Article Text

Download PDFPDF

Early multimodal rehabilitation following lumbar disc surgery: a randomised clinical trial comparing the effects of two exercise programmes on clinical outcome and lumbar multifidus muscle function
  1. Jeffrey J Hebert1,
  2. Julie M Fritz2,
  3. Anne Thackeray3,
  4. Shane L Koppenhaver1,4,
  5. Deydre Teyhen5
  1. 1School of Psychology and Exercise Science, Murdoch University, Murdoch, Western Australia, Australia
  2. 2Department of Physical Therapy, University of Utah and Clinical Research Outcomes Scientist, Intermountain Healthcare, Salt Lake City, Utah, USA
  3. 3Department of Physical Therapy, University of Utah, Salt lake City, Utah, USA
  4. 4U.S. Army-Baylor University Doctoral Program in Physical Therapy, San Antonio, Texas, USA
  5. 5Telemedicine and Advanced Technology Research Center, U.S. Army Medical Research and Material Command, Ft Detrick, Maryland, USA
  1. Correspondence to Dr Jeffrey Hebert, School of Psychology and Exercise Science, Murdoch University, 90 South Street, ECL 4.008, Murdoch, WA 6150, Australia; J.Hebert{at}


Background The optimal components of postoperative exercise programmes following single-level lumbar discectomy have not been identified. Facilitating lumbar multifidus (LM) function after discectomy may improve postoperative recovery. The aim of this study was to compare the clinical and muscle function outcomes of patients randomised to receive early multimodal rehabilitation following lumbar discectomy consisting of exercises targeting specific trunk muscles including the LM or general trunk exercises.

Methods We included participants aged 18 to 60 years who were scheduled to undergo single-level lumbar discectomy. After two postoperative weeks, participants were randomly assigned to receive an 8-week multimodal exercise programme including either general or specific trunk exercises. The primary outcome was pain-related disability (Oswestry Index). Secondary outcomes included low back and leg pain intensity (0–10 numeric pain rating scale), global change, sciatica frequency, sciatica bothersomeness and LM function measured with real-time ultrasound imaging. Treatment effects 10 weeks and 6 months after surgery were estimated with linear mixed models.

Results 61 participants were randomised to receive a general trunk (n=32) or specific (n=29) exercise programme. There were no between-group differences in clinical or muscle function outcomes. Participants in both groups experienced improvements in most outcome measures.

Conclusions Following lumbar discectomy, multimodal rehabilitation programmes comprising specific or general trunk exercises have similar effects on clinical and muscle function outcomes. Local factors such as the individual patient characteristics identified by specific assessment findings, clinician expertise and patient preferences should direct therapy selection when considering the types of exercises tested in this trial for inclusion in rehabilitation programmes following lumbar disc surgery.

  • Exercise rehabilitation
  • Core stability/pelvis/hips, ribs
  • Skeletal Muscle Physiology
  • Physiotherapy
  • Back injuries

Statistics from


Lumbar discectomy is the most commonly performed surgical spine procedure1; however, its clinical outcomes remain poorly defined.2 While the effectiveness of lumbar discectomy largely depends on the technical result of surgery, the postoperative care received by these patients is also likely to impact outcomes.3 The majority of research has focused on the former, while the postoperative management of this population has received relatively little attention.4

Rehabilitation in the form of therapeutic exercise is one aspect of postoperative management thought to improve outcomes following lumbar discectomy. A recent meta-analysis3 of randomised trials concluded that exercise programmes implemented 4–6 weeks after lumbar discectomy improve postoperative pain and disability. However, there is little consistency between exercise protocols3 ,5 ,6 and the most effective rehabilitation approach is unknown.

The timing of rehabilitation may impact outcomes following lumber discectomy. Exercise and physical activity are often avoided after lumbar discectomy owing to fears of reinjury, reherniation and instability.7 However, eliminating activity restrictions and encouraging exercise immediately following lumbar discectomy may allow for an earlier return to function without increasing rates of reherniation or other complications.7 ,8 Despite this evidence, recent surveys of physical therapists5 and spine surgeons6 report large variability in postoperative activity restrictions and exercise recommendations.

The restoration of lumbar multifidus (LM) function may also be an important determinant of clinical outcome following lumbar discectomy. LM deficits are associated with a worsened prognosis following lumbar spine surgery and intraoperative LM injury is associated with the development of ‘failed back syndrome’.9 ,10 Consequently, a common rehabilitation goal for individuals with lumbar spine disorders is the restoration of LM function.11–14 However, previous randomised trials have not examined the effectiveness of an exercise programme focusing on the restoration of LM function following lumbar discectomy.

Therefore, the aim of this study was to compare the clinical and muscle function outcomes of patients randomised to receive early multimodal rehabilitation following single-level lumbar discectomy consisting of exercises targeting specific trunk muscles including the LM or general trunk exercises. We hypothesised that participants randomised to receive the specific exercise programme would experience greater improvements in clinical and muscle function outcomes than participants in the general trunk exercise group.


Study design

This study was a parallel group randomised clinical trial comparing two postoperative rehabilitation protocols following lumbar discectomy. The trial was prospectively registered ( ID: NCT00894972) and reported in accordance with the CONSORT statement.15 ,16

Study participants

We recruited potential participants from academic and private neurological and orthopaedic spine surgery practices in Salt Lake City, Utah. Inclusion criteria were: age 18–60 years, presurgical radiographic confirmation of lumbar disc herniation through MRI or CT and scheduled to undergo single-level lumbar discectomy. Potential participants were excluded if they had prior lumbar spine surgery, surgery at more than one level, a surgical procedure other than discectomy (eg, fusion) or perioperative complications representing a contraindication to exercise. This trial received ethical approval from the Institutional Review Board of the University of Utah. All participants provided written informed consent prior to enrolment.


A random number generator was used to create a permuted-block randomisation list with variable block sizes of 4–6. Sequentially numbered, opaque envelopes containing the participant's group assignment were prepared by research staff not affiliated with this trial. The envelope was opened after the 2-week postoperative assessment by the treating physical therapist. Group assignments were concealed from participants and outcome assessors.

Exercise interventions

The rehabilitation programmes were implemented and supervised by one of three physical therapists (mean (range) experience=12(10–14) years) trained in the research protocol. After two postoperative weeks, all participants initiated an 8-week exercise programme comprising weekly supervised exercise sessions and daily home exercises. Participants were instructed on the importance of limiting sitting, avoiding lumbar flexion and appropriate body mechanics during daily activities. Additionally, participants received an educational pamphlet providing advice regarding the importance of weight management, smoking cessation and stress management. During the first 2–3 weeks of the exercise programme, participants receiving the specific exercise protocol (SPEC) performed exercises not included in the general trunk exercise protocol (GEN). To avoid the potential for attention effects resulting from this extra activity, participants in the GEN performed additional range of motion exercises so that the total treatment time was equivalent between the groups. Participants were considered compliant if they completed at least 80% of the programme in 6 of 8 weeks.

General trunk exercise protocol

The GEN was based on previously reported exercise programmes17 ,18 and designed to increase the strength and endurance of the trunk and lumbar spine musculature.19 ,20 This protocol comprised three components: (1) aerobic exercise, (2) range of motion exercise and (3) strengthening exercise. Consistent with the current guidelines for the treatment of individuals with lumbar spine disorders,21 participants were encouraged to engage in a daily aerobic exercise programme. Walking was the primary mode of aerobic exercise with an initial dosage of 20 min/day and progressing to a goal of 60 min/day towards the end of 8 week period. Range of motion exercises were used as a ‘warm-up’ activity prior to trunk strengthening exercises. Additional details of the GEN are presented in table 1.

Table 1

Multimodal rehabilitation programmes

Specific trunk exercise protocol

The SPEC included all components of the GEN. In addition, participants performed specific trunk muscle exercises similar to protocols used to treat patients with non-specific, non-surgical low back pain (table 1).22 ,23 Initial efforts gave attention to volitional, isometric contractions of the LM.24 This approach also included similar contractions of the transversus abdominis (TrA) using the abdominal drawing-in manoeuvre.25 ,26 The physical therapists instructed participants to perform the abdominal drawing-in manoeuvre in a manner that elicited a contraction of the TrA while attempting to minimise internal oblique muscle activity. Once these skills were acquired and confirmed by the physical therapist through palpation and/or ultrasound imaging, participants were instructed to perform isometric TrA and LM cocontractions. During the supervised exercise sessions, tactile and visual feedback through palpation and real-time ultrasound imaging were used to enhance skill acquisition and the treating physical therapists used this information to ensure appropriate technique.27 The physical therapists re-evaluated these motor skills as needed based on their clinical judgement. While participants in the SPEC performed the same set of strengthening exercises as those in the GEN, they were instructed to maintain a TrA/LM cocontraction with each repetition.

Outcome measures and follow-up procedures

Participants in both groups underwent one preoperative and three postoperative evaluations performed by an examiner blinded to their group assignment. The preoperative assessment took place less than 2 weeks before surgery. The first postoperative assessment occurred 2 weeks after surgery and was immediately followed by an 8 week rehabilitation programme. Participants underwent a 10-week postoperative assessment on completion of the rehabilitation programme. The final postoperative assessment was performed by email or telephone 6 months after surgery.

The primary study outcome was low back pain-related disability. Secondary outcomes included pain intensity, global change, sciatica frequency and bothersomeness and LM muscle function. Low back pain-related disability was assessed with the modified Oswestry Disability Questionnaire (OSW). This version of the questionnaire has demonstrated good test–retest reliability, responsiveness and a minimum clinically important difference of six points.28 ,29

Low back and lower extremity pain scores on a 0–10 Numeric Pain Rating Scale were obtained for current pain intensity as well as the ‘best’ and ‘worst’ pain intensity in the preceding 24 h. The three scores were averaged to estimate overall pain intensity.30–32 Global rating of change (GRC) was assessed with a 15-point Likert-type scale ranging from −7 (“a very great deal worse”) to 0 (“about the same”) to +7 (“a very great deal better”).33 Sciatica frequency and bothersomeness were estimated using the Sciatica Frequency and Sciatica Bothersomeness indices resulting in possible scores of 0–25.34

Muscle function was assessed using brightness-mode, real-time ultrasound images of LM thickness acquired using a Sonosite Titan (Sonosite Inc Bothell, WA) system and a 60 mm, 2–5 MHz curvilinear array. Detailed accounts of this procedure have been reported elsewhere.35 ,36 We examined the per cent change in LM muscle thickness from resting to contracted states. The submaximal contraction strategy was a contralateral arm lift with a hand weight normalised to body mass and resulting in muscle activation approximating 30% of the maximum voluntary isometric contraction.35 Muscle images were acquired on the right and left sides at the L4/5 and L5/S1 spinal levels. Three images of each condition were taken and averaged to reduce variability.37 All images were stored and measured offline using National Institutes of Health (Bethesda, MD) Image J software (V1.43 m). Previous research examining the measurement properties of this technique has reported good levels of reliability36 ,38 and concurrent validity.39 Since this measure is not established for use at the L3/L4 spinal level, patients undergoing surgery at the L3/L4 disc were excluded from this analysis.

Data analysis

The sample size was based on detecting a clinically meaningful difference of 6 points28 on the OSW after 10 postoperative weeks with α=0.05 and a two-tailed hypothesis. Based on our previous work with patients with radicular leg pain, we assumed the SD of change to be 13.7 points.40 Recruiting 70 patients would provide 90% power to detect this treatment effect assuming a correlation between repeated measures of 0.20.

Statistical analyses were performed using IBM SPSS Statistics V.21. Treatment effects were estimated using separate, random-intercept linear mixed models for each outcome variable. Time (2-week postoperative baseline, 10 weeks, 6 months) and exercise group (GEN, SPEC) were modelled as fixed effects. The hypothesis of interest was the time by group interaction which we examined with pairwise comparisons of the estimated marginal means. We included the baseline outcome score as a covariate in each model. For the GRC outcome, the baseline score was the 2-week postoperative measure. For all other outcomes, the baseline score consisted of the preoperative measure. Consistent with the intention-to-treat principle, the linear mixed models estimated values for missing data based on the available scores; therefore, all participants randomised to a treatment group were included in the analyses of clinical outcomes. For the analysis of muscle function, we undertook an available case analysis; data from six patients whose LM muscles could not be visualised and from four patients who were operated at the L3/L4 spinal level were excluded. α-Level was 0.05 for all analyses.


Between April 2009 and July 2012, 105 patients were referred to the study and screened for inclusion. The participant flow through the trial and reasons for exclusion and loss to follow-up are presented in figure 1. Owing to limitations in study resources, enrolment was halted prior to reaching the target of 70 participants. Sixty-one participants were recruited, 29 randomised to the SPEC and 32 to the GEN. The original sample size estimation of 70 participants was based on achieving 90% power. A modified power analysis using the same a priori estimates indicated that 60 participants provided greater than 80% power to detect the identified treatment effects.

Thirty-three participants (54.1%) had surgery performed by one of seven neurological surgeons, and 28 (45.9%) by one of six orthopaedic surgeons. Tables 2 and 3 contain additional preoperative and 2-week postoperative baseline information.

Table 2

Demographic variables and self-report measures at the preoperative and 2-week postoperative baseline examinations

Table 3

Results of the intention-to-treat analysis comparing clinical outcome and muscle function between treatment groups

There were seven perioperative adverse events resulting in four participants requiring an additional surgical procedure. No adverse events were thought to result from the exercise interventions and six of seven participants were able to comply with their assigned exercise protocol. Two participants experienced reherniations at the level of surgery (both GEN). One spontaneous reherniation occurred during the supervised exercise period at postoperative week 3 and the participant was unable to continue with their exercise programme. One participant experienced a reherniation in week 11 resulting from a fall at home. Both participants underwent a second lumbar discectomy. Dural tears were reported in three participants (2 SPEC and 1 GEN). Two dural defects were recognised and repaired during the original surgery, and one participant developed a cerebral spinal fluid leak and spinal headache requiring surgical repair 5 days following the first surgery. One participant experienced short-term local swelling due to a postoperative seroma (SPEC). One participant experienced a short-term rash from an antiseptic solution used during surgery (SPEC).

Four participants in the SPEC and two participants in the GEN (all women, mean (range) age 38.2 [38–48] years, mean (range) body mass index 43.3(38.4–48.7)) were not measured with ultrasound imaging as the expected anatomical landmarks could not be visualised, presumably due to body habitus. Exercise compliance was 69% in the GEN and 59% in the SPEC.

The results of the intention-to-treat analyses revealed no time by group interactions. There were no statistically significant or clinically important between-group differences in disability, pain, global change, sciatica frequency, sciatica bothersomeness or LM muscle function at 10 weeks or 6 months (table 3 and figure 2). There were significant main effects of time (p<0.01) indicating improvements from baseline in disability, pain, sciatica frequency, sciatica bothersomeness and LM function (table 3 and figure 2). There was no difference in the postoperative scores of global change from the 10-week to 6-month follow-up.

Figure 2

Results of the intention-to-treat analysis comparing clinical outcome and muscle function among treatment groups. Data represent unadjusted mean values and 95% CIs. All comparisons are relative to postoperative baseline (2 weeks). *p<0.05. **p<0.01. †p<0.01 for the lumbar multifidus (LM)-specific group and p>0.05 for the general exercise group. ‡p<0.01 for the general exercise group and p>0.05 for the LM-specific group.


This randomised trial found that patients undergoing multimodal specific or general trunk exercise programmes 2 weeks after lumbar discectomy experienced similar short-term and intermediate-term clinical outcomes and changes in LM muscle function. Participants in both groups experienced improvements in disability, pain, sciatica frequency, sciatica bothersomeness and LM muscle function.

There are several plausible explanations for these results. First, optimal rehabilitation following lumbar disc surgery may be more general in nature and not depend as much on specific exercise components. While rehabilitation programmes appear to improve patient outcomes following lumbar disc surgery,3 these treatment effects may result from participation in any general exercise programme and not from an exercise approach targeting specific muscles such as the LM. After lumbar discectomy, patients express little understanding of the surgical procedure, confusion regarding activity limitations and develop a fear of movement and reinjury.41 Elevated fear-avoidance responses following lumbar disc surgery are associated with higher pain-related stress42 and worse clinical outcomes.43 It has been suggested that high-intensity postdiscectomy exercise programmes allow patients to confront their fears and insecurities and that this process may in part explain the effect of postoperative exercise programmes.44 The treatment protocols used in the current study attempted to address the psychosocial component of recovery by providing patients with detailed advice and supervision regarding movement, ergonomics and exercise. If the psychosocial aspects of treatment had a positive effect on clinical outcome, participants in both treatment groups may have benefited from this inclusion in their rehabilitation programme. Given the high prevalence of psychiatric comorbidity and its association with direct and indirect costs following lumbar discectomy,45 additional studies examining recovery from a biopsychosocial perspective are warranted.

Alternatively, understanding the effects of exercise following lumbar discectomy may require the consideration of unique patient attributes related to treatment outcome. While most trials until now have assumed this patient population to be homogeneous in nature, we do not know this to be the case. Preliminary evidence supporting the hypothesis of patient heterogeneity was reported by Flanagan et al46 who implemented a 12-week strength and endurance exercise programme after lumbar microdiscectomy. The authors reported that adaptations in lumbar extensor muscle performance were highly individualised. Moreover, Bouche et al47 reported differences in muscle morphology between patients with and without persistent pain following lumbar discectomy. Patients with persistent pain exhibited smaller muscle cross-sectional area and greater fat cross-sectional area in the LM and psoas muscles, respectively. Finally, Kjellby-Wendt and Styf48 identified clinical outcomes favouring an active training programme among patients with residual leg pain following lumbar discectomy. These findings raise the important question of whether patients with impaired muscle function, altered muscle morphology, residual leg pain or other factors should be treated differently following lumbar discectomy. It is possible that there exist subgroups of patients who require specific postoperative management strategies or, alternatively, who require no additional therapy following discectomy. The primary aim of the current study was not to identify treatment effect modifiers and patient subgroups; however, future trials investigating these questions may be of value.

The rate of lumbar disc reherniation in the current study (3%) was less than the rates reported previously.49 ,50 This finding is consistent with a recent systematic review, concluding that rehabilitation following lumbar disc surgery is not harmful.3 However, it is important to note that study limitations preclude firm conclusions of safety regarding the exercise protocols tested in this trial.

There are several important limitations that should be taken into account when interpreting the results of this study. First, we did not include a no intervention control group and doing so would have aided our understanding of the non-specific effects of treatment in this population. However, evidence from a meta-analysis of previous randomised trials has established that compared to no treatment, exercise programmes lead to decreased pain and disability following lumbar discectomy.3 Second, the large loss to follow-up in the current study may have biased the 6-month estimates, and consequently, these comparisons should be interpreted with caution. Furthermore, we did not follow patients beyond 6 postoperative months, and therefore the long-term effects of the postoperative protocols tested in this trial are unknown. Finally, the method used to estimate LM muscle function may not be well suited to the population studied in this trial. Although previous research has demonstrated ultrasound imaging to be reliable and valid among individuals with non-surgical low back pain, its measurement properties have not been studied in a postoperative population. Alternate measures of muscle morphology and function (eg, MRI, electromyography) may be more appropriate with postoperative patients.

Future research should focus on identifying the optimal components of rehabilitation programmes provided to patients following lumbar discectomy. Specifically, knowledge of different exercise approaches, dosages and timing are important as are the psychosocial aspects of recovery. Additionally, the identification of clinically relevant patient subgroups has the potential to increase the effectiveness and efficiency of care. Finally, knowledge of cost-effectiveness for postoperative rehabilitation will be important to develop health policy for patients undergoing lumbar discectomy.

The results of this trial indicate that multimodal rehabilitation programmes comprising specific or general trunk exercises have similar effects on disability, pain, global change, sciatica frequency and bothersomeness and LM muscle function at 10 weeks and 6 months following single-level lumbar discectomy. The 6-month outcomes should be interpreted with caution owing to participant loss to follow-up. While clinicians are encouraged to offer postoperative exercise programmes to patients following lumbar discectomy, the optimal exercise components have yet to be identified. In the light of these results, we suggest that local factors such as the individual patient characteristics identified by specific assessment findings as well as clinician expertise and patient preferences direct therapy selection when considering the types of exercises tested in this trial for inclusion in rehabilitation programmes following lumbar disc surgery.

What are the new findings?

  • Early specific or general trunk exercises have similar effects on clinical and muscle function outcomes following single-level lumbar discectomy.

  • Participants in both groups improved on most outcome measure.

  • The rate of adverse reactions experienced by participants in either exercise groups was similar to the rates reported previously.

How might it impact on clinical practice in the near future?

  • Clinicians should be directed by the individual patient characteristics identified by specific assessment findings, clinical expertise and patient preferences when considering the types of exercises tested in this trial for inclusion in a rehabilitation programme following lumbar disc surgery.


We wish to thank Faris Al-Odaibi, PhD MSc, PT and Nathan Savage, PhD DPT for their valuable assistance with data collection.


View Abstract


  • Contributors JH was involved in the study conception, study design, data acquisition and analysis, as well as drafting and approval of the final manuscript. JF was involved in the study conception, study design and data analysis, as well as critical manuscript revision and final approval. AT was involved in the data acquisition as well as critical manuscript revision and final approval. SK was involved in the study conception, study design, data acquisition as well as critical manuscript revision and final approval. DT was involved in the study conception, study design, critical manuscript revision and final approval.

  • Funding Orthopaedic Section of the American Physical Therapy Association.

  • Competing interests None.

  • Ethics approval University of Utah Institutional Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.