Antidepressant Drugs for Prevention of Acute and Chronic Postsurgical Pain: Early Evidence and Recommended Future Directions

WITH well more than 40 million surgeries annually in North America alone, postoperative pain causes considerable morbidity and substantially impacts healthcare utilization.¹  Postoperative pain is mediated at multiple neural sites and via multiple mechanisms.²  Thus, different analgesics can only partially reduce postoperative pain. A multimodal analgesic approach is commonly used^3,4 ; however, currently using agents such as opioids,⁵  local anesthetics,⁶  nonsteroidal antiinflammatory drugs,⁷  acetaminophen,^8,9  ketamine,¹⁰  and gabapentin/pregabalin^11–13  have various limitations. Many other agents have been evaluated for efficacy, but evidence has not warranted their routine use. Thus, a continued search for safer, more effective agents for postoperative pain is needed.

Antidepressants are commonly used for various chronic pain conditions^14,15  and are classified according to chemical structure and/or mechanism of action. The most common classes of antidepressants included the following: (1) tricyclic antidepressants, (2) selective serotonin reuptake inhibitors, and (3) serotonin and norepinephrine reuptake inhibitors.¹⁶  Tryptophan, with previously demonstrated antidepressant efficacy,¹⁷  has also been evaluated for postoperative analgesia. Fluradoline has been classified as an antidepressant based on its tricyclic chemical structure and has also been studied in postoperative pain.¹⁸  In addition to serotonin and norepinephrine reuptake inhibitors, tricyclic antidepressants antagonize peripheral sodium channels and spinal N-methyl-d-aspartate receptors.¹⁵  These mechanisms serve to suppress central sensitization which is important in the pathophysiology of acute postoperative pain.^19–21 

Recent studies on antidepressants for postoperative pain have generated, possibly premature, enthusiasm for this potentially new indication. There is a great need for improved treatment options in the management of postoperative pain,²²  and antidepressants could potentially be a valuable addition here. However, safety problems including increased perioperative bleeding,^23–25  serotonin syndrome,²⁶  and other known adverse drug interactions necessitate a rigorous assessment. Thus, this review evaluates efficacy and safety of antidepressants from trials in acute postoperative pain. Prevention of chronic pain after surgery is an emerging goal with fundamental distinctions from acute postoperative pain.^27–29  However, given that studies evaluating the treatment of acute, and prevention of chronic, postoperative pain are conducted in similar perioperative settings, we will also review trials in postoperative chronic pain prevention.

This systematic review was conducted according to guidelines published in the Cochrane Handbook for Systematic Reviews of Interventions.

Given the absence of any previously published reviews of antidepressants for postoperative pain, we conducted a very broad literature search for studies with the following inclusion criteria:

• Placebo-controlled, double-blind, randomized trials (≥10 patients per treatment arm)

• Systemic perioperative administration of an antidepressant agent

• Adults (>18 yr)

• Study patients experiencing pain after any surgical procedure

• Methods included a measure of pain

Validated measures of pain intensity—at rest or with movement—or pain relief assessed during the postoperative period. Trials assessing early (<2 weeks postoperatively) and persistent (≥3 months postoperatively) pain were included, but analysis of early and persistent pain outcomes was to be conducted separately, wherever appropriate.

Treatment-emergent adverse effects, opioid-related side effects, and other outcomes including mood, sleep, and physical function assessed during the postoperative period.

Given the importance of reducing movement-evoked pain for postoperative functional recovery, each trial was evaluated for assessment of pain at rest versus movement-evoked pain.³⁰ 

The Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, CINAHL, and Web of Science (cited reference search from identified studies) were searched from the time of inception of each database until December 4, 2013. The specific search strategy used can be found in appendix 1.

The reference lists of studies that met inclusion criteria, as well as other relevant articles, were searched to identify further trials.

All the review authors made substantive contributions to the development, analysis, and interpretation of this review as well as drafting and approval of the final submission. Two authors (K.W. and I.G.) independently conducted the literature search, identified trials for inclusion, reviewed study quality and risk of bias, and performed data extraction. Between these two authors, no disagreements arose regarding inclusion or exclusion of trials from the review. However, there were some disagreements in ratings of trial quality and risk of bias (most frequently related to study descriptions of randomization and blinding methods) and all of which were resolved by discussion and thus obviating the need for a third adjudicator. All other authors reviewed the results of these judgments and commented as necessary, but no further disagreements arose from this.

The following data were extracted from each study, if available: (1) patient characteristics; (2) study drug, including dose, route, and timing of administration; (3) patient-reported pain intensity at baseline (physician-, nurse-, or care-giver–reported pain was not included in the analysis); (4) patient-reported pain relief expressed at least hourly over 4 to 6 h by using validated pain scales (pain intensity and pain relief in the form of visual analogue scale or categorical scales, or both); (5) patient global assessment of efficacy, using a standard categorical scale; (6) time to use and number of participants requiring rescue medication; (7) number of participants with one or more adverse events; (8) number of participants with a serious adverse event; and (9) number of withdrawals (all cause and adverse event related).

Risk of bias assessment was conducted on each study according to the Cochrane Risk of Bias Tool.³¹  Quality of each trial was assessed using the Oxford Quality Scoring System.³²  The scoring system was used as follows: One point each was scored if the study was randomized and double blind. One point each was scored if procedures for randomization and blinding were reported and appropriate. One point each was deducted if procedures for randomization and blinding were not appropriate. One point was scored if reasons for patient withdrawals and dropouts were described. Given that only randomized and double-blind studies were included, the lowest possible score is 2 and the highest is 5 for any included study.

The primary comparison of interest for this review was between study drug and placebo. Studies would be combined for meta-analysis if they evaluated the same study drug at roughly similar doses and durations of treatment (e.g., a study evaluating a single preoperative drug dose would not be compared with another study evaluating several weeks of treatment with the same drug) and used common outcome measures and time points. RevMan 5.1 (RevMan 2011)³³  was to be used to analyze study data for binary outcomes. Sensitivity analyses would be used to evaluate the robustness of a particular result by repeating primary analyses without any studies considered to be outliers with respect to study quality, drug dose and duration, or pain measurement scales.

Two authors (K.W. and I.G.) independently evaluated differences in participants, interventions, outcomes, study settings, and methodology. Where substantial subjective differences were judged to be present by both reviewers, clinical or methodological heterogeneity was considered to exist. If multiple studies were considered to be adequately homogenous with respect to these features, they would be further evaluated for the presence or absence of statistical heterogeneity.

Subgroup analyses would be performed to compare trial outcomes across different:

1. Surgical procedures

2. Timing of the intervention

3. Duration of intervention

Meta-analysis was to be conducted if the following conditions were met: identification of at least two relevant studies with a low risk of bias and absence of substantial heterogeneity.

Figure 1 describes the flow of this systematic review, which included 16 trials in total (appendix 2).^34–49 Table 1 (acute pain) and table 2 (chronic pain) describe the main features of included trials. Table 3 (acute pain) and table 4 (chronic pain) describe the main results of pain outcomes from included trials.

Table 5 describes the risk of bias of included studies and table 6 describes trial quality, assessment of rest versus dynamic pain, and assessment/reporting of adverse effects/events for the included studies.

Thirteen of the 16 included trials were of good to high quality, but 3 trials were missing important details regarding randomization and blinding methods. Although all studies assessed postoperative pain, only 3 of 16 studies acknowledged the distinction between pain at rest and during movement and these three trials assessed for pain during movement. Only 5 of 16 trials mentioned adverse effect assessment in their Methods section although 9 of 16 trials did provide some adverse effects reporting in their Results section.

Trials excluded from this review (17) and their reason(s) for exclusion are shown in appendix 3.

Inclusion criteria were met (appendix 2) by 15 heterogeneous studies (985 participants) of early postoperative pain involving different antidepressants including amitriptyline (3 trials),^39,42,48  bicifadine (2 trials),^46,49  desipramine (3 trials),^39,43,44  duloxetine (1 trial),⁴¹  fluoxetine (1 trial),⁴⁰  fluradoline (1 trial),⁴⁵  tryptophan (4 trials),^35,37,38,47  and venlafaxine (1 trial).³⁴  Tables 1 and 3 describe the 15 early postoperative pain studies included in the review. Taken together, the studies involving the four drugs that were evaluated with more than one randomized controlled trial (RCT)—amitriptyline, bicifadine, desipramine, and tryptophan—failed to meet our criteria for performing meta-analysis. Although superiority to placebo for pain outcomes was reported in 8 of the 15 included trials,^{34,35,39,43,45,47–49}  only 2 studies reported sufficient data to allow for estimation of standardized effect sizes which were 0.56 for amitriptyline⁴⁸  and 0.78 for fluradoline.⁴⁵ 

No trends in trial outcome were observed across this heterogeneous group of trials to suggest an effect of dose, duration, or timing of treatment. Most studies failed to identify a primary outcome measure and reported treatment group differences not necessarily based on trial primary outcomes. Included studies are described below according to a pharmacological classification.

Following previous small negative trials in third molar extraction⁴³  and hip/knee arthroplasty,⁴²  Vahedi et al.⁴⁸  randomized 200 patients undergoing single-level lumbar discectomy and laminectomy to a single dose of either 25 mg amitriptyline or placebo, 2 h before surgery. Visual analogue scale for pain intensity, pain relief from baseline, and morphine consumption were measured during 24 h, and significantly lower pain intensity was reported in the amitriptyline group at 24 h only.⁴⁸  It should be noted that only one trial of amitriptyline for acute postoperative pain was positive but also that a very narrow range of doses and treatment durations were evaluated.

Max et al.⁴⁴  reported no effect of single doses of desipramine on pain after a variety of different surgical procedures. In contrast to those results, Levine et al.⁴³  and Gordon et al.³⁹  demonstrated superior analgesia in patients receiving opioids after third molar extraction. However, in both of these trials, multiple repeated doses of desipramine were administered for 3 to 7 days before surgery. These are the earliest results to suggest that potential postoperative benefits of antidepressants—in this case, potentiation of postoperative opioid analgesia—may require several days of pretreatment before the surgical procedure.

Gordon et al.⁴⁰  investigated the interaction of fluoxetine with morphine, or pentazocine, in 70 patients undergoing third molar extraction. Patients were randomized to receive either fluoxetine or placebo 7 days preoperatively and either IV morphine or pentazocine postoperatively. No significant fluoxetine–placebo differences in opioid analgesia were reported.

Wang et al.⁴⁹  evaluated the efficacy of bicifadine compared with aspirin and placebo. A total of 100 patients after abdominal or orthopedic surgery were randomized to receive oral placebo, 75 or 150 mg of bicifadine, or 650 mg aspirin. Both bicifadine (at 150 mg only) and aspirin were superior to placebo for pain intensity difference and pain relief. Porter et al. conducted a similar trial, comparing bicifadine with codeine and placebo in patients undergoing lower limb orthopedic surgery. A total of 80 patients were randomized to receive placebo, 150 or 200 mg of bicifadine, or 60 mg of codeine in the immediate recovery period. No significant bicifadine–placebo differences were reported for pain outcomes.

A recent study in 50 patients after total knee arthroplasty by Ho et al.⁴¹  examined the analgesic effect of 60 mg of duloxetine administered preoperatively as well as on postoperative day 1. Pain scores and opioid consumption were evaluated during a 48 h period, as well as a follow-up at 3 and 6 months postoperatively to evaluate the proportion of patients with persistent postoperative pain. Although no significant duloxetine–placebo differences were reported for early postoperative pain outcomes, morphine requirements were significantly lower in the duloxetine group.

Amr and Yousef³⁴  evaluated venlafaxine and gabapentin in 150 patients after partial or radical mastectomy. Patients received venlafaxine 37.5 mg, gabapentin 300 mg, or placebo beginning on the preoperative evening and on each day for the first 10 postoperative days. Pain at rest and active states (visual analogue scale) and consumption of analgesics were measured up to postoperative day 10 with a follow-up at 6 months. When compared with placebo, although no difference in pain at rest was observed in the venlafaxine group, pain with movement was reduced on the 7th to 10th postoperative days. In addition, the use of postoperative analgesics, codeine and paracetamol, on postoperative days 2 to 10 was reduced in the venlafaxine-treated patients.

McQuay et al.⁴⁵  compared fluradoline to aspirin in 120 orthopedic patients with moderate to severe postoperative pain on postoperative day 1 and who were randomized to receive oral placebo, 150 mg fluradoline, 300 mg fluradoline, or 650 mg of aspirin. In this trial, both fluradoline and aspirin were superior for pain intensity and relief outcomes.

Ceccherelli et al.³⁵  evaluated intravenous tryptophan in 45 patients after cholecystectomy who were randomized to receive either placebo or 7.5 or 15 mg/kg IV tryptophan postoperatively. This trial reported significant pain reductions compared with placebo with tryptophan at either dose. Shpeen et al.⁴⁷  randomized 50 patients undergoing endodontic surgery to either 3 g of oral tryptophan or placebo 24 h before the procedure and also observed a significant analgesic effect of tryptophan. In contrast to these two positive trials, two other RCTs by Ekblom et al.³⁷  and Franklin et al.³⁸  failed to demonstrate any significant tryptophan–placebo differences.

Inclusion criteria were met (appendix 2) by three heterogeneous studies (565 participants) of chronic postoperative pain involving different antidepressants including duloxetine (one trial),⁴¹  escitalopram (one trial),³⁶  and venlafaxine (one trial).³⁴  Tables 2 and 4 describe the three chronic postoperative pain studies included in the review. In the duloxetine trial, duloxetine 60 mg was given orally 2 h before surgery and again on the morning of postoperative day 1 and pain was measured at 3 and 6 months after surgery. In the venlafaxine trial, venlafaxine 37.5 mg (or gabapentin) were given orally at bedtime starting the night before surgery and again daily for the first 10 days after surgery and pain was measured 6 months after surgery. In the escitalopram trial, escitalopram 10 mg orally was given daily starting from 2 to 3 weeks before surgery and continued daily up to 6 months after surgery. Clinical heterogeneity of trials with respect to drug, dosing regimen, outcome measure, and surgical procedure precluded any meta-analyses. In the venlafaxine mastectomy trial,³⁴  significantly lower pain was reported upon comparing venlafaxine to both placebo and gabapentin, and the standardized effect size for the venlafaxine–placebo comparison was estimated to be 0.16. In this trial,³⁴  10 days of perioperative study drug administration resulted in a significantly lower 6-month incidence of burning (1 of 50) and stabbing (7 of 50) pain compared with 11 of 50 and 20 of 50 for placebo, respectively.

Only 9 of the 16 included trials reported on adverse effects (table 6). In the trial by Kerrick et al.,⁴²  3 consecutive night-time oral doses of amitriptyline 50 mg, starting on the first night after surgery was associated with a slightly higher (nonsignificant) level of sedation compared with placebo-treated patients. In both bicifadine RCTs by Porter et al.⁴⁶  and Wang et al.,⁴⁹  no significant placebo–bicifadine differences in adverse effects were reported; however, very few details were provided. In the postoperative trial by Max et al.,⁴⁴  measures of sedation and nausea were not significantly affected by desipramine, compared with placebo. Other reported side effects (dry mouth, itching, euphoria, and dizziness) were not statistically compared between desipramine and placebo. No significant adverse effects, compared with placebo, were reported in the single-dose duloxetine trial by Ho et al.,⁴¹  or by the 10-day multidose venlafaxine trial.³⁴  In one of the tryptophan trials, Ceccherelli et al.³⁵  reported no significant differences in adverse effects compared with placebo. In the trial of fluradoline by McQuay et al.,⁴⁵  no significant increases in adverse effects were reported; however, significant increases in blood pressure were observed with the 300 mg dose. Finally, in a rather unique trial evaluating over 6 months of treatment with daily oral escitalopram,³⁶  starting 2 to 3 weeks before cardiac surgery and continuing to 6 months after surgery, reports of overall side effects were significantly more frequent with escitalopram (12.6%) compared with placebo (4.5%) and included diarrhea, constipation, nausea, shivering, somnolence, and tingling of extremities.

A trial registry search for ongoing studies relevant to this review yielded one comparative trial of gabapentin and amitriptyline in the setting of lumbar laminectomy and discectomy (trial status—recruiting; ClinicalTrials.gov Identifier: NCT01014520) and another trial evaluating the efficacy of escitalopram after total knee arthroplasty (trial status—completed; ClinicalTrials.gov Identifier: NCT01430520).

This systematic review revealed 15 RCTs of eight different classified antidepressant drugs for the treatment of acute postoperative pain and 3 RCTs of three different antidepressants for the prevention of chronic postoperative pain. Because of inconsistent results, limitations in the numbers of RCTs for each antidepressant drug, poor procedure specificity, and other limitations in trial size and assessment of clinically relevant outcomes, there is no sufficient evidence to support the clinical use of antidepressants—beyond controlled investigations—for the treatment of acute, or prevention of chronic, postoperative pain. However, the existence of 8 of 15 positive RCTs of antidepressants in the setting of acute postoperative pain suggests the need to further conduct higher-quality, more definitive trials that either confirm or refute the efficacy and clinical utility of antidepressants for this indication.

Overall, included trials were of good to high quality. Most common sources of bias included incomplete descriptions of trial methods with respect to randomization, blinding, and allocation concealment. With respect to potential bias associated with selective outcome reporting,⁵⁰  only five RCTs clearly defined, a priori, a primary outcome measure for the trial. Because small trial size (e.g., <50 patients per parallel treatment arm) may be another source of bias,⁵¹  the observation that only 2 of the 16 RCTs included in this review had 50 or more patients per arm suggests that the vast majority of perioperative antidepressant studies should be considered as smaller proof-of-concept trials rather than more definitive confirmatory trials. Only three RCTs distinguished the difference between pain at rest and pain evoked by movement in their Methods sections. This is important for two reasons: (1) because movement-evoked pain is generally 95 to 226% more intense than pain at rest, failure to control the condition (i.e., at rest vs. during movement) during which pain is assessed during a clinical trial could result in highly variable pain intensity measures within and across trial patients. This increased variability could decrease assay sensitivity and lead to false-negative trial results. (2) If only pain at rest is evaluated in future trials, lack of evidence on the effect of the study medication on movement-evoked pain (which is more severe and may have more functional impact) will limit the clinical relevance of the trial. Given the various potential safety problems associated with perioperative antidepressant use, it is concerning that only six RCTs included a description of safety assessment in their Methods section and only nine RCTs reported any data on adverse effects in their Results section.^52,53  Given the various clinical factors that can differ substantially across surgical procedures, evaluation of postoperative pain treatments is best done in a procedure-specific manner.⁵⁴  Because six of the included RCTs were not exclusive to one specific surgical procedure, greater variability in results from those trials could have further increased variability and reduced assay sensitivity, thus increasing the likelihood of a false-negative result. It should also be noted that except for desipramine in two RCTs of third molar extraction (which could not be combined for meta-analysis because of differences in dosing regimens), no single antidepressant drug was evaluated in more than one surgical procedure. The potential future utility of antidepressant drugs for postsurgical pain also requires the recognition of potential drug interactions with opioids⁵⁵  and other drugs commonly used in the perioperative period. Finally, we observed that 6 of the 15 acute pain treatment RCTs involved only a single dose of study medication suggesting evaluation of treatment during a rather narrow window in the postoperative period.

The search strategy for this review was rather broad and comprehensive, and we also searched for other references using a cited reference search. However, we cannot rule out the possibility that other studies eluded our search. Furthermore, we are unable to locate studies that have never been published. Given previous observations that negative RCTs are less likely to be published, this raises the possibility of publication bias⁵¹  and that there are more negative studies than those we identified. Given the heterogeneity of drugs, doses, and time of administration of the various studies included in this review, we elected not to produce a funnel plot to assess publication bias.

Recent preclinical studies provide some supportive evidence for the positive trials reported in this review and further reinforce the potential for analgesic efficacy of tricyclic antidepressants, serotonin, and norepinephrine reuptake inhibitors for postoperative pain.^56,57  Pharmacological mechanisms of these agents—in particular sodium-channel blockade and N-methyl-d-aspartate receptor antagonism^15,16 —likely have important antinociceptive effects in postoperative pain settings.^19–21  Because analgesic efficacy of antidepressant drugs becomes apparent after days to weeks of gradual dose titration in chronic pain settings, future postoperative analgesic trials may also require a similar duration of dose titration for days to weeks before surgery to demonstrate optimal results. Also, given the potential for adverse drug interactions with other concomitant drugs as well as increased risks of perioperative bleeding, future, more definitive trials should be safely conducted in carefully selected populations so as to avoid these problems. However, if results suggest more convincing evidence of analgesic efficacy, subsequent research will be needed to define appropriate indications and contraindications in surgical patients. Regarding future evaluation of antidepressants for the prevention of chronic postsurgical pain, recently developed methods^58,59  to identify patients at higher risk of this complication may facilitate this goal. Targeting the proposed intervention to patients at highest risk of chronic postsurgical pain, and thus greatest need of prevention, would provide stronger justification for anticipated adverse effects and could also decrease the required patient numbers. Finally, given continued uncertainty about the postsurgical time period during which chronic pain develops,^27–29  consideration should be given to evaluating the preventive effects of antidepressant drugs—given for a longer duration of administration, that is, to continue for days, or even weeks, after postsurgical hospital discharge.

On the basis of currently available studies, there is insufficient evidence to support the clinical use of antidepressants for the treatment of acute postoperative pain. Several positive trial results suggest the potential for therapeutic benefits of antidepressants in certain postoperative clinical settings. Multiple positive trials suggest the therapeutic potential of antidepressants, which need to be replicated. Given current limitations in postoperative pain treatment and the need to more rigorously explore the efficacy of antidepressant drugs, future studies could more definitively characterize the value of antidepressants in postoperative pain management. Future, higher-quality RCTs should address the need for optimal dosing, timing and duration of antidepressant treatment, trial size, safety evaluation and reporting, procedure specificity, and assessment of movement-evoked pain relevant to postoperative functional recovery.

The authors thank R. Andrew Moore, D.Phil. (Oxford University, Oxford, United Kingdom), and Phil Wiffen, M.Sc. (Oxford University, Oxford, United Kingdom), for useful comments made during the preparation of this review.

This work was supported, in part, by a research grant from the Canadian Institutes of Health Research (Ottawa, Ontario, Canada), grant no. MSH-55041.

Dr. Kalso has received consulting fees from Pfizer, Grunenthal, Janssen-Cilag, Orion-Pharmos, and Pharmaleads. Dr. Raja has received research support or consulting fees from Allergan, Alpharma, Schering-Plough, Medtronic, Pfizer, and QRx Pharma. Dr. Gilron has received support from Pfizer, Aventis Pharma, Novopharm, PharmaScience, Apotex, Merck-Frosst, Johnson & Johnson, Ortho-McNeill, and Janssen-Ortho and has received grants from the Canadian Institutes of Health Research (Ottawa, Ontario, Canada), Physicians’ Services Incorporated Foundation, and Queen’s University (Kingston, Ontario, Canada). The other authors declare no competing interests.

1. Postoperative pain.mp. or exp postoperative pain/

2. Antidepressant.mp or exp antidepressant agent/

3. 1 and 2

4. limit 3 to human

1. exp Pain, Postoperative/

2. exp monoamine oxidase inhibitors/or exp adrenergic uptake inhibitors/or exp serotonin uptake inhibitors/or exp serotonin agents/or exp serotonin receptor agonists/or exp antidepressive agents/or exp antidepressive agents, second-generation/or exp antidepressive agents, tricyclic/

3. 1 and 2

4. limit 3 to humans

1. Antidepressive agents or monoamine oxidase inhibitors or serotonin uptake inhibitors or norepinephrine uptake inhibitors and pain, postoperative (limit to humans)

1. “Postoperative Pain”

2. “Antidepressive Agents” OR “Antidepressive Agents, Second Generation” OR “Antidepressive Agents, Tricyclic” OR “Serotonin Agents”

3. 1 and 2

Amr YM, Yousef AA: Evaluation of efficacy of the perioperative administration of venlafaxine or gabapentin on acute and chronic postmastectomy pain. Clin J Pain 2010; 26:381–85.

Ceccherelli F, Diani M, Altafini L, Varotto E, Stefecius A, Casale R, Costola A, Giron GP: Postoperative pain treated by intravenous l-tryptophan: A double-blind study versus placebo in cholecystectomized patients. Pain 1991; 47:163–72.

Chocron S, Vandel P, Durst C, Laluc F, Kaili D, Chocron M, Etievent JP: Antidepressant therapy in patients undergoing coronary artery bypass grafting: The MOTIV-CABG trial. Ann Thorac Surg 2013; 95: 1609–18.

Ekblom A, Hansson P, Thomsson M: l-Tryptophan supplementation does not affect postoperative pain intensity or consumption of analgesics. Pain 1991; 44:249–54.

Franklin K, Abbott F, English M, Jeans M, Tasker R, Young S: Tryptophan–morphine interactions and postoperative pain. Pharmacology biochemistry and behaviour 1990; 35:157–63.

Gordon NC, Heller PH, Gear RW, Levine JD: Temporal factors in the enhancement of morphine analgesia by desipramine. Pain 1993; 53:273–76.

Gordon NC, Heller PH, Gear RW, Levine JD: Interactions between fluoxetine and opiate analgesia for postoperative dental pain. Pain 1994; 58:85–8.

Ho KY, Tay W, Yeo MC, Liu H, Yeo SJ, Chia SL, Lo NN: Duloxetine reduces morphine requirements after knee replacement surgery. BJA 2010; 105:371–76.

Kerrick JM, Fine PG, Lipman AG, Love G: Low-dose amitriptyline as an adjunct to opioids for postoperative orthopedic pain: A placebo-controlled trial. Pain 1993; 52:325–30.

Levine JD, Gordon NC, Smith R, McBryde R: Desipramine enhances opiate postoperative analgesia. Pain 1986; 27:45–49.

Max MB, Zeigler D, Shoaf S, Craig E, Benjamin J, Li SH, Buzzanell C, Perez M, Ghosh B: Effects of a single oral dose of desipramine on postoperative morphine analgesia. J Pain Symptom Manage 1992; 7:454–62.

McQuay HJ, Carroll D, Poppleton P, Summerfield RJ, Moore RA: Fluradoline and aspirin for orthopedic postoperative pain. Clin Pharmacol Ther 1987; 41:531–36.

Porter EJB, Rolfe M, McQuay HJ: Single dose comparison of bicifadine and placebo in postoperative pain. Curr Ther ResClin Exp 1981; 30:156–60.

Shpeen SE, Morse DR, Furst ML: The effect of tryptophan on postoperative endodontic pain. Oral Surg 1984; 58:446–49.

Vahedi P, Salehpour F, Aghamohammadi D, Shimia M, Lotfinia I, Mohajernezhadfard Z, Vahedi Y: Single dose preemptive amitriptyline reduces postoperative neuropathic pain after lumbar laminectomy and discectomy. Neurosurg Quarterly 2010; 20:151–58.

Wang RI, Johnson RP, Lee JC, Waite EM: The oral analgesic efficacy of bicifadine hydrochloride in postoperative pain. J Clin Pharm 1982; 22:160–63.