A comparison between pre-operative carprofen and a long-acting sufentanil formulation for analgesia after ovariohysterectomy in dogs

Veterinary Anaesthesia and Analgesia, 2006, 33, 313 327 doi:10.1111/j.1467-2995.2005.00267.x RESEARCH PAPER A comparison between pre-operative carprofen and a long-acting sufentanil formulation for analgesia after ovariohysterectomy in dogs Louisa S Slingsby* BVSc, PhD, MRCVS, Pamela J Murison* BVMS, cert VA, Dipl ECVA, MRCVS, Lieve Goossens MSc Biostatistics, Marc Engelenà DVM & Avril E Waterman-Pearson* BVSc, PhD, FRCVS, MRCA, DVA, Dipl ECVA *Division of Companion Animals, Department of Clinical Veterinary Science, University of Bristol, Bristol, UK Medisearch International NV, Mechelen, Belgium àjanssen Animal Health BVBA, Beerse, Belgium Correspondence: Dr Louisa S Slingsby, Division of Companion Animals, Department of Clinical Veterinary Science, University of Bristol, Langford House, Langford, Bristol, BS40 5DU, UK. E-mail: louisa.slingsby@bristol.ac.uk Abstract Objective To assess the analgesic efficacy and adverse effects of a novel, long-acting sufentanil preparation in dogs undergoing ovariohysterectomy (OHE). Study design Blinded, positively controlled, randomized field trial with four parallel treatment groups. Animals Eighty client owned dogs undergoing elective OHE randomly allocated into four treatment groups (each n ¼ 20). Materials and methods Three groups received intramuscular (IM) sufentanil (at 10, 15 and 25 lg kg )1, respectively) and the control group received subcutaneous (SC) carprofen 4 mg kg )1 SC plus acepromazine 0.05 mg kg )1 IM as pre-anaesthetic medication. OHE was performed under thiopental/halothane anaesthesia. Visual Analogue Scale (VAS) scores for pain and sedation were awarded and mechanical nociceptive thresholds were measured at the wound and hock before surgery and up to 24 hours after tracheal extubation. Serum cortisol was measured before surgery, during surgery and up to 24 hours after tracheal extubation. Animals with inadequate post-operative analgesia were given rescue medication. Results In the carprofen group, VAS pain scores were significantly higher, wound tenderness was greater and requirement for rescue analgesia was more than in the sufentanil-treated groups. Sufentanil produced dose dependent analgesia and sedation. All treatment groups showed similar patterns of change for cortisol concentrations. Use of the sufentanil preparation was associated with a relatively high incidence of adverse events. Conclusions The long-acting preparation of sufentanil provided excellent post-operative analgesia that was significantly better than that provided by carprofen. However, use of this formulation, in the anaesthetic technique used in the study, resulted in a relatively high incidence of adverse effects. Clinical relevance Full mu (MOP) opioid agonists provide significantly better post-operative analgesia than nonsteroidal anti-inflammatory drugs after moderately painful surgery. However, the widely recognized adverse effects of opioids may preclude the use of these agents. Keywords cortisol, dog, nociceptive thresholds, postoperative analgesia, sufentanil, Visual Analogue Scale scores. Introduction All surgical procedures result in some degree of post-operative pain. Post-operative recovery in humans is compromised by the presence of pain 313

(Kehlet & Holte 2001) and pain prevention is an important ethical issue. Ovariohysterectomy (OHE) results in moderate to severe post-operative pain in dogs (Lascelles et al. 1998; Fox et al. 2000; Slingsby & Waterman-Pearson 2000). As OHE is normally performed electively on healthy, previously pain free animals, all post-operative pain can be attributed to the surgery. Thus OHE is a good model for investigations into analgesia. Carprofen, a nonsteroidal anti-inflammatory drug (NSAID), was chosen as a positive control because it provides effective, long lasting analgesia after orthopaedic and abdominal surgery in dogs (Nolan & Reid 1993; Lascelles et al. 1994, 1998; Slingsby & Waterman-Pearson 2001). At the time of study design it was the only long-acting (duration of up to 24 hours) veterinary analgesic (NSAID or opioid) licensed in the UK for pre-operative administration. No negative control group (not treated with analgesics) was included on welfare grounds and also because carprofen has been demonstrated to be analgesic compared to a negative control (Lascelles et al. 1998). The test drug was a novel, long-acting preparation of sufentanil. A variety of veterinary pharmaceutical products are available as long-acting formulations including anabolic agents, antibacterials, antibiotics, anti-inflammatory agents, corticosteroids and vitamins. Long-acting parenteral preparations are formulated by a number of methods dependant on the characteristics of the pharmaceutical compound involved and the site of injection. Formulation mostly involves physical modification of the drug, for example increasing the viscosity by using polymerbased solutions or by forming suspensions of the polymer-coated drug (micro encapsulation). At the time of writing the formulation of the long-acting sufentanil preparation is not available for public knowledge. Sufentanil is a phenylpiperidine with highly selective mu (MOP) receptor opioid agonist activity. It is five to 15 times as potent as fentanyl (De Castro et al. 1979) and slightly shorter acting. A short-acting sufentanil preparation is commercially available for use in humans where it is mostly used intravenously or epidurally during anaesthesia or for long-term sedation and analgesia in intensive care patients. Constant rate infusions are often used to overcome the short duration of action. Opioids are often thought of as the first tier of analgesic agents used against moderate to severe pain, with morphine considered to be the gold standard. NSAIDs are relegated to use when pain is less severe. However, NSAIDs are effective against mild to moderate pain and have a long duration of action. This is beneficial for veterinary surgeons who prefer not to re-dose analgesics. In addition, most veterinary patients are admitted as day patients and, in the UK, it is unusual to prescribe opioids for use in the home environment. Therefore, development of a long-acting opioid preparation that could provide 24 hours of analgesia would be important. Behavioural scoring is widely used in analgesic research with many studies using the Visual Analogue Scale (VAS) system to assess analgesia and sedation (Welsh et al. 1997; Lascelles et al. 1998; Slingsby & Waterman-Pearson 2001). Nociceptive thresholds may be useful in highlighting central and peripheral changes to the nervous system resulting from surgery and the modifying effects of analgesics (Lascelles et al. 1998; Slingsby & Waterman-Pearson 2001). Serum cortisol increases peri- and postoperatively in dogs (Church et al. 1994; Fox et al. 1994). Administration of analgesics may modify this rise (Fox et al. 1998) although the relationship between cortisol and pain has been questioned (Kehlet 1989). Sufentanil has also been shown to have an anaesthetic sparing effect (Hall et al. 1987; Short 1998). The aim of the study was to assess a novel, longacting sufentanil preparation and compare this to a known control drug (carprofen) in clinical patients undergoing OHE. Materials and methods This study was completed at the University of Bristol under Home Office Licence Numbers 30/1265 (before 13/1/2002) and 30/1883 (after 13/1/2002). It was conducted in compliance with Good Clinical Practice recommendations. Eighty dogs presented for routine OHE were admitted to the study after informed, written owner consent was obtained. Under the Home Office Licence requirements all dogs were required to be permanently identified either with a subcutaneous (SC) microchip or an ear tattoo. If required the dog was micro chipped (with the owner s consent) at the University of Bristol. Dogs were excluded from the study if opioids, a 2 -agonists, NSAIDs or steroids had been administered in the previous 14 days or if an American Society of Anesthesiologists category >1 was assigned. 314 Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327

After clinical examination, dogs were randomly allocated to one of four treatment groups (20 animals per group). Groups 1 3 received sufentanil at 10 (SUF10), 15 (SUF15) or 25 lg kg )1 (SUF25), respectively. Group 4 received 4 mg kg )1 carprofen (CARP). Sufentanil (Sufentanil long-acting preparation 0.5 mg ml )1 injectable solution; Janssen Animal Health B.V.B.A, Beerse, Belgium) was administered intramuscularly (IM) into the quadriceps muscle. Carprofen (Rimadyl Injection, Pfizer Ltd, Sandwich, Kent, UK) was administered SC into the scruff of the neck. Dogs in the carprofen group also received acepromazine (ACP; Novartis Animal Health UK Ltd, Royston, Herts, UK) 0.05 mg kg )1 IM injected into the quadriceps muscle; those in the sufentanil groups did not. At least 60 minutes after pre-anaesthetic medication, anaesthesia was induced with intravenous (IV) thiopental (Thiovet 2.5 g; Novartis Animal Health UK Ltd, Royston, Herts, UK). After endotracheal intubation with a cuffed tube, anaesthesia was maintained using halothane delivered in 100% oxygen via a nonrebreathing system (T-piece or Parallel Lack). Respiratory rate, heart/pulse rate and inspired halothane concentration were monitored during anaesthesia and recorded every 5 minutes. Other monitored variables included ECG, inspired/expired carbon dioxide and oxygen, blood pressure (indirect oscillometric method) and haemoglobin oxygen saturation. All were measured using the Datascope Passport 2 Portable Bedside Monitor (Datascope Medical Co Ltd, Huntingdon, Cambs, UK). Final year veterinary students performed the OHE via a midline incision under the direct and continuous supervision of a veterinary surgeon. No wound dressings were placed over the sutured skin wound. All dogs received antibiotic treatment immediately following induction of anaesthesia with clavulanate potentiated amoxycillin, either SC (Synulox ready to use injection; Pfizer Ltd) at 8.75 mg kg )1 or IV (Augmentin Intravenous; SmithKlineBeecham Pharmaceuticals, Brentford, Middlesex, UK) at 10 mg kg )1. Before recovery dogs were moved to a cage. The endotracheal tube was removed when laryngeal reflexes were restored. Rectal temperature was monitored and appropriate action taken to maintain or restore normothermia by covering the dog with a blanket, maintaining the room temperature between 22 and 26 C and, if required, providing additional warming with an air under blanket heater (Warm Air Hyperthermia System; Cincinnati Sub-Zero Products Inc, Cincinnati, OH, USA). Two investigators performed the VAS scoring. Each animal was scored entirely by one investigator. Both investigators were experienced in the use of the VAS score and preliminary work, performed in tandem, demonstrated a high degree of consistency between the two investigators. The VAS pain scale was anchored at 0 ¼ no pain and 100 ¼ worst possible pain, the sedation scale with 0 ¼ no sedation and 100 ¼ no response to hand clap above head. VAS scoring was performed before test drug administration, after test drug administration but before induction of anaesthesia and postoperatively at 30 minutes, 1, 2, 4, 8, 12, 18 and 24 hours after tracheal extubation. The animals were initially observed from outside the cage, were spoken to and then approached. The animal was handled and encouraged to come out of the cage and walk around. Finally the wound area was palpated. Pain and sedation scores were awarded at the end of this assessment. Mechanical nociceptive thresholds were measured immediately after each VAS scoring procedure. The device used (Constant Rate of Application of Force, CRAF meter) has been previously described (Lascelles et al. 1998). It consists of a blunt ended pin connected to a force sensing load cell mounted in a pen type handle attached to a measuring unit. The pin was pressed perpendicular to the skin on the area tested and force was applied in a constant manner until a response was achieved (withdrawal of test area/flinch/head turn). The two areas tested were the craniolateral area of the tarsus immediately distal to the hock ( hock ) and the ventral midline of the abdomen, halfway between the cranial pubic rim and xiphisternum ( wound ). Blood samples were collected for determination of serum cortisol levels from a cephalic vein catheter before test drug administration, after test drug administration but before induction of anaesthesia, mid surgery (approximately 30 minutes after start of surgery), end of surgery and then 2, 4, 8, 18 and 24 hours after tracheal extubation. Serum was frozen and later dispatched to SCL Biosciences (Gordon Laboratory Group, UK) for analysis. If an animal was deemed to be in unacceptable pain (VAS pain score >60) rescue analgesia was administered. Dogs in the carprofen group received IM pethidine (Pethidine Injection; Arnolds Vet Products Ltd, Shrewsbury, Shropshire, UK) Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327 315

5mgkg )1 while dogs in the sufentanil groups received SC meloxicam (Metacam, Boehringer Ingelheim Ltd, Bracknell, Berks, UK) 0.2 mg kg )1.In these cases the animal was removed from the study and no further VAS scoring, thresholds or blood samples were performed. Both carprofen and sufentanil administration may be associated with adverse events. For carprofen these include typical NSAID side-effects such as vomiting, diarrhoea and gut ulceration. For sufentanil these include typical opioid agonist side-effects such as bradycardia, respiratory depression, sedation or excitement. Any adverse events after test drug administration were recorded. Treatment of adverse events was on a case-by-case basis as required. Statistical testing Statistical tests were performed with the SAS System for Windows, Version 8.2, Copyright 1999 2001 by SAS Institute Inc., Cary, NC, USA. Comparisons between groups for body mass, anaesthetic requirement and timings were made using one-way analysis of variance (ANOVA) followed by post hoc Tukey tests. The rescue analgesia categorical data were analysed using chi-squared tests. The initial VAS scores and nociceptive thresholds, i.e. before administration of test drug and after administration of test drug but before induction of anaesthesia, were compared between groups using Kruskal Wallis ANOVA (VAS) or one-way ANOVA (nociceptive thresholds). Analysis of the post-operative VAS scores and nociceptive threshold data was complicated by missing data at various time points in some animals. Reasons for missing data included rescue treatment, dogs being removed from the study for other reasons, e.g. adverse events, and observations that were missed for logistical reasons. The analysis of post-operative VAS scores and nociceptive threshold data can be performed using two techniques: traditional summary statistics and longitudinal techniques. Use of summary statistics is the most straightforward technique. This analysis is not based on the raw repeated measurements from each subject, but rather on a number of derived parameters, e.g. area under the curve (AUC) calculated from these data. Longitudinal techniques are more powerful when analysing repeated measurements data, especially when dealing with unobserved, missing data. As a result of the lack of clear assumptions, based on the nature of the data, we decided to perform the two most straightforward models. The first model includes the factors group and time. The second model includes an extra factor, the interaction group by time. In addition, more complex models (Thijs 2002) were investigated in cooperation with the University of Diepenbeek, Belgium (these are not reported in this paper). For the traditional analysis the post-operative VAS pain scores at each time point were converted into classes (increasing pain scores, increasing classes) to allow inclusion of data from animals with missing observations. This method of dealing with missing values, rather than recording them as the last observation carried forward ( locf ) method (Verbeke & Molenberghs 1997) was used because there were so many missing values at the later time points (primarily due to the high rescue analgesia requirement in the carprofen group). The actual VAS pain scores were split into four groups corresponding with mild discomfort, moderate discomfort, moderate pain and severe pain. Dogs removed from the study were assumed to have a high degree of pain and were, therefore, put into a higher category than the VAS scores. The inclusion criteria for the classes are shown in Table 1. From the class data the summary parameter, AUC for the entire post-surgery period, was calculated and then compared between groups using two by two Wilcoxon Mann Whitney U-tests. The longitudinal technique involved construction of two linear mixed models. The traditional summary parameter AUC was calculated from the first two hours of post-operative VAS sedation scores and was analysed with Table 1 Descriptors for the allocation of pain scores into classes to enable statistical analysis Class number Description of class 1 VAS scores 0 24 2 VAS scores 25 49 3 VAS scores 50 75 4 VAS scores 75 100 5 Dogs off study for other reasons than analgesia 6 Dogs given rescue analgesia VAS, Visual Analogue Scale. 316 Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327

nonparametric statistics (Wilcoxon Mann Whitney U-tests). Thereafter no further analysis was performed. As more than 20% of the data were imputed the locf method would not reflect clinical reality. Two linear mixed models were constructed for longitudinal analysis. For the post-operative nociceptive thresholds substitutions were made for missing data. The locf method was used for dogs that received rescue treatment or that were removed for other reasons. Dogs that did not respond to the test were assigned a maximum value of 15 N and dogs that would not allow the test were assigned a value of 0 N. Because of the large number of dogs requiring rescue analgesia in the control group only the first 8 hours were investigated. These data were summarized into an AUC for each dog and compared between groups using one-way ANOVA followed by post hoc Tukey s tests. Within-group comparisons were investigated using paired t-tests. The numbers of animals with adverse events were compared between groups by means of Fisher s Exact test. For the serum cortisol concentrations the Kruskal Wallis ANOVA was performed on each time point to test the differences between groups. Within-group comparisons were performed by means of the Signed Rank test. Results The animals body masses ranged from 4.6 to 43.0 kg (mean of 19.0 kg), with no statistically significant difference between groups demonstrated by one-way ANOVA. There was no significant difference between groups with regard to induction dose of thiopental. Mean thiopental doses (mg kg )1 ) (standard deviation) for each group were 13.0 mg kg )1 (3.6), 11.7 mg kg )1 (2.7), 11.5 mg kg )1 (3.6) and 12.8 mg kg )1 (1.8) for SUF10, SUF15, SUF25 and CARP respectively. The overall mean thiopental dose was 12.2 mg kg )1. The overall concentration of halothane was taken as the mean of the inspired halothane concentration recorded every 5 minutes during anaesthesia. Mean overall concentrations (%) (standard deviation) were 1.1% (0.3), 1.2% (0.3), 0.9% (0.3) and 1.3% (0.3) for SUF10, SUF15, SUF25 and CARP respectively. Using one-way ANOVA there was a statistically significant difference between groups (p ¼ 0.015) with a significant Tukey post hoc difference between SUF25 and CARP (p ¼ 0.009), i.e. the overall concentration of halothane for SUF25 was significantly lower than for CARP. This reflects an anaesthetic sparing effect when using sufentanil. There were no significant differences between groups (one-way ANOVA) for the following timings, therefore they are presented as overall mean (standard deviation). Time from injection of the test drug to induction of anaesthesia was 81 minutes (22) and time from induction of anaesthesia to start of surgery was 26 minutes (10). Mean duration of surgery was 52 minutes (18) and time from the end of surgery to tracheal extubation was 14 minutes (10). Rescue analgesia Five dogs (25%) in SUF10 group, five dogs (25%) in the SUF15 group, two dogs (10%) in the SUF25 group and 17 dogs (85%) in the CARP group required rescue analgesia. In addition one dog (5%) in the SUF10 group required further analgesia for a repeat surgery and one dog (5%) in the SUF25 group required further analgesia after administration of naloxone. Using a chi-squared test, there was a statistically significant difference between groups with respect to their requirement for rescue analgesia (p < 0.001). Significantly more dogs from the carprofen group required rescue analgesia when compared with each of the three sufentanil groups (all p < 0.001). VAS scores for pain Traditional statistics of VAS pain data At the first two time points (1st ¼ before administration of the test drug and 2nd ¼ after administration of the test drug and before induction of anaesthesia) VAS pain score was 0 in all groups therefore no comparison between groups was performed. The summary parameter, AUC, (based on the classes) for the whole post-operative period is presented in Table 2 (descriptive statistics) and Table 3 (distribution table). These demonstrated significantly lower values in each of the three sufentanil groups compared with the carprofen group, indicating better analgesia with sufentanil. For the three sufentanil groups most of the dogs (±65%) were located in the first two classes, which suggests that they experienced mild to moderate Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327 317

AUC (whole of post-operative period) Sufentanil Sufentanil Sufentanil Carprofen 10 lg kg )1 15 lg kg )1 25 lg kg )1 4mgkg )1 Table 2 Summary of VAS pain scores for the post-operative period: traditional statistics Median 24.88 24.00 24.00 126.75 Minimum 18.0 18.0 18.0 18.0 Maximum 132.8 120.0 132.8 138.5 Statistics (Wilcoxon Mann Whitney U-tests): carprofen > sufentanil 10 lg kg )1, p ¼ 0.005; carprofen > sufentanil 15 lg kg )1, p < 0.001; carprofen > sufentanil 25 lg kg )1, p < 0.001. VAS, Visual Analogue Scale; AUC, area under the curve. Table 3 Distribution table of Visual Analogue Scale (VAS) pain scores for the post-operative period: traditional statistics Group Description VAS pain score 0 24 VAS pain score 25 49 VAS pain score 50 74 VAS pain score 75 100 Dogs off study other than analgesia Rescue treatment Sufentanil 10 lg kg )1 n (%) 7 (35) 7 (35) 0 (0) 0 (0) 1 (5) 5 (25) n (cum. %) 14 (70) 14 (70) 14 (70) 15 (75) 20 (100) Sufentanil 15 lg kg )1 n (%) 10 (50) 3 (15) 0 (0) 0 (0) 2 (10) 5 (25) n (cum. %) 13 (65) 13 (65) 13 (65) 15 (75) 20 (100) Sufentanil 25 lg kg )1 n (%) 11 (55) 2 (10) 0 (0) 0 (0) 5 (25) 2 (10) n (cum. %) 13 (65) 13 (65) 13 (65) 18 (90) 20 (100) Carprofen 4 mg kg )1 n (%) 1 (5) 2 (10) 0 (0) 0 (0) 0 (0) 17 (85) n (cum. %) 3 (15) 3 (15) 3 (15) 3 (15) 20 (100) discomfort. In contrast, the majority of dogs in the carprofen group (85%) needed rescue treatment. Longitudinal data analysis of VAS pain data The two straightforward linear mixed models that were investigated (Figs 1 & 2) clearly demonstrated higher VAS pain scores in the carprofen group than in all the sufentanil groups. There was less pain with an increasing dose of sufentanil. The conclusion, from the use of traditional statistical methods, was that sufentanil provides a greater analgesic effect, was confirmed by the longitudinal technique. 40 First model: Y= α intercept + β time t + β group Estimate of pain VAS score 30 20 10 0 0 2 4 6 8 10 12 14 16 18 20 22 24 Time after dosing (hours) A: Sufentanil 10 µg kg 1 B: Sufentanil 15 µg kg 1 C: Sufentanil 25 µg kg 1 D: Carprofen 4 mg kg 1 Figure 1 Visual Analogue Scale pain scores for the post-operative period in bitches after ovariohysterectomy using the first linear mixed model. 318 Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327

Second model: Y = β + ( β + β )t (no intercept) group time time * group 60 Estimate of pain VAS score 50 40 30 20 10 Figure 2 Visual Analogue Scale pain scores for the post-operative period in bitches after ovariohysterectomy using the second linear mixed model. 0 0 2 4 6 8 10 12 14 16 18 20 22 24 Time after dosing (hours) A: Sufentanil 10 µg kg 1 B: Sufentanil 15 µg kg 1 C: Sufentanil 25 µg kg 1 D: Carprofen 4 mg kg 1 VAS scores for sedation Longitudinal data analysis of VAS sedation data At the first time point (before administration of the test drug) no dogs were sedated (VAS sedation scores were zero). At the second pre-operative time point sedation was observed in some animals. Median scores (interquartile range IQR) were 0.0 (5.0), 4.5 (8.5), 9.0 (11.5) and 8.0 (16.0) for SUF10, SUF15, SUF25 and CARP groups respectively. Sedation scores were compared between groups using a Kruskal Wallis ANOVA and a significant difference was seen between groups (p ¼ 0.008). Post hoc testing demonstrated significantly lower scores in the SUF10 group compared with the SUF25 and CARP groups. After surgery the individuals VAS sedation scores decreased over time. An AUC for the first 2 hours after extubation was calculated using the locf method to impute missing values. These data are summarized in Table 4. Sedation was significantly greater in the higher dose sufentanil group (SUF25) compared with CARP or SUF10. The middle dose of sufentanil (SUF15) also had a significantly greater sedative effect compared with CARP. Here, the two most straightforward linear mixed models were investigated based on the available case method. These are presented in Figs 3 and 4. This method uses the available data; it simply means that every component of a parameter is estimated using the maximal amount of information available for that component. The two models demonstrated that the highest sufentanil dose caused marked sedation. The other two sufentanil doses produced less sedation comparable with that observed with carprofen, where the degree of sedation was moderate. Mechanical nociceptive thresholds at the hock Between-group comparisons Using a one-way ANOVA, no statistically significant differences were found between the four groups for mechanical nociceptive thresholds measured at the hock before administration of the test drug (p ¼ 0.941, overall mean 5.9 N) and after admin- Table 4 Summary of VAS sedation scores from extubation until 2 hours post-extubation: traditional statistics AUC (until 2 hours post-extubation) Sufentanil Sufentanil Sufentanil Carprofen 10 lg kg )1 15 lg kg )1 25 lg kg )1 4mgkg )1 Median 114.50 131.75 147.88 108.75 Minimum 33.5 9.8 40.5 39.5 Maximum 161.8 178.5 173.0 158.0 Statistics (Wilcoxon Mann Whitney U-tests): sufentanil 25 lg kg )1 > sufentanil 10 lg kg )1, p ¼ 0.012; sufentanil 15 lg kg )1 > carprofen, p ¼ 0.030; sufentanil 25 lg kg )1 > carprofen, p ¼ 0.002. VAS, Visual Analogue Scale; AUC, area under the curve Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327 319

First model: Y = α intercept + β time t + β group Estimate of sedation VAS score 60 50 40 30 20 10 0 0 10 2 4 6 8 10 12 14 16 18 20 22 24 20 30 Time after dosing (hours) A: Sufentanil 10 µg kg 1 B: Sufentanil 15 µg kg 1 C: Sufentanil 25 µg kg 1 D: Carprofen 4 mg kg 1 Figure 3 Visual Analogue Scale sedation scores for the post-operative period in bitches after ovariohysterectomy using the first linear mixed model. Second model: Y= β + ( β + β )t (no intercept) group time time * group Estimate of sedation VAS score 70 50 30 10-10 0 2 4 6 8 10 12 14 16 18 20 22 24-30 -50 Time after dosing (hours) A: Sufentanil 10 µg kg 1 B: Sufentanil 15 µg kg 1 C: Sufentanil 25 µg kg 1 D: Carprofen 4 mg kg 1 Figure 4 Visual Analogue Scale sedation scores for the post-operative period in bitches after ovariohysterectomy using the second linear mixed model. istration of the test drug but before induction of anaesthesia (p ¼ 0.641, overall mean 6.3 N). Using a one-way ANOVA on the AUC calculated for the first 8 hours after extubation, there was no statistically significant difference (p ¼ 0.573, overall mean 58.2 N) between groups for the mechanical nociceptive threshold measured at the hock. Within-group comparisons Although mechanical nociceptive thresholds at the hock tended to increase after administration of the test drug, within each group there were no statistically significant differences between the thresholds before administration of the test drug and after administration of the test drug but before induction of anaesthesia (paired t-tests: p ¼ 0.086, p ¼ 0.450, p ¼ 0.281 and p ¼ 0.974 for SUF10, SUF15, SUF25 and CARP respectively). Therefore, neither carprofen nor sufentanil administered alone (without anaesthesia and surgery) had significant effects on hock mechanical nociceptive thresholds. Mechanical nociceptive thresholds at the wound site Between-group comparisons Using a one-way ANOVA, there were no statistically significant differences between groups for mechanical nociceptive thresholds at the wound site before administration of test drug (p ¼ 0.416, overall mean 4.3 N) and after administration of the test 320 Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327

Table 5 Area under the curve (AUC) for mechanical nociceptive wound threshold data until 8 hours after tracheal extubation: traditional statistics AUC (until 8 hours after extubation) Sufentanil Sufentanil Sufentanil Carprofen 10 lg kg )1 15 lg kg )1 25 lg kg )1 4mgkg )1 Mean 26.5 35.4 33.2 17.7 Standard deviation 15.34 26.56 23.59 14.34 Statistics: one-way ANOVA, p ¼ 0.037; sufentanil 15 lg kg )1 > Carprofen, p ¼ 0.041 (post hoc Tukey test). drug but before induction of anaesthesia (p ¼ 0.825, overall mean 4.1 N). Using a one-way ANOVA on the AUC calculated for the first 8 hours after tracheal extubation (Table 5) there was a statistically significant difference (p ¼ 0.037). The post hoc Tukey test indicated a statistically significant difference (p ¼ 0.041) between the SUF15 and the CARP group. The dogs of the SUF15 groups had higher thresholds than the CARP group, which means that the sufentanil dogs were less sensitive to wound pressure. This difference in means was also present between the SUF25 and the CARP group, but showed only a trend towards significance (p ¼ 0.090). Within-group comparisons Within each group there were no statistically significant differences between the mechanical nociceptive thresholds at the wound site before and after administration of test drug but before induction of anaesthesia (paired t-tests: p ¼ 0.476, p ¼ 0.492, p ¼ 0.233 and p ¼ 0.073 for SUF10, SUF15, SUF25 and CARP respectively). Therefore, neither carprofen nor sufentanil administered alone (without anaesthesia and surgery) had a significant effect on wound thresholds. Adverse events Table 6 presents a summary of the adverse events noted during the study period, excluding inadequate analgesia and requirement for rescue treatment. Treatment for adverse events was on a caseby-case basis. Atropine was used for intra-operative bradycardias and dysrrhythmias such as atrioventricular block. Esmolol was used to treat intraoperative tachycardias. Naloxone was used for post-operative dysphoria. Fluid therapy was used for volume replacement or circulatory support. There were fewer adverse events in the carprofen 4mgkg )1 group than in the sufentanil groups. Results of statistical testing both within groups and post hoc test where applicable are given in Table 6. Serum cortisol concentrations Data from all dogs entering the study were included in this analysis since all the blood samples were obtained before any of the dogs were excluded from the analysis of the VAS and threshold data. No substitutions were made after rescue analgesia or if blood samples were unable to be taken. The values for 24 hours after surgery have been excluded because too few samples were taken at this time point. The data were not normally distributed therefore a nonparametric statistical test was used. Between-group comparisons The descriptive statistics are summarized in Table 7. Kruskal Wallis ANOVAs were performed at every time point between the four groups and no statistical significant differences were found. Within-group comparisons Signed Rank tests were used to compare serum cortisol levels between time points within each group. In all four groups serum cortisol levels did not change significantly after administration of the test drug. In all four groups serum cortisol levels were significantly higher in samples collected during and at the end of surgery, compared with the samples taken before administration of the test drug. This is consistent with the normally observed increase in serum cortisol during surgery. Discussion This study aimed to investigate the use of a novel, long-acting formulation of sufentanil in the dog and compare post-operative outcome measures with a similar group of dogs receiving carprofen. Data Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327 321

Table 6 Summary of adverse events. The first number in each category is the total number of cases, the second number is the number of cases that required pharmacological intervention. Statistical comparisons between groups were made using Fisher s Exact tests with significance set at p < 0.05 Intra-operative adverse events Post-operative adverse events Pre-operative adverse events: salivation dysphoria Potentially drug related: apnoea, hypercapnia, bradycardia ECG anomalies Unlikely to be drug related: haemorrhage Potentially drug related: excess salivation, vomiting, dysphoria, other emergence reactions Unlikely to be drug related: moderate skin haemorrhage, severe abdominal haemorrhage Group Number Treatment required? Number Treatment required? Number Treatment required? Number Treatment required? Number Treatment required? Sufentanil 10 lg kg )1 6 0 7 0 1 1 7 0 1 1 Sufentanil 15 lg kg )1 7 0 14 3 1 1 9 1 1 0 Sufentanil 25 lg kg )1 3 0 11 4 0 0 8 5 1 0 Carprofen 4 mg kg )1 0 0 2 0 2 2 2 0 3 0 Between groups statistics p < 0.05 p < 0.05 Not significant p < 0.05 Not significant Post hoc tests SUF 10 > CARP SUF 15 > CARP SUF 15 > CARP SUF 25 > CARP Not applicable SUF 15 > CARP Not applicable 322 Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327

Table 7 Serum cortisol levels Serum cortisol (nmol L )1 ) [median (minimum maximum) plus number of samples] Time point Sufentanil 10 lg kg )1 Sufentanil 15 lg kg )1 Sufentanil 25 lg kg )1 Carprofen 4 mg kg )1 Pre-test substance 90.0 (33 209) n ¼ 17 101.5 (22 291) n ¼ 16 92.5 (18 307) n ¼ 18 101.0 (13 208) n ¼ 14 Post-test substance 89.5 (18 261) n ¼ 16 120.0 (32 773) n ¼ 19 186.0 (14 297) n ¼ 16 78.0 (34 338) n ¼ 15 Mid surgery 247.0 (158 409) n ¼ 13 258.0 (126 854) n ¼ 17 257.5 (182 448) n ¼ 18 225.0 (152 384) n ¼ 17 End of surgery 264.5 (153 533) n ¼ 14 232.0 (150 820) n ¼ 17 252.0 (166 698) n ¼ 15 320.5 (216 489) n ¼ 16 Extubation + 2 hours 265.0 (89 406) n ¼ 13 216.0 (101 477) n ¼ 13 169.0 (73 286) n ¼ 11 265.0 (92 434) n ¼ 13 Extubation + 4 hours 144.0 (56 212) n ¼ 11 206.0 (65 334) n ¼ 13 166.0 (50 285) n ¼ 9 139.0 (49 223) n ¼ 5 Extubation + 8 hours 107.0 (82 152) n ¼ 5 210.5 (82 240) n ¼ 8 162.0 (61 267) n ¼ 6 139.0 (52 258) n ¼ 3 Extubation + 18 hours 85.5 (71 137) n ¼ 4 144.5 (62 202) n ¼ 8 74.5 (41 133) n ¼ 6 99.0 (99 99) n ¼ 1 Values are median (minimum maximum) and n indicates the number of samples at each time point. analysis proved to be more complicated than initially envisaged because of the number of animals that dropped out of the study (inadequate analgesia or adverse events). For this reason both traditional (summary statistics) and longitudinal methods (repeated measurements data) of data analysis were used. Longitudinal techniques are powerful techniques for analysis of repeated measurement data, especially when dealing with unobserved, missing data (Lindsey 1993; Diggle et al. 1994; Verbeke & Molenberghs 1997). Two linear mixed models were investigated and clearly demonstrated higher VAS pain scores in the carprofen group than in all the sufentanil groups. There is a growing awareness of the need for methods to investigate the sensitivity of the results when assumptions are made to facilitate handling of experimental data, such as using locf techniques or grouping data into classes (as in the current study) (Little & Rubin 1987; Little 1995; Lesaffre & Verbeke 1998; Verbeke et al. 1998; Molenberghs et al. 2001). For this reason the data from the current study underwent further sensitivity analysis (not described here). The conclusions of this sensitivity analysis (Thijs 2002) correspond with the more straightforward methods based on summary statistics described in this paper. In the current study sufentanil administration had some effects on anaesthetic requirement. Dogs receiving sufentanil were in less pain and had less sensitive wounds than those that received carprofen. Plasma cortisol concentrations were not a sensitive indicator of differences between groups. Administration of sufentanil resulted in higher than expected incidence of side-effects before, during and after anaesthesia. Many full and partial opioid receptor agonists have been used for post-operative analgesia in the dog. These include morphine, pethidine, oxymorphone, pentazocine, buprenorphine and butorphanol (Taylor & Houlton 1984; Waterman & Kalthum 1992; Nolan & Reid 1993; Lascelles et al. 1994; Brodbelt et al. 1997; Kyles et al. 1998; Fox et al. 2000). All these opioids provide analgesia of varying duration from 2 hours to over 6 hours depending on the dose and agent used. For prolonged, effective analgesia these agents need frequent administration or infusion. Regular re-dosing has the advantage of greater control and permits agent substitution if desired. In practice animals may not receive further doses of analgesics or the dosing interval may be excessive leading to inadequate analgesia. Repeated injections may also be an aversive experience for the animal. IM injections are more painful than spinal and intravenous injections (Simini 2000; Cupitt & Kasipandian 2004). In addition many animals are discharged to their owners care soon after recovery, preventing repeat opioid administration. Transdermal fentanyl has been used in dogs for post-operative analgesia (Kyles et al. 1998), but patches need to be applied well before surgery in order to achieve effective plasma levels and there are issues of human safety associated with their use. The potential advantages of using a long-acting preparation of an opioid such as sufentanil would be the ability to administer it as pre-anaesthetic medication to provide sedation and an anaesthetic sparing effects. Pre-operative administration would allow it to act as a pre-emptive analgesic. Effective and long-acting analgesia would be provided with- Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327 323

out a second injection and the animal would have analgesia overnight if discharged on the day of surgery. Conventional dosing schedules with opioids may result in a less than ideal pattern of drug peaks and troughs. During the peaks there is increased risk of side-effects and during the troughs analgesia is diminished or lost. The benefit of a sustained-release formulation can be that the high peak, often associated with administering a bolus, can be avoided and the period until the trough occurs is greatly increased. Long-acting preparations are used in veterinary practice for sustained-release of a wide variety of pharmaceutical products in both farm and companion animals (Lee & Putnam 2000; Medlicott et al. 2004). A depot of the drug has to be formed in order to provide a slow and constant release of the active compound from the injection site. Liquid depots (as opposed to solid formulations) may be aqueous-based or oil-based. Although this is the simplest method of providing a controlled release injectable formulation, product stability can be a major limiting factor (Medlicott et al. 2004). These systems can be produced by forming suspensions, gels, emulsions, liposome dispersions or micro emulsions. Sufentanil is rapidly absorbed after IM injection and is effective at low concentrations (Meert 1996) making it an ideal candidate for a sustained-release formulation (Lee & Putnam 2000). The actual technique for producing the long-acting sufentanil used in this study is currently not publicly available but liposome encapsulation has been used to produce sustained release formulations of both morphine (Howell 2001) and oxymorphone (Smith et al. 2004). In a study of this type comparison with a control group of animals is important. A positive control group treated with carprofen was used. Carprofen has been proven to provide post-operative analgesia when compared with no analgesics (Lascelles et al. 1998). It was not considered justifiable or necessary to include a control group without analgesia in this study. Sufentanil at the highest dose (25 lg kg )1 ) studied produced similar sedation to that seen after ACP but did not reduce the induction dose of thiopental (compared with ACP and carprofen). However, it had a significant sparing effect on the inspired percentage of volatile agent required compared with ACP and carprofen. Carprofen has no MAC (minimum alveolar concentration) sparing effects (Alibhai & Clarke 1996) but ACP decreases the MAC of halothane by approximately 40% (Heard et al. 1986). In this study only inspired halothane concentration was measured rather than MAC but the sufentanil decreased the overall inspired concentration to a greater degree than ACP suggesting that it has a greater MAC reducing effect than ACP. This agrees with data from other studies demonstrating a MAC reduction of up to 60 70% when sufentanil is co-administered with enflurane (Hall et al. 1987), isoflurane (Brunner et al. 1994) or halothane (Glenski et al. 1988). Sufentanil did not increase the time required for the dog to regain its laryngeal reflexes at the end of the anaesthetic (time from end of surgery to tracheal extubation). Some animals in all three of the sufentanil-treated groups showed adverse side-effects before induction of anaesthesia, mostly excessive salivation. Salivation and emesis are common sequelae to the use of opioids such as morphine and so were not unexpected. There was a significantly higher incidence of intra-operative complications in the groups given sufentanil at 15 and 25 lg kg )1 compared with the carprofen treated group. These complications were either respiratory (apnoea or hypoventilation) or cardiovascular (bradycardia/tachycardia or dysrhythmias). Significantly more dogs in the sufentanil 15 and 25 lg kg )1 groups required ventilatory support compared with dogs in the carprofen group. Intra-operative administration of opioids, particularly the ultra-short acting, potent opioids such as fentanyl can cause bradycardia, hypotension and respiratory depression (Thurmon et al. 1996; Hall et al. 2001). Sufentanil may cause a reduction in the contribution of central chemoreceptors to ventilatory drive, resulting in decreased ventilation and hypercapnia (Abdul-Rasool & Ward 1989). These respiratory side-effects can be avoided by providing intermittent positive pressure ventilation but this is not commonplace in general veterinary practice, and so it was deemed important to assess the drug during spontaneous ventilation. The cardiovascular side-effects could have been prevented by preanaesthetic medication with an anti-muscarinic agent (Benson et al. 1987). However, these agents also have undesirable side-effects. Post-operatively, all three groups of sufentaniltreated animals experienced significantly less pain than dogs in the carprofen treated group. The superior analgesia was noticeable from the first post-operative time point (30 minutes after tracheal extubation) until the next day. Dogs in the carpro- 324 Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327

fen-treated group required significantly more rescue analgesia. Studies in humans have also demonstrated good post-operative analgesia after intra-operative analgesia with the commercially available short-acting sufentanil preparation (Glenski et al. 1988; Rosseel et al. 1988). All four groups received the analgesics pre-operatively in order to benefit from their pre-emptive effects (Woolf & Chong 1993; Lascelles et al. 1997, 1998; Welsh et al. 1997). In all four treatment groups no obvious secondary hyperalgesia was seen for the first 8 hours after surgery (demonstrated by the hock thresholds remaining at similar values after surgery compared with baseline, Lascelles et al. 1998). However, it may be too early to see development of secondary hyperalgesia at 8 hours. Lascelles et al. (1998) demonstrated hyperalgesia within 24 hours but because of the high occurrence of missing values it was not considered suitable to analyse our data beyond 8 hours. In addition, dogs in the carprofen treated group had significantly more tender wounds than the sufentanil-treated dogs, suggesting that sufentanil may also decrease primary hyperalgesia. The number of animals requiring rescue analgesia in the carprofen group was greater than in previous studies reported by Nolan & Reid 1993; Lascelles et al. 1994, 1998; Welsh et al. 1997 and Slingsby & Waterman-Pearson 2001. This highlights the difficulties inherent in using a subjective system of assessment. In the current study and those cited, the assessors were ignorant of the treatment but here the comparison was between carprofen and what proved to be a very potent long lasting opioid whereas in previous studies comparison was made between carprofen and less potent opioids. In addition there is no doubt that intraobserver variability over time is a factor. Although in all the studies cited above the observers were experienced in the use of VAS scoring systems, over the course of studies observers tend to become more sensitive to the subtle pain behaviours and may give higher VAS pain scores than they would have previously (E.M. Welsh and A.M. Nolan, Personal Communication). The use of multiple observers may also affect pain scores because there may be great variation between different observers in the course of a study even though each individual may be consistent (Holton et al. 1998). However, both assessors in our study were experienced in scoring pain. Furthermore, before starting the trial they scored a number of animals simultaneously but independently; the resulting scores were within 5 mm of each other every time. Additionally, when the data were analysed both assessors had scored an equal proportion of dogs from each test group. Most of the dogs were able to walk, with varying degrees of ataxia, by 4 hours after tracheal extubation. At 8 hours few of the dogs in the sufentanil 10 and 15 lg kg )1 groups and the carprofen group showed any residual sedation. However, dogs in the highest dose sufentanil group were still sedated 8 hours after tracheal extubation, which is clinically relevant. Potentially drug-related side-effects were seen in all groups post-operatively. Some, particularly in the carprofen group, such as salivation or vomiting may have been associated with inadequate analgesia rather than trial drug administration (Holton et al. 2001). The other side-effects, seen only in the sufentanil-treated animals, were dysphoria and some more unusual emergence reactions, e.g. apparent blindness in one case and apparent paresis in another. Many sufentanil-treated animals were sensitive to noise and environmental stimulation. In a few cases treatment with naloxone was required which eliminated the side-effects. Respiratory depression was not seen in any animal post-operatively. In agreement with other reported studies, plasma cortisol levels increased during surgery then fell post-operatively (Church et al. 1994; Fox et al. 1994, 1998; Hansen et al. 1997). However, there were no significant differences between the groups. Hansen et al. (1997) demonstrated reduced cortisol levels intra-operatively in oxymorphone treated dogs compared with a control group. In the current study inclusion of a negative control group might have revealed modification of the surgical stress response by sufentanil or carprofen. There was a wide variation in baseline cortisol levels between individual animals. Most of the dogs presented were presumably undergoing a stressful experience unrelated to the actual surgery and the cortisol release associated with these events may have masked any effects relating to analgesia. Fox et al. (1998) found significantly higher pre-treatment plasma cortisol concentrations in pet dogs (130 nmol L )1 ) compared with purpose-bred dogs (62 nmol L )1 ). Hospitalization for a few days before the study could have decreased the initial cortisol levels but was not practical. Kehlet (1989) concluded that there was no correlation between analgesia and a significant decrease in the surgical stress response in humans Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327 325

(probably as a result of incomplete afferent neural blockade). Plasma cortisol concentrations may, therefore, not provide a useful measure of pain under clinical conditions. In summary, sufentanil, at all three doses tested, provided good, long lasting post-operative analgesia, significantly better than that provided by carprofen. However, the incidence of intra-operative complications, particularly at the two higher doses suggest that sufentanil, in this formulation, would be unsuitable to use pre-operatively in animals following this anaesthetic protocol. Acknowledgements This study was supported by Janssen Animal Health B.V.B.A., Belgium. The authors wish to thank Alan Jones and Rachel Burt for technical assistance and the surgeons: Professor Peter Holt, Alastair Hotson Moore, Richard Coe and Andy Moores for their assistance during this study. References Abdul-Rasool IH, Ward DS (1989) Ventilatory and cardiovascular responses to sufentanil infusion in dogs anesthetized with isoflurane. Anesth Analg 69, 300 306. Alibhai HIK, Clarke KW (1996) Influence of carprofen on minimum alveolar concentration of halothane in dogs. J Vet Pharmacol Ther 19, 320 321. Benson GJ, Thurmon JC, Tranquilli WJ et al. (1987) Intravenous administration of lenperone and glycopyrrolate followed by continuous infusion of sufentanil in dogs: cardiovascular effects. Am J Vet Res 48, 1372 1375. Brodbelt DC, Taylor PM, Stanway GW (1997) A comparison of preoperative morphine and buprenorphine for postoperative analgesia for arthrotomy in dogs. J Vet Pharmacol Ther 20, 284 289. Brunner MD, Briathwaite P, Jhaveri R et al. (1994) MAC reduction of isoflurane by sufentanil. Br J Anaesth 72, 42 46. Church DB, Nicholson AI, Ilkiw JE et al. (1994) Effect of nonadrenal illness, anaesthesia and surgery on plasma cortisol concentrations in dogs. Res Vet Sci 56, 129 131. Cupitt JM, Kasipandian V (2004) Pain and intramuscular injections. Anaesthesia 59, 93. De Castro J, Van der Water A, Wouters L et al. (1979) Comparative study of cardiovascular, neurological and metabolic side-effects of eight narcotics in dogs. Acta Anaesthesiol Belg 30, 5 99. Diggle PJ, Liang K-Y, Zeger SL (1994) Analysis of Longitudinal Data. Oxford Science Publications, Clarendon Press, Oxford, UK. Fox SM, Mellor DJ, Firth EC et al. (1994) Changes in plasma cortisol concentrations before, during and after analgesia, anaesthesia and anaesthesia plus ovariohysterectomy in bitches. Res Vet Sci 57, 110 118. Fox SM, Mellor DJ, Lawoko CRO et al. (1998) Changes in plasma cortisol concentrations in bitches in response to different combinations of halothane and butorphanol, with or without ovariohysterectomy. Res Vet Sci 65, 125 133. Fox SM, Mellor DJ, Stafford KJ et al. (2000) The effects of ovariohysterectomy plus different combinations of halothane anaesthesia and butorphanol analgesia on behaviour in the bitch. Res Vet Sci 68, 265 274. Glenski JA, Friesen RH, Lane GA, et al. (1988) Low-dose sufentanil as a supplement to halothane/n 2 0 anaesthesia in infants and children. Can J Anaesth 35, 379 384. Hall RI, Murphy MR, Hug CC (1987) The enflurance sparing effect of sufentanil in dogs. Anesthesiology 67, 518 525. Hall LW, Clarke KW, Trim CM (2001) Veterinary Anaesthesia (10th edn). WB Saunders, London, UK, pp. 96 101. Hansen BD, Hardie EM, Carroll GS (1997) Physiological measurements after ovariohysterectomy in dogs: what s normal? Appl Anim Behav Sci 51, 101 109. Heard DJ, Webb AI, Daniels RT (1986) Effect of acepromazine on the anaesthetic requirement of halothane in the dog. Am J Vet Res 47, 2113 2115. Holton LL, Scott EM, Nolan AM et al. (1998) Comparison of 3 methods used for assessment of pain in dogs. J Am Vet Med Assoc 212, 61 66. Holton L, Reid J, Scott EM et al. (2001) Development of a behaviour-based scale to measure acute pain in dogs. Vet Rec 148, 525 531. Howell SB (2001) Clinical applications of a novel sustained-release injectable drug delivery system: Depofoam TM technology. Cancer J 7, 219 227. Kehlet H (1989) Surgical Stress: the role of pain and analgesia. Br J Anaesth 63, 189 195. Kehlet H, Holte K (2001) Effect of postoperative analgesia on surgical outcome. Br J Anaesth 87, 62 72. Kyles AE, Hardie EM, Hansen BD (1998) Comparison of transdermal fentanyl and intramuscular oxymorphone on post-operative behaviour after ovariohysterectomy in dogs. Res Vet Sci 68, 245 251. Lascelles BDX, Butterworth SJ, Waterman AE (1994) Postoperative analgesic and sedative effects of carprofen and pethidine in dogs. Vet Rec 134, 187 191. Lascelles BDX, Cripps PJ, Jones A et al. (1997) Postoperative hypersensitivity and pain: the pre-emptive value of pethidine for ovariohysterectomy. Pain 73, 461 471. Lascelles BDX, Cripps PJ, Jones A et al. (1998) Efficacy and kinetics of carprofen, administered preoperatively or postoperatively, for the prevention of pain in dogs undergoing ovariohysterectomy. Vet Surg 27, 568 582. 326 Ó 2006 The Authors. Journal compilation Ó 2006 Association of Veterinary Anaesthetists, 33, 313 327