We assume that the outcome of headache management is the result of additive actions of (1) a specific effect on headache mechanisms, (2) a placebo effect linked to the idea of having received the verum, and (3) a non-specific psychological covert intervention (empathy, kind listening, etc…) which can be at its maximum in some complex therapeutic procedures as acupuncture, touch and manual therapy, biofeedback.
Therefore, to demonstrate a genuine placebo effect for an oral treatment, one should ideally use, at least, a three arm trial design (verum, placebo, and no treatment) [6
]. Differences between the verum and placebo reflect the specific effect. Differences between the placebo and no treatment measure the placebo effect. To analyze the complex procedure effect one should, also, control the non-specific psychological covert intervention by the mean of a pertinent “psychological” control group. Bias may come from a non-convincing sham procedure, or from the negative effect of being included in a “psychological”, or in a non-treated control group.
At this point, it appears interesting to clarify the relationships between placebo, non specific psychological intervention and psychotherapy [35
]. The three act through psychological processes. Placebo effect is simply mostly based on an expectation after the announcement of given therapy and after conditioning, the non-specific psychological intervention is limited to empathy, kind listening without psychological base, and psychotherapy acts mostly through specific intentionally delivered psychological interactions.
Acute treatment for headache attacks
No three arm trials have studied acute treatment [36
]. In migraine, meta-analysis [37
] of placebo arms show that in adult patients, at 2 h, a two point improvement (using a 0–3 scale) is seen in about 28–29% of patients and a pain-free state in 6–9% (verum 58% for improvement and of 29% for pain-free), however, with a high heterogeneity [40
]. These figures demonstrate the specific effect of the verum included in the meta-analysis, and their amplitude suggests that a genuine placebo effect does exist, mostly when a permissive outcome is chosen. A high placebo efficacy is reported in children: meta-analysis of 13 studies found that at 2 h after administration, improvement was seen in 33% (23–43%) of children, and a pain-free state in 14% (9–18%) [43
A meta-analysis of 37 studies about the treatment of tension type headache attacks shows that NSAIDs and acetaminophen have a significant specific effect [44
], but not data are available to analyze a putative placebo effect.
Prophylactic treatment and placebo effect
From a comprehensive meta-analysis of three arm trials [36
], only five studied headache exclusively [45
]; unfortunately, none of these studies considered the actual IHS classification. Only one three arm trial concerned oral medication, given 2 weeks, and did not evidence for a difference in headache score between the placebo an no-treatment [47
In a comprehensive meta-analysis of oral prophylactic treatments RCT of migraine (32 studies) [50
], the percentage of patients presenting a 50% reduction in the number of days with headache reported in the placebo arms is 21% (13–28%). There was a significant heterogeneity. The corresponding data for the verum was 41% (33–49%). The improvement under placebo was greater in parallel compared to cross-over studies and in European compared to North American trials. These studies were performed between 1998 and 2004, usually lasted 12 weeks and did not mention the level of control of the acute attacks. A recent follow-up during 16 months of migraine patients with an optimized attack treatment [51
] showed an improvement with time without differences between placebo alone, beta blockers alone, or placebo associated with behavioral management, suggesting that the administration of any treatment, even a placebo, is sufficient to achieve an apparent therapeutic success [51
In conclusion, a specific effect of the prophylactic oral treatments included in the meta-analysis seems to be demonstrated [52
] only for a relatively short time use. A prophylactic effect of placebo is also suggested [50
] for a short time period by the amplitude of the improvement observed in the placebo arm. Finally, one study suggests a placebo prophylactic efficacy for a long period [51
Meta-analysis of studies of tension type headache prophylaxis with oral treatment provide conflicting results: a lack of superiority of antidepressant medication or myorelaxants over placebo is reported in one study [53
], and a beneficial effect of tricyclic antidepressants in two others [54
]. No data are presented to evaluate a putative placebo effect. Interestingly, the follow-up of four groups of patients with chronic tension-type headache [54
] treated with anti-depressant medication or placebo with or without stress-management therapy, showed that the placebo had a non-different effect compared to the anti-depressant medication or stress-management therapy given alone on headache activity in the sub-group with initial low CTTH severity and on disability in the sub-group without initial mood and anxiety disorders [56
]. This part of the data raises the issue of the placebo efficacy in CTTH of low severity.
The effects of acupuncture in migraine prevention have been evaluated by one meta-analysis [57
]: true acupuncture was not superior to sham acupuncture, but is superior to no treatment up to 4 months after treatment (effect size 0.44 SD). In tension type headache, two meta-analysis [58
] revealed a small advantage of true acupuncture over sham acupuncture, in fact linked to one heavy positive study [60
]. One have to conclude to a lack of specific effect of acupuncture on migraine and to a questionable specific effect of this procedure on tension-type headache. Acupuncture seems to act mostly through a high placebo and non-specific psychological effect.
A cervical pain trial meta-analysis studying manual therapy [61
] found that manipulation (high velocity low amplitude) and mobilization produce similar effects on pain and are not better at short- and intermediate-term than controls for pain relief. Consequently, no specific and no significant placebo effects have been demonstrated. However, interpersonal touch has a major impact in our everyday social interactions [62
], and has been used as a therapy since the dawn of humanity. Touch therapy is more or less codified (healing touch, therapeutic touch, Reiki) and is consistently associated with a special surrounding that can be considered as having a “non-specific” psychological influence. A meta-analysis of studies conducted on touch therapy for pain [63
] includes only one inconclusive study [64
] on tension type headache. Therefore, no conclusion about the specific effect of touch on headache can be drawn.
An interesting three arm study [49
] on chronic headache sufferers treated by soft manual therapy with relaxation (Trager’s technique), controlled “attentional” visits, or no treatment shows a higher improvement of quality of life in the two treated groups compared to the no-treatment group. Consequently, no specific effect can be concluded from this open study. The improvement in the two groups may be due to the psychological non-specific effect.
Interestingly, the beneficial effect of sham acupuncture on headache has been proposed as a model of ritual healing by touch [65
], which provides one way to explain the powerful efficacy of sham acupuncture when compared to no-treatment.
A comprehensive efficacy review of biofeedback (BFB) [66
] concluded that true BFB is not significantly superior to sham BFB in migraine (effect size 0.25, confidence interval 95% 0.49–0.00) but did show a small advantage of true BIB over sham in tension-type headaches (effect size 0.50, confidence interval 95% 0.26–0.75). In both conditions, BFB is superior to the waiting list. Thus, BFB seems to have a specific beneficial effect on tension type headache. The superiority of pseudo BFB on the waiting list may be due to the additive effect of the non-specific psychological effect and the placebo effect of BFB.
In children, an interesting three arm study [48
] (warming BFB associated with cognitive stress management therapy, pseudo BFB associated with an attention therapy, waiting-list) did not find evidence for significant inter-treatment difference and cannot conclude to a specific effect.
A review of the studies of behavioral treatment of headache [67
] reports a 35–55% improvement but also emphasizes many methodological imperfections, including selection bias, credibility of the control procedure, and lack of reproducibility of the results. In addition, most of these studies were performed more than 30 years ago, and a control waiting list group was not reported. Among these studies, an interesting one [46
] compares four treatments for tension headache (relaxation, relaxation + cognitive therapy, pseudo-meditation, and waiting list) and reports a significantly better improvement of a headache index for the two groups treated with relaxation compared to the pseudo-meditation group. Pseudo-meditation consisted of an equal number of sessions in which subjects were engaged in imaging daily activity without becoming relaxed, and is therefore a control of the relaxation. This study provides evidence for a specific effect of relaxation on tension type headache prophylaxis.
As conclusions from this review on RCT
A specific effect of treatment has been demonstrated by meta-analysis in several situation: (1) oral treatment of migraine, and tension type headache attacks, (2) oral treatment for migraine prevention during usually a 12-week administration, regardless the level of control of the attack and the underlying anxio-depressive state. In tension type headache, a questionable specific effect is also reported for acupuncture and for BFB associated with relaxation, and, by one study for relaxation.
A placebo effect is likely associated with every kind of treatment. However the evidences are only indirect. (1) In migraine attack, the amplitude of the improvement in the placebo arms (about half of that observed in the verum arms, if we disregard the improvement due to the natural course) replaces a demonstration. However if we consider the harder outcome of pain free at 2 h, this placebo effect is only about a fifth of that of verum. (2) In oral prophylaxis of migraine, the meta-analysis of short-term RCT reports also an improvement half of that of verum for placebo-treated patients, which is also an indirect proof of a short-term genuine placebo effect. Interestingly one study suggest a long-term placebo effect in chronic tension type headache in patients with a moderate disability or with a low initial anxio-depressive level, and in migraine patient with an optimal attack control.
Both placebo and non specific psychological effect are likely at the origin of the improvement induced by many procedures (migraine prophylaxis by acupuncture or biofeedback, headache in general for manual therapy, touch and behavioral treatment) on the evidences that for these techniques in these precise conditions, the patents improve though no specific effect has ever been demonstrated. The non- specific psychological effects of these complex procedures in headache treatments refer to the “common factors” shared by the various modalities of psychotherapies (see review in [68