Purpose: To verify the actual Partial Weight Bearing (PWB) after Limb Salvage Surgery (LSS) for bone tumours with respect to the orthopaedic surgeon’s prescription by a biofeedback device, and to evaluate if this device can help the patient adhere to the prescribed PWB.
Methods: Eight patients with primary bone tumours who underwent LSS and were prescribed PWB in post-surgery rehabilitation were included. The load was measured in a single experimental session under three different conditions:
- free walking (pre-conditioning gait phase),
- walk with PWB with the audio-feedback device (conditioning gait phase),
- walk with PWB after audio-feedback training (post-conditioning gait phase).
Results: During the pre-conditioning gait phase, seven patients exceeded the PWB by a mean of 42.2%. During the conditioning gait phase seven patients walked below the PWB threshold and one continued to exceed the prescribed PWB. During the post-conditioning gait phase, 4 patients continued to follow the prescribed PWB and 4 patients continue to exceed the PWB threshold, but to a lesser extent (25%) than during the free gait trial (53.7%). All the patients accepted the use of the device well.
Conclusion: Patients rarely observe PWB. The ability to control the correct prescribed PWB, once perceived by the device, is generally reproduced also in the short term without the audio-feedback. The use of an audio-feedback device is suggested to improve the patient’s adherence to the prescription.
Limb Salvage Surgery (LSS) using different skeletal reconstructions (i.e., allografts, special prostheses, and composite techniques) has become firmly established over time in the treatment of bone tumours due to biomedical innovations regarding materials and methods for the reconstruction of resected bone segments, and the introduction of multi-agent chemotherapeutic protocols, that reduce edema and neoplastic mass in the period before surgery [1,2]. The most recent polychemotherapeutic treatments not only allow limb salvage, but also improve patients’ survival and prognosis.
Among the known collateral effects that may appear after chemotherapy, particularly in young patients, growth retardation and bone mass reduction, including osteopenia and osteoporosis, are gaining more and more attention [3-10]. Patient’s age at the time of treatment, compromised nutritional intake , reduced physical activity levels  and also late onset of puberty [13,14] are involved in this process. Limb salvage techniques for bone tumours require a very careful progression of weight bearing on the treated limb after surgery to ensure implant stability and bone growth. The rehabilitation program usually consists of a compromise between an early stimulation of the living bone and a risky overloading which might cause complications such as fractures. A period of Partial Weight Bearing (PWB) is usually prescribed during the post-surgical period while chemotherapy cycles are administered both to allow the implant fixation in time and bone remodelling, to ensure a level of physical activity sufficient to stimulate osteoblast and thus prevent osteoporosis. This effect is crucial both in prosthetic replacements (to drive the bone anchorage to press-fit prosthetic stems) and in biological reconstructions (to facilitate the osteotomy fusion in massive allografts reconstructions or the hypertrophy of vascularised bone grafts) .
However the definition of partial weight bearing (also called “toe-touch weight bearing”) is broad and Rubin et al.,  showed that only 24% of surgeons were able to provide the toe-touch weight bearing definition. This uncertainty often creates doubts in patients and they have no way to check if their motor pattern is correct in observing PWB. Patients are usually taught how to walk with PWB by physical therapists by verbal instructions along with use of bathroom scales [16,17]. Previous studies have shown that patients are not able to walk within the limits of PWB without immediate feedback on actual loading and underloading or overloading often occurs [18-21].
Portable biofeedback devices have been developed to quantify and control objectively the actual load applied on the affected limb. Technical validation studies confirmed the good reliability of these systems and reproducibility of measurements [22-24]. Furthermore, several previous studies were conducted to describe and study the effectiveness of portable biofeedback devices in patients after orthopaedic surgery , fracture fixation , hip arthroplasty [26,27], knee surgery , amputation [29,30], hip arthroplasty , stroke , lower limb fracture [33,34], and patients referred for PWB gait rehabilitation [35,36], and have shown that PWB is generally not observed without biofeedback, particularly for low loading thresholds. However only one previous study  reported data on the effects of the biofeedback training for partial weight bearing loading at the removal of the device, and no studies were addressed to patients with musculoskeletal tumours, often young and with wide muscle removal.
The purpose of this pilot case study was to verify the actual partial weight bearing with respect to the surgeon’s prescription in patients treated by limb salvage surgery for lower limb bone tumours, by a biofeedback device, and to evaluate if this device is useful to educate the patient to observe the prescribed partial weight bearing. Feasibility problems encountered during the use of this assistive technology are discussed to understand how the device can be used to its best in clinical practice.
This feasibility study was performed on 8 consecutive patients (6 female, 2 male, mean age 32 years, range 9-62) referring to the rehabilitation outpatient clinic of Rizzoli Orthopaedic Institute. All patients underwent LSS (2 allografts, 5 prosthesis, 1 pubic-femoral reconstruction) in seven cases for osteosarcoma (one patient at the pelvis, two in the distal femur, four at the proximal tibia), and in one case for chondrosarcoma in the proximal femur. Walking recovery with PWB was prescribed to them by the orthopaedist (Table 1) during the surgical follow up. Inclusion criteria were: ability to walk with crutches for more than 20 m, no musculoskeletal problems in other districts, no comorbilities, no cognitive impairment. At the time of the pilot study all patients were under chemotherapy treatment. However all of them were in a period of recovery from collateral effects after a cycle of chemotherapy, and no symptoms (fatigue, fever, nausea, sedation from medicaments etc.) were present. No patient presented peripheral neuropathy due to chemotherapy. A rehabilitation program for gait recovery with the prescribed partial weight bearing was ongoing at the time of the enrolment.
||Follow-up from surgery (days)
||Body Weight (Kg)
||Prescribed PWB (Kg)
||Prescribed PWB (% BW)
||Device Thresholds (Kg)
The commercial biofeedback device was the Pedar mobile system (NovelGmbH, Munich, Germany), which uses flexible and size-adaptable insoles to measure plantar pressure inside the shoes. Each insole contains 99 capacitive sensors.
During data collection, the patient wears a belt containing the device box, the battery and a start/stop button. The insoles are connected to this box with cables fastened to the legs with Velcro straps.
The software of the device is designed for long-term mobile measurements without direct connection to a computer. The computer is needed to configure the internal memory device and download any data which was stored inside the flash memory of the device during data collection. This software allows two threshold load levels to be defined, usually one upper and the other lower. This is possible in only one of the two feet. The device can also provide two different audio-feedback tones when insole sensors detect loads in excess of those set as thresholds.
The measurements start with a zero deduction, which sets all insole sensors to a zero value. After this, the patient pushes the start/stop button and the measure begins. The same button is used at the end of the procedure to stop data collection.
The study included the quantitative instrumental evaluation through the Pedar device as the main outcome, and a qualitative assessment on patient satisfaction regarding use of the device as the secondary outcome. The Medical Ethical Committee of the institution approved the study protocol. All participants (or their curators) gave their written informed consent to take part in this feasibility study. Then they placed the Pedar insole device correctly inside their own comfortable shoes, and attached the battery, and the necessary cables (Figure 1).Figure 1:
A young patient wearing the device.
Four phases followed:
- Pre-conditioning Gait Phase: a walk of at least twenty steps on a 7-meter-long corridor during which the patient repeated the walking pattern that had been learned during the PWB training period with physiotherapist.
- Weight definition Phase: how the device worked was explained to the patient as follows. The first audio-feedback would have been heard when the load established as the lower threshold was exceeded and a possible second audio-feedback would have been heard when the load established as the upper threshold was also exceeded. Obviously the second audio-feedback means that the patient was not observing the prescribed PWB. Therefore, it was desirable to hear only the first sound.
- Conditioning Gait Phase: a walk of at least twenty steps during which the patient received audio-feedbacks to help observe the PWB.
- Post-conditioning Gait Phase: a walk of at least twenty steps without audio-feedback in which the patient repeated the walking pattern that had been learned during the previous phase.
As a secondary outcome a non-structured satisfaction questionnaire with the following open questions was administered to the patient to investigate the level of satisfaction and usefulness of the device in their opinion:
- Did you appreciate the experimental session with the biofeedback device?
- Did you tolerate the device well?
- Do you think that the use of the device is helpful? Explain why.
Based upon recommended procedures  data from the initial and the last step were discarded. Furthermore, to carry out a more detailed analysis and the comparison of data among different conditions, data were exported to an excel file. The maximum load value for every step was hence identified, and the mean of all values was calculated for each condition (i.e., post-conditioning gait, conditioning gait, post-conditioning gait) to check:
- The correspondence between the prescribed PWB and the real load put on the affected limb during free walking.
- If conditioning with Pedar device enables patients to learn and put into practice the senso-motorial pattern required for walking with PWB.
- If the ability to control the limb loading acquired when using the device is maintained in the short term without audio-feedback.
As for the statistical analysis the paired samples t-test was used to analyse differences among mean values of prescribed PWB, free walking load, walking load with device, and walking load without device. A binomial test was also carried out to check if the proportion of cases having a 25th percentile under and a 75th percentile above an ideal threshold (almost 20% of prescribed PWB) was significantly different from that expected. Statistically significant values were considered for p
The mean number of steps was 34.2 ± 9.6, 38.2 ± 11.8 and 30.6 ± 11.4 respectively for the three gait trials with device.
During the pre-conditioning gait phase, 7 patients walked with a higher load on the treated limb with respect to the prescribed PWB. One patient did not reach the prescribed PWB (mean value -9.6%) (Table 2).
(number of steps 34.2 ± 9.6)
(number of steps 38.2 ± 11.8)
|Post conditioning gait
(number of steps 30.6 ± 11.4)
||Prescribed PWB (Kg)
Data on actual loading in the three gait conditions with respect to prescribed PWB.
During the conditioning gait phase, 7 patients were able to walk with a load below the maximum threshold; one patient continued to overload the treated limb, but with a reduced maximum mean value with respect to that of the pre-conditioning gait phase (28% instead of 43.3%). During the post-conditioning gait phase 4 patients continued to remain under the prescribed PWB while the other 4 patients returned to overload the treated limb, but to a lesser extent.
Figure 2, graphically summarizes the loading values in the three conditions with respect to the PWB. In general, the value of limb loading in the three walking conditions changed on average from 23.5% in the pre-conditioning gait phase, to -13.9% during the conditioning gait phase with the device, to 7.9% in the post-conditioning gait phase, without the device.Figure 2:
Median values and confidence intervals of PWB in Kg in each gait condition, examined with paired t-test.
The paired samples t-test revealed statistically significant differences between the prescribed PWB and the load value registered during the pre-conditioning gait phase (p=0.025), the load value during the pre-conditioning gait phase and the load value during the conditioning gait phase (p=0.008) as well as the load value registered during the conditioning gait phase and the post-conditioning gait phase (p=0.019). No difference was found between the load value during the PWB and the post-conditioning gait phase (Table 3).
||Confidence interval 95%
|Pre-conditioning Gait Phase vs prescribed PWB
|Pre-conditioning Gait Phase vs Conditioning Gait Phase
|Conditioning Gait Phase vs
Post-conditioning Gait Phase
|Post-conditioning Gait Phase vs prescribed PWB
Differences among the three gait conditions.
Finally, the binomial test, performed to check if the proportion of cases that had a 25th percentile under and a 75th percentile above an ideal threshold of ± 3 Kg (almost 20% of prescribed PWB) showed that the mean load value was not significantly below the values of -3 Kg in any of the three trials. Only the mean load value during the pre-conditioning gait phase was significantly above the ideal threshold of +3 Kg (p=0.004) (Figure 3).Figure 3:
Median values differences (delta) and confidence intervals of PWB to check if the proportion of cases that had a 25th
percentile under and a 75th
percentile over an ideal threshold of ± 3 Kg (almost 20% of prescribe PWB) by the Binomial Test. Only the mean loading value during the pre-conditioning gait phase was significantly (p=0.004) over the ideal threshold of +3 Kg.
Concerning the satisfaction of the patients with respect to the use of the device, five patients highly appreciated the experimental session with the device, and three patients said it was useful.
With respect to the device used during the experimental session, two patients stated that the waist belt of the device was too heavy and one of them (the youngest one) also reported that the belt size was too large. One patient said that the two tones of audio-feedback should be more differentiated. Three patients said that the device should be worn at home, and one patient added that the device should be worn during the rehabilitation sessions. Seven patients stated that the use of this device was helpful, because it allowed them to understand the actual load applied on the affected limb thanks to audio-feedback. Only one patient said it is not very useful if its use is not continuous, because having finished the test, the patient finds himself once again without device and unable to know how much load is put on the affected limb.