Eight RCTs explored ketorolac’s efficacy for acute pain in children in the ED and are summarized in Table 1.

The analyzed studies included a total of 941 children and adolescents, with a number of enrolled patients in each trial ranging from 29 to 212. The minimum age of enrolled children was 4 years, and the maximum age was 18 years.

Three RCTs were focused on migraine [7,8,9], two on musculoskeletal trauma [10, 11], one on renal colic [12], one on acute abdominal pain [13], and one on acute vaso-occlusive crisis in sickle cell disease [14].

The route of administration of ketorolac was intravenous in five RCTs [7,8,9, 12, 14], sublingually in two RCTs [11, 13], orally in one RCT [10], and intranasally in another one [7].

Among trials, the dosage of ketorolac ranged between 0.5 and 1 mg/kg.

Ketorolac was compared to sublingual tramadol in two RCTs [11, 13], to oral ibuprofen [10], nebulized fentanyl [12], sublingual paracetamol [13], and intravenous prochlorperazine [9] in one. A single trial compared the combination of intravenous ketorolac and metoclopramide to metoclopramide alone [8]. Another one compared the combination of intravenous ketorolac and morphine to intravenous morphine alone [14]. One trial compared intranasal to intravenous ketorolac [7].

When looking at the potential risk of bias, all the studies were classified in the “low” range (Supplementary Fig. 1). No critical information was missing. Overall, all the RCT described accurately the randomization procedure, the statistical power calculation; moreover, the blinding of the participants and medical staff was asserted. Drop-out was negligible in all studies. Therefore, all the studies are eligible for the following meta-analysis.

In the three RCTs focused on the treatment of migraine, ketorolac was employed intravenously at a dose of 0.5 mg/kg, alone or in combination with metoclopramide [8] and compared to 1 mg/kg intranasal ketorolac [7], 0.2 mg/kg intravenous metoclopramide alone [8], or 0.15 mg/kg intravenous prochlorperazine [9]. These trials showed that the intranasal administration of ketorolac was non-inferior to intravenous ketorolac considering pain scores at 30, 60, and 120 min after the administration [7,8,9]. On the other hand, the administration of ketorolac in combination with metoclopramide did not provide statistically significant changes in pain scores 120 min after the administration [8]. Finally, intravenous prochlorperazine was significantly more effective than intravenous ketorolac in decreasing headache 60 min after administration [9].

Considering musculoskeletal trauma, ketorolac was tested through the oral and sublingual route [10, 11]. One trial compared 0.5 mg/kg oral ketorolac and 10 mg/kg oral ibuprofen in children with severe pain and showed that pain scores were similar 30, 60, and 120 min after administration [10]. Another trial comparing 0.5 mg/kg sublingual ketorolac and 2 mg/kg sublingual tramadol showed a similar reduction in pain scores after 30, 60, and 120 min [11].

A single trial investigated the efficacy of ketorolac for acute abdominal pain in children [13]. In this trial, 0.5 mg/kg sublingual ketorolac was compared to 2 mg/kg sublingual tramadol or to 20 mg/kg of a melt in mouth powder of paracetamol. This trial showed that the three regimens were similarly effective.

The trial focused on children with acute renal colic compared intravenously administered ketorolac (0.9 mg/kg) with 3 mcg/kg nebulized fentanyl and showed that both regimens were able to decrease the intensity of pain at 60 and 120 min [12]. The number of patients with complete pain relief were significantly more in the ketorolac group.

Finally, the trial performed in children with an acute vaso-occlusive crisis in sickle cell disease showed that the combination of intravenously 0.5 mg/kg of ketorolac with 0.1 mg/kg morphine did not provide a superior analgesia compared to the administration of morphine alone [14].

Regarding the meta-analysis, due to the low number of RCT studies available on pediatric patients, the efficacy of ketorolac was evaluated using the random-effects (RE) model. Three studies were excluded from the meta-analysis, Richer et al., since it used a combination of ketorolac and metoclopramide, and Razaei et al., since it included adult patients and did not divide the pediatric results separately, and Tsze et al. because they compared different routes of ketorolac administration.

No major analgesic effect was observed for ketorolac with respect to the other drugs, opioid (RE model, MD = 0.24, 95% CI − 0.37; 0.84) and other NSAID (RE model, MD = 0.48; 95% CI − 0.14; 1.10). Overall, there was also not a different effect between ketorolac and other drugs (RE model, MD = 0.27, 95% CI − 0.57; 1.11). The heterogenicity was low between the studies employing opioid (I2 = 0%, τ2 = 0), medium between the studies using NSAID (I2 = 41%, τ2 = 0.13), and medium–high across all the studies (I2 = 73%, τ2 = 1.00) (Fig. 2).

Forest plot of the analysis for ketorolac effect versus other NSAIDs and opioid. The results from RE (random-effects model), using mean differences (MD), and 95% confidence interval (CI) were reported. Heterogenicity I2 and τ2 were shown. Abbreviations: O, opioids; N, NSAIDs; S, severe pain; M, moderate pain

Then, the studies were categorized based on the main condition treated. Ketorolac was not different to other drugs for abdominal pain (RE, MD = 0.5; CI − 0.15, 1.15), and musculoskeletal trauma (RE, MD = 0.33; CI − 0.32, 0.98). The overall efficacy of ketorolac was not different from the other drugs (RE, MD = 0.27; CI − 0.57, 1.11) (Fig. 3). The heterogenicity was medium among musculoskeletal trauma studies (I2 = 52%, τ2 = 0.17), low in abdominal pain studies (I2 = 0%, τ2 = 0), and medium–high across all studies (I2 = 73%, τ2 = 1.01).

Forest plot of the analysis for ketorolac effect versus other drugs in musculoskeletal trauma (limb trauma), migraine, abdominal pain, and sickle cell vaso-occlusive crises. The results from RE (random-effects model), using mean differences (MD), and 95% confidence interval (CI) were reported. Heterogenicity I2 and τ2 were shown. Abbreviations: O, opioids; N, NSAIDs; S, severe pain; M, moderate pain

Influence analysis showed that the studies by Broasseus et al., Ghirardo et al., and Hardwick et al. may impact on the meta-analysis results; however, we cannot exclude them from the analysis due to the very limited number of the eligible studies (Figure supplementary 2).

Accordingly, to the GRADE system, due to the lack of significant differences in efficacy between ketorolac and other drugs for pain managing in the ED, and the presence of medium–high heterogenicity among the studies, a conditional recommendation was formulated for either the intervention or the comparison (Table 2 and Table supplementary 2).