A Randomized Trial of Ketorolac and Metoclopramide for Migraine in the Emergency Department

Lawrence P. Richer MD, MSc; Samina Ali MD; David W. Johnson MD; Rhonda J. Rosychuk PhD; Amanda S. Newton PhD; Brian H. Rowe MD, MSc


Headache. 2022;62(6):681-689. 

In This Article


Study Design and Setting

A randomized, double-blind, placebo-controlled, parallel group trial was conducted in the EDs of two children's hospitals: Stollery Children's Hospital (Edmonton, Alberta, Canada) and Alberta Children's Hospital (Calgary, Alberta, Canada). The trial was registered with ClinicalTrials.gov (NCT01596166) and this manuscript is the primary analysis of these data. This study received research ethics approval from the University of Alberta and the University of Calgary. This clinical trial was assessed by the sponsoring institution (University of Alberta) to be low risk and, as such, did not require a data safety monitoring committee.


Eligible children presenting to the ED and aged 6 to 17 years were recruited from January 2013 to April 2014. Inclusion criteria were: (a) treatment with usual therapy at home OR at least one dose of oral ibuprofen or acetaminophen in the ED had failed to provide headache relief at the time of recruitment; (b) iv therapy was indicated in the opinion of the treating ED physician; and (c) International Classification of Headache Disorders, 2nd edition[7] criteria for migraine or probable migraine were met. Participants were excluded if they had: (a) ventriculoperitoneal shunt; (b) fever (temperature >38.5°C); (c) meningismus or clinical suspicion of meningitis; (d) head trauma associated with a headache in the past seven days; (e) allergy or other contraindication to metoclopramide or ketorolac; or (f) inability to complete headache pain assessments. Participants were not assessed for a history of medication overuse and were not excluded if ibuprofen was taken prior to enrollment. A screening log was maintained at each site and is documented in Figure 1.

Figure 1.

Consolidated Standards of Reporting Trials (CONSORT) flow diagram. ED, emergency department

A research assistant determined trial eligibility in consultation with the treating ED clinical staff and obtained written informed consent (and assent when appropriate) from each patient and/or caregiver prior to enrollment. Baseline characteristics recorded included age, sex, weight, duration of current migraine attack, duration of typical migraine attack, vomited in last 24 h, number of vomits in last 24 h, fluid intake in last 24 h, use of prophylactic medication, oral medication prescribed in ED prior to enrollment, baseline Visual Analog Scale (VAS) pain score, and baseline Faces Pain Scale - revised (FPS-R) score. The Pediatric Migraine Disability Assessment Scale (PedMIDAS) score, a measure of migraine disability validated in children and adolescents, was also collected. Medications taken by the patient prior to presenting to the ED and prescribed by the ED physician prior to enrollment were documented.

Intervention Groups

The monotherapy group received iv metoclopramide 0.2 mg/kg (maximum 10 mg) with iv placebo (0.9% sodium chloride to a maximum of 1 ml), whereas the ketorolac group (combination therapy) received iv metoclopramide 0.2 mg/kg (maximum 10 mg) with iv ketorolac 0.5 mg/kg (maximum 30 mg; standard concentration was 30 mg/ml). All participants also received 10 ml/kg of iv 0.9% sodium chloride, in keeping with a previous pediatric migraine treatment trial.[8] Intravenous medications were infused over ~ 30 min. Diphenhydramine was not included per protocol as the local standard of practice, assessed by a survey prior to initiating the trial, indicating that most ED physicians only prescribe the medication on an "as needed" basis.

Randomization and Allocation Concealment

Enrolled children were randomly allocated in a 1:1 ratio, using a web-based randomization system and permuted block randomization with variable block sizes at each site. Identically appearing vials labeled with the non-repeating, numeric treatment codes were prepared by a central research pharmacy with ketorolac or placebo and stored in the ED for immediate dispensing. The study pharmacist maintained the treatment code and associated group allocation in the research pharmacy separate from the ED. Metoclopramide was dispensed from the ED in an open-label fashion. The participants, research assistants, research coordinators, treating physicians, and biostatisticians were blinded to group allocation. The treating physician could request unblinding at any time during the study if deemed medically necessary.

Outcome Measures

Primary Outcome. The primary outcome was the mean change in pain intensity from baseline to 120 min after study drug administration. Pain was assessed every 30 min using the VAS (children aged 8 to 17 years) and the FPS-R (children less than 8 years). The VAS is a 100 mm validated pain measurement tool;[9] however, given that the VAS may be less reliable in children under 8 years of age,[9] the FPS-R was also used in younger participants.[10] The mean change from baseline is an accepted method to assess change in pain and has been previously reported in a pediatric ED migraine treatment trial.[8] It was deemed unreasonable to require participants to stay in the ED if their migraine headache had been successfully treated; this need to allow participants to leave the ED was not originally contemplated in the registered protocol. The primary outcome, therefore, was assessed no later than 120 min, and for those who responded early and/or left the ED, the last pain score value was carried forward (LVCF) as the final post-medication measurement. Using LVCF over a period of less than 2 h is unlikely to overestimate the treatment response, as headache recurrence is unusual over such a short timeframe.

Secondary Outcomes for Efficacy. Secondary outcomes measured during the ED visit were: (1) pain freedom defined as a VAS score of 0; (2) headache relief, defined as both 33% or 50% reductions in the VAS score from baseline; (3) presence of nausea; (4) presence of emesis; (5) use of rescue medication at the discretion of the treating ED physician after last assessment period; (6) participant responses to the questions: "I would take the medication again,"[11] and "my headache is a bit better/worse" or "my headache is a lot better/worse"[12] which were used as surrogates for "minimum clinically significant difference"; and (7) I would take this medication again. These outcomes were measured at 120 min or earlier if the participant left the ED.

Follow-up outcomes were assessed via telephone at 24 h after the ED visit. Follow-up outcome measures included: (1) discharged from the ED within 24 h; (2) sustained pain-freedom defined as no recurrence of headache or use of rescue medication within 24 h if pain freedom in the ED; (3) headache recurrence defined as no moderate or severe headache if the child's headache was reported as "a lot better" or had decreased by 33% or 50% from baseline on the VAS in the ED; (4) presence of nausea; (5) presence of vomiting; (6) return to ED within 24 h; (7) satisfaction with treatment in ED; and (8) use of rescue medication in last 24 h.

Secondary Outcomes for Safety. An adverse event (AE) was defined as any symptom, sign, illness, or experience that developed or worsened in severity during the study up to 24 h after leaving the ED. AEs were classified as possibly, probably, or definitely related to the investigational medication by the site investigator who was a qualified physician. Akathisia, a known side effect of metoclopramide and other dopamine receptor antagonists, was defined as an unpleasant sensation/urge to move, especially in the legs, and its presence and severity was determined using an akathisia scoring system collected with every participant.[13] As this scoring system has only been validated in adults, the treating ED physician's diagnosis of akathisia was also recorded. Akathisia and acute dystonic reactions were considered expected adverse drug reactions. A serious AE was defined as any fatal or life-threatening AE that resulted in a prolonged ED stay or hospital admission, resulted in persistent or significant disability or incapacity, or was an important medical event.

Sample Size

The minimum clinically meaningful difference in means between groups was set at 20 mm on the VAS scale. The choice was based on a previous adult ED-based study that determined that 30 mm or more reduction on the VAS scale was clinically important.[14] We chose a more conservative between-group difference to decrease the risk of missing smaller differences. Sample size was calculated for the primary outcome "change in VAS from baseline" using PASS analysis software (NCSS, Kaysville, UT). A sample size of 25 per group provided 90% power to detect a difference between group means of 19.6 mm. This calculation was based on a two-sided type I error of 0.05 and assumed a SD of 21 mm in each group, which was the change from baseline observed in a pilot study that we conducted to inform the current trial.

Data Collection

Research assistants were responsible for data collection and were blinded to treatment allocation when collecting all study data in the ED as well as the 24-h telephone follow-up during which they asked about subsequent admission to the hospital, return to the ED, headache recurrence/relapse, the presence of nausea or vomiting, use of rescue medications, AEs, and satisfaction with treatment.

Statistical Analysis Plan

The data were entered and analyzed blinded to the treatment allocation. All randomized participants that had not withdrawn consent were included in the analysis following the intent-to-treat (ITT) principle. Baseline characteristics and outcome data were summarized by group using descriptive statistics (e.g., mean, standard deviation [SD]; median, interquartile range [represented as IQR = 25th percentile − 75th percentile]; frequency, proportion and 95% confidence interval [CI]), as appropriate. Wilcoxon rank sum test, Pearson's Chi-squared test, or Fisher's exact test were used to assess for important baseline differences. The mean change from baseline on the VAS score was compared by group with pooled-variance, two-sample t-test. Assumptions required to interpret the test statistic on the primary outcome were verified using the Shapiro-Wilk normality test, density plot, and Q-Q plot. Results were considered statistically significant at a two-tailed p value less than 0.05 and the study was powered to detect a minimum clinically important difference of 20 mm between groups. Pearson's Chi-squared test or Fisher's exact test were used for secondary categorical outcomes, such as pain freedom or headache relief. The correlation between VAS and FPS-R for all participants and those less than or equal to 8 years was assessed using the Pearson correlation.

Analysis for important baseline imbalances was performed using simple and multiple linear regression modeling for: (a) the primary outcome—difference in VAS from baseline at 120 min without the last value carried forward and (b) difference in VAS with the full sample, including those who were discharged prior to 120 min with the last value carried forward. Variables showing some association with outcome (p < 0.2 in simple regression model) or other important variables (e.g., study site, baseline pain severity, and time of last pain assessment) were entered into the models and removed if nonsignificant (p > 0.05) or not required for model fit. In addition, a post hoc linear regression to test if the treatment group was a significant predictor of the last pain assessment when controlling for baseline pain severity was undertaken. In addition, a longitudinal linear regression with generalized least squares random effects and time after baseline assessment pain severity was assessed controlling for treatment group and baseline pain on VAS as fixed factors. Analyses were conducted in SAS version 9.4 (SAS, Cary, NC) and R version 4.1.2 (Vienna, Austria).[15] The authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.