Rehabilitation after lower extremity injury or surgery often requires limited weightbearing on the injured limb. In the case of serious traumatic injury or chronic disease of the limb, inability to maintain proper weightbearing restrictions may lead to chronic pain, malunion, disability, or amputation. One of the most common ways to teach patients to limit their weightbearing is to have them place their injured limb on a scale and apply pressure until the desired force limit is learned. Several studies indicate that this method is not reliable.6,11,12 In patients with neuropathy, who lack normal nociceptive feedback, learning to estimate weightbearing is difficult.4
Two types of feedback in gait training are common: knowledge of results feedback and concurrent or augmented feedback. In the first type, knowledge of results feedback, the subject is given feedback from a trainer regarding success in completing the motor task after attempting to limit weightbearing to a set level. In the second type, concurrent or augmented feedback, the feedback occurs simultaneously with the movement and is extrinsic to the subject’s sensory organs, such as an auditory alarm that sounds immediately when a set pressure is exceeded. Using an acoustic alarm to augment partial weightbearing instruction and weight shifting practice onto a scale potentially would aid in gait training, especially for subjects with altered limb sensation or increased risk from excessive weightbearing.
Previous studies have indicated the potential value of other types of concurrent feedback in gait training, although limitations, such as overshooting the weightbearing goal10 and the lack of retention after removal of the feedback,10,12 have been described.6,7,10,11,13
The purpose of this study was to ascertain whether a weightbearing goal could be better achieved with a pressure-sensitive alarm than without the alarm and to compare average peak force (APF) in each group. Four subject groups were evaluated while attempting to limit weightbearing to 20 lb (9 kg) or less with and without the pressure alarm. Results were compared in each group. The limit of 20 lb was chosen because this is a typical target for limited weightbearing at our institution. This investigation also included evaluating the alarm for durability and ability to activate under increased compression frequency.
METHODS AND MATERIALS
The new pressure-sensitive alarm, the Post-Op Pressure Alarm (Aircast, Inc, Summit, NJ), consists of a small, oval-shaped air bladder that measures 3 × 4 inches and is ½ inch thick. A small air hose attaches the air bladder to a valve and whistle device. The valve and whistle are enclosed in a cuff that straps around the ankle. A whistle sounds when a set amount of force is applied to the air bladder. The device was evaluated at 9 kg (20 lb), 25 kg (55 lb), and 40 kg (88 lb).
In the subject trials, the pressure alarm bladder was placed under the shoe and attached with adhesive-backed Velcro tabs. The pressure alarm location was adjusted as needed if the subject bore weight more toward the midfoot or forefoot than in the hindfoot to ensure proper contact. Placement of the alarm bladder under the shoe rather than inside the shoe also allowed for maximum accuracy of in-shoe pressure measurement.
One concern with such an alarm is possible loss of activation after extended use or at increased rates of gait. A limited set of fatigue tests was done to evaluate the durability of the pressure alarm at force settings of 9 kg, 25 kg, and 40 kg.
Six alarms (two at each force setting) were cyclically loaded in an MTS 858 Mini Bionix load frame (MTS Systems, Eden Prairie, MN). One alarm of each force setting was tested at a frequency of 1 Hz, and the second alarm at that force setting was tested at a frequency of 2 Hz, which simulated an increased pace of gait. All alarms were cycled 600,000 cycles. The amount of deformation to cause the alarm to activate was monitored as a function of its fatigability.
From October 4, 2000, to March 31, 2001, 28 consecutive consenting subjects were enrolled in the current study with institutional review board approval. Exclusion criteria included foot ulceration, infection, active Charcot neuroarthropathy, upper extremity weakness, pain, or other conditions making partial weightbearing difficult or impossible.
The 28 subjects were divided into four groups. Group 1 consisted of seven healthy subjects with a mean age of 33 years (range, 26–41 years). Group 2 included seven older adults with a mean age of 59 years (range, 51–75 years) without neurologic or musculoskeletal problems. No age cutoff was predetermined to delineate these groups. However, because a gap in age (41–51 years) occurred randomly, this gap was used to define the two groups with normal sensation. A delineation in the normal groups based on age was considered desirable based on our experience with older patients who have difficulty achieving weightbearing limits. Group 3 included six older subjects with a mean age of 56 years (range, 53–64 years) and without protective sensation as measured with a Semmes-Weinstein 5.07 (10 g) monofilament.3 Group 4 consisted of eight subjects with unilateral transtibial amputation and a mean age of 39 years (range, 24–73 years). These subjects were regular prosthesis users who had protective sensation in their residual limb.
Subjects in all groups were fitted with a postoperative shoe on one foot. In Group 4, this postoperative shoe always was used on the prosthetic foot. The postoperative shoe was used to normalize the surface for the in-shoe sensor and the pressure alarm. A comfortable walking shoe or athletic shoe was used on the opposite side. A licensed physical therapist instructed all subjects in partial weightbearing ambulation using a walker until they were able to show independence in the technique. No obvious postural alteration or gait abnormality was seen that was attributable to the pressure alarm and postoperative shoe combination in comparison to partial weightbearing to 20 lb with a matched shoe. Although full weightbearing on a postoperative shoe produces asymmetry in gait, the limited amount of weight being applied to the limb in this study most likely was the reason for the more typical appearance of gait mechanics.
After gait training, subjects practiced weight shifting for 2 minutes with a dual beam scale set at 20 lb to learn to judge 20 lb of force. A wooden step was used to level the surface between the floor and scale. The walker height was adjusted as necessary to provide support during the practice session on the scale. The foot with the postoperative shoe was placed on the scale, and the other foot was on the matched-height step. Subjects practiced weight shifting side to side for the first minute. During the second minute, the step and scale were offset from each other with the step 12 inches behind the scale. Subjects were asked to shift their weight in a forward and backward direction to simulate walking. Each subject then was weighed to allow calibration of the pressure sensors. For the purposes of this study, any weightbearing under 21 lb was considered to meet the weightbearing limit.
The F-scan (Tekscan, Inc, Boston, MA) in-shoe sensor was used to measure peak force for each footstep during each trial. The in-shoe sensor is 0.18 mm thick with 960 individual pressure-sensing locations before trimming to fit the postoperative shoe. This system has been evaluated for reliability in several studies1,8,9 and the reported results show good reliability of data when a single sensor is used per subject in the same shoe. Increased variability has been seen when different shoes or different sensors are used in comparing data and with prolonged use of a single sensor. In the current study the same type of shoe (a postoperative shoe) and a single sensor were used for each subject. Sensors were evaluated for effects of wear using the manufacturer’s software to analyze saturation pressures for each sensor.
A properly sized postoperative shoe was fit for each subject and the sensor was trimmed with scissors to fit inside the shoe. The tab of the sensor extends laterally above the ankle and attaches to a cuff unit that gathers and processes data from the sensor for transmission to the computer. The cuff unit is held in position by a Velcro ankle band and a shielded cable carries pressure information from the cuff unit to the computer. A waist belt was used to hold the cables in position and allow freedom of movement. The in-shoe sensor was calibrated using the subject’s known weight in accordance with manufacturer specifications.
The pressure alarm air bladder was attached to the center of the bottom of the shoe with Velcro. Two trials were done. In the control trial, a deactivated pressure alarm was applied to the bottom of the shoe to normalize the walking trials. Subjects were informed that this alarm would not sound but was applied to the shoe to make both walking trials feel the same. Subjects were shown starting and ending points that were 30 feet apart. Just before beginning the walking trial, the partial weightbearing sequence was reviewed and subjects again were encouraged to limit weightbearing to 20 lb or less. Subjects then ambulated with the walker for 30 feet while force data were collected. No feedback was given to the subjects after the initial trial. After a 3-minute rest period, a second trial was done. In the second trial, the deactivated pressure alarm was replaced with an activated alarm set to sound at 20 lb and force data again were collected. In both trials, subjects were instructed to walk flat on the postoperative shoe to ensure that the pressure was borne through the pressure alarm. The trials were not randomized because it was not possible to obtain accurate data on standard gait training if the order of the trials was reversed. If the standard gait training trial was second, it would be a measure of retention from the previous concurrent feedback trial. The only way to evaluate true standard gait training was to have no prior feedback.
For the alarm deformation aspect of the current study, a one-way analysis of variance was used to analyze any observed differences in the amount of deformation to cause pressure alarm activation initially and the amount to cause activation after 600,000 cycles.
For the part of the study involving subjects, peak force data were compiled for each footstep and an APF was determined for each trial and each subject. The first and last footstep were omitted from data analysis to avoid including data on partial steps and as a way to normalize the data for velocity. Because data were distributed normally, a two-tailed t test was done. The level of significance in this study was set at p ≤ 0.05.
The amount of deformation to cause activation of the alarm at 1 Hz ranged from a minimum of 0.1 mm (for the 9-kg alarm) to a maximum of 0.58 mm (for the 25-kg alarm) tested at 1 Hz. For the alarms tested at 2 Hz, the deformation ranged from a minimum of 0.1 mm (for the 9-kg alarm) to a maximum of 0.2 mm (for the 40-kg alarm). The differences were not significant (p = 0.99) (Fig 1). After 600,000 loading cycles at the respective test loads, the alarms still were fully functional, emitting an audible alarm when the preset loads were exceeded.
In Group 1, five of the seven subjects were able to limit weightbearing to 20 lb in both conditions (with the activated alarm and with the deactivated alarm). One of the remaining two subjects in Group 1 met weightbearing goals with the activated alarm but not with the deactivated alarm. The overall success rate for these young, healthy subjects was 86% with the alarm and 71% with standard training alone. When using the activated alarm, 43% of these subjects significantly reduced their APF in comparison with the deactivated alarm trial (p = 0.02). The APF for Group 1 with the activated alarm was 10.01 lb (range, 2.22–33.02 lb) versus 13.02 lb (range, 5.13–24.72 lb) with the deactivated alarm (weight training alone).
In Group 2, which included seven older adults with a mean age of 59 years without neurologic or musculoskeletal problems, only one subject was able to achieve weightbearing goals in both conditions. Three subjects in Group 2 were able to limit weightbearing to 20 lb or less with the activated alarm versus deactivated alarm. This represents a success rate of 57% with the activated alarm versus 14% with the deactivated alarm. In this group, 86% of the subjects significantly reduced their APF with the pressure alarm versus standard training alone (p = 0.02). The APF for Group 2 was 31.30 lb (range, 6.20–88.03 lb) with the pressure alarm versus 67.16 lb (range, 16.97–150.08 lb) with the deactivated alarm (standard training alone).
In Group 3 with six older subjects with a mean age of 56 years and without protective sensation, only two (33%) were successful in limiting weightbearing to 20 lb or less when using the activated pressure alarm. No subjects were successful with the deactivated alarm (standard training alone). However, all six subjects (100%) in this group reduced their APF when using the activated alarm. For the group, the APF was 29.78 lb (range, 8.86–47.72 lb) with the activated alarm versus 61.01 lb (range, 11.16–87.53 lb) with the deactivated alarm.
In Group 4, of the eight subjects with transtibial amputation and a mean age of 39 years, only two ambulated with an APF of 20 lb or less with the activated alarm and none achieved the weightbearing goal with the deactivated alarm. When using the activated alarm, 100% of subjects in Group 4 significantly reduced the APF (p = 0.02). For Group 4, the APF was 32.23 lb (range, 14.51–67.08 lb) with the activated alarm versus 64.96 lb (range, 42.52–113.11 lb) with the deactivated alarm (standard training alone).
The ability to achieve limited weightbearing compliance is critical to healing or survival of the limb in the most compromised patients, and this goal is difficult for most patients to achieve. Additionally, the outcome for patients most at risk may be jeopardized severely by one episode of weightbearing that exceeds the limit. Noncompliance with weighbearing limits in older patients or patients with neurologic disorders or amputation may lead to catastrophic results such as infection, nonhealing wounds, surgical failure, and amputation. The use of concurrent (auditory) feedback with alarms may have an important indication in this patient population.
The current study examined the efficacy of concurrent (auditory) feedback to help patients in four different groups to learn to consistently achieve partial weightbearing goals. The study found that performance in each group improved with use of the activated alarm compared with standard weightbearing training alone. The APF was reduced in all four groups and significantly reduced in Groups 1, 2, and 4. Weightbearing compliance consistently improved with use of the pressure alarm in comparison with gait training and practice on a scale alone. None of the subjects in this study with neuropathy or transtibial amputation succeeded in limiting weightbearing to 20 lb without the alarm. Younger patients did relatively well without the alarm (71% success rate), but the success rate in older subjects with normal sensation was only 14% without the alarm. Other studies evaluating the ability of subjects to learn a specific weightbearing task have found similar results.2,4,5 Bohannon and Kelly4 found a discrepancy between the instructed amount of weightbearing and actual weightbearing in subjects with neuropathy and amputation.
Research suggests that concurrent feedback might be a useful clinical tool for achieving partial weightbearing goals. Wannstedt and Herman10 found that 77% of 30 patients with hemiparesis and asymmetric standing posture were taught to achieve standing balance using an auditory limb load monitor. These patients had a stroke 6 or more months before the study and had completed physical therapy. Additionally, 80% of the patients who achieved symmetric standing were able to stand symmetrically without the device after an average of 11 daily sessions with the device. Pataky et al7 used an auditory foot pressure device. This insert contains numerous sensors that continuously record in-shoe plantar pressures for as many as 8 days and provides auditory feedback when the subject exceeds the preprogrammed critical pressure. They described a case in which peak plantar pressure was reduced in a patient with diabetes and a plantar neuropathic ulcer. In the case study, the patient’s ulcer size was reduced by 9 mm in 2 weeks.
One potential drawback of partial weightbearing training with the auditory alarm is the tendency for patients to overshoot the weightbearing goal. In the current study, Groups 1, 2, and 4 had a statistically significant decrease in APF with the activated pressure alarm compared with standard training alone. Despite the marked decrease in APF in these groups with the activated alarm, the subjects still overshot the weightbearing goal by approximately 50%. In a study that compared traditional partial weightbearing training with a scale and an augmented auditory feedback system in 10 healthy female college students, Warren and Lehmann also found that the target load often was exceeded by 50% or more even with auditory feedback.11 These authors suggested that this finding might reflect a delay in motor response after the auditory signal and may be a limitation inherent to this type of feedback.
Retention of learning also may be more efficient with knowledge of results or immediate feedback from a clinician compared with auditory feedback from an alarm when the weightbearing limit is exceeded. In a study by Gray et al,6 61 healthy volunteers were separated into three groups and taught to limit weightbearing to 60 lb using a force platform, a scale, or a method in which the subject placed his or her foot on the hand of a licensed physical therapist, who estimated what the desired force should feel like. All groups received concurrent (auditory) feedback during the training period, when they practiced weight shifting. Only subjects in the platform group received a review of their performance immediately after the fact (knowledge of results feedback). All groups then were analyzed while walking on a force platform without feedback. The platform group had a significantly higher percentage (90.5%) of subjects who were within the desired 30-lb range compared with subjects using the hand (31.6%) and scale (33.3%) methods. Limitations in the study design did not allow definitive conclusions, but the study suggests that knowledge of results feedback might have a substantial effect on learning to limit weightbearing.
In a study comparing concurrent feedback with knowledge of results feedback for learning partial weightbearing ambulation, Winstein et al13 concluded that concurrent (alarm) feedback promotes accuracy and consistency of performance when in use but that accuracy does not continue when the feedback is removed. These investigators suggested that knowledge of results is more effective than concurrent feedback for partial weightbearing training.
Because knowledge of results feedback is permissive of exceeding weightbearing goals in the skill acquisition phase, the clinician is not able to consider this learning method for patients in whom exceeding these goals could have a devastating effect. The pressure alarm may provide the margin of safety needed in the clinical setting, especially with patients whose status might be severely compromised by even brief inadvertent excessive weightbearing on the affected limb. The expected overshooting of the weightbearing limit may be compensated for by decreasing the weightbearing limitation to take the overshooting into account. For example, a patient who could safely apply 30 lb onto the limb may be given a weightbearing limit of 20 lb when using the alarm. Use of the alarm could be continued until the weightbearing limit is increased or until full weightbearing is permitted. Based on the findings of the current study, the pressure-sensitive alarm can be used to achieve improved weightbearing adherence throughout the limited weightbearing period.
Although the current study found that concurrent feedback was more effective than standard gait training in a single trial, the small study numbers limit the ability to make definitive recommendations for use of this learning method. Additionally, one of the three groups in the study (Group 3, older patients without protective limb sensation) was too small to allow statistical analysis of significance, based on a power analysis done after completion of the study. Additional studies are needed to assess the effectiveness of concurrent feedback with a larger study group.
It is desirable to evaluate performance with this learning method during longer periods because patients may become accustomed to the audible feedback and ignore or remove it if it becomes an annoyance. Additional research is required to assess overshooting the weightbearing goal and to develop strategies to compensate for this phenomenon. The efficacy of placement of the alarm bladder in or under the shoe may be another variable worthy of consideration.
Finally, future studies should compare a control group with a concurrent feedback group and a knowledge of results feedback group. If possible, these studies should be done in a clinical setting with retention tests to establish learning effects with time.