The anesthetic mortality rate in horses is between 1% and 19.5%, depending on several factors that include the reason for anesthesia and time of day anesthesia is performed.1,2 Serious injury (most frequently long-bone fracture) during recovery that results in death is responsible for 12.5% to 38% of anesthetic-related deaths in horses.1 Consequently, identification of factors that increase morbidity and fatalities and methods and techniques that improve the quality of anesthetic recovery for horses is warranted. Numerous studies3–9 have been conducted to investigate various methods of improving the quality of the recovery period for horses. Examples of variables that have been evaluated include inhalation and injectable anesthetics, sedatives, analgesics, acetylcholinesterase inhibitors, recovery stall illumination, recovery surface, sling use during recovery, and use of a pool.
Evaluation of the quality of recovery from anesthesia for horses is most often performed by use of subjective visual grading scales. These include a VAS or numeric rating scale, which assigns 1 number to the overall recovery event; an SDS, which assigns a single word to the overall recovery event; or a CGS, which assigns a score to individual aspects of recovery that are summed to yield an overall score.10,11 Because these scales are subjective, they have more sources of systemic errors than do objective measurements.12 Additionally, subjective scoring systems for recovery of horses can have low sensitivity, low repeatability, and high variability; tend to be overly complex; and have institutional and sex bias in certain populations.13–16
Diplomates of the ACVAA are recognized as experts in veterinary anesthesia. They complete rigorous training programs and must pass certification examinations to be awarded diplomate status. Extensive training and demonstrated knowledge in equine anesthesia is a requirement of the training programs. Thus, ACVAA diplomates may have similar opinions and biases regarding quality of recovery of horses from anesthesia. However, it is unknown whether subjective recovery scores assigned by ACVAA diplomates would be similar.
Accelerometry is the quantitative determination of acceleration and deceleration of an object during movement. It is the measure of how rapidly velocity changes and is measured in g or m/s2. It can be used to measure acceleration during physical activity of animals. Three-axis accelerometry is measured with a device that has 3 separate accelerometers mounted orthogonally, which allows for measurement of acceleration in any direction. In human medicine, accelerometry has been widely used, including within the discipline of anesthesia to assess movement disorders of patients after anesthesia and surgery.17 In veterinary medicine, accelerometry has been used to detect estrus in cattle,18 investigate physical activity in dogs,19 assess analgesic treatments in dogs,20 perform gait analysis in dogs,21 and evaluate severity of osteoarthritis in cats.22 For horses, accelerometry has been used to assess lameness,23 racing surfaces,24 load weight,25 and response to flexion testing.26 A system that involves the use of objective accelerometry measurements of horses during recovery from anesthesia would minimize or eliminate subjective biases. However, a standardized system for this purpose has not yet been developed.
Objectives of the study reported here were to evaluate agreement among diplomates of the ACVAA for scores determined by use of an SDS for horses during recovery from anesthesia and evaluate agreement among other diplomates of the ACVAA for scores determined by use of a CGS for horses during recovery from anesthesia. If ACVAA diplomates were not in agreement for scores determined by use of a subjective measurement scale, it was our intent to develop a system that involved the use of 3-axis accelerometry for evaluation of horses during recovery from anesthesia. On the basis that there was common training and testing to achieve diplomate status, hypotheses for the first 2 objectives were that diplomates of the ACVAA would be in agreement when using an SDS or subjective CGS to evaluate horses during recovery from anesthesia.
Materials and Methods
Animals
Twelve healthy mixed-breed adult (> 2 years old) horses (6 castrated males and 6 sexually intact females) were used in the study. Horses were part of a teaching herd at a veterinary college (n = 6) or were owned by clients of a riding stable (6); all horses typically were group housed in outdoor pastures. Horses were of similar body type and stature. Body weight ranged from 443 to 571 kg. Horses were deemed healthy on the basis of a lack of abnormal findings for a routine physical examination, CBC, and serum biochemical analysis. Client consent was obtained for use of all client-owned horses. The study was approved by the University of Illinois Institutional Animal Care and Use Committee.
A minimum sample size of 12 horses was selected on the basis of the authors’ clinical experience and the number of subjects used in previous studies conducted to examine scoring systems13–16 or that involved the use of accelerometers with horses.23–26 We expected that this number of horses would result in a spectrum for mean CGS scores of at least 50 points (CGS scores could range from 11 to 100) and would be adequate to test agreement among diplomates. Additional horses were to be added if a minimum difference of 50 points was not obtained with the initial 12 horses. Additional horses were not needed, as determined on the basis of the mean CGS scores for the initial 12 horses.
Anesthetic protocol and instrumentation
Horses were allowed a minimum of 1 week to acclimate to stall confinement and the indoor environment before initiation of the study. Horses were allowed unlimited access to water, but feed was withheld for 12 hours prior to anesthesia. An area over the left jugular vein was aseptically prepared, and a 14-gauge catheter was placed and secured with suture. Horses were then fitted with a surcingle,a which was located immediately caudal to the elbow joints and tightened to prevent slipping (Figure 1). None of the horses objected to placement of the surcingle because all had been used as riding horses. A factory-calibrated commercially available 3-axis accelerometerb (size, 10 × 4 × 2.5 cm3; weight, 191 g) was attached to each surcingle at the most dorsal aspect of the scapulae (ie, withers). Accelerometers were programmed by use of the manufacturer's software to record Vmax data at 1-second intervals. The Vmax was a single value that quantified acceleration (change in velocity) in 3 axes. It was calculated as follows: Vmax 
, where X, Y, and Z are the axial accelerations for each of the 3 axes (measured in m/s2).
Photograph indicating the location of a surcingle placed on a horse for objective measurement of acceleration during recovery from anesthesia. Notice that the surcingle is placed around the thorax immediately caudal to the elbow joints. A 3-axis accelerometer is attached to the surcingle and located at the most dorsal aspect of the scapulae (ie, withers).
Citation: American Journal of Veterinary Research 78, 6; 10.2460/ajvr.78.6.668
Xylazine hydrochloridec (0.5 mg/kg, IV) was then administered to each horse. Once evidence of sedation was apparent, horses were led into a 4 × 4-m padded recovery stall that had video recording capability. Horses were positioned against one of the padded walls, and anesthesia was induced by IV administration of midazolamd (0.1 mg/kg) and ketamine hydrochloridee (2.2 mg/kg). Horses were positioned in right (n = 8) or left (4) lateral recumbency, depending on the position of their body on the floor after induction of anesthesia. The accelerometer and video recording softwaref were activated, and the recovery stall doors were closed. Ambient sound was not captured on the video recordings.
Horses were allowed to recover from anesthesia unassisted. Horses were left undisturbed, and no other activities occurred in the surrounding area during the entire recovery period to minimize extrinsic influences on recovery variables. Once a horse was standing and deemed to be recovered from anesthesia, the accelerometer was deactivated and the surcingle removed. Horses were considered recovered from anesthesia when they were able to stand without contacting a wall in the recovery stall and could take several steps without falling or colliding with a wall.
Each horse was returned to its stall, and the jugular vein catheter was removed. Videos were saved as .wmv files on a computer and distributed to ACVAA diplomates for rating via a file-sharing cloud drive.g Data acquired from the accelerometers were downloaded onto a software-specific computerh for subsequent analysis. The same procedures were repeated for all 12 horses.
SDS
Video recordings for the 12 horses were viewed separately by each of 8 ACVAA diplomates. Diplomates were not coached on how to evaluate recoveries by use of the SDS. The video file-sharing system could not be programmed to deliver the videos in random order; however, diplomates were not instructed to view videos in any specific order and therefore could view them in the order selected by each diplomate. Each diplomate was asked to assign a score for the recovery of each horse as excellent, good, fair, poor, or unacceptable. Diplomate agreement was determined by use of κ and AC1 statistics, as implemented in a macro for concordance statistics.27–31 The κ statistic is a method commonly used to assess reliability of raters; however, the AC1 statistical method has been developed more recently as an alternative because of instability in κ values in certain circumstances.32 The Landis and Koch scale was used to map κ values for the assigned labels to determine the degree of agreement.33 For the Landis and Koch scale, κ values for agreement were interpreted as follows: poor, < 0; slight, 0 to 0.20; fair, 0.21 to 0.40; moderate, 0.41 to 0.60; substantial, 0.61 to 0.80; and almost perfect, 0.81 to 1.00.
CGS
Video recordings for all 12 horses were viewed separately by 7 ACVAA diplomates who were not involved with grading by use of the SDS. Diplomates were not coached on how to evaluate recoveries by use of the CGS. The video file-sharing system that was used could not be programmed to deliver the videos in random order; however, diplomates were not instructed to view videos in any specific order and therefore could view them in the order selected by each diplomate. Recoveries were scored with a previously described 100-point CGS.4 A Heinze-Zirkler test was used to establish the multivariate normality of the diplomate scores. Variability between diplomate scores of horses was determined by use of a repeated-measures ANOVA. Consistency of individual diplomate scores was determined graphically. Horses were ordered on the basis of total score for the 7 diplomates. Each diplomate's cumulative score was plotted against the total score for the 12 horses. Overall concordance of diplomate CGS scores was determined with the Kendall coefficient of concordance, and pairwise rank correlations between diplomate CGS scores was determined with the Spearman rank correlation coefficient.
Development of a system for evaluating recovery by use of 3-axis accelerometry
To develop a nonsubjective rating system by use of accelerometry data, video recordings of each horse's recovery and the corresponding accelerometer data were viewed. Accelerometry values for each attempt to stand were visually correlated with video attempts to stand (the time signature on the accelerometer data was matched with the time on a stopwatch started at the same time that the accelerometer was activated). The Vmax was identified for each unsuccessful attempt to stand and for the successful attempt to stand. To account for the number of attempts and acceleration of each of the unsuccessful attempts to stand for each horse, Vmax values were summed. For the successful attempt to stand, the corresponding Vmax value was used. An attempt to stand was defined as any movement by a horse whereby the thorax and abdomen were lifted completely free of contact with the floor or walls of the recovery stall so that only parts of the limbs or hooves were in contact with the floor.
Graphs were used to visually determine whether there was a monotonic relationship between accelerometer values and visual CGS scores and to visually determine whether that relationship was linear or not linear. After graphically determining that a nonlinear association existed, mathematical development constrained the model to the range of the CGS (11 to 100). Median CGS scores for the 7 diplomates for each of the 12 horses (augmented by values for the successful attempt to stand [SG], sum of the unsuccessful attempts to stand [∑UG], and the score at the low [1, 1, and 11] and high [15, 30, and 100] limits for SG, ∑UG, and diplomates’ scores, respectively) were fitted to a library of nonlinear surface mathematical models by use of a computer programi to determine the best fit (ie, a response surface of accelerometer scores that provided correspondence with the subjective median scores [CGS] was developed by use of the subjective median score as the dependent variable). Predictor variables were the accelerometer values of the successful attempt to stand (SG) and the sum of the maximum accelerometer scores for the unsuccessful attempts to stand (∑UG). The surface was constrained to the minimum (11) and maximum (100) values that could be obtained from the CGS (ie, accelerometer scores outside the range were scored to the nearest limiting value of 11 or 100). This allowed for the accelerometer continuous variables to be expressed against the verbal descriptors used in the CGS with the terms and assessment criteria currently used in subjective studies.
Statistical analysis
All other analyses were performed with commercially available statistical software.j Values of P < 0.05 were considered significant.
Results
SDS
Overall, agreement among diplomates for the SDS was considered slight (overall κ, 0.19). Additionally, the AC1 value was 0.22, which further indicated minimal agreement among diplomates. Grouping horses on the basis of the frequency of ratings (ie, most of the horses received a specific recovery score) revealed that diplomates had slight agreement for 2 groups (good recovery, κ = 0.12; and fair recovery, κ = 0.03) and fair agreement for 3 groups (excellent recovery, κ = 0.33; poor recovery, κ = 0.29; and unacceptable recovery, κ = 0.25). Scores between pairs of diplomates differed for 17 of 28 (64%) comparisons.
CGS
Mean ± SD scores of the diplomates for the CGS for each of the 12 horses ranged from 19.0 ± 5.2 to 72.7 ± 12.0. There was uneven variability in the diplomate scores for the 12 horses (Figure 2). Overall, diplomates scores for individual horses differed significantly (P = 0.01). Specifically, comparisons between diplomate's scores differed significantly (9/21 [43%] pairs of diplomates; P = 0.03), which suggested a moderate amount of disagreement among diplomates as well as variability between diplomate scores of horses. Within-diplomate scores were generally consistent (ie, a diplomate who assigned a score for a specific horse that was less than the median score of all the diplomates for that horse typically assigned lower scores for all horses). Three diplomates gave scores for most of the horses that were less than the median value, and 4 diplomates gave scores for most of the horses that were greater than the median value. For the 3 diplomates who gave scores less than the median value, scores were less than the median for 32 of 36 (89%) scores assigned. For the 4 diplomates who gave scores greater than the median values, scores were greater than the median for 30 of 48 (63%) scores assigned (Figure 3). The Kendall overall coefficient of concordance for diplomate CGS scores was 0.913. The median Spearman correlation coefficient for pairwise comparison of diplomate CGS scores was 0.91 (range, 0.71 to 0.99 [Table 1]).
Box-and-whisker plots of CGS recovery scores for each of 12 horses during recovery from anesthesia rated by 7 diplomates of the ACVAA by use of a 100-point CGS. The lower limit of each box represents the 25th percentile, the upper limit of each box represents the 75th percentile, the horizontal line in each box represents the median, and the whiskers represent the range (highest and lowest scores).
Citation: American Journal of Veterinary Research 78, 6; 10.2460/ajvr.78.6.668
Graph of CGS recovery score for each of 12 horses during recovery from anesthesia, as rated by 7 diplomates of the ACVAA by use of a 100-point CGS and used to determine consistency among diplomates. Each colored bar represents results for 1 diplomate, and the black bars represent the median value for all 7 diplomates for each horse. For 3 diplomates, values were less than the median for most (32 of 36 [89%]) scores and for 4 diplomates, values were greater than the median for most (30/48 [62%]) scores.
Citation: American Journal of Veterinary Research 78, 6; 10.2460/ajvr.78.6.668
Spearman pairwise rank correlation coefficient values among 7 diplomates of the ACVAA for scores determined by use of a CGS to evaluate quality of recovery of 12 horses from anesthesia.
| Diplomate | |||||||
|---|---|---|---|---|---|---|---|
| Diplomate | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| 1 | — | 0.94 | 0.96 | 0.82 | 0.97 | 0.89 | 0.96 |
| 2 | 0.94 | — | 0.90 | 0.71 | 0.94 | 0.90 | 0.89 |
| 3 | 0.96 | 0.90 | — | 0.85 | 0.95 | 0.92 | 0.95 |
| 4 | 0.82 | 0.71 | 0.85 | — | 0.85 | 0.78 | 0.87 |
| 5 | 0.97 | 0.94 | 0.95 | 0.85 | — | 0.93 | 0.99 |
| 6 | 0.89 | 0.90 | 0.92 | 0.78 | 0.93 | — | 0.91 |
| 7 | 0.96 | 0.89 | 0.95 | 0.87 | 0.99 | 0.91 | — |
Strength of the correlation of ranks is interpreted as follows: very high, 0.90 to 1.00; high, 0.70 to 0.89; moderate, 0.50 to 0.69; low, 0.30 to 0.49; and negligible, 0.00 to 0.29.
— = Not applicable.
Development of a system for evaluating recovery by use of 3-axis accelerometry
Mean ± SD accelerometry Vmax for the 12 horses was 4.38 ± 2.37 m/s2 for SG and 22.57 ± 31.49 m/s2 for ∑UG. The best fit (determined by use of fit statistics and visual verification) for development of a system to evaluate recovery of horses from anesthesia was for a power model (Figure 4). That model resulted in the following equation (r = 0.87; P < 0.001) for a horse's recovery score: recovery score = 9.998 × SG0.633 × ∑UG0174.
Graphic representation for a power model of recovery scores determined by use of 3-axis accelerometry for 12 horses during recovery from anesthesia. Max G (ie, SG) is the maximum gravity (ie, acceleration) measured during the successful attempt to stand. Sum of Max G (ie, ∑UG) is the sum of the maximum gravities (ie, accelerations) measured during all unsuccessful attempts to stand. Median CGS is the median composite score determined by 7 diplomates of the ACVAA by use of a 100-point CGS. Best fit of the 3-axis accelerometry data against CGS scores yielded the following equation: recovery score = 9.998 × SG0.633 × ∑UG0.174. Notice that the color gradient from blue to red indicates an increase in recovery score.
Citation: American Journal of Veterinary Research 78, 6; 10.2460/ajvr.78.6.668
Discussion
Diplomates of the ACVAA generally did not agree on the quality of the recovery of horses from anesthesia when graded by use of an SDS or CGS. For the SDS, whereby a single word was assigned to describe a horse's recovery, diplomates had slight to fair agreement, and pairs of diplomates had different scores for most comparisons (17/28 [64%]), which indicated that their opinions of what constituted a quality recovery differed greatly. For the more complex CGS, diplomates did not agree for many comparisons (9/21 [43%]). When horses were ordered on the basis of ascending score, and scores were then compared between diplomates, ranks for individual diplomates typically were similar to those for other diplomates. This indicated that diplomates were in agreement with the order of horses regarding the severity of recovery quality. This must be interpreted with caution, however, because although ranking orders may have been similar, CGS scores (scale, 11 to 100 points) differed, which indicated that diplomates had different interpretations of the quality of recovery for each horse. Interestingly, diplomates generally were consistent in that if they rated recovery of a horse higher or lower than the median of other diplomates, they typically rated recovery of most horses similarly higher or lower than the median. These findings suggested that diplomates of the ACVAA, despite having to meet the same test standards for board certification, had poor agreement when using subjective methods to score recovery of horses from anesthesia. This could result in anesthesiologists reaching different conclusions about the recovery performance of a horse.
The weak agreement among subjective ratings of diplomates could have been related to differences in training programs, amount of time as a diplomate, experience, gender, or other environmental factors. Little research has been conducted to examine the methods for evaluating quality of recovery of horses. In 1 study13 in which investigators used a 100-point VAS to assess the quality of recovery of 24 horses, female horse owners assigned significantly lower scores than did male horse owners. However, in that same study, there was no gender difference in assigned scores between veterinary anesthesiologists and veterinary surgeons. Additionally, experience did not affect VAS scores in that study.13 To account for variable periods of familiarity with equine anesthesia in that study,13 experience was defined as the number of injuries a specific rater reported had been observed during recovery of horses from anesthesia, and not the amount of time that the rater had been practicing veterinary medicine. In the study reported here, participants were diplomates in the ACVAA for a mean ± SD of 13.1 ± 9.2 years (range, 2 to 30 years). In another study,16 reproducibility and repeatability for 2 commonly used subjective scoring systems and a novel system were tested. In that study, 12 raters viewed 10 videos of horses during recovery and scored each of them with each of the 3 scoring systems; 6 months later, the raters repeated the process. Various degrees of intrarater variability and consistency were found among the 3 systems. Although there was some degree of reliability and reproducibility for the 3 systems, there were limitations associated with practicality, simplicity, and imprecision that suggested that an alternative system should be investigated. These findings further support the need for a system that limits the use of subjective indices and is not dependent on the experience of the raters.
Use of subjective measurements in other areas of veterinary medicine has been studied. For example, studies conducted to investigate the ability of therapeutic medications to alleviate pain have relied heavily on subjective assessment. As in the study reported here, subjective assessment of pain has similar variability in observer results. In a study34 in which 4 veterinarians with postgraduate training in anesthesia used 3 subjective methods to assess pain in 50 dogs after surgery, agreement among the veterinarians was fair, with κ values similar to those for the study reported here. More recently, a study35 conducted to assess agreement between veterinary students and anesthesiologists regarding postoperative pain in dogs revealed that experience plays a role in the perceptions that affect subjective scoring, as was indicated by differences in scores between the students and anesthesiologists. Additionally, the level of agreement among the anesthesiologists was only moderate in that study,35 which indicates that factors other than educational experience alone play a role in subjective assessment. Findings for pain assessment studies indicate that on the basis of subjective assessment, individuals can draw different conclusions about an animal's status.
Use of subjective measurement in clinical medicine, errors, and sources of those errors have been investigated and defined.12,36–38 One of the main inherent limitations appears to be the knowledge base and competence of the person rendering a judgment, which suggests that expertise and experience greatly influence results of subjective testing.36 Additionally, the type and presentation of a subjective rating system can affect results. For example, length of a VAS, use of right-angle stops, distribution of scale marks and descriptors, and amount of experience with a VAS can affect the distribution and uniformity of scores.37,38 Other errors and bias can occur with the use of subjective measurements.12,39–41 Different behavior occurs when subjective measurements are affected by intentional or emotional behaviors. Selective perception occurs when observers allow themselves to perceive only what is most important to them. Intrusive measurement occurs when the reality of a measurement could not be fully revealed because of intentional or accidental deviation or intentional noncompliance. Memories of previous assessments result in errors attributable to use of past assessments to prevent an accurate assessment of the reality of the current measurement. Availability is a type of error that occurs when memories of past experiences influence the available knowledge of an individual providing a judgment or assessment. Similarity occurs when there is great similarity with a measurement of others that results in a tendency to repeat already known measurements. Automated response occurs with measurements obtained without further analysis or evaluation and for which the results do not necessarily reflect the reality of the situation. Anchoring and adjustment occur when a person tends to adjust the answer based on some initial value that serves as an anchor. Clearly, any or all of these errors can occur when evaluating horses during recovery from anesthesia, which supports the development of an objective system for evaluation.
Diplomates in the present study were not coached or trained on how to score horses by use of the SDS or CGS. Coaching or training can potentially introduce availability, similarity, automated response, and anchoring and adjustment bias. Additionally, because all ACVAA diplomates presumably complete similar training programs and the certification examination to achieve diplomate status, we wanted to test agreement among a presumably relatively uniform population.
Accelerometers have been used extensively in human medicine to obtain objective data in areas such as anesthesia, pain management, and exercise physiology.17,42,43 Accelerometers are being used more frequently in veterinary medicine for data collection to enhance clinical applications and investigations. Some examples of the use of accelerometers include collection of objective data to monitor behavior in beef and dairy cattle,18,44 to assess gait analysis and the amount of physical activity in dogs,19–21,45,46 and to evaluate osteoarthritis in cats.22 In a manner similar to that for their use on horses of the present study, accelerometers have been affixed to elastic bands placed around the thorax of adult horses to aid in developing algorithms to account for movement artifacts on ECGs and to investigate lying behavior in horses during periods of recumbency.47,48 Similar to results for horses of the present study, accelerometers were well tolerated by all horses in those previous studies.47,48 In recumbent horses in 1 study,48 predictability, sensitivity, and specificity were all > 99%, which further supports the potential usefulness of accelerometers in evaluating animals during recovery from anesthesia.
Location for placement of the accelerometer on the horses was considered during the planning of the present study. Areas considered for placement included the head, distal aspect of the limbs, and various locations on the thorax and abdomen. Locations on the head and limbs were eliminated because movement of the head and limbs can occur unrelated to an attempt to stand, and wide swings of the head and limbs could have resulted in accelerations beyond the limits of measurement of the accelerometer. The withers was chosen because it allowed accessibility to the accelerometer regardless of the position (right or left lateral recumbency) of a horse after induction of anesthesia and was along the anatomic pathway of the surcingle. As advances are made in accelerometer technology, a miniaturized accelerometer that can be attached to a horse without the need for a surcingle should be developed.
In the present study, an algorithm was developed by use of measuring movement via 3-axis accelerometry of all unsuccessful attempts and the successful attempt of a horse to stand during recovery from anesthesia to assign an objective numeric score for a horse's recovery. The accelerometer recovery scores were plotted against the CGS scores, with the constraints that a recovery score < 11 was interpreted as a score of 11 (excellent) and a recovery score > 100 was interpreted as a score of 100 (unacceptable). The objective accelerometer scoring system may be used to replace subjective scoring systems for recovery assessment and, in multicenter trials, would potentially reduce variability and bias among centers. It is likely that scores that differ significantly but that are numerically similar (eg, 35 vs 38) would represent clinically irrelevant differences. Therefore, a categorical grouping of scores may be useful for clinical and research purposes. On the basis of results for the present study, preliminary categorization of recovery scores would have been as follows: 11 to 30 as excellent, 31 to 50 as good, 51 to 70 as fair, 71 to 90 as poor, and ≥ 91 as unacceptable. However, recovery score ranges that distinguish differences in recovery will need to be developed by use of a larger sample of horses, institutions, and ACVAA diplomates as well as inclusion of other veterinarians and, possibly, veterinary students.
Use of an accelerometer method will require that 2 assumptions must be met. One is that a lower acceleration of a horse during recovery equates to a higher quality recovery. This appears logical because a horse that has greater and more coordinated muscular control after it achieves a standing position would have less acceleration than a horse with a poor ability to control its mass movement and that repeatedly falls or staggers and collides with a wall of the recovery stall. This assumption was supported by results for an accelerometer placed on 1 nonanesthetized horse resting in sternal recumbency in a stall (data not shown). Results indicated that once that horse stood during 1 controlled attempt, the accelerometer data (which were inputted into the equation derived in the study reported here) yielded the lowest possible score of 11. However, testing of this method on several unanesthetized horses as they transition from a sternal to a standing position should be performed to support validation of the use of the accelerometer method. The second assumption is that fewer attempts to stand also equate to a better quality recovery. Again, this is a logical assumption because repeated attempts to stand likely indicate poor coordination and muscular control.
Accelerometers could be used in research on recovery from anesthesia. Data obtained with an accelerometer during recovery from anesthesia could be used along with other objective data, such as time to first movement, number of attempts to stand, and time to stand, to construct a more complete characterization of the recovery process for comparison of various recovery methods.
Limitations associated with the study reported here included the inability to ensure randomization of the order in which videos were viewed by diplomates, use of a surcingle on the horses, and no criterion-referenced standard for comparison of outcomes. For the present study, we used subjective data to determine agreement of the m/s2-based quantities (ie, SG and ∑UG) with values for familiar subjective scoring systems; however, in the longer term, we propose that an accelerometer-based method be developed to replace the subjective methods.
Evaluation of results of the present study indicated that subjective assessment of horses during recovery from anesthesia by use of an SDS or CGS resulted in significant variability in agreement among diplomates and supported the need for development of a nonsubjective evaluation system. Additionally, details were provided on the development of a system that involved the use of 3-axis accelerometry to evaluate recovery of horses, which is a method that may eliminate subjective error and variability. Further development and testing of accelerometry is necessary before recommendations on use of such methods can be made. Future studies should include horses during recovery from various methods of anesthesia and horses anesthetized for various surgical procedures. The repeatability, sensitivity, and specificity of 3-axis accelerometry should be assessed. These data will be necessary to develop a database that can be used to form a consensus on the interpretation of m/s2-derived values as, for example, acceptable or unacceptable. In the present study, we devised an algorithm that was an initial link between an objective method and a clinical verbal description of a horse's recovery from anesthesia.
Acknowledgments
Presented in abstract form at the 12th World Congress of Veterinary Anaesthesiology, Kyoto, Japan, September 2015.
ABBREVIATIONS
| ACVAA | American College of Veterinary Anesthesia and Analgesia |
| CGS | Composite grading scale |
| SDS | Simple descriptive scale |
| VAS | Visual analogue scale |
| Vmax | Maximum velocity |
Footnotes
Exselle surcingle, National Bridle Shop Inc, Lewisburg, Tenn.
GP1-L programmable accelerometer, Sensr, Elkader, Iowa.
AnaSed, Akorn, Decatur, Ill.
Hospira Inc, Lake Forest, Ill.
Ketaset, Fort Dodge, Fort Dodge, Iowa.
Replay video capture, Applian Technologies Inc, San Anselmo, Calif.
OneDrive, Microsoft Corp, Redmond, Wash.
Sensware, version 1.1.1.0, Sensr, Elkader, Iowa.
CurveExpert Pro, DG Hyams, Madison, Ala.
SAS, version 9.3, SAS Institute Inc, Cary, NC.
References
1. Senior JM. Morbidity, mortality, and risk of general anesthesia in horses. Vet Clin North Am Equine Pract 2013; 29: 1–18.
2. Dugdale AHA, Taylor PM. Equine anaesthesia-associated mortality: where are we now? Vet Anaesth Analg 2016;43:242–255.
3. Chesnel MA, Clutton RE. A comparison of two morphine doses on the quality of recovery from general anesthesia in horses. Res Vet Sci 2013; 95: 1195–1200.
4. Clark-Price SC, Posner LP, Gleed RD. Recovery of horses from general anesthesia in a darkened or illuminated recovery stall. Vet Anaesth Analg 2008; 35: 473–479.
5. Posner LP, Kasten JI, Kata C. Propofol with ketamine following sedation with xylazine for routine induction of general anaesthesia in horses. Vet Rec 2013; 173: 550.
6. Ray-Miller WM, Hodgson DS, McMurphy RM, et al. Comparison of recoveries from anesthesia of horses placed on a rapidly inflating-deflating air pillow or the floor of a padded stall. J Am Vet Med Assoc 2006; 229: 711–716.
7. Wiese AJ, Brosnan RJ, Barter LS. Effects of acetylcholinesterase inhibition on quality of recovery from isoflurane-induced anesthesia in horses. Am J Vet Res 2014; 75: 223–230.
8. Woodhouse KJ, Brosnan RJ, Nguyen KQ, et al. Effects of postanesthetic sedation with romifidine or xylazine on quality of recovery from isoflurane. J Am Vet Med Assoc 2013; 242: 533–539.
9. Steffey EP, Brosnan RJ, Galuppo LD, et al. Use of propofol-xylazine and the Anderson sling suspension system for recovery of horses from desflurane anesthesia. Vet Surg 2009; 38: 927–933.
10. Donaldson LL, Dunlop GS, Holland MS, et al. The recovery of horses from inhalant anesthesia: a comparison of halothane and isoflurane. Vet Surg 2000; 29: 92–101.
11. Hubbell JAE. Recovery from anaesthesia in horses. Equine Vet Educ 1999; 11: 160–167.
12. Bispo CAF, Cazarini EW. A study on the foundations of the occurrence of errors in subjective measurements. Int J Bus Hum Technol 2012; 2: 173–183.
13. Farmer E, Chase-Topping M, Lawson H, et al. Factors affecting the perception of recovery quality in horses after anaesthesia. Equine Vet J 2014; 46: 328–332.
14. Portier KG, Séna A, Senior M, et al. A study of the correlation between objective and subjective indices of recovery quality after inhalation anaesthesia in equids. Vet Anaesth Analg 2010; 37: 329–336.
15. Suthers JM, Christley RM, Clutton RE. A comparison of scoring systems for assessing recovery quality after general anaesthesia in horses, in Proceedings. 47th Br Equine Vet Assoc Congr 2008; 140–141.
16. Suthers JM, Christley RM, Clutton RE. Quantitative and qualitative comparison of three scoring systems for assessing recovery quality after general anaesthesia in horses. Vet Anaesth Analg 2011; 38: 352–362.
17. Swan BD, Grill WM, Turner DA. Investigation of deep brain stimulation mechanisms during implantable pulse generator replacement surgery. Neuromodulation 2014; 17: 419–424.
18. Valenza A, Giordano JO, Lopes, Jr. G, et al. Assessment of an accelerometer system for detection of estrus and treatment with gonadotropin-releasing hormone at the time of insemination in lactating dairy cows. J Dairy Sci 2012; 95: 7115–7127.
19. Morrison R, Penpraze V, Beber A, et al. Associations between obesity and physical activity in dogs: a preliminary investigation. J Small Anim Pract 2013; 54: 570–574.
20. Wernham BG, Trumpatori B, Hash J, et al. Dose reduction of meloxicam in dogs with osteoarthritis-associated pain and impaired mobility. J Vet Intern Med 2011; 25: 1298–1305.
21. Pillard P, Gibert S, Viguier E. Development of a 3D accelerometric device for gait analysis in dogs. Comput Methods Biomech Biomed Engin 2012; 15(suppl 1):246–249.
22. Guillot M, Moreau M, d'Anjou MA, et al. Evaluation of osteoarthritis in cats: novel information from a pilot study. Vet Surg 2012; 41: 328–335.
23. Keegan KG, Kramer J, Yonezawa Y, et al. Assessment of repeatability of a wireless, inertial sensor-based lameness evaluation system for horses. Am J Vet Res 2011; 72: 1156–1163.
24. Crevier-Denoix N, Pourcelot P, Hoden-Douilly L, et al. Discrimination of two equine racing surfaces based on forelimb dynamic and hoof kinematic variables at the canter. Vet J 2013; 198(suppl 1):e124–e129.
25. Matsuura A, Irimajiri M, Matsuzaki K, et al. Method for estimating maximum permissible load weight for Japanese native horses using accelerometer-based gait analysis. Anim Sci J 2013; 84: 75–81.
26. Marshall JF, Lund DG, Voute LC. Use of a wireless, inertial sensor-based system to objectively evaluate flexion tests in the horse. Equine Vet J Suppl 2012; 43: 8–11.
27. Fleiss JL, Nee JC, Landis JR. Large sample variance of kappa in the case of different sets of raters. Psychol Bull 1979; 86: 974–977.
28. Fleiss JL, Levine B, Paik MC. The measurement of interrater agreement. In: Fleiss JL, Levine B, Paik MC, eds. Statistical methods for rates and proportions. 3rd ed. New York: John Wiley & Sons Inc, 2003; 598–626.
29. Kendall MG. Ranks and variate values. In: Kendall MG, ed. Rank correlation methods. 2nd ed. New York: Hafner Press, 1955; 123–143.
30. Advanced Analytics LLC website. Computing inter-rater reliability with the SAS system. Available at: irrhbk.advance-danalyticsllc.com/research_papers/inter_rater_reliability_with_sas.pdf. Accessed Feb 10, 2016.
31. SAS Institute Inc. mAgree macro: sample 25006: Compute estimates and tests of agreement among multiple raters. Available at: support.sas.com/kb/25/006.html. Accessed Feb 6, 2016.
32. Wongpakaran N, Wongpakaran T, Wedding D, et al. A comparison of Cohen's kappa and Gwet's AC1 when calculating inter-rater reliability coefficients: a study conducted with personality disorder samples. BMC Med Res Methodol 2013; 13: 61.
33. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 159–174.
34. Holton LL, Scott EM, Nolan AM, et al. Comparison of three methods used for assessment of pain in dogs. J Am Vet Med Assoc 1998; 212: 61–66.
35. Barletta M, Young CN, Quandt JE, et al. Agreement between veterinary students and anesthesiologists regarding postoperative pain assessment in dogs. Vet Anaesth Analg 2016; 43: 91–98.
36. Sonnenberg A. We only see what we already know—a modified Bayes' formula to explain inherent limitation of diagnostic tests. Med Hypotheses 2004; 63: 759–763.
37. Wewers ME, Lowe NK. A critical review of visual analogue scales in the measurement of clinical phenomena. Res Nurs Health 1990; 13: 227–236.
38. McCormack HM, Horne DJ, Sheather S. Clinical applications of visual analogue scales: a critical review. Psychol Med 1988; 18: 1007–1019.
39. Tversky A, Kahneman D. Judgment under uncertainty: heuristics and biases. Science 1974; 185: 1124–1131.
40. Kahneman D. A perspective on judgment and choice: mapping bounded rationality. Am Psychol 2003; 58: 697–720.
41. Hicks EP, Kluemper GT. Heuristic reasoning and cognitive biases: are they hindrances to judgments and decision making in orthodontics? Am J Orthod Dentofacial Orthop 2011; 139: 297–304.
42. Caruso JF, Olson NM, Taylor ST, et al. Front squat data reproducibility collected with a triple-axis accelerometer. J Strength Cond Res 2012; 26: 40–46.
43. Cliff DP, Reilly JJ, Okely AD. Methodological considerations in using accelerometers to assess habitual physical activity in children aged 0–5 years. J Sci Med Sport 2009; 12: 557–567.
44. White BJ, Coetzee JF, Renter DG, et al. Evaluation of two-dimensional accelerometers to monitor behavior of beef calves after castration. Am J Vet Res 2008; 69: 1005–1012.
45. Vitger AD, Stallknecht BM, Nielsen DH, et al. Integration of a physical training program in a weight loss plan for overweight pet dogs. J Am Vet Med Assoc 2016; 248: 174–182.
46. Clark K, Caraquel C, Leahey, et al. Evaluation of a novel accelerometer for kinetic gain analysis in dogs. Can J Vet Res 2014; 78: 226–232.
47. Lanata A, Guidi A, Baragli P, et al. A novel algorithm for movement artifact removal in ECG signals acquired from wearable systems applied to horses. PLoS One 2015; 10:e0140783.
48. DuBois C, Zakrajsek E, Haley DB, et al. Validation of triaxial accelerometers to measure the lying behavior of adult domestic horses. Animal 2015; 9: 110–114.
