Factors Affecting Disability With Nonspecic Low Back Pain in Different Subgroups: A Hierarchical Linear Regression Analysis

Purpose: To systematically explore how disability is inuenced with layers (demographic level, pain level and psychosocial factors) in nonspecic low back pain (NSLBP) in different subgroups. Methods: This is a cross-sectional study that compared two different subgroups in NSLBP at two hospitals. Hierarchical multiple regression analysis was performed to analyse factors affecting disability in different groups (overall group, acute group and subacute/chronic group). Results: In the overall group (n = 235), explanatory power increased with each additional variable in the order of demographic characteristics, pain intensity and psychosocial factors. Pain intensity (ß = 0.219), Pain Catastrophising Scale (PCS) (ß = 0.175) and Pain Self-Eciency Questionnaire (PSEQ) (ß = −0.370) were signicantly associated with disability. In the acute group (n = 65), explanatory power improved with each additional variable for the disability in the order of demographic characteristics, pain intensity and psychosocial factors. Ultimately, pain intensity and PSEQ had signicant explanatory power, with pain having the most inuence. However, in the subacute/chronic group (n = 170), explanatory power increased with each additional variable in the order of demographic characteristics, pain intensity and psychosocial factors and all, including psychosocial factors, had a strong impact, with self-ecacy having the most substantial impact on disability. Conclusion: Depending on the duration of the disease, the factors affecting the disability differed, with pain having more inuence than psychosocial factors in the acute phase and psychosocial factors having more inuence in the chronic phase.


Introduction
Low back pain (LBP) has become a global problem, and medical costs are increasing every year [1]. In LBP, 80-90% are classi ed as nonspeci c low back pain (NSLBP), which cannot be attributed to a structural problem [2]. In NSLBP, one of the most important outcomes is disability, and it is widely known that pain in uences disability [3]. Pain intensity at the onset can contribute to chronicity [4]. Also, psychosocial factors in uence the disability of LBP, especially chronic low back pain (CLBP) [5].
Speci cally, psychosocial factors include fear of movement, catastrophic thoughts of pain and selfe cacy. Alma et al. reported that the factors that contributed to chronicity after one year were high fear of movement and older age as well as the intensity of pain at the beginning of the disease [6]. Fear of movement is known as the fear-avoidance model [7] and is believed to contribute to functional impairment and chronicity of LBP [8].
Fear of movement is an important predictor of failure to return to work [9]. Furthermore, a study on Saudi Arabians with CLBP reported that there was an association between psychosocial factors, including fear of exercise and disability [10]. In terms of self-e cacy, not only is it related to pain intensity and disability but also a correlation between low self-e cacy and physical function has been reported [11]. For example, in LBP patients, pain intensity and lumbar potential and stability during lifting movements were reduced [12]. In terms of catastrophic thinking for pain, in a systematic review of LBP by Wertli et al., they reported that catastrophic thinking is associated with disability and may contribute to chronicity [13].
Thus, from previous studies, it is clear that many factors, such as pain intensity, psychosocial factors and age, are involved in the disability of LBP. However, no studies have systematically examined which factors have more substantial impact on the disability in NSLBP among pain, various psychosocial factors and even demographic data such as age. Furthermore, most of such studies were on CLBP, and few studies examine how pain intensity and psychosocial factors affect the disability in NSLBP as a whole or acute LBP. That is to say, depending on the duration of the disease, factors affecting the disability of LBP may differ, but we have not found any studies that examine the magnitude of the effects of pain intensity and psychosocial factors in different duration of the disease.
Therefore, this study aimed to systematically explore how disability is in uenced with layers (demographic level, pain level and psychosocial factors) in NSLBP using hierarchical multiple regression analysis and to examine them by the duration of the disease to see if there are any different effects. Our hypotheses were as follows: First, pain intensity in uences the disability, and psychosocial factors still in uence the disability, even based on pain intensity. Second, the in uence of psychosocial factors would be more remarkable in CLBP, whereas psychosocial factors are less in acute LBP.

Materials And Methods
Study design This is a cross-sectional study that compared two different subgroups in LBP at two hospitals was performed according to the STROBE statement. Written informed consent was obtained from the patients before the study. The study was conducted in compliance with the Declaration of Helsinki and was conducted with the approval of the Ethics Committee of Sapporo Maruyama Orthopaedic Hospital (no. 0039) as well as the Faculty of Health Sciences, Hokkaido University (approved no. 20-58).

Participants
This study included patients collected from the two medical institutions between January 2019 and December 2020. The inclusion criteria of the participants were as follows: (1) over 20 years old; (2) diagnosed as NSLBP by an orthopaedic surgeon, with pain occurring in the lumbosacral region with radiation limited above the knee, with or without referred pain and signs of nerve root compromise [2]; (3) no use of painkillers at the initial physiotherapy session and (4) being able to understand the Japanese language well enough to complete the questionnaires independently. The exclusion criteria included (1) presence of spinal deformities; (2) any history of surgery on the spine; (3) pregnancy; (4) rheumatological and/or in ammatory disease; and (5) other serious pathologies (e.g., pyogenic spondylitis and cauda equina syndrome). Subjects were divided into three groups: the "all LBP group," the "acute LBP group" with an onset of less than four weeks (28 days) and the "subacute/CLBP group" with an onset of more than 28 days.

Measures
Pain intensity NRS was used for LBP intensity. The 11-point NRS was used to measure pain intensity [14][15] to determine average LBP intensity on the day of evaluation. A score of 0 indicates no pain, and a score of 10 indicates the worst imaginable pain. Alghadir et al. assessed knee pain intensity with multiple scales, including the NRS, and demonstrated that NRS was valid [19].

Disability/functional status
The Japanese version of the RMDQ, which contains 24 dichotomous scales (No = 0; Yes = 1), was used [20]. A higher total score indicates more signi cant impairment due to LBP. Concurrent validity of the Japanese RMDQ, high internal consistency (Cronbach's alpha for all items = 0.86) and high test-rest reliability (0.95) have been clari ed [20].

Pain catastrophising
To assess pain catastrophising, the Japanese version of the PCS, which consists of 13 items from 0 to 4, was used [21]. A higher total score indicates more signi cant de cits of pain catastrophising, with a total score of 30. This is the cut-off point for a clinically appropriate level of catastrophising [16]. The construct validity and high internal consistency of the Japanese version of the PCS (Cronbach alpha for all items = 0. 89) were shown [22].

Fear of movement
The Japanese version of the TSK, which consists of 17 items from 1 to 4, was used. A higher total score indicates strong fear of movement, with a total score of 37 being the cut-off for strong fear of movement [23]. The concurrent validity and high internal consistency of the Japanese version of the TSK (Cronbach alpha for all items = 0.85) have been con rmed [23].
Pain self-e ciency The Japanese version of PSEQ, which consists of 10 items from 0 to 6, was used for pain self-e ciency [24]. A high total score indicates stronger self-e cacy for pain.

Data analysis
We excluded subjects with missing data in the demographic characteristics and measures. Descriptive analyses were used to understand the participants' characteristics. The test differences were compared between the two groups of patients with acute and subacute/chronic LBP for categorical variables using the Chi-square test. In contrast, Welch's test was conducted for continuous variables. The correlation of each outcome was examined for different groups (overall, acute and subacute/chronic groups) using Pearson's product-moment correlation coe cient to evaluate the association of other factors with disability. In this study, p-values of ≦ 0.40, 0.40-0.75 and ≧ 0.75 were considered to indicate weak or no correlation, moderate correlation and strong correlation, respectively [25].
Furthermore, factors affecting disability in different groups were analysed through hierarchical multiple regression analysis. In multiple regression analysis, the dependent variable included the RMDQ for disability due to LBP, and the independent variables included the characteristics of the participants, pain intensity and psychological factors (PCS, TSK and PESQ). For all LBP group, the acute LBP group and the subacute/chronic LBP group, hierarchical multiple regression analysis was performed. Demographic characteristics including age, body weight and BMI were rst entered (step 1) as control variables, followed by pain intensity in the second step (step 2) and nally, the psychological factors including the scores of PCS, TSK and PESQ were added in the third step (step 3) to test unique associations between the psychological factors and patients' disability beyond the effects of patients' demographic characteristics and pain intensity of LBP. HAD ver. 16.1 (Hiroshi Shimizu, Nishinomiya, Japan) was used to perform all statistical analyses [26]. All p-values are two-sided. The alpha level was set at 5%.

Demographic data
In this study, the total number of participants was 235 (65 participants in the acute group and 170 participants in the subacute/chronic group). Table 1 shows the participants' demographic characteristics, including psychosocial factors. Comparison of intergroup pain intensity, disability and psychosocial factors Table 2 presents a comparison of pain intensity, disability and psychosocial factors between the acute and subacute/chronic groups. Only the value of TSK showed a signi cant difference between the two groups. Correlation of disability, pain intensity and psychological factors Table 3 shows a matrix of the correlations among the overall, acute and subacute/chronic groups. In the subacute/chronic LBP group, the highest correlations were found between RMDQ and PSEQ, r = − 0.510.  Hierarchical multiple regression analysis   Tables 4-6 show the hierarchical multiple regression analysis results for the overall LBP, acute LBP and subacute/chronic LBP groups.

Overall LBP group
The overall LBP group consisted of 235 participants. Table 4 shows the results. In step 1, the demographic variables were added. The results of hierarchical linear regression showed that participants' demographic factors, tested in step 1, explained 1.3% of the variance in the disability. There was no signi cant association between all demographic factors and worse disability. In step 2, we included pain intensity as a variable. The results of the hierarchical linear regression showed that the pain intensity, tested in step 2, explained an additional 7.6% of the variance in disability. Furthermore, there was a signi cant association between pain intensity and worse disability (p < 0.05). In step 3, psychological factors, which are TSK, PCS and PESQ, were included. The results of the hierarchical linear regression showed that psychological factors, tested in step 3, explained an additional 23.3% of the variance in disability. Furthermore, those that were signi cantly associated with disability were LBP (ß = 0.219), PCS (ß = 0.175), and PSEQ (ß = −0.370). This nal model explained 32.2% of the variance in participant's disability. The acute LBP group consisted of 65 participants. Table 5 presents the results. In step 1, the demographic variables were added. The hierarchical linear regression results showed that patients' demographic factors, tested in step 1, explained 1.3% of the variance in disability. There was no signi cant association between all demographic factors and worse disability. In step 2, we included pain intensity as the independent variable. The results of hierarchical linear regression showed that the pain intensity, tested in step 2, explained an additional 7.2% of the variance in disability. There was a signi cant association between pain intensity and worse disability (ß = 0.275; p < 0.05). In step 3, psychological factors, which were TSK, PCS and PESQ, were included as the independent variables. The results of hierarchical linear regression showed that psychological factors, tested in step 3, explained an additional 14.5% of the variance in disability. Finally, those signi cant relations to the disability were LBP (ß = 0.320) and PSEQ (ß = −0.316). This nal model explained 23.0% of the variance in patients' disability. The subacute/chronic LBP group consisted of 170 participants. Table 6 presents the results. In step 1, the demographic variables were added. The results of hierarchical linear regression showed that patients' demographic factors, tested in step 1, explained 11.0% of the variance in disability. All demographic factors were not statistically signi cant. In step 2, pain intensity was included as variables. The results of hierarchical linear regression showed that the pain intensity, tested in step 2, explained an additional 8.2% of the variance in the disability. There was a signi cant association between pain intensity and worse disability (p < 0.001). In step 3, psychological factors, which are TSK, PCS and PSEQ, were included. The results of hierarchical linear regression showed that psychological factors, tested in step 3, explained an additional 29.1% of the variance in disability. Ultimately, those that were signi cantly related to disability

Discussion
In this study, how basic characteristics, pain intensity and psychosocial factors were related to disability using the hierarchical multiple regression analysis for NSLBP was systematically examined. Overall, with the addition of variables in demographic characteristics, pain intensity and psychosocial factors, the explanatory power (predictive accuracy) increased. Pain intensity had a signi cant effect in the acute LBP group and all psychosocial factors had a signi cant effect, especially the PSEQ value, in the subacute/chronic LBP group. The results also showed that pain and PSEQ values were affected in all groups, suggesting that pain intensity and self-e cacy are key factors affecting the disability in patients with NSLBP.
In all groups, pain self-e cacy had a signi cant effect on disability. In [27], several other studies showed the relationship between self-e cacy and pain intensity and disability. A study on spine disorders, including LBP patients, found that low self-e cacy signi cantly in uenced the disability of LBP and may also affect postoperative outcomes [27]. In a report on LBP subjects in Italia, more than half of the patients had low self-e cacy, and the group had a high level of disability [28]. It has been reported that pain self-e cacy is a more critical factor in the disability than fear of movement [29]. Additionally, among fear of movement, disability and pain intensity, pain self-e cacy serves as a mediator [30]. Self-e cacy was also reported to affect functional aspects such as postural stability and range of motion at the spine; the higher the self-e cacy, the greater the stability and range of motion [12]. Like other previous studies, pain self-e cacy had the most substantial effect on the disability in this study compared to several psychological factors, but especially in the subacute and chronic group. However, even in the acute phase, the impact of self-e cacy on disability is one of the new ndings. To date, most of the studies have focused on chronic LBP. Pain self-e cacy has been reported to be more critical in the recovery of LBP than other psychosocial factors [31], but these participants were only for chronic LBP. Indeed, Ferrari et al. reported that self-e cacy was associated with disability and the degree of pain intensity, indicating that it is an important factor, but in the context of chronic LBP [32]. Thus, the suggestion that pain selfe cacy in uences the disability for overall NSLBP, including the acute phase, is highly informative and suggests that self-e cacy is critical in implementing the management of NSLBP based on the biopsychosocial model, regardless of the duration of the disease.
Pain self-e cacy, pain, catastrophic thinking and fear of movement also in uenced the disability in the subacute and chronic phases. Many psychological factors have been reported to have an impact on chronic NSCLP. Ogunlana et al. showed that those with higher PCS values have a more severe disability in catastrophic pain thinking [33]. Wertli et al. showed in their systematic review that catastrophising was a prognostic factor for LBP [13]. However, it has also been shown that the mechanism is still unclear. Besides catastrophic thoughts of pain, Leeuw et al. reported that there was an association between painrelated fear and functional impairment in CLBP patients [34]. Wertli et al. found that there was an association between fear-avoidance beliefs and LBP disability developed within six months, which is one crucial factor in chronicity [35]. These two factors are also predictors of chronicity of LBP [36]. Based on the results of previous studies and the present study, it is highly likely that these factors impact the disability for CLBP.
On the other hand, among the psychosocial factors, it was only self-e cacy that had an impact on the acute group, while fear of movement and catastrophic thoughts of pain had little impact on the disability.
Pain intensity was indicated to be the most in uential factor in disability for the acute group. This suggests that pain intensity itself, rather than psychosocial factors, has the most substantial impact on LBP in the acute phase. Grotle et al. reported less impact on the acute phase than in the chronic phase regarding fear of movement [37]. In previous studies, severe pain is a factor in the transition from the acute to the chronic phase [4], so how to improve pain in the acute phase will be the key to management of preventing chronicity.

Limitations
While important ndings were obtained, there are some limitations to this study. First, the sample size for the acute phase was relatively small, with only 65 participants. This was due to the stricter de nition of acute LBP as being less than 4 weeks from the onset. This may somewhat reduce the precision of the results of the hierarchical multiple regression analysis. Second, subacute and chronic groups were combined. Comparing the three following groups (acute, subacute and chronic groups) might have provided more meaningful results. However, the de nition of "subacute" is ambiguous, and for this study, we wanted to know the difference in the characteristics between the acute and the following groups, so this grouping was considered appropriate for this study. Despite the several limitations mentioned above, this study includes a large sample size of 255 subjects to examine the impact of functional disability in NSLBP and analyse them by the onset time with multiple psychosocial factors. This was a strong point of this study and should provide signi cant insights into the factors that in uence the disability in NSLBP.

Clinical implication and further research
This study suggests that, regardless of the duration of the disease, pain self-e cacy has an impact and that improving pain self-e cacy from early on in the onset of LBP can improve disability. Improving pain self-e cacy means giving patients the con dence that they can manage their LBP. For this purpose, the management based on biopsychosocial models is necessary, and cognitive behavioural therapy is one of the usual methods, which is effective for CLBP [38]. Furthermore, recently, cognitive functional therapy (CFT) has been proposed [39]. CFT has been shown to improve patients' belief and self-management and to have long-term effects on disability and psychosocial factors, so it can be affected by increasing the patient's own self-e cacy [40]. Combining functional and psychological components, as in CFT, is ideal and will be an essential concept for future management of LBP. To achieve this, we rst need to change the way we think about management, not only from the patient's perspective but also from physical therapists and other healthcare professionals. The reality is that many healthcare providers tend to prefer "biomedical explanations and interventions" [43]. It has also been reported that healthcare professionals have sometimes found di culties to accurately assess the psychosocial factors of patients [45]. Therefore, healthcare professionals need to be trained to implement practices based on the biopsychosocial model to improve the self-e cacy of patients with LBP. Also, the patient may need to be trained to understand them by healthcare professionals as well.

Conclusions
In this study, the explanatory power increased with each additional variable for the disability in the order of essential characteristics, pain intensity and psychosocial factors in the overall, acute and subacute/chronic groups. Depending on the duration of the disease, the factors affecting the disability differed, with pain having more in uence than psychosocial factors in the acute phase and psychosocial factors having more in uence in the chronic phase. Furthermore, the results suggest that pain selfe cacy signi cantly impacts all groups, including the acute phase. The result suggests the need for management based on a biopsychosocial model that enhances self-e cacy at an earlier stage. Future studies on how disability is changed by interventions based on biopsychosocial models that enhance self-e cacy and investigate the effects of the impact in more different groups may help manage NSLBP.

Declarations
Funding: No funding source was provided for this study.
Ethics statement : Written informed consent was obtained from the patients before the study. The study was conducted in compliance with the Declaration of Helsinki and was conducted with the approval of the Ethics Committee of Sapporo Maruyama Orthopaedic Hospital (no. 0039) as well as the Faculty of Health Sciences, Hokkaido University (approved no. 20-58).