Upper body movement analysis of multiple limb asymmetry in 367 clinically lame horses

Background: Compensatory lameness is common in horses and evaluation can be challenging. Objectives: To investigate patterns of compensatory movements in clinical cases with fore- or hindlimb lameness before and after diagnostic analgesia. Study design: Retrospective clinical study. Methods: Multiple limb lameness of 367 horses was characterised by type (push-off, impact or mixed), limb (fore- or hindlimb in predominant lameness) and side (ipsi- or contralateral in concurrent lameness) using a body-mounted inertial sensor (BMIS). Diagnostic analgesia was performed until the percentage improvement of the vector sum in forelimb lameness and the mean difference of the maximum or minimum pel vic height (PD max or PD min ) in hindlimb lameness was ≥50%. Linear mixed model and post-estimation of effects were performed by contrast command with multiple comparisons adjusted by Bonferroni method. Correlation of pre- and post-analgesia of all head and pelvis asymmetry parameters was tested with Spearman's rank correlation. Results: Improvement in vector sum per mm after diagnostic analgesia in forelimb impact lameness positively correlated with decrease in PDmax in contralateral mixed lameness (0.187 mm, r = .58, P < .05). Improvement in PD min per mm after diagnostic analgesia in hindlimb mixed and PD max in hindlimb push-off lameness decreased vector sum in ipsilateral forelimb impact lameness by 0.570 and 0.696 mm, respectively ( P < .05), with no positive correlation. diagnostic analgesia in the affected forelimb and, vertical head movement asymmetry, due to compensation of a predominant hindlimb lameness, decreases after positive diagnostic analgesia in the affected hindlimb.


| INTRODUC TI ON
Multiple limb lameness is a common finding during orthopaedic examination of horses. Although lameness can be present independently in more than one limb, compensatory movements can be misinterpreted as a true lameness. To achieve definitive diagnosis and initiate therapy, it is important to identify the site of the primary lameness. [1][2][3][4] Compensatory movement is caused by weight-shifting from the primary painful limb. In consequence, there is convincing evidence of a "compensatory" or "false" lameness if it improves simultaneously with positive analgesia of the predominant lameness which can be called "primary" or "true" accordingly. 1,5 Patterns of compensatory movements have been comprehensively evaluated in experimental studies. 1,6-10 In a small number of horses with naturally occurring forelimb lameness, maximum pelvic height difference decreased after diagnostic analgesia in horses with contralateral compensatory hindlimb lameness, 11 whereas head movement asymmetry decreased significantly in horses with hindlimb lameness and ipsilateral compensatory forelimb lameness. 12 The type of the primary and associated compensatory lameness, namely, impact (during weight-bearing phase of the stride), push-off and mixed (impact and push-off, occurring at different moments of the stride cycle) 13,14 has not been investigated in detail. A body-mounted inertial sensor system, BMIS (Lameness locator ™ ), has been used previously to objectively assess lameness, effects of diagnostic analgesia and patterns of compensatory lameness. 11,[15][16][17][18] The aim of this study was to objectively evaluate compensatory patterns of primary fore-or hindlimb lameness in a large number of clinical cases, characterise lameness by the phase of the stride and analyse the effect before and after diagnostic analgesia. We hypothesised, (I), vertical pelvic movement asymmetry, due to compensation of a predominant forelimb lameness, decreases after positive diagnostic analgesia in the affected forelimb and, (II), vertical head movement asymmetry, due to compensation of a predominant hindlimb lameness, decreases after positive diagnostic analgesia in the affected hindlimb.

| Animals
Data were analysed retrospectively from clinical cases presented to the Equine Clinic, Freie Universität Berlin, Germany, from April 2011 to December 2015 for lameness investigation. Horses were included in the study, if (1) evaluation in trot could be performed with the initial body-mounted inertial sensor system (BMIS), (2) two or more limbs had asymmetry values calculated by the BMIS equipment equal or higher than the threshold values described below. In an asymmetric horse, the threshold of the absolute mean is greater than 8.5 mm for vector sum, 6 mm for HD min , HD max and 3 mm for PD min , PD max . These thresholds might represent at least weak evidence lameness, 19,20 (3) the percentage improvement of the vector sum in forelimb lameness and the mean difference of the maximum or minimum pelvic height (PD max or PD min ) in hindlimb lameness were more than 50% and (4) the asymmetry did not switch to another limb following diagnostic analgesia. Horses were excluded from the analysis if predominant lameness was not definable (ie BMIS did not identify a predominant lameness). Inclusion and exclusion criteria are summarised in a flow chart ( Figure 1).

| Lameness evaluation
Lameness examination was performed subjectively by a veterinarian specialising in orthopaedics and objectively with a BMIS (LAMENESS LOCATOR ™ : EQUINOSIS ® ) system. 16,17,21 Horses were trotted in a straight line over a concrete surface and at least 25 strides were included in each data trial.
The vertical acceleration of the head was recorded and the mean difference in maximum head height (HD max ) and the mean difference in minimum head height (HD min ) during the stance phase of the left and right limbs. The displacement differences were calculated in millimetres. The vector sum was calculated as a measure of head movement asymmetry. 20 In hindlimbs, the mean difference in maximum pelvic height (PD max ) and the minimum pelvic height (PD min ) during the stance phase of the left and right limb were calculated accordingly. 20 The sensor on the right forelimb contains a gyroscope and is used for step division. The side, type and grade of lameness were defined by positive and negative values of head or pelvic height differences respectively. 20 The predominant lameness was determined by an algorithm combining an evidence score The percentage improvement after diagnostic analgesia in both fore-and hindlimb was calculated as: [(Parameter before analgesia−Parameter after analgesia)/Parameter before analgesia]. 11,22 Diagnostic analgesia was performed until it was considered positive Abbreviations: Capital letter -F, predominant forelimb lameness; C, concurrent contralateral lameness; I, concurrent ipsitateral lameness; Lowercase letter -i, impact; p, push-off; m, mixed lameness; PD max , maximum pelvic height difference; PD min , minimum pelvic height difference; HD min , minimum head height difference a Positive correlation (tested by Spearman's rank analysis) of improvement of vector sum and PD max b Positive correlation (tested by Spearman's rank analysis) of percentage improvement between vector sum and PD min (p < 0.05).
c Analysis the effect of before and after diagnostic analgesia on asymmetry variables with adjusted covariates (sex, breed, age) by Linear Mixed model.
with the percentage improvement of vector sum in forelimb lameness and PD max or PD min in hindlimb lameness were more than 50%. 20

| RE SULTS
The inclusion criteria were met by 367 horses of which 199 had predominant forelimb and 168 had predominant hindlimb lameness.
Cases included 159 females, 32 males and 176 geldings with a mean age of 11 years (SD 5.1 years). Warmblood horses represented 68% of the study population and there was a range of other breeds.
Predominant forelimb with concurrent hindlimb lameness was seen in 15 patterns ( Table 1). The most common type in forelimbs was impact lameness (94%), with the concurrent hindlimb lameness found contralaterally (54%) and these are described in more detail in Table 1.
After diagnostic analgesia of the forelimb lameness, the PD min of concurrent hindlimb lameness decreased significantly in all three patterns (FiC, FiI and FiIC; P < .05), PD max significantly decreased in the FiC group (P < .001).  (Table 1, Table S1). The change in vector sum and PD max was positively correlated in FiCm (r = .58, P = .007).

| The percentage improvement of fore-and concurrent hindlimb lameness
After a positive analgesia of forelimb lameness, the percentage improvement was >50% of PD max in groups FiCp, FiCm and FiIm, and of PD min in all groups were less than 50% improvement (Table 1)

| Predominant hindlimb lameness with the type of concurrent forelimb lameness
When the type of concurrent forelimb lameness was considered, the HD max , HD min and vector sum of compensatory forelimb TA B L E 2 Comparison of kinematic variables following hindlimb analgesia in 168 horses with predominant hindlimb lameness and concurrent lameness in other limbs

Group
No. Abbreviations: Capital letter -H, predominant hindlimb lameness; C, concurrent contralateral lameness; I, concurrent ipsitateral lameness; Lowercase letter -i, impact; p, push off; m, mixed lameness; PD max , maximum pelvic height difference; PD min , minimum pelvic height difference; HD min , minimum head height difference; NA, not applicable. a A positive correlation (tested by Spearman's rank analysis) of percentage improvement between PD max and HD min .

Comparison of variables (mm) between before -after analgesia (P-value) c following hindlimb analgesia
b Positive correlation (tested by Spearman's rank analysis) of improvement of PD max and HD min (P < .05).
c Analysis of the effect of before and after diagnostic analgesia on asymmetry variables with adjusted covariates (sex, breed, age) by Linear Mixed model. lameness decreased significantly in HmIi ( Table 2, Table S3).
HD max only decreased in HiCi and HpIp. HD min was significantly decreased in HiIi, HpIi, HmIi and HmIp. There was a significant positive correlation after hindlimb diagnostic analgesia between HD min and the decrease in PD max ( Table 2, r = .44, P < .05) in HmIp.
No further correlations were identified between the change in PD max , PD min and vector sum or HD max after hindlimb diagnostic analgesia in any group.

| The percentage improvement of predominant hindlimb and concurrent forelimb lameness
After a positive analgesia of hindlimb lameness, the percentage improvement of vector sum of concurrent forelimb lameness improved <50% in all groups ( Table 2) and there was no correlation between percentage improvement of PD max or PD min and vector sum, but there was a positive correlation of percentage improvement between PD max and HD min in HmIi.

| The results of the linear mixed model, second model
The effect of decrease in predominant asymmetry on the concurrent asymmetry variables is presented in Table S4 and Table S5. A

| D ISCUSS I ON
Assessment of large number of clinical cases during BMIS evaluation of horses trotting in a straight line revealed that combined fore and hindlimb lameness is much more common than identification of forelimb-only and hindlimb-only lameness. 18 In our study, we only included horses with a combined fore-and hindlimb lameness pattern. In our cases, predominant impact lameness was more common in forelimbs than in hindlimbs and forelimb impact lameness 186/199 (93.5%) was more frequent than push-off lameness 13/199 (6.5%). Horses are likely prone to be injured at the impact phase of the stride as larger forces act on the musculoskeletal system during the stance phase on the forelimb. 23 During the stance phase, the forelimb is loaded with 60%-65% of the bodyweight. 13,24 The version of the LAMENESS LOCATOR ™ used in this study does not indicate mixed lameness in forelimb as HD max and HD min are considered as a single value (the vector sum) in determination of side and type of lameness. In hindlimb lameness, impact and push-off are determined and reported separately, and mixed lameness can be registered.
In predominant forelimb lameness and concurrent hindlimb lameness, the most obvious reduction in pelvic movement asymmetry after the positive forelimb analgesia occurred in the push-off and mixed type of lameness of the contralateral hindlimb. These were also the most common patterns and there were significant positive correlations between the percentage improvement in vector sum and PD max following diagnostic analgesia for these patterns. The PD max improved more than 50% in both groups providing evidence that the hindlimb asymmetry was a compensatory movement in these horses. The consistent improvement of PD max in compensatory lameness with a push-off component (FiCp, FiCm) indicates that the primary forelimb In the current study, forelimb impact lameness was also seen in combination with other types of hindlimb lameness. In horses showing a contralateral mixed (FiCm) lameness, the PD min , representing the impact component of hindlimb lameness, improved after predominant forelimb analgesia. It is possible that this is in fact a secondary lameness.
In a smaller number of horses, an ipsilateral hindlimb lameness improved after diagnostic analgesia in the forelimb suggesting a compensatory movement (FiIi, FiIm). Other studies also reported different compensatory patterns for forelimb lameness including both contralateral and ipsilateral compensatory hindlimb lameness. 7,[9][10][11] Our results are in contrast to the study by Maliye et al, 11  Vector sum did not decrease in concurrent forelimb lameness after diagnostic analgesia in hindlimbs in our study. However, Maliye and Marshall 12 recognised a significant improvement of vector sum in the concurrent ipsilateral forelimb lameness after diagnostic analgesia of the hindlimb but no correlation. They suggested that other kinematic parameters might provide an explanation and, therefore, further examinations are required. Horses could respond to hindlimb lameness by decreasing the push-off phase of stride of the lame hindlimb more than by increasing the load of the compensatory limb. 30 Keegan 29 described differences in head movement asymmetry between horses with predominant hindlimb lameness and suggested that this can be caused by an individual body conformation.
The results of the current study indicate that at least a part of the concurrent forelimb lameness observed with predominant hindlimb lameness is caused by a true lameness. The lameness may be independent from the hindlimb lameness or represent a secondary injury caused by longstanding compensation. In our cases no further workup of the forelimb lameness was performed because the lameness was very mild to moderate. 19 In horses with multiple limb lameness, with subjective evaluation, it is challenging to determine which are the lame limbs or if there is a compensatory lameness present. Examination is usually conducted on a straight line and circle to combine as much information as possible. 31 Evaluation is particularly difficult if the horse moves with a short, shuffling gait rather than overt lameness.
The retrospective design was a limitation of this study and the data could not be controlled. In horses presented with lameness in more than one limb, a primary and a compensatory lameness was identified with the help of diagnosis analgesia. However, there is a risk that horses with a true multiple limb lameness were included because in some cases a certain degree of asymmetry persisted after positive diagnosis analgesia of the primary lameness. There were many subgroups with variable numbers of horses of naturally occurring lameness patterns combined with concurrent lameness. Some of these subgroups were small, limiting the power of statistical analysis.

| CON CLUS IONS
Various asymmetry patterns of concurrent lameness were seen in horses with naturally occurring primary forelimb impact lameness, contralateral compensatory hindlimb lameness with a push-off and mixed component being the most common. In horses with hindlimb lameness, compensatory movements were seen in ipsilateral forelimbs, mostly as an ipsilateral push-off or impact lameness, during straight line trot was classified by the sensor system algorithm based on motion analysis.

E TH I C A L A N I M A L R E S E A RCH
This study was approved by the Ethics Committee of the Freie Universität Berlin.

OWN E R I N FO R M E D CO N S E NT
Owners gave consent for their animals' inclusion in the study.

ACK N OWLED G EM ENTS
We would like to thank our colleagues who were involved with helping during lameness examinations undertaken at Equine Clinic, Freie Universität Berlin. Open access funding enabled and organized by ProjektDEAL.

CO N FLI C T O F I NTE R E S T S
No competing interests have been declared.

AUTH O R CO NTR I B UTI O N S
S. Phutthachalee; principal author, study design, data collection, preparation of the manuscript and final approval of the manuscript, K.
Mählmann; manuscript preparation and revising the content, and final approval of the manuscript, S. Seesupa; study design, data analysis and interpretation, C. Lischer; senior author, contributed to the study design, revising the content and final approval of the manuscript.

DATA ACCE SS I B I LIT Y S TATE M E NT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/evj.13367.