The foot fault scoring system to assess walking adaptability in rats and mice: a reliability study

The foot fault scoring system of the ladder rung walking test is used to assess walking adaptability in rodents. However, the reliability of the ladder rung walking test foot fault score has not been properly investigated. This study was designed to address this issue.Two independent and blinded raters analyzed 20 rat and 20 mice videos. Each video was analyzed twice by the same rater (80 analyses per rater). The intraclass correlation coefficient (ICC) and the Kappa coefficient were employed to check the accuracy of agreement and reliability in the intra- and inter-rater analyses of the ladder rung walking test outcomes. Excellent intra- and inter-rater agreement was found for the forelimb, hindlimb and both limbs combined in rats and mice. The agreement level was also excellent for total crossing time, total time stopped and number of stops during the walking path. Rating individual scores in the foot fault score system (0 to 6) ranged from satisfactory to excellent, in terms of the intraclass correlation indexes. Moreover, we showed experienced and inexperienced raters can obtain reliable results if supervised training is provided. We conclude the ladder rung walking test is a reliable and useful tool to study walking adaptability in rodents and can help researchers address walking-related neurobiological questions.


INTRODUCTION
Walking adaptability can be defined as a complex sensory-motor function, qualified 56 or required to control and coordinate various degrees of freedom in joints, in a variety of 57 environmental contexts, or that interfere with locomotion [1][2][3]. Gait is influenced by the 58 temporal and spatial integration of the cognitive and neuromusculoskeletal neural systems 59 [4]. Moreover, the ability to adapt gait according to environmental context is a crucial aspect 60 in maintaining body stability and preventing falls [5][6][7][8]. 61 Whilst several studies into walking adaptability have focused on human 62 biomechanics [2,9,10], animal models can usefully provide neurobiological insights at the 63 cellular and molecular level [11][12][13]. For instance, the Ladder Rung Walking Test (LRWT) 64 has been used to assess walking adaptability [14,15] in unilateral ischemic injury in the 65 motor cortex [12,16]; spinal cord injury [17,18]; dopaminergic depletion induced by 6-66 hydroxydopamine (a model of Parkinson's disease) [19]; neonatal white matter injury [20] 67 and stress-related conditions [7,13,21]. 68 The LRWT can assess walking patterns by using measures of inter-foot 69 coordination, foot support, fore and hindlimb kinematics, step and gait cycles, gait speed, and 70 the ability to adapt walking by applying a foot-fault score [16,17]. The test provides 71 measures of gait adaptability with emphasis in forelimb and hindlimb function by applying 72 the foot-fault score [15]. The foot-fault score system is widely used in the literature since it 73 requires only a hand camera and a minimally trained researcher to analyze the video and 74 apply the foot-fault score [13,14]. This method may avoid common pitfalls that occur when 75 using reflective markers on the flexible skin of rodents [22,23] and gives a measure of the 76 success in adapting walking [7,13].

77
The foot-fault score system is a 7-point category scale in which the quality and 78 appropriateness of foot placement is judged by analyzing a video recording, frame-by-frame, 79 of rodents walking along a 1-meter long horizontal ladder. The rungs are arranged in a pattern 80 that requires murine ability to adapt walking [14,15]. However, to the best of our knowledge, 81 this test has not been properly assessed regarding its intra-rater and inter-rater reliability and 82 reproducibility, which is a source of uncertainty. Current studies usually elect a single rater to 83 analyze all videos in an attempt to minimize bias, which is scientifically insufficient. The 84 present study sought to provide scientific information regarding the external validity of the 85 LRWT findings in rodents, thus contributing to advancements in the field of neurobiology of 86 walking adaptability.

89
We randomly select 40 video recordings of rodents from our lab database ( 100 We used two LRWT apparatus, one for rats and another adapted for mice. Both 101 consisted clear Plexiglas side walls (100 cm long and 20 cm high). The diameter of the metal 102 rungs varied, being 3 mm for rats and 2 mm for mice. The minimum and maximum gaps 103 between the rungs also varied, being from 1 to 5 cm for rats and from 0.5 to 2.5 cm for mice.

104
In both cases, the ladders were elevated horizontally 30 cm above the ground, with a neutral 105 cage placed in the starting position and the animal's home cage placed at the opposite end of 106 the ladder (Figure 1). The between-wall distance was adjusted leaving 1 cm wider than the 107 size of rodent to prevent the animal turning around during the crossing [13,14,24].  Symmetrical Pattern). In irregular patterns, rungs are randomly repositioned in each trial to 116 prevent the rodents learning the rung sequence. Thus, irregular patterns are more useful when 117 studying walking adaptability ( Figure. 1, Asymmetrical Pattern) [7,13,14]. In this study, 118 only irregular rung patterns were analyzed.

119
In the test, the animals were placed at the beginning of the ladder, walked along it, 120 adapting their foot placement on the rungs until reaching the home cage ( Figure 1). While 121 performing the test, we filmed the rodents using a camera (GoPro Hero 4, 12 megapixels). An 122 acquisition rate of 240 frames per second (FPS) in a lateral view was adopted allowing a 123 post-hoc frame-by-frame video analysis.

126
To assess the fore and hindlimb placement on the rungs, which requires precise and 127 coordinated foot positioning as well as stride and inter-limbic coordination a quantitative foot 128 fault scoring system [14] derived from a categorical analysis was used. In the system, a 129 frame-by-frame video recording analysis is performed to identify the steps in each limb and 130 qualify foot placement using a 7-point category scale [14,15] (Table 1). The score 0 is given 131 when the limb did not touch the rung (missed a rung) and resulted in a fall (total miss). A fall 132 is considered when the limbs fell between rungs and the animal's posture and balance are 133 disturbed. Score 1 is given when the limb slipped off a rung and a fall occurred (deep slip).

134
Score 2 is given when the limb slipped off a rung during weight bearing, but a fall did not 135 occur and the rodent interrupts walking (slight slip). Score 3 is given when, before weight 136 bearing the limb on a rung, the rodent quickly lifted and placed the foot on another rung 137 (replacement). Score 4 occurs when the limb is clearly about to be placed on a rung, but the 138 rodent quickly changes the feet placement to another rung without touching the first rung 139 (correction). Score 4 is also given when the limb is placed on a rung, but the animal removes 140 the foot and repositions it on the same rung. Score 5 is given when the limb is placed on the 141 rung either using the wrist or digits for the forelimb or heel or toes for the hindlimb (partial 142 placement). Finally, score 6 is given when the full body weight bearing is applied on a rung 143 with the midportion of the foot (correct placement) (Table 1).

145
The score given in each category is then multiplied by the frequency of foot 146 placements in the same category. Afterwards, the sum of all the categories provides the total 147 combined score (sum of the forelimb plus the hindlimb scores). The fully explained video 148 protocol and all technical details to apply the foot fault score were previously published by 149 Metz and Whishaw (2009). In this study the following outcomes in the LRWT were assessed for inter-rater and

Foot placement reliability between inter-and intra-rater 169
In order to asses inter-and intra-rater reliability, two independent and blinded raters for scores 2, 5 and 6 was also considered satisfactory (Table 2).
204 Table 2. Agreement between raters I and II regarding the outcomes obtained in the LRWT in rat.

Inter-rater reliability for mice 217
The inter-rater reliability score system for mice is shown in Table 3. We observed a 218 strong agreement between the raters in the combined total score (ICC=0.954/P=0.0001), total  Table 3. Agreement between raters I and II regarding the outcomes (scores) recorded in the LRWT in mice.   Table 4 shows the intra-rater analyses in rat. We found excellent agreement in the 230 combined total score, total crossing time, number of stops and total time stopped for both 231 raters. Regarding score evaluation, rater I obtained excellent agreement in all the scores for 232 the forelimb (ICC between 0.899 to 0.989 / p=0.0001). Rater II achieved excellent agreement 233 in all scores for forelimb (ICC between 0.787 to 0.920), except for score 6, which was 234 considered satisfactory (ICC=0.652 / p=0.13). In relation to hindlimb agreement, rater I obtained a similar excellent degree of 237 agreement to that for the forelimb, ranging from ICC 0.838 to 1 / p=0.0001. Whilst rater II 238 achieved a lower agreement than rater I, the ICC was very good, ranging from 0.637 to 1, 239 with only score 5 graded as satisfactory (ICC 0.637). Moreover, both raters obtained 240 excellent intra-rater scores in the outcomes: combined total score, total crossing time, number 241 of steps, total time stopped and total score for forelimb and hindlimb, ranging from ICC 242 0.806 to 0.993 for rater I and ICC 0.915 to 0.981 for rater II.  245 Overall, the intra-rater reliability for mice was excellent for both raters (Table 5). For 246 rater I, in the forelimb foot placement agreement for all the 7 scores were excellent (ICC 247 0.939 to 1 / p=0.0001). For rater II, the agreement was also excellent, varying between ICC 248 0.778 and 0.968 for scores 0 to 5. However, score 6 was considered satisfactory (ICC 0.488 / 249 p=0.077). Regarding the hindlimb placement, similar results were found, with the raters only 250 differing in score 6 (rater II obtained a lower ICC: 0.749 / p=0.002) ( Table 5).   According to the authors, the test is a sensitive skilled task for assessing slight impairments of 264 walking function and is useful when assessing functional recovery following brain or spinal 265 cord injury and the effectiveness of rehabilitative therapies [14,30]. Locomotion during the 266 ladder rung walking test is known to depend on ascending and descending neural pathways, 267 since accurately crossing the rungs requires finely adjusted motor control, balance, limb 268 coordination and muscle control [7,13,14]. 269 However, to determine the psychometric properties of behavioral tests it is essential to 270 obtain reliable, consistent and scientifically valid findings [31]. Both, intra-and inter-rater 271 agreement are important metrics to ensure reliability and reproducibility [32]. Here, we 272 sought to assess intra-and inter-rater agreement in the foot fault score of the ladder rung 273 walking test using two strains of rodents -Wistar rats and C57BL/6 mice. Two independent 274 researchers (with and without previous experience using the test's scoring system) analyzed 275 the videos. Our findings suggest the foot fault score system of the ladder rung walking test is 276 a useful, reliable and consistent tool for studying skilled walking performance in rodents. We 277 also found excellent inter and intra-rater reliability for "total crossing time", "number of 278 stops" and "total time stopped". The agreement measures provided by this study suggest data 279 obtained by different research groups using the ladder rung walking test should be 280 comparable [33] and encourage the use of the test in further studies.

281
The ladder rung walking test is an interesting option for researchers investigating 282 neural mechanisms involved in the ability to adapt walking [7,13,15,34]. Since the score 283 reflects the animal's ability to adapt limb placement and position in a contextual environment 284 [14,29], the foot fault score system is useful to study walking adaptability in rodents [7,13].

285
Whilst traditional biomechanical models of walking analysis require expensive devices, 286 constant animal handling for placing reflective markers and development of signal-processing 287 routines [35,36], the ladder rung walking test provides walking adaptability assessment using 288 a fast, simple and inexpensive method.

289
Whereas we observed satisfactory to excellent intraclass correlation indexes in rating 290 individual scores (0 to 6), caution is necessary when using the foot fault score system.

291
Individual scores present subtle differences that may confuse untrained raters. For example, 292 differentiating between scores 3 (replacement) and 4 (correction) requires attention to 293 identify whether the rodent touched the rung before completing paw placement. Moreover, in 294 some situations, the rodent supports a single paw simultaneously on two rungs that are placed 295 too close each other. This may cause confusion in scoring 5 (partial placement) or 6 (correct 296 placement). In addition, rodents sometimes place their paw on the acrylic wall to help 297 walking forward, a behavior that is not considered in the foot fault score system.

298
Furthermore, the subtle differences between score 1 (deep slip) and 2 (slight slip) may cause 299 uncertainty for untrained raters. Finally, the speed of the video recording may also change the 300 perception of the raters during the gait cycle analysis [37]. Thus, the present results suggest 301 experienced and inexperienced raters can get reliable results if appropriate training is 302 provided. We highly recommend the careful study of the article and videos previously 303 published by Metz and Whishaw [14,15] and supervised practice before using the foot fault 304 scoring system.

305
Despite being originally designed for rats, the ladder rung walking test can be used in 306 mice with some adjustments to the apparatus, namely a) the diameter of the rungs should be 307 reduced to allow a proper grip and paw placement; and b) the minimal and maximal between-308 rung interval should be changed, as previously described [13,24]. Our findings show these 309 adaptations are valid to obtain reliable results in C57BL/6 mice and may be valid for other 310 mice strains.

311
This study has some limitations. First, only two rodent strains were assessed.

312
Anyway, the current findings provide evidence of the accuracy and reliability of the foot fault 313 score in both Wistar rats and C57BL/6 mice. Second, we did not compare specific injury 314 models. Despite which, all individual scores (0 to 6) in the foot fault score were found in the 315 studied videos, which minimize this concern. 316 5. CONCLUSION 317 We conclude the foot fault score of the ladder rung walking test is a reliable and 318 useful tool to study walking adaptability in rodents. Moreover, experienced and 319 inexperienced raters can obtain reliable results if previous supervised training is provided.