Thirty individuals (n = 15 in each of the CLBP and HC groups) were recruited to participate in this study through advertisements on social platforms and at outpatient services. The inclusion criteria of the CLBP group were a medical diagnosis of non-specific LBP with pain and symptoms persisting for >3 months; pain quantified at >3 cm on the visual analog scale (VAS), for which medical treatment had been sought; age 18–40 years; and no history of other diseases. Participants were excluded if they were not aged 18–40 years or had LBP of traumatic or structural origin, LBP with neurological symptoms, or pain radiating to the lower leg(s), previous back surgery, spinal tumors or infections, or neurological and/or musculoskeletal disorders unrelated to LBP (26). The diagnosis of CLBP was based on the diagnostic guidelines published by the American College of Physicians and the American Pain Society (27).

The inclusion criteria for the HC group were the absence of LBP in the preceding 2 years and no history of other diseases (28). The exclusion criteria for HCs were age outside of the specified range, pregnancy even if the person who matched the inclusion criteria, or history of other diseases or surgery.

The study was approved by the Human Subjects Ethics Subcommittee of the first affiliated hospital (Registration Number: 2020425) and was performed in adherence to the tenets of the Declaration of Helsinki. The study had been recorded in Chinese Clinical Trial Registry (Registration Number: ChiCTR2100043113). All participants were enrolled in the study after obtaining written informed consent.

The degree of pain was quantified using a VAS, comprising a 10-cm scale with scores ranging from 0 (no pain) to 10 (severe pain). Participants in the CLBP group were instructed to complete the ODI (29) and RDQ (30) for the assessment of their pain-related disability and dysfunction in daily life. The ODI comprised 10 items on pain and activities of daily living (pain intensity, lifting, walking, social life, personal care, sitting, standing, sleeping, sex life, and traveling), and each item was scored from 0 to 5. The total score was calculated by summing the individual item scores. A higher total score indicated greater disability. The RDQ score was also calculated by adding the scores of the 24 items in the questionnaire. Each question had a response of “yes” (1 point) or “no” (0 points), with the total scores ranging from 0 (no disability) to 24 (maximum disability).

A multifrequency body composition analysis system (Imp SFB7, Inc., Sydney, NSW, Australia) was used to measure the electrical impedance of the participants' LPMs. This medical multifrequency device is capable of applying currents simultaneously at various frequencies from 2 kHz to 1 MHz. Two systematically trained rehabilitative clinicians were involved in data collection but were blinded to the participants' groups. The electrode arrangement for LPM testing referred to the principles described by Rutkove (13) and Hu et al. (17) and followed the standardized measurement methods of the Imp SFB7 for measuring the electrical properties of the LPM (31, 32). The Imp SFB7 is a linear EIM device, which was different from the device used in Ching et al.'s study (21).

The participants were instructed to lie in the prone position with a soft pillow placed below the abdomen to keep the lumbar arch straight; thus, fully exposing the lower back. Before testing, the skin covering the LPM area was cleaned with 75% alcohol and was left to dry thoroughly. On each side, the upper edge of the third electrode was placed on the skin, 2 cm adjacent to the spinal process, and at the horizontal level of the spinal process to the anterior superior iliac crest. In the longitudinal direction, the 1st, 2nd, 3rd, and 4th electrodes (1st and 4th are current electrodes and 2nd and 3rd are voltage electrodes) were arranged vertically along the ridge process and were separated by 20, 30, and 20 mm, respectively (Figure 1A). The phase angle (PA), resistance (R), reactance (X) and Z values were recorded at 5, 50, 100, and 200 kHz. The contralateral LPM was similarly assessed (Figure 1B). The model of the electrode plates was L-DEX (Lymphedema Dual Tab Electrode, Pinkenba QLD 4008, Australia). To avoid possible bias, the parameters were measured three times in each participant. The operators were two physiotherapists with more than 3 years of clinical experience and were blinded to the participants' groups.

Impedance can be described mathematically. The EIM parameters commonly comprise R, X, PA, and Z. X consists of a combination of the two forms of reactance, capacitive reactance X_C, and inductive reactance X_L. The PA is calculated via standard trigonometric relationships, as follows (12): (1)PA = arctan (X/R) Z represents the complex impedance of the circuit, which is a combination of its inherent R and its X. The latter, produced by the presence of capacitors and/or inductors in the circuit, obstructs the current flow (12). Z is calculated as follows: (2)Z=R2+(XC-XL)2 In medical applications, this equation can be simplified by ignoring the X_L term, because inductance is thought to play a minimal role in standard bioimpedance measurements, and the X_C term can be simply written as X (12). In this study, all these parameters were processed using the ImpediMed SFB7 multifrequency analysis system.

Statistical analysis of the parameters was performed using SPSS version 26.0 (IBM Statistics, New York, NY, USA). The Kolmogorov–Smirnov test was used to determine normality of data distribution. The results showed that parameters in the two groups were normally distributed (the Kolmogorov–Smirnov test; p > 0.05), except for age and body mass index in the HC group, and RDQ and X at 5 kHz in the CLBP group. Independent sample t-tests were used to compare the basic characteristics of the two groups, and the chi-square test was used to compare sex. As age was significantly different between the CLBP and HC groups (Table 1), it was regarded as a covariate in the analysis of variance (ANOVA). One-way ANOVA with age as covariate was used to compare R, X, Z, and PA between the bilateral LPM in the two groups at each current frequency. Spearman's rank correlation coefficient was used to assess the correlation between the functional indexes and EIM parameters. Two-tailed p-values < 0.05 were considered statistically significant.

This study included 30 participants (15 in each of the CLBP and HC groups). There was no statistically significant difference in sex, height, weight, and body mass index between the two groups (p = 0.247, p = 0.979, p = 0.905, and p = 0.805, respectively). However, the average age differed between groups (p = 0.002). The baseline characteristics of the two groups are shown in Table 1.

Since the participants' ages differed between the two groups, it was regarded as a covariate factor in the analysis of variance used to calculate the electrical impedance parameters. We found that the R values of the left LPM were higher in the CLBP group than in the HC group at the current frequencies of 5 (p = 0.047), 50 kHz (p = 0.019), 100 (p = 0.017), and 200 kHz (p = 0.016). Furthermore, those of the right LPM differed between the two groups at 50 (p = 0.047), 100 (p = 0.043), and 200 (p = 0.044), but not at 5 kHz (p = 0.084).

The X values of the bilateral LPM showed no significant difference between the CLBP and HC groups at any of the four current frequencies (p > 0.05 for all).

The PA values of the bilateral LPM were lower in the CLBP group than in the HC group at 5 (left, p < 0.001; right, p = 0.005), 50 (left, p = 0.003; right, p = 0.007), 100 (left, p = 0.019; right, p = 0.015), and 200 kHz (left, p < 0.001; right, p = 0.024).

The trend of the Z values was similar to that of the R values. The average Z values of the bilateral LPM were higher in the CLBP group than in the HC group at 50 (left, p = 0.020; right, p = 0.050), 100 (left, p = 0.017; right, p = 0.044), and 200 kHz (left, p = 0.017; right, p = 0.045), as was the Z value of the left LPM at 5 kHz (p = 0.048). However, this was not the case on the right side at 5 kHz (p = 0.085). The details are shown in Table 2 and are visually represented in Figure 2.

We calculated the correlation of the VAS, ODI, and RDQ scores with R, X, and PA in the CLBP group. The Kolmogorov–Smirnov test showed that the VAS, ODI, and EIM data of each side of the LPM at 50 kHz were normally distributed (p > 0.05 for each parameter), but the RDQ scores were not (p < 0.05). Because VAS, ODI, and RDQ are discontinuous variables, Spearman's rank correlation coefficient was applied in the correlation analysis.

The RDQ scores had a significant negative correlation with R of the LPM on both sides (left, r = −0.547, p = 0.035; right, r = −0.582, p = 0.023). ODI scores showed a similar negative correlation with R of the LPM on both sides (left, r = −0.523, p = 0.045; right, r = −0.581, p = 0.023). RDQ scores showed a positive correlation with PA of the right LPM (r = 0.521, p = 0.046) but no significant correlation with that of the left LPM (r = 0.428, p = 0.112) (Figure 3). No significant correlation was observed between the questionnaire scores and the other EIM data (p > 0.05).

Electrical impedance myography can be used to assesses disease-induced changes to muscle's normal composition and architecture, including myocyte atrophy and loss, edema, reinnervation, and the deposition of endomysial connective tissue and fat (13). Skin, subcutaneous fat, and bone underlying the voltage electrodes do not significantly affect the results. The inherent resistivity of skeletal muscle is over an order of magnitude lower than that of fat and two orders of magnitude lower than that of bone, and together with the high detector impedance, these, therefore, have a minimal effect on the muscle impedance signatures (33). The impact of skin-subcutaneous fat layer thickness on EIM measurements had been previously reported by Tarulli et al. (34). They found that skin-subcutaneous fat layer thickness did not contribute substantially to the phase measured by linear EIM. The intracellular and extracellular matrices of the muscle tissue act as resistors, and any atrophy that reduces the CSA of muscle tissue increases the R value (35). On the other hand, changes in the X value are related to changes in the features of the cell membrane and layers of tissue (36). Coutinho et al. (37) suggested that changes in X were likely to be caused by changes in the features of the cellular membranes. In this study, the R and Z values increased significantly at 5 kHz in the CLBP group. However, no differences in the X and PA values were observed between the two groups. Because of the membrane permeability effects that occur at low frequencies, only a high-frequency current can flow through the intracellular compartment, and the current flow can be interrupted at 5 kHz. Our study showed that the R and Z values in the CLBP group were increased, and that the PA was decreased, as compared to the HC group. There was no significant difference in the X value at 50, 100, or 200 kHz. This was partly consistent with the results of the study by Ching et al. (21), who assessed the electrical properties of the LPM in patients with acute LBP and reported that the intracellular R value at 100 kHz was larger in these patients than in healthy participants.

The R and Z values that were elevated in the CLBP group indicated that the LPM might not have significantly atrophic changes or that changes were induced by localized edema of the muscle. Statland et al. performed a prospective study on 16 bilateral limb and trunk muscles in 35 genetically defined and clinically affected patients with facioscapulohumeral muscular dystrophy, using EIM at 50 and 100 kHz (18). They showed that the R and PA did not change, but that X showed differences. R did not increase in the patients with facioscapulohumeral muscular dystrophy. These results are in contrast to those of this study. The volume of the muscle has an impact on the EIM measurement (38) and localized oedema and inflammation which reduce the resistance often appeared in acute or subacute autoimmune conditions (25). CSA negatively correlated with the muscle's inherent impedance, and the decrease in the CSA score resulted in an increase in R (38). Sweeney et al. reported that the size and percentage change in the thickness of the lumbar multifidus muscle in patients with CLBP at the L4/L5 level did not differ between the two sides or from that in HCs (39). We also found no significant differences in EIM parameters between the two sides of the LPM in our patients (Figure 2). Thus, we concluded that the volumes of the LPM were bilaterally symmetrical in our patients. A previous study assessed EIM sensitivity to experimental inflammation induced by localized intramuscular injection of λ-carrageenan. Substantial reductions in phase and X in 5 kHz EIM values persisted at 48 and 72 h in the λ-carrageenan-treated group (40). However, changes in muscle composition can also affect the R and Z values. Increased fatty infiltration or connective tissue will have an impact on the measured electrical impedance (13).

The phenomenon of increased R and Z in the study may be due to extracellular remodeling, such as deposition of fat or and connective tissue. Most chronic myopathies are accompanied by substantial compositional changes in the muscle, including the deposition of fat and connective tissue (12). Goubert et al. (41) used MRI to evaluate differences in muscle structure and function in patients with RLBP, non-continuous CLBP, and continuous CLBP. They found that fat CSA and lean muscle fat index were significantly higher in the multifidus and erector spinae in cases with continuous CLBP than in those with non-continuous CLBP and recurrent LBP. Moreover, the fat area within the multifidus in patients with CLBP was larger than that in the controls (42). Capacitance, due to membrane structure and function, is a function of reactance and causes the current to lag behind the voltage, creating a phase shift or PA13 that is quantified as the angular transformation of the ratio of reactance to resistance (38). Based on these observations, the R value was increased and the X value was not changed in the CLBP group, indicating that the cellular membranes of the LPM in these patients might be undamaged, and that the number of cells was possibly similar to that in healthy individuals. This corroborates our finding that the peak value did not shift at different current frequencies.

To minimize the effect of muscle anatomy on impedance analysis, PA was used to describe the relationship between R and X (43). This evaluates the membrane oscillation properties of the muscle. In this study, the PA values were decreased in the CLBP group, which indicated that the membrane oscillation properties of the LPM were changed in patients with CLBP. Li et al. (44) suggested that the decrease in PA can be demonstrated by two representative patterns of the R and X association. The first pattern shows larger R values in the muscles of patients with CLBP than in the HCs and similar X values in the muscles of both groups across multiple frequencies. In the second pattern, the X values were smaller in the muscles of patients with CLBP than in the controls, whereas R remained similar in the muscles of both groups. In this study, the results demonstrated the first pattern. This indicated that the LPM in young adults with CLBP might have increased fatty infiltration, and the cellular membrane oscillation properties were decreased. However, the cellular membranes remained intact, and the number of cells did not change. Hu et al. have analyzed the correlation between muscle structures and electrical properties of the tibialis anterior in subacute stroke survivors (17). They found the decreased PA and elevated R and X in the affected tibialis anterior muscle of the subacute stroke survivors related to the lost muscle fibers. Therefore, the muscle fibers of LPM in the patients with CLBP in this study might also have been lost to some degree, with increased fatty infiltration and connective tissue formation. Alternatively, there were multiple types of pathological changes that contributed to the results. Nonetheless, this is an inference based on the results of this study and requires further research.

Multifrequency EIM is an extension of the linear EIM technique, which opens a previously unexplored dimension for EIM testing, potentially enhancing its ability to evaluate neuromuscular conditions (33). Multifrequency EIM overcomes the drawbacks of using a single frequency and provides a method for further assessment of the disease states of the muscle. In some muscle disease states, the frequency spectrum showing changes may differ (45). A previous study on R, X, and PA vs. frequency drew comparisons between healthy individuals and similarly aged patients with inclusion body myositis and found changes in the cellular structure due to the disease. However, the study reported that the disease created a shift in the peak phase to much higher frequencies (13). In this study, we investigated the electrical properties of the LPM of the two groups, measured at four different current frequencies, to explore whether there were results differed under different frequency conditions. The peak phase did not exhibit a significant shift at different current frequencies (Figure 2). This result indicates that the cellular structure of the LPM in patients with CLBP was not significantly altered as compared to that in the healthy individuals.

The ODI is an effective method for measuring disability in patients with back pain presenting with varying degrees of severity and different causes. This index has stood the test of time, and it has been used in various clinical situations in most developed countries worldwide (20). On the other hand, the RDQ is a self-rated assessment of physical function in patients with back pain. It is a short and simple method, making it suitable for combining with other measures of function (e.g., work disability or psychological) in research settings. The ODI and RDQ have good content validity, construct validity, test–retest reliability, and responsiveness (30). These two methods are often used to assess the activities of daily life functions in patients with CLBP (46).

A standardized current frequency of 50 kHz is commonly used (12). In this study, we found that the ODI scores of the patients with CLBP correlated negatively with R, although both were increased in the CLBP group as compared to the HCs, at 50 kHz. Correlation analysis showed that RDQ scores correlated negatively with R, and positively with PA, whereas there was no significant correlation between X and RDQ scores in the CLBP group (Figure 3). These results were somewhat unexpected. The correlation plots indicated that a lower R was associated with worse disease, but patients had higher R values than did the HCs. This implied that the disease status of the LPM might not be as serious in the less-disabled patients. We considered two explanations for these results. One is that the patients with scores indicating greater disability might be at the “early stage” of CLBP; thus, the structural changes in the muscle were less than in those who had lower scores. Patients with CLBP lose muscle fibers or have fatty infiltration and increased connective tissue in the LPM, which tends to generate compensatory adaptation strategies for daily living (47), preventing significant disability in these persons. Since self-evaluated questionnaires and the electrical properties of the LPM are independent phenomena, a negative correlation between these parameters might be caused by multiple factors. Another explanation is that the pathological relationship between function and the LPM electrical properties was observed in only a small number of patients. Changes in the LPM physiology or pathology may be due to a combination of factors (48). Given the limited sample size and population of this study, this result might be not applicable to all patients with CLBP. Nonetheless, the results of correlation analysis in the study offered basic evidence that the EIM technique can be applied for electrical property measurements in CLBP. As an objective and easily operable measurement method, EIM should be widely applied for the diagnosis of CLBP in future.

This study also had some limitations. First, the small sample size could have limited these findings and might present pathological characteristics of the LPM in only a part of this patient population. Second, the age of the participants included in the study was a limitation. Measurements were taken only for the LPM in young adults, thus limiting the generalizability of the results. Difference in the age and sex of the patients with CLBP may consequently present different structural or functional status in their LPMs (49, 50). Third, the anatomical structure of the lumbar crest did not permit the measurement of the EIM parameters in the transverse direction; hence, the anisotropy ratios of each variable were not calculated. Muscle anisotropy represents the degree of columnar order in the arrangement of the fibers (51); however, the arrangement of the LPM fibers in CLBP was not assessed in this study. Fourth, the location of the pain in the low back might result in different electrical impedance values. Since some of the patients in the study could not state the location of their pain accurately, we did not distinguish patients into unilateral and bilateral pain groups. This should be investigated in future.

The electrical properties of the bilateral LPMs differed between the patients with CLBP and HCs, regardless of the current frequency used. The R and Z values were elevated, PA values were decreased, and no significant changes occurred in the X values in young adults with CLBP, as compared to HCs. This may indicate that the LPM of the patients with CLBP has fewer muscle fibers with increased fatty infiltration or connective tissue, and that the cellular membranes are not damaged, but that their oscillation properties are altered. The altered electrical properties of the LPM correlated with disability in daily life. This suggests that exploring the electrical properties of the LPM in patients with CLBP is meaningful and can contribute to discovering the pathological characteristics of the muscle composition. EIM, as a novel technique for assessing CLBP, should have widely applications in future.