Abstract
Background Several studies have investigated differences in paraspinal muscle morphology between patients with low back pain (LBP) and control patients. However, inconsistencies in the results of some of these studies may limit generalizations.
Objective The purpose of this study was to systematically review studies evaluating paraspinal muscle morphology in patients with LBP and control patients, with a focus on the effects of blinding.
Data Sources An electronic search was performed with the use of relevant databases. Study quality was evaluated by means of the Newcastle-Ottawa Quality Assessment Scale.
Study Selection Case-control studies investigating paraspinal muscle size between patients with LBP and control patients who were healthy were included. Studies that compared paraspinal muscle size between symptomatic and asymptomatic sides of patients with unilateral LBP also were included.
Data Extraction Studies investigating the same outcome—at the same spinal level and for the same muscle and population—were pooled. Mean differences with 95% confidence interval were calculated for each study.
Data Synthesis Eleven studies were included. With 1 exception, all pooled results were significantly different statistically between groups, suggesting that paraspinal muscles are smaller in patients with chronic LBP than in control patients and on the symptomatic side of patients with chronic unilateral LBP. In patients with acute unilateral LBP, there was no significant difference between sides. A qualitative examination demonstrated a trend toward an increased effect size when outcome assessors were unblinded.
Limitations Limitations of this review include the small number of studies included and their small sample size. Misclassification of blinding status may have occurred when the study did not report blinding status.
Conclusions Evidence suggests that paraspinal muscles are significantly smaller in patients with chronic LBP than in control patients. Although no definite conclusion could be taken as to the effects of blinding, future imaging studies should consider the use of blinded outcome assessors.
Several low back pain (LBP) studies have emphasized the important role of paraspinal muscle morphology on the etiology, prognosis, and management of patients with this condition. Theories on the role of these muscles arose from imaging studies suggesting that patients with LBP have smaller multifidus muscle cross-sectional area (CSA)1–4 and more fatty infiltration compared with asymptomatic control patients who are healthy.2,5,6 However, there are inconsistencies among the results of these studies, with some suggesting a significant difference in paraspinal muscle morphology between patients with LBP and control patients, whereas others suggest no differences between patients with LBP and control patients.1,3,7 Inconsistencies also can be found in the results of studies evaluating paraspinal muscle morphology between the symptomatic and asymptomatic sides of patients with unilateral LBP.8–10
As with all scientific studies, specific methodological safeguards must be followed to minimize bias in study results. A specific methodological criterion that has been reported to be often lacking in the report of imaging studies looking at patients with LBP is blinding.11 Blinding of the outcome assessors is important to ensure the internal validity of a study.12,13 The use of unblinded outcome assessors has been reported, in other research designs, to consistently overestimate the treatment effect as much as 17% to 40%.13–15 Similarly, it has been shown that when diagnostic studies are conducted without adequate blinding of assessors, the diagnostic odds ratio can be overestimated as much as 30%.16
Although some studies and systematic reviews have reported on the effect of the use of unblinded assessors in diagnostic accuracy tests, we are unaware of any systematic reviews that evaluated the impact of this methodological issue in imaging studies looking at quantitative measures and, most importantly, on the morphology of paraspinal muscles. Performing measurements unblinded, that is, knowing patient's clinical history or having access to radiologist reports, can potentially influence the results and conclusions of the studies. Therefore, in an effort to bridge the aforementioned knowledge gaps, the purposes of this systematic review were: (1) to review the literature regarding the association between paraspinal muscle morphology and LBP and (2) to assess the difference between blinded and unblinded assessors when performing quantitative paraspinal muscle measurements with the use of imaging modalities.
Method
Data Sources and Searches
This systematic review was conducted following the PRISMA statement to ensure transparency and complete reporting of the results.17 To identify relevant articles, an electronic database search was conducted with the use of the earliest record to the current day on MEDLINE (1948 to February 2012), EMBASE (1980 to February 2012), and Scopus (1960 to February 2012). Key words for LBP and paraspinal muscle atrophy were included in the database search by use of MeSH (Medical Subject Headings of the National Library of Medicine) terms, as well as word truncations specific for each database. Search strategies and terms used for each database search are presented in the Appendix. There were no language limits in our search. Experts in the area of LBP and paraspinal muscle measurement were contacted and asked to review the list of selected studies and to identify any missed or unpublished study.
One reviewer performed the electronic database searches and 2 independent reviewers screened titles and abstracts for eligibility. Full articles were obtained and reviewed by 2 independent reviewers on the basis of a standardized inclusion criteria form. In the case of disagreement regarding whether a study met an inclusion criterion, the specific criterion was reviewed and discussed until a common consensus was reached. If additional information was needed to determine the inclusion of a study, the authors were contacted. We contacted 3 authors about the eligibility of 4 different studies.18–21 All authors responded to our request, and, on the basis of the responses, 1 study was deemed eligible for inclusion in the review.20 The bibliographies of all eligible studies were scrutinized to identify possible studies missed by the electronic search. Electronic databases also were searched to identify possible relevant systematic reviews that were previously published. Citation tracking with the use of ISI Web of Science and manual searches of the reference lists of previous reviews also were performed.
Study Selection
Studies were considered for inclusion if they met all of the following criteria: (1) case-control studies (patients with LBP versus control patients) or studies evaluating patients with unilateral LBP pain and comparing paraspinal muscle size between the symptomatic side and asymptomatic side (“normal” control); (2) patients with nonspecific LBP (with or without leg pain) or with specific LBP (eg, radiculopathy, disk herniation, sciatica, spinal stenosis, spondylitis, spondylolysis, spondylolisthesis, osteoarthritis, or facet joint osteoarthritis); (3) patients older than 18 years of age; (4) studies evaluating patients with acute LBP (<6 weeks), patients with subacute LBP (between 6 and 12 weeks), or patients with chronic LBP (>12 weeks) (mixed population studies were included only if data were available for each patient group separately); (5) use of ultrasound, magnetic resonance imaging or computed tomography (CT) scan to perform paraspinal muscle measurements; (6) inclusion of at least 1 of the following outcomes: multifidus or paraspinal muscle group (multifidus and erector spinae combined) CSA or functional CSA (FCSA) (fat-free area); (7) assessment at either L4–L5 or L5–S1, or at the level “below” or “same” level in cases of symptomatic and asymptomatic studies looking at a specific pathological spinal level. Exclusion criteria were: (1) previous spine surgery, (2) cauda equina syndrome, (3) spine fracture, (4) malignant or inflammatory disease, or (5) pregnancy.
Data Extraction and Quality Assessment
Methodological quality of the included studies was assessed by 2 independent reviewers through the use of the Newcastle-Ottawa Quality Assessment Scale for case-control studies. Any disagreement was resolved through discussion. Methodological quality score was not considered as an inclusion criterion. The Newcastle-Ottawa Quality Assessment Scale for case-control studies allows evaluation of studies on 3 specific criteria: (1) subject selection, (2) comparability between cases and control patients, and (3) exposure. A maximum of 10 stars can be awarded to each study.
Two independent reviewers extracted the data from the selected studies, which included sample size and mean (standard deviation) muscle size (CSA or FCSA) for each muscle at the levels of interest through the use of a standardized form. Authors were contacted if not enough information was provided to allow quantitative data analysis.1,9
Data Synthesis and Analysis
An effect estimate (mean difference) with 95% confidence interval (95% CI) was calculated for each study. The effect estimate was calculated to facilitate the interpretation of the results and allow for a simple quantification of the difference between the LBP group and the control group. An effect estimate also was calculated to determine the difference between the blinded and presumed unblinded studies. A meta-analysis with the use of RevMan 5.1 from the Cochrane Library (Copenhagen, Denmark: The Nordic Cochrane Centre, The Cochrane Collaboration [www.Cochrane.org]) was performed by pooling the results of homogenous studies. Case-control studies and studies that used patients with unilateral LBP, comparing muscle morphology of the asymptomatic with the symptomatic side, were evaluated separately. Results were pooled when studies used similar outcomes for the same muscle, at the same spinal level for a similar patient population. Weighted mean differences were used when the outcome measures were the same for the pooled studies, and standardized mean differences were used when the outcome measures were different. The effect estimate for each muscle of interest and selected spinal level was evaluated separately because previous reports suggested that paraspinal muscle atrophy is generally more important at L5–S1 than at L4–L5.3,22
When case-control studies reported the mean muscle size for the right and left sides, the mean measurement and standard deviations for both sides were pooled to calculate the effect estimate. The same calculation was performed if the authors reported separate mean measurements for men and women. In 2 particular studies, the authors compared 2 groups of patients with LBP (eg, unilateral LBP and bilateral LBP, moderate LBP, and severe LBP) with a group of control patients who were healthy.2,3 For the purpose of this study, the data for all LBP groups were pooled to calculate the effect estimate.
Homogeneity of the studies was determined with the use of I2 calculated by use of RevMan 5.1. As suggested by the Cochrane Handbook for Systematic Reviews of Interventions,23 I2 was used to assess statistical heterogeneity. An I2 value >50% indicates considerable heterogeneity, and thus studies with an I2>50% were pooled with the use of a random-effects model, whereas homogenous studies (I2<50%) were pooled with the use of a fixed-effects model.23 Both random and fixed-effects meta-analyses were performed with the use of the inverse variance method.
We had initially planned to run a meta-regression to evaluate the effects of blinding as a single covariate. However, we did not conduct this analysis because the small number of studies included in this review, particularly evaluating the same outcome (same muscle and spinal level), would lead to a low power and consequently increased chance of type II error. Therefore, all evaluations of the effects of blinding were performed qualitatively.
Results
Study Selection
The electronic search yielded a total of 145 studies after removal of duplicates. After reviewing titles and abstracts, 28 studies were potentially eligible for inclusion. One additional study was deemed potentially eligible after a Web of Science search of these articles.24 Therefore, a total of 28 studies were considered for inclusion, but only 11 original studies 1–4,7–9,20,25–27 were found to fulfill the inclusion criteria (Fig. 1).
Flowchart of systematic review inclusion and exclusion of case-control and symptomatic/asymptomatic studies. LBP=low back pain.
Methodological Quality
Methodological quality of the included studies is presented in the Table. The mean methodological score of all studies was 6 of a total of 10. Seven studies used blinded outcome assessors1,4,8,9,25–27 and 4 studies were deemed as unblinded.2,3,7,20 Studies were considered blinded when the authors clearly stated that the outcome assessor was blinded to all participants' clinical history or radiologist report at the time of imaging assessment. Studies were presumed to be unblinded when: (1) the authors did not explicitly state that the outcome assessor was blinded to participants' clinical histories and radiologist reports, (2) the outcome assessor was not blinded to group allocation, or (3) no information was mentioned concerning the blinding status of the outcome assessor.
Description of the Included Studiesa
Study Characteristics
Seven of the included studies were case-control studies, all comparing paraspinal muscle group or multifidus muscle size between patients with chronic nonspecific LBP and control patients.1–4,7,20,25 Three of the case-control studies were conducted with the outcome assessor blinded, and 4 were presumed as not blinded. Four studies were conducted with the use of patients with a clinical presentation of unilateral LBP, in which muscle size was compared between the symptomatic and asymptomatic sides.8,9,26,27 One of these studies looked at patients with acute symptoms,26 another compared a group of patients with acute and chronic symptoms,27 and the other 2 studies included patients with chronic pain.8,9 All unilateral LBP studies were blinded, and therefore comparison of the effect estimate between presumed unblinded and blinded studies was not possible. The pathological level (eg, disk herniation, degenerative disk disease) of the unilateral cases was identified through imaging. In these cases, muscle measurements were taken above, below, or at the same spinal level of the pathology. Refer to the Table for the study characteristics.
Case-Control Studies
Four case-control studies with a total of 111 patients with chronic LBP and 110 control patients compared the multifidus muscle CSA at L4.3,4,7,25 Three studies were conducted with the use of ultrasound3,7,25 and 1 with the use of CT scan.4 When authors provided measurements of the multifidus muscle CSA at the upper and lower L4, the data from the lower L4 were used.4 The methodological appraisal of these studies ranged from 2 and 6 on a 10-point scale. The pooled weighted mean difference of multifidus muscle CSA (cm2) at L4 was −1.08 (95% CI=−1.44, −0.72); thus, the multifidus muscle was significantly smaller in patients with chronic LBP than in control patients (Fig. 2A). The individual weighted mean difference of the 2 blinded studies was −1.09 (95% CI=−1.54, −0.64) compared with −1.11 (95% CI=−2.19, −0.02) for the 2 presumed unblinded studies. Thus, the mean difference between blinded and presumed unblinded studies was negligible (0.02 difference, 1.84% increase). The 95% CI of the presumed unblinded studies was much wider than that of the blinded studies. The results of the 2 presumed unblinded studies were contradictory, which could be the reason for the difference in 95% CI values. Hides et al3 found that the multifidus muscle was significantly smaller in patients with chronic LBP than in control patients, whereas Lee et al7 reported no significant difference between the 2 groups. Other possible reasons for the wide 95% CI include difference in sample size and measurement error.
Mean differences and 95% confidence intervals (95% CIs) of blinded and unblinded studies evaluating multifidus muscle cross-sectional area (in square centimeters) at (A) L4 and (B) L5 between patients with chronic low back pain (LBP) and control patients. IV=inverse variance.
Three studies compared the multifidus muscle CSA (cm2) at L5 in patients with chronic LBP, with a total of 75 patients with chronic LBP and 76 control patients,3,7,25 and all used ultrasound as imaging modality. The pooled weighted mean difference was −1.67 (95% CI=−3.15, −0.19); thus, the multifidus muscle also was significantly smaller in patients with chronic LBP than in control patients at L5 (Fig. 2B). The individual pooled mean difference of the blinded study was statistically significantly different: −1.75 (95% CI=−2.52, −0.98), whereas the pooled weighted mean difference of the 2 presumed unblinded studies was not statistically different: −1.61 (95% CI=−4.32, 1.09). Again, the 2 presumed unblinded studies included in this comparison had contradictory results. Hides et al3 found a significant difference between the 2 groups, whereas Lee et al7 reported no significant difference.
The paraspinal muscle group CSA (cm2) at L4–L5 (or lower L4) was evaluated in 2 studies, with a total of 84 patients with chronic LBP and 94 control patients.2,4 One study used CT scan4 and the other used MRI2 as the imaging modality. The pooled weighted mean difference was statistically significant: −1.93 (95% CI=−2.90, −0.95), suggesting that the paraspinal muscle group is smaller in patients with chronic LBP compared with control patients (Fig. 3). As shown in Figure 3, the mean difference for the blinded study was −1.71 (95% CI=−2.97, −0.45) compared with −2.25 (95% CI=−3.80, −0.70) for the presumed unblinded study. Although both studies showed that the paraspinal muscle group CSA was significantly smaller in the chronic LBP group, the effect estimate was larger (0.54 difference, 31.6% increase) for the presumed unblinded study. Because we did not perform a meta-regression, it is not possible to determine whether this difference is significant and related to the blinding status or whether other methodological issues are at the origin of this difference.
Mean differences and 95% confidence intervals (95% CIs) of blinded and unblinded studies evaluating paraspinal muscle group cross-sectional area (in square centimeters) at L4–L5 or lower L4 between patients with chronic low back pain (LBP) and control patients. IV=inverse variance.
In an attempt to facilitate comparisons between individuals, some authors provided the L4–L5 (or lower L4) paraspinal muscle group CSA muscle to bone or disk ratio, meaning that the muscle CSA was divided by the vertebral body or disk CSA of the same spinal level. Thus, another comparison was made for this group, which included 2 studies with a total of 43 patients with chronic LBP and 33 control patients.1,20 Lee et al20 conducted an MRI study to compare paraspinal muscle size between patients with degenerative flat back (degenerative deformity characterized by a loss of lumbar lordosis) and “normal” control patients, whereas Danneels et al1 compared patients with chronic mechanical LBP with a group of control patients. The pooled weighted mean difference was statistically significant (−0.26 [95% CI=−0.38, −0.14]), whereas the individual mean difference of the blinded study was −0.26 (95% CI=−0.48, −0.04) compared with −0.26 (95% CI=−0.40, −0.12) for the presumed unblinded study (Fig. 4). These results show that both studies demonstrated the paraspinal muscle group CSA was statistically significant smaller in patients with chronic LBP compared with the control patients and suggest that the effects of blinding were negligible.
Mean differences and 95% confidence intervals (95% CIs) of blinded and unblinded studies evaluating paraspinal muscle group cross-sectional area ratio at L4–L5 or lower L4 between patients with chronic low back pain (LBP) control and control patients. IV=inverse variance.
Symptomatic/Asymptomatic Comparison Between Sides
Most symptomatic and asymptomatic studies included in this review used paired t tests to evaluate the difference between sides, which control for the effect of intra-individual variation and take larger consideration of the within-group difference. However, we used the mean and standard deviations provided in the original manuscript to calculate mean difference and 95% CI estimates presented in this review as suggested in the Cochrane Handbook for Systematic Review of Interventions.23 Thus, it is important to consider that the 95% CI estimates presented in this review are more conservative and are not representative of the original study results9,26,27 because they ignore the intra-individual correlation and are based on different statistical analyses and assumptions.
Only 2 studies looking at patients with acute unilateral LBP were eligible for this review, with a total of 82 patients.26,27 Both studies were conducted with the use of MRI and measured multifidus muscle CSA (cm2) at the same spinal level as the pathology (disk herniation). The pooled mean difference was not significantly different between sides: 0.25 (95% CI=−0.31, 0.82) (Fig. 5A). This finding suggests that there is no significant difference between the multifidus muscle CSA of the symptomatic side and asymptomatic side in patients with acute unilateral LBP.
Mean differences and 95% confidence intervals (95% CIs) for symptomatic/asymptomatic studies comparing (A) multifidus muscle cross-sectional area (CSA) (in square centimeters) between sides in patients with acute low back pain (LBP) at the pathologic spinal level, (B) multifidus muscle CSA or functional CSA (in square centimeters) between sides in patients with chronic LBP at level below lumbar pathology, and (C) multifidus muscle CSA or functional CSA (in square centimeters) between sides in patients with chronic LBP at the pathologic spinal level. IV=inverse variance.
Two blinded MRI studies of patients with chronic LBP, with a total of 72 patients, compared the multifidus muscle CSA or FCSA (cm2) for the spinal level below the pathology.8,9 Because the studies included in this comparison investigated either CSA or FCSA, a standardized mean difference was obtained. The pooled standardized mean difference demonstrated a statistically significant different between sides: −0.40 (95% CI=−0.73, −0.07), although results of each study were contradictory (Fig. 5B). Note that the original mean multifidus muscle CSA presented by Barker et al8 varied between 46.6–64.6 mm2, which is very small and improbable for multifidus muscle measurement. An error when transforming units could have been the cause of this inaccuracy.
Three blinded MRI studies of patients with chronic LBP (N=125) compared the multifidus muscle CSA or FCSA (cm2) between sides at the pathological spinal level.8,9,27 The standardized mean difference was obtained because the outcomes were different (CSA or FCSA) between studies. The pooled standardized mean difference was −0.43 (95% CI=−0.68, −0.18) and statistically significant different between sides (Fig. 5C). However, the results are contradictory because 1 study8 showed a significant difference between muscle sides, whereas 2 studies did not show a significant difference.9,27
Discussion
The aims of this systematic review were: (1) to review the literature regarding the differences in paraspinal muscle morphology between patients with LBP and control patients who were healthy and between symptomatic and asymptomatic sides and (2) to assess the difference between blinded and unblinded assessors when taking quantitative paraspinal muscle measurements in patients with LBP. Our findings revealed that the multifidus and paraspinal muscle groups are smaller in patients with chronic LBP compared with control patients and that the multifidus muscle is significantly different between sides for patients with unilateral chronic LBP but not patients with acute LBP. When looking at the individual effect estimate of the presumed unblinded studies, there was a tendency for the effect estimate to be greater compared with blinded studies, but this difference did not change the direction of the results. This trend, however, was not observed in all of our comparisons. Caution should be taken when interpreting the difference in effect estimates between the blinded and presumed unblinded studies because the small sample size and small number of studies included in this review limit the evidence to suggest that blinding status could potentially lead to systematic error or differential bias. However, because blinding of the outcome assessors, for the most part, is easily achievable in this field, future related imaging studies should use blinded assessors to increase their internal validity.
The results of most studies suggest that multifidus and paraspinal muscle groups are smaller in patients with chronic LBP than in control patients who are healthy because all pooled estimates were statistically significant. In addition, our results showed that patients with chronic LBP appear to have more multifidus muscle atrophy at L5 than L4 because the pooled effect estimates were greater for the L5 multifidus muscle CSA comparison. Overall, the CSA of the multifidus muscle at L4 and L5 was reported as between 3.47 and 7.08 cm2 for patients with chronic LBP3,4,7,25 and between 4.61 and 7.65 cm2 for control patients,3,4,7,25 with a mean difference between groups of 1.08 and 1.67 cm2. Whether this difference is clinically significant is unclear, but, because the multifidus muscle acts primarily as a spinal stabilizer and controls the intersegment motion of the individual vertebrae,40 any atrophy or injury to the multifidus muscle is expected to compromise spinal function. Similar findings also were true for studies comparing muscle size between the symptomatic painful side and asymptomatic side of patients with chronic LBP. There were some conflicting results across studies, but the pooled effect estimates (standardized mean difference) were statistically significant for both the pathological spinal level and the level below, demonstrating that patients with chronic unilateral LBP have smaller muscles on the symptomatic side compared with the asymptomatic side. However, there is some evidence to suggest that patients with acute (<6 weeks) unilateral LBP have no significant difference in multifidus muscle CSA between the symptomatic side and the asymptomatic side. Among the studies comparing the paraspinal muscle group size between patients with chronic LBP and control patients, all 4 studies found that paraspinal muscle as a group was statistically significantly smaller in patients with chronic LBP.
Disuse, muscle denervation, and reflex inhibition have been proposed as possible mechanisms for muscle atrophy in patients with LBP.28,41 However, whether muscle atrophy, asymmetry, or fatty infiltration evaluated by imaging modalities results from LBP and pathology or represents a risk factor is still being argued.6 Thus, future prospective studies should focus on identifying whether muscle atrophy is a cause or a result of LBP. More studies are needed to better understand the role of the multifidus and paraspinal muscle groups in the etiology and management of common spinal disorders.
When looking at case-control comparisons of the multifidus muscle CSA at L4 and L5, our attention was drawn to the notably smaller effect estimate and nonsignificant results obtained from the Lee et al study.7 It is noteworthy that the results of that study are somewhat questionable because the reported multifidus muscle mean CSA was considerably larger than that in all of the other studies included in this comparison. The case definition and mean age of the chronic LBP group were similar to those characteristics in the other 3 studies; the major difference was that all patients (patients with chronic LBP and control patients) were active laborers, which is unlikely to be the only reason to explain the much larger multifidus muscle mean CSA reported by Lee et al.7 Moreover, the multifidus muscle becomes larger as we move caudally down the lumbar spine, which was not observed in the study by Lee et al.7
The use of different imaging modalities could have introduced bias into the results of this review, which may partly explain some of the conflicting results found in this review. Magnetic resonance imaging technology provides higher image resolution compared with ultrasound and CT scan and allows better detection of soft tissues such as fat and muscle.42,43 Atrophied muscles have more irregular boundaries and fatty infiltration, which greatly increase the level of difficulty when tracing the borders of the muscle of interest.43 Moreover, ultrasound does not allow the differentiation of muscle and fat tissues; thus, accurate distinction of muscle tissues from fat borders is virtually impossible.25,44,45 Furthermore, there is some evidence to suggest that intrarater and interrater reliability are better when lumbar paraspinal measurements are obtained with MRI compared with CT scan.43 Differences also exist between the image acquisition of ultrasound and MRI (or CT scan); ultrasound imaging is generally performed in a prone position and requires short acquisition time as opposed to MRI, which is performed in a supine position and necessitates considerably longer acquisition time. However, a previously published study comparing multifidus muscle CSA measurements obtained with MRI and ultrasound45 suggested that the 2 modalities can be used interchangeably. Although the latter study included a small sample of only 10 young women who were healthy (21–31 years old) and such measures have yet to be validated in older individuals with LBP conditions, we have decided to perform a mixed analysis in this study, pooling the results of the 3 imaging modalities.
To our knowledge, this is the first meta-analysis reviewing the literature on multifidus and paraspinal muscle group morphology changes associated with LBP. Moreover, this is the first study evaluating blinding when performing quantitative imaging measurement of paraspinal muscles. An extensive database search with the use of a strict standardized method is one of the primary strengths of this review. There are some limitations to this review that should be acknowledged. First, only a small number of studies met the inclusion criteria. Second, most of the included studies had small sample sizes, which decreased the study power and increased the possibility of making a type II error. Additionally, we might have misclassified some blinded studies as unblinded when authors did not provide information about blinding of the outcome assessor. Finally, because of the small number of studies, we were not able to perform a meta-regression to quantitatively evaluate the effects of blinding on the results of the studies.
Conclusion
The results of this systematic review suggest that multifidus and paraspinal muscle groups are significantly smaller in patients with chronic LBP than in control patients who are healthy and on the symptomatic side of patients with chronic unilateral LBP compared with the asymptomatic side. Thus, the role of paraspinal muscle morphology on the etiology, prognosis, and treatment of patients with LBP must be further investigated. When examining the estimates of the individual studies, there seems to be a trend toward an increase in effect size when the outcome assessor is presumed to be unblinded.
Future related imaging studies should use blinded outcome assessors to increase their internal validity. How blinding is achieved should be clearly stated in the method section of the studies because this aspect often is inadequately described in publications.46 A clear description of the assessor's experience also should be provided to help the reader judge the level of skill of the assessor and facilitate future clinical comparisons.11
Appendix.
Search Strategies
Footnotes
Both authors provided concept/idea/research design, writing, data collection and analysis, and consultation (including review of manuscript before submission). Ms Fortin provided project management. The authors thank Michele Crites Battié for her review of this work and helpful comments.
Ms Fortin received support from the European Union Community's Seventh Framework Programme (FP7, 2007–2013; grant HEALTH F2-2008-201626; project GENODISC). Dr Macedo is supported by the Canadian Institutes of Health Research and Alberta Innovates Health Solutions.
- Received November 14, 2012.
- Accepted March 11, 2013.
- © 2013 American Physical Therapy Association