Nocebo response in human models of migraine: A systematic review and meta-analysis of randomized,double-blind,placebo-controlled,two-way crossover trials in migraine without aura and healthy volunteers

Abstract

Background

Human models of migraine have been used for the past 30 years to test putative ‘trigger’ molecules and ascertain whether they induce migraine attacks in humans. However, nocebo effects using this model have never been systematically explored.

Objective

To assess the nocebo response rate in randomised clinical trials conducted at the Danish Headache Center, and in which human models of migraine were used.

Methods

In this systematic review and meta-analysis, we searched PubMed for studies of human models of migraine with a randomised, double-blind, placebo-controlled, two-way crossover design that included data on the incidence of migraine attacks or headache after infusion of placebo. A total of 943 articles were screened by title and abstract. Of these, 27 studies met the inclusion criteria (published between 1994 and 2020) and were included in the qualitative and quantitative analysis. We performed a random effects meta-analysis for the incidence of migraine attacks or delayed headache after placebo infusion.

Results

Twenty-seven studies were eligible for inclusion: 12 studies reported data for adults with migraine (n = 182), whereas 16 studies reported data on healthy volunteers (n = 210). For adults with migraine, the incidence of migraine attacks after placebo was 8.1% (95% CI = 2.5–15.5%, I² = 50.8%). The incidence of delayed headache was 25.9% (95% CI = 18.5–34.1%, I² = 18.9%). For healthy volunteers, the incidence of migraine attacks after placebo was 0.5% (95% CI = 0.0–3.6%, I² = 0.0%) while the incidence of delayed headache was 10.5% (95% CI = 4.8–17.6%, I² = 45.2%).

Conclusion

The nocebo response in randomised, placebo-controlled two-way crossover trials with intravenous infusions of placebo in migraine is negligible. Future studies using human models of migraine can be conducted by assuming a nocebo response rate of 15.5%.

Keywords

Human models of migraine headache provocation experimental headache clinical trial pain

Introduction

Human models of migraine have been used for decades to study signaling pathways responsible for the genesis of migraine pain (1). The rationale is to investigate whether endogenous signaling molecules or other putative ‘trigger’ molecules can induce migraine attacks in humans (2,3). Interest in these ‘trigger’ molecules have directly facilitated the development of mechanism-based therapies, targeting calcitonin gene-related peptide (CGRP) or its receptor (3,4).

A randomised, double-blind, placebo-controlled, two-way crossover design has been the gold standard for human models of migraine (2). Participants are randomly allocated to receive a putative ‘trigger’ molecule or placebo (e.g. isotonic saline). A unique feature of these models is that only individuals with migraine develop “triggered” attacks, while no more than mild headache is reported by healthy volunteers (2). Numerous studies have been carried out with this design and it is, therefore, timely to ascertain the nocebo response rate in individuals with migraine without aura and healthy volunteers. Such efforts may improve the design of future human models of migraine and contextualise findings from previously published studies. Here, we provide a systematic review and meta-analysis of the nocebo response rate in randomised clinical trials (RCTs) conducted at the Danish Headache Center, and in which human models of migraine were used.

Methods

Data sources

No protocol was registered for this study. A systematic search was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guideline. The PubMed database was searched for article on human models of migraine on June 25, 2020, with the following search string and no restrictions in terms of date of publication: “Migraine AND Denmark”. We also searched ClinicalTrials.gov for published, unpublished, and ongoing RCTs. The search features were “Migraine”, “Danish Headache Center”, and “Denmark”. If available, we obtained missing or unpublished data by contacting the relevant study authors.

Eligibility criteria

The PICOS (population, intervention, comparison, outcome and study design) structured approach was used in this systematic review. The population of interest was adult individuals with migraine without aura and healthy volunteers. The intervention of interest was intravenous infusion of placebo (isotonic saline) and an infusion time of less than 120 minutes. The search was limited to studies conducted at the Danish Headache Center. Reference lists of identified primary articles were reviewed to identify studies that may have been missed in the search process. Studies were deemed eligible if they included data from randomised, double-blind, placebo-controlled, two-way crossover trials. Furthermore, we only included studies in which the study population of interest were adults with migraine without aura according to any edition of the International Classification of Headache Disorders (5 –7) or healthy volunteers. The key outcomes of interest were the incidence of migraine attacks after intravenous infusion with placebo (isotonic saline) in individuals with migraine without aura and healthy volunteers.

Study selection and data extraction

Two investigators (HG and AI) independently screened all articles by title and abstract. Articles deemed relevant were selected for full-text review to determine eligibility for inclusion. Data was independently extracted by two investigators (HG and AI) and cross-checked by a third investigator (TPD) to ensure the integrity of the contents. Discrepancies were resolved by discussion between the three investigators (HG, AI, and TPD). For each of the included studies, the investigators (HG, AI, and TPD) extracted the following data: Age, gender, experimental criteria for migraine attacks, personal and/or family history of headache disorders, medication history, headache prior to the experimental day, incidence of migraine attacks and delayed headache in after placebo in individuals with migraine without aura or healthy volunteers, and incidence of delayed headache after placebo in healthy volunteers.

Statistical analysis

We performed a random-effects meta-analysis for the primary outcome (incidence of migraine attacks after intravenous infusion with placebo (isotonic saline)) and secondary outcome (incidence of delayed headache after placebo) in both individuals with migraine without aura and healthy volunteers. Both outcomes were defined according to the authors’ case definition. Effect sizes were calculated using the inverse variance method on data transformed with Freeman-Tukey Double arcsine transformation while tau² (the between-study variance) was calculated with the Restricted Maximum Likelihood method. The I² statistic was used to assess between-study heterogeneity, with values ≥75% indicating considerable heterogeneity. Effect sizes are presented as the incidence (%) and the 95% confidence interval. Statistical analyses were performed using Microsoft Excel 2016 (Microsoft Corp) and R (version 3.6.3) with the “meta” (version 4.11) and “metaphor” (version 2.1) packages.

Risk of bias assessment

The risk of bias in the included studies was independently assessed by two investigators in accordance with the Cochrane Handbook for Systematic Reviews of Interventions (8). The potential risk of systemic error was categorised as low, high or unclear. The following seven factors, selection bias, performance bias, reporting bias, detection bias, attrition bias and ‘other’ sources of bias, were assessed.

Results

The database search identified 937 citations. An additional six articles were identified through a manual search of identified primary articles and the database search on ClinicalTrials.gov. A total of 943 articles were screened by title and abstract. Of these, 27 studies met the inclusion criteria (1994 and 2020) and were included in the qualitative and quantitative analysis (9 –35) (Figure 1). Twelve studies (10 published and two studies under review) reported data on the incidence of migraine attacks after placebo in individuals with migraine without aura (9 –17,31,33), whereas 10 studies did so in healthy volunteers (18,19,22,23,25,27,29,30,32,34). Furthermore, 16 studies reported data on the incidence of delayed headache after placebo in healthy volunteers (13,18–30,32,34). A summary of the included studies is presented in Table 1 and Table 2.

Table 1.

Summary of nocebo responses in trials with individuals with migraine.

Author (year), study drug, infusion time	Mean age (range)*	Female participants (%)*	Mean migraine frequency (range)	Medication	Headache prior to experimental day	Criteria for experimental migraine attack	Migraine after placebo	Incidence (95% CI)	Delayed headache after placebo⁺	Incidence (95% CI)
Thomsen et al. (1994), glyceryl trinitrate, 20 min	36.3 years (range: 21–51 years)	8/12 (66.7%)	21 attacks per year (range: 6–76 attacks per year)	- No daily medication apart from oral contraceptives - No medication intake ≥12 h prior - No triptans intake≥24 h prior - No ergotamine intake ≥48 h prior	†	Migraine without aura, (ICHD 1)	1/11	9.1% (0.0–26.1%)	1/11	9.1% (0.0–26.1%)
Lassen et al. (2002), calcitonin gene-related peptide, 20 min	39.5 years (range: 31–47 years)	11/12 (91.7%)	6–72 attacks per year (†)	- No daily medication apart from oral contraceptives - No medication intake ≥12 h prior - No triptan intake≥24 h prior - No ergotamine intake ≥48 h prior	- No headache on arrival	Migraine without aura, (ICHD 1)	0/9	0.0% (0.0–0.0%)	1/9	11.1% (0.0–31.6%)
Rahmann et al. (2008), vasoactive intestinal polypeptide, 25 min	31 years (range: 22–40 years)	12/14 (85.7%)	1–6 attacks per 8 weeks (†)	- No daily medication apart from oral contraceptives - No medication intake within 4 plasma half-lives	- No headache≥24 h prior	†	1/12	8.3% (0.0–24.0%)	2/12	16.7% (0.0–37.8%)
Petersen et al. (2009), adrenomedullin, 20 min	41 years (range: 28–50 years)	10/12 (83.3%)	≤6 attacks per month (†)	- No intake of concomitant medication that might influence outcome measures - No triptan intake≥48 h prior - No ergotamine intake ≥72 h prior	- No migraine attack ≥48 h prior - No tension-type headache≥48 h prior	†	4/12	33.3% (6.7–60.0%)	4/12	33.3% (6.7–60.0%)
Schytz et al. (2009), carbachol, 25 min	31 years (range: 19–44 years)	18/18 (100%)	25 attacks per year (range: 6–54 attacks per year)	†	- No headache on arrival - No migraine attack ≥5 days prior	Migraine attack, fulfilling either 1 or 2: 1. Headache fulfilling criteria C and D for migraine without aura (ICHD II) 2. Headache described as mimicking usual migraine attack and treated with a triptan	6/18	33.3% (0.0–55.1%)	7/18	38.9% 16.4–61.4%
Schytz et al. (2009), pituitary adenylate cyclase-activating polypeptide-38, 20 min	31 years (range: 20–47 years)	11/12 (91.7%)	1–4 attacks per month (†)	†	- No headache on arrival - No migraine attack ≥5 days prior - No tension-type headache≥48 h prior	Migraine attack, fulfilling either 1 or 2: 1. Headache fulfilling criteria C and D for migraine without aura (ICHD II) 2. Headache described as mimicking usual migraine attack and treated with a triptan	0/12	0.0% (0.0–0.0%)	3/11	27.3% (1.0–53.6%)
Wienecke et al. (2009), prostaglandin I₂, 25min	35.3 years (range: 20–53 years)	8/12 (66.7%)	†	- No daily medication apart from oral contraceptives	- No migraine attack ≥5 days prior - No tension-type headache ≥5 days prior	Migraine attack, fulfilling either 1 or 2: 1. Headache fulfilling criteria C and D for migraine without aura (ICHD II) 2. Headache described as mimicking usual migraine attack and treated with a triptan	2/12	16.7% (0.0–37.8%)	3/12	25.0% (0.5–49.5%)
Antonova et al. (2012), prostaglandin E₂, 25 min	29.8 years (range: 18–52 years)	9/12 (75%)	†	- No daily medication apart from oral contraceptives	No headache≥24 h prior No migraine attack ≥48 h prior	Migraine attack, fulfilling either 1 or 2: 1. Headache fulfilling criteria C and D for migraine without aura (ICHD II) 2. Headache described as mimicking usual migraine attack and treated with a triptan	0/12	0.0% (0.0– 0.0%)	†	†
Al-Karagholi et al. (2019), levcromakalim, 20 min	28 years (range: 18–49 years)	13/16 (81.3%)	1–5 attacks per month (†)	- No daily medication apart from oral contraceptives	- No headache≥48 h prior	Migraine attack, fulfilling either 1 or 2: 1. Headache fulfilling criteria C and D for migraine without aura (ICHD II) 2. Headache described as mimicking usual migraine attack and treated with a triptan	1/16	6.3% (0.0– 18.1%)	7/16	43.8% (19.4–68.1%)
Ghanizada et al. (2019), pituitary adenylate cyclase-activating polypeptide-27, 20 min	29 years (range: 18–52 years)	15/20 (75%)	≥2 attacks per month (3.55 range 2–6 attacks per month)	- No daily medication apart from oral contraceptives	- No headache≥48 h prior - No migraine attack ≥5 days prior	Migraine attack, fulfilling either 1 or 2: 1. Headache fulfilling criteria C and D for migraine without aura (ICHD II) 2. Headache described as mimicking usual migraine attack and treated with a triptan	2/20	10.0% (0.0–23.1%)	4/20	20.0% (2.5–37.5%)
Al-Karagholi et al. 2020, (under peer review), 20 min	23 years (range: 18–37 years)	19/22 (86,4%)	1–5 attacks per month (†)	- No daily medication apart from oral contraceptives	- No headache≥48 h prior - No migraine attack ≥5 days prior	Migraine attack, fulfilling either 1 or 2: 1. Headache fulfilling criteria C and D for migraine without aura (ICHD II) 2. Headache described as mimicking usual migraine attack and treated with a triptan	0/22	0.0% (0.0–0.0%)	5/22	22.7% (5.2–40.2%)
Ghanizada et al. 2020, (under peer- review), 20 min	25 years (range: 19–38 years)	18/20 (90%)	≥2 attacks per month (3.3 range 2–5 attacks per month)	- No daily medication apart from oral contraceptives	- No headache≥48 h prior - No migraine attack ≥3 days prior	Migraine attack, fulfilling either 1 or 2: 1. Headache fulfilling criteria C and D for migraine without aura (ICHD II) 2. Headache described as mimicking usual migraine attack and treated with a triptan	3/20	15.0% (0.0–30.6%)	9/20	45.0% (23.2–66.8%)

*Baseline data; †Not reported; ⁺Any headache in post-hospital phase considered as delayed headache.

Table 2.

Summary of nocebo responses in trials with subjects with healthy volunteers.

Author (year), study drug, infusion time	Mean age (range)*	Female participants (%)*	Headache history	Family history of primary headache	Medication history	Headache prior to experimental day	Migraine after placebo	Incidence (95% CI)	Delayed headache after placebo⁺	Incidence (95% CI)
Ashina et al. (2000), glyceryl trinitrate, 20 min	37 years (23–52 years)	13/16 (81.3%)	- No history of migraine - < 12 headache days per year	†	- No daily medication apart from oral contraceptives - No analgesics ≥24 hours prior	- No headache ≥12 hours prior	0/16	0% (0–0%)	0/16	0% (0–0%)
Kruuse et al (2000), pentoxifylline, 60 min	24.5 years (21–43)	7/10 (70%)	- No history of migraine	†	†	†	†	NA	0/10	0% (0–0%)
Birk et al. (2005), adenosine, 20 min	25¹ (22–28)	6/12 (50%)	- No history migraine	No history of close relatives with migraine or tension-type headache > 1 day per week	- No daily medication apart from oral contraceptives - No medication intake within 4 plasma half-lives	- No headache on arrival	0/12	0% (0–0%)	0/12	0% (0–0%)
Hansen et al. (2006), vasoactive intestinal polypeptide, 25 min	24 years (22–28 years)	7/14 (50%)	No history of migraine or any other type of headache (except for episodic tension-type headache less than once per month)	†	- No daily medication apart from oral contraceptives	- No headache on arrival	0/14	0% (0–0%)	2/14	14.3% (0% - 32.6%)
Hansen et al. (2007), glyceryl trinitrate,	25 years (23–28 years)	5/12 (41.7%)	No history of migraine or any other type of headache (episodic tension-type headache no more than once a month)	†	- No daily medication apart from oral contraceptives	- No headache≥12 h prior	0/12	0% (0–0%)	4/12	33.3% (6.7–60.0%)
Wienecke et al. (2008), prostaglandin I₂, 25 min	24.4 years (19–40 years)	6/12 (50%)	No history of migraine or any other type of headache (episodic tension-type headache no more than once a month)	†	No daily medication apart from oral contraceptives	- No headache on arrival	0/11	0% (0–0%)	0/11	0% (0–0%)
Lindholt et al. (2009), norepinephrine, 90 min	24 years (20–29)	7/10 (70%)	No history of migraine or any other type of headache (episodic tension-type headache no more than once a week)	†	†	†	†	NA	3/10	30.0% (1.6–58.4%)
Schytz et al. (2009), carbachol, 25 min	23 years (20–29 years)	7/12 (58.3%)	No history of migraine or any other type of headache (except tension-type headache < 5 days/month)	†	- No daily medication apart from oral contraceptives	- No headache on arrival	0/11	NA	1/11	18.2% (0–41.0%)
Schytz et al. (2009), pituitary adenylate cyclase-activating polypeptide-38, 20 min	25 years (20–31 years)	7/12 (58.3%)	No history of migraine or any other type of headache (except tension-type headache < 5 days/month)	†	- No daily medication apart from oral contraceptives	- No headache on arrival - No tension-type headache ≥2 days prior - No migraine attack ≥5 days prior	0/12	0% (0–0%)	2/12	16.7% (0–37.8%)
Wienecke et al. (2009), prostaglandin E₂, 25 min	22.1 years (19–26 years)	6/14 (42.9%)	No history of migraine or any other type of headache (episodic tension-type headache no more than once a month)	1/14 subjects had a family history of migraine	No daily medication apart from oral contraceptives	- No headache on arrival	†	NA	3/11	27.3% (1.0–53.6%)
Antonova et al. (2011), prostaglandin F_2α, 20 min	24.4 years (19–33 years)	7/12 (58.3%)	No history of migraine or any other type of headache (episodic tension-type headache no more than once a month)	†	†	- No headache on arrival	†	NA	2/12	16.7% (0.0–37.8%)
Wienecke et al. (2011), prostaglandin D₂, 25 min	24 years (18–33 years)	6/12 (50%)	No history of migraine or any other type of headache (episodic tension-type headache no more than once a month)	2/12 subjects had a family history of migraine	No daily medication apart from oral contraceptives	- No headache on arrival	†	NA	1/12	8.3% (0.0–24.0%)
Arngrim et al. (2014), acetazolamide, 2 min	25 years (19–33 years)	16/16 (100%)	No history of migraine or any other type of headache (except tension-type headache < 5 days/month)	No first-degree relative with migraine	No daily medication apart from oral contraceptives	- No headache ≥48 h prior	1/12	8.3% (0.0–24.0%)	4/12	33.3% (6.7–60%)
Hougaard et al. (2019), endothelin, 20 min	26.1 years (†)	8/14 (57.1%)	No history of any primary headache disorders except episodic tension-type headache < 2 days per month during the last year	No first-degree family relative with migraine	No daily medication apart from oral contraceptives	- No headache ≥72 h prior	†	NA	1/13	7.7% (0.0–22.2%)
Pellesi et al. (2020), vasoactive intestinal polypeptide, 120min	25.8 years (23–29)	6/12 (50%)	No history of any primary headache disorders except episodic tension-type headache≤ 1 days per month	No first-degree family relative with migraine or other headache	No daily medication apart from oral contraceptives	- No headache ≥48 h prior	0/12	0% (0–0%)	0/12	0% (0–0%)
Al-Karagholi et al. (2020), maxipost, 20min	24 years (18–39)	12/20 (60%)	No history of any primary headache disorders except episodic tension-type headache≤ 1 days per month	^†	No daily medication apart from oral contraceptives	- No headache ≥48 h prior	1/20	5.0% (0.0–14.6%)	6/20	30.0% (9.9–50.1%)

*Baseline data; ¹Median; †Not reported; ⁺Any headache in post-hospital phase considered as delayed headache.

Migraine attacks after placebo in individuals with migraine without aura

Twelve studies were included in the primary analysis, comprising a total of 182 adults with migraine without aura (9 –17,31,33). Age ranged from 18–53 years while the proportion of female participants ranged from 67–100%. The random effects meta-analysis showed an incidence of 8.1% (95% CI = 2.5–15.5%, I² = 50.8%). In terms of methodology and design, all studies required that participants were free of headache upon arrival on the experimental day and had no daily intake of medications other than oral contraceptives. Furthermore, all studies used randomisation with 1:1 allocation and a wash-out period of at least 7 days between the two experimental days. Carry-over and/or period effects were assessed in seven of 12 included studies, in which none were found (12,13,15–17,31,32) (Table 1) (Figures 2, 3).

Delayed headache after placebo in individuals with migraine without aura

Eleven studies were included in the primary analysis, comprising a total of 163 adults with migraine without aura (9 –17,31,33). The random effects meta-analysis showed that the incidence of delayed headache after placebo was 25.9% (95% CI = 18.5–34.1%, I² = 18.9%). The applied methodology and design were similar in the studies, in which the incidence of migraine attacks had been investigated in adults with migraine without aura (Table 1, Figures 4, 5).

Migraine attacks after placebo in healthy volunteers

Eleven studies were included in the primary analysis, comprising a total of 138 healthy volunteers (13,18,19,22,23,25,27,29,30,32,34). The random effects meta-analysis showed that the incidence of migraine attacks after placebo was 0.5% (95% CI = 0.0–3.8%, I² = 0.0%). In terms of methodology and design, all studies required that participants were free of headache upon arrival on the experimental day and had no personal history of primary headache disorder, except episodic tension-type headache less than 5 days per month. In addition, all studies used a 1:1 randomisation and 7-day wash-out period (13,18,19,22,23,25,27,29,30,32,34) (Table 2) (Figures 4, 5).

Delayed headache after placebo in healthy volunteers

Seventeen studies were included in the primary analysis, comprising a total of 210 healthy volunteers (13,18–30,32,34). The random effects meta-analysis showed that the incidence of delayed headache after placebo was 10.5% (95% CI = 4.8–17.6%, I² = 45.2%). The applied methodology and design were similar to the aforementioned studies, in which the incidence of migraine attacks had been investigated in healthy volunteers. Carry-over and/or period effects were assessed in nine of 17 included studies, in which none were found (13,19–21,23,24,27) (Table 2) (Figures 4, 5).

Discussion

This is the first systematic review and meta-analysis to ascertain the incidence of migraine attacks and delayed headache after intravenous infusion of placebo. In individuals with migraine without aura, the incidence of migraine attacks was 8.1% after placebo, whereas the corresponding incidence for delayed headache was 25.9%. In healthy volunteers, the incidence of migraine attacks was 1.5%, whereas the corresponding incidence for delayed headache was 10.5%.

Our data demonstrate that nocebo effect is a confounding factor in delayed incidence of headache in human migraine model. We found that nocebo effect is more noticeable in individuals with migraine compared to healthy volunteers. Interestingly, the incidence of migraine attacks remains low after infusion of placebo. Our study shows that nocebo effect clearly lacks migraine-inducing properties. Furthermore, the magnitude of nocebo response differs across the studies reviewed in this study, thus showing variation in the ways that participants respond to nocebo effect, and with proper methodological strategies these effects can be mitigated. Recently, the effect of nocebo on generation of headache pain was reviewed (35). This study shows that psychosocial factors might play an important role in nocebo response (35). Here, we provided the first empirical evidence that nocebo affects individuals with migraine. Greater attention is needed to reduce the magnitude of nocebo in the experimental setups and clinical practice in individuals with migraine.

Furthermore, an important question raised by our findings is whether migraine attacks reported after placebo can be attributed to methodological limitations or simply the probability of developing a spontaneous attack within a 12-h observational period. Human models of migraine use a randomised, double-blind, placebo-controlled, two-way crossover design (2). This approach eliminates the between-subject variability and is the best suited design to assess short-lived effects of an intervention (36). A well-described limitation of crossover designs is, however, the potential influence of carry-over and period effects. In this context, it should be highlighted that such effects have never been reported in human models of migraine despite consistent testing in most of the included studies (13,19–21,23,24,27). This could be attributed to the standardised use of a 7-day wash-out period, but a remaining unknown is then whether a shorter wash-out period might suffice. Furthermore, we systematically assessed the risk of bias in the included studies. Unclear reporting of blinding of participants/personnel and allocation concealment were the main reasons for unclear risk of bias. However, it seems that proper and effective blinding has been carried out, but the authors did not properly document it in the method section. No high risk of bias was detected in the studies. The overall risk of bias in the studies remained low. In this review, only studies with intravenous administration of substances in migraine without aura and healthy volunteers were included. All participants were informed and gave their informed consent to receive an active and a placebo infusion. However, a hidden (infusion pump) design was used for infusion of substances in the cross-over design. Interestingly, previous studies showed that a hidden design might minimise the placebo/nocebo responses in clinical trials (37 –39). Taken together, the lack of carry-over and period effects suggest that methodological issues are less important reasons, if at all, for the development of migraine attacks after placebo. It is, therefore, relevant to consider what the probability would be for developing a spontaneous attack within a 12-h observational period. Of note, all included studies had exclusively enrolled individuals with episodic migraine, who reported between six and 76 attacks per year (9–35). Thus, it may be hypothesised that the incidence of migraine attacks after placebo is comparable to the probability of developing spontaneous attacks within the same observational period. This possibility should be further explored and would require theoretical modelling, in which the conditional probability of a migraine attack within a given observational period is calculated. Equivalence of the calculated probability with the observed incidence would yield a discussion of whether a placebo arm is even needed. This is truly an intriguing matter, but such theorical models are currently not available and a placebo arm is likely to be considered a necessity until otherwise proven. To date, most studies have assumed a 20% nocebo response rate; that is, a 20% induction rate of migraine attacks after placebo in individuals with migraine without aura (9–35). This should be revised considering our findings, and an assumption of a 15.5% nocebo response rate would be appropriate for future investigations. The implications would then maximise feasibility and minimise costs.

Limitations and strengths

Our study has some limitations. We have included studies with intravenous infusion only, which were performed in our centre. All studies were performed with slightly different definition of migraine frequency using ICHD I-III. Our findings were also limited to the small trial sample size. Furthermore, individual studies were performed using different migraine trigger substances with different physiological, biochemical and pharmacological properties, which may possess differential placebo/nocebo effects. Despite these limitations, our study is the most comprehensive analysis of nocebo effect in the human models of migraine.

Conclusions

Migraine attacks are seldom reported by individuals with migraine without aura after intravenous infusion with placebo and never by healthy volunteers. Future sample size considerations should assume a nocebo response of 15.5% in those with migraine.

Article highlights

In individuals with migraine without aura, the incidence of migraine attacks after placebo was 8.1%.

In healthy volunteers, the incidence of migraine attacks after placebo was 0.5%.

An assumption of a 15.5% nocebo response rate (upper limit of the 95% confidence interval) should be used for future investigations using human models of migraine.

Figure 1.

Flow chart of the study selection process.

Figure 2.

Visual abstract showing migraine attack induction rate after placebo infusion.

Figure 3.

Forest plot: (a) Migraine attacks after placebo in individuals with migraine without aura. (b) Delayed headache after placebo in individuals with migraine without aura.

Figure 4.

Visual abstract showing delayed headache induction rate after placebo infusion.

Figure 5.

Forest plot: (a) Migraine attacks after placebo in healthy volunteers. (b) Delayed headache after placebo in healthy volunteers.

Footnotes

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: MA is a consultant, speaker or scientific advisor for Allergan, Amgen, Alder, Eli Lilly, Lundbeck, Novartis, and Teva, primary investigator for Alder, Amgen, ElectroCore, Novartis and Teva trials. MA has no ownership interest and does not own stocks of any pharmaceutical company. MA serves as associate editor of Cephalalgia, Headache, and the Journal of Headache and Pain. MA is the President of the International Headache Society. MAK has acted as an invited speaker for Novartis and received travel grant from ElectroCore and FMA is a lecturer or scientific adviser for Novartis and Teva. HG, AI, HA, and TPD declare they have no conflicts of interest.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study received financial support from Lundbeck Foundation (R155-2014-171).

ORCID iDs

Hashmat Ghanizada

Mohammad Al-Mahdi Al-Karagholi

References

Ashina M. Migraine. N Engl J Med 2020; 383: 1866--1876.

Ashina

Hansen

Olesen

Pearls and pitfalls in human pharmacological models of migraine: 30 years’ experience. Cephalalgia 2013; 33: 540–553.

Ashina H, Schytz HW and Ashina M. CGRP in human models of primary headaches. Cephalalgia 2018; 38: 353--360.

Ashina M, Hansen JM, Do TP, et al. Migraine and the trigeminovascular system-40 years and counting. Lancet Neurol 2019; 18: 795--804.

Headache Classification Committee of the International Headache Society. The International Classification of Headache Disorders, 2nd edition. Cephalalgia 2004; 24: 9–160.

Headache Classification Committee of the International Headache Society. The International Classification of Headache Disorders, 3rd edition. Cephalalgia 2018; 38: 1–211.

Headache Classification Committee of the International Headache Society. Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Cephalalgia 1988; 8: 1–96.

Higgins JPT, Savović J, Page MJ, et al. Assessing risk of bias in a randomized trial. In: Higgins JPT, Thomas J, Chandler J, et al. (eds.) Cochrane Handbook for Systematic Reviews of Interventions, version 6.1. Cochrane, 2020. www.training.cochrane.org/handbook

Thomsen

Kruuse

Iversen

, et al. A nitric oxide donor (nitroglycerin) triggers genuine migraine attacks. Eur J Neurol 1994; 1: 73–80.

10.

Lassen

Haderslev

Jacobsen

, et al. CGRP may play a causative role in migraine. Cephalalgia 2002; 22: 54–61.

11.

Rahmann

Wienecke

Hansen

, et al. Vasoactive intestinal peptide causes marked cephalic vasodilation, but does not induce migraine. Cephalalgia 2008; 28: 226–236.

12.

Schytz

Wienecke

Olesen

, et al. Carbachol induces headache, but not migraine-like attacks, in patients with migraine without aura. Cephalalgia 2010; 30: 337–345.

13.

Schytz

Birk

Wienecke

, et al. PACAP38 induces migraine-like attacks in patients with migraine without aura. Brain 2009; 132: 16–25.

14.

Antonova

Wienecke

Olesen

, et al. Prostaglandin E2 induces immediate migraine-like attack in migraine patients without aura. Cephalalgia 2012; 32: 822–833.

15.

Al-Karagholi

MA-M

Hansen

Guo

, et al. Opening of ATP-sensitive potassium channels causes migraine attacks: A new target for the treatment of migraine. Brain 2019; 142: 2644--2654.

16.

Ghanizada

Al-Karagholi

MA-M

Arngrim

, et al. PACAP27 induces migraine-like attacks in migraine patients. Cephalalgia 2020; 40: 57–67.

17.

Wienecke

Olesen

Ashina

Prostaglandin I2 (epoprostenol) triggers migraine-like attacks in migraineurs. Cephalalgia 2010; 30: 179–190.

18.

Ashina

Bendtsen

Jensen

, et al. Nitric oxide-induced headache in patients with chronic tension-type headache. Brain 2000; 123: 1830–1837.

19.

Birk

Petersen

Kruuse

, et al. The effect of circulating adenosine on cerebral haemodynamics and headache generation in healthy subjects. Cephalalgia 2005; 25: 369--77.

20.

Hansen

Sitarz

Birk

, et al. Vasoactive intestinal polypeptide evokes only a minimal headache in healthy volunteers. Cephalalgia 2006; 26: 992–1003.

21.

Hansen

Pedersen

Larsen

, et al. Magnetic resonance angiography shows dilatation of the middle cerebral artery after infusion of glyceryl trinitrate in healthy volunteers. Cephalalgia 2007; 27: 118–127.

22.

Schytz

Wienecke

Oturai

, et al. The cholinomimetic agent carbachol induces headache in healthy subjects. Cephalalgia 2009; 29: 258–268.

23.

Wienecke

Olesen

Oturai

, et al. Prostaglandin E2 (PGE2) induces headache in healthy subjects. Cephalalgia 2009; 29: 509–519.

24.

Wienecke

Olesen

Oturai

, et al. Prostacyclin (epoprostenol) induces headache in healthy subjects. Pain 2008; 139: 106–116.

25.

Antonova

Wienecke

Olesen

, et al. Pro-inflammatory and vasoconstricting prostanoid PGF2α causes no headache in man. Cephalalgia 2011; 31: 1532–1541.

26.

Wienecke

Olesen

Ashina

Discrepancy between strong cephalic arterial dilatation and mild headache caused by prostaglandin D 2 (PGD 2). Cephalalgia 2011; 31: 65–76.

27.

Arngrim

Schytz

Asghar

, et al. Association of acetazolamide infusion with headache and cranial artery dilation in healthy volunteers. Pain 2014; 155: 1649–1658.

28.

Pellesi

Al-Karagholi

Hannibal

, et al. Two-hour infusion of vasoactive intestinal polypeptide induces delayed headache and extracranial vasodilation in healthy volunteers. Cephalalgia 2020; 40: 1212–1223.

29.

Kruuse

Jacobsen

Thomsen

, et al. Effects of the non-selective phosphodiesterase inhibitor pentoxifylline on regional cerebral blood flow and large arteries in healthy subjects. Eur J Neurol 2000; 7: 629–638.

30.

Lindholt

Petersen

Tvedskov

, et al. Lack of effect of norepinephrine on cranial haemodynamics and headache in healthy volunteers. Cephalalgia 2009; 29: 384–387.

31.

Ghanizada H, Al-Karagholi MA, Arngrim N, et al. Adrenomedullin induces migraine-like attacks in patients with migraine without aura. In: Proceedings of EHF Congress, Berlin, Germany, 1 July 2020. European Headache Federation.

32.

Al-Karagholi

MA-M

Ghanizada

Nielsen

CAW

, et al. Opening of BK Ca channels alters cerebral hemodynamic and causes headache in healthy volunteers. Cephalalgia 2020; 40: 1145–1154.

33.

Petersen

Birk

Kitamura

, et al. Effect of adrenomedullin on the cerebral circulation: Relevance to primary headache disorders. Cephalalgia 2009; 29: 23–30.

34.

Hougaard

Younis

Iljazi

, et al. Cerebrovascular effects of endothelin-1 investigated using high-resolution magnetic resonance imaging in healthy volunteers. J Cereb Blood Flow Metab 2020; 40: 1685–1694.

35.

Benedetti

Frisaldi

Barbiani

, et al. Nocebo and the contribution of psychosocial factors to the generation of pain. J Neural Transm 2020; 127: 687–696.

36.

Kenward MG and Jones B. Design and analysis of cross-over Trials. In: Rao CR, Miller JP and Rao DC (eds) Epidemiology and Medical Statistics Handbook of statistics, vol 27. Amsterdam, Oxford: North-Holland (an imprint of Elsevier), 2008, pp. 464--490.

37.

Wager

Atlas

LY.

The neuroscience of placebo effects: Connecting context, learning and health. Nat Rev Neurosci 2015; 16: 403–418.

38.

Evers

AWM

Colloca

Blease

, et al. Implications of placebo and nocebo effects for clinical practice: Expert consensus. Psychother Psychosom 2018; 87: 204–210.

39.

Benedetti

Maggi

Lopiano

, et al. Open versus hidden medical treatments: The patient’s knowledge about a therapy affects the therapy outcome. Prev Treat 2003; 6: 1a.