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


Baseline Characteristics

Figure 1 outlines screening, enrollment, and follow-up and Table 1 outlines baseline characteristics divided by treatment group. Overall, the age range of participants was 6 to 17 years with a mean age of 13 (SD: 3) years; 35/53 (66%) were girls. The median duration of migraine attack prior to ED presentation was 24 (IQR: 8, 96) hours. The median PedMIDAS score was 21 (IQR: 6, 40) which corresponds to a mild degree of disability, and 15/53 (28%) were taking migraine preventative medications. Most participants (47/53 [89%]) had taken their usual medication at home. The most common usual treatments taken within 24 h of presentation were ibuprofen (n = 16), acetaminophen (n = 11), or acetaminophen and ibuprofen in combination (n = 15), naproxen sodium (n = 2), diclofenac potassium (n = 1), almotriptan (n = 1), zolmitriptan (n = 1), and ergotamine-caffeine (n = 1). One participant had received a codeine-acetaminophen combination in addition to acetaminophen and ibuprofen. Six of 53 (11%) participants were prescribed oral medications in the ED (acetaminophen n = 3 and ibuprofen n = 3). The overall baseline pain severity on VAS was 67 (SD: 20). Overall, 36 of 53 participants (68%) received some form of an NSAID prior to enrollment with 19 of 27 participants (70%) in the monotherapy group and 17 of 26 participants (65%) in the ketorolac group.

Primary Outcome

Thirty-one of 53 (58%) participants had their last pain measurement at 120 min and the remainder (22/53, 41%) had their last pain assessment at an earlier time point before leaving the ED. The mean pain difference among participants assessed at 120 min is visualized in Figure 2B and was −44 mm (SD: 24, 95% CI: 32–57) for the monotherapy group (metoclopramide + placebo) and −36 mm (SD: 24, 95% CI: 23–49) for the ketorolac group (metoclopramide + ketorolac) with a difference in means between groups of 8 mm (95% CI: −9 to 25; p = 0.355). The Pearson correlation between FPS-R and VAS at baseline was good for all participants (r = 0.82) but low (r = 0.36) for the five participants who were ≤8 years of age.

Figure 2.

(A) Box plot of VAS (mm) by treatment group at baseline, 30, 60, 90, and 120. (B) Overall mean decrease from baseline on VAS divided by treatment group without the last value caried forward for participants who left the ED earlier. ED, emergency department; VAS, Visual Analog Scale

As it was deemed unreasonable to require participants to stay in the ED if their migraine headache had been successfully treated, 22 patients had their last pain assessment prior to the planned 120 min. The mean difference in pain intensity at each timepoint is visualized in Figure 2A. In total, 11 participants (42%) had their last assessment at 90 min or less in the monotherapy group and 11 in the combination therapy group (41%) and summarized in Table S2. The overall difference in mean pain intensity from baseline to the last pain assessment carried forward (i.e., using the VAS at 120 min or an earlier time point if the patient had improved sufficiently to be discharged from the ED) was not significantly different with −38 mm (SD: 21, 95% CI: 29–46) for monotherapy group and −42 mm (SD: 24, 95% CI: 32–51) for the ketorolac group with a difference in means between groups of 4 mm (95% CI: −16 to 8, p = 0.525).

Simple regression analysis was performed without the last pain assessment carried forward to assess the impact of the observed imbalances in the baseline characteristics duration of current migraine attack and other variables showing some association with outcome (p < 0.2). Migraine attack duration, syncope, ibuprofen or acetaminophen taken in ED, and baseline nausea were included in the model, but none were significant predictors of the difference in VAS from baseline. The overall regression was not statistically significant (R 2 = 0.2612, F(df 4, 25) = 2.21, p = 0.097) and was similar when the last value was carried forward for those who were discharged prior to 120 min.

In a post hoc analysis, multiple linear regression was used to test if baseline pain severity, time of last assessment, and treatment group significantly predicted the last pain assessment. When controlling for baseline pain severity and the time of the last pain assessment, there was no significant difference with treatment allocation (β = 0.6, 95% CI: −11–14, p = 0.8). The post hoc longitudinal linear regression model with time after baseline assessment (i.e., 30, 60, and 120 min) and with treatment group and baseline pain as fixed factors was significant (R 2 = 0.34, Wald χ 2 = 50.78, p < 0.0001). Baseline pain intensity was again the only significant predictor for VAS with a coefficient of 0.81 (SE: 0.1, p < 0.001, 95% CI: 0.6–1), whereas the estimate for the ketorolac group versus monotherapy was not significant and 3.7 (SE: 4.6, p = 0.423, 95% CI: −5.3 to 12.6).

Secondary Outcomes

Efficacy. Table 2 outlines planned secondary outcome measures reported in the ED for efficacy. There was no evidence of a statistically significant difference between groups for any of the secondary outcomes for efficacy. Additional medications not included in the study protocol were prescribed to a total of 10 participants with six in the monotherapy group and four in the combined therapy group. Medications administered after study completion in the ED included dexamethasone (n = 5), diphenhydramine (n = 2), ondansetron (n = 2), and oxycodone (n = 1). An additional dose of metoclopramide was given to two participants and ketorolac to three participants.


In total, nine of 52 (17%) participants reported one or more AE, but only three of these were reported to have occurred after signing the consent form. There was no statistically significant difference in AEs between groups, with five participants in the monotherapy group reporting one or more AEs and four in the ketorolac group. The reported AEs are summarized in Table S1. In total, there were five AEs reported in the monotherapy group and seven in the ketorolac group. Reported AEs included runny nose (n = 1), restlessness (n = 2), vomiting (n = 2), sedation (n = 3), headache recurrence (n = 2), and diarrhea (n = 1). These AEs were categorized as mild (n = 7), moderate (n = 4), and severe (n = 1). AEs determined to be "possibly" related to the study medications included sedation (n = 2), restlessness (n = 1), runny nose (n = 1), and recurrence of headache (n = 1). The one serious AE was an admission to the hospital for continuing headache treatment and determined not to be related to the study medications.

Using the validated akathisia measurement instrument, definite akathisia was not recorded for any participants. Probable akathisia was diagnosed when a participant reported the urge to move, but the subjective report was not corroborated by the independent motor assessment by the treating ED physician. There were more participants diagnosed with probable akathisia in the monotherapy group with nine of 25 (36%) compared with two of 22 (9%) in the ketorolac group (p = 0.030). Patients reporting akathisia rated the severity as mild in six of 11 cases (54%), moderate in two of 11 (18%), and severe in three of 11 (27%). No dystonic reactions were observed nor reported as an AE.

Follow-up Outcomes

Follow-up outcomes are presented in Table 3. Overall, 28 of 46 participants (61%) reported their headache had improved a lot in the ED, but six of 47 (13%) participants had sustained pain-freedom at 24 h and five of 48 (10%) returned to the ED. Although more participants in the ketorolac group (five of 23 (22%)) had sustained pain-freedom compared with one of 24 (4%) in the monotherapy group, the difference was not significant (p = 0.100).