Before the intervention | — | — | 1 | 0.929h | 0.902h | –0.562e | |||||
|
After the intervention | — | — | — | 0.892h | 0.905h | –0.733h | |||||
|
||||||||||||
|
Before the intervention | — | — | — | 1 | 0.946h | –0.514e | |||||
|
After the intervention | — | — | — | — | 0.910h | –0.616e | |||||
|
||||||||||||
|
Before the intervention | — | — | — | — | 1 | –0.465 | |||||
|
After the intervention | — | — | — | — | — | –0.694h | |||||
|
||||||||||||
|
Before the intervention | — | — | — | — | — | 1 | |||||
|
After the intervention | — | — | — | — | — | — |
aHF-HRV: high-frequency heart rate variability (a frequency domain index that can indicate parasympathetic nervous system activity).
bRMSSD: root mean square of successive differences in normal heartbeat contractions or interbeat intervals measured in milliseconds (a time domain heart rate variability index that can be associated with parasympathetic nervous system activity).
cSDNN: SD of the normal heartbeat contractions; that is, normal-to-normal interbeat intervals measured in milliseconds.
dHR: heart rate measured in beats per minute.
e
fNot applicable.
gAutism spectrum disorder symptomatology measured using the Social Communication Questionnaire [
h
Across the study, 474 web-based surveys were collected providing data on sources of stress and use of biofeedback devices throughout the intervention period. Participants who had difficulties using their biofeedback device or ECG recorder had their monitoring data included, provided they had consented.
Participants were asked during the intervention period whether the device had helped them when used. The majority (311/474, 65.6%) of the reports provided indicated that the biofeedback device had been used and that “it helped” when used. Only 4% (19/474) of the reports indicated that the device “didn’t help” when used. Regarding use, 88% (417/474) of the reports indicated that the device had been used for between 0 and 10 minutes, with 12% (57/474) of the reports indicating that the device had been used for >10 minutes.
Participants were also asked to report the reasons for not using the device. A summary of 144 reports provided by participants detailing a range of reasons is presented in
Participant reported effect of biofeedback device when used (n=144).
Reported reason for not using device | Responses, n (%) |
“Not stressed” today | 65 (45.1) |
No clear reason given | 24 (16.7) |
“Busy” or “away” | 20 (13.9) |
“Device not with me” | 17 (11.8) |
“Forgot to use” | 9 (6.2) |
“Practicing later” | 6 (4.2) |
“Error on device” | 3 (2.1) |
The most frequent reason reported for not using the device was simply that the participant did not feel stressed. Other reasons reported were “too busy” (20/144, 13.9%), “device not with me” (17/144, 11.8%), “forgot to use it” (9/144, 6.3%), “practicing the device later” (6/144, 4.2%), and “error on device” (3/144, 2.1%). Of all reports submitted, 16.7% (24/144) offered no clear reason for not using the device.
Usability ratings were captured on the 2 biofeedback devices, an ECG recorder, and a web-based survey (
Participant ratings of 2 types of biofeedback devices, an electrocardiogram (ECG) recorder, and SMS text monitoring using the System Usability Scale (SUS) [
Type of equipment useda | SUS score, mean (SD; range) | Benchmark SUS scoreb, mean (SD) |
StressEraser (n=9) | 76.60 (16.25; 47.5-92.5) | 68 (12.5) |
Inner Balance (n=9) | 83.61 (8.94; 70.0-92.5) | 68 (12.5) |
Actiwave ECG recorder (n=9) | 70.00 (7.80; 60.0-77.5) | 68 (12.5) |
SMS text message survey (n=15) | 78.50 (9.05; 65.0-100.0) | 68 (12.5) |
aA total of 18 ratings of biofeedback devices were completed: 9 for StressEraser and 9 for Inner Balance; 9 ratings were completed regarding the electrocardiogram recorder; and 15 ratings were completed regarding the SMS text message report.
bBenchmark calculation of average SUS scores [
During the debriefing at the end of the study, participants and carers reported perceived benefits and any problems. Unexpectedly, many of the participants reported cold fingers during initial training, with 70% (7/10) of those initially allocated to the StressEraser reporting difficulty using this device at some point during the study.
Clinical disclosures included previously unrecognized cardiac irregularities (3/20, 15%) and severe mental health difficulties (1/20, 5%). These were identified
A range of positive benefits were reported by both participants and their carers, with the most frequent benefit reported being feeling
Participant reported problems of using biofeedback device (n=24).
Reported problems | Responses, n (%) |
Finger sensor errors | 5 (21) |
Difficulty using while stressed | 5 (21) |
Lack of practice | 4 (16) |
Device functions difficult | 3 (13) |
Needed reminders | 3 (13) |
Didn’t find it helped | 2 (8) |
Ear sensor difficulty | 2 (8) |
Participant reported benefits of using biofeedback device (n=43).
Reported benefits | Responses, n (%) |
Felt calm or helped | 11 (25) |
Ease of use | 9 (21) |
Helped breathing | 5 (11) |
Helped sleep | 4 (9) |
Good video tutorials provided | 3 (7) |
Visual or can “see” results | 3 (7) |
Helped when not using | 2 (5) |
Helped focus | 2 (5) |
Portable | 2 (5) |
Efficient | 2 (5) |
This pilot study highlights potential positive effects of HRV biofeedback in people with ASD, with fewer symptoms of anxiety being reported after using HRV biofeedback devices at home. Carers also reported fewer behavioral
A perceived strength of this study was the experimentation of a new intervention in a population where there is high need and where little research exists on issues that affect the lives of people with ASD [
We note the discovery of previously undetected cardiac irregularities and deteriorating mental health. This highlights the mental and physical health vulnerabilities in this population [
Several limitations are acknowledged in this pilot study. A necessary limitation was the small sample size and the inherent risk of error and bias when using self-report measures. It was not possible to determine the exact reasons for nonparticipation in the study because of the ethical constraints regarding contacting those who declined to participate. All participants contacted had already attended services for anxiety. It may be that those who did not participate were not now experiencing problems with anxiety or that they were concerned about participating in research into an untested intervention. Future studies should address this issue to further assess whom this treatment might be beneficial for.
It is possible that reductions in self-reported anxiety may have been related to other outside factors or to nonspecific therapeutic variables. However, it is of note that no time was spent with participants talking about their anxiety or providing any other type of intervention—the training given only involved a review of existing instruction guidelines in the use of each biofeedback device. It is also possible that unconscious bias may have occurred within carer interviews and participant reports because of a wish for the treatment to succeed; however, the debriefing reports indicated that several participants and their carers did report at the end of the intervention that the device did not help despite their initial hopes that it would be beneficial, and participant reports and carer reports were in concordance regarding the changes noted.
This study attempted to carry out before-and-after HRV assessment analysis by means of a single-lead ECG recorder using a psychophysiological stress test within participants’ homes. This type of test paradigm assessment proved unsuccessful because of the lack of standardized test conditions inherent within multiple home environments. This type of stress assessment is one which is unlikely to be useful outside of clinical settings. Before embarking on home intervention programs, future interventions using physiological monitoring such as ECG assessments may require an initial clinic appointment to carry out full mental and physical health checks [
In addition, individuals with ASD have also been found to exhibit both hyperarousal and hypoarousal responses to stress tasks, suggesting that the classic paradigm of a stress profile assessment designed for non-ASD populations is unlikely to provide a clear picture when used in people with ASD [
Usability assessment indicated that both ECG monitoring and remote smartphone monitoring of stress levels were found to be acceptable in people with ASD. Small wireless ECG monitors can now be used for 24-hour recordings, and the approach of remote stress monitoring combined with longer ECG recordings could be used in future studies to provide much needed data on anxiety and the physiological profile of people with ASD.
A review of HRV biofeedback studies [
Importantly, some participants in this pilot may not have been able to develop the specific type of resonant frequency breathing using home trainer biofeedback devices, which is argued to increase HRV [
This preliminary work has provided vital information for further studies, which could now test effectiveness of HRV biofeedback for home-based remote management of anxiety in an adequately powered randomized trial using a comparator intervention with matched intervention time. This approach was used with non-ASD populations in 2 recent studies that reported positive effects from biofeedback in comparison with control interventions, such as mindfulness and walking [
Studies investigating comparative interventions could use a breathing pacer app for smartphone use, which could assess paced breathing alone in comparison with HRV biofeedback. Future investigations should capture device data, enabling information to be gathered regarding length and type of breathing practice, which may help to address questions relating to any dose-response relationship for HRV biofeedback. Finally, this initial study only targeted young people aged 13 to 24 years with no known learning disability. A key step in further work will be to assess this methodology and intervention in other groups of people with ASD, such as older adults and people with intellectual disability.
ASD is now a common condition. Reports suggest high costs of supporting people with ASD; yet, little research has been undertaken into new types of interventions specifically designed to meet their needs [
Conditions such as ASD pose a significant cost to individuals, health care providers, and society as a whole [
Digital technology may represent a useful method of engaging people with ASD by using some of their characteristic strengths and interests, without the complex social and communication demands of traditional cognitive and behavior therapies [
The application of home-based solutions to difficulties experienced by people with ASD also represents what has been termed a “naturalistic developmental behavioral intervention” that can help with the generalization of skills because of their use in real-world interactions [
Systematic reviews have outlined developments in biofeedback across a range of modalities as well as some of the challenges to be addressed in future investigations [
Participant demographic information.
Measures.
CONSORT (Consolidated Standards of Reporting Trials) 2010 checklist of information to include when reporting a randomized trial.
The Template for Intervention Description and Replication checklist.
Study protocol.
autism spectrum disorder
Consolidated Standards of Reporting Trials
electrocardiogram
heart rate variability
photoplethysmography
System Usability Scale
HC was supported by a Health & Social Care Doctoral Fellowship (2013-2018) from the R&D Division of the Public Health Agency, Northern Ireland, United Kingdom. The authors wish to offer thanks to Professor Richard Gevirtz, Alliant University, for his help and guidance; Assistant Professor Keri Heilman, University of North Carolina, for her advice on assessment; and Dr Cathal Breen, Ulster University, for his help with electrocardiogram recording. This study would not have been possible without the direct and forthright contributions of people with autism spectrum disorder and the support of South Eastern Health and Social Care Trust.
HC was responsible for the conception of this study. HC, MD, and WGK were responsible for the study design. HC and MD drafted the article, which was critically revised by all the other authors. HC was responsible for data collection and for clinical evaluations. HC, MD, and JM were responsible for data analysis.
None declared.