Effect of displaced versus non-displaced pelvic fractures onlong-term racing performance in 31 Thoroughbred racehorses

SE Hennessy,a* MA Muurlink,a,b GA Anderson,c TN Puksmanna and RC Whittona

Objective To evaluate the long-term racing prognosis forThoroughbred racehorses with displaced versus non-displacedfractures of the pelvis identified by scintigraphy.

Design Retrospective case analysis.

Methods Medical records of 31 Thoroughbred racehorses pre-senting to the University of Melbourne Equine Centre with fracturesof the pelvis that were identified by scintigraphy were reviewed.Pelvic fracture site was determined and defined as displaced ornon-displaced based on ultrasound and/or radiographic findings.Race records were analysed for each horse, with a minimum of24 months’ follow-up, and correlated with fracture type to deter-mine long-term prognosis for racing. Results are expressed asmedian and range.

Results Fractures at a single site were more common (n = 22)than fractures involving two sites (n = 9) and the ilial wing was themost commonly affected (n = 12). Thoroughbred racehorses withdisplaced pelvic fractures at any site (n = 12) raced fewer timeswithin 24 months of diagnosis than horses with non-displaced frac-tures (n = 19) (median 0.5, range 0–13 vs 7, 0–24; P = 0.037), butthere was no clear statistical difference in race earnings betweenthe two groups (median A$0, range A$0–$123,250 vs A$14,440,A$0–$325,500, respectively; P = 0.080). Four horses with displacedfractures (33%) were euthanased on humane grounds becauseof persistent severe pain. When these horses were excluded fromthe analysis, there were no differences in performance variablesbetween horses with a displaced or non-displaced pelvic fracture.

Conclusion Thoroughbred racehorses with a displaced or non-displaced pelvic fracture that survive the initial post-injury periodhave a good prognosis for racing.

Keywords fractures; pelvis; racehorses; Thoroughbreds

Abbreviations CI, confidence interval; HR, hazard ratioAust Vet J 2013;91:246–250 doi: 10.1111/avj.12053

Pelvic fractures that are the result of direct trauma occur inall types of horses, whereas those that are a fatigue injury areprimarily a problem of racehorses. Direct trauma can lead to

fractures anywhere in the pelvis,1 but the wing of the ilium is predis-posed to fatigue fractures, which often originate on its caudal aspectand extend over the sacroiliac joint craniodorsally or craniolaterally.2

Cyclic loading during locomotion results in bone fatigue at sites ofhigh load, which may progress from microcracking to incomplete,non-displaced fractures and finally to complete displaced fractures.3

With the widespread availability of scintigraphy and ultrasound,incomplete fatigue fractures of the ilium are being increasingly rec-ognised. In immature racing Thoroughbreds they are a common causeof hindlimb lameness and poor performance.4–6 The prognosis forreturn to function of horses with ilial fractures is regarded as good,provided there is no sacroiliac or coxofemoral joint involvement;however, detailed data on prognosis and factors affecting it arelimited.6–8

In this study, we assessed the differences in the return to function, timeto return, race earnings, and number of races and places betweenhorses with displaced or non-displaced pelvic fractures using a popu-lation of racing Thoroughbred horses presented to the University ofMelbourne Equine Centre. We hypothesised that horses with non-displaced fractures of the pelvis would return to work sooner and racemore successfully than horses with displaced pelvic fractures.

Materials and methods

We included 31 Thoroughbred racehorses that had been presentedto the University of Melbourne Equine Centre between 2 November2003 and 20 April 2010 and that had scintigraphic evidence of a pelvicfracture. A pelvic fracture was defined as a clear change in the normalpelvic contour on scintigraphic images or focal, moderate to intenseradiopharmaceutical uptake within the pelvis that was separate orextended from the normal areas of high uptake at the tuber sacrale,tuber coxae or tuber ischii.

All horses were in full-time race training when the injury occurred.Medical records, including nuclear scintigraphy, ultrasonography andradiology reports, were reviewed.

Physical and lameness examinations (including grading on the lame-ness scale of the American Association of Equine Practitioners) wereperformed.9 Where there was scintigraphic evidence of a fracture,ultrasonography (Siemens Acuson Sequioa 512, Siemens Plc USA) ofthe pelvis was performed percutaneously using a 3.5-MHz curvedarray transducer to determine if the fracture was displaced. Radio-graphy was only used to determine coxofemoral joint involvement(standing and recumbent ventrodorsal views, 150 kV, 250–300 mA).Fractures were classed as displaced or non-displaced based on ultra-sound and/or radiographic findings. Treatment protocols and restperiods were clinician-dependent, but in all cases the horse was restedfor at least 3 months.

*Corresponding author: University of Liverpool, Philip Leverhulme Equine Hospital,Leahurst, Chester High Road, Cheshire, CH64 7TE, UK; simonhennessy@hotmail.comaUniversity of Melbourne, Equine Centre, Werribee, Victoria, AustraliabUniversity of Queensland, Equine Hospital, Gatton Campus, Gatton, Queensland,AustraliacUniversity of Melbourne, Faculty of Veterinary Science, Werribee, Victoria, Australia

bs_bs_banner

EQUINE

EQU

INE

© 2013 The AuthorsAustralian Veterinary Journal © 2013 Australian Veterinary AssociationAustralian Veterinary Journal Volume 91, No 6, June 2013246

A standard scintigraphic protocol was used for each horse. An18-gauge catheter was placed in the left jugular vein and 12.7–15MBq/kg of 99Tc-HDP administered (standard 7 GBq/14 MBq/kg dosefor an average Thoroughbred racehorse). Frusemide (0.7 mg/kg IV)was injected 60 min prior to scanning the pelvis. Dynamic (35 ¥ 2-sframes), delayed-phase standing lateral, dorsal, caudal and obliquepelvis images were acquired 3 h after radiopharmaceutical injectionwith a large field-of-view gamma camera (Phillips Argus Epic) usinga 128 ¥ 128 matrix. Data were transferred to a workstation (SunUltra Workstation) and processed with motion correction software(University College London, Institute of Nuclear Medicine).

Race and stud book records were accessed in April 2012. Only horseswith at least 24 months’ follow-up were included. Data obtainedincluded sex, age, prize money, races and places achieved before andafter diagnosis.

Statistical analysisFisher’s exact test was used to compare proportions. Preliminaryanalysis of the residuals of the post-diagnosis race performance vari-ables showed that almost all were not normally distributed (Shapiro-Wilk’s test). The Mann-Whitney U-test with exact P values was usedto compare the mean ranks of post-diagnosis race performance vari-ables between horses with displaced or non-displaced fractures. Thelog-rank test and the hazard ratio (HR) from a Cox proportionalhazard model were used to compare the distribution of time to firstrace post-diagnosis for displaced and non-displaced fractures,together with the Kaplan-Meier estimate of median time to first racefor all horses.10 Two-tailed P < 0.05 was considered to be statisticallysignificant. The statistical software used was Stata 12.1 for Windows(StataCorp, College Station, TX, USA) and IBM SPSS 20 (IBM Corp.,New York, NY, USA) was used for Mann-Whitney exact P-values.

Results

All 31 horses included in the study (mean age 3.3 years, range1–6 years; 14 females (45%), 13 geldings (42%), 4 entire males (13%))

had been referred for scintigraphic examination. Of them, two horsespresented with a history of trauma and both had displaced ischialfractures, six horses presented with a history of poor performance andall had non-displaced fractures of the ilial wing, eight horses presentedfor lameness immediately following a race or a training gallop, sevenhorses were referred for mild to moderate hindlimb lameness thatcould not be localised with distal limb diagnostic analgesia by theattending veterinarian and eight horses presented with severe lame-ness (grade 3/5 or more) without associated trauma, of which four haddisplaced fractures.

All horses had scintigraphic examination of their hindlimbs,pelvis andthoracolumbar spine. Areas of increased radiopharmaceutical uptake,not associated with the pelvis, included the dorsal spinous processesof the thoracic and lumbar vertebrae, the tarsometatarsal joints, thetibial diaphysis, the lateral proximal hind sesamoid bone, the fourthmetatarsal bone and the lateral condyle of the third metatarsal bone.

There were 22 fractures involving a single site within the pelvis and9 fractures involving two sites. The most common site of fracture wasthe ilial wing, followed by the ischium (Table 1); 19 fractures werenon-displaced and 12 were displaced (Figures 1, 2). Of those withdisplaced fractures, four horses (33%) were euthanased at 2, 2, 3 and18 days after diagnosis on humane grounds because of persistent,severe pain. Of the four horses, two were presented 6–8 weeks after theinitial injury, one with fractures of the acetabulum and the ilial wing,and the other with fractures of the acetabulum and ilial shaft. Bothof them had severe gluteal muscle atrophy and grade 4/5 lameness thatwas noted to worsen with box rest. The third horse had bilateral ilialwing fractures with cranial displacement of the pelvis and associatedmuscle atrophy and severe lameness. The fourth horse presentedacutely after fast work with a severe comminuted oblique fracture ofthe ilial wing and shaft with grade 5/5 lameness. Among the 27 survi-vors, 12 horses were rested for 3 months, 12 for 6 months and 3 wererested for unspecified periods longer than 3 months.

With respect to racing, five horses that were in full race training at thetime of injury did not race before or after diagnosis, four horses raced

Table 1. Pelvic fracture distribution and outcome in 31 Thoroughbred racehorses

Fracture site n Fracture type Median no. of races No. euthanased

Displaced Non-displaced Pre-diagnosis Post-diagnosis

Single site

Ilial wing 12 2 10 7.5 13.0 0

Ilial shaft 2 0 2 0.5 5.0 0

Acetabulum 1 0 1 0.0 0.0 0

Ischium 7 5 2 0.0 18.0 0

Multiple sites

Bilateral ilial wings 3 1 2 7.0 2.0 1

Ilial wing and shaft 3 1 2 1.0 2.0 1

Acetabulum and ischium 1 1 0 0.0 0.0 0

Acetabulum and ilial wing 1 1 0 0.0 0.0 1

Acetabulum and ilial shaft 1 1 0 8.0 0.0 1

EQUINE

EQU

INE

© 2013 The AuthorsAustralian Veterinary Journal © 2013 Australian Veterinary Association Australian Veterinary Journal Volume 91, No 6, June 2013 247

before diagnosis but not after, 9 horses raced after diagnosis but notbefore, and 13 horses raced before and after diagnosis. The mediantime to first race based on all horses was 9.9 months (95% confidenceinterval (CI) 7.7–12.3).

The majority of horses with non-displaced pelvic fractures (16/19)raced within 24 months of diagnosis compared with only half of thehorses with displaced pelvic fractures (6/12) (Table 2). Horses with adisplaced pelvic fracture at any site (n = 12) raced fewer times within24 months of diagnosis than horses with non-displaced fractures(n = 19) (median 0.5, range 0–13 vs 7, 0–24; P = 0.037), but there wasno clear difference in race earnings between the two groups (medianA$0, range A$0–$123,250 vs A$14,440, A$0–$325,500, respectively;P = 0.080) (Table 2). When the four horses that were euthanased wereexcluded from the analysis, there were no differences in the perfor-mance variables between horses with a displaced or non-displaced

Figure 1. Left oblique (top) and dorsal scintigraphy images of the pelvisof a horse with an incomplete ilial stress fracture. Note the increasedradiopharmaceutical uptake lateral to the left tuber sacrale on the caudalsurface of the ilium (black arrows). Ultrasonographic examinationrevealed no displacement of fracture fragments.

Figure 2. Right oblique (top) and dorsal (middle) scintigraphy imagesof the pelvis of a horse with a complete fracture of the right ilium (blackarrows). Ultrasound examination (bottom) revealed marked displace-ment of fracture fragments.

EQUINE

EQU

INE

© 2013 The AuthorsAustralian Veterinary Journal © 2013 Australian Veterinary AssociationAustralian Veterinary Journal Volume 91, No 6, June 2013248

fracture (Table 3). The median time to first race for horses withdisplaced pelvic fractures was 10.4 months (95% CI 6.8–16.7) and9.3 months (95% CI 6.9–11.5) for horses with a non-displaced pelvicfracture (log-rank test P = 0.50). The HR did not differ from 1 for timeto first race for horses with a displaced compared with a non-displacedfracture (HR 0.73, 95% CI 0.28–1.86; P = 0.51).

Of the 4 horses presenting with fractures involving the acetabulum,none raced post diagnosis (2 were euthanased), whereas 6 of 7 horseswith ischial fractures raced post diagnosis and 11 of 12 horses withunilateral ilial wing fractures raced post diagnosis, including both ofthe horses with displaced fractures. One of two horses that presentedwith unilateral fractures of the ilial shaft raced post diagnosis. Of the6 horses with fractures in multiple sites and no acetabular involve-ment, 4 raced post diagnosis (2 were euthanased).

Discussion

Our study results demonstrated that Thoroughbred racehorses withdisplaced or non-displaced pelvic fractures that survive the initialpost-injury period have a good prognosis for racing. We found noevidence that horses with non-displaced pelvic fractures returned toracing sooner than those with displaced fractures. Pelvic fractures inThoroughbred racehorses have a variety of presentations and shouldbe considered in the differential diagnosis of poor performance andhindlimb lameness.

Our findings are consistent with those of other studies of pelvicfractures, which have demonstrated that 61–75% of horses withoutacetabular and ilial shaft involvement return to racing after diagno-

sis.8,11,12 Less success has been reported for Thoroughbred racehorseswith fractures of the ilial wing identified by scintigraphy, with 6 of 10,7

and 1 of 413 returning to racing in two separate studies. Studies thathave previously reported a good prognosis for displaced traumaticpelvic fracture relate to successful salvage of animals for potentialbreeding.11,14 A comparison of displaced and non-displaced pelvicfractures, and the prognosis for return to racing, has not been previ-ously examined. The importance of determining the site of pelvicfracture is highlighted in this study. Similar to other studies, fracturesinvolving the acetabulum had a poor prognosis for return to racing,whereas both ilial and ischial fractures had a good prognosis.8

A number of factors have the potential to affect future race perform-ance in horses with displaced fractures. In the case of ilial fractures,these include changes in the sacroiliac joint and subsequent sacroiliacdisease. Many displaced fractures are associated with the developmentof sacroiliac osteophytes and degenerative changes in the ilial articu-lar surface.2 Following a displaced fracture, the fracture fragmentsre-attach in a new anatomical arrangement that can influence thetime required for return to function as the horse adapts to the newanatomy.15 Also, the greater pain resulting from a displaced fracturewill often result in more muscle atrophy, which can delay the return toracing.1,6 The effect of the degree of muscle atrophy on the time toreturn to racing was not examined in this study. Based on our find-ings, these factors are less important and uncontrollable pain oracetabular involvement are the primary factors affecting long-termoutcome.

Scintigraphy is commonly used to identify pelvic fractures, but itis rarely possible to determine if a fracture is displaced using this

Table 2. Post-diagnosis performance for displaced and non-displaced fractures of 31 Thoroughbred racehorses

Outcome Fracture type P value

Non-displaced(n = 19)

Displaced(n = 12)

Raced in 24 months post-diagnosis 16 (84%) 6 (50%) 0.056

Median no. races in 24 months post-diagnosis (mean; range) 7 (7.5; 0–24) 0.5 (3; 0–13) 0.037

Median no. places in 24 months post-diagnosis (mean; range) 2 (2.9; 0–8) 0 (1.6; 0–8) 0.13

Median prize money in 24 months post-diagnosis (A$: mean; range) 14,440 (50,783; 0–325,500) 0 (16,129; 0–123,250) 0.080

Table 3. Post-diagnosis performance for displaced and non-displaced fractures of Thoroughbred racehorses after the exclusion of four horseseuthanased on humane grounds

Outcome Fracture type P value

Non-displaced(n = 19)

Displaced(n = 8)

Raced in 24 months post-diagnosis 16 (84%) 6 (75%) 0.62

Median no. races in 24 months post-diagnosis (mean; range) 7 (7.5; 0–24) 3.5 (4.5; 0–13) 0.34

Median no. places in 24 months post-diagnosis (mean; range) 2 (2.9; 0–8) 0.5 (2.4; 0–8) 0.61

Median prize money in 24 months post-diagnosis (A$: mean; range) 14,440 (50,783; 0–325,500) 575 (24,194; 0–123,250) 0.41

EQUINE

EQU

INE

© 2013 The AuthorsAustralian Veterinary Journal © 2013 Australian Veterinary Association Australian Veterinary Journal Volume 91, No 6, June 2013 249

technique.16 Fracture displacement or callus formation can beobserved using ultrasound, in the majority of cases, whereas non-displaced fractures are difficult to identify.17–19 Therefore, the combi-nation of scintigraphy and ultrasound enabled diagnosis in mostcases of pelvic fracture in this study. Standing radiography of thepelvis was used in this study to determine coxofemoral joint involve-ment in some cases, because of the difficulty of assessing acetabularinvolvement using ultrasound alone.17 The overlying sacrum andintestinal contents do not allow a detailed view of the whole ilium onradiographs.14,17,20,21

In the present study, it was often not possible to discriminatebetween traumatic fractures and displaced fatigue fractures becausein only a minority of cases was there a history of trauma or an acuteonset of lameness following fast work. However, it is likely that manyof the cases were fatigue fractures, as it has been reported thatilial fractures in a British Thoroughbred flat racehorse populationaccounted for 15.5% of the total fractures seen in the population,with fatigue fractures accounting for 87% of the ilial fractures.22 Inthe present study, 50% (3/6) of displaced ilial fractures occurredduring training of young horses that had never raced. Similarly, it hasbeen reported that in humans the majority of tibial and femoralstress fractures occur early during military training, suggesting thatbone adaptation is protective.23 As found in our study, ilial fracturesare most commonly observed in young, immature Thoroughbredracehorses aged up to 4 years of age.6,12,15,22 Others have described ilialwing fractures in older horses;1,2,5 however, two of these were post-mortem studies from which it was difficult to determine the time offracture.2,5

Study limitationsWe had a low number of cases, resulting in a lack of power, and thestudy population included four horses that were euthanased afterdiagnosis, making it impossible to know whether these horses couldhave returned to racing if the problem had been appropriatelymanaged. However, all four horses were euthanased because of per-sistent, severe pain rather than for economic reasons. Also, not allhorses in this study had full-body scintigraphy, so although it is pos-sible that other problems affecting long-term outcome may have beenoverlooked, the majority were able to perform well.

In conclusion, the poorer prognosis for Thoroughbred racehorseswith displaced pelvic fractures compared with non-displaced pelvicfractures appears to be because of an inability to manage severe painin some affected horses and/or involvement of the acetabulum.Accurate localisation of the fracture with a combination of scinti-graphy and ultrasound will allow an appropriate prognosis to begiven.

References

1. Dyson SJ. Lameness associated with the stifle and pelvic regions. Am AssocEquine Pract 2002;48:387–411.2. Haussler KK, Stover SM. Stress fractures of the vertebral lamina and pelvis inThoroughbred racehorses. Equine Vet J 1998;30:374–381.3. Riggs CM. Fractures: a preventable hazard of racing Thoroughbreds? Vet J2002;163:19–29.4. Bathe AP. 245 fractures in Thoroughbred racehorses: results of a 2-year pro-spective study in Newmarket. Am Assoc Equine Pract 1994;40:175–176.5. Carrier TK, Estberg L, Stover SM et al. Association between long periodswithout high–speed workouts and risk of complete humeral or pelvic fracture inThoroughbred racehorses: 54 cases (1991–1994). J Am Vet Med Assoc 1998;212:1582–1587.6. Van Wessum R. The pelvis. In: Baxter GM, editor. Adam’s and Stashak’s lamenessin horses. 6th edn. Wiley Blackwell, Oxford, 2011:840–847.7. Pilsworth RC, Shepherd MC, Herinckx BMB et al. Fractures of the wing of theilium, adjacent to the sacroiliac joint in Thoroughbred racehorses. Equine Vet J1994;26:94–99.8. Fuller A, Beever E, Fraser B et al. A retrospective study of 80 cases of pelvicfracture diagnosed by gamma scintigraphy in Thoroughbred racehorses. BrEquine Vet Assoc 2008;47:142–143.9. Ross MW. Movement. In: Ross MW, Dyson SJ. Diagnosis and management oflameness in the horse. 1st edn. Saunders, St Louis, 2002:60–73.10. Hosmer D, Lemeshow S. Applied survival analysis: regression modeling of timeto event data. 2nd edn. John Wiley & Sons, New York, 2008.11. Rutkowski JA, Richardson DW. A retrospective study of 100 pelvic fractures inhorses. Equine Vet J 1989;29:256–259.12. Shepherd MC, Pilsworth RC, Hopes R et al. Clinical signs, diagnosis, manage-ment and outcome of complete and incomplete fracture to the ilium. Am AssocEquine Pract 1994;40:177–180.13. Davenport-Goodall CLM, Ross MW. Scintigraphic abnormalities of the pelvicregion in horses examined because of lameness or poor performance: 128 cases(1993–2000). J Am Vet Med Assoc 2004;224:88–95.14. Little C, Hilbert B. Pelvic fractures in horses: 19 cases (1974–1984). J Am VetMed Assoc 1987;190:1203–1206.15. Pilsworth RC. Diagnosis and management of pelvic fractures in the Thorough-bred racehorse. In: Ross MW, Dyson SJ, editors. Diagnosis and management oflameness in the horse. 2nd edn. Elsevier Saunders, St Louis, 2011:564–571.16. Hornof WJ, Stover SM, Koblik PD et al. Oblique views of the ilium and thescintigraphic appearance of stress fractures of the ilium. Equine Vet J 1996;28:355–358.17. Geburek F, Rotting AK, Stadler PM. Comparison of the diagnostic valueof ultrasonography and standing radiography for pelvic-femoral disorders inhorses. Vet Surg 2009;38:310–317.18. Driver AJ, Nagy A. Fracture of the ischium in an eight-year-old Arabiangelding: a diagnostic challenge. Equine Vet Educ 2008;20:127–130.19. Tomlinson JE, Sage AM, Turner TA et al. Detailed ultrasonographic mapping ofthe pelvis in clinically normal horses and ponies. Am J Vet Res 2001;62:1768–1775.20. May SA, Patterson LJ, Peacock PJ et al. Radiographic technique for the pelvis inthe standing horse. Equine Vet J 1991;23:312–314.21. Barrett EL, Talbot AM, Driver AJ et al. A technique for radiography in thestanding horse. Equine Vet J 2006;38:266–270.22. Verheyen KLP, Wood JLN. Descriptive epidemiology of fractures occurring inBritish Thoroughbred racehorses in training. Equine Vet J 2004;36:167–173.23. Finestone A, Milgrom C, Wolf O et al. Epidemiology of metatarsal stress frac-tures versus tibial and femoral stress fractures during elite training. Foot Ankle Int2011;32:16–20.

(Accepted for publication 24 September 2012)

EQUINE

EQU

INE

© 2013 The AuthorsAustralian Veterinary Journal © 2013 Australian Veterinary AssociationAustralian Veterinary Journal Volume 91, No 6, June 2013250