The study subjects consisted of patients who visited the rehabilitation outpatient clinic of our medical center between July 20, 2015 and March 8, 2016. Patients with unilateral or bilateral lower-limb amputation and Syme amputation or higher who had been wearing prostheses for longer than 6 months were considered for the study. Among these patients, those who complained of low back pain that lasted for more than 3 months and agreed to participate in the study were selected prospectively. Patients who had spine surgery were also considered, and those who were able to walk with an activity level of K1 or higher were selected. The exclusion criteria were: inability to walk with an activity level of K0; inability to perform exercise due to cognitive impairment; hemiplegia and quadriplegia due to stroke or traumatic brain injury resulting in loss of the ability to walk or balance; upper limb amputation; and exercise treatment for low back pain at another hospital. The participants provided written informed consent and the protocol was approved by the Ethics Committee at Veterans Health Service Medical Center (IRB No. BOHUN 2015-05-007).

The goal of the exercise program was to strengthen the deep lumbar stabilizing muscles (transversus abdominis, lumbar multifidi, and internal obliques). The program was run twice a week for 8 weeks, with a total of 16 sessions. The exercise duration was 30 minutes including: 5 minutes of warming up, 20 minutes of the main exercise, and 5 minutes of cooling down. The exercise was performed under the supervision of a rehabilitation doctor and two physical therapists, in a group exercise format. Each warm-up and cooling down exercise was performed for 5 minutes within a range of motion without pain and included free gymnastics as well as static and dynamic stretching. The exercise program was developed by analyzing the characteristics of lower-limb amputees that differ from those of regular people, selecting exercises that could be performed by the amputees (Fig. 1). The program was based on the lumbar stabilization exercises used in the studies by Aluko et al. [10] and Moon et al. [4]. Among the participants included in this study, a model patient was selected to test the feasibility of the exercise. Twelve of the 14 exercises were separated into four sets of 3 exercises each and one set was performed in each session. The other 2 exercises, hollowing of the lower abdomen and lifting the trunk while prone, were performed at the beginning of each session. The first set of exercises was performed twice in the first week, and the second set was performed twice in the second week, and so on. After the 4th week, the exercises were repeated in the same order, resulting in a total of 16 sessions. During each exercise, the last movement was held for 10 seconds and this process was repeated five times [11]. Normal breathing was maintained during the exercise [12]. The therapist provided detailed verbal instructions for each routine beforehand, and the participants performed the exercises while watching the instructional video. All the subjects participated fully in the exercise program and none of them complained of aggravated low back pain or other discomfort.

A total of two evaluations were conducted; the first was carried out 1 week before the beginning of the program, and the second after completion of the 8-week program. To evaluate the severity of low back pain, the visual analog scale (VAS) scores were assessed. The Korean version of the Oswestry Disability Index (K-ODI) was used to evaluate the functional disability and quality of life [13]. The Thomas test was used to measure the iliopsoas muscle length in a physical examination [14]. When a patient pulled the back of one thigh towards their chest and lowered the other thigh passively, the angle between the thigh and the ground was measured. A trunk-raising test was used to measure abdominal muscle strength [6]. The trunk-raising test was graded according to the system, which states that the applied load varies according to the position of the arm when the examinee lifts the trunk while lying down. Back muscle strength was measured using the prone-lying trunk-raising test [6]. For stabilization, a strap was tied to the buttock and tibial region, and as needed, the examiner pressed on the patient's upper-thigh region for extra stabilization. Scores ranged from 0 to 5. All the examiners were instructed to successfully attempt a grade 3 test first, followed by other grades. Back extensor muscle endurance was evaluated using the Sorensen test [14]. Patients adopted a prone position, and the lower limbs were tied with a strap with the superior side of the iliac crest. First, stabilization was used while the examinees' arms rested on a chair in front of the bed. After starting the test, the examinees were instructed to lift their arms from the chair and lift their upper body into a horizontal position. The duration of stable position was recorded with a stop watch. Extra stabilization was achieved as the test proceeded while the examiner pressed on the patient's upper thigh. Physical examination of the affected and unaffected sides was carried out after removing the prosthesis. All the physical examinations were blindly conducted by two experienced rehabilitation doctors.

For objective evaluation of muscle strength, the Cybex (CSMi, Stoughton, MA, USA) examination equipment was used to measure the peak torque and total work at the angular velocities of 60°/s and 120°/s [15,16,17,18]. The examinees stood on the foothold and the region where the extension line of the iliac crest met the spine was identified as lumbar 4–5. Using this as a reference, the height of the foothold was adjusted so that the axis of rotation was positioned between lumbar 5 and sacral spine 1. The waist was firmly stabilized with a pelvic girdle, and the height of the popliteal pad was adjusted. Both lower limbs were safely stabilized with femoral and tibial pads, and a sacral pad was moved from the front to the rear so that the axillary line passed through the center of the rotational axis. The scapular pad was positioned at the center of the scapula and connecting links at both ends of the chest pad were attached to the scapular pad to completely stabilize the upper body. The patients were instructed to hold the handle in front of the chest pad with both hands. Before the examination, preliminary tests were performed three times at each angular velocity so that the patients were familiar with the isokinetic strength test [16,18].

To analyze the results, statistical analysis was conducted using SPSS version 18 (SPSS Inc., Chicago, IL, USA) for Windows and the Wilcoxon matched-pairs signed-rank test was performed to comparatively analyze the results before and after the exercise. The Mann-Whitney U-test was used to compare the results of clinical and isokinetic muscle strength tests after the exercise between the bilateral and the unilateral amputation groups based on the differences before and after exercise. Values were presented as mean±standard deviation, and p<0.05 was considered statistically significant.

All 19 subjects who visited the rehabilitation outpatient clinic for chronic low back pain were enrolled prospectively. Mean age was 63.9±7.4 years and the duration after amputation was 39.6±7.5 years. Mean K-level was 2.6±0.6 and the cause of amputation was trauma in all the patients. Amputation was performed on both sides for 3 patients, on the right side for 5 patients, and on the left side for 11 patients. The bilateral procedures included right and left transfemoral, right transfemoral and left transtibial, and long transfemoral amputations. Among the unilateral amputations, 5 were transfemoral, 1 was knee disarticulation, 9 were transtibial, and 1 was Syme amputation. The general characteristics of the subjects are summarized in Table 1.

VAS showed a statistically significant decrease after 8 weeks of exercise. The K-ODI showed a decrease after 8 weeks of exercise but the difference was not statistically significant (p=0.343). Iliopsoas muscle length showed no significant differences in both amputation and intact sites. Abdominal muscle strength and back extensor strength were significantly increased after the exercise (p<0.05). Back extensor endurance also showed an increase from 22.3±20.7 seconds before the exercise to 46.8±35.1 seconds after the exercise with a statistical significance (p<0.001) (Table 2).

The respective peak torque of the lumbar spine for the flexor and extensor muscles before the exercise was 146.4±76.7 Nm and 32.4±219.7 Nm at 60° of angular velocity, and 123.4±81.9 Nm and 35.3±23.5 Nm at 120° of angular velocity (Table 3). After the program, the respective peak torques were 208.0±61.2 Nm and 72.8±24.0 Nm at 60° of angular velocity, and 189.4±62.0 Nm and 56.8±31.3 Nm at 120° of angular velocity, suggesting a statistically significant increase in peak torques of the flexor and extensor muscles at 60° and 120° of angular velocity. Total work at the lumbar spine for the flexor and extensor muscles at 60° and 120° of angular velocity also showed a statistically significant increase (p<0.05).

The mean value of the VAS, abdominal muscle strength, back extensor muscle strength, back extensor muscle endurance and isokinetic muscle strength test was improved in both bilateral and unilateral amputation groups. However, there were no significant differences before and after exercise between bilateral and unilateral amputation groups.