The full economic costs associated with colic surgery in racehorses are not known. The short-term costs in terms of medical bills can be estimated by veterinarians offering the services, and the losses caused by time off racing can be best estimated by the owner or trainer for each specific horse; however, the potential for decreased performance and therefore loss of expected winnings has not been described, to the authors’ knowledge.
Little information is available regarding performance after exploratory celiotomy for colic. In populations of horses from various disciplines and ages, 68% to 84% of horses that survived 6 months after surgery have been reported to perform at or above their intended level of use following colic surgery.1,2 Specifically in Thoroughbred racehorses, those that underwent surgery at < 2 years of age were reported to have a reduced likelihood to race but no impairment of performance if they did race, compared with maternal siblings.3 There is no information describing an objective measure of postoperative performance in horses with a preestablished performance level in the literature, to the authors’ knowledge.
The purpose of the study reported here was to determine racing performance after colic surgery in Thoroughbreds. We hypothesized that undergoing colic surgery may make a horse less likely to race again but would not affect an adult Thoroughbred racehorse's performance in terms of earnings, starts, or earnings per start.
Materials and Methods
Case and control selection—Thoroughbreds admitted because of colic at Chino Valley Equine Hospital from 1996 to 2010 that had raced at least once prior to admission, underwent exploratory celiotomy, and survived to discharge were included in the surgical cohort. The reference (control) cohort included 2 horses from the last race in which each horse that underwent surgery competed prior to surgery that were not known to have been treated for colic. Because age, sex, racing class, and track surface significantly affect race earnings and starts,4 reference horses were chosen as the 2 horses closest in placing to the horse that underwent surgery that were matched for age and sex. Because the reference horses were race matched, they were also matched for class and track surface as of the last race prior to surgery. If records were not available from that race (eg, if the horse raced in another country prior to surgery), then reference horses were chosen from the first race after surgery If there were no horses that were age and sex matched, then horses within 1 to 2 years of age or another sex were chosen as reference horses.
Medical records review—For horses that underwent surgery, data acquired from medical records included the sex, age at surgery, site of primary gastrointestinal tract lesion, and whether a resection and anastomosis was performed. For both cohorts, the race earnings, starts, and earnings per start were obtained for 6-month intervals for each horse for the 24 months preceding the surgery date and up to 42 months following surgery. The surgery date for each horse that underwent surgery and matched reference horse was defined as the end of the month in which surgery was performed on the horse that underwent surgery. Horses that had a single start prior to surgery were not excluded from the analysis because the large data set and versatile analysis protected against the effects of low-frequency observations in the data. Analysis of the preoperative period validated that the reference cohort was adequately matched. All race records were obtained from an online database.a For each horse that underwent surgery that did not return to racing, the trainer or owner was contacted by telephone to determine the reason that the horse did not race.
Definitions—The first 6-month period after surgery, during which the horses that underwent surgery were recovering, was defined as the recovery period. The postoperative period was defined as the 36 months following the recovery period. Performance analysis excluded the recovery period.
Outcomes were defined as return to racing, career longevity, and performance. Return to racing was defined as having 1 or more starts in the postoperative period. Career longevity was the period until the horse stopped racing or was censored and was measured in 6-month periods. The time at which horses stopped racing was defined as the first 6-month period with no starts after which the horse did not race again. Censoring occurred when < 42 months of follow-up were available and horses were still racing. Performance included analysis of earnings, starts, and earnings per start for the total pre- and postoperative periods.
Statistical analysis—Return to racing and performance were evaluated as separate questions such that analysis of postoperative racing performance excluded horses that did not return to racing. Return to racing was evaluated by use of logistic regression analysis, with the exposure variable being surgery, to quantify the association with an OR. Age, sex, site of gastrointestinal tract lesion, and resection were considered as potential confounders to the outcome-surgery relationship. Each variable was added individually, and variables that changed the OR by > 10% or resulted in a significant (P ≤ 0.05) outcome-exposure association were considered to indicate a confounding effect.5 The surgical cohort was divided by the site of gastrointestinal tract lesion into those undergoing surgery because of small intestinal versus large intestinal lesions. A χ2 test evaluating rate of return to racing was performed between each surgical cohort and its matched reference cohort and then between the 2 surgical cohorts. χ2 Tests were applied to the reference group to determine whether there was a baseline effect of sex or age on rate of returning to racing. Because of low sample size for some horse ages, the ages were collapsed into 3 groups as 2 and 3 years old, 4 years old, and ≥ 5 years old.
The total earnings, starts, and earnings per start before and after surgery were subjected to a Tukey ladder test to find an appropriate transformation of the data to generate outcomes with a normal distribution. No consistent transformation was identified; therefore, performance variables were compared between the horses that underwent surgery and reference horses via bootstrapping-based regression with 1,000 samples drawn from the subject-level population. Results were confirmed with a standard unpaired, unequal variance t test and a robust regression method as suggested by Campbell.6 Although all divergences were investigated, only results obtained by use of the bootstrap method are reported because of its less restrictive assumptions. Postoperative performance was evaluated by comparison between the surgical and reference groups rather than by comparison from before to after surgery within a single group to account for the effects of increasing age through the study period. Descriptive data are presented as median and IQR values because of the nonnormality of the distribution.
Career longevity of the horses that returned to racing was investigated by use of Kaplan-Meier analysis. All statistical tests were performed with statistical software,b and significance was set at P ≤ 0.05. All 95% CIs are based on the SEM derived from the mean squared error of the bootstrap.
Results
Records of 85 horses that underwent surgery were obtained for analysis. These were 42 (49%) females, 12 (14%) sexually intact males, and 31 (37%) geldings. Ages ranged from 2 to 7 years at the time of surgery, with a median age of 3 years. There were 31 (36%) small intestinal lesions, of which 11 underwent resection, and 54 (64%) large intestinal lesions, of which 2 underwent resection. The horses with small intestinal lesions comprised 10 (32%) females, 6 (19%) sexually intact males, and 15 (49%) geldings. The horses with large intestinal lesions comprised 32 (59%) females, 6 (11%) sexually intact males, and 16 (30%) geldings. The proportion of geldings was significantly (P = 0.03) higher in the small intestinal cohort. The proportion of age groups was not significantly different between the small intestinal and large intestinal surgical groups. Earnings peaked immediately prior to surgery, then slowly decreased over time; earnings per start slowly decreased through the entire study period (Figure 1).
Median race earnings (A), starts (B), and earnings per start (C) per 6 months in 85 racehorses after exploratory celiotomy for colic (surgical cohort) and a matched reference cohort of 170 racehorses. Error bars represent the IQR. As horses stopped racing, they were excluded rather than counted as having zero earnings or zero starts.
Citation: Journal of the American Veterinary Medical Association 243, 4; 10.2460/javma.243.4.532
Median race earnings (A), starts (B), and earnings per start (C) per 6 months in 85 racehorses after exploratory celiotomy for colic (surgical cohort) and a matched reference cohort of 170 racehorses. Error bars represent the IQR. As horses stopped racing, they were excluded rather than counted as having zero earnings or zero starts.
Citation: Journal of the American Veterinary Medical Association 243, 4; 10.2460/javma.243.4.532
Median race earnings (A), starts (B), and earnings per start (C) per 6 months in 85 racehorses after exploratory celiotomy for colic (surgical cohort) and a matched reference cohort of 170 racehorses. Error bars represent the IQR. As horses stopped racing, they were excluded rather than counted as having zero earnings or zero starts.
Citation: Journal of the American Veterinary Medical Association 243, 4; 10.2460/javma.243.4.532
Return to racing and longevity—Fifty-nine of 85 (69%) horses that underwent surgery returned to racing after the recovery period versus 125 of 170 (73%) reference horses (OR, 0.81; P = 0.49). Age, sex, site of gastrointestinal tract lesion, and resection each altered the OR by < 10% and were therefore not considered confounders of the effect of surgery on return to racing.
In both the surgical and reference cohorts, there was a significant effect of sex on return to racing but no effect of age. Neither the effect of sex nor that of age was significantly different between cohorts. Geldings were significantly more likely to return to racing than females (P < 0.001) or sexually intact males (P < 0.001). The difference in return to racing between females and sexually intact males was not significant. There was no significant difference in proportion of geldings with increasing age in either cohort. In the reference cohort, 54 of 61 (89%) geldings, 14 of 25 (56%) sexually intact males, and 57 of 84 (68%) females returned to racing. In the surgical cohort, 27 of 31 (87%) geldings, 9 of 12 (75%) sexually intact males, and 23 of 42 (55%) females returned to racing.
For the analysis of site of lesion, there was no difference in return to racing when comparing each subset of horses that underwent surgery with its reference cohort; however, there was a difference when comparing the surgical small intestinal cohort and surgical large intestinal cohort. The surgical cohort with small intestinal lesions included 26 of 31 (85%) horses that returned to racing, which was significantly (P = 0.009) greater than the proportion of horses that underwent surgery with large intestinal lesions that returned to racing (33/54 [61%]; Table 1). Neither surgical cohort had significantly different rates of return to racing from their reference cohorts, which were 47 of 62 (76%) and 78 of 108 (72%) for the small intestinal and large intestinal reference cohorts, respectively.
Number (%) of racehorses treated (yes) or not treated (no) via intestinal resection during exploratory celiotomy for colic caused by a lesion in the small intestine or large intestine and number or proportion (No. [%]) that returned to racing.
| Variable | Resection | Total horses | Horses that raced |
|---|---|---|---|
| All horses (n = 85) | Yes | 13 (15) | 11 (85) |
| No | 72 (85) | 48 (67) | |
| Total | 85 (100) | 59 (69) | |
| Small intestinal lesion (n = 31) | Yes | 11 (35) | 9 (82) |
| No | 20 (65) | 17 (85) | |
| Total | 31 (37) | 26 (84) | |
| Large intestinal lesion (n = 54) | Yes | 2 (4) | 2 (100) |
| No | 52 (96) | 31 (60) | |
| Total | 54 (63) | 33 (61) |
Of the 26 horses that underwent surgery that did not race after surgery, telephone contact was obtained with the owners of 20 horses. The 6 horses with no telephone follow-up were all mares, and 4 went on to have at least 1 foal recorded by the Jockey Club—2 in the first and 2 in the second breeding season after surgery. Of the 20 horses for which telephone follow-up was available, 6 stopped racing because of complications with colic or surgery. Of these, 4 were euthanized because of repeated colic, 1 was retired because of laminitis, and 1 was retired because of an incisional hernia.
Five horses were started back into training but were subsequently retired or euthanized because of musculoskeletal injury (n = 4) or pneumonia (1) before their first race after surgery Nine of 20 horses were retired, with no attempt to return them to training. Six of these were retired to breeding. With regard to site of the lesion, 5 horses with small intestinal lesions did not return to racing. Of these, 2 had no telephone follow-up, 2 were euthanized because of repeated colic, and 1 was retired without returning to training. Of the 21 horses with large intestinal lesions that did not return to racing, 4 had no telephone follow-up available, 4 were euthanized or retired because of colic or surgery-associated complications, 8 were retired without returning to training, and 4 were returned to training before being retired because of injury or illness.
Career longevity analysis of the pattern of horses that ceased racing did not detect any difference between horses that underwent surgery and reference horses (Figure 2). Career longevity was considered both from the surgery date (ie, including all horses entered in the study) and separately from the start of the postoperative period (ie, including only those horses that returned to racing), and both methods yielded no difference between cohorts. There was no difference in career longevity between the small intestinal and large intestinal surgical cohorts and their respective reference cohorts.
Results of Kaplan-Meier analysis for race career longevity of the same horses as in Figure 1. Time 0 was considered as the end of the month in which the horse underwent surgery. The decrease at 6 months represents the horses that did not return to racing.
Citation: Journal of the American Veterinary Medical Association 243, 4; 10.2460/javma.243.4.532
Results of Kaplan-Meier analysis for race career longevity of the same horses as in Figure 1. Time 0 was considered as the end of the month in which the horse underwent surgery. The decrease at 6 months represents the horses that did not return to racing.
Citation: Journal of the American Veterinary Medical Association 243, 4; 10.2460/javma.243.4.532
Results of Kaplan-Meier analysis for race career longevity of the same horses as in Figure 1. Time 0 was considered as the end of the month in which the horse underwent surgery. The decrease at 6 months represents the horses that did not return to racing.
Citation: Journal of the American Veterinary Medical Association 243, 4; 10.2460/javma.243.4.532
Performance—In the postoperative period, the median earnings of horses that returned to racing was $22,600 for horses that underwent surgery and $25,808 for reference horses. Median number of starts was 9 for horses that underwent surgery and 9 for reference horses. Median earnings per start was $1,712 for horses that underwent surgery and $2,314 for reference horses (Table 2).
Postoperative mean, median, and range of earnings, starts, and earnings per start for 59 racehorses that returned to racing 6 months after exploratory celiotomy for colic (surgical group) and 125 race-matched reference racehorses (reference group).
| Cohort | Mean (95% CI) | Median (IQR) | Range |
|---|---|---|---|
| Surgical group | |||
| Earnings ($) | 52,934 (30,205–75,662) | 22,600 (9,190–55,138) | 0–566,900 |
| No. of starts | 12.1 (9.68–14.6) | 9 (3.5–20) | 1–34 |
| Earnings per start ($) | 4,853 (2,616–7,091) | 1,712 (1,010–5,051) | 0–62,989 |
| Reference group | |||
| Earnings ($) | 60,800 (40,208–81,392) | 25,808 (6,359–70,994) | 0–881,726 |
| No. of starts | 12.4 (10.8–13.9) | 9 (5–19) | 1–40 |
| Earnings per start ($) | 4,883 (3,243–6,522) | 2,314 (890–5,541) | 0–79,135 |
Bootstrapping-based regression revealed no difference between horses that underwent surgery and reference horses for any performance outcome (earnings, starts, or earnings per start) in either the pre- or postoperative period. The mean postoperative earnings of horses that returned to racing was $52,934 for horses that underwent surgery and $60,800 for reference horses. Mean number of starts was 12.1 for horses that underwent surgery and 12.4 for reference horses. Mean earnings per start was $4,853 for horses that underwent surgery and $4,883 for reference horses (Table 2). This meant that, in the 36-month postoperative period, reference horses earned a mean of $7,866 more (95% CI, –$22,825 to $38,558; P = 0.62), had a mean of 0.26 more starts (95% CI, −2.6 to 3.1; P = 0.86), and had a mean earnings per start of $29 more than horses that underwent surgery (95% CI, –$2,755 to $2,814; P = 0.98). Significance values determined by use of the t test and robust regression were in agreement with bootstrapping-based regression for all tests evaluated.
Performance analysis was repeated with the cohorts allocated by the site of gastrointestinal tract lesion in the horses that underwent surgery. No significant difference in performance was found between horses that underwent surgery and reference horses in either group before surgery. After surgery, no difference was detected between horses with large intestinal lesions that underwent surgery and their reference cohort (Table 3). In the small intestinal cohorts, the starts and earnings per start were not significantly different between the horses that underwent surgery and reference cohorts (Table 4). However, the postoperative earnings were lower in the surgical than the reference cohort as determined via bootstrapping regression. The bootstrapping-based small intestinal surgical cohort earnings were $23,867, and reference cohort earnings were $57,357, with an estimated difference in earnings of $33,490 (95% CI, $1,170 to 65,809) more for the reference horses (P = 0.042).
Postoperative mean, median, and range of earnings, starts, and earnings per start for 33 racehorses that returned to racing 6 months after exploratory celiotomy for colic caused by large intestinal lesions (surgical group) and 78 race-matched reference racehorses (reference group).
| Cohort | Mean (95% CI) | Median (IQR) | Range |
|---|---|---|---|
| Surgical group | |||
| Earnings ($) | 75,835 (39,869–111,800) | 24,340 (11,090–80,246) | 0–566,900 |
| No. of starts | 11.9 (8.85–15.0) | 8 (5–19) | 1–34 |
| Earnings per start ($) | 7,184 (3,233–11,134) | 2,967 (1,382–7,533) | 400–62,989 |
| Reference group | |||
| Earnings ($) | 62,875 (34,310–91,439) | 25,504 (4,862–77,197) | 0–881,726 |
| No. of starts | 12.5 (10.4–14.6) | 9.5 (4–19.75) | 1–40 |
| Earnings per start ($) | 4,884 (3,219–6,549) | 2,344 (797–5,847) | 0–44,086 |
Postoperative mean, median, and range of earnings, starts, and earnings per start for 26 racehorses that returned to racing 6 months after exploratory celiotomy for colic caused by small intestinal lesions (surgical group) and 47 race-matched reference racehorses (reference group).
| Cohort | Mean (95% CI) | Median (IQR) | Range |
|---|---|---|---|
| Surgical group | |||
| Earnings ($) | 23,867 (13,297–34,437) | 17,565 (1,660–37,651) | 0–127,094 |
| No. of starts | 12.4 (8.32– 16.5) | 9.5 (3–20.75) | 2–33 |
| Earnings per start ($) | 1,896 (1,105–2,686) | 1,466 (441–2,475) | 0–8,902 |
| Reference group | |||
| Earnings ($) | 57,357 (26,642–88,072) | 27,066 (8,217–67,623) | 282–712,212 |
| No. of starts | 12.3 (10.0–14.5) | 9 (7–17) | 1–31 |
| Earnings per start ($) | 4,880 (1,628–8,132) | 2,023 (1,064–4,968) | 58–79,135 |
Discussion
This study objectively evaluated the athletic performance of Thoroughbred racehorses following exploratory celiotomy for colic. No decline in performance attributable to the surgery was detected in the surgical cohort as a whole, although horses with small intestinal lesions may have had reduced performance after surgery.
The use of a reference cohort to evaluate the baseline attrition rate in racehorses was essential to the understanding of this study. Horses stop racing for a variety of reasons, including injuries, breeding, or a decline in performance, which may not be related to a given medical procedure.7 Only 73% of the reference cohort in this study was still racing 6 months beyond the time at which the matched horses that underwent surgery received surgery. Without this information, a return to racing rate of 69% in the horses that underwent surgery may have been seen as an important limiting effect of surgery.
The horses that underwent surgery with small intestinal lesions were more likely to return to racing than the horses that underwent surgery with large intestinal lesions. This was most likely because of the difference in proportion of geldings in each group. However, the horses with small intestinal lesions that underwent surgery had a higher rate of return to racing than their reference cohort and the horses with large intestinal lesions that underwent surgery had a lower rate of return to racing than their reference cohort, despite each surgical-reference pairing having the same proportion of geldings. The difference between each surgical cohort and its reference cohort was not significant in this study, and a larger sample size would be required to determine whether this result reflects a true difference.
Investigation of the reasons that horses that underwent surgery did not return to racing revealed that the majority (14/20) were retired for reasons unrelated to the colic episode or surgery. Four of the 83 (5%) horses with follow-up were euthanized because of recurrent colic after discharge from the hospital. This was in keeping with the reported rate of death of 5% between 10 to 100 days after colic surgery.8 Other causes of not returning to racing were laminitis and incisional herniation, which have also been identified as risk factors for a decreased rate of return to use in previous studies.1,2
The career longevity curves (ie, Kaplan-Meier curves) were similar between cohorts, suggesting that major performance-limiting complications associated with the surgery did not appear to be a problem. Other variables did not have an effect on the OR of returning to racing, which suggested that each age and sex was equally affected by exposure to surgery and there was not a differentially worse prognosis for a particular group. For example, undergoing colic surgery did not seem to precipitate the end of the career of an older horse nor did it seem to result in females being retired to breeding earlier than they might have otherwise. The career length of horses that underwent surgery with small intestinal lesions was not significantly shorter than their reference cohort, suggesting that the cause of their potentially decreased earnings was not an inability to race and a shortened career. The lack of significant findings within subsets of the surgical cohort may be because of low sample sizes in the subgroups and should be investigated further.
Earnings increased sharply immediately prior to surgery, followed by a slow decrease after surgery that was similar between the surgical and reference cohorts (Figure 1). The immediate preoperative period was intentionally matched in earnings by choosing race-matched, placing-matched reference horses. That the reference cohort had a sharp decrease in earnings immediately after this time, despite nothing occurring to those horses at time 0, suggests that the 6 months before the surgery date was the natural peak of their careers. The fact that the horses that underwent surgery often had colic surgery at the peak of their careers suggests 2 possibilities. Either peak performance (and associated stressors, feeding practices, and management) is a risk factor for colic, or owners are more likely to pursue surgery for horses that are performing well.
The issue of racing performance effect on equine health management arises somewhat frequently in equine veterinary research and often presents an analytic challenge. Performance data, especially earnings, are generally nonparametric and highly positively skewed, given that negative earnings are not possible. Ideally, this type of data would be transformed into a normal population (eg, by log transformation). However, transformation may not be possible, as with this data set. Bootstrap methods have been used in medical research for economic evaluations and are recommended for data with skewed distributions because they have the advantage of making no assumption of the true shape of the population.6,9–11 Two other methods were used in concert with bootstrapping to examine the effect of surgery on outcome. These were the unpaired t test with Satterthwaite correction and robust regression with the Huber-White error estimator. We performed this 3-way exploration with the following understanding: if outliers were present, then the robust regression result may be different from the other 2; if nonnormality (skewness) were present, then the bootstrap result may be different from the other 2; and if variance difference per se were influencing the statistics, the correction for heteroscedasticity in the Satterthwaite t test may lead to its separation from the other 2. Agreement between the methods lends confidence to the inferences made. In this investigation, all 3 methods agreed in level of significance, with 1 exception: the small intestinal group's postoperative earnings. Bootstrap and t test methods yielded extremely close values with marginal significance (P = 0.042, and P = 0.051 respectively). The robust regression removed the effect of the high earners and revealed no difference (P = 0.115); however, the robust regression method is the most susceptible to nonnormality of data. Because bootstrap methods required fewer assumptions, reporting of the bootstrap method was preferred.6 However, the finding of reduced earnings in horses that underwent surgery with small intestinal lesions, compared with their reference cohort, requires further validation with a larger sample size.
A major limitation of this study was the small sample size in relation to the large variation in earnings, which was reflected in wide 95% CIs. Because of this, it is possible that subtle differences between populations were not detected. This was especially true for the analysis of subgroups (age, sex, site of lesion, and resection). However, there was good agreement between the findings of this study and the existing reports on postoperative athletic performance. Of the 77 horses known to have survived 6 months after surgery, 77% returned to racing, which was quite similar to reported return-to-use rates of 68% to 84% in horses that survived 6 months.1,2 Also, the present study did not identify any loss of performance following colic surgery, which was in agreement with a previous investigation of juvenile Thoroughbreds.3
ABBREVIATIONS
| CI | Confidence interval |
| IQR | Interquartile range |
EQUIBASE [database online]. Lexington, Ky: Equibase Co LLC, 2012. Available at: www.equibase.com/. Accessed Oct 15, 2012.
Stata Statistical Software, release 12, StataCorp LP, College Station, Tex.
References
1. Christophersen MT, Tnibar A, Pihl TH, et al. Sporting activity following colic surgery in horses: a retrospective study. Equine Vet J Suppl 2011;(40):3–6.
2. Davis W, Fogle CA, Gerard MP, et al. Return to use and performance following exploratory celiotomy for colic in horses: 195 cases (2003–2010). Equine Vet J 2013; 45:224–228.
3. Santschi EM, Slone DE, Embertson RM, et al. Colic surgery in 206 juvenile Thoroughbreds: survival and racing results. Equine Vet J Suppl 2000;(32):32–36.
4. Cheetham J, Riordan AS, Mohammed HO, et al. Relationships between race earnings and horse age, sex, gait, track surface and number of race starts for Thoroughbred and Standardbred racehorses in North America. Equine Vet J 2010; 42:346–350.
5. Woodward M. Epidemiology: study design and data analysis. 2nd ed. Boca Raton, Fla: Chapman & Hall/CRC, 2005.
6. Campbell MJ. Statistics at square two. London: BMJ Books, 2006; 133.
7. Wilsher S, Allen WR, Wood JL. Factors associated with failure of Thoroughbred horses to train and race. Equine Vet J 2006; 38:113–118.
8. Proudman CJ, Smith JE, Edwards GB, et al. Long-term survival of equine surgical colic cases. Part 1: patterns of mortality and morbidity. Equine Vet J 2002; 34:432–437.
9. Good PI. A practitioner's guide to resampling for data analysis, data mining, and modeling. Boca Raton, Fla: CRC Press, 2012.
10. Lambert CM, Hurst NP, Forbes JF, et al. Is day care equivalent to inpatient care for active rheumatoid arthritis? Randomised controlled clinical and economic evaluation. BMJ 1998; 316:965–969.
11. Vittinghoff E. Regression methods in biostatistics: linear, logistic, survival, and repeated measures models. 2nd ed. New York: Springer, 2012.
