The authors compared 10mg intravenous metoclopramide as a sole analgesic agent to a combination of metoclopramide and 20mg intravenous tenoxicam, and to tenoxicam alone, for the relief of pain in emergency department patients aged 18-65 with clinically suspected or radiologically proven renal colic. Participants were block randomised with a block size of 1 and a ratio of 1:1:1 by a well-described, simple manual approach to allocation concealment (pulling their group allocation out of a bag.) The study was double blind, with placebo medication used to maintain blinding in the sole agent groups.

Pain was measured on a 100mm Visual Analogue Scale (VAS) at baseline, 10, 20 and 30 minutes after trial medication administration.

They screened 397 patients and randomised 240 participants, with appropriate sample size calculations presented for a superiority study. There were no significant differences in participant baseline characteristics or pain scores.

All groups had a reduction in pain score at all time points compared to baseline, and this reduction increased with time. There was a statistically significant difference between groups at the 20min time point, with the greatest pain relief occurring in the combined tenoxicam and metoclopramide group. There was no statistically significant difference between groups at the other time points. Pain relief in the groups can be seen in Table 1.

There was no mandated radiological confirmation of diagnosis, and no details were given concerning the number of participants with radiological confirmation; multiple approaches to adjudicating the diagnosis were described, including clinical suspicion, urinalysis, ultrasound and CT.

The authors describe the possibility that participants could have received opiate medication prior to randomisation; this is not quantified or further assessed.

There was a very long interval between data collection and manuscript publication (6 years); the reason for this is not clear.

There was a low risk of bias in all domains using the RoB2 tool, although it is noted that the allocation concealment is not particularly robust, and that staff performing randomisation processes could have ascertained which groups were represented by the employed code system over time. Furthermore, it is not possible to establish whether one group may have received more opioids than the other prior to randomisation, although there is no evident reason why this may have occurred. Overall the raters felt that there were some concerns about bias in this trial.

The authors compared 20mg intravenous metoclopramide to -fen, a combination intravenous analgesic containing metylscopolamine nitr. 0.15mg, papaverine hydrochloride 20mg, morphine hydrochloride 6.6mg, noscapine hydrochloride 3mg and codeine chloride 0.4mg. The study was double blind, and medication was presented in unlabelled vials administered in sequence. The method for generation of the sequence is not explicit.

Pain was documented on a 100mm VAS at baseline, 10, 20 and 30 minutes by both the patient and the treating nurse; authors describe a highly significant correlation (p <0.01) between patient and nurse pain assessment but are not explicit about which results are used for the final analysis. No response to contact attempts with the corresponding author was received. Therefore, it is not possible to combine the results from this study with those from Baloglu Kaya et al. (2015) to undertake meta-analysis.

They recruited 40 emergency department patients with clinical suspicion of renal colic, and later excluded 1 in whom this could not be proven radiologically.

Patients were excluded if they required supplementary analgesia; this led to the exclusion of 3 patients in the metoclopramide group and 4 patients in the Spasmofen group. A further patient in the Spasmofen group was excluded due to “inability to cooperate”. This left 17 patients in the metoclopramide group and 14 in the Spasmofen group for final analysis.

The pain scores in each group at each time point can be seen in Table 2; the authors describe no significant difference between groups but do not present the results of their statistical analysis.

The method for generating the randomisation sequence is not described, and whilst the article suggests that both patient and treating clinician were blinded, it is not possible to assess the rigour of this given the limited description provided.

Using nurse measurement of pain severity is not currently considered good practice, and it is not clear whether the patient or the nurse measurements were used for the final analysis. Furthermore, the authors describe correlation between patient and nurse values, rather than agreement; this calls the utility of the results further into question.

The sample size was small, with a high drop-out rate; it is particularly concerning that patients with a supplemental analgesic requirement were excluded in a study of analgesic efficacy. No sample size calculations are presented, and it is not clear whether the study was intended as a superiority or non-inferiority study.

The risk of bias in many domains is difficult to assess due to the brevity of the article. Lack of description of allocation sequence generation and of group baseline characteristics gives some concerns in Domain 1 (randomisation process). The high drop-out rate, in part directly attributable to the effectiveness of the study intervention, leads to a high risk of bias in Domain 3 (missing outcome data). The lack of clarity concerning whose measurement of pain constituted the primary outcome leads to some concern in Domain 4 (measurement of the outcome), and similarly gives concern that the chosen metric for the primary analysis may have been decided once the results were known. Overall, the raters felt that the paper was at a high risk of bias, although the direction of this bias was not clear.