Phenylbutazone and flunixin meglumine used singly or in combination in experimental lameness in horses


  • J. H. FOREMAN,

    Corresponding author
    1. Department of Veterinary Clinical Medicine, College of Veterinary Medicine, University of Illinois at Urbana-Champaign, Urbana, Illinois, USA
    Search for more papers by this author

    1. Department of Veterinary Clinical Medicine, College of Veterinary Medicine, University of Illinois at Urbana-Champaign, Urbana, Illinois, USA
    Search for more papers by this author

  • Dr Ruemmler's current address is Boston Equine Associates, 28 Tremont Street, Rehoboth, Massachusetts 02769, USA.



Reason for performing study: Using an adjustable heart bar shoe model of foot pain, the objective of this study was to test the hypothesis that the combined use of phenylbutazone (PBZ) and flunixin meglumine (FM) would prove more efficacious in alleviating lameness than either drug alone.

Materials and methods: One hour after induction of lameness at weekly intervals, 8 healthy adult Thoroughbred horses randomly underwent one of 4 i.v. treatments: saline (SAL) placebo (1 ml/45 kg bwt), PBZ (4.4 mg/kg bwt), FM (1.1 mg/kg bwt) or PBZ+FM (at the same dosages as given individually). Heart rate (HR) and lameness score (LS) responses were assessed in a blinded manner every 20 min for 5 h after lameness induction and then hourly for 12 h after treatment. Jugular venous blood samples were obtained at -1, 0, 0.05, 1, 2, 4, 6, 8, 10 and 12 h and subsequently analysed for drug concentrations. Repeated measures ANOVA and post hoc Tukey's test were used to identify analgesic effects at a significance level of P<0.05.

Results: Heart rate was lower in all nonsteroidal anti-inflammatory drug (NSAID)-treated trials from 2 h to 10 h post treatment (P<0.05). Analgesic effects of FM and PBZ+FM, as evidenced by decreases in HR, lasted for 12 h post treatment (P<0.05). Lameness score decreased earlier in PBZ and PBZ+FM trials than in FM trials (P<0.05) and the analgesic effect on LS lasted for 12 h post treatment for all NSAID trials (P<0.05). Peak PBZ plasma concentration was 73.7 ± 6.0 and 77.9 ± 5.5 µg/ml. Peak FM concentration was 12.0 ± 0.8 and 13.7 ± 1.0 µg/ml

Conclusions: It was concluded that the combination of PBZ+FM was not more effective than either PBZ or FM alone. These data do not support the hypothesis that the combination is more efficacious at these dosages than either drug alone in this model of acute foot pain.


Governing jurisdictions for equestrian competitions have widely varying rules and regulations regarding the use of nonsteroidal anti-inflammatory drugs (NSAIDs) on the day of competition. The Fédération Equestre Internationale (FEI) has a zero tolerance policy for the detectable presence of any NSAIDs on competition day. Some states (such as Illinois) allow the presence of one NSAID (phenylbutazone [PBZ]) on race day, but only below certain approved threshold concentrations. Other racing jurisdictions have previously allowed the use of more than one NSAID on race day, sometimes in any plasma concentration on race day (Dirikolu et al. 2009). The unregulated use of multiple NSAIDs on competition day may confer on some horses an unfair advantage, while simultaneously putting those same horses at risk for NSAID toxicity. One recent toxicity study on the combined use of PBZ (orally) and flunixin meglumine (FM) i.v. for 5 days documented substantially greater toxicity (including gastric ulcers visible on endoscopy in 4 of 4 horses examined) in horses receiving combination therapy (Reed et al. 2006).

Unfortunately, there is a dearth of objective data regarding the clinical efficacy of combined PBZ and FM therapy in horses. The one published efficacy study combining these 2 drugs used 5 days of twice daily half-dose oral PBZ (2.2 mg/kg bwt) accompanied by twice daily single doses of FM i.v. (1.1 mg/kg bwt) and showed increased efficacy of the combination over the half dose of PBZ (Keegan et al. 2008). While the Reed et al. (2006) and Keegan et al. (2008) method of administration may be common in some horse show stables, it is more common in others to administer both compounds i.v. at full doses on the day of, or day prior to, competition or on training days in many racing and horse showing stables (S. Waterman, personal communication). Furthermore, practitioners often administer or enquire about the efficacy of administering PBZ and FM simultaneously i.v. to patients suffering from acute onset severe lameness including laminitis. Analgesic efficacy for concurrent, acute administration of both PBZ and FM i.v. has not previously been reported.

In earlier studies of these NSAIDs administered singly using this model of acute foot pain, FM tended to have an earlier onset of efficacy than PBZ, whereas PBZ tended to last longer in efficacy than FM (Foreman et al. 1994, 2008, 2010). The hypothesis for the current study was that simultaneous i.v. administration of PBZ and FM would provide greater analgesia than either drug used alone because the combination was hypothesised to act synergistically to work earlier (FM effect) and last longer (PBZ) than the individual compounds. The specific objective of this experiment was to compare the i.v. efficacy of PBZ, FM, PBZ and FM combined and isotonic saline (IS: negative control) in a reversible model of foot lameness in the horse. Variables monitored in a masked (blinded) manner included heart rate (HR), lameness score (LS) and plasma drug concentration.

Materials and methods

All materials and methods used for this study were performed under the approval and authority of the Institutional Animal Care and Use Committee.


Eight Thoroughbred horses (mean ± s.e. age 4.6 ± 0.6 years, median 4 years, range 3–7 years; 6 geldings, 2 females) weighing 443.2–511.4 kg (mean ± s.e. 473.4 ± 8.7 kg, median 476.5 kg) were studied for 4 weeks. Complete physical and lameness examinations were performed before shoeing to ensure that each subject was normal before experiments were begun. Each subject's left front foot had an adjustable heart bar shoe applied. The right front foot had a simple keg shoe of a similar weight applied for balance. A minimum of 7 days of stall rest were allowed after shoeing and before any drug trials were begun.

Lameness induction

A set screw was placed in the heart bar shoe to induce lameness 1 h prior to administration of treatment. Lameness was scored using a previously described decimal grading system (Foreman and Lawrence 1987; Foreman et al. 1990, 1994, 2008, 2010; Seino et al. 2003). Previous studies have shown that degree of HR elevation is a valid indicator of severity of lameness in horses with naturally occurring (Foreman et al. 1990) and experimentally induced lameness (Foreman and Lawrence 1987). Given that HR is a primary variable of interest in this model (Foreman and Lawrence 1987; Seino et al. 2003; Foreman et al. 2008, 2010), movement or exercise outside the stall confounds the use of HR measurement as an objective quantifiable determinant of lameness or response to medication. All lameness grading in this model therefore occurred in the individual box stall without jogging any horse and without walking any individual horse other than observing spontaneous walking and turning movements within the stall (Foreman et al. 2010). Lameness grades were based originally on the American Association of Equine Practitioners (AAEP) Lameness Scale (Foreman and Lawrence 1987), but were modified for use solely in the stall, similar to the Obel laminitis scale. Lameness grades were: Grade 0 (sound or undetectable lameness), Grade 1 (barely detectable lameness; horse rarely looked lame at a walk in the stall, mainly when turning and/or pointed the lame toe forward intermittently and rarely), Grade 2 (mild lameness; horse more consistently lame at a walk in the stall, had a mild head bob when walking or turning in the stall and pointed its toe more consistently), Grade 3 (moderate lameness but not nonweightbearing; horse had more obvious head bob at a walk, toe pointing more frequently), Grade 4 (nonweightbearing 50% of the time, severe head bob, toe pointing whenever not walking but not always 3-legged lame at a walk) and Grade 5.0 (nonweightbearing 100% of the time). Lameness that did not meet all the expressed criteria for a given grade were judged to be between 2 major points on the scale (e.g. a lameness might have been graded as Grade 3.3 rather than Grade 3 or 4 as the only possibilities for that horse at that point in time), thus rendering the scale continuous rather than a 6-point ordinal scale. A lameness grade of 5.0 was achieved initially in each horse in each trial.


Four treatments were studied. The combination of PBZ and FM was chosen because it was suspected to have good efficacy in daily use for years in Kentucky Thoroughbred racing and training (S. Waterman, personal communication; Dirikolu et al. 2009). Phenylbutazone1 at 4.4 mg/kg bwt i.v.; flunixin meglumine (Banamine)2, at 1.1 mg/kg bwt i.v.; both PBZ (4.4 mg/kg bwt i.v.) and FM (1.1 mg/kg bwt i.v.) in combination and saline (SAL) (negative control: 1 ml/45 kg bwt i.v.) were administered on a weekly basis in a randomised order. Treatments were administered by a coinvestigator who had no input on HR and LS determinations. Treatments were assigned using a Latin Square design so that all treatment order permutations were represented. During each treatment period, 2 horses received each treatment. Treatment periods were at least 7 days apart to ensure adequate washout of the previously administered treatment.

Sampling intervals

Parameters were monitored by an investigator unaware of treatment assignments. Heart rate and LS were determined every 20 min for the first 5 h after lameness induction and then hourly to 12 h post treatment. Heparinised jugular venous blood samples were obtained at rest (time -1), immediately pretreatment (time 0), 0.05 (immediately post treatment), 1, 2, 4, 6, 8, 10 and 12 h after treatment and were chilled, separated by centrifugation and frozen for transport and subsequent determination of plasma drug concentrations. Tension on the heart bar was released after 13 h on trial and each subject examined at a walk and trot for any adverse effects of the induced lameness. These techniques and monitoring intervals have been used previously successfully and with minimal lasting adverse effects on soundness in previous experiments in Thoroughbred horses undergoing racetrack exercise (Foreman et al. 1990) and in research horses with experimentally induced foot lameness (Foreman and Lawrence 1987; Foreman et al. 1994, 2008, 2010; Seino et al. 2003).

Plasma drug concentrations

Plasma drug concentrations were determined by an analyst who was masked with regard to sample collection times, but not to the identity of the administered drug. The analytical method used in this study to determine plasma drug concentrations was modified from a previously published method (Luo et al. 2004). Briefly, the method consisted of liquid/liquid extraction of the plasma sample followed by LC-MS-MS analysis. Ketoprofen was chosen as the internal standard because it is of similar polarity to the analytes of interest, yet it is inert under the extraction and detection conditions, it does not co-elute with the analytes and it is unlikely to be present in these samples collected from research horses in a controlled environment. The intra- and inter-assay coefficients of variation were less than 5%. Accuracy was 95% at plasma PBZ concentrations of 15 µg/ml and FM concentrations of 1 µg/ml.

Data analysis

Mean (± s.e.) HR, LS and plasma drug concentrations were determined for each treatment at each sampling interval. Lameness score was treated as a continuous variable since a continuous decimal grading scale was used. Significance of differences between treatments was analysed by repeated measures multivariate analysis of variance. When repeated measures documented significant differences between treatments over time, Tukey's test was used post hoc to identify those discreet points in time at which a significant effect between specific treatments existed. Significance level was set at P<0.05.


All treatments were administered at 0 h on the x axis (denoted by the arrow in Figs 1–4).

Figure 1.

Mean±s.e. heart rate (HR) in phenylbutazone (PBZ) trials was lower than saline (SAL) placebo from 2 through 10 h post treatment (P<0.05). Heart rate with flunixin meglumine (FM) alone and in PBZ+FM combined trials was lower from 2 to 12 h post treatment (P<0.05). Lameness was induced at -1 h and treatments administered at 0 h (denoted by arrow) on the x axis.

Figure 2.

Mean ±s.e. lameness score (LS) for all nonsteroidal treatment combinations was lower compared to saline (SAL) placebo trials from 1.3 to 12 h post treatment (P<0.05). The LS for phenylbutazone (PBZ) and PBZ+flunixin meglumine (FM) trials decreased more quickly, with onset of effect at 1 h post treatment, 20 min earlier than for FM alone (P<0.05). Lameness was induced at -1 h and treatments administered at 0 h (denoted by arrow) on the x axis.

Figure 3.

Mean±s.e. plasma phenylbutazone (PBZ) concentration was not different between trials where PBZ was administered either singly or in combination with flunixin meglumine (FM: P>0.05). Lameness was induced at -1 h and treatments administered at 0 h (denoted by arrow) on the x axis.

Figure 4.

Mean±s.e. plasma flunixin meglumine (FM) concentration was not different between trials where FM was administered either singly or in combination with phenylbutazone (PBZ: P>0.05). Lameness was induced at -1 h and treatments administered at 0 h (denoted by the arrow) on the x axis.

Heart rate (Fig 1)

Mean pretreatment HR ranged from 43.5 ± 2.2 to 49.5 ± 1.5 beats/min and was not different between any of the 4 treatment groups (P>0.05). After treatment, mean HR in SAL placebo trials was 39.0 ± 1.7–49.5 ± 1.5 beats/min. In PBZ trials mean HR was lower (range 36.5 ± 1.0–39.5 ± 1.7 beats/min) than SAL placebo (39.0 ± 1.7–48.5 ± 2.8 beats/min) from 2 to 10 h post treatment (P<0.05). Mean HR with FM alone (33.5 ± 1.2–40.0 ± 1.5 beats/min) and in PBZ+FM combined trials (33.5 ± 1.7–39.5 ± 2.3 beats/min) was lower than SAL placebo from 2 to 12 h post treatment (P<0.05).

Lameness score (Fig 2)

Mean pretreatment LS ranged from 4.8 ± 0.1 to 5.0 ± 0.0 and was not different between any of the 4 treatment groups (P>0.05). Mean LS for all NSAID treatments (range 0.3 ± 0.3–4.4 ± 0.2) was lower compared with SAL placebo trials (range 3.9 ± 0.7–4.9 ± 0.1) from 1.3 to 12 h post treatment (P<0.05). Lameness score for PBZ and PBZ+FM trials decreased more quickly, with onset of effect at 1 h post treatment, 20 min earlier than for FM alone (P<0.05).

Plasma phenylbutazone concentration (Fig 3)

Mean plasma PBZ concentration (Fig 3) peaked 5 min post injection at 73.7 ± 6.0 µg/ml (combination)−77.9 ± 5.5 µg/ml (PBZ alone). Mean trough plasma PBZ concentration at 12 h post administration was 12.2 ± 1.3 µg/ml (PBZ alone)−15.6 ± 3.5 µg/ml (combination). There was no difference in PBZ concentrations at any time point between trials where PBZ was administered either singly or in combination with FM (P>0.05).

Plasma flunixin meglumine concentration (Fig 4)

Mean plasma FM concentration (Fig 4) peaked 5 min post injection at 12.0 ± 0.8 µg/ml (FM alone)−13.7 ± 1.0 µg/ml (combination). Mean trough plasma FM concentration at 12 h post administration was 0.2 ± 0.0 µg/ml (FM alone)−0.4 ± 0.1 µg/ml (combination). There was no difference in FM concentrations at any time-point between trials where FM was administered either singly or in combination with PBZ (P>0.05).


This foot lameness model, combined with intermittent plasma drug concentration sampling, comprises a simple, repeatable pharmacodynamic model that more clearly delineates plasma analgesic efficacy thresholds for commonly used analgesic drugs such as PBZ, FM and ketoprofen (Foreman et al. 2008). Modelling to determine the degree of plasma protein binding of the administered NSAID is not necessary in this model, nor was it a goal of this project. Plasma samples for determination of drug concentrations were obtained exactly as they would be for any horse undergoing drug testing at a racetrack or at a horse show, where no consideration is given to plasma protein binding.

Phenylbutazone concentrations in this experiment remained higher for a longer period of time than in previous studies (Foreman et al. 2008). In one of the first studies on PBZ use in horses, Piperno et al. (1968) showed that plasma concentrations decreased to approximately 5 and 3 µg/ml at 6 and 9 h post administration, respectively, in keeping with Foreman et al. (2008). While it might be tempting to ascribe the apparently longer-lasting PBZ concentrations in this experiment to be due to the simultaneous use of FM, PBZ concentrations were no different between the single NSAID alone or PBZ+FM treated trials (Fig 3). Neither drug seemed to be affected by the presence or absence of the other based on the similar plasma concentrations for each drug regardless of single or simultaneous administration (Figs 3, 4). This finding is in keeping with previous work that has shown no alteration in the pharmacokinetics of either drug when administered singly or concurrently i.v. in the same dosages used in our experiment (Semrad et al. 1993).

The data presented here show that the combination of PBZ and FM produced no better analgesia than PBZ or FM alone. We are of the opinion that the demonstration of greater efficacy with the FM+PBZ combination by Keegan et al. (2008) was primarily a factor of their having chosen their control group to be those horses treated with half-dose PBZ whereas we compared full-dose FM, PBZ or the combination of FM+PBZ to one another. Previous work with this model has shown that half doses of PBZ (2.2 mg/kg bwt i.v.) have limited efficacy (only about 2 h) compared with full or single doses (4.4 mg/kg bwt i.v.) (J.H. Foreman, unpublished data). In our model, double doses (8.8 mg/kg bwt i.v.) have no greater efficacy than single doses, in keeping with previous work by Hu et al. (2005), demonstrating that a single, full therapeutic dose maximises cyclo-oxygenase blockade. We believe that the positive results of the combination study by Keegan et al. (2008) were primarily due to their choice of a control medication (PBZ 2.2 mg/kg bwt per os b.i.d.) given at a dose unlikely to provide complete efficacy for 12 h after dosing (due to incomplete cyclo-oxygenase saturation). When a combination of half-dose PBZ and full-dose FM (1.1 mg/kg bwt i.v. b.i.d.) was compared with half-dose PBZ (Keegan et al. 2008), the combination would be likely to be additive and to have greater efficacy than the half-dose control treatments. Based on previous experiments with both these drugs individually in this model (Foreman et al. 1994; 2008, 2010), we hypothesised that FM would cause earlier onset of efficacy than PBZ alone, and that PBZ would cause a longer duration of effect in combination than would FM alone. When we controlled for dose in this experiment, so that each medication was administered in a full dose whether administered singly or in combination, we saw no greater efficacy from the combination than from either drug alone. The expected synergy (based on previously observed onset or duration activities of the individual compounds) did not prove to be true using these doses in this manner in this model.

Concerns regarding the overuse of NSAIDs in competition are 2-fold (MacAllister et al. 1993; Foreman et al. 2008, 2010). First, it is possible that NSAID use may provide an unfair advantage in that an unsound horse may, with medication, be able to compete on an equal plane with one which is innately sound and that requires no external medication. This concern is one of fairness and in the lay press recently has been referred to as masking, or the use of an NSAID to hide a lameness (as opposed to the more traditional use of the term masking to refer to the use of one medication to inhibit detection or hide the use of another prohibited medication, e.g. frusemide used to dilute urine). The second concern is one of toxicity. There is a considerable amount of data which shows that NSAIDs in normal or excessive dosages can precipitate a wide array of clinical problems in horses and ponies (Snow et al. 1979, 1980, 1981; Traub et al. 1983; Collins and Tyler 1985; MacAllister et al. 1993; Reed et al. 2006). These diseases include gastric ulceration, right dorsal colitis, renal papillary necrosis and renal tubular disease (MacAllister et al. 1993).

Other experimental models of pain in horses require the permanent disfigurement or death of the experimental subjects. These models include dental dolorimetry (Brunson and Majors 1987; Brunson et al. 1987), caecal balloon catheters (Lowe 1978; Pippi and Lumb 1979; Pippi et al. 1979; Kalpravidh et al. 1984a,b; Muir and Robertson 1985) and induction of irreversible arthritis with either chemical or bacterial injection of the joint or cartilage damaged induced by surgical intervention (McIlwraith and Van Sickle 1981; Bertone et al. 1988; Frisbie et al. 2007; Kawcak et al. 2007). A model describing the use of heat generated either from topical light sources focused on the skin surface, or from deeply implanted electrodes, has been previously described (Pippi et al. 1979; Kalpravidh et al. 1984a,b) but its correlation to clinical musculoskeletal pain relief has been questioned (Foreman et al. 2008). As we have stated previously (Foreman et al. 2008), ‘clearly the shoeing model is more humane in that it does not require the “irreversible consumption” of large numbers of horses with surgical implantations or devices, or induction of irreversible arthritis, to arrive at objective assessments of musculoskeletal pain responsiveness'.

As to the effectiveness of the model in mimicking the NSAID response of horses with acute or chronic pain related to inflammation (of the joint, for instance), we can state that, although rare, we have inadvertently created foot abscesses in subjects using this model for studying analgesic efficacy (J.H. Foreman, unpublished data and personal observation). While it has been suggested that our model does not create inflammation, one must question the accuracy of that statement if we have created, however inadvertently, foot abscesses with weekly application of toe pressure for no more than 3–4 consecutive weeks. Instron testing data from cadaver specimens have shown that the mechanical heart bar shoe concavely deforms the palmar surface of the foot and distal phalanx, thus likely provoking an inflammatory response in the soft and bony tissues of the foot (C. Koblik, unpublished data). Even pressure alone results in inflammation. We make no claims here that our model mimics chronic mild joint inflammation that might receive similar treatments by practitioners but these data presented here clearly show that acute lameness in this model is attenuated for 12 h by the use of one or both of the NSAIDs studied here.

We conclude from these data that the combination of PBZ+FM was not more efficacious than PBZ or FM alone using this model of acute pain. These data do not support the hypothesis that the combination of FM and PBZ is more efficacious than a single NSAID when administered in this manner. If there is no greater analgesic efficacy with the combination and there is a demonstrated increase in toxicity with a combination of oral half-dose PBZ and i.v. single-dose FM (Reed et al. 2006), then it is logical to limit the use of both compounds simultaneously to prevent proven toxicity without demonstrable efficacy.

Authors' declaration of interests

No conflicts of interest have been declared.

Source of funding

Funded by the Grayson/Jockey Club Foundation, Keeneland Racing Association, Thoroughbred Owners and Breeders Association, Kentucky Thoroughbred Association, and the Maria Caleel Fund for Equine Sports Medicine Research at the University of Illinois.


The authors thank Drs. Jason Bundy and Keith Zientek for technical assistance.

Manufacturers' addresses

1 Butler Co, Columbus, Ohio, USA.

2 Schering-Plough Animal Health, Kenilworth, New Jersey, USA.