Journal of Gerontology: MEDICAL SCIENCES 1990. Vol.45. No. I. M12-19 Copyright I WO by The Gerontological Society of America Gait Assessment in the Elderly: A Gait Abnormality Rating Scale and Its Relation to Falls Leslie Wolfson,1 Robert Whipple,1 Paula Amerman,1 and Jonathan N. Tobin2 Departments of 'Neurology and 2Epidemiology and Social Medicine, Albert Einstein College of Medicine, Bronx, NY. We evaluated the gait of 49 nursing home residents (27 of whom had a history of recent falls), and 22 controls. Measures consisted of stride length and walking speed, as well as a videotape-based analysis of 16 facets of gait. The study demonstrates that stride length, walking speed, and the assessment of videotaped gait correlated well with each other and were significantly impaired infallers compared to controls. Arm swing amplitude, upper-lower extremity synchrony, and guardedness of gait were most impaired in fallers. Although subjects who fell were more often demented than controls, it is likely that this represents a selection bias in nursing homes. Visual rating of gait features in the nursing home population is a simple and useful alternative to established methods of gait analysis. WALKING comprises repetitive movements that are programmed as coordinated patterns involving the extremities and trunk. The patterned ankle, knee, and hip movements that make up the swing (the foot is lifted and swings forward) and stance (the foot is planted and moves backwards relative to the trunk) phases of walking may be produced by a central pattern generator, which in cats is located in the spinal cord (1). This spinal cord activity is modulated by brainstem locomotor centers, basal ganglia, cerebellum, and motor cortex, as well as afferent input (proprioceptive, vestibular, and visual) (2). Locomotor con- trol mechanisms and their connections, widely distributed throughout the nervous system, may be compromised at many sites by a variety of disease processes. In normal aging many of these patterned movements become less vigorous and, therefore, stride length shortens and walking speed slows (3). In healthy 80-year-old people the changes are modest (20% decrease by comparison with younger individuals), but in house-bound elderly the decre- ments in stride length and walking are more prominent (4). The gait of elderly with a history of falls is also compro- mised, with decreased walking speed, shorter stride lengths, and variability in the length of successive steps by compari- son with others (5). The relationship of other gait abnormali- ties to falling has, however, not been studied in a systematic fashion. Furthermore, there is a need for a simple method of gait evaluation that is neither overly time-consuming nor requires costly instrumentation. This study assessed the validity of such a simplified Gait Abnormality Rating Scale (GARS) against the traditional indices of gait quality (stride length and walking velocity) in a group of nursing home residents with a history of falls, compared to a group of controls. We developed quantitative stride length and walk- ing speed criteria that may serve as predictors of future falls as well as demonstrating that the GARS is also a valid correlate of falls. METHODS Definition of a fall. ��� "Fallers" were defined as residents who had experienced at least two unexplained falls during the previous year. Unexplained falls are defined as endoge- nous in nature and not attributable to environmental hazards. Exogenous precipitants included environmental hazards such as surface, footwear, lighting, eyeglasses, and cloth- ing. The details of an unexplained fall, which was almost invariably unwitnessed, were confirmed through the incident report filled out by nursing staff on duty at the time of the fall. A research nurse facilitator then filled out a supplemen- tary form and further clarified the role that environmental hazards might have had in promoting the fall. Subjects. ��� Twenty of the fallers were enrolled over a two-year period from Morningside House Nursing Home, a facility with 196 skilled nursing facility beds and 190 health- related facility beds. We reviewed all 1324 incident reports filled out by a research nurse facilitator at the time of a fall during 1983 and 1984. A review of the reports indicated that 607 falls were explainable in terms of environment or medi- cal causes. The remaining 717 unexplained falls were cate- gorized according to person and date of occurrence. Subjects with two or more unexplained falls within one year (N = 137) were potentially eligible as cases. Of these, 117 failed to qualify because they did not meet the selection criteria. The most frequent reasons for exclusion were: untestable due to dementia (31%) and nonambulatory status (13%). The remaining 20 subjects were recruited. In addition, during the second year of the study another 7 subjects were recruited from the Margaret Tietz skilled nursing and health-related facility, using the same selection criteria. Twenty-two non- fallers were recruited into the control group from the Morningside House Nursing Home. Residents (fallers and controls) were excluded from par- ticipation if they met one or more of the following exclusion criteria: terminal illness behavioral or comprehension prob- lem severe enough to prevent the subject from carrying out simple test-related instructions (e.g., get up from the chair, walk, turn around, sit down, stand still, etc.) a nonambula- tory status, the required use of a wheelchair, walker, or orthosis amputations arthritis or orthopedic ailments of posture or gait (e.g., scoliosis, leg length discrepancies), or pain of an intensity that compromised the ability to stand M12 at UB der LMU Muenchen on August 8, 2012 http://geronj.oxfordjournals.org/ Downloaded from

GAIT ABNORMALITY RATING SCALE M13 unsupported, or to ambulate (antalgic gait) blindness (visual acuity worse than 20/70 or visual field defects) impairment of neurological function secondary to previously diagnosed neurologic disease (e.g., Parkinson's disease, stroke, myelopathy) history of orthostatic hypotension, seizures, transient ischemic attacks, syncope, or cardiac arrhythmias by EKG. Medical evaluation. ��� A comprehensive neurologic ex- amination was carried out by a neurologist who was part of the project. The examination was a structured evaluation of neurologic functions including: basal ganglia (6), cerebellar (7), sensory and pyramidal systems. A 5-level rating scale of severity of dysfunction (0 = normal, 1 = minimal, 2 = mild, 3 = moderate, 4 = severe) was used, and those subtests that most reflected each category of neurologic function were chosen for categorical analysis (bradykinesia, Parkinsonian gait, heel-knee-shin, dysdiadokinesia, large toe joint position sense, and vibration perception at ankle). A general impression (Yes-No) was also given regarding the presence or absence of "extrapyramidal" findings. The Blessed Mental Status (8), a measure of dementia severity, was also administered. A detailed medical history was obtained by structured interviews administered to subjects and by review of the medical charts by a nurse researcher. Medication usage was obtained from a review of the patient charts, which con- tained records of all medications taken by subjects. The types of medications coded included hypnotics, antihista- mines, tranquilizers, antidepressants, sedatives, and antihy- pertensives. A detailed physical examination for the presence of osteo- arthritis was performed by a nurse practitioner trained espe- cially for this evaluation by an orthopedic surgeon (9). Stride length and gait velocity assessment. ��� The gait course consisted of a 10m X lm length of hard-pile indus- trial carpeting with a slip-proof undersurfacing. The edges of the carpet were marked with short sections of reference tape at 0.5m intervals. Subjects were encouraged to wear their most comfortable pair of walking shoes, as long as heel height did not exceed two inches. Subjects were then asked to walk the length of the course and return at their usual walking speed. This was done twice in order to permit videotaping from both side and antero-posterior views. No cues were provided to subjects regarding the speed with which they were expected to walk. A cylindrical length of chalk (2" x 3") was then attached with tape to the rear of each shoe (blue on the left and pink on the right) so that the end of the chalk was in the plane of the heel. The subject was again videotaped when walking the course. For safety purposes one of the investigators walked beside the subject. Walking aids were not permitted during the trials, but physical assistance was extended, if necessary, in the form of contact guarding. Upon conclusion of the trial the chalk imprints on the carpeting were used as reference points for the measurement of stride length. The means of 10 left and 10 right strides were measured. Measurements were taken from the central three-quarters of the gait course in order to avoid inclusion of strides reflecting acceleratory and deceleratory phases. During a playback of the tape, each rater independently timed the duration of a subject's second gait trial with a stopwatch. The times were averaged and then divided by 10 meters (the distance tra- versed) in order to derive the gait velocity. Gait Abnormality Rating Scale (GARS). ��� An objective of this study was to devise a gait rating system that could be easily and quickly carried out in the clinic, and yet would require no added instrumentation expense beyond a standard videocamera and monitor. To that end, a rating system was developed that entailed evaluating gait from the videotaped recordings. On replaying the tapes, two judges (a neurologist and a physical therapist) independently rated the subjects' gait according to 16 variables (Table 1 A-C) using a 4-point scale (0 = normal, 1 = mildly impaired, 2 = moderately impaired, 3 = severely impaired). The GARS was calcu- lated by summing each of the individual variables. A higher score signifies a more impaired gait. The choice of gait components included in the GARS was made on the basis of prior studies describing gait in the elderly (3) and subjects with Parkinsonism (10). The determinations of "normal" and of "severely impaired" were also derived from the above studies, as well as from the personal clinical experi- ences of each rater. These extreme scores were easier to formulate than were the intermediate grades of "mildly" and "moderately impaired." The scoring was based on ranking of the extent of dysfunction of a particular variable compared to the normal. The criteria are defined in Table 1. Using six pilot subjects, the two raters participated in a series of video-rating training sessions in order to enhance consistency of scoring. The recordings were played back repeatedly in slow-motion and frame-by-frame modes while the raters discussed the scoring system. Scores at the oppo- site ends of a variable ("0" and "3") were easily agreed upon, while the intermediate grades were less clear-cut. Agreement was more difficult with gait components that varied along a continuum (e.g., variability or guardedness) than for those that were more easily distinguishable because of discernible reference points (foot contact, weaving, stag- gering). Once the study began, the raters were unaware of each other's assessments during the tape reviews, as well as to group identity of the subject. One of the raters (LW) had contact with some of the subjects during clinically relevant neurologic evaluations. Statistical analysis. ��� Comparisons of categorical data were done using corrected chi square and/or Fisher's exact tests. Continuous variables were compared by Mests. Inter- rater reliability for the qualitative gait rating was assessed using Spearman rank-order correlations. Although Kappa was designed as an index of assessment, it is most appropri- ate for nominal level data. Use of Kappa for ordinal and interval scale measures is inappropriate (11) and loses the ordering information contained in the scale. The degree to which rank order deviates from a perfect correlation (r = 1) demonstrates the rate of agreement and acknowledges "par- tial credit" for independent ratings that fall near each other on the ordinal scale. The Wilcoxon Signed Ranks Test was at UB der LMU Muenchen on August 8, 2012 http://geronj.oxfordjournals.org/ Downloaded from