Secondary Logo

Comparison of Lower Limb and Back Exercises for Runners with Chronic Low Back Pain


Medicine & Science in Sports & Exercise: December 2017 - Volume 49 - Issue 12 - p 2374–2384
doi: 10.1249/MSS.0000000000001396

Introduction This single-blind randomized trial was conducted to compare the treatment effect of lower limb (LL) exercises versus conventional lumbar extensor (LE) and lumbar stabilization (LS) exercises in recreational runners with chronic low back pain (cLBP), because there is currently no specific protocol for managing runners with cLBP.

Methods Eighty-four recreational runners with cLBP were allocated to three exercise groups (LL, LE, LS) for an 8-wk intervention. Outcome measures included self-rated pain and running capability, LL strength, back muscle function, and running gait. Participants were assessed at preintervention, mid-intervention, and end-intervention; selected outcomes also followed up at 3 and 6 months. Generalized estimating equation was adopted to examine group–time interaction.

Results The LL group improved 0.949 points per time point in Patient-Specific Functional Scale (P < 0.001), which was higher than the LE (B = −0.198, P = 0.001) and LS groups (B = −0.263, P < 0.001). All three groups improved on average 0.746 points per time point in Numeric Pain Rating Scale for running-induced pain (P < 0.001). Knee extension strength increased 0.260 N·m·kg−1 per time point (P < 0.001) in the LL group, which was higher than the LE (B = −0.220, P < 0.001) and LS groups (B = −0.206, P < 0.001). The LL group also showed a greater increase in running step length (2.464 cm per time point, P = 0.001) compared with LS group (B = −2.213, P = 0.013). All three groups improved similarly in back muscle function.

Conclusion LL exercise therapy could be a new option for cLPB management given its superior effects in improving running capability, knee extension strength, and running gait.

1Physical Education and Sports Science Academic Group, National Institute of Education, Nanyang Technological University, SINGAPORE; and 2Physiotherapy, Rehabilitation Department, Ng Teng Fong General Hospital, Jurong Health Services, SINGAPORE

Address for correspondence: Pui Wah Kong, Ph.D., Physical Education and Sports Science Academic Group, National Institute of Education, Nanyang Technological University, 1 Nanyang Walk, Singapore 637616; Email:

Submitted for publication March 2017.

Accepted for publication July 2017.

The prevalence of chronic low Back Pain (cLBP) among recreational runners was reported as high as 13.6% in the United States (56). Low back injuries accounted for approximately 7% of all running injuries (15), but there is no runner-specific exercise therapy protocol in the current management of cLBP. The clinical management of runners with cLBP is largely based on protocols for the general population which were built upon the understanding of fatigability (28,47) and strength deficit (16,23) of the lumbar extensor (LE) muscles, and/or compromise in motor control of the lumbar stabilizing muscles (12,22). Unfortunately, randomized trials investigating the treatment effect of LE and lumbar stabilization (LS) exercise therapies have shown inconclusive results in managing cLBP in the general population (20,26,35). On the other hand, lower limb (LL) muscle impairments associated with cLBP, such as knee extensor inhibition and reduction in hip extensor activation and fatigability, are often reported in this population (24,31,52). More specific to runners, one recent study showed that recreational runners with cLBP exhibited diminished knee extensor strength compared with asymptomatic runners (4). The authors postulated that weakness of the knee muscles may reduce capacity for shock attenuation, transmitting higher forces to the low back during running. This suggests that strengthening LL muscles may contribute to reducing loading and pain at the low back. Given that the treatment effects of conventional LE and LS exercise therapies remain inconclusive, it is worth considering LL exercise therapy as an alternative approach to treat recreational runners with cLBP for potentially better rehabilitation outcomes.

To fill the research gap of lacking evidence-based exercise therapy protocols to treat runners with cLBP, this study aimed to evaluate the effectiveness of LL exercises compared with conventional LE exercises and LS exercises in managing cLBP in recreational runners. We hypothesized that specific LL exercises would be more effective in reducing running-induced pain and improving self-rated running capability, LL strength, back muscle function, and running gait than the conventional back exercises.

Back to Top | Article Outline


A single-blind randomized trial was conducted to evaluate the effectiveness of LL exercises compared with conventional LE exercises and LS exercises in managing cLBP in recreational runners. Ethical approvals were granted from the Nanyang Technological University Institutional Review Board and the National Healthcare Group Domain-Specific Review Board. Ninety-one potential participants were screened, of which 84 (male, 42) with a mean age of 27.3 (5.5) yr (Table 1) enrolled while waiting for physiotherapy service in a hospital from November 8, 2013, to September 6, 2014. The targeted sample size of 84 was calculated using G*Power 3.1 (medium effect size, 0.25; power, 80%; α = 0.05; three groups for repeated measures). Written informed consents were obtained and the rights of participants were protected. Using a block randomization table, participants were evenly allocated into one of the three treatment groups (n = 28 per group with even sex distribution): 1) LL exercises, 2) LE exercises, and 3) LS exercises. The study flowchart of participants is shown as Figure 1.





The inclusion criteria were as follows: 1) were 21–45 yr old, 2) had a body mass index between 18 and 25 kg·m−2, 3) had cLBP localized below the costal margin and above the inferior gluteal folds for >3 and <36 months, 4) were running 2–5 times per week for ≥2 km per session, and 5) started running 6 months before the study and have reached stable training intensity for at least 3 months before the study. The exclusion criteria were as follows: 1) average pain intensity for the past 1 wk <2 or >4 out of a 10-point Numeric Pain Rating Scale (NPRS); 2) specific low back pain (LBP), for example, spine fracture, disk herniation, and nerve root compression; 3) history of spine surgery; 4) current/history of LL conditions, for example, fracture, ankle sprain, patellar femoral pain syndrome, anterior tibial stress syndrome, hip and knee arthritis, and ligament laxity; 5) high fear-avoidance beliefs as determined by the Fear-Avoidance Beliefs Questionnaire (14,55) with a physical activity score of higher than 12 or a work score of higher than 19 (13); 6) work involving regular heavy lifting or hard physical work; or 7) use of pain medication.

Outcome measures in the current study consist of self-rated pain and running capability, LL strength, back muscle function, and running gait. Participants were asked to rate their running-induced pain using NPRS (0–10) according to the average rating during the past 1 wk. Similarly, self-rated running capability over the past 1 wk was measured using the Patient-Specific Functional Scale (PSFS; 0 stands for unable to perform running and 10 stands for able to perform running at the same level as that before the cLBP condition). The PSFS has been reported to be a more responsive functional outcome measurement tool compared with other scales for cLBP rehabilitation in various studies (17,38,42), especially in low-activity limitation population (17), such as the runners recruited in the current study.

LL strength was assessed using an isokinetic dynamometer (Biodex system 4 Pro; Biodex Corp., Shirley NY) to measure the peak concentric torque at 60°·s−1 for knee extension, hip extension, and hip abduction. The test speed of 60°·s−1 was reported to demonstrate high test–retest reliability for knee extension (intraclass correlation coefficients (ICC), 0.95) (49), hip abduction (ICC, 0.89), and hip extension (ICC, 0.90) (8) isokinetic strength measurements. Both left and right limbs were tested, with orders randomized. Details of the protocols are described elsewhere (4). The peak torque values were normalized to body mass before being used for analysis.

Back muscle function was evaluated by lumbar stabilizing muscle activation and LE muscle fatigability, following the procedures previously used to assess runners with cLBP (4). The transversus abdominis (TrA) and lumbar multifidus (LM) activations, reflected by muscle percent thickness changes between resting and submaximal contraction (25), were measured using a rehabilitative ultrasound image (RUSI) device (LOGIQ P5; GEHC, Milwaukee, WI) by a RUSI-certified physiotherapist. The test–retest reliability of this physiotherapist was excellent for both TrA (ICC = 0.96, 95% confidence interval (CI) (0.89–0.98); minimal detectable changes (MDC, 95% CI) = 16.55%) and LM (ICC (95% CI) = 0.97, 95% CI (0.92–0.99); MDC (95% CI) = 6.54%) measurements. To reflect LE muscle fatigability, surface electromyography (EMG) signals of the bilateral iliocostalis and longissimus during a 2-min Sorensen test were also recorded at 1000 Hz (Bagnoli™ Desktop EMG system; Delsys®, Boston, MA). Raw EMG data were band-pass filtered at 20–450 Hz and then analyzed in the frequency domain using built-in software (EMGworks® Software; Delsys®). The medium frequency was determined from the power density spectrum obtained using the fast Fourier transform technique with a Hamming windowing of 0.1 s. Finally, the medium frequency slope (MFS) was calculated as the slope of the medium frequency plotted over time for each muscle.

Spatiotemporal running gait parameters were measured using the OptoGait system comprising two parallel bars (100 cm × 8 cm) mounted on each side of the treadmill (Microgate S.r.I, Bolzano, Italy), which has been shown valid and reliable for gait analysis, with accuracy within 1 cm (29,30). Step length was calculated as the distance between the tip (of the toe) of two successive foot contacts. The minimal contact time and flight time were both set at 10 ms. Each participant was instructed to run at his/her usual comfortable speed for 10 min. Running gait data for the eighth to ninth minutes were collected at 1000 Hz as per the manufacturer’s specification. Subsequently, four gait parameters were extracted for analysis: self-selected speed, step length, flight time, and contact time.

Participants were assessed on all outcome measures at preintervention, mid-intervention, and end-intervention by a dedicated therapist who was blinded to treatment groups in the physiotherapy clinic. For self-rated pain and running capability, as well as running gait, additional follow-ups were done at 3 and 6 months after intervention. Participants were blinded from any previous ratings and results.

Participants were requested to attend supervised exercise sessions (LL, LE, or LS) with their physiotherapists twice per week that were spread at least 2 d apart for 8 wk. They were also asked to perform home exercises on other days of the week, guided by an instruction sheet. For each supervised session, all participants performed a standardized warm-up comprising general stretching exercises and stationary bicycling for 15 min, before their 30-min specific exercise session with the therapist.

For the LL exercises group, participants performed resistance exercises targeting the knee and hip muscles for 8 wk that has been shown to be effective in increasing muscle strength of the targeted muscles (1). During the supervised exercise sessions conducted in the physiotherapy department, a hip resistance training device was used to strengthen the hip abductors (Fig. 2A) and extensors (Fig. 2B), and a leg press machine was used to train the hip and knee extensors (Fig. 2C). Taking into account of participants’ safety and minimizing the risk of overexercising, for each exercise, the participants performed 3 sets of 10 repetitions at an intensity of 10 repetition maximum (RM) with 2 min of rest in between each set. The training intensity and volume adopted here were recommended for muscle strengthening (27). The 10 RM was reestimated at week 5, and the resistance was adjusted on the basis of the new 10 RM for the remaining 4 wk of training. The training volume and frequency remained the same. For the home exercises, single-leg squat (Fig. 2D) and wall sit (Fig. 2E) were prescribed to participants instead. The single-leg squat was reported to produce 82.3% of maximal voluntary isometric contraction (MVIC) for hip extensor and 71.0% for hip abductor (2). The wall sit as a close kinetic chain knee exercise was reported to produce 46% to 80% of MVIC of knee extension (50). Participants were asked to perform 3 sets of 10 repetitions of home exercises on days when there was no supervised exercise session. From week 5 onward, participants were instructed to hold a 2.5-kg weight during single-leg squat and to hold a 5-kg weight during wall sit.



For the LE exercises group, participants were prescribed an 8-wk progressive back extensors training program to achieve physiological changes in muscle fatigability (7,44). To take care of participants’ safety and to prevent excessive physical and psychological stress deriving from the exercise program, a progressive approach was used. For the first week of training, participants performed leg raise in a 4-point kneel position with the lumbar spine in a neutral position during the leg flexion and extension (Fig. 3A). In the second week, participants performed contralateral leg and arm raises (Fig. 3B). This arrangement enabled participants to reach approximately 40% of MVIC at the beginning of the second week without increasing the risk of injury to their low back muscles (43). Three sets of 10 repetitions per session were performed for all exercises, which recommended for muscle endurance and fatigability improvement (1). Isometric contraction was also added to the end of each repetition because it was reported as an essential component to improve the LE fatigability (7,39). Participants were instructed to hold the end position for 5 s and rest for 2 s before the next repetition. Two minutes of rest was given in between sets. To reach the recommended intensity of approximately 60% of MVIC of LEs for improvement in fatigability (39), 0.5 kg of ankle weight was added at week 3 and 0.5 kg of wrist weight was added at week 4. Subsequently, an increment of 0.5 kg every week for the ankle and 0.5 kg every 3 wk for the wrist were suggested to the participants (Fig. 3C). Prone back extension (Fig. 3D) was introduced to replace the 4-point kneeling exercises in week 5 because the percentage of MVIC produced by this exercise is greater than 65% (10). Home exercises were identical to those in the supervised session, except for no prone back extension from week 5 onward (Fig. 3D).



For the LS exercises group, participants received a series of TrA and LM muscle activation and motor control training as previously described by Koumantakis et al. (26). There were three stages of training: stages 1 and 2 were approximately 2 wk and stage 3 was approximately 4 wk in duration. Participants were allowed to progress to the next stage without being restricted by the timeline as soon as they were able to complete the current stage of exercises satisfactorily. In stage 1, participants were instructed to conduct low-load activation of the lumbar stabilizing muscles, TrA and LM, with no movement (isometrically) and in minimal loading positions of sitting and standing (Fig. 4A, B). The RUSI was used to provide visual feedback for TrA and LM activation. Excessive effort causing incorrect muscle activation in the global muscles or spinal movement at the initial stages was discouraged. Progressively, the holding time for each contraction was increased to at least 60 s and the duration of each exercise session was increased up to 10 min (45). In stage 2, integration of the lumbar stabilizing muscle activity into light dynamic functional tasks was added to participant’s exercise programs as shown in Figure 4C and D. The participants were instructed to practice with the same holding time and exercise duration as the first stage. In stage 3, heavier-load functional tasks were progressively introduced to participants as shown in Figure 4E, with the resistance from the theraband during shoulder external rotation in 70°–90° abduction and in Figure 4F with resistance during shoulder abduction to 90°. For this stage, participants were instructed to practice the exercises for 10 min twice a day (41). All exercises were used for both supervised exercise sessions and home exercise, except the sitting balance integration component (Fig. 4C), in which the gym ball was substituted with a chair with cushion during home exercise.



All participants were informed that they should not feel exacerbation of their back pain during training and that the body reaction to exercise should be limited to “aching” or “soreness.” Otherwise, the exercise intensity should be reduced or the program should be terminated immediately. Participants were encouraged to continue their regular running but refrain from heavy gym weight training during the entire 8-wk exercise training and 6-month follow-up periods. Exercise logs were provided to document their home exercise sessions, running frequency, and running distance. After completing the 8-wk intervention, participants were asked to stop their home exercises.

Statistical analyses in the current study consist of covariate screening and treatment effect comparisons. The participants’ characteristics, exercise compliant rate, and running distance were compared among the three groups using a one-way ANOVA (for parametric data) or the Kruskal–Wallis test (for nonparametric data). Initially, variables with a P value less of than 0.10 were planned to be treated as covariates for the comparison of the treatment effects, but none of them had a P value of less than 0.10 (Table 1).

A generalized estimating equation (GEE) approach using SPSS 21.0 was used to compare the treatment and interaction effects. Given GEE’s capability of handling outcomes with missing data and various correlations between time points (32), we could include all 84 participants’ data in the analysis. The dependent variables entered to GEE were the NPRS score, PSFS score, peak isokinetic torques, TrA and LM percent thickness change, MFS for iliocostalis and longissimus, and running gait parameters. Targeted main effects (group, time, sex) and interactions (group–time, group–sex, group–body side) were entered as the independent variables to form the GEE models. A backward elimination approach (α = 0.05) was applied during the model formation. Except for group and time, independent variables that did not significantly contribute to the model were removed. The model was then rerun with those significant independent variables.

Back to Top | Article Outline


In total, 74 participants completed all measurement sessions by April 27, 2015 (Fig. 1). In the LL group, one participant injured her back during fifth week of the intervention period, and her subsequent follow-up data were not included in the analysis. The back injury was not related to the study. There were one participant loss of contact after his preintervention measurement session and one participant who missed out her final measurement (6-month follow-up) due to migration. In the LE group, one participant injured his ankle in the third week of the intervention period by a minor traffic accident, and his subsequent follow-up data were not included in the statistical analysis. There were also one participant loss of contact after his initial measurement session, one participant loss of contact after his mid-intervention measurement session (end of week 4), and one participant who missed out his end-intervention measurement session (end of week 8) due to busy schedule. In the LS group, one participant injured her back during the seventh week of the intervention period, and her subsequent follow-up data were not included in the analysis. Another participant injured his ankle during the eighth week of the intervention period, and thereafter, he was not located due to loss of contact; thus, only his preintervention and mid-intervention measurement data were included in the analysis. Both injuries were not related to the study. There was also one participant in this group that missed out his end-intervention measurement session (end of week 8) due to his busy schedule.

Among the three exercise groups, there was no significant difference in the participants’ characteristics, compliant rate, or running habit (Table 1). The means (SD) for all outcome measures during the intervention and follow-up periods are presented in Table 2. Because group–body side did not contribute to any GEE model, the averaged readings from both sides are presented in Table 2.



GEE analyses of all outcome measures are presented as follows. For the NPRS score of average running-induced pain during the past 1 wk, there was a main effect of time (P < 0.001). Participants in all three groups achieved an average rate of improvement of 0.746 points over each time point (B = 0.746; 95% CI, −0.799 to −0.693; P = 0.001). Mean NPRS score differed across the three groups (P = 0.009). The LL group had 0.273 points lower mean NPRS score than did the LE group (B = −0.273; 95% CI, 0.041–0.505; P = 0.021) and 0.329 points lower than did the LS group (B = −0.329; 95% CI, 0.088–0.570; P = 0.008).

For PSFS score, the changes in score significantly differed across the three groups over time (group–time interaction, P < 0.001). The LL group achieved an average rate of improvement of 0.949 points over each time point (B = −0.949; 95% CI, 0.877–1.021; P < 0.001), which was 0.198 (B = −0.198; 95% CI, −0.316 to −0.080; P = 0.001) and 0.263 (B = −0.263; 95% CI, −0.406 to −0.120; P < 0.001) more than the LE and LS groups, respectively.

For LL isokinetic strength, there was a significant group–time interaction (P = 0.001) in peak knee extension torque. The LL group improved on average 0.260 N·m·kg−1 over each time point (B = 0.260; 95% CI, 0.193–0.326; P < 0.001), which was 0.220 N·m·kg−1 (B = −0.220; 95% CI, −0.307 to −0.133; P < 0.001) and 0.206 N·m·kg−1 (B = −0.206; 95% CI, −0.306 to −0.105; P < 0.001) higher than the LE and LS groups, respectively. Peak knee extension torque differed between sexes (P < 0.001), with male participants presenting 0.276 N·m·kg−1 higher compared with the female participants (B = 0.276; 95% CI, 0.115–0.437; P = 0.001). Peak hip extension torque increased over time (0.078 N·m·kg−1 per time point; 95% CI, 0.042–0.113; P < 0.001) but did not differ across the three groups (P = 0.154). Similarly, peak hip abduction torque increased over time (0.106 N·m·kg−1 per time point; 95% CI, 0.075–0.137; P < 0.001), with no between-group difference (P = 0.363).

For lumbar stabilizing muscle activation, TrA percent thickness changes increased on average 11.4% over each time point (B = 11.4; 95% CI, 8.30–14.40; P < 0.001), with no differences among the three groups (P = 0.061). There was a main effect of body side (P = 0.001), with the dominant side exhibited 8.4% greater thickness change than the nondominant side (B = 8.4; 95% CI, 3.50–13.40; P = 0.001). LM percent thickness changes improved on average 9.2% over each time point (B = 9.2; 95% CI, 2.20–16.20; P < 0.001) but did not differ among the three groups (P = 0.188).

For LE muscle fatigability measured using longissimus MFS, all three groups slightly improved over each time point by 0.023 (B = 0.023; 95% CI, 0.007–0.040; P = 0.005). Means of MFS were different between male and female participants (P < 0.001), and the LE group presented lower MFS than did the LL group (B = −0.058; 95% CI, −0.101 to −0.015; P = 0.008). For iliocostalis, the MFS differed across the three groups over time (P = 0.033). Although there was no change in MFS over time in the LL (B = 0.006; 95% CI, 0.008–0.021; P = 0.398) and LS groups (B = 0.010; 95% CI, −0.027 to 0.047; P = 0.609), the LE group improved their MFS by 0.055 more over each time point (B = 0.055; 95% CI, 0.014–0.097; P = 0.009).

For running gait, self-selected running speed did not differ across the three groups (P = 0.444) or change over time (P = 0.185). There was a main effect of sex (P < 0.001), with male participants running 2.366 km·h−1 faster than their female counterparts (B = 2.366; 95% CI, 1.875–2.856; P < 0.001). Changes in running step length differed across the three groups over time (P = 0.046). Participants in the LL group achieved an average increase of 2.464 cm in step length over each time point (B = 2.464; 95% CI, 0.953–3.975; P = 0.001), which was similar to the LE group (B = −1.690; 95% CI, −3.639 to 0.260; P = 0.089) but greater by 2.213 cm per time point compared with the LS group (B = −2.213; 95% CI, −3.959 to −0.468; P < 0.001). Step length differed between sexes (P < 0.001), with longer step length in male participants (B = 26.12; 95% CI, 21.384–31.839; P < 0.001). Flight time remained stable with no changes over time (P = 0.208) and no difference across the three groups (P = 0.931). Similarly, contact time also did not change over time (P = 0.356) or differ among the three groups (P = 0.371).

Back to Top | Article Outline


This single-blinded randomized trial was conducted to evaluate the effectiveness of the LL exercises, as compared with conventional back exercises, in managing nonspecific cLBP in the recreational runner population. The study hypothesis that LL exercises would be more effective in improving rehabilitation outcomes was partially supported by our key findings: 1) greater improvement in self-rated running capability and knee extension strength in the LL group than in the LE and LS groups, 2) greater increase in running step length in the LL and LE groups than in the LS group, and 3) similar reduction in running-induced pain and improvement in back muscle function across all three exercises groups.

Pain reduction is a key rehabilitation outcome in the treatment and management of cLBP. This study showed that running-induced pain improved over time for all participants regardless of the exercise groups. At 6 months after intervention, the total reduction in NPRS score was 2.984 points (0.746 × 4 time points). This improvement exceeded the MDC (95% CI) of 2.0 (6) and hence can be considered clinically significant. Although there are no studies directly comparing LL exercise with conventional back exercises on pain reduction in runners with cLBP, others have shown that general exercise (which included LL components) reduced back pain to a similar extent compared with specific LE or LS exercises (11,33,37). Thus, the reduction in running-induced pain observed among runners with cLBP in the present study is likely due to the general effect of exercise rather than a specific type of exercise.

Using the PSFS to assess self-rated running capability, participants in the LL group improved 3.796 (0.949 × 4 time points) at 6-month follow-up compared with 3.004 for LE and 2.744 points for the LS groups, respectively. Although all three groups responded positively to the exercise treatments, it is important to note that only the LL group had achieved clinically significant improvement by exceeding the MDC (95% CI) of 3.521 points for single activity PSFS (calculated from the MDC, 3.0 (90% CI)) (51). This demonstrates that LL exercise therapy is more effective than conventional back exercises in improving self-rated running capability and therefore is a promising approach to treat cLBP among runners. Previous studies on older, less active population revealed mixed results when comparing the treatment effect of general exercise and LS exercise on self-rated general functional outcome measured using Oswestry Low Back Pain Disability Index or Roland Morris Disability Questionnaire (11,33,37). To our best knowledge, the present study is the first to use PSFS running as a specific functional outcome to evaluate the effectiveness of exercise therapy in managing cLBP. Among the various survey instruments commonly used to evaluate back pain, PSFS was reported to be more responsive (effect size, 1.7) and specific for population with low physical activity limitation (17), and hence, this tool was chosen for the runners recruited in the current study. Using a running-specific functional outcome measurement, it is convincing to note the superior treatment effect on the improvement in self-rated running capability achieved by the LL group than the other groups.

Regarding the LL isokinetic strength, we initially hypothesized greater improvements in all hip and knee muscle strengths in the LL group compared with the LE and LS groups. This hypothesis was partially supported by our findings that peak knee extension torque increased more in the LL group than in the two conventional approaches, but similar improvements in hip extension and hip abduction torque were observed. By the end of the 8-wk intervention, peak knee extension torque increased by 31.82 N·m in total (0.260 N·m·kg−1 × 61.2 kg (mean body weight, Table 1) × 2 time points) compared with 4.8 N·m for the LE group and 6.5 N·m for the LS group, respectively. The improvement in the LL group is of clinical importance because it has far exceeded the MDC (95% CI) of 17.88 N·m (49). Compared with the knee, the overall improvements in hip muscle strength were too small to be practically meaningful (hip extension: 9.55 N·m (MDC (95% CI) = 28.82 N·m), hip abduction: 12.97 N·m (MDC (95% CI) = 34.00 N·m)) (8).

It is interesting to note that among all muscle functions tested, LL exercises only induced greater improvement in knee extensor strength compared with conventional back exercises. Other functions including hip muscle strength, lumbar stabilizing muscle activation, and LE muscle fatigability were similarly affected regardless of type of exercises prescribed. This suggests that the higher self-rated running capability observed in the LL group is likely related to greater gain in knee strength, supporting a previous speculation that weak knee extensors may compromise one’s ability to absorb impact shock during running and hence transmitting higher forces to the low back (4). Thus, improving knee extensor strength may be the step needed to break the vicious cycle of knee and back muscle dysfunction previously reported in LBP population (19,46).

In the current study, similar improvements in lumbar stabilizing muscle activation are seen across all three groups for both muscles. By the end of the 8-wk intervention, TrA activation improved by 22.8%, which was approaching the MDC (95% CI) of 25.4% (25). Similarly, LM activation improved by 18.4% at the end of 8 wk, and this change overcame the MDC (95% CI) of 11.0% (25). One previous study reported non–exercise-specific improvement in TrA activation when comparing LE and LS exercises in the general LBP population (54). LL weight-bearing exercises were also observed to induce similar TrA activation as back muscle exercises (21). For the LM, similar EMG activations were reported for different types of exercises that closely resembled those adopted in the current study (18,40,53). Collectively, our results are in line with the literature that clinically meaningful improvements in lumbar stabilizing muscle activation can be achieved via exercise training in general. However, improvement in longissimus fatigability, despite showing statistical significance, is unlikely clinically meaningful because the changes in MFS by 0.046 (0.023 × 2 time points) after 8-wk training were too small to overcome the MDC (95% CI) ranging from 0.11 to 0.17 (9). The lack of improvement in longissimus fatigability parallels with previous studies examining the training effect of LE (26,39), stabilization (26), and general LL exercises (36). Interestingly, improvement in iliocostalis fatigability in the LE group is clinically meaningful, with MFS changes of 0.11 (0.055 × 2 points) after 8 wk. This finding contradicted a previous study reporting no change in muscle fatigability after a 12-wk LE isoinertial exercise intervention (36). Differences in exercise protocol (isometric vs isoinertial) and EMG electrodes placement (parallel vs 45° to the spine) (36) may explain the lack of agreement between the previous and present studies. In summary, the results from the present study suggest that LL exercises are equally effective as back exercises in improving lumbar stabilizing muscle activation in runners with cLBP. On the other hand, LE muscle fatigability was responsive to LE exercises but not LL or LS exercises.

Running step length as the only running gait parameter significantly changed by exercise training in current study increased by 9.856 cm (2.464 cm × 4 time points) by the end of the 6-month follow-up in the LL group, which was similar to that in the LE group but significantly more than that in the LS group. Because the running speeds on treadmill are rather stable over time (Table 2), increased step length would have resulted in reduced step frequency given that speed is a product of step length and step frequency. Taking fewer steps to complete the same distance may have reduced the number of impacts on the spine during the ground contact (5), contributing to reduced pain and improved running capability reported by the participants. Although the OptoGait system has been shown to be valid and reliable for measuring spatiotemporal parameters during walking (29,30), there is no established MDC to determine the clinical relevance of any observed changes during treadmill running. Nevertheless, it is clear that LL exercises lead to the greatest increase in running step length among our participants.

There are a few limitations to the current study. First, the low attendance of supervised exercise session (mean, 5.6 (4–7) visits of 16) was less than the minimal frequency of once a week for successful treatment in cLBP patients (3). Home exercise compliance rate was much better (mean, 29.3 (13–52) sessions), suggesting that any improvements observed were most likely attributed to home exercises rather than supervised training. In the literature, home exercise compliance rather than formal physiotherapy session attendance was found to be correlated with the reductions in pain and self-reported disability in cLBP patients (34). Future studies can investigate whether home exercise alone is sufficient to successfully treat cLBP conditions. Second, only spatiotemporal gait variables during a treadmill test were measured. Since altered trunk posture during running in cLBP population has been reported (48), additional kinematics and kinetic data will be useful for a more comprehensive biomechanical evaluation of running gait. Lastly, the current findings should be applied with caution to older, less active individuals because our participants were younger (age, 27.3 (5.5) yr), recreational runners.

Back to Top | Article Outline


LL exercise therapy has shown to be a promising approach to the clinical management of nonspecific cLBP in recreational runners. Compared with conventional back exercises, LL exercise therapy was more effective in improving key rehabilitation outcomes including self-rated running capability, knee extension strength, and running step length. All exercise therapies were equally effective in reducing running-induced pain and improving back muscle function.

The authors thank the National Institute of Education Academic Research Fund, Singapore, for providing the financial support for this project and the management of Allied Health Division, Jurong Health Services, Singapore, for providing experiment venue and clinical hours to complete the project.

The authors thank Mr. Johnny Wong from Clinical Research Unit, Ng Teng Fong General Hospital, Jurong Health Services, Singapore, for his valued advice on the statistical analysis of the current study.

This study was funded by the National Institute of Education Academic Research Fund. The funding source did not play a role in the investigation. We affirm that we have no financial affiliation (including research funding) or involvement with any commercial organization that has a direct financial interest in any matter included in this article. The results of the present study do not constitute endorsement by the American College of Sports Medicine. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.

Back to Top | Article Outline


1. American College of Sports Medicine. ACSM’s Guidelines for Exercise Testing and Prescription. 9th ed. New York: Lippincott Williams & Wilkins; 2013.
2. Boren K, Conrey C, Le Coguic J, Paprocki L, Voight M, Robinson TK. Electromyographic analysis of gluteus medius and gluteus maximus during rehabilitation exercises. Int J Sports Phys Ther. 2011;6(3):206–23.
3. Bruce-Low S, Smith D, Burnet S, Fisher J, Bissell G, Webster L. One lumbar extension training session per week is sufficient for strength gains and reductions in pain in patients with chronic low back pain ergonomics. Ergonomics. 2012;55(4):500–7.
4. Cai C, Kong PW. Low back and lower-limb muscle performance in male and female recreational runners with chronic low back pain. J Orthop Sports Phys Ther. 2015;45(6):436–43.
5. Cappozzo A. Force actions in the human trunk during running. J Sports Med Phys Fitness. 1983;23(1):14–22.
6. Childs JD, Piva SR, Fritz JM. Responsiveness of the numeric pain rating scale in patients with low back pain. Spine (Phila Pa 1976). 2005;30(11):1331–4.
7. Chok B, Lee R, Latimer J, Tan SB. Endurance training of the trunk extensor muscles in people with subacute low back pain. Phys Ther. 1999;79(11):1032–42.
8. Claiborne TL, Timmons MK, Pincivero DM. Test–retest reliability of cardinal plane isokinetic hip torque and EMG. J Electromyogr Kinesiol. 2009;19(5):e345–52.
9. Dedering A, Gnospelius A, Elfving B. Reliability of measurements of endurance time, electromyographic fatigue and recovery, and associations to activity limitations, in patients with lumbar disc herniation. Physiother Res Int. 2010;15(4):189–98.
10. De Ridder EM, Van Oosterwijck JO, Vleeming A, Vanderstraeten GG, Danneels LA. Posterior muscle chain activity during various extension exercises: an observational study. BMC Musculoskelet Disord. 2013;14:204.
11. Ferreira ML, Ferreira PH, Latimer J, et al. Comparison of general exercise, motor control exercise and spinal manipulative therapy for chronic low back pain: A randomized trial. Pain. 2007;131(1–2): 31–7.
12. Ferreira PH, Ferreira ML, Hodges PW. Changes in recruitment of the abdominal muscles in people with low back pain: ultrasound measurement of muscle activity. Spine. 2004;29(22):2560–6.
13. Flynn T, Fritz J, Whitman J, et al. A clinical prediction rule for classifying patients with low back pain who demonstrate short-term improvement with spinal manipulation. Spine. 2002;27(24): 2835–43.
14. George SZ, Fritz JM, Erhard RE. A comparison of fear-avoidance beliefs in patients with lumbar spine pain and cervical spine pain. Spine (Phila Pa 1976). 2001;26(19):2139–45.
15. Glick JM, Katch VL. Musculoskeletal injuries in jogging. Arch Phys Med Rehabil. 1970;51(3):123–6.
16. Grabiner MD, Jeziorowski JJ. Isokinetic trunk extension discriminates uninjured subjects from subjects with previous low back pain. Clin Biomech (Bristol, Avon). 1992;7(4):195–200.
17. Hall AM, Maher CG, Latimer J, Ferreira ML, Costa LO. The Patient-Specific Functional Scale is more responsive than the Roland Morris disability questionnaire when activity limitation is low. Eur Spine J. 2011;20(1):79–86.
18. Hamlyn N, Behm DG, Young WB. Trunk muscle activation during dynamic weight-training exercises and isometric instability activities. J Strength Cond Res. 2007;21(4):1108–12.
19. Hart JM, Weltman A, Ingersoll CD. Quadriceps activation following aerobic exercise in persons with low back pain and healthy controls. Clin Biomech (Bristol, Avon). 2010;25(8):847–51.
20. Helmhout PH, Harts CC, Viechtbauer W, Staal JB, de Bie RA. Isolated lumbar extensor strengthening versus regular physical therapy in an army working population with nonacute low back pain: a randomized controlled trial. Arch Phys Med Rehabil. 2008;89(9): 1675–85.
21. Hides JA, Wong I, Wilson SJ, Belavy DL, Richardson CA. Assessment of abdominal muscle function during a simulated unilateral weight-bearing task using ultrasound imaging. J Orthop Sports Phys Ther. 2007;37(8):467–71.
22. Hodges P, Holm AK, Hansson T, Holm S. Rapid atrophy of the lumbar multifidus follows experimental disc or nerve root injury. Spine. 2006;31(25):2926–33.
23. Holmström E, Moritz U, Andersson M. Trunk muscle strength and back muscle endurance in construction workers with and without low back disorders. Scand J Rehabil Med. 1992;24(1):3–10.
24. Kankaanpaa M, Taimela S, Laaksonen D, Hanninen O, Airaksinen O. Back and hip extensor fatigability in chronic low back pain patients and controls. Arch Phys Med Rehabil. 1998;79(4):412–7.
25. Koppenhaver SL, Hebert JJ, Fritz JM, Parent EC, Teyhen DS, Magel JS. Reliability of rehabilitative ultrasound imaging of the transversus abdominis and lumbar multifidus muscles. Arch Phys Med Rehabil. 2009;90(1):87–94.
26. Koumantakis GA, Watson PJ, Oldham JA. Trunk muscle stabilization training plus general exercise versus general exercise only: randomized controlled trial of patients with recurrent low back pain. Phys Ther. 2005;85(3):209–25.
27. Kraemer WJ, Adams K, Cafarelli E, et al. American College of Sports Medicine position stand. Progression models in resistance training for healthy adults. Med Sci Sports Exerc. 2002;34(2):364–80.
28. Lee JH, Hoshino Y, Nakamura K, Kariya Y, Saita K, Ito K. Trunk muscle weakness as a risk factor for low back pain. A 5-year prospective study. Spine (Phila Pa 1976). 1999;24(1):54–7.
29. Lee M, Song C, Lee K, Shin D, Shin S. Agreement between the spatio-temporal gait parameters from treadmill-based photoelectric cell and the instrumented treadmill system in healthy young adults and stroke patients. Med Sci Monit. 2014;20:1210–9.
30. Lee MM, Song CH, Lee KJ, Jung SW, Shin DC, Shin SH. Concurrent validity and test–retest reliability of the OPTOGait photoelectric cell system for the assessment of spatio-temporal parameters of the gait of young adults. J Phys Ther Sci. 2014;26(1):81–5.
31. Leinonen V, Kankaanpää M, Airaksinen O, Hänninen O. Back and hip extensor activities during trunk flexion/extension: effects of low back pain and rehabilitation. Arch Phys Med Rehabil. 2000;81(1): 32–7.
32. Ma Y, Mazumdar M, Memtsoudis SG. Beyond repeated-measures analysis of variance: advanced statistical methods for the analysis of longitudinal data in anesthesia research. Reg Anesth Pain Med. 2012;37(1):99–105.
33. Macedo LG, Latimer J, Maher CG, et al. Effect of motor control exercises versus graded activity in patients with chronic nonspecific low back pain: a randomized controlled trial. Phys Ther. 2012;92(3): 363–77.
34. Mannion AF, Helbling D, Pulkovski N, Sprott H. Spinal segmental stabilisation exercises for chronic low back pain: programme adherence and its influence on clinical outcome. Eur Spine J. 2009;18(12):1881–91.
35. Mannion AF, Muntener M, Taimela S, Dvorak J. Comparison of three active therapies for chronic low back pain: results of a randomized clinical trial with one-year follow-up. Rheumatology (Oxford). 2001;40(7):772–8.
36. Mannion AF, Taimela S, Muntener M, Dvorak J. Active therapy for chronic low back pain part 1. Effects on back muscle activation, fatigability, and strength. Spine. 2001;26(8):897–908.
37. Marshall PW, Kennedy S, Brooks C, Lonsdale C. Pilates exercise or stationary cycling for chronic nonspecific low back pain: does it matter? a randomized controlled trial with 6-month follow-up. Spine. 2013;38(15):E952–9.
38. Maughan EF, Lewis JS. Outcome measures in chronic low back pain. Eur Spine J. 2010;19(9):1484–94.
39. Moffroid MT, Haugh LD, Haig AJ, Henry SM, Pope MH. Endurance training of trunk extensor muscles. Phys Ther. 1993;73(1): 10–7.
40. Nuzzo JL, McCaulley GO, Cormie P, Cavill MJ, McBride JM. Trunk muscle activity during stability ball and free weight exercises. J Strength Cond Res. 2008;22(1):95–102.
41. O’Sullivan PB, Phyty GD, Twomey LT, Allison GT. Evaluation of specific stabilizing exercise in the treatment of chronic low back pain with radiologic diagnosis of spondylolysis or spondylolisthesis. Spine (Phila Pa 1976). 1997;22:2959–67.
42. Pengel LH, Refshauge KM, Maher CG. Responsiveness of pain, disability, and physical impairment outcomes in patients with low back pain. Spine (Phila Pa 1976). 2004;29(8):879–83.
43. Pollock ML, Graves JE, Carpenter D, Foster D, Leggett SH, Fulton M. Muscle. In: Hochschuler S, Cotler H, editors. Rehabilitation of the Spine: Science and Practice. Philadelphia (PA): Mosby; 1993.
44. Pollock ML, Leggett SH, Graves JE, Jones A, Fulton M, Cirulli J. Effect of resistance training on lumbar extension strength. Am J Sports Med. 1989;17(5):624–9.
45. Richardson CA, Jull G, Hodges P, Hides J. Therapeutic Exercises for Spinal Segmental Stabilization in Low Back Pain. Edinburgh: Churchill Livingstone; 1999.
46. Rojhani Shirazi Z, Biabani Moghaddam M, Motealleh A. Comparative evaluation of core muscle recruitment pattern in response to sudden external perturbations in patients with patellofemoral pain syndrome and healthy subjects. Arch Phys Med Rehabil. 2014;95(7):1383–9.
47. Roy SH, De Luca CJ, Synder-Mackler L, Emley M, Crenshaw RL, Lyons JP. Fatigue, recovery, and low back pain in varsity rowers. Med Sci Sports Exerc. 1990;22(4):463–9.
48. Seay JF, Van Emmerik RE, Hamill J. Trunk bend and twist coordination is affected by low back pain status during running. Eur J Sport Sci. 2014;14(6):563–8.
49. Sole G, Hamren J, Milosavljevic S. Test–retest reliability of isokinetic knee extension and flexion. Arch Phys Med Rehabil. 2007;88:626–31.
50. Stensdotter AK, Hodges PW, Mellor R, Sundelin G, Hager-Ross C. Quadriceps activation in closed and in open kinetic chain exercise. Med Sci Sports Exerc. 2003;35(12):2043–7.
51. Stratford PW, Gill C, Westaway M, Binkley J. Assessing disability and change on individual patients: a report of a patient specific measure. Physiother Can. 1995;47:258–63.
52. Suter E, Lindsay D. Back muscle fatigability is associated with knee extensor inhibition in subjects with low back pain. Spine. 2001;26(16):E361–6.
53. Tarnanen SP, Siekkinen KM, Hakkinen AH, Malkia EA, Kautiainen HJ, Ylinen JJ. Core muscle activation during dynamic upper limb exercises in women. J Strength Cond Res. 2012;26(12): 3217–24.
54. Teyhen DS, Rieger JL, Westrick RB, Miller AC, Molloy JM, Childs JD. Changes in deep abdominal muscle thickness during common trunk-strengthening exercises using ultrasound imaging. J Orthop Sports Phys Ther. 2008;38(10):596–605.
55. Waddell G, Newton M, Henderson I, Somerville D, Main CJ. A Fear-Avoidance Beliefs Questionnaire (FABQ) and the role of fear-avoidance beliefs in chronic low back pain and disability. Pain. 1993;52(2):157–68.
56. Woolf SK, Barfield WR, Nietert PJ, Mainous AG 3rd, Glaser JA. The Cooper River Bridge Run Study of low back pain in runners and walkers. J South Orthop Assoc. 2002;11(3):136–43.


© 2017 American College of Sports Medicine