Introduction
Migraine is the second largest cause of years lived with disability and the 14th leading cause of disability-adjusted life years globally (Steiner et al., Reference Steiner, Stovner, Jensen, Uluduz and Katsarava2020). Chronic migraine, defined as headaches occurring on at least 15 days a month, with eight of them having migraine features for more than three months (Olesen, Reference Olesen2018), results in substantially greater disability than episodic migraine (Stokes et al., Reference Stokes, Becker, Lipton, Sullivan, Wilcox, Wells, Manack, Proskorovsky, Gladstone and Buse2011). It affects relationships, careers, finances, and general health (Buse et al., Reference Buse, Murray, Dumas, Adams, Reed, Fanning and Lipton2019). Over half (57%) of people with chronic migraine miss five or more days of household activities over a three-month period (Bigal et al., Reference Bigal, Serrano, Reed and Lipton2008). People with chronic migraine access more medical resources than people with episodic migraine, with healthcare costs being nearly five times greater (Negro et al., Reference Negro, Sciattella, Rossi, Guglielmetti, Martelletti and Mennini2019). Topiramate, onabotulinumtoxin A (Botox), and calcitonin gene-related peptide (CGRP) monoclonal antibodies (MAbs) are of proven effectiveness for reducing headache/migraine days and, except for topiramate, improving headache-related quality of life for people with chronic migraine (Naghdi et al., Reference Naghdi, Underwood, Madan, Brown, Duncan, Matharu, Aksentyte, Davies, Rees, Cooklin, Grove and Mistry2023). The effect sizes are, however, modest, typically no more than 2.0 – 2.5 headache/migraine days per month. This means that even with adequate preventive drug treatment many people continue living with chronic migraine. In common with other chronic pain disorders there may be a role for educational, behavioural, cognitive, and self-management support interventions for people with chronic migraine (Probyn et al., Reference Probyn, Bowers, Mistry, Caldwell, Underwood, Patel, Sandhu, Matharu and Pincus2017). Their role, if any, has not yet been established. A 2019 Cochrane review of psychological therapies for prevention of migraine, episodic or chronic, concluded that it was not possible to determine if psychological interventions were effective for migraine because of the absence of high-quality research (Sharpe et al., Reference Sharpe, Dudeney, de C Williams, Nicholas, McPhee, Baillie, Welgampola and McGuire2019).
Method
Study registration
We wrote the protocol in May 2021 and prospectively registered it with the International Prospective Register of Systematic Reviews (CRD42021260376). We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for reporting the review.
We used the PICO framework to guide the development of the research question. Population: adults meeting the International Classification of Headache Disorders, 3rd edition (ICHD-3) criteria for chronic migraine, with or without aura. Interventions: educational, behavioural, and cognitive self-management support interventions. Control: usual therapy, typically involving pharmacological management. Outcomes: headache frequency, headache-related disability, pain intensity or severity of symptoms, quality of life, psychological wellbeing, and medication consumption.
Search
We developed search strategies (Supplementary Material 1) with expert librarian support to retrieve peer-reviewed randomised controlled trials (RCT), comparing educational, behavioural, cognitive, and self-management support interventions against usual care or sham interventions. We used Medical Subject Headings (MeSH) and free text searching, considering alternative spellings and truncations. As criteria for ‘chronic migraine’ were first introduced in 2004. We searched Cochrane, Embase, Medline, PsychINFO, Scopus, and Web of Science databases for studies published between 2004 and 18th November 2021. We updated this on the 27th of July 2023. We also carried out backward citation tracking. We retrieved full-text articles published in English language in peer-reviewed scientific journals, including electronic versions. We excluded grey literature and conference articles.
Eligibility criteria
Inclusion criteria
-
Adults ≥ 18 years old
-
International Classification of Headache Disorders-3 (ICHD-3) definition of chronic migraine with or without aura, either stated by authors or evident from the recruited population
-
Studies with mixed types of headaches, with at least 75% of the participants with chronic migraine
-
Chronic migraine with or without medication overuse
-
Behavioural, cognitive, educational, self-management support interventions
-
Behavioural interventions with an exercise component
Exclusion criteria
-
Headache due to secondary causes
-
Headaches <15 days a month
-
Headaches with no features of migraine (National Institute for Health and Care Excellence, 2021)
-
Exercise only trials
-
Interventions involving use of any apparatus, i.e., acupuncture, transcutaneous electrical stimulation
-
Sham interventions
-
Trials with <15 participants per treatment arm
Selection process and data extraction
Two reviewers (AA, ND) independently screened all records by title and abstract, followed by full-text analysis of the selected trials. We discussed our decisions. A third reviewer (MU) acted as an arbitrator. We contacted authors for clarification where there was ambiguity about whether the trial met the eligibility criteria. The excluded articles following full-text appraisal are listed in Supplementary Material 2.
For each included study, AA and ND independently extracted data on country, size, intervention type, duration, delivery mode, materials for data collection, follow up, and declaration of interest.
Risk of bias assessment
AA and ND independently assessed the risk of bias, as per the Cochrane Risk of Bias Tool v2.0 (Sterne et al., Reference Sterne, Savović, Page, Elbers, Blencowe, Boutron, Cates, Cheng, Corbett, Eldridge, Emberson, Hernán, Hopewell, Hróbjartsson, Junqueira, Jüni, Kirkham, Lasserson, Li, McAleenan, Reeves, Shepperd, Shrier, Stewart, Tilling, White, Whiting and Higgins2019), with MU reviewing any disagreements. For one trial, for which MU is the chief investigator, an independent reviewer (JB) considered any disagreements. We classified the papers as having low, some concerns, or high risk of bias, as per the Cochrane Handbook for Systematic Reviews (Higgins et al., Reference Higgins, Thomas, Chandler, Cumpston, Li, Page and Welch2019).
Data synthesis
We classified the studies based on how the authors described the interventions – psychological, behavioural, or educational interventions. We used the general framework for synthesis outlined in section 9.2A in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins et al., Reference Higgins, Thomas, Chandler, Cumpston, Li, Page and Welch2019). We extracted mean, standard deviation, confidence intervals, effect size, and p-values for treatment and control groups for the following outcomes: headache frequency, pain intensity/severity of symptoms, quality of life, psychological wellbeing, and medication consumption. We compared the values at baseline and at follow-up. We assessed whether the data were appropriate for statistical pooling for meta-analysis and produced a narrative synthesis of the results when pooling was not feasible.
Deviations from protocol
In error, we omitted headache-related disability as a specified outcome from our original protocol. This has been identified as being of similar importance to people with chronic migraine as headache/migraine days (Haywood et al., Reference Haywood, Potter, Froud, Pearce, Box, Muldoon, Lipton, Petrou, Rendas-Baum, Logan, Stewart, Underwood and Matharu2021). We have therefore also extracted and presented these data.
Results
In November 2021, we identified 3947 articles through searching databases and four trials following backward citation searching. After screening, we assessed 34 full-text articles and included four studies (Figure 1).
Study flow diagram November 2021.
We updated our search in July 2023 and found two eligible studies published since November 2021 (Figure 2). One study (Grazzi et al., Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022) was published as two separate publications, with three-month and 12-month follow up.
Study flow diagram July 2023.
Three studies were done in the USA (Calhoun and Ford, Reference Calhoun and Ford2007a; Seng et al., Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021; Smitherman et al., Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016), and one each in Iran (Rashid-Tavalai et al., Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015), Italy (Grazzi et al., Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022), and the UK (Underwood et al., Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a, Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins, Mistry, Mistry, Newton, Nichols, Norman, Padfield, Patel, Petrou, Pincus, Potter, Sandhu, Stewart, Taylor and Matharu2023b). Two studies (Seng et al., Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021; Underwood et al., Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a) included mixed populations of episodic and chronic migraine and reported some results separately just for those people with chronic migraine. The six trials included 713 participants with chronic migraine, with sample sizes ranging from 31 (Smitherman et al., Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016; Seng et al., Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) to 396 (Underwood et al., Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a). No two studies used similar interventions: Calhoun and Ford (Reference Calhoun and Ford2007a), behavioural sleep modification; Grazzi et al., (Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022), mindfulness and education about migraine as an addition to medication withdrawal in people with medication overuse headache; Rashid-Tavalai et al. (Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015), coping skills training; Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) mindfulness-based cognitive therapy; Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016), cognitive-behavioural therapy for insomnia (CBTi); Underwood et al., (Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a), an educational and self-management support intervention. Two studies (Calhoun and Ford, Reference Calhoun and Ford2007a; Smitherman et al., Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) used sham interventions for their control group. The rest used usual care controls. All studies recruited participants from physician referrals or clinic attendance; additionally, Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) used advertisements. Underwood et al. (Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a) allowed participants to self-refer; however, most participants were recruited by writing to them.
The participants’ mean age ranged from 30 to 47 years, and they were predominantly female (83%–100%). Underwood et al. (Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a), Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021), and Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) reported that their population was predominantly white (80%, 77%, and 81% respectively). No other studies reported data on ethnicity. There were 40% participants with a formal diagnosis of depression in the study by Calhoun and Ford (Reference Calhoun and Ford2007a), whereas all participants in the study by Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) had insomnia. Underwood et al. (Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a) reported that of all participants, including those with chronic tension-type headache and episodic migraine 53% had probable anxiety and 22% had probable depression. Medication overuse headache was present in 100% participants in Grazzi et al. (Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022), 74% in Calhoun and Ford (Reference Calhoun and Ford2007), 56% in Underwood (data on all participants (Underwood et al., Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a)), and 0% in Smitherman et al. Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains(2016). Neither Rashid-Tavalai et al. Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour(2015) or Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) reported proportion with medication overuse headache.
We found protocols on clinicaltrials.gov for four studies (Grazzi et al., Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022; Seng et al., Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021; Smitherman et al., Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016; Underwood et al., Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins, Mistry, Mistry, Newton, Nichols, Norman, Padfield, Patel, Petrou, Pincus, Potter, Sandhu, Stewart, Taylor and Matharu2023b). The remaining two studies (Calhoun and Ford, Reference Calhoun and Ford2007; Rashid-Tavalai et al., Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015) did not refer to a protocol.
Risk of bias
We judged two trials to be at high risk of bias (Calhoun and Ford, Reference Calhoun and Ford2007; Rashid-Tavalai et al., Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015) and four had some concerns (Grazzi et al., Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022; Seng et al., Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021; Smitherman et al., Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016; Underwood et al., Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a) (Figure 3, Supplementary material 3).
Risk of bias assessment of the included studies.
The heterogenous nature of the interventions tested and outcome reporting mean that statistical pooling of results was not appropriate. Both Underwood et al. Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry(2023a) and Grazzi et al. Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi(2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022) measured HIT-6 at 12 months. However, the different analyses presented mean data pooling was not possible. We therefore present a narrative summary. No studies reported separately on migraine days (Table 1). For reporting purposes, we have categorised the studies broadly as educational interventions (Rashid-Tavalai et al., Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015; Underwood et al., Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a), psychological interventions (Grazzi et al., Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022; Seng et al., Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) and behavioural interventions (Calhoun and Ford, Reference Calhoun and Ford2007; Smitherman et al., Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016). We based this on authors’ descriptions in the original studies.
Characteristics of included studies
Headache frequency
Psychological interventions
The results of all the studies are summarised in Table 2. At three months, Grazzi et al. (Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022) reported a difference in headache frequency; the control group had 31.3 (SD 20.3) headache days, and the intervention group had 26.6 (SD 21.8). They found a statistically significant time by group interaction (p = 0.029). They also report that 45/83 (54%) of control group and 60/79 (76%) of intervention had ≥50% reduction in headache frequency (p=0.004). At 12 months, there was a statistically significant reduction in the headache frequency in the intervention group when compared to the control group (p < 0.0001); 69/88 (78%) of individuals in the intervention group achieved a 50% or more headache frequency reduction, compared to 43/89 (48%) in the control group.
Summary of results; NS = not significant; * = significant at p = .05, **significant at p = .025, ‘-‘ = data not reported
Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) did not find any effect of mindfulness-based cognitive therapy on headache frequency at four months. Mean difference in headache days/30 days was 0.18 (95% CI −4.63, 4.99).
Behavioural interventions
Calhoun and Ford (Reference Calhoun and Ford2007) reported a statistically significant reduction in mean headache frequency at six weeks, following behavioural sleep modification training: 17.4 vs 23.9 headache days in the previous 28 days (p = 0.01). Furthermore, none of the participants in the control group reverted to episodic migraine (p = 0.029) at follow-up.
Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) reported that at two weeks post-intervention, the control group showed a greater reduction in headache frequency than the CBTi (Cognitive Behavioural Therapy intervention) group (36% and 27% respectively). The odds ratio of a participant having a headache after the CBTi was 1.06 (p = 0.883, 95% CI 0.52 to 2.15). At six weeks, the headache frequency reduced by 49% in the CBTi group and by 25% for controls. The odds ratio of a participant having a headache after CBTi was 0.40 (p = 0.028, 95% CI 0.17 to 0.91) in comparison to control. As they assessed two endpoints of their primary outcome, they used a criterion of p < 0.025 to determine significance, which their results did not meet. These analyses were based on total person days with/without headaches reported by participants rather than headache days per person. It is unclear how this was accounted for in the analyses. A 50% reduction in headache frequency at follow up was reported by 5/15 (33%) of control 7/16 (44%) of the CBTi group (no statistical test performed).
Headache-related disability
Educational interventions
Underwood et al. Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry(2023a) reported no significant effects of their educational and self-management support intervention at 12 months (p = 0.325, mean difference (−0.7, 95% CI −1.97 to 0.65)) using the Headache Impact Test (HIT-6, scale range 36–78) (Kosinski et al., Reference Kosinski, Bayliss, Bjorner, Ware, Garber, Batenhorst, Cady, Dahlöf, Dowson and Tepper2003). The limits of the 95% confidence interval exclude the target difference of 2.0 set for the trial.
Psychological interventions
Grazzi et al. (Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023) reported improved outcomes on Migraine Specific Quality of Life Questionnaire (MSQ) and Migraine Disability Assessment Test (MIDAS) at 12 months only, with no difference at intermediate timepoints at any timepoint on HIT-6 and World Health Organisation Disability Assessment Schedule (WHODAS-12). These data are presented graphically only. They further report statistically significant difference in proportion with ≥6 point improvement in HIT-6; 29/78 (37%) of the control group and 45/76 (59%) of the intervention group [chi squared p = 0.006, odds ratio 2.45, 95% CI 1.28 to 4.69) had achieved the reduction at 12 months.
Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) reported that the estimated proportion of people reporting severe disability (MIDAS score of ≥21) was reduced by 16.4% in participants receiving mindfulness-based cognitive therapy and increased by 8.4% in participants receiving usual care. The mean difference in MIDAS scores (0-270) was −0.7 (95% CI: −1.27, −0.08). The limits of the 95% confidence interval exclude the target difference of 1.29 set for the trial. Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) also used the Headache Disability Inventory (Jacobson et al., Reference Jacobson, Ramadan, Aggarwal and Newman1994) to measure headache disability, but no significant differences were found.
Behavioural interventions
Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) reported no differences in MIDAS and HIT-6 scores at two weeks or six weeks after correcting for baseline scores. However, no analysis is presented.
Pain intensity/severity of symptoms
Educational interventions
Rashid-Tavalai et al. (Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015) used the Migraine Headache Index (Hamedanizadeh et al., Reference Hamedanizadeh, Ebadi, Zandi and Kachuyee2010) and reported that the intervention did not have a significant effect on pain severity (p = 0.26).
Psychological interventions
Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) found that the average attack intensity over 30 days did not differ between groups at four months; difference −0.02 (95%CI −0.16, 0.13, scale range 0–3).
Behavioural interventions
Calhoun and Ford (Reference Calhoun and Ford2007) expressed pain intensity as a headache index score. They calculated it by multiplying the number of days with severe headache by 3, moderate by 2 and mild by 1. After summating the score, the authors reported a significant difference between groups (p = 0.001). Between-group difference is not presented.
Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) determined headache severity by asking participants to rate each headache using a 0 to 10 scale. The authors stated that there were no differences between groups when controlling for baseline; however, no statistical analyses were presented.
Quality of life
Educational interventions
Rashid-Tavalai et al. (Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015) used the World Health Organisation Quality of Life (WHO-QoL) questionnaire (Saxena et al., Reference Saxena, Orley and Group1997). When comparing the scores between groups at seven weeks they reported a statistically significant difference (p = 0.049).
Psychological interventions
Grazzi et al. (Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023) reported that the participants in the intervention group better quality of life scores at 12 months, but the results were not statistically significant.
Psychological wellbeing
Behavioural interventions
Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) reported no difference between CBTi and control groups at six weeks in Patient Health Questionnaire (PHQ-9) (Kroenke et al., Reference Kroenke, Spitzer and Williams2001) scores for depression (p = 0.054), but not in the Generalised Anxiety Disorder (GAD-7) (Spitzer et al., Reference Spitzer, Kroenke, Williams and Löwe2006) scores for anxiety (p = 0.420).
Medication consumption
Behavioural interventions
Calhoun and Ford (Reference Calhoun and Ford2007) reported that at baseline 74% of all participants overused medication; however, 12 weeks after the intervention this was stopped in all patients who reverted to episodic migraine (48%) and in 60% of participants that did not revert.
Psychological interventions
Grazzi et al. (Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023) reported that the participants in the intervention group showed a significantly lower consumption of medications (including total drug intake and non-steroidal anti-inflammatory drugs) than the control group at 12 months, but not the triptans. The data was presented graphically only.
Discussion
Summary of the main findings
In this systematic review, we investigated the effectiveness of educational, behavioural, psychological, and self-management support interventions for chronic migraine. We analysed two educational, two psycho-educational and two behavioural trials. It is disappointing how few eligible studies we identified, with so few participants (N = 713), when chronic migraine is such a common and sometimes profoundly disabling condition. It is further disappointing that the published trials are not adequately powered to identify important benefits for people with chronic migraine. Two, Calhoun and Ford (Reference Calhoun and Ford2007) (N = 43) and Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) (N = 31) are explicitly pilot studies; two, Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) (N = 31) and Underwood et al. (Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry2023a) (N = 396), only report on the sub-set of the overall trial population with chronic migraine; and one, Rashid-Tavalai et al. (Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015) (N = 35), does not provide a priori sample size calculation. Only Grazzi et al. (Reference Grazzi, D’Amico, Guastafierro, Demichelis, Erbetta, Fedeli, Nigri, Ciusani, Barbara and Raggi2023, Reference Grazzi, Raggi, Guastafierro, Passavanti, Marcassoli, Montisano and D’Amico2022) (N = 177) present a sample size calculation relevant to a chronic migraine population, based on showing an improvement on portion achieving a ≥50% reduction in headache days from 48% to 68%. By way of a benchmark for the size of trials needed to show difference in headache-related quality of life, a simple trial designed to show a 2.0 difference in HIT-6, the worthwhile difference set by Underwood (Patel et al., Reference Patel, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Mistry, Mistry, Nichols, Petrou, Pincus, Potter, Sandhu, Stewart, Taylor, Underwood and Matharu2020) with the observed baseline standard deviation of 5.5, from that trial, with 80% power at the 5% significance level requires data on 240 participants. It is even more disappointing that there is insufficient good-quality evidence of effectiveness for any of the treatment approaches to support their use outside of a research context.
Underwood et al. Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry(2023a) included both people with chronic migraine and people with chronic tension-type headache and episodic migraine to assess the effectiveness of their educational intervention. Only the primary outcome of HIT-6 at 12 months is reported separately for people with chronic migraine. The limits of the 95% confidence interval for HIT-6 at one year excluded the pre-specified target difference, allowing them to conclude that the intervention tested was not effective on HIT-6. The authors’ overall conclusion for the main study, including people with episodic migraine, is that their data excludes the possibility that the intervention tested might be effective. There would seem to be little justification for any further evaluation of approach used by Underwood et al. Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry(2023a). For the remaining trials where there is at least some evidence, with exception of Grazzi, the reporting of the interventions tested is inadequate to allow replication. To our knowledge, none have used the Template for Intervention Description and Replication (TIDieR) checklist (Hoffmann et al., Reference Hoffmann, Glasziou, Boutron, Milne, Perera, Moher, Altman, Barbour, Macdonald, Johnston, Kadoorie, Dixon-Woods, McCulloch, Wyatt, Phelan and Michie2014).
Calhoun and Ford (Reference Calhoun and Ford2007) found a significantly reduced headache frequency and pain severity in the behavioural intervention group at six weeks when compared to controls. As sleep problems can exacerbate chronic migraine (Schwedt, Reference Schwedt2014), facilitating sleep hygiene through behavioural modifications could effectively improve headache-related outcomes for individuals with chronic migraine. Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) also reported reductions in the odds of experiencing a headache following behavioural CBTi, but no statistical significance between groups. With both behavioural interventions in this review being almost identical, it is possible that Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) failed to reach significance as their participants had comorbid insomnia, and simple behavioural sleep instructions may not be sufficient to achieve a clinically meaningful change in this population. Lastly, Smitherman et al. Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains(2016) reported borderline significance between groups for the reduction in depression scores at six weeks. Though not specific for chronic migraine, CBT was effective in reducing depression scores for individuals with migraine in a 2012 trial (N = 213) (Bromberg et al., Reference Bromberg, Wood, Black, Surette, Zacharoff and Chiauzzi2012). Given the frequent co-occurrence of depression in people with migraine (Yang et al., Reference Yang, Zhao, Heath, Madden, Martin and Nyholt2016), managing negative thoughts could be a valuable tool in improving their mental health (Martin et al., Reference Martin, Aiello, Gilson, Meadows, Milgrom and Reece2015). However, Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) did not focus on the cognitive component of CBT, which may explain why they did not find significant differences in depression scores.
The studies classified as psychological assessed the effectiveness of mindfulness-based cognitive therapy (Seng et al., Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) and coping skills (Rashid-Tavalai et al., Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015). In line with a Cochrane review of psychological therapies for migraines (Sharpe et al., Reference Sharpe, Dudeney, de C Williams, Nicholas, McPhee, Baillie, Welgampola and McGuire2019), Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) reported no significant difference in headache frequency between groups at four months. Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) also reported no significant improvements in the average intensity of pain, whereas a systematic review demonstrated that interventions with mindfulness components had small effects on pain intensity in participants with migraines (Probyn et al., Reference Probyn, Bowers, Mistry, Caldwell, Underwood, Patel, Sandhu, Matharu and Pincus2017). However, Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) reported a statistically significant reduction in headache-related disability using MIDAS questionnaire. Their results highlight that headache frequency and headache-related disability are not co-dependent. This is supported by the Delphi consensus process, where the authors recommended that non-pharmacological interventions should prioritise headache-related disability over headache frequency (Luedtke et al., Reference Luedtke, Basener, Bedei, Castien, Chaibi, Falla, Fernández-De-Las-Peñas, Gustafsson, Hall and Jull2020). This is because interventions, such as mindfulness-based therapy, target the ability to function with pain, not eliminate the occurrence of the disease (Cherkin, Reference Cherkin2021).
Completeness and applicability of the evidence
We used the International Classification of Headache Disorders 3rd edition (ICHD-3) (Olesen et al., Reference Olesen, Goadsby and Steiner2003) criteria for chronic migraine, which requires individuals to have 15 headaches a month, with eight migraines for longer than three months. These strict criteria meant that very few studies could be included in our analysis. This led us to exclude nearly all the data from Underwood et al. Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry(2023a) where they had identified a population using an epidemiological definition of chronic headache rather than ICHD-3. The authors justified the inclusion of this population by explaining that in practice they would have access to this intervention and are therefore a clinically relevant population. Additionally, we excluded a trial with an average of 12 headache days a month (Cousins et al., Reference Cousins, Ridsdale, Goldstein, Noble, Moorey and Seed2015), but a longitudinal study with over a thousand participants with chronic migraine showed that nearly 75% of those with chronic migraine at baseline dropped below this diagnostic boundary at least once in 12 months (Serrano et al., Reference Serrano, Lipton, Scher, Reed, Stewart, Adams and Buse2017). Furthermore, participants just below the required diagnostic threshold for chronic migraine were reported to have comparable levels of headache-related disability (Chalmer et al., Reference Chalmer, Hansen, Lebedeva, Dodick, Lipton and Olesen2020). Identifying such individuals would ensure that appropriate treatment is given according to their needs.
The trials included in this study had a majority female population. This is due to the higher prevalence of chronic migraine in women in comparison to men; a systematic review of the global epidemiology of chronic migraine found the prevalence was 2.5–6.5 times higher in women (Natoli et al., Reference Natoli, Manack, Dean, Butler, Turkel, Stovner and Lipton2010). Consequently, our analysis is less applicable to males with chronic migraine. Furthermore, of the three studies that reported on ethnicity, 77%–81% of the population were white reducing the generalisability of this data. We only included trials of adults aged 18 or over.
Educational, behavioural, and psychological interventions could become tools to manage chronic migraine long-term; however, we cannot determine any long-term effects from these studies.
Quality of the evidence
No trial had low risk of bias, with reasons listed in Supplementary Material 3. We found the presentation of results by Seng et al. (Reference Seng, Conway, Grinberg, Patel, Marzouk, Rosenberg, Metts, Day, Minen and Buse2021) to be difficult to interpret due to lack of clarification in table legends. Rashid-Tavalai et al. (Reference Rashid-Tavalai, Bakhshani, Amirifard and Lashkaripour2015) used a scale called ‘Migraine Headache Index’, which we were unable to access. Additionally, no raw data were reported, leaving us unable to determine how pain intensity had been calculated. This affected the credibility of the conclusions drawn by the authors. The use of an odds ratio to report headache frequency by Smitherman et al. (Reference Smitherman, Walters, Davis, Ambrose, Roland, Houle and Rains2016) to be a non-standard approach, as ‘headache days’ is a continuous variable. Additionally, migraine days have been shown to cluster; thus, the probability of having a migraine on any day is affected by the occurrence of migraines on previous days (Barra et al., Reference Barra, Dahl, Vetvik and MacGregor2020). We also noted the lack of clarity in reporting the effect of intervention on headache frequency by Calhoun and Ford Reference Calhoun and Ford(2007); the analysis was stated as being between-groups, but the supplementary graph seemed to indicate a within-group analysis, comparing baseline with follow-up. We contacted the authors for clarification but had no success, leaving this ambiguity unresolved. Grazzi present the majority of their secondary analyses in a graphical form only, limiting their interpretation.
Some studies did not specify what they meant by ‘usual care’ for their control groups. For instance, Rashid-Tavalai described it as ‘pharmacotherapy’, whereas Seng defined it as ‘standard migraine care’. These two trials were carried out in different countries; therefore, ‘usual’ treatment may mean different pharmacological regimes and may include additional strategies, such as headache diaries. Further clarification would be beneficial to facilitate comparison. Smitherman and Calhoun & Ford used sham instructions in addition to pharmacological management for their control group. These included five-step instructions, such as performing acupuncture twice daily and scheduling a consistent meal time. Underwood referred the control group back for usual care from their general practitioner along with recommendations for appropriate drug treatment and relaxation compact disc. Grazzi provided the most detail regarding their control group, specifying that their ‘treatment as usual’ condition consisted of overused medication withdrawal, education on proper medication use and prescriptions of suitable pharmacological agents. They offered a comprehensive summary of the methods they utilised to facilitate withdrawal as well as the exact medication and dosage that was used as usual migraine prophylaxis. This offers clarity and facilitates replication in future studies.
Practical implications
Three interventions utilised graduate-level psychologists to deliver the interventions. A 2017 systematic review found no evidence that psychological interventions delivered by a specialist were more effective than those delivered by an allied healthcare professional (Probyn et al., Reference Probyn, Bowers, Mistry, Caldwell, Underwood, Patel, Sandhu, Matharu and Pincus2017). This has important implications for the cost of delivering the intervention. In contrast, a feasibility study found that it was not practical to include lay migraineurs, as the unpredictability of migraine attacks prevented them from being able to commit to delivering the sessions (White et al., Reference White, Potter, Patel, Nichols, Haywood, Hee, Mistry, Carnes, Taylor and Underwood2019).
Whilst largely focused on drug trials, ICHD guideline for prophylactic treatment of chronic migraine is a useful tool to inform trial design (Silberstein, Reference Silberstein, Tfelt-Hansen, Dodick, Limmroth, Lipton, Pascual and Wang2008). To reduce variability between trials in future, authors should adhere to the ICHD-3 criteria when recruiting the participants and clearly define their population. To facilitate reproducibility and critical appraisal, we encourage the use of the CONSORT checklist (Hopewell et al., Reference Hopewell, Chan, Collins, Hróbjartsson, Moher, Schulz, Tunn, Aggarwal, Berkwits, Berlin and Bhandari2025). A published protocol, adhering to the SPIRIT guidelines would further enhance interpretation (Chan et al., Reference Chan, Boutron, Hopewell, Moher, Schulz, Collins, Tunn, Aggarwal, Berkwits, Berlin and Bhandari2025). Better descriptions of both control and intervention groups, including the underpinning theory informing the content of the active intervention is needed. This may be best addressed by publication of separate intervention development papers, as done by Underwood (Patel, Reference Patel, Potter, Matharu, Carnes, Taylor, Nichols, Pincus, Underwood and Sandhu2019). We were unable to do a meta-analysis due to the heterogeneity of the reported outcomes. Future trials could utilise the two-domain core outcome, headache frequency and headache-related disability identified as part of a core outcome set. This will allow for meaningful pooling of data in future reviews.
Strengths and limitations
We followed the PRISMA checklist for the reporting of this systematic review and prospectively registered our protocol. We assessed the trials using the Cochrane Risk of bias tool. We could not perform any meta-analyses to estimate the overall effect of the interventions due to the heterogeneity of the trials. Moreover, we only included articles published in English and may have missed relevant trials in other languages.
Our search was conducted in July 2023; therefore, we re-ran the original search in PubMed in June 2025 to identify any newly published trials. The updated search yielded 26 new studies; however, none met our predefined eligibility criteria. Therefore, our findings remain current and up to date.
Conclusion
Chronic migraine is a disabling condition for which pharmacological treatment is not always sufficient. Therefore, non-pharmacological support interventions could complement the existing treatment and improve headache-related outcomes. We found some weak evidence for the effectiveness of educational, behavioural, and psychological interventions in reducing headache frequency, headache-related disability, and pain intensity. However, research by Underwood et al. Reference Underwood, Achana, Carnes, Eldridge, Ellard, Griffiths, Haywood, Hee, Higgins and Mistry(2023a) demonstrated that in a large randomised controlled trial, with long-term follow-up, educational/behavioural treatment for chronic migraine had no clinically meaningful improvement for patients. These interventions cannot yet be considered part of routine care. Nevertheless, patients ought to be actively involved in the decision-making process when determining the most appropriate management plan for chronic migraine, including how to make the best use of medication for migraine prophylaxis. Focus should also be placed on finding alternative treatments to better support individuals living with this debilitating condition.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1463423625100571
Data availability statement
Supplemental material for this article is available online (CRD42021260376).
Acknowledgements
The authors would like to thank Sam Johnson who has reviewed our search strategy and advised us on how to adapt it for different databases to ensure accurate and comprehensive literature retrieval.
Author contributions
AA - developed the search strategy, undertook the searches, initial screening and retrieving papers, assessed against the inclusion criteria, appraised the quality of papers, abstracted data from papers for synthesis; ND - developed the search strategy, undertook the searches, initial screening and retrieving papers, assessed against the inclusion criteria, appraised the quality of papers, abstracted data from papers for synthesis; MR - provided guidance on data retrieval to ensure accurate data extraction for analysis; MU - provided expert clinical guidance on all parts of the project (except on appraising and extracting data from the paper where he was a published author), checked the accuracy of data extraction; MM - checked the accuracy of data extraction, provided expert clinical guidance on all parts of the project and helped critically appraise the included literature; All authors contributed towards developing the research question; All authors reviewed the included papers; All authors were involved in writing draft and final versions of the report; All authors approved the submitted version; All authors agree to be personally accountable for their own contributions to the work.
Funding statement
The authors received no financial support for the research, authorship, or publication of this article.
Competing interests
MM is the President of the medical advisory board of the CSF Leak Association. He has received consulting fees from AbbVie, TEVA, Lundbeck, Eli Lilly, Salvia, Pfizer. He has received payment for the development of educational presentations from AbbVie, Pfizer and Eli Lilly and support for attending a meeting from Pfizer. He has is on the advisory board for AbbVie, TEVA, Lunbeck, Eli Lilly, Salvia and Pfizer. He has the following patent issued WO2018051103A1: System and method for diagnosing and treating headaches. He has stock options with Tesla, Adobe, Nvidia, META and Microsoft. He has received grants from Abbott, Medtronic, and The Ehlers Danlos society. He is co- investigator on multiple NIHR grants. He has published multiple papers on headache disorders, some of which are cited in this paper.
MU is chief investigator or co-investigator on multiple previous and current research grants from the UK National Institute for Health Research and is a co-investigator on grants funded by the Australian NHMRC and Norwegian MRC. He was an NIHR Senior Investigator until March 2021. He is a director and shareholder of Clinvivo Ltd that provides electronic data collection for health services research. He is part of an academic partnership with Serco Ltd, funded by the European Social Fund, related to return-to-work initiatives. He receives some salary support from University Hospitals Coventry and Warwickshire. He is a co-investigator on two current and one completed NIHR funded studies that have, or have had, additional support from Stryker Ltd. He has published multiple papers on headache disorders, some of which are cited in this paper.
Ethical standards
Not applicable.
Open access
