Perioperative Dextromethorphan as an Adjunct for Postoperative Pain: A Meta-analysis of Randomized Controlled Trials

N-METHYL-d-ASPARTATE (NMDA) receptor antagonists have become widely used adjuncts for postoperative analgesia.^1,2  Ketamine, a well-studied NMDA antagonist, has been shown to decrease postoperative pain when administered preemptively,^3,4  intraoperatively,⁵  and postoperatively,^6,7  without causing an increase in sedation but with a notable increase in hallucinations and nightmares.⁸  Dextromethorphan, an NMDA receptor antagonist that is most routinely used as an oral antitussive, has also been extensively studied for its use as a perioperative analgesic adjunct.^9–29  Dextromethorphan has previously undergone systematic review without quantitative meta-analysis in which the authors determined that the drug was a potentially useful analgesic adjunct, but there still remained significant questions about the consistency of findings between studies.³⁰  Since that systematic review was accepted for publication in 2005, there have been more than 10 additional studies^{9–14,31–34} on dextromethorphan for postoperative pain control. A meta-analysis of the results of studies that investigate dextromethorphan for its effect on postoperative pain and opioid reduction has not yet been published. Therefore, we performed a meta-analysis on the use of preoperative dextromethorphan and its effects on opioid consumption and postoperative pain scores.

This study is a meta-analysis of existing, publicly available literature, did not involve the collection of new human or animal data, and is exempt from the institutional review board review. The Cochrane specifications for systematic reviews was used to guide the construction of this meta-analysis.³⁵  A systematic search was performed in PubMed, Web of Science, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Pubget, and EMBASE on August 4, 2014. The following search terms were used: (dextromethorphan) AND acute pain; (dextromethorphan) AND postoperative pain; (dextromethorphan) AND pain. Trials were only included if they were randomized, double-blinded, placebo-controlled, and published in English. Unpublished abstracts and reports were excluded. Pediatric trials on patients less than 12 yr of age were also excluded. Authors of the trials were not contacted for original data.

To ensure the quality of included trials, each trial was scored based on a modified validated scale previously used for meta-analysis.³⁶  The scale was designed to evaluate the quality of placebo-controlled, randomized trials and includes the following parameters:

1. Randomization: a point was given for stating the trial was randomized. An additional point was given if randomization was described and appropriate, such as the use of a random number generator.

2. Blinding: a point was given if the trial was stated to be double-blind. If blinding method was described and appropriate, such as the use of identical placebo pills, an additional point was given.

3. Withdrawals: a point was given if patient withdrawals and the reasons for withdrawals were reported.

4. Pain intensity: to ensure that the trial evaluated clinically significant pain, a point was given if mean visual analog pain scores were greater than 30 mm or greater than 3 of 10 on a numeric rating scale.

5. Power analysis: a point was given if sample size was determined through the use of a power analysis.

Thus, the minimum requirements for inclusion would be a score of 2 points and the maximum score would be 7 points.

To be included in the meta-analyses, we required all trials to have a treatment arm in which IV, intramuscular, or per os dextromethorphan was administered before surgery—if treatment groups also received intraoperative or postoperative doses of dextromethorphan (table 1), they were included as well. Only test groups from studies in which dextromethorphan was administered preoperatively were included for analysis. If test groups were administered dextromethorphan only intraoperatively or postoperatively, they were not included for analysis. If multiple dextromethorphan dosages were administered in an included study, the highest dose group was used for the comparison. However, as a sensitivity analysis, all comparisons were recalculated, where possible, using the lowest dose groups.

The outcome variables we sought were postoperative opioid consumption, pain scores, and incidence of side effects. The investigation of published studies led to the a posteriori selection for analysis of total opioid consumption for 24 to 48 h postoperatively, numeric pain scores at 1, 4 to 6, and 24 h, and the incidence of opioid- and dextromethorphan-related side effects. For comparison of postoperative opioid use, studies were included if they tracked total use of opioids over a 24- or 48-h period. If an opioid other than IV morphine was used, such as meperidine, the reported values were converted into IV morphine equivalents using an online calculator.³⁷  Inclusion required sole use of opioids as a pro re nata analgesic. Comparisons between groups that received the same nonopioid intervention (such as a single dose of a nonsteroidal antiinflammatory drug [NSAID] in both control and dextromethorphan groups) were also included.

Studies were eligible for pain score comparisons if they reported pain scores on a standardized 0 to 10 numeric rating scale, such as the visual analog scale. Numeric pain score comparisons were performed at three time points: 1, 4 to 6, and 24 h postoperatively. For the 1-h group, studies were included if they reported pain scores within 1 h postoperatively. Thus, studies were also included in this group if they did not report pain scores at 1 h but did report in the first hour in the postanesthetic care unit. For the 4- to 6-h group, studies were included if they reported pain scores at 4 or 6 h. If a study reported pain scores at both times, the score at 4 h was used.

We intended to compare the incidence of opioid-related side effects, such as nausea and itching, as well as dextromethorphan-related side effects, such as nausea and euphoria, but this was not feasible due to the small number of events reported. Thus, rather than report meta-analysis of side effects, we systematically reviewed the included trials for reported side effects.

Statistical analyses were performed with Review Manager version 5.3 (The Nordic Cochrane Centre, Denmark). All calculations required knowledge of the mean and SD for the compared parameters. Some studies represented mean and SD graphically—in these cases, the computer program Plot Digitalizer³⁸  was used to estimate the values at the set time points. As mean and SD were used for comparison calculation, the effect size is expressed as difference in means (MD). By convention, MDs favoring dextromethorphan were considered negative and those favoring control considered positive. To account for anticipated heterogeneity, a random-effects model³⁹  was used for all calculations. We also used the I² statistic to assess the degree to which differences between trials were due to heterogeneity.⁴⁰  Alpha was set at 0.05 and, after performing a Bonferroni correction accounting for four total comparisons, the significance criterion set at 0.0125. All comparisons are presented graphically in this article by using forest plots.⁴¹ 

The selection process is summarized in figure 1. Table 1 lists all studies used in the comparisons including pertinent aspects of their design and subgroups. A total of 40 studies were identified and a total of 19 were excluded, leaving 21 studies that were used in at least one comparison. The median quality score of these studies was 5 of 7 with an interquartile range of 2.

A total of 14 trials reported mean and SD of opioid consumption for the first 24 or 48 h postoperatively and a total of 848 patients were included in the comparison. MD favored dextromethorphan (MD, −10.51 mg of IV morphine equivalents; 95% CI, −16.48 to −4.53 mg; P = 0.0006; fig. 2). Three studies in particular, reported by Weinbroum et al.,¹⁹  Helmy and Bali,²²  and Wu et al.,²⁵  were statistical outliers with an MD of less than −20. To ensure that these three studies alone had not resulted in the comparison’s significance, a sensitivity analysis was performed with them excluded. The overall effect was lessened, but the comparison remained significant (MD, −4.45 mg IV morphine equivalents; 95% CI, −7.47 to −1.43 mg; P = 0.004) and I², although still high, was reduced from 97 to 88%.

Pain scores at 1 h were reported as mean and SD in 13 studies with a total of 884 included patients. MD favored dextromethorphan (−1.60; 95% CI, −1.89 to −1.31; P < 0.00001; fig. 3). Weinbroum et al.¹⁹  was an outlier with an MD of less than −4. After exclusion, the overall effect was lessened, but the comparison remained significant (−1.50; 95% CI, −1.78 to −1.22; P < 0.00001) and I² decreased from 91 to 90%.

Pain scores at 4 to 6 h were reported in 13 studies with a total of 950 included patients. Three subgroups in two studies (reported by Suski et al.¹⁰  and Weinbroum²⁰ ) reported pain scores at both 4 and 6 h and the values recorded at the 4 h were used in the comparison. MD favored dextromethorphan (−0.89; 95% CI, −1.11 to −0.66; P < 0.00001; fig. 4) with an I² of 88%.

Pain scores at 24 h were reported in 12 studies with 797 included patients. Dextromethorphan was also favored at this time point (MD, −0.92; 95% CI, −1.24 to −0.60; P < 0.00001; fig. 5) with an I² of 92%.

A total of three studies in the opioid consumption, 24-h pain score, and 1-h pain score comparisons and two studies in the 4- to 6-h pain score comparison included multiple dosing regimens of dextromethorphan. Of note, two of these studies (reported by Wu et al.²⁵  and Weinbroum et al.²¹ ) were completed by groups that only used their highest dose in subsequent studies—thus, their highest dosing groups best approximated the most common dextromethorphan doses in the study and were used for the initial comparisons. When comparisons were recalculated using the low-dose instead of high-dose groups for comparison, all results remained significant although with a lower magnitude of effect (opioid consumption MD, −10.05 mg of IV morphine equivalents; 95% CI, −15.79 to −4.31 mg, P = 0.0006; pain at 1 h: MD, −1.50; 95% CI, −1.79 to −1.21, P < 0.00001; pain at 4 to 6 h: MD, −0.87; 95% CI, −1.11 to −0.64, P < 0.00001; and pain at 24 h: MD, −0.65; 95% CI, −0.95 to −0.35; P < 0.0001).

Eighteen of 21 trials included in our meta-analyses tracked the incidence of side effects, which for both opioids and dextromethorphan primarily consist of nausea, vomiting, dizziness, and lightheadedness. Ten studies reported either no side effects or a nonsignificant difference between groups.^{9,10,13,14,16,19,20,22,24,27,29}  Five studies did, however, report a decrease in side effects in groups receiving dextromethorphan.^{15,17,18,25,26}  One study²³  found a higher incidence of nausea in the dextromethorphan group, with rating mild-to-moderate nausea reported by patients at 31 time points in the dextromethorphan group compared with 20 time points in the control group although no patients reported severe nausea at any time. Weinbroum et al.²¹  tracked sedation using a standardized scale and found an increase in sedation in the placebo group.

A variety of study designs in multiple hospital settings and countries have attempted to elucidate the value of perioperative dextromethorphan as an adjunctive analgesic. In a prior report, these efforts were synthesized in a qualitative systematic review of NMDA receptor antagonists’ role in decreasing postoperative pain and opioid consumption, which demonstrated a significant benefit from dextromethorphan in 67% of included studies.²  In addition, a separate qualitative systematic review of dextromethorphan that shared in common 15 of the studies used in this analysis suggested that dextromethorphan had potential as an adjunct to postoperative opioid analgesics but did note variability among the analyzed studies.³⁰  Here, we have systematically searched the published literature on the preoperative use of dextromethorphan to decrease postoperative pain and opioid use. Ultimately, we identified 21 trials published between 1998 and 2013 that addressed these metrics and were suitable for quantitative meta-analysis. The results of our meta-analyses suggest that, when used preoperatively, dextromethorphan significantly decreases pain and opioid use in the postoperative period.

To objectively index included trials by design quality, we scored each trial based on a quality index. The majority of studies in our meta-analyses scored in the 5 to 7 range. These studies demonstrated a high degree of transparency in their study designs and sampling processes. A minority of studies scored in the 2 to 4 range, with a score of 2 representing the minimum requirements of being a randomized blinded trial. Although we did not weigh trials based on their scores, the average scores of the trials do reflect the average high quality of the studies from which we draw our conclusions.

As an NMDA receptor antagonist,⁴²  dextromethorphan has been proposed to exert its effects as a preemptive analgesic by preventing NMDA-mediated calcium current and subsequent modulation of nociception in spinal pain fibers and the central nervous system. This in turn prevents a pain phenomenon known as “windup” that results in amplified subsequent responses to painful stimuli and poorer responses to opioids.^43–46  In previous trials, dextromethorphan has shown benefit in various chronic pain conditions including diabetic neuropathy, postherpetic neuralgia,⁴⁷  and phantom limb pain.^48,49  Effects on cancer pain have also been investigated in at least two trials with mixed results.^50,51 

Multimodal preemptive analgesic adjuncts, including NMDA receptor antagonists, local anesthetic infiltration, NSAIDs, epidural analgesia, and preemptive opioids, have been the subject of a prior meta-analysis.³⁶  This study found benefit with preemptive NSAIDs, epidural analgesia, and local anesthetic infiltration, but its comparisons for both ketamine and dextromethorphan were equivocal. In contrast, the same year, a meta-analysis of perioperative IV ketamine use reported a mean of 15.7 mg less morphine consumption at 24 h and mean pain score improvements of 0.89 at 6 h, 0.42 at 12 h, 0.35 at 24 h, and 0.27 at 48 h.⁵²  These results are remarkably similar to our own. A more recent meta-analysis of perioperative IV ketamine found benefits for opioid consumption and time to first analgesic but did note increased hallucinations and nightmares.⁸  The statistic used for this analysis was the standardized mean difference rather than mean difference, making direct comparison to effect observed in our own study difficult.

However, although ketamine is widely used as a multimodal adjunct worldwide, our anecdotal experience from multiple institutions is that dextromethorphan does not appear to share the same level of popularity and is very rarely used as an adjunct for postoperative analgesia. On the basis of our findings, the use of dextromethorphan perioperatively could potentially provide similar benefits to preemptive ketamine therapy in a simple oral, intramuscular, or IV formulation. Further investigation, particularly a head-to-head randomized trial alongside placebo, may help clarify whether the different NMDA antagonists provide similar levels of relief with a similar incidence of dysphoric or other side effects, or not. Additional research may also explore whether there is benefit from the simultaneous use of more than one NMDA receptor antagonist as it is unclear whether this would result in an additive, synergistic, or antagonistic effect.

Well-documented dextromethorphan side effects and concerns include dose-related tachycardia, respiratory depression, gastrointestinal symptoms, and abuse potential.⁵³  Although its recreational abuse potential is clear, dextromethorphan dependence has only been rarely described,^54,55  and its abuse is best described in adolescents.^56,57  Recent work has described dose-dependent hallucinogenic properties of dextromethorphan as well as acute changes in memory and cognition^58,59  although these effects typically occurred at doses well in excess of those used in the included studies. Thus, it seems reasonable to avoid doses above 2 mg/kg per os, which has been described as a dose above which dissociative effects are typically seen,⁶⁰  in order to prevent neurologic disturbances before surgery. However, there exists, to our knowledge, no evidence that a single dose of dextromethorphan for preemptive analgesia would increase potential for postoperative abuse, and indeed, review of the included trials revealed a minimal incidence of dextromethorphan-related adverse effects.

Although opioids are a mainstay of effective perioperative analgesia, their use is nonetheless frequently associated with side effects that can increase hospital costs and length of stay.⁶¹  Multimodal analgesia has been proposed as a way to improve pain control while reducing side effects,⁶²  but to date, little evidence exists to link opioid-sparing analgesic regimens to reduced opioid-related adverse effects. The available studies were insufficient for meta-analysis on the incidence of side effects with dextromethorphan, but our qualitative review of the literature suggests that most studies saw minimal change in the incidence of side effects. Ketamine, in contrast, was shown in prior meta-analysis to increase the risk of hallucinations when administered in awake patients although the incidence of opioid-related side effects was also unchanged.⁵²  This difference highlights the fact that different NMDA receptor antagonists are not necessarily interchangeable, and therefore, continued exploration into other agents such as dextromethorphan and memantine is still warranted. Larger studies may clarify whether opioid-sparing doses of dextromethorphan are able to quantitatively decrease the incidence of opioid-related side effects without causing hallucinations at similar rates to ketamine.

Similar to the systematic review by Duedahl et al.,³⁰  we observed a high degree of heterogeneity, with an I² greater than 80% in each comparison. This is likely a reflection of the variability between study designs, such as differences in type of surgery, dextromethorphan dosing regimens, dextromethorphan administration routes, and postoperative analgesic regimens. We had anticipated this and therefore used a random-effects model for all of our calculations. The high heterogeneity does, nonetheless, demonstrate the variability in findings among dextromethorphan studies and highlights the need for a larger study with a standardized protocol to clarify dextromethorphan’s role in the perioperative setting. Important details to clarify include the optimal perioperative dextromethorphan dose and duration of use, the incidence of side effects, and whether or not the perioperative use of dextromethorphan improves outcomes such as hospital length of stay.

Our analysis is also limited by the fundamental reliance of meta-analyses on the existing data and the reporting mechanisms of the original studies. Many high-quality studies needed to be excluded from the quantitative analyses due to reporting results in forms other than mean and SD, tracking opioid use over periods less than 24 h, or reporting pain scores in forms other than fixed intervals (such as only reporting the worst recorded). In a small number of studies with multiple dextromethorphan dosing arms, we also had to exclude groups in order to avoid duplicating control patients in our comparisons. As a result, our quantitative analyses do not necessarily represent the full body of literature on the perioperative use of dextromethorphan. In addition, due to the heterogeneity of published studies, this is an a posteriori–derived analysis of total opioid consumption for 24 to 48 h postoperatively and pain scores at 0 to 1 h, 4 to 6 h, and 24 h postoperatively.

Despite these limitations, our comparisons do nonetheless represent a cross-section of several hundred patients in the available randomized controlled trials on the effects of preoperative dextromethorphan on postoperative pain control with significantly favorable results. To date, no large randomized controlled trial has been conducted on this topic. Our quantitative meta-analyses of the existing randomized controlled studies of dextromethorphan for postoperative pain control demonstrated a significant reduction in postoperative opioid use for 24 to 48 h after surgery as well as pain represented by pain scores up to 24 h after surgery. Due to high heterogeneity between the existing trials and the lack of a single large randomized study on this topic, further evidence is required to definitively determine a benefit.

The authors thank Hang Lee, Ph.D. (Harvard Medical School and the Massachusetts General Hospital Biostatistics Center, Boston, Massachusetts), for statistical consultation.

This work was conducted with support from Harvard Catalyst/The Harvard Clinical and Translational Science Center (National Center for Research Resources and National Center for Advancing Translational Sciences, National Institutes of Health [Bethesda, Maryland] Award UL1 TR001102) and financial contributions from Harvard University (Cambridge, Massachusetts) and its affiliated academic healthcare centers.

The authors declare no competing interests.