Many yearling Thoroughbreds are purchased at public auctions from July through September for the purpose of racing them the following year as 2-year-old horses. The price of a yearling Thoroughbred racehorse prospect can vary depending on their pedigree, conformation, and disclosed medical information, such as history of arthroscopic surgery and presale radiographic findings.1–3 Because abnormalities detected via presale radiography may affect the ability of a yearling to race as a 2-year-old or limit racing performance, veterinarians are typically hired by prospective horse buyers to evaluate radiographs obtained before a sale.
The association between presale radiographic findings for yearling Thoroughbreds and future racing performance has been investigated in 2 studies1,2 conducted in the United States; however, conclusions in the reports of those studies were different. In one of those studies,1 1,162 yearling Thoroughbreds offered for sale at 1 of 2 sales in Kentucky between 1993 and 1996 were included; results of that study indicated yearlings with moderate or severe supracondylar lysis of the palmarodistal aspects of third metacarpal bones, abnormalities in dorsomedial aspects of middle carpal joints, enthesophytes on forelimb proximal sesamoid bones, or osteochondral fragments of hind limb P1s were less likely to start a race as 2- or 3-year-old horses, compared with yearlings without these radiographic findings. However, results of that other study,2 which included 348 yearling Thoroughbreds offered for sale in Texas between 2002 and 2003, indicated none of the radiographic findings evaluated were associated with future racing performance. In both of those studies,1,2 horses were not selected by use of a randomization procedure and radiographs of yearlings were obtained from private veterinary practices; characteristics of horses were not similar between those 2 studies. For example, the median sales prices of yearling Thoroughbreds in the studies conducted in Kentucky1 and Texas2 were $40,000 and $9,000, respectively. In addition, the numbers of horses with osteochondral fragments of hind limb P1s in the studies conducted in Kentucky1 and Texas2 were 25 and 10, respectively; this was a limitation that may have caused false-negative results in the study2 conducted in Texas. Because results of those studies were different, the associations between presale radiographic findings for yearling Thoroughbreds in the United States and future racing performance are not well established. The objective of the study reported here was to determine the relationships between various presale radiographic abnormalities and future racing performance via evaluation of data for Thoroughbreds selected with a randomization procedure that were offered for sale at a September 2006 yearling sale at a Thoroughbred sales facility in Lexington, Ky.
Materials and Methods
Animals—Of the yearling Thoroughbreds (n = 5,161) offered for sale at the September 2006 yearling sale at a Thoroughbred sales facility in Lexington, Ky,a 397 horses were selected for inclusion in the study. These 397 horses were also included in another study3 in which the prevalence of various presale radiographic findings and association between those findings and sales prices of horses were determined. The horses were selected via a randomization procedure with software.b Because 33 of these horses had osteochondral fragments of hind limb P1s and 362 did not have this finding, the present study had 80% power to detect significant (P < 0.05) differences in data of 50% (horses without P1 fragments that failed to start a race) and 75% (horses with P1 fragments that failed to start a race).c
Study design—This study was designed as a retrospective observational cohort study. Horses were grouped on the basis of detection (yes vs no) of various presale radiographic findings for forelimbs and hind limbs. The outcome variables for racing performance included having started a race (yes vs no), having placed (ie, finished a race in first, second, or third place) at least once in a race (yes vs no), total amount of money earned (high vs low), and amount of money earned per start (high vs low); effects of various presale radiographic findings on these variables were determined. Potential confounding factors (sex, history of presale arthroscopic surgery, purchase price at the yearling sale, and day of the sale on which a horse was sold [ie, sale day]) that may have affected racing performance were included in the analysis. Only data for the 2-year-old racing year were evaluated.
Radiographic findings—For each horse, repository digital radiographs were available, which had been obtained before the sale. Radiographs were evaluated by an equine veterinarian with 26 years of experience with prepurchase examination of yearling horses. Procedures for evaluation of radiographs and categorization of radiographic findings were the same as those used in another study.3 For each horse, 32 radiographic views were available. These views were the standard radiographic views obtained for yearlings at the salea and had been selected by a panel of veterinarians on the basis of guidelines provided by the American Association of Equine Practitioners. Forelimb radiographic findings evaluated in the present study included detection of osteochondral P1 fragments, proximal sesamoid bone osteophytes or enthesophytes, and middle carpal joint osteophytes or enthesophytes. Hind limb radiographic findings evaluated included detection of osteochondral P1 fragments and proximal sesamoid bone osteophytes or enthesophytes. In addition, osteochondritis dissecans lesions of the lateral and medial trochlear ridges of femurs and subchondral bone cysts of medial femoral condyles of femurs were included as 1 variable (OCD of stifle joints). These 6 radiographic findings were selected for inclusion in the study because they affect yearling sales price3 and are predictors of future racing performance1,2 of horses.
Racing performance—Racing performance data were collected by use of an electronic databased and included the following information: yearling sale identification number, whether a horse started a race, whether a horse placed in a race, total amount of money earned (in US dollars), and amount of money earned per start during the 2-year-old racing year. Amount of money earned per start was calculated as the total amount of money earned during the 2-year-old racing year divided by the total number of races started during that year.
Statistical analysis—Data for continuous variables (total amount of money earned, amount of money earned per start, and purchase price at the yearling sale) were categorized as high or low on the basis of the median value for the data; values less than or equal to the median value were categorized as low, and values greater than the median value were categorized as high. Data for sale day were allocated to 1 of 2 categories (early [days 1 through 4] vs late [days 6 through 14]) on the basis of the day of the sale on which horses were sold. Categories for sale day did not include day 5 because horses were not sold on that day. Horses with a history of arthroscopic surgery (yes vs no) were further categorized on the basis of the anatomic location of the surgery (eg, no arthroscopic surgery, arthroscopic surgery for MTP joints, or arthroscopic surgery for joints other than MTP joints). Data regarding anatomic location of surgery were classified in this manner because presale arthroscopic surgery in MTP joints is a common surgical procedure for yearling Thoroughbreds, and Thoroughbred industry professionals typically believe that this procedure does not have a detrimental effect on future racing performance of horses.
Median yearling sales prices for horses that started a race and those that did not were compared via the Wilcoxon rank sum test. Data for the variables sex (colt vs filly), history of arthroscopic surgery (yes vs no), purchase price at the yearling sale (high vs low), sale day (1 through 4 vs 6 through 14), and detection (yes vs no) of various radiographic findings for forelimbs and hind limbs were compared between horses that did or did not race as 2-year-olds and between horses that did or did not place in a race as 2-year-olds via a χ2 test.
For horses that started a race as 2-year-olds, median total amount of money earned and median amount of money earned per start were compared for the variables sex, history of arthroscopic surgery, purchase price at the yearling sale, sale day, and detection of various radiographic findings for forelimbs or hind limbs via the nonparametric Wilcoxon rank sum test. The relationships between radiographic findings and the outcome variables total amount of money earned and amount of money earned per start were not analyzed via multivariable ANOVA because these 2 outcome variables did not meet assumptions for normality (including when data were ranked or log 10, natural log, or square root transformations were performed). Therefore, these relationships were analyzed via logistic regression.
The odds of starting a race (yes vs no), placing in a race (yes vs no), having a high or low total amount of money earned, and having a high or low amount of money earned per start were determined via unconditional multivariable logistic regression models. Associations between detection of each type of radiographic finding and starting a race (yes vs no) and placing in a race (yes vs no) during the 2-year-old racing year were determined via univariate χ2 tests; variables with a value of P < 0.20 were tested for inclusion in the multivariable logistic regression models. To determine the best-fitting model, the variable with the smallest P value in the univariate χ2 test was the first variable included in the logistic regression model. Then, each of the other variables were included in the model to determine whether inclusion of each variable significantly improved the fit of the model. Variables were included in the model in descending order of the value of the likelihood ratio statistic (determined via a χ2 test with 1 degree of freedom); only variables with a value of P ≤ 0.05 were retained in the model. Following determination of the primary multivariable logistic regression model, the interaction term between the variables forelimb proximal sesamoid bone osteophytes or enthesophytes and hind limb osteochondral P1 fragments was tested for significance via the likelihood ratio statistic. Confounding for the variables sex, history of arthroscopic surgery, purchase price at the yearling sale, and sale day was evaluated via inclusion of each of these variables in the model and determination of the subsequent change in the adjusted OR of the remaining variables in the model; variables were retained in the model if the value of the adjusted OR changed by > 10% when that variable was added to the model. The multivariable logistic regression model was assessed for goodness of fit via the Hosmer-Lemeshow test, and the ability of the model to discriminate between horses that started a race versus those that did not was determined via the receiver operating characteristic curve.4 For the final model, adjusted ORs and 95% CIs were reported.
For linear regression analysis, osteochondral fragments on the proximodorsal and proximopalmar aspects of forelimb P1s were considered as 1 variable (forelimb osteochondral P1 fragments). Similarly, osteochondral fragments on the proximodorsal and proximoplantar aspects of hind limb P1s were considered as 1 variable (hind limb osteochondral P1 fragments). For all analyses, values of P ≤ 0.05 were considered significant.
Results
Number of horses that started a race—Of the 397 horses included in the study, 192 (48%) started at least 1 race during the 2-year-old racing year. Median yearling sales price of the 397 horses in the study was $35,000 (interquartile range, $10,000 to $85,000). Median yearling sales price was not significantly (P = 0.39) different between horses that started a race ($37,000) and those that did not ($30,000). During the 2-year-old racing year, 38 of the 60 (63%) horses with forelimb proximal sesamoid bone osteophytes or enthesophytes did not start a race and 167 of the 337 (50%) horses without those radiographic abnormalities did not start a race (Table 1); results of a univariate χ2 test indicated these proportions were significantly (P = 0.04) different. Twenty-two of the 33 (67%) horses with hind limb osteochondral P1 fragments did not start a race, and 183 of the 364 (50%) horses without those radiographic abnormalities did not start a race; results of a univariate χ2 test indicated these proportions were significantly (P = 0.07) different. Forty-six of the 78 (59%) horses with 2 radiographic abnormalities did not start a race as 2-year-olds, and 159 of the 319 (50%) horses with 1 radiographic abnormality did not start a race as 2-year-olds; results of a univariate χ2 test indicated these proportions were not significantly (P = 0.14) different. The most common combination of radiographic abnormalities for horses with ≥ 2 abnormalities was forelimb and hind limb proximal sesamoid bone osteophytes or enthesophytes (39/78 [50%]); the next most frequently detected combination of radiographic abnormalities was hind limb osteochondral P1 fragments and forelimb proximal sesamoid bone osteophytes or enthesophytes (16/78 [21%]).
Effects of various variables on whether 397 yearling Thoroughbreds sold at a Thoroughbred sales facility in Lexington, Ky, during September 2006 did (n = 192 horses) or did not (205) start at least 1 race during the 2-year-old racing year.
Variable | Category | Started a race (No. [%] of horses) | P value* | |
---|---|---|---|---|
Yes | No | |||
Sex | Colt | 101 (52) | 94 (48) | 0.17 |
Filly | 91 (45) | 111 (55) | ||
History of arthroscopic surgery† | No | 163 (47) | 182 (53) | 0.25 |
Yes | 29 (56) | 23 (44) | ||
History of arthroscopic surgery: MTP joint or other joint† | No | 163 (47) | 182 (53) | 0.17 |
Yes | 12 (71) | 5 (29) | ||
Other | 17 (49) | 18 (51) | ||
Sale day‡ | Early | 52 (54) | 45 (46) | 0.23 |
Late | 140 (47) | 160 (53) | ||
Purchase price at the yearling sale§ | High | 101 (51) | 98 (49) | 0.33 |
Low | 91 (46) | 107 (54) | ||
≥ 2 radiographic abnormalities‖ | No | 160 (50) | 159 (50) | 0.14 |
Yes# | 32 (41) | 46 (59) | ||
Forelimb radiographic abnormalities‖ | ||||
MCP joint P1 OC fragments (dorsal or palmar) | No | 187 (49) | 198 (51) | 0.63 |
Yes | 5 (42) | 7 (58) | ||
MCP joint P1 OC fragments (dorsal) | No | 188 (49) | 199 (51) | 0.59 |
Yes | 4 (40) | 6 (60) | ||
MCP joint P1 OC fragments (palmar) | No | 191 (48) | 204 (52) | 0.96 |
Yes | 1 (50) | 1 (50) | ||
Proximal sesamoid bone osteophytes or enthesophytes | No | 170 (50) | 167 (50) | 0.04 |
Yes | 22 (37) | 38 (63) | ||
Osteophytes or enthesophytes in middle carpal joints | No | 175 (48) | 191 (52) | 0.45 |
Yes | 17 (55) | 14 (45) | ||
Hind limb radiographic abnormalties‖ | ||||
MTP joint P1 OC fragments (dorsal or plantar) | No | 181 (50) | 183 (50) | 0.07 |
Yes | 11 (33) | 22 (67) | ||
MTP joint P1 OC fragments (dorsal) | No | 187 (49) | 194 (51) | 0.16 |
Yes | 5 (31) | 11 (69)** | ||
MTP joint P1 OC fragments (plantar) | No | 186 (49) | 193 (51) | 0.19 |
Yes | 6 (33) | 12 (67)** | ||
Proximal sesamoid bone osteophytes or enthesophytes | No | 120 (50) | 122 (50) | 0.54 |
Yes | 72 (46) | 83 (54) | ||
OCD of stifle joints¶ | No | 171 (47) | 190 (53) | 0.20 |
Yes | 21 (58) | 15 (42) |
P values were calculated via a χ2 test.
Analysis for history of arthroscopic surgery was performed with data categorized for horses that did or did not have arthroscopic surgery of any joint before the yearling sale (yes vs no) and with data categorized for horses that did or did not have arthroscopic surgery of MTP joints (yes, no, or other; horses in the other category had arthroscopic surgery of joints other than the MTP joint).
Data regarding day of the yearling sale on which horses were sold were categorized as early (horse sold on days 1 through 4 of the sale) or late (horse sold on days 6 through 14 of the sale). Categories for sale day did not include day 5 because horses were not sold on that day.
Data regarding purchase price for horses at the yearling sale were categorized as low or high on the basis of the median purchase price for the horses. For each horse, values less than or equal to the median value were categorized as low and values greater than the median value were categorized as high.
The only radiographic abnormalities evaluated included forelimb MCP joint osteochondral P1 fragments, forelimb proximal sesamoid bone osteophytes or enthesophytes, middle carpal joint osteophytes or enthesophytes, hind limb MTP joint osteochondral P1 fragments, hind limb proximal sesamoid bone osteophytes or enthesophytes, and OCD of stifle joints (including lateral or medial trochlear ridge OCD lesions or medial femoral condyle bone cysts).
Horses with OCD of stifle joints had lateral or medial trochlear ridge OCD lesions or medial femoral condyle bone cysts.
Of the 78 horses with multiple radiographic abnormalities, 39 (50%) had fore- and hind limb proximal sesamoid bone osteophytes or enthesophytes.
One of these horses had osteochondral fragments of both the proximodorsal and proximoplantar aspects of P1.
OC = Osteochondral.
After adjustment for detection of hind limb osteochondral P1 fragments in radiographs, results of logistic regression indicated the odds of failure to start a race as a 2-year-old for horses with forelimb proximal sesamoid bone osteophytes or enthesophytes were 1.78 times as great as the odds for horses without this radiographic abnormality (P = 0.04; Table 2). After adjustment for detection of forelimb proximal sesamoid bone osteophytes or enthesophytes in radiographs, the odds of failure to start a race as a 2-year-old for horses with hind limb osteochondral P1 fragments were 2.02 times as great as the odds for horses without this radiographic abnormality, although this finding was not significant (P = 0.06). However, most of the 95% CI (0.95 to 4.31) values for this finding were higher than the null value (1.0). Inclusion of the variables sex, history of arthroscopic surgery, purchase price at the yearling sale, or sale day in the model did not change the ORs for failure to start a race by > 10% for horses with forelimb proximal sesamoid bone osteophytes or enthesophytes or those with hind limb osteochondral P1 fragments; this finding indicated the associations between these 2 radiographic abnormalities and failure to start a race were not confounded by sex, history of arthroscopic surgery, purchase price at the yearling sale, or sale day. Inclusion in the model of the interaction term between forelimb proximal sesamoid bone osteophytes or enthesophytes and hind limb osteochondral P1 fragments did not produce a significant likelihood ratio statistic. Results of the Hosmer-Lemeshow goodness-of-fit test indicated that the model fit the data (χ2 < 0.001; 1 degree of freedom; P = 0.98). However, the model did not have a high ability to discriminate between horses that started a race versus those that did not (area under the receiver operating characteristic curve, approx 0.60).
Results of logistic regression analysis regarding associations between various radiographic abnormalities and failure to start at least 1 race during the 2-year-old racing year for 397 yearling Thoroughbreds sold at a Thoroughbred sales facility in Lexington, Ky, during September 2006.
Variable | Category | Adjusted OR (95% CI) | P value |
---|---|---|---|
Forelimb proximal sesamoid bone osteophytes or enthesophytes | No | Referent | NA |
Yes | 1.78 (1.01–3.16) | 0.04 | |
Hind limb MTP joint P1 OC fragments (dorsal or plantar) | No | Referent | NA |
Yes | 2.02 (0.95–4.31) | 0.06 |
The multivariable logistic regression model was assessed for goodness of fit via the Hosmer-Lemeshow test (χ2 < 0.001; 1 degree of freedom; P = 0.98). Results for each variable indicate results for that variable adjusted for effects of the other variable.
NA = not applicable.
Number of horses that placed in a race—During the 2-year-old racing year, 17 of the 21 (81%) horses with OCD of stifle joints placed (ie, finished in first, second, or third place) in a race and 105 of the 171 (61%) horses without OCD of stifle joints placed in a race (Table 3); results of a univariate χ2 test indicated these proportions were not significantly (P = 0.08) different. Results of logistic regression indicated sex, history of arthroscopic surgery, purchase price at the yearling sale, and sale day did not have an influence on the association between having OCD of stifle joints and whether a horse placed in a race it had started during the 2-year-old racing year.
Effects of various variables on whether 192 yearling Thoroughbreds sold at a Thoroughbred sales facility in Lexington, Ky, during September 2006 did (n = 122) or did not (70) place (ie, finish first, second, or third) in at least 1 race they started during the 2-year-old racing year.
Variable | Category | Placed in a race (No. [%] of horses) | P value* | |
---|---|---|---|---|
Yes | No | |||
Sex | Colt | 63 (62) | 38 (38) | 0.72 |
Filly | 59 (65) | 32 (35) | ||
History of arthroscopic surgery† | No | 102 (63) | 61 (37) | 0.51 |
Yes | 20 (69) | 9 (31) | ||
History of arthroscopic surgery: MTP joint or other joint† | No | 102 (63) | 61 (37) | 0.36 |
Yes | 14 (78) | 4 (22) | ||
Other | 6 (55) | 5 (45) | ||
Sale day‡ | Early | 29 (56) | 23 (44) | 0.17 |
Late | 93 (66) | 47 (34) | ||
Purchase price at the yearling sale§ | High | 64 (63) | 37 (37) | 0.95 |
Low | 58 (64) | 33 (36) | ||
≥ 2 radiographic abnormalities‖ | No | 100 (63) | 60 (37) | 0.50 |
Yes | 22 (69) | 10 (31) | ||
Forelimb radiographic abnormalities‖ | ||||
MCP joint P1 OC fragments (dorsal or palmar) | No | 120 (64) | 67 (36) | 0.26 |
Yes | 2 (40) | 3 (60) | ||
MCP joint P1 OC fragments (dorsal) | No | 120 (64) | 68 (36) | 0.32 |
Yes | 2 (50) | 2 (50) | ||
MCP joint P1 OC fragments (palmar) | No | 122 (64) | 69 (36) | 0.18 |
Yes | 0 (0) | 1 (100) | ||
Proximal sesamoid bone osteophytes or enthesophytes | No | 107 (63) | 63 (37) | 0.23 |
Yes | 15 (68) | 7 (32) | ||
Osteophytes or enthesophytes in middle carpal joints | No | 112 (64) | 63 (36) | 0.67 |
Yes | 10 (59) | 7 (41) | ||
Hind limb radiographic abnormalities‖ | ||||
MTP joint P1 OC fragments (dorsal or plantar) | No | 113 (62) | 68 (38) | 0.19 |
Yes | 9 (82) | 2 (18) | ||
MTP joint P1 OC fragments (dorsal) | No | 118 (63) | 69 (37) | 0.43 |
Yes | 4 (80) | 1 (20) | ||
MTP joint P1 OC fragments (plantar) | No | 117 (63) | 69 (37) | 0.30 |
Yes | 5 (83) | 1 (17) | ||
Proximal sesamoid bone osteophytes or enthesophytes | No | 75 (63) | 45 (37) | 0.69 |
Yes | 47 (65) | 25 (35) | ||
OCD of stifle joints¶ | No | 105 (61) | 66 (39) | 0.08 |
Yes | 17 (81) | 4 (19) | ||
See Table 1 for key. |
Total amount of money earned—Results of the Wilcoxon rank sum test indicated the median total amount of money earned during the 2-year-old racing year was lower for horses with any type of forelimb osteochondral P1 fragments ($114) than it was for horses without such radiographic abnormalities ($4,540), but this difference was not significant (P = 0.19; Table 4). Results of logistic regression indicated sex, history of arthroscopic surgery, purchase price at the yearling sale, and sale day did not have an influence on the association between detection of forelimb osteochondral P1 fragments and total amount (high vs low) of money earned during the 2-year-old racing year.
Effects of various variables on median (interquartile range) total amount of money earned during the 2-year-old racing year for 192 yearling Thoroughbreds sold at a Thoroughbred sales facility in Lexington, Ky, during September 2006.
Variable | Category | No. of horses | Total amount of money earned ($) | P value* |
---|---|---|---|---|
Sex | Colt | 101 | 3,094 (416–13,613) | 0.16 |
Filly | 91 | 5,165 (1,194–14,673) | ||
History of arthroscopic surgery† | No | 163 | 4,080 (520–13,700) | 0.89 |
Yes | 29 | 5,200 (1,377–15,419) | ||
History of arthroscopic surgery: MTP joint or other joint† | No | 163 | 4,080 (520–13,700) | 0.57 |
Yes | 18 | 7,418 (1,669–18,325) | ||
Other | 11 | 2,174 (110–7,620) | ||
Sale day‡ | Early | 52 | 2,011 (289–12,257) | 0.22 |
Late | 140 | 4,690 (856–14,203) | ||
Purchase price at the yearling sale§ | High | 101 | 5,565 (406–19,335) | 0.34 |
Low | 91 | 3,280 (840–10,805) | ||
≥ 2 radiographic abnormalities‖ | No | 160 | 4,464 (476–13,423) | 0.48 |
Yes | 32 | 4,863 (691–20,175) | ||
Forelimb radiographic abnormalities‖ | ||||
MCP joint P1 OC fragments (dorsal or palmar) | No | 187 | 4,540 (620–13,853) | 0.19 |
Yes | 5 | 114 (100–26,972) | ||
MCP joint P1 OC fragments (dorsal) | No | 188 | 4,532 (579–13,815) | 0.45 |
Yes | 4 | 1,687 (95–38,826) | ||
MCP joint P1 OC fragments (palmar) | No | 191 | 4,523 (565–13,853) | ND |
Yes | 1 | 114 (NA) | ||
Proximal sesamoid bone osteophytes or enthesophytes | No | 170 | 3,968 (543–13,389) | 0.25 |
Yes | 22 | 8,675 (595–26,262) | ||
Osteophytes or enthesophytes in middle carpal joints | No | 175 | 4,600 (565–14,635) | 0.39 |
Yes | 17 | 3,940 (450–8,679) | ||
Hind limb radiographic abnormalities | ||||
MTP joint P1 OC fragments (dorsal or plantar) | No | 181 | 4,405 (538–13,406) | 0.30 |
Yes | 11 | 10,270 (905–34,800) | ||
MTP Joint P1 OC fragments (dorsal) | No | 187 | 4,523 (557–13,700) | 0.73 |
Yes | 5 | 3,940 (1,632–34,774) | ||
MTP joint P1 OC fragments (plantar) | No | 186 | 4,242 (548–13,505) | 0.29 |
Yes | 6 | 12,472 (734–44,325) | ||
Proximal sesamoid bone osteophytes or enthesophytes | No | 120 | 4,226 (422–13,179) | 0.31 |
Yes | 72 | 5,190 (722–20,175) | ||
OCD of stifle joints¶ | No | 171 | 3,995 (450–13,853) | 0.32 |
Yes | 21 | 8,365 (1,927–13,418) |
Money earned per start—Results of the Wilcoxon rank sum test indicated the median amount of money earned per start during the 2-year-old racing year was lower for horses with any type of forelimb osteochondral P1 fragments ($114) than it was for horses without this radiographic abnormality ($1,560), but this difference was not significant (P = 0.13; Table 5). Results of logistic regression indicated sex, history of arthroscopic surgery, purchase price at the yearling sale, and sale day did not have an influence on the association between detection of osteochondral P1 fragments and amount (high vs low) of money earned per start.
Effects of various variables on median (interquartile range) amount of money earned per start during the 2-year-old racing year for 192 yearling Thoroughbreds sold at a Thoroughbred sales facility in Lexington, Ky, during September 2006.
Variable | Category | No. of horses | Amount of money earned per start ($) | P value* |
---|---|---|---|---|
Sex | Colt | 101 | 1,215 (250–3,603) | 0.08 |
Filly | 91 | 1,850 (409–4,220) | ||
History of arthroscopic surgery† | No | 163 | 1,508 (349–4,025) | 0.92 |
Yes | 29 | 1,680 (370–4,076) | ||
History of arthroscopic surgery: MTP joint or other joint† | No | 163 | 1,508 (349–4,025) | 0.56 |
Yes | 18 | 2,359 (391–4,253) | ||
Other | 11 | 1,524 (110–2,174) | ||
Sale day‡ | Early | 52 | 1,627 (250–5,524) | 0.97 |
Late | 140 | 1,510 (398–3,510) | ||
Purchase price at the yearling sale§ | High | 101 | 1,848 (312–5,485) | 0.11 |
Low | 91 | 1,220 (398–2,831) | ||
> 2 radiographic abnormalities‖ | No | 160 | 1,565 (355–4,017) | 0.95 |
Yes | 32 | 1,391 (287–4,317) | ||
Forelimb radiographic abnormalities‖ | ||||
MCP joint P1 OC fragments (dorsal or palmar) | No | 187 | 1,560 (397–4,067) | 0.13 |
Yes | 5 | 114 (100–3,824) | ||
MCP joint P1 OC fragments (dorsal) | No | 188 | 1,560 (382–4,056) | 0.30 |
Yes | 4 | 259 (95–5,532) | ||
MCP joint P1 OC fragments (palmar) | No | 191 | 1,560 (377–4,067) | ND |
Yes | 1 | 114 (NA) | ||
Proximal sesamoid bone osteophytes or enthesophytes | No | 170 | 1,511 (348–4,035) | 0.42 |
Yes | 22 | 2,044 (442–4,895) | ||
Osteophytes or enthesophytes in middle carpal joints | No | 175 | 1,575 (349–4,133) | 0.34 |
Yes | 17 | 1,364 (325–1,890) | ||
Hind limb radiographic abnormalities‖ | ||||
MTP joint P1 OC fragments (dorsal or plantar) | No | 181 | 1,524 (362–3,867) | 0.47 |
Yes | 11 | 2,567 (226–14,580) | ||
MTP joint P1 OC fragments (dorsal) | No | 187 | 1,560 (348–4,025) | 0.94 |
Yes | 5 | 788 (204–10,866) | ||
MTP joint P1 OC fragments (plantar) | No | 186 | 1,519 (370–4,001) | 0.31 |
Yes | 6 | 4,592 (197–15,285) | ||
Proximal sesamoid bone osteophytes or enthesophytes | No | 120 | 1,488 (342–3,473) | 0.30 |
Yes | 72 | 1,812 (397–4,318) | ||
OCD of stifle joints¶ | No | 171 | 1,468 (300–4,085) | 0.27 |
Yes | 21 | 2,091 (957–3,880) |
Discussion
Results of the present study provided new information regarding the associations between yearling presale radiographic findings and racing performance during the 2-year-old racing year for Thoroughbreds. In contrast to other studies1,2 in which data regarding 2- and 3-year-old racing performance of Thoroughbreds were combined, only data regarding the 2-year-old year were evaluated in the present study. Findings of this study indicated detection of forelimb proximal sesamoid bone osteophytes or enthesophytes or hind limb osteochondral P1 fragments was associated with failure to start a race during the 2-year-old racing year for Thoroughbreds.
During their 2-year-old racing year, 192 of the 397 (48%) horses included in this study started at least 1 race. This proportion was higher than the proportion (11,197/35,045 [32%]) of Thoroughbreds born in the United States in 2005 that started a race,c which was the same year in which horses in the present study had been born. Yearlings offered for sale at the Thoroughbred sales facilitya at which horses in the present study had been sold have a higher median sales price versus yearlings sold at other facilities.1–3 Other authors1 have proposed that horses with a high sales price likely enter higher quality training stables and have better racing opportunities than horses with a low sales price. Those authors1 found that 946 (81%) of the horses included in that study raced during the 2- or 3-year-old racing years. Median sales price of the horses included in that study was significantly higher than that of all horses offered for sale at 1 of 2 sales facilities in Kentucky between 1993 and 1996.1 Those authors1 suggested that pedigree, conformation, and other unidentified factors related to future racing performance may have contributed to the high proportion of horses in the study that started a race. Comparison of results of the present study with those of that other study1 was difficult because the proportion of horses that started a race during the 2-year-old racing year was not reported in that other study.
Results of the present study did not support the conclusion of other authors1,2 that horses with a high yearling sales price have better opportunities to race than horses with a low sales price. Median yearling sales price was not significantly different between horses that started a race ($37,000) and those that did not start a race ($30,000) in the present study. This finding may be attributable to the method (randomization procedure) used to select horses for inclusion in the study, which could have avoided an overrepresentation of horses with a high yearling sale purchase price. However, results of a recent study5 that included 51 yearling Thoroughbreds trained in Florida suggested that Thoroughbreds with high commercial values as yearlings may have better racing opportunities than Thoroughbreds with low commercial values. In that study,5 lameness of yearling Thoroughbreds was associated with reduced amount of galloping exercise during training, particularly for lame horses with a high yearling sales price (median sales price, $225,000; median amount of galloping exercise, 250 furlongs [1 furlong is 1/8 mile; approx 50 km]) versus lame horses with a low yearling sales price (median sales price, $20,000; median amount of galloping exercise, 412 furlongs [approx 82 km]). This finding suggests that some trainers may exercise horses with a high commercial value less than they exercise horses with a low commercial value as a means of preventing musculoskeletal injury. As a result, more horses with high commercial values may have opportunities to start a race than horses with low commercial values because management standards regarding detection and treatment of lameness may be more stringent for the horses with high values.
Results of this study indicated the odds of failure to start a race as a 2-year-old were 1.78 times as great for yearling Thoroughbreds with forelimb proximal sesamoid bone osteophytes or enthesophytes as they were for yearling Thoroughbreds without such radiographic abnormalities. This finding of the present study supported findings of another study1 that the odds of starting a race were 3 times as low for yearlings with forelimb proximal sesamoid bone enthesophytes as they were for horses without that radiographic finding. One difference between the present study and that other study1 was that the percentage of horses with forelimb proximal sesamoid bone osteophytes or enthesophytes in the present study (15%) was higher than the percentage of horses with similar radiographic abnormalities in that other study1 (1%); therefore, the findings of the present study regarding such radiographic abnormalities may be more clinically relevant than those of that other study. In the present study, detection of osteophytes or enthesophytes was considered as one radiographic abnormality; this method was intended to prevent misclassification of such radiographic abnormalities and was used because yearling Thoroughbreds have a low prevalence of such abnormalities.1,2,6 Results of the present study supported findings of those other studies1,2,6 that yearling Thoroughbreds with such radiographic abnormalities have worse future racing performance than yearling Thoroughbreds without those radiographic abnormalities. However, the effects of such radiographic findings on soundness of a horse are unknown because such findings could be associated with MCP joint osteoarthritis or suspensory ligament injury.
Results of this study indicated the odds of failure to start a race as a 2-year-old were 2.02 times as great for yearling Thoroughbreds with hind limb osteochondral P1 fragments as they were for yearling Thoroughbreds without such radiographic abnormalities. These results supported results of another study1 conducted in Kentucky but did not support results of a study2 conducted in Texas. The study2 conducted in Texas had several limitations, including a small number of horses (n = 10) with osteochondral P1 fragments. We identified several explanations for the finding of the present study that hind limb osteochondral P1 fragments were associated with failure to start a race as a 2-year-old. Equine MTP (and MCP) joints are more commonly affected with traumatic and degenerative lesions versus any other joints of the appendicular skeleton.7 Osteochondral fragments in joints cause inflammation and damage to joint tissues via metalloproteinases, aggrecanases, and cathepsins released from cartilage and synovial tissue.7 Such proteases cleave type II collagen, which disrupts cartilage extracellular matrix, resulting in fluid distention of affected joints, clinical signs of mild to moderate lameness, signs of pain during palpation, and progression toward osteoarthritis.7 The MTP joints are high motion joints, which may worsen inflammation in such joints with osteochondral fragments, especially in racehorses that are prone to hyperextension of these joints at high speed. In addition, the proportion of yearling Thoroughbreds with radiographic abnormalities or a history of arthroscopic surgery of MTP joints is higher than the proportion of such horses with radiographic abnormalities or a history of arthroscopic surgery of MCP joints.1,6 Results of another study8 suggested lameness is more commonly detected for hind limbs of yearling Thoroughbreds than it is for forelimbs of such horses, and such lameness is typically attributable to joint injury.8 Also, clinical management options for yearlings with osteochondral fragments of the proximoplantar aspect of P1 are not well established because such osteochondral fragments are often embedded in fibrous tissue, the clinical importance of such findings is not well established, and surgical removal of such fragments can be difficult.7 Furthermore, osteochondral P1 fragments in horses can be surgically removed after yearling sales and before training is started. Results of another study9 conducted in a training center in Florida indicated that the median number of days lost during training was higher for the 8 yearling Thoroughbreds in that study with osteochondral P1 fragments that were surgically removed after yearling sales but before training was started (38 days lost) versus the 32 yearlings without P1 fragments (19 days lost); prevalence of osteochondral fragments on dorsal and plantar aspects of P1 was not reported separately in that study. A high number of days lost during training because of recovery from surgery or a combination of recovery from surgery and other causes of training failure could detrimentally affect the future ability of a yearling Thoroughbred to start a race as a 2-year-old. Results of the present study indicated a history of arthroscopic surgery (performed before the yearling sale) was not associated with whether a horse started a race as a 2-year-old. However, history of arthroscopic surgery performed after the yearling sale and before training started was not evaluated for yearlings with osteochondral P1 fragments in this study.
Among the 192 (48%) horses in this study that started a race, the outcomes for placing in a start, total amount of money earned, and amount of money earned per start during the 2-year-old racing year were not different between horses with or without forelimb proximal sesamoid bone osteophytes or enthesophytes or between horses with or without hind limb osteochondral P1 fragments. These findings may indicate that such presale radiographic findings have a greater effect on performance of yearling Thoroughbreds in training than they have on racing performance of 2-year-old Thoroughbreds. Such radiographic findings in combination with risk factors for reduced athletic performance not evaluated in this study may prevent yearlings from starting a race as 2-year-old horses. For example, the present study and other studies1,2 had limitations because known risk factors (ie, lameness and exercise history)5,8,9 for reduced athletic performance of yearling Thoroughbreds during training (ie, before the 2-year-old racing year) and the associations of such risk factors with future racing performance were not evaluated. In addition, although results of the present study indicated yearling presale radiographic findings of forelimb proximal sesamoid bone osteophytes or enthesophytes and findings of hind limb osteochondral P1 fragments were associated with failure of horses to start a race as 2-year-old Thoroughbreds, further evaluation would be required to determine the long-term (ie, during the 3-year-old racing year) effects of such radiographic findings on future racing performance.
This study had several limitations. In another study,3 the sample size (n = 397 horses) was determined for estimation of the prevalence of various radiographic findings for yearling Thoroughbreds and was not calculated for determination of the relationship between detection of various radiographic abnormalities and future racing performance. For example, the number (n = 12) of horses in this study with forelimb osteochon dral P1 fragments detected during yearling prepurchase examinations was too small to adequately evaluate the relationship between this variable and 2-year-old racing performance. Results of another study1 indicated fewer yearling Thoroughbreds with forelimb osteochondral P1 fragments started a race during 2- or 3-year-old racing years versus yearlings without such radiographic findings. Another limitation of the present study was that all radiographs were evaluated by 1 veterinarian, and the accuracy (sensitivity and specificity) of the method of radiographic evaluation was not assessed.3 Sales day (early [days 1 to 4] vs late [days 6 to 14]) was used as a proxy variable for evaluation of the relationship between horse quality (pedigree and conformation) and future racing performance. Although this approach for determination of horse quality has some value, assessment of the accuracy of the method would require further evaluation. Another limitation in this and other studies1,2 was that exercise histories prior to racing of horses as 2-year-old Thoroughbreds were not available for evaluation. Although the logistic regression model for the outcome of failure to start a race (which included findings for 2 radiographic abnormalities [forelimb proximal sesamoid bone osteophytes or enthesophytes and hind limb MTP joint P1 OC fragments]) fit the data well, the model fitness assessment was calculated for 3 risk groups (instead of ≥ 6 risk groups, which is recommended4 for assessment of model fit). This likely affected the ability of the model to discriminate between horses that started a race and those that did not. Inclusion of a larger sample size of horses with various radiographic abnormalities and other factors that affect athletic performance (ie, exercise history during training before horses start racing as 2-year-olds) would likely improve performance of the model. Results of this study apply only to yearling Thoroughbreds sold at the September 2006 yearling sale at a Thoroughbred sales facilitya in Lexington, Ky, and extrapolation of these results for yearling Thoroughbreds sold at that facility more recently should be performed with caution. To the authors' knowledge, the present study is the third study in which associations between various presale radiographic findings for yearling Thoroughbreds in the United States and future racing performance of those horses were determined. Despite the limitations of this study, results supported findings of another study1 that detection of proximal sesamoid bone osteophytes or enthesophytes in forelimbs or proximal osteochondral P1 fragments in hind limbs of yearling Thoroughbreds is associated with failure to start a race during the 2-year-old racing year.
ABBREVIATIONS
CI | Confidence interval |
MCP | Metacarpophalangeal |
MTP | Metatarsophalangeal |
OCD | Osteochondritis dissecans |
P1 | Proximal phalanx |
Keeneland Thoroughbred Racing and Sales, Lexington, Ky.
Research Randomizer, Geoffrey C. Urbaniak and Scott Plous, Middletown, Conn. Available at: research randomizer.org. Accessed Mar 20, 2011.
SamplePower, version 2.0, SPSS Inc, Chicago, Ill.
The Jockey Club Information Systems. Online factbook. Available at: www.jockeyclub.com/factbook.asp. Accessed Mar 20, 2011.
References
1. Kane AJ, McIlwraith CW, Park RD, et al. Radiographic changes in Thoroughbred yearlings. Part 2: associations with racing performance. Equine Vet J 2003;35:366–374.
2. Cohen ND, Carter GK, Watkins JP, et al. Association of racing performance with specific abnormal radiographic findings in Thoroughbred yearlings sold in Texas. J Equine Vet Sci 2006;10:462–474.
3. Preston SA, Zimmel DN, Chmielewski TL, et al. Prevalence of various pre-sale radiographic findings and association of findings with sales price in Thoroughbred yearlings sold in Kentucky. J Am Vet Med Assoc 2010;236:440–445.
4. Hosmer DW, Lemeshow S. Assessing the fit of the model. In: Hosmer DW, Lemeshow S, eds. Applied logistic regression. 2nd ed. Hoboken, NJ: John Wiley & Sons, 2000;143–202.
5. Preston SA, Brown MP, Chmielewski TL, et al. Effects of yearling sale purchase price, exercise history, lameness, and athletic performance on purchase price of Thoroughbreds at 2-year-old in-training sales. J Am Vet Med Assoc 2012;241:1499–1504.
6. Kane AJ, Park RD, McIlwraith CW, et al. Radiographic changes in Thoroughbred yearlings. Part 1: prevalence at the time of the yearling sales. Equine Vet J 2003;35:354–365.
7. Pool RR. Pathologic manifestations of joint disease in the athletic horse. In: McIlwraith CW, Trotter GW, eds. Joint disease in the horse. Philadelphia: WB Saunders Co, 1996;87–104.
8. Preston SA, Trumble TN, Zimmel DN, et al. Lameness, athletic performance, and final returns in yearling Thoroughbreds bought for the purpose of resale for profit. J Am Vet Med Assoc 2008;232:85–90.
9. Hernandez J, Hawkins DL. Training failure among yearling horses. Am J Vet Res 2001;62:1418–1422.