Mary Dimmock has led the way on this issue, and is co-author of this post.
In response to requests by U.S. patient organizations and advocates, the U.S. Agency for Healthcare Research and Quality (AHRQ) has issued an Addendum to its 2014 ME/CFS evidence review. This Addendum downgrades the conclusions on the effectiveness of cognitive behavioral therapy (CBT) and graded exercise therapy (GET), and this has tremendous implications for medical education and treatment recommendations.
At Long Last, AHRQ Undertakes Reanalysis
The 2014 AHRQ evidence review had originally reported that treatment with CBT and GET improved fatigue, function, global improvements, and employment in ME/CFS patients. This conclusion was based on an analysis that combined studies using any of seven CFS and ME definitions as though they represented the same disease. We highlighted this fatal flaw in our comments on the draft evidence review. The review itself acknowledged that CFS definitions did not require hallmark symptoms of the disease, that no treatment studies used ME/CFS or ME definitions, and that the Oxford definition was so broad that it would include patients with other diseases and should be retired. The 2015 NIH Pathways to Prevention report concurred that Oxford should be retired, stating that it could “impair progress and cause harm.” And yet, the 2014 AHRQ evidence review maintained its conclusion that CBT and GET could benefit some “ME/CFS” patients, based in substantial part on studies that used the Oxford definition.
Following the publication of David Tuller’s PACE analysis in October 2015, ME organizations and advocates called on AHRQ to reanalyze the PACE trial, to analyze the results of Oxford studies separate from studies using other definitions, and to analyze CBT studies separately from other forms of counseling. Following requests in November 2015 and again in February 2016, AHRQ agreed to ask the authors to reanalyze studies by definition and to separate out CBT from other counseling studies. They declined to reanalyze the PACE trial, which they had ranked a good trial with “undetected” reporting bias.
This new analysis was published in late July as an Addendum to the original report (see pages 1-16 of the linked report). This updated review reanalyzed the effectiveness of GET, CBT, and other forms of counseling, first with Oxford studies included and then after Oxford studies were excluded. As we predicted, breaking out the Oxford studies forced a significant downgrade in the review’s conclusions.
GET is Ineffective for ME Patients
The 2014 evidence review had combined four GET studies and two other exercise studies (qigong and orthostatic training) in its analysis and concluded, “GET improved measures of fatigue, function, and clinical global impression of change compared with controls.”
The Addendum reanalyzed just the four GET studies, three of which used the Oxford definition. When all four GET studies were analyzed together, the Addendum reported a moderate strength of evidence of improved function and global improvement and low strength of evidence of improved fatigue and employment. But when Oxford studies were excluded from the analysis, the Addendum reported that there was insufficient evidence of effectiveness of GET on any outcome.
|Study Outcome||Results with Oxford Studies||Results Excluding Oxford Studies|
|Function||Moderate strength of evidence||Insufficient evidence|
|Fatigue||Low strength of evidence||Insufficient evidence|
|Quality of Life||Insufficient evidence||Insufficient evidence|
|Employment||Low strength of evidence||Insufficient evidence|
|Global Improvement||Moderate strength of evidence||Insufficient evidence|
CBT Is Barely Effective
The 2014 evidence review combined CBT and other counseling therapies in its analysis and concluded that collectively, these therapies improved fatigue, function, quality of life, and global improvement. Acknowledging that “CBT is a unique approach with disputable underlying rationale regarding the fear avoidance theory contributing to the perpetuation of symptoms in ME/CFS,” the Addendum reanalyzed the seven CBT studies separately from the other counseling studies. The Addendum also analyzed the Oxford CBT studies separately from the Fukuda CBT studies, although one study (Deale 1997, 2001) appears to have been incorrectly classified.
When all seven CBT studies were reanalyzed, the Addendum reported low strength of evidence for improved function, fatigue, and global improvement; insufficient evidence of improvement in employment; and low strength of evidence that CBT did not improve quality of life. When Oxford studies were excluded from the reanalysis, the Addendum found insufficient evidence of effectiveness of CBT on function, employment and global improvement and a low strength of evidence of improved fatigue.
|Study Outcome||Results with Oxford Studies||Results Excluding Oxford Studies|
|Function||Low strength of evidence of improvement||Insufficient evidence|
|Fatigue||Low strength of evidence of improvement||Low strength of evidence of improvement|
|Quality of Life||Low strength of evidence that CBT does not impact quality of life||Low strength of evidence that CBT does not impact quality of life|
|Employment||Insufficient evidence||Insufficient evidence|
|Global Improvement||Low strength of evidence of improvement||Insufficient evidence|
Implications and What Next
The Addendum reaches a damning conclusion, cloaked in dry language:
This addendum has delineated differences in treatment effectiveness and harms according to case definitions, highlighting studies that used the Oxford case definition and how these studies impacted our conclusions. Additionally, results of studies evaluating CBT have been considered independently from other counseling and behavioral therapies. Our sensitivity analysis would result in a downgrading of our strength of evidence on several outcomes which can be attributed to the decrease in power, dominance of one large trial, or lack of trials using criteria other than the Oxford case definition for inclusion. Blatantly missing from this body of literature are trials evaluating effectiveness of interventions in the treatment of individuals meeting case definitions for ME or ME/CFS.
In other words, there is no evidence that supports the conclusion that CBT and GET are effective treatments or ME or ME/CFS patients.
And THAT is a ground breaking conclusion.
CBT and GET still remain the most common treatment recommendations across mainstream “evidence-based” medical education today. Recommendations for CBT and GET have been based directly on Oxford studies like PACE, or indirectly through evidence reviews such as AHRQ and Cochrane that have relied heavily on Oxford studies. At the same time, these treatment recommendations have ignored numerous patient surveys that have reported harms from CBT and GET. The resultant medical education, including that from CDC and key medical education providers such as UpToDate, confuses medical providers on the nature of the disease and puts patients at significant risk of harm from inappropriate treatment.
Recommendations for CBT and GET have become so pervasive that they are also found in the educational plans of medical societies. The American Academy of Family Physicians issued a 2016 needs assessment for education of members that specified that CBT and GET “improved fatigue, work and social adjustment, anxiety, and postexertional malaise” and called out the need for more training of doctors in the use of exercise therapies.
The problem, as this reanalysis found, is that when you remove the Oxford studies, the positive effect of CBT and GET disappears. The Addendum notes that “using the Oxford case definition results in a high risk of including patients who may have an alternate fatiguing illness or whose illness resolves spontaneously with time.” CBT and GET appears to help these patients to a limited extent. However, the recommendation of either of these therapies for ME and ME/CFS patients is inappropriate.
This Addendum highlights a key issue that has perverted medical education and evidence reviews in this field for a long time. It is medically inappropriate to make recommendations for disease treatment based on studies in patients who do not have the disease. CDC, UpToDate, AAFP, and other medical education sources must change their recommendations for ME/CFS treatments. Furthermore, NIH must address the gap in treatment trials for ME and ME/CFS patients by funding intervention studies on these patients.
There is no evidence that CBT and GET are effective treatments for us, and therefore, these treatments can no longer be recommended. If CDC and others persist in recommending treatments for which there is no evidence of effectiveness in ME/CFS patients, it will not only perpetuate confusion but also put patients at risk. Such an unscientific recommendation goes against the principles of evidence-based medicine and is not accepted in other diseases. It will not be tolerated here.
Edited 8/24/16 to add: Some medical education sites may be using these terms to refer to some other form of talk therapy and “exercise,” not PACE style CBT and GET. But the ambiguity in what is actually being recommended and its expected benefit creates medical misunderstanding and a risk of harm for patients, especially given the lack of published evidence for these other approaches and the volume of evidence for PACE style CBT and GET. Recommendations for talk therapy and any kind of “exercise” or activity management must be precise in terms of the recommended approach, the expected impact and benefit, the potential harm, and any contraindications, particularly for severely ill patients.