2012 AHS/AAN Guidelines for Prevention of Episodic Migraine
The 2012 AHS/AAN guidelines for episodic migraine prevention provide a welcome summary of the evidence that underpins commonly used treatments for migraine. Most of the drugs deemed to have the highest level of evidence in the 2000 guidelines remain in that category in 2012. Although the methods used to locate and appraise evidence and link it to recommendations varied among the 3 guidelines we reviewed, there was remarkable consistency in the ratings of drugs for first-line use.
In contrast, recommendations diverged substantially for gabapentin and feverfew. This divergence is potentially confusing for clinicians and patients. It may be related to differences in search strategies or methods for selecting the evidence. In our view, however, it is most likely due to the way in which recommendations were formulated. The AHS/AAN ratings were assigned solely on the basis of an assessment of efficacy, while the Canadian and EFNS categorizations sought to balance efficacy and side effects. In the case of gabapentin, which is widely believed to be well tolerated, it is therefore not surprising that the Canadian and EFNS guidelines place the drug in their second and third tiers, respectively, while the AHS/AAN guidelines downgrade the drug because of conflicting and poor-quality evidence of efficacy.
We believe this example points out a serious shortcoming in rating methods that seek to incorporate both benefits and harms. First, although willingness to use treatments is influenced by both benefits and harms, ratings assigned by others can never hope to correctly capture the views of different patients. Second, it is well known that potential harm and adverse events are not systematically sought or reported in clinical trials, so that published evidence of harms is likely to be an underestimate. And finally, it is questionable whether a drug with low or uncertain efficacy should be recommended for widespread use on the basis of tolerability when more effective drugs are available. The quality and credibility of clinical trial evidence for gabapentin has recently been called into considerable question. This evidence quality problem is better addressed by methods that use a purer approach in generating recommendations, one that is based principally on assessments of efficacy.
On the other hand, a pure efficacy-based system of recommendations raises the question of what constitutes a clinically meaningful treatment effect, especially in relation to side effects. Some drugs that have statistically significant evidence of benefit in well-designed and conducted trials may nonetheless provide very marginal benefits when applied to routine clinical practice. An example of this situation is frovatriptan, where the magnitude of clinical benefit is quite small in comparison with treatment burden and cost. One compromise might be to present information about both efficacy and side effects but to refrain from incorporating them in a composite measure. This allows clinicians using the guidelines to individualize treatment decisions based on both efficacy and patient preferences regarding specific risks.
The quality of the AHS/AAN and Canadian guidelines, as assessed with the AGREE-II tool, was better than that of guidelines in other specialties, which is heartening. Both are recommended for use on the basis of the AGREE-II quality assessment. It is likely that the Canadian guidelines scored particularly high on the AGREE measure because they were developed in accordance with AGREE recommendations. In contrast, the EFNS guidelines do not appear to meet widely accepted standards for guideline quality and are not recommended for use.
Future efforts are needed to ensure that guidelines are regularly updated and that guideline developers make use of methods to locate and incorporate unpublished clinical trial evidence. There are many reasons that it can be difficult to locate clinical trial evidence. Some have to do with problems in tagging studies as RCTs in MEDLINE or the need to condense information to meet word limits imposed by medical journals. It is clear, however, that much clinical trial evidence has never been published or has been incompletely reported. Unfortunately, there is substantial reason to believe that this problem of missing or manipulated evidence affects the clinical evidence when one evaluates migraine therapy. The incorporation of unpublished evidence into meta-analyses has been shown to alter conclusions about treatment efficacy. To ensure the integrity, validity, and credibility of migraine clinical practice guidelines, developers should make strenuous efforts to locate all relevant evidence. This and other recommendations for the development of future migraine clinical practice guidelines are listed in Box 1.
In summary, the 2012 updated AHS/AAN guidelines for preventive treatment of episodic migraine provide a welcome and comprehensive overview of the breadth and quality of existing evidence. They confirm the benefits of many widely used therapies, identify drugs that should be avoided, and remind clinicians of emerging evidence for a wide array of newer treatment choices.
Discussion
The 2012 AHS/AAN guidelines for episodic migraine prevention provide a welcome summary of the evidence that underpins commonly used treatments for migraine. Most of the drugs deemed to have the highest level of evidence in the 2000 guidelines remain in that category in 2012. Although the methods used to locate and appraise evidence and link it to recommendations varied among the 3 guidelines we reviewed, there was remarkable consistency in the ratings of drugs for first-line use.
In contrast, recommendations diverged substantially for gabapentin and feverfew. This divergence is potentially confusing for clinicians and patients. It may be related to differences in search strategies or methods for selecting the evidence. In our view, however, it is most likely due to the way in which recommendations were formulated. The AHS/AAN ratings were assigned solely on the basis of an assessment of efficacy, while the Canadian and EFNS categorizations sought to balance efficacy and side effects. In the case of gabapentin, which is widely believed to be well tolerated, it is therefore not surprising that the Canadian and EFNS guidelines place the drug in their second and third tiers, respectively, while the AHS/AAN guidelines downgrade the drug because of conflicting and poor-quality evidence of efficacy.
We believe this example points out a serious shortcoming in rating methods that seek to incorporate both benefits and harms. First, although willingness to use treatments is influenced by both benefits and harms, ratings assigned by others can never hope to correctly capture the views of different patients. Second, it is well known that potential harm and adverse events are not systematically sought or reported in clinical trials, so that published evidence of harms is likely to be an underestimate. And finally, it is questionable whether a drug with low or uncertain efficacy should be recommended for widespread use on the basis of tolerability when more effective drugs are available. The quality and credibility of clinical trial evidence for gabapentin has recently been called into considerable question. This evidence quality problem is better addressed by methods that use a purer approach in generating recommendations, one that is based principally on assessments of efficacy.
On the other hand, a pure efficacy-based system of recommendations raises the question of what constitutes a clinically meaningful treatment effect, especially in relation to side effects. Some drugs that have statistically significant evidence of benefit in well-designed and conducted trials may nonetheless provide very marginal benefits when applied to routine clinical practice. An example of this situation is frovatriptan, where the magnitude of clinical benefit is quite small in comparison with treatment burden and cost. One compromise might be to present information about both efficacy and side effects but to refrain from incorporating them in a composite measure. This allows clinicians using the guidelines to individualize treatment decisions based on both efficacy and patient preferences regarding specific risks.
The quality of the AHS/AAN and Canadian guidelines, as assessed with the AGREE-II tool, was better than that of guidelines in other specialties, which is heartening. Both are recommended for use on the basis of the AGREE-II quality assessment. It is likely that the Canadian guidelines scored particularly high on the AGREE measure because they were developed in accordance with AGREE recommendations. In contrast, the EFNS guidelines do not appear to meet widely accepted standards for guideline quality and are not recommended for use.
Future efforts are needed to ensure that guidelines are regularly updated and that guideline developers make use of methods to locate and incorporate unpublished clinical trial evidence. There are many reasons that it can be difficult to locate clinical trial evidence. Some have to do with problems in tagging studies as RCTs in MEDLINE or the need to condense information to meet word limits imposed by medical journals. It is clear, however, that much clinical trial evidence has never been published or has been incompletely reported. Unfortunately, there is substantial reason to believe that this problem of missing or manipulated evidence affects the clinical evidence when one evaluates migraine therapy. The incorporation of unpublished evidence into meta-analyses has been shown to alter conclusions about treatment efficacy. To ensure the integrity, validity, and credibility of migraine clinical practice guidelines, developers should make strenuous efforts to locate all relevant evidence. This and other recommendations for the development of future migraine clinical practice guidelines are listed in Box 1.
In summary, the 2012 updated AHS/AAN guidelines for preventive treatment of episodic migraine provide a welcome and comprehensive overview of the breadth and quality of existing evidence. They confirm the benefits of many widely used therapies, identify drugs that should be avoided, and remind clinicians of emerging evidence for a wide array of newer treatment choices.
SHARE
