Friday, October 14, 2011

Does Exercise Prevent Migraines as Good as Meds?

Migraine A newly-reported clinical trial from Sweden found that exercise is as effective as drug therapy, or as relaxation techniques, in preventing migraines. However, reductions in monthly frequency of migraines with all 3 treatments appear rather modest and may be disappointing to some patients. Yet, there were limitations of this research and the true value of these treatment approaches might not have been fully explored.

Treatments for migraine prevention are an important component of care for patients suffering this serious and often debilitating condition. Pharmacotherapy — such as with the antiseizure medication topiramate — is often a first line treatment, and nondrug behavioral therapies, such as structured relaxation, have been documented as providing some benefits. Much less is known about the beneficial role exercise might play in preventing the frequency and severity of migraine attacks and how it might compare to the other two therapies.

Therefore, researchers at the University of Gothenburg, Sweden, conducted a single-center, prospective, randomized clinical trial in adult migraineurs to compare the 3 modalities as preventative measures [Varkey et al. 2011]. They enrolled 91 subjects and acquired baseline data during a 4 to 12 week period. The mean age of participants was 44 years, 92% were female, the mean duration of migraine disorder was 26 years, and subjects experienced 4 migraine attacks/month on average. Following the baseline period, subjects were randomized to 1 of 3 groups for 12 weeks of treatment:

  1. Supervised group aerobic exercise (indoor cycling) for 40 minutes three times a week.

  2. A once per week, individually supervised relaxation program — relaxation and breathing exercises, and stress-management techniques — with at-home practice between sessions using an instructional CD.

  3. Once daily topiramate, titrated upward 25 mg/week until highest tolerated dose was achieved (maximum allowed, 200 mg/d).

The primary efficacy outcome variable was the mean reduction in frequency of migraine attacks during the final month of treatment compared with baseline. Secondary measures included migraine status (eg, pain intensity, acute medication use), quality of life, level of physical activity, and oxygen uptake. Followup data were also collected on these measures at 3- and 6-months posttreatment.

Writing in an early online edition of the journal Cephalalgia from the International Headache society, the authors report that in an intention-to-treat (ITT) analysis the mean reduction in number of monthly migraine attacks was (A) 0.93 (95% confidence interval (CI), 0.31 to 1.54) in the exercise group, (B) 0.83 (95% CI, 0.22–1.45) in the relaxation group, and (C) 0.97 (95% CI, 0.36–1.58) in the topiramate group. The differences between groups on this primary measure were not statistically significant (P = 0.95) [which is confirmed by overlapping CIs, but notice the small effect size estimates at the lower ends of CI ranges].

Among the secondary outcome measures, there was a significant reduction in migraine pain favoring drug therapy (P = 0.04), and in oxygen uptake favoring the exercise group (P = 0.008) as might be expected, but there were no significant differences between groups on the other measures. Adverse treatment effects, none classified as serious, were experienced by a third of subjects in the topiramate group, but none were reported in the exercise or relaxation groups.

The authors conclude that exercise may be an important option for the prophylactic management of migraine in patients who do not benefit from or do not want to take daily medication or to participate in a relaxation techniques program. However, in all 3 groups the absolute changes from baseline and the treatment effect sizes were modest, and there also are some questions regarding the interpretation and clinical importance of study outcomes.

COMMENTARY: Overall, this was a well-designed study, with appropriate enrollment and randomization procedures, straightforward treatment conditions, and clear-cut measurements of outcome variables. Another important feature was that the researchers performed assessments using both intention-to-treat (ITT) and per-protocol (PP) methodologies.

An ITT analysis accounts for data from all subjects enrolled in a trial, including last-known data for subjects who did not adhere to the study plan or who dropped out for any reason. Often, subjects are considered treatment failures if they withdraw or do not fully participate, so this approach usually reduces the magnitude of treatment effects; however, it is a more naturalistic approach in that real-world patients are similarly often not fully compliant with treatment regimens. Conversely, a PP analysis considers only subjects who were fully compliant with the study plan, or protocol, and completed the treatment regimen; this theoretically provides a more accurate portrayal of absolute effects associated with treatment.

In this study, there were at least slightly greater improvements in most data measures reflected in the PP analyses as compared with the ITT analysis; however, only two were of statistical significance. Topiramate therapy significantly reduced migraine pain intensity during the treatment period (15 points on a 0-100 Visual Analog Scale; P = 0.01), and at 6-months posttreatment topiramate reduced acute medication use by about 5 doses/month (P = 0.05). In almost all other instances, PP and ITT analyses were comparable for all measures across all 3 treatment groups — and, most importantly, improvements from baseline on all measures were of only modest proportions.

From an evidence-based pain management perspective, some additional research points are worth noting…

  • The investigators report having conducted a power analysis and determined that, to detect a statistically and clinically significant reduction of 1.0 migraine attack per month, they would need at least 30 subjects per group. This assumed a standard deviation of 1.2 and 80% power with a 2-sided alpha of 0.05 [statistical power concepts were discussed in an UPDATE here].

    The authors do not explain why 1 less attack/month would be sufficiently of clinical importance. In this study, with subjects averaging 4 attacks per month, it would represent only a 25% reduction in frequency that may or may not generally be considered as adequate by patients.

  • The researchers did not make allowances for study drop-outs and there were only 25 subjects in the exercise group, 26 in the relaxation group, and 21 in the topiramate group who completed treatment. Therefore, at this point, the study was underpowered to detect significant differences between groups even if some actually did exist (that is, a Type II error).

  • In their ITT analysis of the primary outcome (reduction in attack frequency/month) none of the groups achieved the 1.0 difference from baseline considered as clinically important. In the PP analysis, only the relaxation group (at 1.17) achieved this threshold, but differences between groups were still not statistically different in this underpowered analysis. It is impossible to know if distinctions between groups might have been more robust and significant if more subjects were enrolled.

  • From another perspective, the authors include a table indicating response rates on the primary efficacy variable (migraine frequency), using both ITT and PP analyses. From 48% to 65% of participants were “not clinically improved (≤25% improvement in migraine frequency),” depending on the treatment group and type of analysis. Although, from 23% to 38% were responders with ≥50% improvement. Overall, however, at least half of participants did not benefit to a clinically important extent from any of the 3 therapies.

    Interestingly, topiramate therapy performed worst of all three therapies in the ITT analysis (65% not clinically improved) but best in the PP analysis (38% responders with ≥50% improvement). This highlights how, when it comes to medication therapy, strict adherence to the prescribed regimen can make quite a large impact on outcomes. If the authors had only reported an ITT analysis, topiramate would have looked like the most inferior of the 3 treatments.

  • Some of the other differences from baseline to posttreatment were of only modest proportions, and of questionable clinical significance. For example, in the PP analysis the 15-point reduction in migraine pain afforded by topiramate on a 0 to 100 point VAS, noted above, was significantly superior to exercise (5.1 points) or relaxation (3.6 points), but the absolute sizes of any of these reductions could be of questionable value to patients; eg, In the best case, would a migraine pain-score reduction from, say, 80 to 65 be perceived by patients as an important outcome of therapy?

    The authors acknowledge that, “The treatment effects in our study were somewhat smaller than expected and the effects in the medicine group are smaller compared to other studies of topiramate.” They also point out that, contrary to their own findings, in at least one other study exercise was found to actually be a “trigger factor” for migraines, which might encourage some migraineurs to avoid exercise entirely.

  • The occurrence of adverse effects with topiramate might be of some concern. Upon learning of the drug they were going to receive, 7 subjects (23%) randomized to the topiramate group immediately dropped out of the study. Among the remaining 24 who took at least one dose of the drug, a third of them experienced adverse effects and 3 of them withdrew from the study due to these. Although none of the effects were considered serious by the investigators, some patients experienced multiple adverse effects of medication therapy.

  • The authors insist throughout their published report that this study was adequately powered; yet, even with the minimally-required 30 subjects per group, as in the ITT analysis, one must question if Type II errors were present, failing to significantly distinguish one therapy from the other, and/or if these small numbers of patients are sufficiently representative of a typical population with migraine to assume external validity.

Taken together, this study has many limitations; however, it suggests that exercise, relaxation, or topiramate therapies might be of some benefit in some patients with multiple monthly migraine attacks; although, one treatment does not appear to be spectacularly superior to the other. The choice may come down to patient preference for type of treatment, with drug therapy being easiest for many but also having potentially bothersome adverse effects. Exercise could be the most time/effort intensive but with added benefits in terms of overall well-being (unless exercise is a migraine trigger). Perhaps most important, practitioners and patients should have realistic expectations, since positive effects on migraines gained by each of the 3 therapies may be only modest in magnitude and scope, with only about half or fewer patients achieving more than a 25% reduction in migraine frequency.

REFERENCE: Varkey E, Cider A, Carlsson J, Linde M. Exercise as migraine prophylaxis: A randomized study using relaxation and topiramate as controls. Cephalalgia. 2011(Sep), online ahead of print [abstract here].