There has been a long-standing discussion amongst many equine veterinarians about the confidence they feel using the naked eye to evaluate and assess lame horses. Many feel certain that they have a keen eye and can discern even subtle gait asymmetries, and think that there is little need for objective assessment using inertial sensor modalities.
A recent webinar presented by Maarten Oosterlinck, DVM, PhD, DECVSMR, DECVS and Sandra Starke, BSc, MSc, PhD cited interesting data on how such self-assessed confidence in visual assessment may not be all it seems.
At the start of the discussion, Oosterlinck notes that it is difficult to develop a gold standard for subjective evaluation – in many cases, it boils down to a range of opinions. For more obvious cases of gait asymmetry diagnostic of lameness, he remarks that assessors can be 80-90% accurate. However, with more subtle lameness, the percentage of correct assessment decreases substantially to inadequate diagnostic accuracy. The question therefore becomes what constitutes ‘obvious’ and ‘subtle’ lameness? Respectively, the amount of gait asymmetry may mark a threshold beyond which the human eye is reliable in noticing movement abnormalities.
Limitations to Assessment
Certain characteristics of using visual evaluation for lameness pose limitations. Resolution is a primary concern with both temporal and spatial limitations. If the movement frequency occurs too quickly for the human eye to observe, the observer cannot see it – this is a temporal limitation. If there is only a minimal amount of change in gait asymmetry, similarly, it is difficult to see visually – this is a spatial limitation. The head of a sound horse at trot moves several centimeters up and down twice per stride, which is easy to see, remarks Oosterlinck. A lame horse may show a difference between the right and left stride with the head movement modified by only a couple of millimeters, which can be difficult to discern. Stride-to-stride variation, especially on uneven ground, can further complicate things.
Other effects that produce artifact or limit observation have to do with the lighting conditions under which the exam is conducted, and also the distance the horse is trotted away and how acute the observer’s vision. Both lighting and distance can influence the ability to observe small changes in vertical displacement of the horse’s head.
Oosterlinck mentioned another pitfall based on the “concept” of lameness. For example, a rider may have a “feeling” that the horse is “off.” However the rider’s interpretation is often a lameness in a different limb identified as lame after veterinary examination. He also points out that there is variability between observers even if experienced at lameness assessment. And, different grading systems are used to describe lameness. Normal human bias also can occur, as for example a veterinarian’s assessment after intervention with diagnostic nerve blocks. The time of day or day the vet performs the procedure may influence the assessment – a workup on a Friday afternoon or Saturday morning may find the vet declaring a positive response from a nerve block in a subconscious desire to move the exam along.
What are We Looking At?
For the forelimb, a lower position of the head when the sound limb contacts the ground compared to the lame limb is considered an indication of lameness, and most often corresponds to an impact lameness. An exception, which is more rare, is an extreme push-off lameness where there is only a difference in the high positions of the head.
For the hindlimb, most veterinarians are taught to look at total movement amplitude of the tuber coxae (left vs right), which increases on the lame side. Some caution is necessary in assessment since movement of the tuber coxae is affected by conformation, muscle asymmetry, or an old injury such as fracture of the tuber coxae.
Because the sacrum is on the midline, its vertical displacement is not as affected by conformational asymmetries. However, because the amplitude of sacral movement is smaller than tuber coxae movement, it can be harder for the eye to discern sacral movement.
Competence in Equine Lameness Recognition
A recent study had students evaluate videos of horses trotting on a straight line out and back, and on a lunge line to determine if the horse is lame or not lame, and if lame, then whether it is left or right limb. The more experienced students achieved a higher percentage of correct classifications. [Starke S. and May S. Veterinary student competence in equine lameness recognition and assessment: A mixed methods study. Veterinary Record 2017, 181:168; doi: 10.1136/vr.104245.].
The study also evaluated what landmarks on the horse the students were looking at in assessing lameness by using a computer to track students’ eye motions while watching horse movement on a straight line. Oosterlinck explains that the inexperienced observers looked at everything – topline, legs, feet – and therefore were confused by too many variables, many of them not being a reliable indicator of lameness. The experienced observers looked mostly at the horse’s topline, i.e. head and pelvis. Focusing more on this one area of anatomy likely increased the percentage of correct observations, as it is the most robust indicator for lameness: it is directly proportional to the changes in vertical force taken by each limb as the horse unloads the painful limb.
Yet even with experience, correct assessment of which limb was lame occurred only about 75% of the time. Correct identification that a horse is sound only occurred 50% of the time. There is a lot of room for improvement, says Oosterlinck.
A second recent study compared accuracy in visual lameness assessment with veterinary expertise and experience. [Starke S. and Oosterlinck M. Reliability of equine visual lameness classification as a function of expertise, lameness severity, and rater confidence. Vet Record Sept 2018, 184 (2); doi; 10.1136/vr.105058]
Near-realistic animations of sound and lame horses were presented to observers using six sound horses (3 shown from the front, 3 from the rear) and 12 horses with gait asymmetry ranging from 10 to 60 percent (6 forelimb lame horses shown from the front and 6 hindlimb lame horses shown from the rear). A total of 89 attendees at a veterinary seminar were invited to participate in a live scoring session, with veterinarians of various levels of expertise. They were grouped by experience and amount of caseload, with an average of 25 respondents viewing and assessing each clip. They were asked to determine if the horse is sound or lame, and to assign which limb is lame. For both fore and hindlimbs, it was more difficult for the observers to assign the correct limb as lame compared to deciding whether or not the horse is lame. The more subtle the asymmetry, the greater the variation in assessment.
The approximated detection threshold (at least 50% of participants correct) for identifying whether the horse was lame or not lame was around 25% asymmetry for the forelimb and 15% asymmetry for the hindlimb.
Importantly, assessment of hindlimb lameness resulted in a large proportion of perfectly symmetrical horses being mistaken for lame, skewing the detection threshold for hindlimb lameness.
The approximated detection threshold for also identifying the correct limb as lame was around 35% asymmetry.
“You will get better with experience” is often stated as veterinarians begin evaluating lameness in horses. The study proves otherwise, as no systematic effect of expertise on assessment accuracy was found. With regard to caseload, there is a significant difference and improvement of accurately identifying a lameness if a practitioner sees more than 10 cases per week, but this applies only for 60% gait asymmetry.
Declaring a horse "sound” turns out not to be an easy task despite a practitioner’s years of experience and caseload numbers. Experience does not significantly influence a vet’s ability to correctly declare a horse sound; instead, the observer often sees lameness where there was none. Correct identification by the viewers of a sound horse occurred in 75% for forelimbs, and only 30% for hindlimbs – many sound horses were incorrectly identified as lame in the hindlimbs. The study summarized, “Visual gait assessment may overall be unlikely to reliably differentiate between sound and mildly lame horses irrespective of an assessor’s background.”
Looking at many horses for lameness evaluation also does not necessarily improve one’s skill level or accuracy. Oosterlinck comments that this may be especially true for a solo practitioner with no-one else present to “adapt” an opinion. With years of experience, it is possible to become more precise, i.e. perform a repeatable assessment, however practitioners may simply be repeatably wrong. Accuracy (identifying close to a true value) can only improve if there is some level of comparison of one’s assessment – this can be accomplished through case discussion with a colleague or by using an objective lameness measurement system.
Errors in Assessing Lameness
The latter study also proved that a practitioner’s self-rated confidence about his or her conclusions is not well correlated to correctness for either forelimb or hindlimb lameness assessment. Evaluation of the most common errors showed that most struggled to recognize the presence of lameness in the forelimb but with assigning lameness to the correct limb in hindlimb lameness.
Better opportunities for self-assessment and deliberate practice can help people develop perceptual skills and become aware of their own abilities and limitations. A website – www.lamenesstrainer.com – can help with this.
At least 70% of sport horses are considered by their owners to be “normal,” yet have significant measured movement asymmetry. Similar findings were noted in a study using a comprehensive subjective evaluation. Potential problems are often beyond an owner’s ability to observe. Technology such as objective measurement devices (inertial sensors) can help measure and track such subtle movement asymmetry over time and to evaluate location and degree of lameness.
Equine biomechanics researcher Dr. Sandra Starke notes that it is important for a veterinarian to differentiate normal versus abnormal gaits. Incorrect classification of either is equally impactful. If a sound horse is declared lame, this has a financial impact on the owner regarding future sale of the horse as well as the expense of unnecessary diagnostic workups by a veterinarian for a non-existent problem. Conversely, Starke adds that overlooking early pathology can have a profound impact on a horse’s performance and athletic longevity by missing out on addressing a problem as early as possible.
Oosterlinck considers the inertial sensor system as an extra “colleague” that gives another, but “objective”, point of view. The inertial sensor system improves recognition of both subtle and compensatory lameness, which in turn improves performing diagnostic blocks on the correctly assigned lame limb. Most likely, many of the so-called “mystery lameness” that “cannot be blocked” are, in fact, compensatory movements by a primary lameness in another limb. In addition, a major advantage of a sensor system is the quantification of partial improvements after diagnostic blocks, which is a very common issue.