Exercise interventions in migraine patients: a YouTube content analysis study based on grades of recommendation

Introduction

Migraine is a common health problem, with an estimated global age-standardized prevalence of 14.4%, the greatest impact of which is felt by women aged 15–49 years (Stovner et al., 2018). Migraine is defined as a recurrent headache disorder manifesting in attacks lasting 4–72 h, with characteristics of unilateral location, pulsating quality, moderate to severe intensity, aggravation by routine physical activity and association with nausea, photophobia and phonophobia (Olesen, 2018).

Migraine is the second leading cause of disability in the world, just after low back pain, with a major impact on work, family and society (Vos et al., 2017; Buse et al., 2019). Most patients perceive its negative impact on personal relationships, parenthood, careers, quality of life and health (Buse et al., 2019). Migraine’s impact on quality of life includes reduced physical activity, thereby resulting in high levels of sedentary behavior (Farris et al., 2018a). The disability generated by migraine incurs a high economic cost in developed countries, derived from healthcare evaluations, work absenteeism and low work efficiency (Berg & Stovner, 2005; Vos et al., 2017).

The primary treatment modality for migraine is pharmacological intervention based on the strongly recommended triptans and non-steroidal anti-inflammatory drugs for acute migraine management. For patients with chronic migraine, other pharmacological interventions, such as beta-blockers, valproic acid and topiramate, are recommended for migraine prophylaxis (Evers et al., 2009; Cameron et al., 2015; Ashina, 2020). Recently, the increased use of erenumab (a fully human monoclonal antibody) and botulinum toxin have had good results (Lattanzi et al., 2019; Herd et al., 2019; Ashina, 2020). However, none of these pharmacological interventions are free from adverse events (Evers et al., 2009; Lattanzi et al., 2019; Ashina, 2020).

There has therefore been increased interest in recent years in studying non-pharmacological interventions such as exercise to decrease the intake of medication and enhance the effects of both interventions in migraine symptoms. Aerobic exercise can reduce the intensity, frequency and duration of migraine episodes and improve the quality of life of patients with migraine (La Touche et al., 2020). Unlike pharmacological interventions, aerobic exercise does not provoke adverse events, although there is a widespread belief among patients with migraine that exercise can trigger a migraine episode, a belief associated with intentional avoidance of physical activity (Farris et al., 2018a).

Online health information can influence how patients deal with their health problems, and there is an increasing use of the Internet and social media to obtain health information. A study by AlMuammar et al. (2021) showed that 92.6% of the participants used the Internet to obtain medical information, 42% of them used this information because they did not want to go to the hospital, and YouTube was the second most frequently used social media for all searches. YouTube is one of the most ubiquitous forms of social media, with abundant information on various health topics. However, most of the sources that provide this information are not from healthcare professionals, influencing the videos’ quality. For example, only 9% of the authors of migraine videos on YouTube are healthcare professionals, and 44% of the non-healthcare practitioner videos showed complementary and alternative medicine as valid treatments for migraine (Saffi et al., 2020). The same could be true for videos on exercise interventions for migraine, which might not achieve sufficient quality and might hinder patients from exercising or could even generate adverse effects. With all this information, the main objective of the present study was to review and grade the current evidence regarding exercise in migraine. The second objective was to perform a content analysis of YouTube videos on exercise for migraine. The third objective was to evaluate the videos included in this study based on the grades of recommendation for exercise interventions for migraine.

Materials and Methods

Results

Selected videos from YouTube

Initially, 367 videos were selected for possible inclusion in this study. After going through the eligibility process (Fig. 1), 129 videos were ultimately included for the qualitative analysis. The kappa coefficient indicating agreement between the examiners in the evaluation of the GQS and DISCERN scales was high for both (κ = 0.831 and κ = 0.735, respectively).

The frequency data of the videos by GQS and DISCERN scores, author and subject are shown in Table 4. The descriptive statistics obtained from the videos are presented in Table 5, with their median and quartiles due to the data’s non-normal distribution.

Table 4:

Distribution of videos by GQS and DISCERN scores, author and subject categories.

		No.	%
GQS	Poor quality	92	71.3
	Generally poor quality	19	14.7
	Moderate quality	13	10.1
	Good quality	4	3.1
	Excellent quality	1	0.8
	Total	129	100
DISCERN	Very poor	100	77.5
	Poor	15	11.6
	Average	9	7
	High	5	3.9
	Very high	0	0
	Total	129	100
Author	Healthcare professional	38	29.5
	Exercise professional	15	11.6
	Alternative therapist	30	23.3
	Non-healthcare	43	33.3
	Patient	3	2.3
	Total	129	100
Subject	Aerobic exercise	24	18.6
	Strength exercise	6	4.7
	Stretching/Mobility	19	14.7
	Yoga/Tai Chi	26	20.2
	Meditation/Breathing/Relaxation	14	10.9
	Alternative therapies	9	7
	Self-massage/Posture	7	5.4
	Lifestyle changes regarding exercise	18	14
	Vestibular rehabilitation	6	4.7
	Total	129	100

DOI: 10.7717/peerj.14150/table-4

Table 5:

Descriptive statistics of the video characteristics in median and quartiles.

	Median	Quartile 1	Quartile 3
Likes	64	4.5	816.5
Dislikes	0	0	17
Video power index	99.42%	94.88%	100%
Visits	2,679	187	27,203.5
Days online	827	454.5	1,702
View ratio	3.4914	0.2794	42.1551
Comments	8	0.75	56.75
Duration	399	211.5	786.5

DOI: 10.7717/peerj.14150/table-5

The distribution of the GQS and DISCERN scores for the two author categories (healthcare/exercise professional authors and alternative medicine/patient authors) is shown in Table 6. The chi-squared test showed a statistically significant difference between these two categories (p < 0.001), with healthcare/exercise professional authors showing a larger percentage of high scores than alternative medicine/patient authors.

Table 6:

Distribution of GQS and DISCERN scores within healtcare/exercise professional authors and alternative medicine/patient authors categories.

		Healthcare/exercise professional authors	Alternative medicine/patient authors
GQS	Poor quality	25 (47.2%)	67 (88.2%)
	Generally poor quality	12 (22.6%)	7 (9.2%)
	Moderate quality	11 (20.8%)	2 (2.6%)
	Good quality	4 (7.5%)	0 (0%)
	Excellent quality	1 (1.9%)	0 (0%)
	Total	53 (100%)	76 (100%)
DISCERN	Very poor	29 (54.7%)	71 (93.4%)
	Poor	10 (18.9%)	5 (6.6%)
	Average	9 (17%)	0 (0%)
	High	5 (9.4%)	0 (0%)
	Very high	0 (0%)	0 (0%)
	Total	53 (100%)	76 (100%)

DOI: 10.7717/peerj.14150/table-6

The Mann–Whitney U test also showed that the alternative medicine/patient authors’ videos obtained more likes, dislikes, visits, visits per day (VR) and comments than the healthcare/exercise professional videos, with statistically significant differences (p < 0.05). In contrast, there were no differences in VPI, days online or duration between groups (Table 7).

Table 7:

Differences between healthcare/exercise professionals authors and alternative medicine/patient authors in the quantitative characteristics of the videos.

		Median	Q1	Q3	P
Likes	Healthcare/exercise	21	2	138	0.002
	Alternative/patient	129.5	10.5	1,116
Dislikes	Healthcare/exercise	0	0	4	0.007
	Alternative/patient	4	0	39.75
VPI	Healthcare/exercise	100%	94.77%	100%	0.262
	Alternative/patient	98.75%	95.06%	100%
Visits	Healthcare/exercise	892	83.5	7,577.5	0.004
	Alternative/patient	8,269	239.5	48,983.5
Days online	Healthcare/exercise	978	470	1839	0.352
	Alternative/patient	717	406.75	1,544.25
VR	Healthcare/exercise	1.24	0.11	10.26	0.002
	Alternative/patient	7.46	0.53	77.71
Comments	Healthcare/exercise	2.5	0	20.5	0.008
	Alternative/patient	15.5	1.75	115.25
Duration	Healthcare/exercise	305	149.5	726	0.138
	Alternative/patient	432.5	237.5	835.25

DOI: 10.7717/peerj.14150/table-7

Note:

Healthcare/exercise, healthcare/exercise professional authors; Alternative/patient, alternative medicine/patient authors; VPI, Video Power Index; VR, view ratio.

The correlation analysis showed strong correlations between GQS and DISCERN. There were statistically significant negative weak correlations between GQS and likes, dislikes, visits, VR and comments. Moreover, there were statistically significant negative weak correlations between DISCERN and likes, dislikes, visits, VR and comments (Table 8).

Table 8:

Correlation analysis of the quality scales and the characteristics of the videos.

	GQS	DISCERN	Likes	Dislikes	VPI	Visits	Days online	VR	Comments
DISCERN	0.89**
Likes	−0.39**	−0.35**
Dislikes	−0.26**	−0.20*	0.86**
VPI	−0.04	−0.12	−0.24**	−0.59**
Visits	−0.37**	−0.33**	0.95**	0.88**	−0.36**
Days online	0.04	−0.03	0.10	0.18*	−0.22*	0.30**
VR	−0.40**	−0.32**	0.95**	0.85**	−0.30**	0.94**	−0.02
Comments	−0.35**	−0.31**	0.92**	0.80**	−0.32**	0.90**	0.96	0.91**
Duration	−0.16	−0.14	0.29**	0.19*	0.03	0.21*	−0.23**	0.29**	0.28**

DOI: 10.7717/peerj.14150/table-8

Notes:

VPI, Video Power Index; VR, view ratio.

* P < 0.05.

** P < 0.001.

Evaluation of the videos based on grades of recommendation

The evaluation of the videos regarding the grades of recommendation based on SIGN included 90 of the 129 videos. The videos excluded from the SIGN evaluation showed interventions that have been discarded from the classification into grades of recommendation due to lack of evidence.

The evaluation of the included videos showed a B grade of recommendation for those videos about general advice on aerobic exercise or moderate-intensity aerobic exercise, yoga and lifestyle/behavior changes regarding exercise. These videos accounted for approximately 71% of all videos included in the SIGN evaluation and close to 50% of the total videos included in the current study. The percentage of videos regarding their interventions and grades of recommendation are shown in Fig. 2.

Regarding the distribution of the grades of recommendation of the videos based on their authors’ category, there was no statistically significant difference between both categories (p = 0.198). This distribution is shown in Fig 3.

Discussion

The purpose of this study was to evaluate the quality of YouTube videos regarding prescribed exercise and advice for patients with migraine. We also evaluated the evidence levels and grades of recommendation for prescribed exercise for migraine, which made it possible to assess the quality of the videos based on objective and evidence-based criteria about modalities and parameters regarding exercise for migraine.

The results of this study showed that YouTube videos on exercise for migraine are generally of low quality and poor reliability. These findings contrast with the results for other YouTube videos on prescribed exercise for various health-related problems, in which most of the videos reached a high- or moderate-quality score (Kocyigit et al., 2019; Koçyiğit, Okyay & Akaltun, 2020; Culha et al., 2021; Chang & Park, 2021; Rodriguez-Rodriguez et al., 2021). However, other authors have found results similar to those of this study, with most of the videos reaching a low-quality score, although scales other than GQS and DISCERN have been used to evaluate the quality and reliability of the videos (Villafañe et al., 2018; Lee et al., 2018). The studies that found better quality scores in their included videos also found that most of the authors were healthcare professionals, which could explain the difference in scores.

Although the percentage of low-quality videos in the healthcare/exercise professional author category was higher than the percentage of high-quality videos, this category obtained more videos with higher quality scores in both GQS and DISCERN than the alternative medicine/patient author category, with statistically significant differences. These results agree with those of previous studies with higher quality scores in the videos from healthcare professionals, academics, universities and health institutions and lower quality scores in those of non-healthcare professionals, patients and independent users (Lee et al., 2018; Kocyigit et al., 2019; Koçyiğit, Okyay & Akaltun, 2020; Koçyiğit, Akyol & Şahin, 2021; Chang & Park, 2021; Rodriguez-Rodriguez et al., 2021). In the study by Culha et al. (2021), no differences were found in GQS scores between healthcare professionals/institutions and independent health information websites. However, the explanation for this outcome appears to be that most of the independent websites belonged to physical therapists, who are healthcare professionals and should therefore have been included in this category.

Regarding the video characteristics, there were also significant differences between both authors’ categories, with higher numbers of likes, dislikes, views, VR and comments in the alternative medicine/patients’ videos. Likewise, a negative correlation was found between the quality of the videos and the number of likes, dislikes, views, VR and comments. In contrast to these findings, other articles found no statistically significant differences in video characteristics regarding video sources or quality categories (Kocyigit et al., 2019; Culha et al., 2021; Koçyiğit, Akyol & Şahin, 2021; Heisinger et al., 2021; Chang & Park, 2021).

The lowest-quality migraine exercise videos appear to generate a greater impact on the population, with higher viewership and interaction rates. There were no differences between both authors’ categories in VPI, and there was no correlation between the quality scores and VPI, which could be interpreted as viewers being unable to discriminate between low- and high-quality information regarding exercise for migraine.

The high prevalence of headaches, including migraines (Vos et al., 2017), and the lack of awareness regarding prescribed exercise and its usefulness in this disorder facilitate the spread of misinformation in media resources such as YouTube, without being correctly evaluated by the users. YouTube has no filter or peer-review process for screening videos based on their source, quality or evidence (Delli et al., 2016). This context makes it possible to generate and spread misconceptions about exercise for migraine, as observed in this article, which could result in patients performing interventions that could fail to improve their condition or even worsen it.

There were no differences between the two categories regarding time the videos have been online and their duration. These characteristics could therefore not affect the differences in the other features.

In terms of grades of recommendation, aerobic exercise obtained a B grade for migraine treatment. Moderate-intensity aerobic exercise appears to be the most studied modality of prescribed aerobic exercise for migraine and also obtained a B grade, followed by high-intensity aerobic exercise and low-to-moderate continuous aerobic exercise, both obtaining a C grade. Aerobic exercise can improve migraine in various ways: it can activate anti-inflammatory systems that improve the activity of neuropeptides and other inflammatory mediators involved in triggering the trigeminal neurovascular system (Hasbak et al., 2002; Bigal et al., 2007; Beavers et al., 2010); aerobic exercise can activate central mechanisms of pain modulation, such as the descendent pain inhibitory system and endocannabinoid system (Raichlen et al., 2012; Rice et al., 2019); and aerobic exercise can alter psychological and social dimensions, improving outcome expectancies, self-efficacy and self-management strategies and reducing anxiety and depression symptoms (Salmon, 2001; Trivedi et al., 2011; Bromberg et al., 2012; Amin et al., 2018).

Yoga also obtained a B grade. The mechanisms by which yoga ameliorates migraine symptoms could be related to improved physical status and its psychological effects in reducing stress and anxiety, both of which are involved in the triggering and poorer prognosis of migraine (Wren et al., 2011; Probyn et al., 2017). Similar effects on ameliorating migraine symptoms by reducing stress could be achieved with relaxation and breathing exercises (Kaushik et al., 2005), which obtained a C grade.

In terms of lifestyle/behavior changes regarding exercise, which obtained a B grade, the introduction of better mealtime patterns or diet interventions, sleep behaviors, correct hydration and regular physical activity could improve migraine symptoms. The alteration or disruption of these patterns is related to the triggering and worsening of migraine, and the implementation of regular schedules and healthy behaviors could reduce the symptoms and impact of migraine (Khan et al., 2021).

Resistance training, which obtained a C grade of recommendation in favor of the intervention, achieves improvements in other chronic conditions, such as fibromyalgia and low back pain (Larsson et al., 2015; Wewege, Booth & Parmenter, 2018). Activating central pain modulation mechanisms and the improvement in psychological dimensions could explain the efficacy of this intervention in these conditions (Wewege, Booth & Parmenter, 2018; Rice et al., 2019). However, more research regarding the effects of resistance training for migraine treatment is needed to reduce the uncertain evidence in these patients. Moreover, strengthening of the neck muscles appears to be ineffective for reducing migraine pain, obtaining a C grade of recommendation against this intervention (Benatto et al., 2022).

Qigong and Tai Chi could improve migraine symptoms, with a D grade of recommendation. These interventions are based on case series and narrative reviews, and better study designs need to be developed to improve the conclusions as to their efficacy (Elinoff et al., 2009; Song & Chu, 2021).

Evaluating these results for grades of recommendation, almost 71% of the 90 videos included in the SIGN evaluation obtained a B grade of recommendation for their interventions. These videos were almost exclusively about aerobic exercise, yoga and lifestyle/behavior changes regarding exercise. However, most of the videos had low quality and low reliability regarding the flow and information offered. Moreover, there were no differences between both author categories’ videos regarding grades of recommendation. Healthcare and exercise professionals do not seem to have shared better evidence-based information than alternative medicine therapists and patients based on these results. However, most of grade B videos from alternative medicine/patient authors were about yoga. On the one hand, this intervention has achieved a high grade of recommendation, but most academics and healthcare professionals seem to have some negative bias toward yoga. Its eastern origin and the mystic aura that defines part of this intervention could have led to its rejection by most of the scientific community. On the other hand, the introduction of some mystical and pseudo-scientific concepts in the yoga videos evaluated in the present study has influenced the low scores obtained on the quality scales by these videos.

Thirty percent of the videos showed interventions with a C and D grade of recommendation, such as relaxation and breathing exercises, resistance exercises, Qigong and Tai Chi. Neck strengthening was shown as an effective intervention for migraine treatment in 2% of the videos included in the evaluation. However, the current classification into grades of recommendation has shown that neck strengthening is not an effective intervention for migraine pain improvement, with a C grade of recommendation. Only one study evaluated the efficacy of neck strengthening in migraine treatment (Benatto et al., 2022). If research in this field progresses, the grade of recommendation could change in the future.

The videos not included in the evaluation with grades of recommendation showed interventions that had been discarded from the classification due to lack of evidence, such as self-massage exercise and stretching of the cervical spine region. These interventions lacked evidence for reducing migraine pain and should not be presented as effective interventions on YouTube.

Conclusions

Prescribed exercises based on aerobic exercise, yoga and lifestyle/behavior changes regarding exercise are recommended, with a grade B according to the results of our classification into grades of recommendation. Nearly 50% of the total videos included in this study obtained this grade of recommendation, although most of them were of low quality and low reliability regarding information on the effects of exercise for migraine based on the quality evaluation. Therefore, better quality videos regarding exercise for migraine should be shared on YouTube, with better scoping strategies by healthcare and exercise professionals and with more rigorous YouTube policies for uploading videos. Moreover, better evidence-based information should be disseminated to the population to increase the quality of the content on social media.

Supplemental Information