Use of gabapentin for perioperative pain control – A meta-analysis

Department of Anesthesiology and Pain Management, 1Toronto Western and 2General Division, University Health Network, University of Toronto, Toronto, Ontario Correspondence: Dr Philip WH Peng, McL 2-405, Toronto Western Hospital, 399 Bathurst Street, Toronto, Ontario M5T 2S8. Telephone 416-603-5118, fax 416-603-6494, e-mail philip.peng@uhn.on.ca PWH Peng, DN Wijeysundera, CCF Li. Use of gabapentin for perioperative pain control – A meta-analysis. Pain Res Manage 2007;12(2):85-92.

D espite the use of new drugs and delivery modalities, studies have shown that acute postoperative pain continues to be undermanaged (1,2).Approximately three of four patients experience acute pain after surgery, and 80% of these have moderate to extreme pain.Opioid analgesics are the cornerstone of pharmacological postoperative pain management (3), although they also contribute to increased morbidity and associated hospital costs (4,5).A balanced, multimodal approach of providing postoperative analgesia has been suggested to improve analgesia and minimize opioid-related adverse events (6).Various agents or techniques have been well investigated (7), including acetaminophen (8), nonsteroidal anti-inflammatory drugs (9), cyclooxygenase 2 inhibitors (10), ketamine (11) and regional anesthetic techniques (12).
Gabapentin, an anticonvulsant widely used for the treatment of various neuropathic pain syndromes, has recently been investigated as an analgesic agent in the perioperative setting.Although it is not an analgesic itself, early experience suggests this medication can improve analgesia and reduce opioid requirements (13).An important component of the opioid-sparing effect is to demonstrate a reduction in opioid-related side effects (14).The sample size estimates of all the perioperative gabapentin studies were based on analgesic effectiveness and reduction in opioid consumption.
Thus, these studies may not have the power to detect differences in side effects.Similarly, these studies may not have the power to detect possible increased rates of side effects of gabapentin, such as dizziness, which may impact on postoperative recovery.
The objective of the present study was to examine the analgesic effectiveness, opioid-sparing effects and side effects associated with the perioperative use of gabapentin.

METHODS
The present meta-analysis adhered to the recommendations of the Quality of Reporting of Meta-analyses recommendations (15).

Inclusion and exclusion criteria
Eligible studies were parallel-group, randomized, controlled trials enrolling adults (18 years or older) undergoing surgery under general anesthesia, regional anesthesia and monitored anesthesia care.The interventions assessed were the preoperative (within 24 h), intraoperative or postoperative (within 24 h) administration of gabapentin.Trials had to report any of the following outcomes: cumulative analgesic consumption (24 h), visual analogue scale (VAS) pain scores or adverse effects (dizziness, sedation, respiratory depression, nausea, vomiting, pruritus or urinary retention).Studies incorporating a local anesthetic technique or nerve block as part of the anesthetic regimen were excluded.

Search strategy
Eligible trials were identified using MEDLINE (1966  Register, TEXTMED, Science Direct, IngentaConnect and Google Scholar.The search used the following subject headings and text terms: "gabapentin", "Neurontin", "postoperative care", "postoperative period", "perioperative care", "surgery", "pain treatment", "analgesic" and "analgesia" (16).Bibliographies of included articles and published reviews were also searched.No language restrictions were applied.Unpublished trials were not sought, but authors of included studies were contacted to provide additional data.Studies that were reported only as abstracts or letters were excluded.

Methods of review
All three authors (PP, DW, CL) independently performed literature searches and assessed all identified full papers for inclusion.Reasons for exclusion were documented for all excluded studies.The following data were abstracted by two authors (DW, CL) onto data abstraction forms: patient demographic data, surgery, treatments, follow-up, 24 h cumulative analgesic consumption, time to first analgesic, VAS pain scores (rest and movement at 2 h, 4 h, 12 h and 24 h after surgery), dizziness, sedation, respiratory depression, nausea, vomiting, pruritus and urinary retention.Outcome definitions used by the original researchers were accepted.When the data on pain and adverse effects were incomplete, or when the pain score was presented as a median rather than mean ± SD, the first authors of those articles were contacted for further information.Three reviewers also independently rated study quality on the basis of the adequacy of randomization (method reported), allocation concealment, double-blinding and follow-up.All discrepancies were resolved by consensus.

Statistical analyses
Analyses were performed using Review Manager 4.2.8 (Cochrane Collaboration, Denmark).Initially, heterogeneity was measured using the I 2 statistic.This value is the proportion of total variation that is explained by between-study variation as opposed to chance (17,18).Higher I 2 statistics imply more heterogeneity among pooled studies than would be expected by chance alone.In cases of low heterogeneity (I 2 =0%), data were pooled under the fixed effects model; otherwise, the random effects model was used (19).Statistical significance was defined by a two-tailed P≤0.05.
Binary outcomes were expressed as pooled relative risks (RR) with 95% CIs.The RR was chosen to aggregate binary data because of its easy interpretability and stability (in contrast to risk difference and number needed to treat) with varying baseline risks (20).Pooled effects on time to first analgesic were expressed as weighted mean differences.The pooling of data on 24 h cumulative analgesic use was complicated by the differing postoperative analgesic regimens in the included studies.Aggregating data as weighted mean differences entailed standardizing all analgesic doses (ie, conversion to 'morphine-equivalents'), which was not possible given that a wide variety of analgesic agents was used and that some of them, such as tramadol, had no reliable equivalent ratio.Treatment effects on cumulative 24 h opioid use were therefore expressed as the more clinically meaningful 'ratio of means'; namely, the ratio of mean analgesic use in the gabapentin versus control arms.The methods of Friedrich et al (21) were used to calculate the pooled ratios of means.Treatment effects on mean VAS pain scores at rest and movement were calculated as both ratio of means and weighted mean differences at four prespecified time points after surgery: 2 h, 4 h, 12 h and 24 h.

Sensitivity analyses
Sensitivity analyses were planned a priori to determine whether the choice of included studies influenced pooled treatment effects.The first sensitivity analysis assessed the relationship between study quality and treatment effects.The meta-analyses were repeated in high-quality studies, which were defined as studies fulfilling all of the four criteria: adequate randomization with method reported, proper allocation concealment, double-blinding and accounting for dropouts.The second sensitivity analysis involved using funnel plots to assess for publication bias (22).The final sensitivity analysis assessed the relationship between severity of postoperative pain and treatment effects.Previous studies have suggested that VAS pain scores must exceed 30 mm for a study to detect analgesic effects (23).The meta-analyses were therefore repeated in studies that reported mean VAS pain scores at rest above 30 mm in the control arm at any of the prespecified time points (2 h, 4 h, 12 h or 24 h).
The most common dose of gabapentin assessed was 1200 mg daily (12 studies), with some studies using doses as low as 300 mg daily (Table 1).Eleven studies (25,(28)(29)(30)(31)(32)(33)36,(38)(39)(40) administered gabapentin as a single dose within 1 h to 2 h before surgery; the remainder involved initiating therapy on the day before surgery or continuing it for up to 10 days after surgery (Table 1).With the exception of Rorarius et al (29), all studies used an inactive placebo (Table 1).The study that used an active control was excluded from analyses pertaining to sedation and respiratory depression.

Adverse effects
Gabapentin was associated with a significant increase in dizziness (RR 1.40, 95% CI 1.06 to 1.84), with minimal heterogeneity (Table 3).It was also associated with an increase in sedation of borderline significance (RR 1.65, 95% CI 1.00 to 2.74), albeit with significant heterogeneity (I 2 =83.3%).With regard to the opioid-sparing effects, gabapentin significantly reduced vomiting and pruritus (Table 3).The heterogeneity observed in analyses pertaining to sedation and nausea was not explained by subgroup analyses based on surgical procedure, gabapentin dose or study quality.

Sensitivity analyses
Aside from pain with movement at 12 h and 24 h after surgery, the results were not affected by study quality or severity of postoperative pain (Table 4).Funnel plots revealed no obvious publication bias with regard to opioid consumption, pain scores or adverse effects.

DISCUSSION
Gabapentin was introduced initially by Pfizer Inc in the early 1990s as an anticonvulsant, but is better known as an effective treatment for neuropathic pain (43).It is not considered an analgesic per se.A study published in 2002 (25) prompted investigators to rethink that gabapentin may be a broad-spectrum analgesic (44).Since then, 19 studies have been published.In the present meta-analysis, we have shown that gabapentin reduces pain scores, both at rest and with movement following various surgeries, lengthens the time for analgesic rescue, decreases the consumption of opioids and, most importantly, lowers rates of opioid-related side effects.The pain score at rest was reduced by 27% to 39% (VAS 7.2 mm to 14.3 mm on a scale of 0 mm to 100 mm) during the first 24 h, and the pain score with movement was reduced by 18% to 28% (VAS 8.2 mm to 10.2 mm) in the first 12 h.
At the time of publication, three meta-analyses on this topic have been published (45)(46)(47).However, we have performed the most extensive search, and the number of studies included in this meta-analysis was 18, compared with eight (45), 12 (46) and 16 (47) studies in the other meta-analysis.Furthermore, we were able to contact all except two authors for further information.Ho et al (47) excluded the data on pain score in four studies because the pain score was presented as a median or was not presented.However, we were able to get those data directly from the authors.As a result, we had the most extensive data collection for analysis of side effects, pain scores at rest and pain scores associated with movement.The latter were not presented in any of those three metaanalyses.
Our findings led to the question of how gabapentin works in reducing postoperative pain.In the past few decades, there has been a substantial improvement in the understanding of the mechanism of surgical pain (48)(49)(50).The neural pathway is no longer viewed as a 'hard-wire system' -it demonstrates neural plasticity, the neurobiological means by which changes in the nervous system can modulate responses to any stimulus.Following surgery, the intense influx of pain signal from tissue trauma can amplify and modify pain through a process called sensitization (central and peripheral sensitization).Of particular interest is central sensitization, in which the excitability and responsiveness of dorsal horn neurons to pain transmission is enhanced (50).The results of these changes contribute to abnormal pain responses such as pain sensation to innocuous stimuli (allodynia), enhanced pain responses (hyperalgesia) and spread of hypersensitivity beyond injured tissue.Because some of these modulation processes may turn into modification such as enhancement of gene expression, the pain may persist beyond apparent tissue healing, contributing to chronic postsurgical pain syndromes (51,52).Despite extensive investigations, the mechanism of how gabapentin works is still unknown.Although it is an analogue of gamma-aminobutyric acid (GABA), it is neither active at GABA A or GABA B receptors nor metabolically converted to GABA (53,54).Its known main site of action is the alpha 2 -delta subunit of presynaptic voltage-gated calcium channels (55).However, other targets, such as sodium channels, N-methyl-Daspartic acid receptors, monoaminergic and opioid pathways, have been implicated (53,54).Gabapentin does not affect the calcium ion influx of normal neurons.It reduces calcium ion influx at presynaptic terminals in hyperexcited neurons (51), which may lead to reduction of release of excitatory neurotransmitters such as glutamate, substance P and noradrenaline.Thus, gabapentin appears to reduce the hyperexcitability of dorsal horn neurons that is induced by tissue damage.
In the acute pain setting, a meaningful improvement in pain score has been investigated.In patients with moderate baseline pain, the minimal improvement, as seen on the VAS, is 13 mm on a scale from 0 mm to 100 mm, or a 20% reduction from baseline (56).Although the improvements in pain scores in the present analysis were modest (7.2 mm to 14.3 mm at rest during the first 24 h, and 8.2 mm to 10.2 mm in the first 12 h with movement), these reductions represented more than 20% reduction from the baseline pain score at rest.The improvement in analgesia is far better than those achieved with ketamine (11) and is comparable with cyclooxygenase 2 inhibitors (41).
Despite the benefits of gabapentin, one has to be cautious about its side effects.The two most common side effects of gabapentin in the management of chronic pain are somnolence and dizziness (43).In this meta-analysis, gabapentin also increased the rates of dizziness and sedation in surgical patients.Similar to most acute pain studies, the quality of reporting of side effects was generally poor (57).For instance, most included gabapentin studies did not describe the methods for reporting the occurrence and severity of side effects.For studies that reported dizziness, only three assessed its severity (25,29,34).Most of the reported dizziness episodes were mild (25,29) and did not affect patient satisfaction (42).Furthermore, the most common dose regimen was a single dose of 1200 mg before surgery (25,(28)(29)(30)(31)(32)34,38,42), which is a very high initial dose compared with the recommended dose for the management of chronic neuropathic pain (58).In the chronic pain setting, the starting dose usually does not exceed 300 mg daily to avoid dizziness and sedation.The dose is then gradually titrated upward to the desired therapeutic dose.However, adjusting the dose in the preoperative period is impractical, and thus, a higher dose is chosen.Interestingly, a dose-response study suggested no benefit of a starting dose above 600 mg (39).Further studies are required to elucidate the optimal dose and severity of side effects.Despite this, physicians should be cautious about the potential for sedation and dizziness, especially in ambulatory surgical patients.

Limitations
There are several limitations in this meta-analysis, mostly related to the methodological weakness of the original studies.Although we used the more conservative random effects model to aggregate data from the gabapentin randomized, controlled trials, the underlying heterogeneity among the included trials limited the strength of our overall conclusions.This called for the need of a properly designed, randomized, controlled trial.The results of the present meta-analysis provide valuable data for improving the design of future trials.We are not able to comment on the dose-response for efficacy because of the diverse regimen used.Before surgery, 15 studies prescribed a single dose (300 mg to 1200 mg) and three studies prescribed repeated doses.Following surgery, only four studies prescribed repeated doses, varying from 1200 mg once daily for two days to 400 mg three times a day for 10 days.Only two studies (26,39) examined the analgesic effects of gabapentin in more than one dose group.The other limitation is related to the reporting of side effects.Most studies mainly focused on analgesic effectiveness and did not systemically evaluate adverse effects.Reporting of the methods of assessment, severity of adverse effects and patient withdrawal related to the side effects is recommended in analgesic trials (59).

Suggestion for further studies
Further studies should improve the reporting of side effects and their severity (57).When evaluating the efficacy of a drug, the frequency, severity and the impact (distress) of different adverse effects should be considered.A composite scoring system, the Opioid-related Symptoms Distress Scale, has been suggested for reporting the opioid-related side effects (59).
So far, there is insufficient information to recommend a specific dose regimen for gabapentin.Only one study examined dose-response efficacy and suggested that the highest efficacy was a single preoperative dose of 600 mg (39).However, analgesic effects can be procedure specific (60,61), and more studies are needed to contribute to a dose recommendation.
Gabapentin has been shown to reduce mechanical hyperalgesia in several preclinical pain models (62)(63)(64).However, studies showing that it reduces hyperalgesia in postoperative patients are lacking.Further studies are recommended to examine the effects of gabapentin on mechanical hyperalgesia on and around the wound.This will give valuable information of how gabapentin works in the acute pain setting.
There are other advantages of using gabapentin in the postoperative setting.It has been used to relieve anxiety and improve sleep quality, both of which are important to patient care in the perioperative period (65,66).However, only one study examined the effect of gabapentin on relieving anxiety.Further studies are recommended to explore these clinical advantages of gabapentin with well-validated measurement tools.
Finally, antisensitization medications such as ketamine have been demonstrated to reduce the prevalence of chronic postsurgical pain (67).More studies are encouraged to examine the effects of gabapentin on the prevalence and severity of chronic postsurgical pain in different surgical models.

CONCLUSION
Gabapentin may be a new class of analgesic medication that targets the sensitization process.In our meta-analysis, we were able to show that gabapentin improves the analgesic efficacy of the opioids both at rest and with movement, and reduces opioid consumption and opioid-related adverse effects.However, it also increases rates of dizziness and sedation.Further studies are required to study the impact of these side effects, the optimum analgesic dose and the other nonanalgesic effects of gabapentin, such as anxiolysis and sleep improvement.

Figure 3 )Figure 2 )
Figure3) Per cent reduction in visual analogue scale (VAS) pain scores with movement by gabapentin over the first 24 h following surgery.The closed circles are the point estimates for pooled treatment effects, expressed as ratios of means (random effects model) at 2 h, 4 h, 12 h and 24 h following surgery.The error bars are the corresponding 95% CIs.The table below the graph reports the heterogeneity (I 2 statistic) for these treatment effects.The text adjacent to each point estimate is the same treatment effect expressed as the absolute reduction in VAS scores (weighted mean difference in random effects model) with 95% CI ACKNOWLEDGEMENTS: DN Wijeysundera is supported in part by the Canadian Institutes of Health Research (Ottawa, Ontario).

TABLE 2
*See reference 68 for details.+ Information present; -Information absent

TABLE 1
Characteristics of the studies included in the present meta-analysis (29)itional two arms that assessed celecoxib (alone and in conjunction with gabapentin) were excluded; † Additional arm that assessed mexilitene was excluded *One study(29)that used an active control (oxazepam 15 mg) was excluded from this analysis

TABLE 4
Sensitivity analyses based on study quality and severity of postoperative pain -the second sensitivity analysis involved repeating analyses in studies that reported a mean control arm visual analogue scale (VAS) pain score at rest ≤ ≤ 30 mm at any prespecified time point