Email: firstname.lastname@example.org. S.H. Franklin's present address is: School of Animal and Veterinary Sciences, The University of Adelaide, Roseworthy Campus, Roseworthy, South Australia 5371, Australia.
Reasons for performing study: Overground endoscopy is being performed with increasing frequency in the UK. A previous study has shown that the type of exercise test may affect the diagnosis of upper respiratory tract (URT) obstructions. The successful clinical application of overground endoscopy systems requires understanding of appropriate field exercise testing protocols.
Objectives: The aim of this study was to report the exercise test parameters used during overground endoscopy in UK Thoroughbred racehorses and to investigate potential effects of these on the diagnosis of URT obstructions.
Methods: The exercise test parameters used and the endoscopic observations of the URT during exercise were recorded for 140 Thoroughbred racehorses referred for the investigation of abnormal respiratory noise and/or poor athletic performance.
Results: The exercise test parameters varied widely between horses. The presenting complaint was easiest to reproduce in horses with a history of abnormal noise in training and hardest to reproduce in those referred for investigation of poor performance with no abnormal noise. No associations between the presence or not of an URT obstruction and exercise test parameters was identified. For horses referred for abnormal noise during racing DDSP was more likely to be observed when longer distances were performed.
Conclusions: It is difficult to standardise exercise tests in the field when multiple premises are used and when training gallops differ markedly to racecourses. It was not possible to establish exercise test protocols which should be used for all Thoroughbred racehorses. Therefore, at present the best advice for horses which make abnormal noise during a race or have poor race performance, can only be to replicate race conditions as closely as possible. For horses that race over longer distances this will require the use of a circular gallops (e.g. racecourse) if only short gallops are available at the trainer's premises.
Several studies have documented that dynamic upper respiratory tract (URT) obstructions cannot be diagnosed accurately during a resting endoscopic examination and that endoscopy during exercise is required to make a definitive diagnosis (Kannegieter and Dore 1995; Parente and Martin 1995; Tan et al. 2005; Lane et al. 2006b). Exercising endoscopy can be undertaken either on a high-speed treadmill or in the field.
Overground (field) endoscopy is being performed with increasing frequency in the UK. A previous study comparing overground endoscopy and treadmill endoscopy suggested that the type of exercise test may affect the ability to make a diagnosis of dynamic URT obstructions (Allen and Franklin 2010). In particular, the occurrence of dorsal displacement of the soft palate (DDSP), was more likely to occur under more strenuous exercise testing conditions (Allen and Franklin 2010). In many circumstances overground endoscopy is performed at the trainer's premises over routine training speeds and distances. However, in contrast to racing in some other countries, many trainers in the UK only undertake training on short inclined gallops and therefore the speeds and distances experienced during training may not be the same as those experienced during racing.
The successful clinical application of overground endoscopy systems requires the development of appropriate field exercise testing protocols. The aim of this study was to report the exercise test parameters used during overground endoscopy in UK Thoroughbred racehorses and to investigate potential effects of these on the diagnosis of URT obstructions.
Materials and methods
This study was restricted to Thoroughbred racehorses referred for overground endoscopy for the investigation of abnormal respiratory noise and/or poor athletic performance. Only horses in which the overground endoscopy was performed on a gallops or a racecourse were included. The first 50 horses in this study have also been described in a previous study (Allen and Franklin 2010). The technique and equipment has previously been described (Franklin et al. 2008; Allen and Franklin 2010). The history and presenting complaints were recorded. For horses with a history of abnormal respiratory noise during training, the endoscopy was performed during a normal gallop training session. For those that were referred with a history of abnormal noise or poor performance only during races the trainers were also asked to perform a strenuous training session. If no abnormality was observed further exercise was undertaken at the trainer's discretion. The speeds, distances and inclines that the exercise test was performed over were recorded using a GPS monitor (Garmin Forerunner 305)1. The distance was recorded from the start of exercise to the end of exercise and therefore includes the distance for acceleration and deceleration. When more than one interval was performed, the distance of each interval was added to calculate a total distance and this was approximated to the nearest 100 m. The average speed of the exercise test was calculated by dividing the exercise test distance with the time taken. Heart rates (HR) were recorded1 concurrently throughout the exercise test. After the exercise test the veterinary surgeons assessed whether the presenting complaint had likely been reproduced during the exercise test, taking account of information from the trainer and jockey.
For horses that had raced previously the distance the horse last raced over was recorded (http://www.racingpost.com) to permit comparisons with the exercise test. In addition, the average speed for the race was calculated by the time taken by the winning horse to complete the race. A test race ratio was calculated, whereby the total exercise test distance was divided by the distance the horse last raced over. Therefore, horses undergoing exercise tests at equal or greater distances than the horses racing distance had values greater than or equal to one.
Statistical analysis was performed using PASW 17.0. Preliminary statistics were undertaken to establish whether data were normally distributed. Nonparametric tests were used for skewed data. Horses were divided into 3 groups according to presenting complaint: abnormal noise in training, abnormal noise in racing and poor race performance without abnormal noise. Chi-square tests were used to compare the presence of URT abnormalities or whether the presenting complaint had been reproduced between the 3 groups. Spearman or Pearson correlations were used to assess correlations between individual exercise test parameters. Independent t tests or Mann Whitney tests were used to assess exercise test parameters and the presence of URT obstructions. Wilcoxon signed rank tests were used to compare the test and race data. Statistical significance was set at P<0.05.
The inclusion criteria were met by 140 horses, 80 flat and 60 National Hunt (NH) racehorses. The ages ranged from 2–11 years (median 4 years). There were 19 mares, 51 colts and 70 geldings.
The exercise test parameters used are shown in Table 1. Thirty-five different tracks were used. The total exercise test distance varied from 800–6100 m (median 1900 m). The test was performed in more than one increment in 50 horses. The exercise test was performed on a circular/oval gallops or racetrack in 19 horses. The straight gallops were all inclined to varying degrees, whereas the circular gallops used were flat (P<0.001). The median test distance undertaken on circular gallops was 3200 m and the median test distance on straight gallops was 1600 m (P<0.001). There was no significant difference in mean peak speed between circular gallops (57 km/h) and straight gallops (56 km/h) (P = 0.46). For straight gallops, peak speed was negatively correlated with incline (r =−0.266 P<0.001). There was also a negative correlation between incline and distance (r =−0.403 P<0.001), as inclined gallops tended to be shorter.
Table 1. The exercise test parameters used during the overground endoscopy procedure in 140 Thoroughbred racehorses
Exercise test parameter
Total test distance (m)
Distance of first or only increment (m)
Peak speed (km/h)
Average speed (km/h)
A total of 125 horses had raced prior to the endoscopic examination. For flat racehorses there was no significant difference between median test distance (1600 m) and median race distance (1600 m) (P = 0.44). However, for NH racehorses median test distance (2500 m) was significantly lower than race distance (4200 m) (P<0.001). There was a significant difference between peak speed achieved by flat horses (58 km/h) compared with NH horses (53 km/h) (P<0.001). The average speeds of the exercise tests for both flat and NH horses were significantly lower than the average speeds needed to win the race (P<0.001). Table 2 shows the average race speeds for the winning horse for different race lengths and the peak and average speeds for the exercise test for that category.
Table 2. The winning speeds for the last race prior to referral for overground endoscopy and comparisons with the speeds encountered during the exercise test
Race distance miles/furlongs
Average racing speed km/h
Peak test speed km/h
Average test speed km/h
Flat up to 1mile (= 1609 m)
54–63 (mean 58)
50–69 (mean 59)
30–50 (mean 41)
Flat greater than 1 mile
53–60 (mean 56)
50–64 (mean 57)
32–47 (mean 41)
NH up to 2.5 miles (= 4022 m)
45–55 (mean 50)
45–66 (mean 55)
32–50 (mean 39)
NH greater than 2.5 miles
44–52 (mean 48)
37–61 (mean 53)
30–54 (mean 41)
The peak heart rates achieved during the exercise tests varied from 178–258 beats/min (median 219 beats/min, mean 220 beats/min). In many horses high heart rates were observed prior to exercise and only 8 horses had peak heart rates less than 200 beats/min. Five horses were considered to have an abnormally elevated heart rate response during exercise (>250 beats/min), of which 2 horses were confirmed to have paroxysmal atrial fibrillation.
The URT abnormalities identified and the numbers of horses observed with each abnormality are shown in Figure 1. Normal URT function was observed in 27 horses, a single URT abnormality in 48 and multiple abnormalities in 65 horses. Palatal dysfunction was the most common abnormality observed in 103 horses; 84 had palatal instability (PT) during exercise, of which 11 experienced DDSP on pulling up and 19 had DDSP during strenuous exercise. Axial deviation of the aryepiglottic folds (ADAF) was also observed commonly and there was a significant association with palatal dysfunction (P = 0.001).
There were no significant differences in exercise test parameters between horses with and without a diagnosis of URT obstruction. The median exercise test distance was 1900 m (P = 0.84) and median incline 3% (P = 0.33) for both groups. The mean speed was 56 km/h for horses diagnosed with an URT obstruction and 55 km/h for horses without a URT obstruction (P = 0.64).
Horses reported to make abnormal noise in training
Sixty-five horses were reported to make abnormal noise in training. In sixty (92%) horses one or more abnormalities of the URT were observed (Fig 2). The presenting complaint was reproduced in 53 (82%) horses. Further details of the 12 horses where the presenting complaint was not reproduced are shown in Table 3.
Table 3. Details of racehorses in which the presenting complaint was not reproduced during the exercise test
Category of presenting complaint
Number of horses that presenting complaint was not reproduced
Reason and endoscopic observation
Abnormal noise during training
7 horses reported to specifically make ‘gurgling’ noise, which was not reproduced during the exercise test. Palatal instability was observed in all cases but DDSP did not occur. 4 horses unspecified ‘abnormal noise’ was reported, but no abnormality was observed and no noise was heard during the test. 1 horse a ‘loud snoring noise’ was reported. Epiglottic retroversion was observed in this horse during walk, but did not occur during faster speeds and only a ‘whistle’ was heard at faster speeds which was associated with vocal cord collapse.
Abnormal noise during racing
7 horses were reported to make unspecified ‘abnormal noise’, yet no noise was heard during the test. In 5 no endoscopic abnormalities were observed and in 2 only mild palatal instability was observed. 3 horses, a ‘gurgling’ noise was reported which was not reproduced during the test. 2 showed palatal instability and in 1 horse no abnormality was observed. 2 horses were reported to make abnormal noise concurrently when the horse stopped abruptly in a race, but during the test only a low grade inspiratory noise was heard however both horses performed well and this noise was not associated with stopping suddenly.
Poor performance with no abnormal noise
For horses referred for pulling up in races, stopping in races and slowing suddenly in races this was never reproduced during testing. For horses with disappointing or poor race form this was either not reproduced, i.e. the horse worked well in the test or was unanswered as this was found to be very difficult to assess.
Horses reported to make abnormal noise in racing only
Thirty-one horses were reported to make abnormal noise during racing. In 26 (84%) horses one or more abnormalities of the URT were observed (Fig 2). The presenting complaint was reproduced in 19 (61%) horses. Further details of the 12 horses where the presenting complaint was not reproduced are shown in Table 3.
Further analysis was undertaken to establish whether exercise test parameters affect the likelihood of horses progressing from PI to DDSP. Horses diagnosed with DDSP were exercised over longer distances (mean 3000 m) than those diagnosed with PI (mean 2000 m) (P = 0.04). Dorsal displacement of the soft palate was more likely to be diagnosed when the exercise test was undertaken on circular gallops than on straight gallops (P = 0.016). There were no significant differences in peak speed between horses with DDSP and those with PI. There was a trend for the test race ratio to be greater in horses diagnosed with DDSP (mean 0.9) compared with those diagnosed with PI (mean 0.7), however this was not significant (P = 0.29).
Horses reported with poor performance and no noise
Forty-four horses presented with poor racing performance with no abnormal noise reported by the trainer. Twenty horses were referred for poor or disappointing race performance, 16 for pulling-up or stopping and 5 for slowing, fading or failing to finish. In 27 (61%) horses an abnormality of the URT was observed (Fig 2). However, it was difficult to reproduce the presenting complaint for horses in this category (Table 3). In the 3 horses (7%) where the complaint was considered to have been reproduced, 2 were reported to have excessive blowing post exercise as well as poor performance and one was reported to have become wobbly and to have EIPH as well as racing poorly. In all 3 cases a nonURT reason for the poor performance was identified.
Between group comparisons
There was a significant difference between type of racehorse (flat/NH) and presenting complaint group (P = 0.014). A greater proportion of flat horses were reported to make noise in training.
There was a significant difference in whether the presenting complaint was reproduced between the groups (P<0.001), the presenting complaint was easiest to reproduce in horses referred for abnormal noise in training and hardest to reproduce in those referred for poor performance. There was a significant difference in whether a URT abnormality was observed between the groups (P<0.001); URT abnormalities were most likely observed in horses that made abnormal noise during training and least likely in those referred for poor performance without abnormal noise. Horses were significantly less likely to have a URT abnormality observed during the test if the presenting complaint was not reproduced (P<0.001).
There were no significant differences between the groups in the proportion of horses with DDSP (DDSP) (P = 0.31), PI (P = 0.30), ADAF (P = 0.52) or pharyngeal wall collapse (PWC) (P = 0.81) (Fig 2). There was a significant difference for arytenoid cartilage collapse and vocal fold collapse (ACC+VCC) (P = 0.021), which was less common in horses referred for poor performance without abnormal noise.
The aim of this study was to report the exercise test parameters used during overground endoscopy in UK Thoroughbred racehorses and to identify whether these affected the ability to diagnose URT abnormalities.
The results of this study show there was marked variation in the exercise test parameters that were used during overground endoscopy. The exercise test was most frequently performed at the trainers' premises, therefore tests were highly dependent on the facilities available at that training yard and are subsequently very difficult to standardise. The advantages of undertaking exercise tests on high-speed treadmills are that the speeds, distances and inclines can be standardised between horses and, as they can be decided upon prior to the test, are under exact control of the veterinary surgeon. Whereas during overground endoscopy the exercise tests were more dependent upon the trainer, jockey and facilities and the veterinary surgeon was less able to specify exact parameters of the exercise test.
For many horses a normal ‘gallop’ training session was performed and this type of field exercise test was used in Thoroughbreds in 2 recent publications (Gramkow and Evans 2006; Vermeulen and Evans 2006). However, it has previously been shown that gallop speeds vary considerably between trainers (Dyson et al. 2003). Peak speeds during testing were similar to the average speeds obtained from winning horses during racing. However, these were maintained only briefly during the exercise test and hence average speeds during testing were significantly lower. For flat horses, the distances performed during training may be similar to race distances whereas for NH horses training distances are markedly shorter than the distances encountered during racing. Interestingly, this probably explains why a significantly larger proportion of flat horses were referred for abnormal noise in training compared with NH horses. The differences in the training and racing distance explain why NH horses were often referred for abnormal noise in racing that was not present during training.
In contrast to racing in other countries, many UK trainers often only train on short inclined gallops. In exercise test studies from other countries, horses are often trained at racetracks (Davie and Evans 2000; Vermeulen and Evans 2006) or on oval training tracks (Courouce et al. 1999; Lindner 2010). It has previously been shown that undertaking similar exercise tests on different oval tracks did not result in differences in physiological or locomotor responses between the 2 tracks (Courouce et al. 1999). It is likely that exercise tests similar to the demands of racing are more likely to be replicated on oval training gallops than on short inclined straight gallops, particularly for NH horses and also for flat horses that race over longer distances. When only short gallops were available it was often necessary to perform the exercise test in intervals. However, the rest periods between intervals may allow for partial recovery to occur and may delay the onset of fatigue (Midgley et al. 2007). This study suggests that in many cases undertaking exercise tests at trainer's premises is unlikely to be a valid test of racing.
The HR response to exercise was often of limited use in determining whether the exercise test was sufficiently strenuous because it may be influenced by other factors including underlying fitness, excitement, the presence of the endoscope, and also by the underlying disorder causing the poor performance (Courouce et al. 1999). High heart rates were readily achieved irrespective of the exercise test parameters.
There were no significant differences in exercise test parameters between horses with and without a diagnosis of URT obstruction. It is likely that URT collapse occurs when a combination of critical negative airway pressure is reached and when fatigue of the upper airway dilator muscles occurs. The finding that similar obstructions were observed irrespective of whether horses were referred for abnormal noise in training or abnormal noise in racing suggests between horse variations may be very important. For example, the inspiratory pressures and degree of fatigue required to induce an URT obstruction in one horse may be different than those required to induce the same abnormality in another horse.
Similar to other treadmill and overground endoscopy studies, palatal dysfunction was the most common URT abnormality observed (Morris and Seeherman 1991; Kannegieter and Dore 1995; Martin et al. 2000; Franklin et al. 2006; Lane et al. 2006a; Desmaizieres et al. 2009; Pollock et al. 2009). However, as previously reported (Allen and Franklin 2010), although PI was frequently observed, DDSP was diagnosed less frequently than expected. In this study DDSP was observed in only 20% of horses (14% during exercise, 6% on pulling-up), in contrast to 40% of cases in a review of Thoroughbred racehorses referred for treadmill endoscopy to the same centre (Lane et al. 2006a). For horses referred with abnormal noise in racing it was found that DDSP was more likely if longer distances were performed, which may necessitate a circular or racecourse gallop. As peak speeds were similar for circular and straight gallops, this suggests that DDSP more likely occurs as a result of fatigue (i.e. further distances) than by more negative inspiratory pressures (i.e. faster speeds). There were a high proportion of horses observed to have PI in the poorly performing group. Palatal instability is often associated with abnormal inspiratory noise (Lane et al. 2006b). However, this is quieter than the noises associated with other forms of URT collapse (Franklin 2002) and hence may not always be detected by the jockey or trainer. Further understanding of the importance of PI and its effects on airflow and performance are required. In addition, understanding whether it is possible to predict which horses observed to have PI would experience DDSP under more strenuous conditions such as racing would be beneficial.
The differences in the proportion of horses observed to have a URT abnormality when referred for abnormal noise compared to those referred for poor performance without a history of abnormal noise was not surprising. If abnormal noise is present some form of collapse or turbulence within the respiratory tract must exist to create this. It was more difficult to assess horses referred for poor racing performance without abnormal noise. If normal URT function was observed this could either be because the exercise test had not reproduced the URT problem, or that the URT was normal and there was another cause of the poor performance. It was difficult to reproduce the presenting complaint in horses referred for poor race performance. For the majority of horses referred for pulling up, stopping or slowing suddenly in races, this was not reproduced during the exercise test. For horses with poor or disappointing race performances this was either not reproduced, i.e. the horse worked well during the test or was unanswered due to the subjective nature of assessing individual horse's performance. This study confirmed that care should be taken interpreting a normal airway if the presenting complaint was not reproduced during the exercise test.
In conclusion, it is difficult to standardise exercise tests in the field when multiple premises are used and when training gallops differ markedly to racecourses. It was not possible to establish exercise test protocols which should be used for all Thoroughbred racehorses. Overground endoscopy performed at a trainer's premises is appropriate for investigation of abnormal noise during training, irrespective of the type of racing performed. For horses that only show clinical signs during racing, the exercise test must be representative of racing in order to establish a definitive diagnosis. This study suggests that undertaking exercise tests over race distances is a key factor. For many horses which race over longer distances this will require the use of a circular gallops (e.g. racecourse) if only short gallops are available at the trainer's premises. Until it becomes possible to standardise field exercise tests better, overground exercise tests should be tailored for each individual horse, particularly taking into account the presenting complaint and the race distance for that horse.
The authors would like to thank the referring veterinary surgeons and trainers of all cases.