Substantial variability exists in the interpretation of survey radiographs among equine veterinarians

Abstract

Background

Radiographs are a common diagnostic tool utilised during pre‐purchase examinations, yet differences surround their interpretation among equine veterinarians.

Objectives

(1) To determine veterinarians' subjective impressions of a spectrum of radiographic abnormalities regarding suitability for purchase; (2) To determine how veterinarians' years in practice, primary discipline or board certification status influence their subjective impression of radiographic abnormalities, and (3) To determine how horses' discipline, current level of work (working at intended level vs. future prospect) and buyer intent (sell vs. retain ownership) influence veterinarians' subsequent recommendation.

Study design

Worldwide electronic survey of veterinary professionals.

Methods

Four radiographic series of navicular bones, tarsi, and stifles, representing varying degrees of pathological change (normal, mild, moderate and marked) were presented. Respondents (n = 211) graded their level of concern from 1 (no concern) to 10 (very concerned) based on the radiographs and case scenarios presented.

Results

Overall, as the grade of pathological change increased for each anatomical location, so did the level of veterinarian concern, but a wide range of answers were submitted for all grades of pathology. Board certified practitioners had a higher level of concern over marked radiographic changes of the navicular bone for both western performance and English sport horses as well as prospect horses (mean = 7.2 vs. 6.1; P = 0.009; 95% CI for difference = [0.3, 2.0]). Board certified respondents had a lower level of concern over marked radiographic changes of the tarsus regardless of the performance status of the horse (mean = 7.1 vs. 8.0; P = 0.008; 95% CI for difference = [−1.5, −0.2]). Western performance horse practitioners had a lower level of concern over mild radiographic changes of the tarsus regardless of the performance status of the horse (mean = 3.3 vs. 4.2; P = 0.009; 95% CI for difference = [−1.6, −0.2]).

Main limitations

Only one example of each pathological grade per anatomical region was presented.

Conclusions

On survey radiographs, the perceived clinical relevance of radiographic abnormalities among equine practitioners differs widely, making purchase recommendations highly variable.

1. INTRODUCTION

In equine practice, pre‐purchase examinations (PPEs) and associated survey radiographs have played a crucial role in the buying, selling and financial negotiations of equine athletes across all disciplines. Regardless of the number of years a veterinarian has been practicing or the discipline of a horse, the equine veterinarian's role in the PPE is to identify medical and orthopaedic issues that may limit immediate or future performance. Aside from a thorough physical and musculoskeletal examination, radiographs are the most common diagnostic tool utilised during a PPE. ¹ Specifically, radiographs are used to characterise current or potential orthopaedic abnormalities to help assess future performance capabilities and suitability for purchase. ¹ , ² , ³ The veterinarian is regularly requested to speculate about a horse's potential to perform an intended discipline based on their subjective PPE findings, including radiographic interpretation. ³ Inherent to any subjective assessment, there is often ambiguity surrounding the interpretation of the clinical relevance of radiographic abnormalities within the PPE setting. There are many studies documenting the prevalence of radiographic abnormalities in various breeds and disciplines but fewer that objectively correlate radiographic findings and future performance. Disagreement among equine veterinarians regarding interpretation of PPE findings may arise due to differences in athletic expectations, or in the consequences of pathological changes, between different disciplines. Additionally, differences in interpretation may be influenced by the experience level of the practitioner, past clinical experiences, additional post graduate veterinary training, high monetary value of the horse being evaluated and/or potential legal disputes against PPE assessments. ⁴ , ⁵ , ⁶ , ⁷ , ⁸ Cumulatively, these differences make it very difficult to predict how another veterinarian will interpret the clinical relevance of survey radiographs when multiple reviews by different veterinarians are solicited by buyers.

The prevalence of radiographic changes based on presale or survey radiographs of several disciplines of horses, particularly Thoroughbred and Standardbred racehorses, have been well described. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ Reports regarding English sport horse survey radiographs have also been published. ¹ , ² , ¹⁴ Similar reports for western performance horses are lacking however, with only one previous report describing survey radiographic findings in cutting horses. ¹⁵ Contino et al. evaluated PPE repository radiographs of 458 yearling and 2‐year‐old cutting horse prospects and found that 89% of horses had radiographic abnormalities. ¹⁵ Out of those horses, the tarsi had the majority of abnormalities identified (69.4%), followed by the stifles (44.5%), hind and fore fetlocks (43.7% and 36.3%, respectively) and carpi (7.9%). ¹⁵ Front feet and/or navicular bones were not evaluated in that particular study. ¹⁵

The clinical relevance of various radiographic abnormalities has been reported to be highly subjective according to a veterinarian's experience level and primary sport of practice (e.g., English vs. western). ¹ , ² , ³ , ⁴ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ Furthermore, despite many previous reports for racehorses, ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ sport horses ¹ , ² , ¹⁴ and cutting horses, ¹⁵ , ¹⁷ we were unable to identify previous studies where a direct comparison of veterinarian subjective impressions of various radiographic findings according to experience level and discipline were made. Therefore, the objectives of this study were to: (1) determine subjective impressions regarding suitability for purchase of normal, mild, moderate and marked radiographic changes for various anatomical regions (navicular bones, tarsi and stifles); (2) determine how a veterinarian's experience level, primary discipline or board certification status influence their subjective impression of radiographic abnormalities, and (3) determine how a horse's discipline (English sport horse vs. western performance horse), current level of work (performing at intended level vs. future prospect) and buyer intent (sell vs. retain ownership) influenced the veterinarians' subsequent recommendation. We hypothesised that more experienced and board‐certified practitioners would perceive radiographic changes to be of less risk than that of less experienced or non‐board‐certified practitioners. Additionally, we hypothesised that radiographic abnormalities of horses already competing at the desired level would have lower perceived risk than prospect horses for all equine practitioners.

2. MATERIALS AND METHODS

2.1. Veterinary survey

A 41‐question survey of equine practitioners worldwide was performed using a web‐based survey (Surveymonkey.com©) (Text S1). The survey was advertised through various American Association of Equine Practitioners discussion boards, bimonthly member emails, equine practitioner social media platforms and through the authors' personal connections. Responses were collected over 50 days, from March to May 2022. All individual responses to survey questions were confidential and blinded to authors performing data analysis. Only three anatomical locations were selected for this study as they are the most commonly radiographed regions in a western performance horse PPE based on the authors' clinical experience. Additionally, a previous report found that the front feet and tarsi were the two areas most frequently radiographed in a PPE. ¹⁸ The limited regions were also intended to prevent survey fatigue for participants. The foot, specifically the navicular bone, the tarsi and the stifles were included. For the purpose of this study, the focus of the tarsi was the distal intertarsal (DIT) and tarsometatarsal (TMT) joints and, for the stifle, the medial femoral condyle. Images were originally obtained during clinical investigations and were selected from a medical record archived and one radiographic series for each degree of severity (normal, mild, moderate, and marked) of the three anatomical locations was included in the survey, resulting in a total of 12 series. All images were reviewed, graded for severity of pathology, and deemed acceptable for positioning and image resolution by an equine board‐certified radiologist. For each series of radiographs, respondents were blinded to the severity of the case and were provided the images in random order. All radiographs were provided as JPEG images to respondents. Furthermore, respondents were presented with the same three case scenarios for each series of radiographs.

The first scenario (scenario 1) was a mid‐career, Quarter Horse gelding used for western all‐around events that presents for PPE. The potential buyer would like to compete the horse at the same level of competition that it is now with the option for re‐sale in 2–3 years. The second scenario (scenario 2) was a mid‐career, Warmblood gelding used as an English sport horse that presents for PPE. The potential buyer would like to compete the horse at the same level of competition that it is now with the option for re‐sale in 2–3 years. The final scenario (scenario 3) was a prospect gelding that presents for PPE. The potential buyer would like to compete the horse at a higher level of competition than it is now with the option for re‐sale in 2–3 years. In all scenarios, it was stated that all physical exam findings were within normal limits, the horse was sound on both arena and a hard surface at exam, flexion tests were all insignificant, and no sensitivity to hoof testers was identified. These 3 scenarios for each of the 12 radiographic series resulted in a total of 36 questions to be answered by each respondent. For these 36 questions, the respondents were asked to grade their level of concern from 1 (no level of concern) to 10 (very concerned) based on the radiographs and case scenario presented.

Questions 37–41 were demographic questions asking the respondent how much of their practice was equine, how many years they had been in equine practice, if they were board‐certified, and if so, in which specialty(ies), the geographic location of their primary work (state, country, or province), and primary discipline of the horses in their practice. Respondents were asked to select from various ranges (<5 years, 5–10 years, 11–20 years or >20 years) their number of years in practice; for the purposes of this study, an experienced practitioner was defined as one who has practiced for more than 10 years. Level of experience was defined by the years in practice only and not number of cases or other contributing factors. Based on the response to listed board certification(s) only Diplomates of the American College of Veterinary Surgeons (DACVS) and/or of the American College of Veterinary Sports Medicine and Rehabilitation (DACVSMR) were included in the statistical analysis for board certification versus not.

An ‘English sport horse’ was defined as a horse performing in show hunters or any of the three Olympic equestrian disciplines (show jumping, dressage and/or eventing). ¹⁹ For the purposes of this study, a veterinarian that primarily practices on English sport horses was referred to as an English sport horse veterinarian. The term ‘western performance horse’ has evolved from the Quarter Horse breed and its many genetic lineages actively competing or intending to compete in various associations. ²⁰ The versatility of a western performance horse has generated many different disciplines under the broad term including, but not limited to reining, cutting, barrel racing, roping or all‐around horses. ²⁰ For the purposes of this study, a veterinarian that primarily services western performance horses was referred to as a western performance veterinarian. Both definitions were presented to survey respondents when asked what discipline of horses they practiced on most frequently.

2.2. Radiographic acquisition and inclusion

The radiographs used in this study were reviewed and collected in a retrospective manner from PPEs presented to a private practice between December 2020 and January 2022. The images were all acquired using the same type of digital radiography machine and system (Sound Sprint II Equine DR™) and were captured using the same technique for each anatomical location based off the system's technique chart (Sound® SmartDR Premier Technique Guideline). All patient and client identifying information was removed from the radiographs for the survey. The radiographs used in this study were taken from stock breeds (Quarter or Paint horses).

2.3. Navicular bone

A lateromedial, horizontal dorsopalmar, dorso 65° proximal–palmarodistal oblique (obtained using the upright pedal technique), and palmaro 45° proximal–palmarodistal oblique (navicular skyline) view were included for each navicular region series. Respondents were asked to evaluate the navicular bone. A total of four series representing normal, mild, moderate and marked pathological changes were included in the survey and presented in random order (Figures 1 and S1). As stated previously, the respondents were unaware of the various severity grades assigned to any of the series of radiographs.

FIGURE 1.

Palmaroproximal–palmarodistal oblique (navicular skyline) radiographic view of the navicular bone provided in the survey. (A) Normal. (B) Mild changes. (C) Moderate changes. (D) Marked changes.

2.4. Tarsus

A dorsoplantar, dorso 45° medial–plantarolateral oblique, lateromedial and dorso 45° lateral–plantaromedial oblique view were included for each tarsal series. For each tarsal series, respondents were specifically asked to focus on the DIT and TMT joints and cuboidal bones. A total of four radiographic series of the tarsus were provided in random order, representing normal, mild, moderate and marked pathological changes (Figures 2 and S1).

FIGURE 2.

Lateromedial radiographic view of the tarsus tarsi provided in the survey. (A) Normal. (B) Mild changes. (C) Moderate changes. (D) Marked changes.

2.5. Stifle

A caudocranial, caudal 45° lateral–craniomedial oblique and lateromedial view were included for each stifle series. For each stifle series, respondents were specifically asked to focus on the medial femoral condyle. A total of four stifles series were provided in random order, representing normal, mild, moderate and marked pathological changes (Figures 3 and S1).

FIGURE 3.

Caudal 45° lateral–craniomedial oblique stifle radiograph images used within survey. (A) Normal. (B) Mild changes. (C) Moderate changes. (D) Marked changes.

2.6. Data analysis

Summary statistics, including numerical answers and percentages, were calculated by the survey program. The raw means, standard deviations, medians, interquartile ranges (IQR) and ranges were reported in Tables 1, 2, 3. To reflect where the majority of the data lie, ranges excluding outliers were also reported. Here, outliers were defined as responses more than 1.5 IQR above the third quartile or more than 1.5 IQR below the first quartile, as well as survey questions that were unanswered. Note that all other data summaries and statistical analyses were still based on the full data (including outliers). A visualisation of the raw data is provided in Text S2. Levels of veterinarian concern were analysed using a linear mixed model. A Likert scale (1–10) to rank the responses from the veterinarians was utilised as previously described ²¹ and the intervals between grades were assumed to be equal. The anatomical region, grade of pathology and case scenario were included as fixed effects. Each individual respondent was included as a random effect to account for the correlation among responses from the same individual. To study the influence of a veterinarian's experience, board certification status and stated specialty discipline on response, these characteristics were included in the model as a fixed effect one‐at‐a‐time. The linear mixed model was used to test significance and construct confidence intervals of the effects. Moreover, following the main analysis, a stratified analysis within each anatomical region and each grade of pathology, and further within each case scenario, was performed to identify heterogeneous effects of a respondent's characteristics on the level of concern across different anatomical region, pathology grade, and case scenario. In other words, the model was first fitted to a subset of the questions associated with a specific anatomical region and pathology grade, and then to a further restricted subset of the data within a specific anatomical region, pathology grade, and case scenario. All statistical analyses, summary and stratified, were conducted using the lme4 package ²² within the R statistical software (The R Foundation©). ²³ The details of the model specification are provided in Text S2. A residual analysis (Text S2) demonstrated that the residuals were reasonably normally distributed and supported the validity of the linear mixed model. A P value of 0.05 was designated as the limit of statistical significance and 95% confidence intervals (CIs) were provided. Due to the exploratory nature of the analyses, no multiplicity adjustments were performed, and the P‐values should be interpreted for descriptive purposes. ²⁴

TABLE 1.

Survey respondents' level of concern for various survey radiographs of the navicular bone graded from 1 (no level of concern) to 10 (high level of concern).

	Navicular bone series
	Normal	Mild	Moderate	Severe
All responses	Range 1–10 (n = 211)	Range 1–10 (n = 211)	Range 1–10 (n = 211)	Range 1–10 (n = 211)
Excluding outliers	Range 1–6 (n = 205)	Range 1–9 (n = 211)1	Range 1–10 (n = 211)	Range 1–10 (n = 211)

	Normal			Mild			Moderate			Severe
	Mean ± SD	Med	IQR	Mean ± SD	Med	IQR	Mean ± SD	Med	IQR	Mean ± SD	Med	IQR
Responses
Overall	2.6 ± 1.9	2	2.3	3.8 ± 2.1	4	3	6.1 ± 2.3	6	3	6.3 ± 2.7	7	4
Experience
≤10 years (n = 65)	2.8 ± 2.1	2	3	4.0 ± 2.1	4	3	6.5 ± 2.2	7	3	6.1 ± 2.7	6	4
>10 years' (n = 146)	2.5 ± 1.9	2	2	3.8 ± 2.1	4	3	5.9 ± 2.4	6	4	6.5 ± 2.7	7	3
Board Certification
DACVS/DACVSMR (n = 41)	1.9* ± 1.4	1	1	3.4 ± 1.7	3	3	6.2 ± 2.3	7	3	7.2* ±2.3	8	3
Non DACVS/DACVSMR (n = 170)	2.8* ± 2.0	2	3	3.9 ± 2.2	4	3	6.1 ± 2.3	6	3	6.1* ±2.7	6	4
Discipline
Western performance (n = 37)	2.5 ± 1.8	2	2	3.4 ± 1.8	3.5	2.3	5.7 ± 2.4	6	4	6.5 ± 2.4	7	3
Non‐western performance (n = 174)	2.6 ± 1.9	2	3	3.9 ± 2.1	4	3	6.2 ± 2.3	6	3	6.3 ± 2.7	7	4
English performance (n = 120)	2.5 ± 1.9	2	2	3.8 ± 2.1	4	3	6.2 ± 2.3	6	3	6.4 ± 2.7	7	5
Non‐English performance (n = 91)	2.7 ± 2.0	2	2	3.9 ± 2.1	4	3	5.9 ± 2.4	6	4	6.3 ± 2.6	6.5	4

Note: Values reported are range, mean and respective standard deviation (SD), median (med) and interquartile range (IQR). Outliers were defined as a response more than 1.5 IQR above the third quartile or more than 1.5 IQR below the first quartile. Statistically significant (P < 0.05) comparisons are denoted by an asterisk. Note that the range with and without outliers is different while the number of respondents is the same. This occurred because the data is a summary across all three scenarios such that a respondent's answer could have been an outlier in one scenario (and thus the response excluded from the reported range) but not an outlier on another scenario, causing the respondent to still be counted in the final n.

TABLE 2.

Survey respondents' level of concern for various survey radiographs of the tarsi graded from 1 (no level of concern) to 10 (high level of concern).

	Tarsal series
	Normal	Mild	Moderate	Severe
All responses	Range 1–10 (n = 211)	Range 1–10 (n = 211)	Range 1–10 (n = 211)	Range 1–10 (n = 211)
Excluding outliers	Range 1–3 (n = 194)	Range 1–9 (n = 211)1	Range 1–10 (n = 211)	Range 3–10 (n = 209)

	Normal			Mild			Moderate			Severe
	Mean ± SD	Med	IQR	Mean ± SD	Med	IQR	Mean ± SD	Med	IQR	Mean ± SD	Med	IQR
Responses
Overall	2.0 ± 1.8	1	1	4.0 ± 2.1	4	3	5.1 ± 2.1	5	4	7.8 ± 2.2	8	3
Experience
≤10 years (n = 65)	1.8 ± 1.4	1	1	4.0 ± 2.1	3	3.3	5.1 ± 1.9	5	2	8.0 ± 2.0	8	3
>10 years' (n = 146)	2.1 ± 2.0	1	1	4.0 ± 2.1	4	3	5.1 ± 2.1	5	4	7.8 ± 2.4	8	4
Board Certification
DACVS/DACVSMR (n = 41)	1.7 ± 1.6	1	1	3.8 ± 1.8	4	2	5.3 ± 1.9	5	3	7.1* ±2.4	8	3
Non DACVS/DACVSMR (n = 170)	2.1 ± 1.9	1	1	4.1 ± 2.2	4	3.5	5.1 ± 2.1	5	4	8.0* ±2.1	8	3
Discipline
Western performance (n = 37)	1.7 ± 1.6	1	1	3.3* ±1.9	3	3	4.8 ± 2.1	5	3	7.3* ±2.4	8	3
Non‐western performance (n = 174)	2.0 ± 1.9	1	1	4.2* ±2.1	4	3	5.2 ± 2.0	5	3	8.0* ±2.1	8	3
English performance (n = 120)	1.9 ± 1.8	1	1	4.3* ±2.1	4	2	5.3 ± 2.1	5	3	7.9 ± 2.2	8	3
Non‐English performance (n = 91)	2.1 ± 1.9	1	1	3.7* ±2.1	3	3	4.9 ± 2.0	5	3	7.8 ± 2.1	8	3

Note: Values reported are range, mean and respective standard deviation (SD), median (med) and interquartile range (IQR). An outlier was defined as a response more than 1.5 IQR above the third quartile or more than 1.5 IQR below the first quartile. Statistically significant (P < 0.05) comparisons are denoted by an asterisk. Note that the range with and without outliers is different while the number of respondents is the same. This occurred because the data is a summary across all three scenarios such that a respondent's answer could have been an outlier in one scenario (and thus the response excluded from the reported range) but not an outlier on another scenario, causing the respondent to still be counted in the final n.

TABLE 3.

Survey respondents' level of concern for various survey radiographs of the stifle graded from 1 (no level of concern) to 10 (high level of concern).

	Stifle series
	Normal	Mild	Moderate	Severe
All responses	Range 1–10 (n = 211)	Range 1–10 (n = 211)	Range 1–10 (n = 211)	Range 1–10 (n = 211)
Excluding outliers	Range 1–3 (n = 189)	Range 1–10 (n = 211)	Range 1–10 (n = 211)	Range 2–10 (n = 209)

	Normal			Mild			Moderate			Severe
	Mean ± SD	Med	IQR	Mean ± SD	Med	IQR	Mean ± SD	Med	IQR	Mean ± SD	Med	IQR
Responses
Overall	1.9 ± 1.8	1	1	5.4 ± 2.3	6	3	4.4 ± 2.5	4	4	7.5 ± 2.2	8	3
Experience
≤10 years (n = 65)	2.2 ± 1.9	1	2	5.9* ±2.3	6	4	4.1 ± 2.5	4	3	7.8 ± 2.1	8	3
>10 years' (n = 146)	1.8 ± 1.8	1	1	5.3* ±2.3	5	4	4.5 ± 2.5	4	4	7.4 ± 2.2	8	3
Board Certification
DACVS/DACVSMR (n = 41)	1.6 ± 1.5	1	1	5.4 ± 2.4	5	3	4.9 ± 2.3	5	4	7.6 ± 2.2	8	2.3
Non DACVS/DACVSMR (n = 170)	2.0 ± 1.9	1	1	5.5 ± 2.3	6	3	4.2 ± 2.5	5	4	7.5 ± 2.2	8	3
Discipline
Western performance (n = 37)	2.5* ± 2.7	1	3	5.2 ± 2.4	5	4	4.5 ± 2.3	4	3	7.3 ± 2.3	8	3
Non‐western performance (n = 174)	1.8* ± 1.6	1	1	5.5 ± 2.3	6	3	4.3 ± 2.5	4	4	7.5 ± 2.2	8	4
English performance (n = 120)	1.8 ± 1.6	1	1	5.6 ± 2.3	6	3	4.4 ± 2.5	4	4	7.6 ± 2.2	8	4
Non‐English performance (n = 91)	2.1 ± 2.1	1	1	5.3 ± 2.2	5	3	4.3 ± 2.4	4	4	7.4 ± 2.2	8	3

Note: Values reported are range, mean and respective standard deviation (SD), median (med) and interquartile range (IQR). Outliers were defined as a response more than 1.5 IQR above the third quartile or more than 1.5 IQR below the first quartile. Statistically significant (P < 0.05) comparisons are denoted by an asterisk. Note that the range with and without outliers is different while the number of respondents is the same. This occurred because the data is a summary across all three scenarios such that a respondent's answer could have been an outlier in one scenario (and thus the response excluded from the reported range) but not an outlier on another scenario, causing the respondent to still be counted in the final n.

3. RESULTS

3.1. Respondent demographics

A total of 211 survey responses were received, 203 of which were fully completed. The majority of responses were received from across the United States (162/211; 76.8%), but also from practitioners in Canada (20/211; 9.5%) and Europe (16/211; 7.6%). Fourteen out of the 211 respondents did not specify their geographical location. Of the 211 respondents, 84 (39.8%) reported practicing veterinary medicine for over 20 years, 62 (29.4%) reported practicing for 11–20 years, 48 (22.7%) reported practicing for 5–10 years, and 17 (8.1%) reported practicing for less than 5 years (Table 4). The largest percentage of respondents (78.7%) indicated that their practice was equine‐exclusive and an additional 13.3% spent the majority of their time (>75%) dedicated to equine practice. The remaining respondents reported spending their time in mixed animal practices. More than half 56.9% (120/211) of respondents worked primarily on English sport horses while an additional 17.5% (37/211) of respondents worked primarily on western performance horses. All other categories of disciplines were <10% of the total responses and a few (<5%) stated they work on an equal amount of English sport and western performance horses. Regarding board certification status, 8.5% (18/211) respondents were DACVS, 6.6% (14/211) were DACVSMR and 4% (9/211) were diplomates of both specialties. No board‐certified imaging diplomates completed the survey. On average, the survey program reported it took respondents 17 min to complete.

TABLE 4.

Survey respondent demographics.

	% Total	(95% CI)	Total respondents (n = 211)
Type of practice
Exclusively equine	78.7%	(72.5, 84.0)	166
>75% Equine	13.3%	(9.0, 18.6)	28
Discipline
English sport horse	56.9%	(49.9, 63.7)	120
Western performance horse	17.5%	(12.7, 23.4)	37
Years in practice
>20	39.8%	(33.2, 46.8)	84
11–20	29.4%	(23.3, 36.0)	62
5–10	22.7%	(17.3, 29.0)	48
<5	8.1%	(4.8, 12.6)	17
Board Certification
DACVSMR only	6.6%	(3.7, 10.9)	14
DACVS only	8.5%	(5.1, 13.1)	18
Both DACVSMR/DACVS	4.3%	(2.0, 7.9)	9
Total Board Certified	19.4%	(14.3, 25.4)	41

When evaluated overall, survey responses for level of concern across anatomical regions and grades of pathology among practitioners were numerically different but not significantly different (Table 5). When the authors' report statistical values, it is implied that the demographic of focus is compared with the rest of the survey respondents. Some respondents answered with 0 instead of 1 as per the survey instructions which was considered for statistical analyses. It was extrapolated that generally all scenario 3 questions were answered with higher perceived risks grades compared with scenarios 1 and 2.

TABLE 5.

Survey respondents' level of concern, graded from 1 (no level of concern) to 10 (high level of concern), for radiographic series of all anatomical regions and grades of pathological change.

Responses	Level of concern
All respondents (n = 211)	4.8 ± 2.9
Experience
≤10 years' experience (n = 65)	4.9 ± 2.9
>10 years' experience (n = 146)	4.7 ± 2.9
Board Certification
DACVS and DACVSMR (n = 41)	4.7 ± 2.9
Non DACVS or DACVSMR (n = 170)	4.8 ± 2.9
Discipline
Western performance (n = 37)	4.6 ± 2.8
Non‐western performance (n = 174)	4.8 ± 2.9
English performance (n = 120)	4.8 ± 2.9
Non‐English performance (n = 91)	4.7 ± 2.8

Note: Values reported are means and respective standard deviation. Outcomes across anatomical regions and grades of pathology were numerically different but not significant. No statistically significant (P < 0.05) differences were identified in the overall comparison.

3.2. Navicular bone

Excluding outliers, the level of concern among respondents ranged from 1 to 6 for normal and 1 to 10 for marked radiographic changes (Table 1). Board‐certified (DACVS and DACVSMR only) equine practitioners had a statistically higher level of concern (mean = 7.2 vs. 6.1, respectively; P = 0.009; 95% CI for difference = [0.3, 2.0]) for marked radiographic changes of the navicular bone, but a statistically lower level of concern (mean = 1.9 vs. 2.8, respectively; P = 0.006; 95% CI for difference = [−1.5, −0.3]) for normal radiographic findings of the navicular bone regardless of the performance status of the horse (Table 1). Board‐certified equine practitioners had a statistically higher level of concern regarding marked radiographic changes of the navicular bone for both scenario 1 (mean = 6.8 vs. 5.7, respectively; P = 0.02; 95% CI for difference = [0.2, 1.9]) and scenario 2 (mean = 6.7 vs. 5.7, respectively; P = 0.03; 95% CI for difference = [0.1, 1.9]). Also, board‐certified equine practitioners had a statistically lower level of concern regarding normal radiographic findings of the navicular bone (mean = 1.8 vs. 2.6, respectively; P = 0.008; 95% CI for difference = [−1.4, −0.2]), (mean = 1.7 vs. 2.6, respectively; P = 0.008; 95% CI for difference = [−1.5, −0.2]) for the same two scenarios (scenario 1 and 2, respectively). Furthermore, board certified equine practitioners had a statistically higher level of concern (mean = 8.2 vs. 7.0, respectively; P = 0.006; 95% CI for difference = [0.4, 2.1]) for marked radiographic changes and a statistically lower level of concern (mean = 2.2 vs. 3.1, respectively; P = 0.01; 95% CI for difference = [−1.6, −0.2]) for normal radiographic findings of the navicular bone for a future prospect horse (scenario 3).

Regarding experience level, experienced practitioners had a statistically lower level of concern (mean = 6.7 vs. 7.4, respectively; P = 0.04; 95% CI for difference = [−1.4, −0.0]) for moderate radiographic changes of the navicular bone, specifically for a prospect horse.

There were no statistically significant differences in interpretation of radiographic changes of the navicular bone when comparing western performance horse or English sport horse practitioners to others.

3.3. Distal tarsal joints and cuboidal bones

Excluding outliers, the level of concern among respondents ranged from 1 to 3 for normal and 3 to 10 for marked radiographic changes (Table 2). Board‐certified equine practitioners had a statistically lower level of concern (mean = 7.1 vs. 8.0, respectively; P = 0.008; 95% CI for difference = [−1.5, −0.2]) for marked radiographic changes of the tarsus regardless the performance status of the horse (Table 2). Additionally, this same group of veterinarians had a statistically lower level of concern for marked radiographic changes of the tarsus for scenario 1 (mean = 6.4 vs. 7.5, respectively; P = 0.007; 95% CI for difference = [−1.8, −0.3]), scenario 2 (mean = 6.7 vs. 7.7, respectively; P = 0.01; 95% CI for difference = [−1.7, −0.2]) and scenario 3 (mean = 8.2 vs. 8.9, respectively; P = 0.03; 95% CI for difference = [−1.3, −0.1]).

Western performance horse practitioners had a statistically lower level of concern regarding mild (mean = 3.3 vs. 4.2, respectively; P = 0.009; 95% CI for difference = [−1.6, −0.2]) and marked (mean = 7.3 vs. 8.0, respectively; P = 0.05; 95% CI for difference = [−1.4, −0.0]) radiographic changes of the tarsus regardless the performance status of the horse (Table 2). Additionally, western performance horse practitioners had a statistically lower level of concern for mild radiographic changes of the tarsus in scenario 1 (mean = 2.8 vs. 3.5, respectively; P = 0.02; 95% CI for difference = [−1.4, −0.1]) and scenario 3 (mean = 3.9 vs. 5.2, respectively; P = 0.002; 95% CI for difference = [−2.1, −0.5]). Finally, western performance horse practitioners had a statistically lower level of concern (mean = 8.2 vs. 8.9, respectively; P = 0.03; 95% CI for difference = [−1.4, −0.1]) for marked radiographic changes of the tarsus in future prospect horses (scenario 3).

Conversely, English sport horse veterinarians had a statistically higher level of concern (mean = 4.3 vs. 3.7, respectively; P = 0.032; 95% CI for difference = [0.1, 1.1]) for mild radiographic changes of the tarsus regardless of the performance status of the horse (Table 2). Additionally, this same group of practitioners had a statistically higher level of concern for mild (mean = 5.4 vs. 4.4, respectively; P = 0.002; 95% CI for difference = [0.4, 1.6]) and moderate (mean = 6.4 vs. 5.7, respectively; P = 0.01; 95% CI for difference = [0.2, 0.8]) radiographic changes of the tarsus specifically for a future prospect horse (scenario 3).

There were no statistically significant differences when looking at experienced practitioners compared with those practitioners with fewer than 10 years' experience for radiographic findings of the distal tarsal joints and cuboidal bones.

3.4. Medial femoral condyle of stifle

Excluding outliers, the level of concern among respondents ranged from 1 to 3 for normal and 2 to 10 for marked radiographic changes (Table 3). Experienced equine practitioners had a statistically lower level of concern (mean = 5.3 vs. 5.9, respectively; P = 0.04; 95% CI for difference = [−1.2, −0.0]) regarding mild radiographic changes in the stifle (Table 3). Additionally, experienced practitioners had a statistically lower level of concern for scenario 1 (mean = 4.6 vs. 5.3, respectively; P = 0.03; 95% CI for difference = [−1.3, −0.1]) and scenario 3 (mean = 6.2 vs. 7.0, respectively; P = 0.03; 95% CI for difference = [−1.4, −0.1]) when considering mild radiographic changes of the stifle.

Western performance horse veterinarians had a statistically higher level of concern (mean = 2.5 vs. 1.8, respectively; P = 0.028; 95% CI for difference = [0.1, 1.4]) for normal radiographic findings of the stifle compared with all other practitioners for all three scenarios (Table 3). This same group of practitioners had a statistically higher level of concern compared with English sport horse practitioners regarding normal radiographic findings of the stifle for both a western (mean = 2.3 vs. 1.6, respectively; P = 0.02; 95% CI for difference = [0.1, 1.3]) or an English sport horse (mean values = 2.4 vs. 1.7, respectively; P = 0.04; 95% CI for difference = [0.1, 1.3]) performing at its expected level (scenarios 1 and 2). Furthermore, the same group of practitioners had a statistically higher level of concern (mean = 2.9 vs. 2.1, respectively; P = 0.04; 95% CI for difference = [0.0, 1.6]) for normal radiographic findings of the stifle for a future prospect horse (scenario 3).

4. DISCUSSION

This study, to our knowledge, is the first to objectively describe veterinarians' perception of survey (PPE) radiographs and the factors that may influence them. In general, except for the mild and moderate radiographic changes of the stifle, as the grade of pathological abnormality increased for each anatomical location, so did the level of concern for practitioners regarding purchase suitability. However, the level of concern ranges among practitioners also varied based on evaluated demographics making universal interpretation of the same set of radiographs inconsistent. This known discrepancy among veterinarians' interpretations of the same radiographs has substantial implications for horses that are being purchased for resale. While many statistically significant results were identified among the various demographics of practitioners that participated in the survey, the differences are not necessarily clinically significant. Therefore, the authors will only discuss those statistically significant differences that were at least 10% different (IE a difference of 1 point or more on the 10‐point grading scale).

It should be noted that a wide range of answers were submitted for all grades of pathological change which underscores the variability in radiographic interpretation that exists among equine veterinarians. For example, some of the questions were answered with scores ranging from 1 to 10. Thus, for the same exact set of radiographs, some veterinarians had no concern over purchase of the horse while the other veterinarians had the highest level of concern. This information may make it very difficult to predict what another veterinarian will conclude on the same series of radiographs.

One of the largest discrepancies between respondents' level of concern surrounded navicular bone radiographs with either normal radiographic findings or marked pathological changes. The range of responses for a normal navicular bone radiograph series was 1–6 and the range for marked navicular pathological change was 1–10. It is not known if this discrepancy suggests that some practitioners are not concerned about marked navicular bone pathological changes in terms of soundness or if the radiographs were not interpreted as normal versus marked, respectively. Additionally, the level of concern for marked pathological changes of the navicular bone was lower than the mean level of concern for marked pathological changes of both tarsi and stifles. It remains unknown whether marked navicular bone changes are perceived as less performance limiting than marked changes of the tarsi and stifles, or if the radiographic abnormalities were under interpreted. Another potential explanation is that only one standard projection of the palmaroproximal–palmarodistal oblique view of the navicular bone was provided in the survey. A recently published study that assessed detection of flexor cortical lysis found that an additional palmaroproximal–palmarodistal oblique projection improved the accuracy of lesion recognition. ²⁵ Similar to the study herein, investigators also surveyed veterinarians and veterinary students of various experience levels, but found an increase in confidence level in interpretation with additional views. ²⁵ Presentation of multiple palmaroproximal–palmarodistal oblique projections of the navicular bone may have improved lesion categorisation in the current study, as the marked navicular bone series highlighted a flexor cortical erosion of the navicular bone.

Board certified equine practitioners (DACVS and DACVSMR) had a statistically higher level of concern regarding marked radiographic changes of the navicular bone and a statistically lower level of concern for normal radiographic changes of the navicular bone for all scenarios. A higher level of concern surrounding marked pathological changes of the navicular bone from board certified respondents is consistent with previous reports that higher grades of navicular bone pathological changes are commonly career/performance limiting. ³ , ¹⁶ In a retrospective evaluation of equine pre‐purchase evaluations, Van Hoogmoed et al. reported that higher pathological changes of the navicular bone were associated with lameness. ³ Advanced imaging of the equine digit has demonstrated that radiographs are unable to depict all pathological changes present. Additionally, when advanced navicular bone pathological change is present on radiographs, a soft tissue component is often involved as well, ²⁶ which may risk the soundness and performance capabilities of a horse. Therefore, greater pathological change(s) of the navicular bone identified on radiographs may lead practitioners to a higher degree of concern regarding the athletic potential or future soundness of the horse.

Comparing the tarsus and stifle to the navicular bone responses, a much narrower respondent range was reported for level of concern regarding normal tarsi and stifle radiographs. The reported range for normal tarsi and normal stifles was 1–3 when outliers were excluded, compared with 1–6 for navicular bones. This may suggest that practitioners are more comfortable interpreting normal tarsi and stifle radiographs as such and/or may suggest that practitioners have less concern in general regarding pathological changes of the tarsus and stifle. In the same study conducted by Van Hoogmoed et al., in contrast to navicular bones that had higher correlation between pathological changes and lameness, higher pathological changes in the tarsus were less likely to be associated with lameness. ³ For tarsal radiographs, the biggest difference between respondents was between the practitioners in different disciplines. Western performance horse practitioners had a lower level of concern regarding mild radiographic changes of the tarsus in prospect horses compared with all other practitioners. Conversely, sport horse veterinarians compared with all other practitioners had higher levels of concerns regarding mild radiographic changes of the tarsus regardless of career level or prospect status of the horse. Murray et al. reports a high incidence of tarsal injury in dressage horses which could be due to increased tarsal loading and joint compression in the collected paces predisposing dressage horses to pathological tarsal changes. ²⁷ This could be one reason for sport horse veterinarians expressing a higher level of concern over pathological changes of the tarsi. Barrett et al. reported that the presence of mild distal tarsal osteophytes was correlated with lower mean earnings in 4‐year‐old cutting horses, ¹⁷ yet Western performance horse veterinarians were less concerned about mild distal tarsal joint pathology. This may suggest that veterinarians do not perceive these changes as performance limiting and/or that these horses can be sufficiently managed clinically.

The range of responses for level of concern regarding the stifle radiographs was similar to the tarsus. Level of concern over normal stifle radiographs ranged from 1 to 3 while marked stifle radiographs ranged from 2 to 10. For the stifle, none of the differences between demographics were considered clinically significant (all differences were <10%). Due to the frequency at which western performance horses suffer from stifle disease, it was surprising that western practitioners did not show a more substantial level of concern compared with other practitioners. It has been reported that stifle disease is a frequent source of lameness in Quarter Horses especially those within western performance disciplines, with the medial femorotibial joint compartment being the most commonly affected. ²⁸ A retrospective review on lameness localisation of western performance horses reported 9% of lamenesses were localised to the stifle, which was the third most common location followed by the distal forelimb and distal tarsus/proximal metatarsal regions. ²⁰ Therefore, more emphasis and scrutiny is typically placed on hindlimb joints in western disciplined horses. In contrast, Barrett et al. reported radiographic abnormalities of the medial femoral condyle were not significantly associated with performance outcome in cutting horses despite the authors' hypothesising moderate to marked lesions of the medial femoral condyle would be associated with a poorer performance outcome. ¹⁷ Additionally, a study of radiographic findings in 2‐year‐old Thoroughbreds at in‐training sales reported that none of the stifle joint lesions were associated with a decreased likelihood of racing, earning money, or racing in a stakes race compared with horses that had no radiographic lesion. ²⁹ Similar to radiographic interpretation of the navicular bone region, the complexity of soft tissues that support the stifle joint (e.g., meniscal instability) do not always manifest as radiographically apparent abnormalities.

It should be noted that respondents had a higher level of concern over the series of stifle radiographs that were categorised as mild (Text S1, Images 11.1–11.3) as compared with the stifle radiographs that were categorised as moderate (Text S1, Images 9.1–9.3). Retrospective review of the images demonstrates that the small subchondral bone defect in the medial femoral condyle is more visible in the caudocranial view on the mild series compared with the moderate series. However, when comparing the caudolateral–dorsomedial oblique projection, the subchondral bone defect is clearly larger in the moderate series compared with the mild series. While it cannot be proven based on the information obtained from the survey, the authors hypothesise that respondents utilised primarily the caudocranial projection to evaluate the medial femorotibial joint and likely were interpreting the mild series as having more pathological change compared with the moderate series. Regardless, ultimately, respondents had fairly similar levels of concern over the mild and moderate stifle series and, as would be expected, had the lowest level of concern over the normal stifle series and the highest level of concern over the marked stifle series.

Furthermore, consistent with our hypothesis, the survey responses suggested that more experienced practitioners were often more lenient of radiographic abnormalities than less experienced practitioners. This may be attributed to less experienced veterinarians not being as comfortable performing such examinations unless they have been exposed to a high number of them during their clinical practice. In the authors' experience, it takes time and/or a large caseload to gauge what is ‘normal’ and/or ‘acceptable’ for both imaging findings and the musculoskeletal exam. Therefore, it was not surprising that experienced practitioners tended to express less concern over mild or even moderate radiographic changes. The authors herein acknowledge that experience in this study was only measured based on years in practice and not caseload of the practitioner.

Likewise, the authors' hypothesised that board certified equine practitioners would perceive radiographic abnormalities to be of less risk as compared with the perception of risk assigned by non‐board‐certified practitioners. However, this was not supported by the results of the navicular bone series in which board‐certified practitioners responded with higher concern over marked radiographic changes of the navicular bone.

The authors' hypothesis that radiographic abnormalities of horses already competing at the desired level would have lower perceived risk than prospect horses was supported by the results of this study. The general response to all scenario 3 (future prospect horse) questions were answered with higher perceived risks grades compared with scenarios 1 and 2 (western and English sport horse performing at its expected level).

As reported in the results, average survey completion time was 17 minutes which does not allow for a long time to review 12 sets of radiographs (Figure S1) and answer the corresponding questions. It is possible that if respondents felt rushed that it could have contributed to the large amount of variability among practitioners' answers. Conversely, when under time constraints respondents likely answered with their immediate ‘gut’ response which is often reflective of their actual opinion.

The authors acknowledge that there were several limitations to this study. The findings of this study may have been further solidified had more survey responses been collected. To keep the survey reasonably feasible, only a few anatomical areas were evaluated and analysed. Future studies with the same study design and scenarios focusing on different pathological changes of the same regions, or different anatomical regions of the horse could provide additional insight. Additionally, only 3 or 4 views were provided of each anatomical region. Some practices or individual practitioners may perform additional radiographic views, especially when concerned about pathological changes of a certain region.

Another limitation to the study is that only one example of each grade of pathological change for each anatomical region was presented. Additionally, respondents were blinded to the grade assigned by the equine radiologist to each series. It is not known how participants would have responded had they been given the radiologist's assigned pathological grade for each series of radiographs. Despite this, the presented study scenario more closely mimics the reality of PPEs in which survey radiographs do not come with a pre‐assigned grade. A radiologist was used to screen the inclusion radiographs to ensure an appropriate range of changes were presented that spanned from normal to marked. The authors grant that the categorisations were assigned by one radiologist and should not be interpreted as the gold standard. The goal of this study was not to determine accuracy or precision of the respondents' interpretation of the radiographs compared with an imaging diplomate, but rather was to get their opinion within the PPE setting. Finally, all radiographs used in this study were taken from stock breeds (Quarter or Paint horses) but should still be applicable to any horse and are representative of pathological changes seen in horses presented for a PPE.

Furthermore, all images in the survey were provided in JPEG format which did not allow for resolution, size or contrast manipulation and thus may have limited the interpretation experience. Despite this, the images provided to all respondents were identical so may have actually served to more clearly elucidate the variation in practitioners' interpretation.

In conclusion, our results demonstrate that practitioners have higher purchase concern as the level of radiographic pathological changes of the navicular region, tarsi and stifles increase. Statistically significant differences were found between respondents according to board certification status and discipline. These findings highlight the variability in radiographic interpretation that exists among equine veterinarians. On survey radiographs, the perceived clinical relevance of radiographic abnormalities among equine practitioners differs widely, making purchase recommendations and future re‐sale potential for buyers highly variable. This brings ambiguity surrounding the clinical relevance of interpreted radiographic abnormalities within a PPE setting. Further studies that attempt to correlate radiographic findings with performance in all disciplines are warranted.

AUTHOR CONTRIBUTIONS

Initial study was conceptualised by Erin K. Contino and further refined by Myra F. Barrett, Angie M. Esselman, David D. Frisbie and Sherry A. Johnson. Data was acquired by Myra F. Barrett, Erin K. Contino, Angie M. Esselman, David D. Frisbie and Sherry A. Johnson. Data analysis and interpretation was done by Myra F. Barrett, Erin K. Contino, Angie M. Esselman, David D. Frisbie, Sherry A. Johnson and Tianjian Zhou. All authors contributed to the drafting and editing of the manuscript and gave their final approval. Angie M. Esselman had full access to all data and takes responsibility for the integrity of the data and accuracy of the data analysis.

FUNDING INFORMATION

There are no sources of funding to disclose.

CONFLICT OF INTEREST STATEMENT

No competing interests have been declared.

DATA INTEGRITY STATEMENT

No new data were created or analysed in this study.

ETHICAL ANIMAL RESEARCH

Ethical committee oversight is not required by this journal; radiographs were obtained for clinical purposes and participants were veterinarians aware of the study design and goals.

INFORMED CONSENT

Participation in the survey was taken as consent for inclusion in the study.

PEER REVIEW

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1111/evj.14045.

Supporting information

Figure S1. PPE radiographs.

Text S1. Questionnaire.

Text S2. Statistics supplement.

Esselman AM, Johnson SA, Frisbie DD, Barrett MF, Zhou T, Contino EK. Substantial variability exists in the interpretation of survey radiographs among equine veterinarians. Equine Vet J. 2025;57(1):169–182. 10.1111/evj.14045

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.