Materials and Methods

Animals—Dogs were selected from those admitted to 18 veterinary practices in France and 7 veterinary practices in Germany. All dogs were scheduled to undergo major orthopedic surgery. Because all dogs were client-owned animals, they were housed, managed, and fed as pets. Informed consent was obtained from owners of all dogs. The study was approved by the French and German regulatory authorities and Novartis committees after consideration of scientific, ethical, and animal welfare guidelines. The study was conducted in compliance with standards for good clinical practice¹³ and published guidelines for studies conducted on efficacy of NSAIDs.¹⁴ The report for the study was prepared by taking into account the CONSORT statement recommendations on reporting of noninferiority and equivalence randomized trials.¹⁵

Inclusion criteria for the study included dogs ≥ 6 weeks old (regardless of breed or sex) that weighed between 2.5 and 80 kg and were undergoing major orthopedic surgery. Exclusion criteria included dogs known to be pregnant or lactating, dogs with severe concomitant disorders of the kidneys, liver, or gastrointestinal tract that may have interfered with the evaluation of response to treatment, dogs that had received locally or systemically administered NSAIDs or opioids within 24 hours before inclusion in the study, or dogs that received corticosteroids within 30 days before inclusion in the study. In addition, once dogs were admitted to the study, they could be excluded if they had concomitant disorders (eg, trauma) that may have interfered with the evaluation of response to treatment, received a forbidden concomitant treatment (eg, rescue analgesia), or had adverse events that required cessation of treatment or if there was failure of compliance by the owner to administer drugs in accordance with the treatment protocol. Drugs likely to affect assessment of the main efficacy variables were disallowed; these drugs included analgesics (opioids and α₂-receptor agonists), other NSAIDs, corticosteroids, macrolides, and tetracyclines. All other concomitant treatments were recorded. Standard variables relating to anesthetic premedication, induction, maintenance, and recovery were recorded.

Study design—The study was a multicenter, prospective, parallel-group, randomized, blinded field study Dogs were allocated into 2 treatment groups (robenacoxib and meloxicam) in a ratio of 2:1 (robenacoxib:meloxicam). Allocation to treatment groups was via a randomization list prepared by one of the authors (WS). No separate randomization was performed for age, body weight, or sex of dogs. Case allocation was stratified according to each attending veterinarian and the anticipated duration of surgery (which was dichotomized to ≤ 1 hour and > 1 hour). Maximum number of dogs assessed by any attending veterinarian was 12 for robenacoxib and 6 for meloxicam.

Because robenacoxib and meloxicam formulations differed, blinding was maintained via a doubleinvestigator technique: the attending veterinarian was responsible for clinical assessments, whereas a second person (ie, the dispenser) was responsible for treatment prescription and drug administration. Four times during the study, the attending clinician accidentally gained knowledge of the treatment administered to a dog; each was during the final clinical examination of a dog. However, all dogs were included in the analysis because treatment was revealed to attending veterinarians after completion of the last clinical examination. Owners had knowledge of the treatments administered orally.

Two sets of numbered randomization envelopes (1 for surgeries with a predicted duration ≤ 1 hour and 1 for surgeries with a predicted duration > 1 hour) were prepared for each investigation center. After a dog was entered into the study by an attending veterinarian, the dispenser selected the envelope bearing the lowest number in the set corresponding to the anticipated duration of surgery and opened it to reveal the treatment allocation. Predicted durations were ≤ 1 hour for 38 dogs assigned to receive robenacoxib and 11 dogs assigned to receive meloxicam and > 1 hour for 59 dogs assigned to receive robenacoxib and 32 dogs assigned to receive meloxicam.

Treatments—Dogs in the robenacoxib group received an initial dose of robenacoxib^a (2 mg/kg, SC) at the time of baseline assessment (prior to surgery) and then received additional doses of robenacoxib^b (1 to 2 mg/kg, PO, q 24 h) beginning at 24 hours after extubation (day 1) and continuing for up to 15 days. The tablets contained 5 or 20 mg of robenacoxib/tablet and could not be divided to more closely approximate the correct dosage; tablets were administered with or without food.

Dogs in the meloxicam group received an initial dose of meloxicam^c (0.2 mg/kg, SC) at the time of baseline assessment (prior to surgery) and then received additional doses of meloxicam^d (0.1 mg/kg, PO, q 24 h) beginning on day 1 and continuing for up to 15 days. Meloxicam was mixed with food for oral administration.

Efficacy examinations and collection of samples—An efficacy examination was performed on all dogs before induction of anesthesia (baseline). Additional examinations were performed after surgery at 1, 2, 4, 8, and 24 hours after extubation (day 1) and at day 12.

Efficacy assessment—The primary end point for efficacy comprised the sum of scores for a Glasgow pain scale¹⁶ (Appendix), which was assessed by a clinical investigator at each of 6 time points (1, 2, 4, 8, and 24 after extubation and 12 days after surgery [time of suture removal]). Maximum score for the Glasgow pain scale is 24; however, mobility (score B) could not be assessed in 29 dogs at early time points (total of 74 missing data points). Therefore, the primary end point of the study was a modified Glasgow score (maximum score, 20) that consisted of parts A, C, and D of the Glasgow pain scale but excluded part B.

Six secondary efficacy end points were assessed by clinical investigators and owners. Clinical investigators assessed pain in resting dogs by use of a VAS with a scale ranging from 0 (no pain) to 100 (severe pain manifested by vocalization, aggression, and refusal to allow examination) and also during gentle palpation or manipulation of the affected limb or joint by use of a VAS with a scale ranging from 0 (no pain elicited) to 100 (severe pain manifested by vocalization, aggression, and refusal to allow examination).⁴ These VAS assessments were performed 7 times: at the time of administration of the robenacoxib or meloxicam prior to induction of anesthesia (baseline); at 1, 2, 4, 8, and 24 hours after extubation; and on day 12. Clinical investigators also made a global assessment of the efficacy (overall pain control) at 1 and 24 hours after extubation by use of a 4-point scale (0 = excellent, 1 = good, 2 = fair, and 3 = poor). Clinical investigators made an assessment of inflammation on day 12 on the basis of swelling of the affected joint or limb, local heat, redness, or paresis by use of a VAS with a scale ranging from 0 (no inflammation) to 100 (major inflammation).

Owners had knowledge of the treatments administered orally. Therefore, owner assessments of efficacy were secondary end points. Owners provided their perception of the severity of their dog's condition assessed daily from day 1 to day 12 by use of variables consistent with the Glasgow pain scales for demeanor (0 = normal, 1 = slightly modified, 2 = moderately modified, 3 = markedly modified, and 4 = severely modified) and mobility (0 = normal, 1 = slightly impaired, 2 = moderately impaired, 3 = markedly impaired, and 4 = severely impaired). Mobility could not be assessed in all dogs.

Plasma cortisol concentration—Venous blood samples were collected into tubes containing EDTA at baseline, extubation, and 1, 2, 4, and 8 hours after extubation and used for measurement of plasma cortisol concentrations. Plasma samples were stored at −20°C prior to analysis. Cortisol was measured by use of a commercial radioimmunoassay kit^e by personnel at the National Veterinary School of Toulouse, France. Within-day and between-day precisions were < 14%, and the accuracy ranged from 93% to 109%. The limit of quantitation of the assay was 10 ng/mL.

Safety assessment—The safety of administered treatments was assessed on the basis of adverse events, local tolerance, results of hematologic and clinical biochemical analyses, and buccal mucosal bleeding time. An adverse event was defined as any observation that was unfavorable or unintended, regardless of whether it was considered to be a product-related event. Local tolerance was assessed by the dispenser as evidence of pain at the injection site at the time of the SC injection (baseline) on a scale ranging from 0 (no pain) to 3 (severe pain).

Venous blood samples (2 mL) were collected into tubes containing EDTA at baseline, day 1, and day 12 and used for measurement of hematologic variables (RBC counts, WBC counts, differential WBC counts, Hct, and hemoglobin concentration). Additional venous blood samples (5 mL) were collected at the same 3 times into serum tubes and used for measurement of serum activities of alanine aminotransferase, alkaline phosphatase, aspartate aminotransferase, and γ-glutamyltransferase and concentrations of total protein, urea, and creatinine.

Buccal mucosal bleeding times were measured at baseline and day 1.

Statistical analyses—Statistical tests were performed by use of commercially available software.^f All evaluations were performed as 2-sided tests with significance set at 5%. The 95% CIs were calculated for the assessment of noninferiority. No correction for multiple tests was conducted because there was a single primary end point on which conclusions of efficacy were based, and the 6 secondary end points were used as supportive information only.

The experimental unit was each dog. All dogs that were allocated into 1 of the 2 groups were used for data analysis (intention-to-treat population), except for 1 dog that was excluded because no surgery was performed (pharyngitis did not allow intubation). As defined in the protocol, the primary analyses were conducted on the intention-to-treat population, and unless stated otherwise, results were for the intention-to-treat population.

Results

Baseline data and dosages administered—A total of 141 dogs were included in the study. One dog was removed because it did not undergo surgery (pharyngitis prevented intubation); thus, 140 dogs (97 received robenacoxib and 43 received meloxicam) were used for the intention-to-treat analysis. Of the 140 dogs, 136 were in the per-protocol population because 4 dogs had major deviations from the protocol (postinclusion removal because of forbidden concomitant treatment [n = 3 dogs] and an adverse event unrelated to treatment [ie, trauma;1 dog]). Data for these dogs were included in intention-to-treat analyses, although a separate perprotocol analysis was also performed for the primary end point.

Results for demographic and baseline variables plus types of orthopedic surgery were calculated (Table 1). There were no significant differences between the 2 groups at baseline, except for the frequency of surgery during the preceding 6 months and the frequency of surgeries categorized as other. Baseline (prior to anesthesia) pain score did not differ significantly between the groups, either when dogs were at rest (P = 0.39) or during palpation or manipulation of the affected joint or limb (P = 0.21). The duration of surgery (mean, 1.27 hours) and intubation (mean, 2.1 hours) did not differ significantly between groups. The most frequent types of surgery were performed for repair of ruptured cruciate ligaments (44/140 [31%]) and fractures (35/140 [25%]).

The targeted dosage of robenacoxib for SC administration was 2.00 mg/kg. Actual mean dosage administered was 2.01 mg/kg (minimum, 1.68 mg/kg; maximum, 2.29 mg/kg). The targeted dosage of meloxicam for SC administration was 0.20 mg/kg; actual mean dosage administered was 0.22 mg/kg (minimum, 0.17 mg/kg; maximum, 0.50 mg/kg).

The targeted dosage of robenacoxib for oral administration was 1 to 2 mg/kg administered once daily. Because the tablets could not be divided, actual dosages varied. The mean dosage administered orally was 1.40 mg/kg (median, 1.36 mg/kg; minimum, 0.98 mg/kg; and maximum, 1.98 mg/kg). The tablets were administered with or without food, as determined by each owner, for up to 15 days (mean ± SD, 11.5 ± 2.7 days; median, 12 days). Eighty-eight of 97 (91%) dogs received orally administered doses for ≥ 10 days. In the remaining 9 (9%) dogs, the shorter duration of treatment was attributable to owner error or withdrawal of the dog from the study. For the 97 dogs, robenacoxib tablets were administered in the food (21 [22%] dogs), with a small quantity of food (42 [43%] dogs), directly in the mouth other than at feeding time (31 [32%] dogs), or not specified (3 [3%] dogs).

The targeted dosage of meloxicam for oral administration was 0.1 mg/kg. Meloxicam was administered (in accordance with the manufacturer's instructions) mixed with food for up to 15 days (mean, 11.3 ± 1.8 days; median, 11 days). Mean dosage orally administered was 0.10 mg/kg (median, 0.10 mg/kg; minimum, 0.09 mg/kg; and maximum, 0.15 mg/kg). Forty of 43 (93%) dogs were orally administered meloxicam for ≥ 10 days. In the other 3 (7%) dogs, the shorter duration of administration was attributable to owner errors or withdrawal of the dog from the study.

For both treatment groups, dosing compliance was monitored by the dispenser. Owners were required to complete and return accounting forms and to return all unused products and empty packets.

Rescue treatment and concomitant treatments—Attending clinicians were authorized to provide rescue treatment during the first 24 hours after surgery. However, no dogs required such treatment.

Attending clinicians were advised to avoid concomitant treatment, if possible. Concomitant treatments included drugs used for anesthesia that were permitted in the study protocol. Premedicants administered included no drugs (n = 33 dogs), acepromazine (70), diazepam (34), glycopyrollate (4), and atropine (1). Opioids and α₂-receptor agonists were not allowed as premedicants. Anesthesia was induced with thiopentone (n = 115 dogs) or propofol (25) and maintained with halothane (79) or isoflurane (61). There were no significant (P = 0.94) differences between the 2 groups with regard to the drugs used and no significant (P = 0.37) difference in the amount of time that elapsed between SC administration of meloxicam or robenacoxib and extubation.

Additional allowed concomitant treatments were primarily fluids (n = 122 dogs) and antimicrobials administered via injection (121) or orally (111). There were no significant differences between the groups with regard to distribution of these treatments.

Three dogs (all in the robenacoxib group) were administered 4 disallowed drugs (dexamethasone, morphine, ketamine, and meloxicam). All drugs were administered as single doses. Data from these dogs were retained in the intention-to-treat analysis but were not included in the per-protocol analysis.

Efficacy—The primary end point defined in the study protocol was the arithmetic sum of the 6 scores assigned by a clinical investigator for the Glasgow scale at 1, 2, 4, 8, and 24 hours after extubation and at day 12. The maximum possible score was 24. However, the protocol provided for the possibility that the mobility score would not be assessed if locomotion was not possible at the time of assessment. Thus, mobility was not assessed in 29 dogs (maximum possible score for these dogs was 20). Because it is not valid to compare dogs for which data with mobility were scored or were not scored, it was decided that the main primary end point for the study would be the modified Glasgow score without the mobility assessment. However, similar results were obtained for assessments performed with the entire Glasgow scale and with the modified Glasgow scale with mobility omitted (Table 2).

The fact that pain was well controlled in the immediate postoperative period by both robenacoxib and meloxicam was indicated by mean scores for the Glasgow pain scale of < 5 (111 dogs with mobility assessment excluded; maximum score = 20; Figure 1) and < 7 (140 dogs with mobility assessment included; maximum score = 24) at 1 and 2 hours after extubation. Thereafter, scores decreased steadily, and mean values for both groups were < 1 (111 dogs with mobility assessment excluded) and < 2 (140 dogs with mobility assessment included) by day 12.

Mean ± SD scores for a modified Glasgow pain scale (ie, without the score for mobility) assessed by clinical investigators as the primary end point for dogs treated with robenacoxib (n = 97 [black circles]) or meloxicam (43 [white circles]) to alleviate pain associated with orthopedic surgery. Each dog in its respective treatment group received an initial dose (2 mg of robenacoxib/kg or 0.2 mg of meloxicam/kg, SC) before surgery and daily doses (1 to 2 mg of robenacoxib/kg or 0.1 mg of meloxicam/kg, PO) for up to 15 days after surgery. The score represents an NRS ranging from 0 to 20.

Citation: American Journal of Veterinary Research 72, 2; 10.2460/ajvr.72.2.184

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Mean ± SD scores for a modified Glasgow pain scale (ie, without the score for mobility) assessed by clinical investigators as the primary end point for dogs treated with robenacoxib (n = 97 [black circles]) or meloxicam (43 [white circles]) to alleviate pain associated with orthopedic surgery. Each dog in its respective treatment group received an initial dose (2 mg of robenacoxib/kg or 0.2 mg of meloxicam/kg, SC) before surgery and daily doses (1 to 2 mg of robenacoxib/kg or 0.1 mg of meloxicam/kg, PO) for up to 15 days after surgery. The score represents an NRS ranging from 0 to 20.

Citation: American Journal of Veterinary Research 72, 2; 10.2460/ajvr.72.2.184

• Download Figure
• Download figure as PowerPoint slide

Mean ± SD scores for a modified Glasgow pain scale (ie, without the score for mobility) assessed by clinical investigators as the primary end point for dogs treated with robenacoxib (n = 97 [black circles]) or meloxicam (43 [white circles]) to alleviate pain associated with orthopedic surgery. Each dog in its respective treatment group received an initial dose (2 mg of robenacoxib/kg or 0.2 mg of meloxicam/kg, SC) before surgery and daily doses (1 to 2 mg of robenacoxib/kg or 0.1 mg of meloxicam/kg, PO) for up to 15 days after surgery. The score represents an NRS ranging from 0 to 20.

Citation: American Journal of Veterinary Research 72, 2; 10.2460/ajvr.72.2.184

• Download Figure
• Download figure as PowerPoint slide

The ANOVA estimates for differences in overall group means and pairwise group mean values were determined (Table 2). For both primary and secondary efficacy variables, mean values were similar between the groups and there were no significant differences.

A noninferiority analysis was performed (Table 3). Noninferiority of robenacoxib relative to meloxicam was indicated by CI values > 0.80 for the primary end point (ie, the sum of Glasgow pain scale scores with or without mobility assessment). In addition to the main intention-to-treat analysis for 140 dogs, noninferiority analysis was repeated on the per-protocol population of 136 dogs. Similar results were obtained with both analyses. For the primary end point of the per-protocol population, the reciprocal of the scores for the roben-acoxib-to-meloxicam ratio was 1.12 (95% CI, 0.97 to 1.30), whereas it was 1.16 (95% CI, 0.98 to 1.37) for the intention-to-treat population. Therefore, there was noninferiority for both populations.

Plasma cortisol concentration—Plasma cortisol concentrations were measured in samples obtained at baseline and up to 8 hours after extubation (Table 4). Significant (P < 0.001) and quantitatively similar increases were obtained for both groups, when compared with preanesthesia values. However, concentrations in both groups were similar to the preanesthesia values by 4 hours after extubation. Comparison of the groups by use of an ANOVA indicated no significant (P = 0.31) differences between the robenacoxib and meloxicam groups.

Safety—The frequency of dogs with recorded adverse events did not differ significantly (P = 0.45) between groups (robenacoxib, 24/97; meloxicam, 10/43). Most adverse events were classified as benign. Most reports concerned the gastrointestinal tract (vomiting, diarrhea, dark feces, or soft feces). A single life-threatening event (lymphangiectasia and edema of the hind limbs of a dog following surgery to remove cement from a hip joint prosthesis) was classified as unrelated to treatment.

Pain at the site of SC injections was minimal. It was reported as absent in 92 dogs, slight in 4 dogs, and was not assessed in 1 dog (robenacoxib group) and was absent in 40 dogs and slight in 3 dogs (meloxicam group). Mean ± SD scores for pain at injection sites were 0.04 ± 0.20 for robenacoxib and 0.07 ± 0.26 for meloxicam and did not differ significantly (P = 0.68) between groups.

Mean ± SD buccal mucosal bleeding time after administration of robenacoxib was not significantly (P = 0.78) different at 24 hours after extubation (137 ± 48 seconds), compared with the value before surgery (138 ± 55 seconds). For meloxicam, bleeding times also did not differ significantly (P = 0.057) between values obtained before surgery (131 ± 51 seconds) and 24 hours after extubation (117 ± 45 seconds).

Analysis of hematologic and clinical biochemical data revealed no changes that were regarded as biologically relevant. However, there were several variables whose values differed significantly from the baseline values, and these generally were detected in both treatment groups. The RBC count, Hct, and γ-glutamyltransferase activity decreased in both groups at 1 and 24 hours after extubation. The hemoglobin concentration decreased in the robenacoxib group at 1 and 24 hours after extubation. Numbers of eosinophils and platelets increased in both groups at day 12. The WBC count and number of neutrophils increased at 1 and 24 hours after extubation in both groups and at day 12 in the meloxicam group. Numbers of lymphocytes and monocytes and the alkaline phosphatase activity increased at 1 and 24 hours after extubation and at day 12 in both groups. The alanine aminotransferase activity decreased in the robenacoxib group at day 12, where-as the aspartate aminotransferase activity increased at 1 and 24 hours after extubation in both groups and decreased at day 12 in the robenacoxib group. Total protein concentration decreased at 1 hour after extubation and at day 12 in both groups. Urea concentration decreased at 1 and 24 hours after extubation in both groups and increased at day 12 in the meloxicam group. Finally, the creatinine concentration decreased at 1 and 24 hours after extubation in both groups and increased at day 12 in the robenacoxib group.

Discussion

The principal finding of the study reported here was that a treatment regimen consisting of a single SC injection of robenacoxib before surgery followed by once-daily oral administration of robenacoxib tablets for up to 15 days had similar efficacy and tolerability for the management of pain and inflammation associated with orthopedic surgery when compared with results for a positive control treatment (ie, meloxicam injection followed by oral administration of meloxicam). Both robenacoxib and meloxicam provided good efficacy, which was evident from the lack of rescue treatments required for any dog and the low scores for pain and inflammation at all time points after surgery. Therefore, the results are clinically relevant and provide statistical proof of the similarity of effect.

A major strength of this study was that it was designed as a multicenter, prospective, randomized, parallel-group, blinded field trial. The study had adequate power because the similarity of robenacoxib and the positive control treatment was proven statistically for the primary end point. Three major factors contributed to the power in the study. First, the mean efficacy of robenacoxib was numerically higher than for meloxicam (1.16 for the ratio of the reciprocals of the primary end point). Second, there was moderate variability (SEs were < 10% of the mean) for the primary end point. Third, there were few serious protocol deviations that resulted in premature withdrawal of dogs from the study.

It can be difficult to clinically assess pain in animals. In the study reported here, the primary end point was a clinical investigator's assessment of pain by use of a Glasgow pain scale for 6 separate indices. At the time of the study, the weight for the factors to be applied to the components of the scale had not yet been published^1,17; thus, each variable of the scale was assessed as an NRS. This assessment was supported by 6 secondary end points (evaluation of pain at rest, pain during palpation or manipulation, and inflammation assessed by a clinical investigator via a VAS; a global assessment of efficacy conducted by a clinical investigator via an NRS; and the owner's evaluation of demeanor and mobility over the duration of the study).

The study did have limitations. First, the number of dogs included (97 for robenacoxib and 43 for meloxicam) was not perfectly balanced in accordance with the ratio of 2:1 planned in the study design. This was the result of the fact that 10 veterinarians each recruited < 4 dogs for the study and 6 veterinarians each recruited only a single dog for the study. Because the protocol did not define a minimum number of dogs per site, all dogs were included in the analysis.

Second, although the dogs were assessed fairly intensively by a veterinarian during the first 24 hours after surgery, follow-up assessments (from day 2 until the final visit to a veterinarian) were less intensive and relied on each owner's daily assessments of their dog's demeanor and mobility. This technique was used because it was judged unacceptable to hospitalize the dogs unnecessarily beyond 24 hours after surgery or to require extra visits to a veterinarian.

Third, although the assessments made by the veterinarians were conducted in a blinded manner via the use of a dispenser to manage the test treatments, dog owners were aware of the treatment. Therefore, the primary end point of the study was based on assessments performed by a veterinarian and not on assessments performed by the owners.

Finally, the limitations of noninferiority studies that involve use of positive control treatments have been established.¹⁵ In the study reported here, use of a placebo would have been unethical because a number of NSAIDs are registered in the European Union for perioperative use in dogs and are widely used. Meloxicam was selected as the positive control treatment because it is used extensively in dogs and has proven efficacy for alleviation of perioperative pain and inflammation.

The efficacy of meloxicam in the management of pain and inflammation associated with orthopedic surgery in dogs has been evaluated in several studies. A single dose of meloxicam (0.2 mg/kg, IV) administered in combination with butorphanol prior to surgery resulted in significantly lower scores for an adjusted pain score at some time points, compared with the effects for butorphanol alone, in 40 dogs undergoing cruciate surgery.⁸ In another field study,¹⁰ there was slight but not significant efficacy for meloxicam versus a negative control treatment in 20 dogs undergoing cruciate repair and epidural anesthesia. In 3 additional field studies,^9,11,12 meloxicam was reported to have efficacy equivalent to that of carprofen, fentanyl, and ketoprofen. In addition, the efficacy of meloxicam at the dosages used in the present study has been established, compared with results for a placebo, in animals with experimentally induced urate synovitis. Administration of 0.2 mg of meloxicam/kg via SC injection¹⁸ and repeated oral administration of 0.1 mg of meloxicam/kg¹⁹ had significantly better efficacy, compared with results for a placebo. Therefore, we conclude that there is satisfactory evidence for the efficacy of meloxicam in control of pain and inflammation in dogs undergoing orthopedic surgery. However, the Glasgow pain scale was not used in any of these aforementioned studies. In optimally designed noninferiority studies, the methods and outcome measures should be similar to those used in the original evaluation of the active control product.¹⁵

We are not aware of any published data on the efficacy of orally administered meloxicam for the treatment of postoperative pain and inflammation for up to 15 days, as was used in the present study. Although the pain associated with osteoarthritis in dogs may differ in character and intensity from that after orthopedic surgery, it was found in a meta-analysis of 16 clinical trials that only for meloxicam did a high degree of comfort exist in relation to efficacy for the treatment of osteoarthritis in dogs.²⁰ As mentioned previously, oral administration of meloxicam at a dosage of 0.1 mg/kg had significant efficacy in animals with experimentally induced urate synovitis.¹⁹ Therefore, there is sufficient evidence for the efficacy of meloxicam to justify its use as the positive control treatment in the study reported here. The safety of meloxicam for use during surgery in dogs has also been evaluated, with no reported adverse effects on the cardiovascular system and kidneys.^21–23

An additional limitation of the noninferiority design of the present study was the choice of a noninferiority threshold (ie, δ) value of 0.20, the same value recommended in guidelines for bioequivalence studies. Noninferiority margins should be the largest margin that can be clinically acceptable; to our knowledge, specific recommendations for noninferiority margins for veterinary NSAIDs have not been published. In fact, the results indicated that efficacy of robenacoxib similar to that of the positive control treatment would also have been achieved if we had defined a value of δ = 0.05 for the primary end point (ie, a maximum difference of 5% for modified Glasgow pain scores).

Clinical investigators were instructed to provide rescue treatment if they judged it to be necessary. None of the dogs required rescue treatment, which may indicate that the measurements were insensitive. However, the low scores for pain at all times after surgery support an alternative conclusion that both treatment regimens were effective in preventing most of the perioperative pain and inflammation so that additional treatment was not necessary.

Plasma cortisol concentrations at the time of extubation in both groups were twice as high as the preanesthesia values. However, the increase was transient, and concentrations returned to baseline values by 4 hours after extubation. The extent and duration of the observed increase in cortisol concentrations are consistent with results of other studies^24–26 in which investigators reported increases in concentrations (by a factor of 2 to 3) that lasted 2 to 4 hours. In addition to anesthesia, the stress of surgery (especially major surgical intervention involved in most orthopedic cases), with associated cardiovascular and body fluid changes, may also increase circulating cortisol concentrations. In these clinical cases, the mean duration of surgery was 75 minutes. There were no significant differences between the treatment groups with regard to cortisol concentrations. In another study,⁸ significantly lower plasma cortisol concentrations were detected in dogs undergoing cruciate repair and receiving meloxicam than in dogs that did not receive an NSAID. However, no changes in plasma cortisol concentrations were detected after meloxicam administration in dogs undergoing mammary gland excision in yet another study.²³

Hematologic and clinical biochemical data, results of examinations by the clinical investigators, and reported adverse events indicated that both NSAIDs investigated in the present study were tolerated well. Although a range of adverse effects were reported, these were all assessed as mild or benign (except for 1 adverse effect that was not a treatment-related event), and most were judged to be not related to treatment or only questionably related to treatment. The most frequently reported adverse effects involved the gastrointestinal system. Such adverse effects as occasional vomiting and loose feces are commonly reported in trials of this kind, including those that have incorporated a placebo treatment.

No change in buccal mucosal bleeding time before and after surgery was observed with robenacoxib. This is not surprising because robenacoxib used at recommended dosages does not inhibit COX-1 in dogs.^6,18 Robenacoxib used at elevated dosages in safety studies did not affect buccal bleeding time or hematologic results.²⁷ Buccal bleeding time also did not increase with meloxicam treatment, as reported previously in dogs undergoing abdominal or orthopedic surgery.^3,28,29

We concluded that a treatment regimen consisting of a single SC injection of robenacoxib (2 mg/kg) before surgery followed by once-daily oral administration (1 to 2 mg/kg) of tablets after surgery for 10 to 15 days had similar efficacy and tolerability, compared with results for meloxicam, for the control of pain and inflammation in dogs undergoing orthopedic surgery.