Secondary Logo

Journal Logo


The effect of triple vs. double nonopioid therapy on postoperative pain and functional outcome after abdominal hysterectomy

A randomised double-blind control trial

Gilron, Ian; Tu, Dongsheng; Dumerton-Shore, Deborah; Duggan, Scott; Rooney, Rachel; McGrath, Michael; Orr, Elizabeth

Author Information
European Journal of Anaesthesiology (EJA): April 2015 - Volume 32 - Issue 4 - p 269-276
doi: 10.1097/EJA.0000000000000190



The most predictable acute pain condition is postsurgical pain.1 Over the past 60 years, recognition of the limited efficacy and adverse effects of commonly used analgesic treatments for postoperative pain has led to the evolution of multimodal analgesia.2 Opioids administered via patient-controlled delivery systems continue to be an important modality in clinical situations where regional techniques are not utilised. However, coadministered nonopioid analgesics are increasingly recognised to reduce opioid-related side effects, in part, due to opioid dose sparing.3,4 Some evidence suggests that combining two different nonopioid analgesics, in addition to opioid analgesia, can further improve analgesia and reduce opioid-related adverse effects,5–7 which reinforces the longstanding practice of multimodal analgesia.2 Recent examples of randomised controlled trials (RCTs) demonstrating the additional benefits of double combinations of nonopioid analgesics include a dexamethasone/pregabalin combination for pain after tonsillectomy8 and a pregabalin/ketamine combination for pain after hip arthroplasty.9 Narrative reviews on postoperative multimodal analgesia often mention over half a dozen different classes of nonopioid agents [e.g. acetaminophen, NSAIDs, N-methyl-D-aspartate receptor (NMDA) antagonists, gabapentinoids, alpha-2 adrenergic agonists, systemic local anaesthetics and cannabinoids],10,11 but as yet, no RCTs have rigorously evaluated the added benefit of combining more than two nonopioids for postoperative pain. The purpose of this study was, therefore, to compare the efficacy of a triple combination of acetaminophen, a NSAID (meloxicam) and a gabapentinoid (gabapentin) to each of the three respective double combinations of these drugs (i.e. acetaminophen and meloxicam; acetaminophen and gabapentin; and meloxicam and gabapentin). The hypothesis of this trial was that a triple combination of nonopioid drugs (acetaminophen, meloxicam and gabapentin) would reduce postoperative evoked pain to a greater degree than any of the three respective double-drug combinations (acetaminophen and meloxicam; acetaminophen and gabapentin; or meloxicam and gabapentin)

Materials and methods

Ethical approval for this study (study code: ANAE-152-09; Federalwide assurance #IRB00001173) was provided by the Queen's University Health Sciences Research Ethics Board, Queen's University, Kingston, Ontario, Canada (Chair Dr Albert Clark) in March 2009. All enrolled study patients provided written informed consent. Eligible patients who were enrolled were aged at least 18 years with a BMI less than 40 kg m−2, American Society of Anesthesiologists’ (ASA) class I or II and scheduled for elective open abdominal hysterectomy, which could also include removal of cervix, ovaries, fallopian tubes and which could be for the treatment of either benign or malignant disease. Trial exclusions were as follows: hypersensitivity to any agents to be used in the study; persistent preoperative pain; daily intake or intake within 48 h prior to surgery of any glucocorticoid agents, NSAIDs or other analgesics, not including daily administration of 325 mg of aspirin or less for cardiovascular prophylaxis; evidence of substance or alcohol abuse; major psychiatric disorder; bleeding disorder; peptic ulcer disease; moderate to severe asthma; seizure disorder requiring treatment with an anticonvulsant; and language barrier to communication with research staff.

Study design

This was a single-centre, parallel, randomised, double-blind trial with four treatment groups (1 : 1 : 1 : 1 allocation ratio) involving the administration of acetaminophen 4 g 24 h−1 (1 g four times daily), meloxicam 15 mg 24 h−1 (once daily), gabapentin 1.8 g 24 h−1 (in divided doses three times daily; 1.2 g on day of surgery) (AMG); acetaminophen 4 g 24 h−1 (1 g four times daily), meloxicam 15 mg 24 h−1 (once daily); meloxicam 15 mg 24 h−1 (once daily), gabapentin 1.8 g 24 h−1 (in divided doses three times daily; 1.2 g on day of surgery); or acetaminophen 4 g 24 h−1 (1 g four times daily), gabapentin 1.8 g 24 h−1 (in divided doses three times daily; 1.2 g on day of surgery). Across all treatment groups, all patients received red (meloxicam or placebo) capsules, white (acetaminophen or placebo) capsules and grey (gabapentin or placebo) capsules as per a triple dummy design starting 1 h preoperatively on the day of surgery and continuing through to 10 p.m. on postoperative day 2. All similarly coloured capsules were identical across treatment groups so as to preserve double-blinding and the treatment group allocation schedule was prepared by a nonstudy pharmacist to preserve allocation concealment. At study commencement, a nonstudy pharmacist prepared a concealed, computer-generated random treatment allocation schedule that randomised these four treatment groups, in blocks of four, to a series of patient numbers. Enrolled patients were assigned the next consecutive patient number and the corresponding set of study medications were dispensed by the pharmacy. No one other than the nonstudy pharmacists were aware of any patient's treatment allocation during the trial. While in hospital, nursing staff administered and observed study medication consumption.

Trial protocol

All study patients received routine perioperative care. Preoperatively, enrolled study patients received study information from a research nurse and completed baseline pain, peak expiratory flow rate (PEFR) and timed up and go (TUG) test measurements. Participants subsequently received study medications 1 h prior to surgery. Intraoperatively, patients received a balanced anaesthetic with an intravenous (i.v.) induction agent, neuromuscular blocker, inhaled desflurane and nitrous oxide at the discretion of the attending anaesthesiologist (who was blinded to treatment group), i.v. fentanyl (within a dosage range of 3 to 5 μg kg−1 administered within the first 30 min of surgery) and i.v. morphine (within a dosage range of 0.05 to 0.15 mg kg−1 administered over 30 min, starting 30 min before the anticipated completion of surgery). No local or regional anaesthesia, any other analgesic drugs or any corticosteroids (other than dexamethasone 4 mg i.v. at the beginning of surgery) were allowed during surgery. For the prevention of postoperative nausea and vomiting, all patients received dexamethasone 4 mg i.v. at the beginning of surgery and ondansetron 4 mg i.v. 30 min before the anticipated completion of surgery. Following surgery, all study patients received patient-controlled analgesia (PCA) with i.v. morphine as the only nonstudy drug analgesic as per the routine protocols of the Kingston General Hospital Acute Pain Management Service. Patients with a previous history of intolerability to morphine could use hydromorphone as the PCA opioid drug. The PCA was commenced in the postanaesthetic care unit (PACU). Throughout their stay in the PACU, patients were asked whether they required pain relief. Following an affirmative response, patients received opioid medication (fentanyl, morphine or hydromorphone) in incremental doses from a nurse blinded to treatment group. Nurse-delivered opioid doses were administered as deemed necessary until patients were alert enough to use the PCA pump and all nurse-delivered opioid doses were included in the reporting of opioid consumption. The PCA pump was first set to deliver a 1.0 mg morphine bolus with a lockout interval of 5 min, with no continuous background infusion. If pain relief was inadequate at these settings at any time during the study, the protocol allowed for the adjustment of the PCA pump to deliver up to 2.0 mg and, if necessary, the lockout interval could be increased up to 10 min. The PCA pump could be continued on the patient's hospital ward for 48 h postoperatively with morphine as the only nonstudy drug analgesic to be used. All patients received nausea and vomiting prophylaxis with ondansetron 4 mg i.v. every 8 h for 36 h. If patients developed intolerable nausea or vomiting despite this regimen, the protocol allowed for the added administration of dimenhydrinate 25 to 50 mg i.v. every 4 h as needed. Itching could be treated with diphenhydramine 25 mg i.v. every 4 h as needed.


The primary outcome for this trial was cough-evoked pain intensity measured on the day of surgery (2, 4 and 6 h postoperatively and 8 a.m., 12 p.m. and 4 p.m. on postoperative days 1 and 2). All pain intensity measures in this study were rated using a 0 to 10 numerical rating scale (NRS), with 0 defined as ‘no pain’ and 10 defined as ‘worst pain imaginable’. Timing of outcome measures was selected, in part, to accommodate work schedules of research staff and the majority of surgeries were scheduled early in the day as per clinical routines of the Kingston General Hospital. Secondary outcomes included: pain intensity at rest (2, 4 and 6 h postoperatively and 8 a.m., 12 p.m. and 4 p.m. on postoperative days 1 and 2); pain intensity during sitting and with forced expiration (8 a.m., 12 p.m. and 4 p.m. on postoperative days 1 and 2); retrospective ratings of average nocturnal pain intensity from 10 p.m. on the previous day to 8 a.m. on the current morning on postoperative days 1 and 2; PEFR; total opioid consumption; time to PACU discharge; presence and severity (mild, moderate or severe) of self-reported adverse events (evaluated by open-ended questioning) including 12 opioid-related symptoms (nausea, vomiting, constipation, difficulty passing urine, difficulty concentrating, drowsiness or difficulty staying awake, feeling lightheaded or dizzy, feeling confused, feelings of general fatigue or weakness, itchiness, dry mouth and headache); time to bladder catheter removal; time to first urination after bladder catheter removal; time to first bowel movement; TUG test (assessed between 12 p.m. and 4 p.m. on the afternoon of postoperative day 2); modified Brief Pain Inventory (mBPI) (assessed between 12 p.m. and 4 p.m. on the afternoon of postoperative day 2); time to discharge from hospital; time to return to regular activities; and time return to work (in those working outside of the home). At 30 days after surgery, telephonic follow-up calls were made to assess persistent postoperative surgical pain intensity (present rest pain, present cough pain and worst pain in past week). Study outcome measures were evaluated and recorded by the research study nurses who were blinded to treatment group assignments. At each designated time point during the trial, pain intensity was measured as per the following progression, which took approximately 15 min to complete: pain at rest; pain evoked by sitting, in a standardised fashion, from the supine position, followed by a 120 s rest period; pain evoked by peak expiration using a peak flow meter followed by a 120 s rest period; and pain evoked by cough. Reporting of opioid consumption included i.v. opioids administered by nurses in the PACU, PCA opioid consumption and nurse-administered opioids on the hospital ward. Opioid consumption was reported by converting all administered opioids to i.v. morphine equivalents using standard opioid conversions.12

Statistical analysis

Statistical analyses were conducted by the trial's statistician (D.T.). Sample size calculation was based on the null hypothesis of no pain difference across all groups. Our sample size estimation was based on previous estimates of variability in pain scores following hysterectomy,5 and using the Bonferroni method to adjust for three multiple pairwise comparisons (i.e. AMG vs. acetaminophen and meloxicam, AMG vs. meloxicam and gabapentin and AMG vs. acetaminophen and gabapentin) resulted in an estimate of 36 patients per treatment group to detect a mean difference between treatment groups of 1.5 cm on a 0 to 10 NRS with a power of 0.8 and an alpha level of 0.05. The primary analysis compared cough-evoked pain across the following three pairwise comparisons: AMG vs. acetaminophen and meloxicam; AMG vs. meloxicam and gabapentin; and AMG vs. acetaminophen and gabapentin. Secondary analyses compared study outcomes across the following three pairwise comparisons: meloxicam and gabapentin vs. acetaminophen and meloxicam; meloxicam and gabapentin vs. acetaminophen and gabapentin; and acetaminophen and gabapentin vs. acetaminophen and meloxicam. Repeated continuous measures, such as the primary outcome measure of cough-evoked pain and all other pain measures, morphine consumption and PEFR were analysed using a linear mixed model with terms of treatment and interaction between treatment and time being first fitted with the data.13 If the interaction term was significant, the treatment groups were then compared at each time point by testing appropriately formed contrasts of the model parameters. In all other cases, a model with treatment variable only was refitted. Continuous data measured at one single point, such as TUG test time, pain on day 30 and pain interference were analysed using nonparametric analysis of variance (ANOVA) with the Wilcoxon test as a special case when there were two groups.14 Proportion data such as frequency of adverse events were analysed using Fisher's exact method.15 For all analyses, the three pairwise comparisons at 0.05 level were only made if the overall test for the difference among all groups was significant at the 0.05 level. This Fisher's least significant difference (LSD) method provides protection for multiple comparisons.16 The original trial protocol stipulated that study data were to be analysed once upon completion of 144 study patients with no planned subgroup analyses. As discussed in the Results section below, an unplanned interim analysis led to the premature discontinuation of the trial. Exploratory correlational analyses were conducted to evaluate relationships between pain (at rest and with movement) and postsurgical functional outcomes as measured by the TUG test and interference subscales of the mBPI. As a secondary goal of this trial, we conducted exploratory analyses to explore the relationship between pain (at rest and also evoked by sitting, forced expiration and coughing) and postsurgical functional outcomes (TUG test and pain interference subscales of the mBPI). For the relationship between pain measures and functional outcomes (TUG and mBPI), each pain measure, for each patient, was averaged over the three timepoints (8 a.m., 12 p.m. and 4 p.m.) on postoperative day 2, as the mBPI and TUG test were measured between 12 p.m and 4 p.m. on this day. Thus, scores for each pain measure (rest, sitting, forced expiration and coughing) were separately plotted against TUG test outcome and each interference subscale of the mBPI. Correlation analyses were performed using all available data from trial patients regardless of treatment group assignment and using Pearson correlation coefficients to assess the relationship between each pain measure (rest, sitting, forced expiration and coughing) and the functional outcomes of interest (TUG and each interference subscale of the mBPI). Given that these were treated as exploratory analyses, all statistical tests were considered significant if they met a threshold of P value less than 0.05.



Figure 1 shows the patient flow through the trial, which was conducted between November 2009 and July 2013. During the course of the trial, the emerging clinical practice of postoperative transversus abdominis plane (TAP) injections17 for rescue analgesia on the day of surgery led to an increasing number of trial dropouts due to protocol violations. This situation led to challenges with trial patient retention and, therefore, an unplanned interim analysis (after enrolment of 87 patients) was conducted in order to appropriately plan for trial continuation and timing of completion. Results of the interim analysis led to premature discontinuation of the trial. Regardless of the timing of trial exit, all patients receiving study drug were tracked for at least 30 days for adverse events and major adverse reactions. Outcomes were included for analysis from all patients until their withdrawal from, or completion of, the trial. Table 1 describes the characteristics of patients included in the analysis.

Fig. 1
Fig. 1:
CONSORT flow diagram. AG, acetaminophen-gabapentin; AM, acetaminophen-meloxicam; AMG, acetaminophen-meloxicam-gabapentin; DOS, day of surgery; MG, meloxicam-gabapentin; PACU, postanaesthesia care unit; POD, postoperative day; TAP, transversus abdominis plane.*Patients included in interim analysis contributed pain outcome data for at least one postoperative timepoint on day of surgery. All patients receiving at least one dose of study drugs were included in safety analysis.
Table 1
Table 1:
Demographic characteristics of study patients

Interim analysis results

Results of the interim analysis conducted after enrolment of 87 patients demonstrated no significant differences between treatment groups from the day of surgery through to postoperative day 2 for the primary outcome of cough-evoked pain (P = 0.45 between 4 groups, Fig. 2 and online supplement Appendix 1, as well as for pain at rest (P = 0.19), pain evoked by sitting (P = 0.13) and pain evoked by a forced expiratory manoeuvre (P = 0.36). There were no statistically significant differences in the percentage of patients with moderate to severe cough pain.

Fig. 2
Fig. 2:
Mean and standard error of cough-evoked pain intensity scores for each of the four treatment groups in the trial. AG, acetaminophen-gabapentin; AM, acetaminophen-meloxicam; AMG, acetaminophen-meloxicam-gabapentin; DOS, day of surgery; MG, meloxicam-gabapentin; NRS, numerical rating scale; POD, postoperative day.

Appendix 2 (online supplement, shows the calculated conditional powers for the three primary comparisons (AMG vs. acetaminophen and meloxicam; or meloxicam and gabapentin; or acetaminophen and gabapentin) for the primary endpoint, cough-evoked pain intensity, at each time point. These results indicate that, based on the trend of difference, there would be almost no chance of reaching a statistically significant difference between AMG and all the other three treatment groups if the study had recruited to full sample size. In terms of any trends suggesting superiority of the triple-therapy group, these were not consistent over time nor did the triple therapy group consistently appear to be superior to the other double therapy groups. There were also no significant differences between treatment groups from the day of surgery through to postoperative day 2 for pain at rest or pain evoked by sitting or peak expiration (online supplement Appendix 1, as well as postoperative opioid consumption (morphine or hydromorphone; Fig. 3 and online supplement Appendix 3, and PEFR, TUG test results, mBPI interference subscales and pain intensity at 30 days after surgery (online supplement Appendix 4, There were no significant treatment group differences for nocturnal pain intensity (P = 0.29), time to first urination (P = 0.16), time to hospital discharge (P = 0.18) or time to return to work (P = 0.32). The most commonly reported postoperative treatment side effects during the study included drowsiness/sedation, dizziness/lightheadedness and nausea. Appendix 5, indicates no significant differences between treatment groups for the incidence of side effects.

Fig. 3
Fig. 3:
Mean and standard error of opioid consumption (mg) for each of the four treatment groups in the trial. DOS, day of surgery. AMG, acetaminophen-meloxicam-gabapentin; AM, acetaminophen-meloxicam; AG, acetaminophen-gabapentin; MG, meloxicam-gabapentin; POD, postoperative day.

The time difference (increase) from baseline in the TUG test was significantly and positively correlated with all three evoked pain measures as well as pain at rest (Table 2). Pain interference (from the mBPI) with general activity, walking and sleep was also significantly and positively correlated with all three evoked pain measures as well as pain at rest (Table 2). Only pain evoked by forced expiration was significantly and positively correlated with pain interference with mood and none of the pain measures were significantly correlated with mBPI pain interference in relation to others (Table 2).

Table 2
Table 2:
Correlations between pain and timed-up-and-go test and modified Brief Pain Inventory outcomes


This trial failed to meet necessary recruitment targets and also failed to demonstrate the superiority of a gabapentin/meloxicam/acetaminophen combination over each of the three respective double-drug combinations. To the best of our knowledge, this is the first double-blind, RCT comparing a triple nonopioid analgesic combination with each of its three respective double-drug combinations. Clinical implementation of a new analgesic modality (TAP nerve blocks) challenged the conduct of this clinical trial, which prohibited any nonstudy drug analgesic interventions. In response to this unanticipated change in practice that was delaying trial completion, we conducted an unplanned interim analysis in order to plan trial management appropriately. Results of this interim analysis (following enrollment of 87 patients) indicated no significant differences between the triple nonopioid regimen (AMG) and any of the three double regimens (acetaminophen and meloxicam; meloxicam and gabapentin; acetaminophen and gabapentin). Furthermore, estimation of conditional powers suggested almost no chance of demonstrating superiority of the triple combination over all three double combinations.

Although these results fail to support the rationale for an acetaminophen/meloxicam/gabapentin combination, this is not necessarily a conclusive result and future studies should be conducted to replicate this incomplete study and to further explore the merits of triple analgesic combinations. One possible explanation for the lack of evidence supporting improved analgesia with a triple combination is that there could be some partial overlap of analgesic mechanisms, meaning that the cumulative analgesic effect is less than the sum of the analgesic effect of each individual agent. In our previous postoperative analgesic combination trials, combining gabapentin with rofecoxib5 and gabapentin with meloxicam7 did not appear to result in a fully additive analgesic effect. Another possible explanation could relate to interactions with concomitantly administered PCA. As previously highlighted by Kissin,18 some sedating analgesic agents (e.g. gabapentin) might reduce opioid consumption not only by reducing pain and analgesic requirements but also possibly by causing sedation and interfering with the ability of the patient to press the PCA handset.

In this small study, all four treatments were relatively well tolerated with no trend of more frequent adverse event rates in the triple combination group. However, previous literature suggests that, in various different settings, polypharmacy can be associated with higher rates of adverse drugs interactions, drug errors and drug-related adverse events.19,20 Therefore, the potential risks of more frequent adverse events with the addition of more classes of analgesic drugs in the postoperative setting emphasise the need for stronger evidence supporting various multidrug regimens.

This trial involved the assessment of various secondary functional outcomes and thus provided the opportunity to evaluate the impact of spontaneous vs. evoked pain on these outcomes. In our previous studies, we reported that pain was significantly correlated with PEFR after hysterectomy21 and laparoscopic cholecystectomy.22 In the present study, exploratory analyses suggested that more intense postoperative pain was associated with slower walking (TUG test) and that greater pain was associated with more interference with general activity, walking and sleep (mBPI). These observations further expand the apparent impact of postoperative pain on functional outcomes assuming that those outcomes are largely related to pain. Whereas we previously observed that pain after hystertectomy was more consistently correlated with PEFR than pain at rest, the present study suggests that both evoked and spontaneous pain were similarly correlated with outcomes on the TUG test and mBPI. One possible explanation for this is our observation that pain at rest is correlated with pain evoked by movement. However, this does not rule out the possibility that pain treatments that selectively reduce evoked pain may also improve functional outcomes. For example, we previously reported on a clinical trial of the experimental AMPA (2-Amino-3-hydroxy-5-Methyl-4-isoxazole-Proprionic Acid)/kainate antagonist, LY293558, which had little effect on spontaneous pain, although it significantly reduced movement-evoked pain after third molar surgery.23 More recently, preclinical investigations of the DNA transcription factor decoy, AYX1, demonstrated improved weight-bearing and spontaneous rearing in a rat knee surgery model.24

The strengths of our study design are that it was a comprehensive comparison between a triple nonopioid analgesic combination and all three of its respective double-drug combinations included a systematic assessment of pain at rest as well as pain evoked by movement. However, this study is also limited by incomplete recruitment as well as concomitant patient-controlled opioid administration, which may equalise appreciable differences in efficacy of the four studied treatment groups.

In conclusion, we report the results of a clinical trial interim analysis suggesting almost no chance that a combination of acetaminophen, meloxicam and gabapentin is superior to all three of the respective double-drug combinations. Demonstration of significant correlations between postoperative pain and an expanded set of functional outcomes (TUG test and mBPI) further reinforce the rationale to develop new treatments, which more effectively reduce pain that interferes with postoperative function and recovery.

Acknowledgements relating to this article

Assistance with the article: none.

Financial support and sponsorship: this work was supported by a Physicians’ Services Incorporated Foundation Operating Grant (PSI Grant #08-46) and Canadian Institutes of Health Research (CIHR Grant #MSH-55041). These public research funding agencies provided peer review and financial support of this study but had no other input or influence on the study.

Conflicts of interest: IG has received support from Pfizer, Aventis Pharma, Novopharm, Pharmscience, Apotex, Merck, Johnson & Johnson, Ortho-McNeill, Astra-Zeneca and Janssen-Ortho. None of the other authors have any relevant financial conflicts with the subject matter discussed in the manuscript.

Presentation: none.


1. Kehlet H, Dahl JB. Anaesthesia, surgery, and challenges in postoperative recovery. Lancet 2003; 362:1921–1928.
2. Kehlet H, Dahl JB. The value of ‘multimodal’ or ‘balanced analgesia’ in postoperative pain treatment. Anesth Analg 1993; 77:1048–1056.
3. Elia N, Lysakowski C, Tramèr MR. Does multimodal analgesia with acetaminophen, nonsteroidal anti-inflammatory drugs, or selective cyclooxygenase-2 inhibitors and patient-controlled analgesia morphine offer advantages over morphine alone? Meta-analyses of randomized trials. Anesthesiology 2005; 103:1296.
4. Marret E, Kurdi O, Zufferey P, Bonnet F. Effects of nonsteroidal antiinflammatory drugs on patient-controlled analgesia morphine side effects: meta-analysis of randomized controlled trials. Anesthesiology 2005; 102:1249–1260.
5. Gilron I, Orr E, Tu D, et al. A placebo-controlled randomized clinical trial of perioperative administration of gabapentin, rofecoxib and their combination for spontaneous and movement-evoked pain after abdominal hysterectomy. Pain 2005; 113:191–200.
6. Hyllested M, Jones S, Pedersen JL, Kehlet H. Comparative effect of paracetamol, NSAIDs or their combination in postoperative pain management: a qualitative review. Br J Anaesth 2002; 88:199.
7. Gilron I, Orr E, Tu D, et al. A randomized, double-blind, controlled trial of perioperative administration of gabapentin, meloxicam and their combination for spontaneous and movement-evoked pain after ambulatory laparoscopic cholecystectomy. Anesth Analg 2009; 108:623–630.
8. Mathiesen O, Jørgensen DG, Hilsted KL, et al. Pregabalin and dexamethasone improves postoperative pain treatment after tonsillectomy. Acta Anaesthesiol Scand 2011; 55:297–305.
9. Martinez V, Cymerman A, Ben Ammar S, et al. The analgesic efficiency of combined pregabalin and ketamine for total hip arthroplasty: a randomised, double-blind, controlled study. Anaesthesia 2014; 69:46–52.
10. Elvir-Lazo OL, White PF. The role of multimodal analgesia in pain management after ambulatory surgery. Curr Opin Anaesthesiol 2010; 23:697–703.
11. Young A, Buvanendran A. Recent advances in multimodal analgesia. Anesthesiol Clin 2012; 30:91–100.
12. Hurley RW, Cohen SP, Wu CL. Acute pain in adults. In: Ballantyne JC, Fishman SM, editors. Bonica's management of pain, chapter 51. 4th ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2010. 700–723
13. Verbeke G, Molenberghs G, editors. Linear mixed models in practice: a SAS-oriented approach. New York: Springer; 1997.
14. Lehmann EL. Nonparametrics: statistical methods based on ranks. San Francisco, CA: Holden-Day; 1975.
15. Agresti A. Categorical data analysis. New York: John Wiley & Sons, Inc; 1990.
16. Snedecor GW, Cochran WG. Statistical methods. 7th ed.Ames, IA: The Iowa State University Press; 1980.
17. McDonnell JG, O’Donnell B, Curley G, et al. The analgesic efficacy of transversus abdominis plane block after abdominal surgery: a prospective randomized controlled trial. Anesth Analg 2007; 104:193–197.
18. Kissin I. Patient-controlled-analgesia analgesimetry and its problems. Anesth Analg 2009; 108:1945–1949.
19. Calderón-Larrañaga A, Poblador-Plou B, González-Rubio F, et al. Multimorbidity, polypharmacy, referrals, and adverse drug events: are we doing things well? Br J Gen Pract 2012; 62:e821–e826.
20. Rollason V, Vogt N. Reduction of polypharmacy in the elderly: a systematic review of the role of the pharmacist. Drugs Aging 2003; 20:817–832.
21. Gilron I, Tod D, Goldstein DH, et al. The relationship between evoked versus spontaneous pain and peak expiratory flow after abdominal hysterectomy. Anesth Analg 2002; 95:1702–1707.
22. Erb J, Orr E, Mercer CD, Gilron I. Interactions between pulmonary performance and movement-evoked pain in the immediate postsurgical period: implications for perioperative research and treatment. Reg Anesth Pain Med 2008; 33:312–319.
23. Gilron I, Max MB, Lee G, et al. Effects of the 2-amino-3-hydroxy-5-methyl-4-isoxazole-proprionic acid/kainate antagonist LY293558 on spontaneous and evoked postoperative pain. Clin Pharmacol Ther 2000; 68:320–327.
24. Mamet J, Klukinov M, Yaksh TL, et al. Single intrathecal administration of the transcription factor decoy AYX1 prevents acute and chronic pain after incisional, inflammatory, or neuropathic injury. Pain 2014; 155:322–333.
© 2015 European Society of Anaesthesiology