*2.8. Measurement*

#### 2.8.1. Measurements of Spinal Alignment

All subjects underwent a whole-spine lateral radiographic examination in a standing position with the hands on the clavicles [17] for the measurement of the sagittal vertical axis (SVA) as an indicator of total spinal alignment as well as pelvic incidence (PI) and lumbar lordosis (LL). A PI minus LL (PI-LL) mismatch was defined as PI-LL >10◦ [18]. The spinal alignment measurements were performed by 2 board-certified spine surgeons and a trained staff member. The calculated inter-rater reliability scores for each parameter were 0.95 for SVA, 0.80 for PI, and 0.65 for LL [10]. The calculated intra-rater reliability scores for each parameter were 0.91 for SVA, 0.97 for PI, and 0.96 for LL. For validity, our previous study demonstrated that our measurements were comparable to those of previous reports [10].

#### 2.8.2. Evaluation of BMD and SMI

All subjects underwent dual-energy X-ray absorptiometry (GE Prodigy, GE healthcare, Chicago, IL, USA) of the lumbar spine. Osteoporosis was defined as a T-score ≤ −2.5 [19]. Skeletal muscle mass was calculated as the sum of the skeletal muscle mass of the arms and legs, assuming that the mass of lean soft tissue was representative of skeletal muscle mass. SMI was calculated as four-limb lean soft tissue mass in kilograms divided by height in meters squared.

#### 2.8.3. Clinical Evaluation of Subjects

All subjects were evaluated for the degree of LBP by VAS scores (0–100 mm). In this study, subjects with moderate to severe LBP, defined as VAS > 50 mm, were considered as having LBP [20].

#### *2.9. Statistical Methods*

Welch's *t*-test was used to compare the mean values of continuous variables. Fisher's exact test was adopted to evaluate the differences between categorical variables. We employed a logistic regression model with the existence of moderate or severe LBP (i.e., VAS > 50 mm) as a response variable and subject-specific factor candidates as explanatory variables. Univariate and multivariate analyses using the forced entry method included the factors of sex, BMI, SMI, smoking, BMD, osteoporosis (i.e., T-score ≤ −2.5), SVA >50 mm, PI-LL mismatch, and aging as potential confounding factors of LBP according to previous reports [8,9]. Factors with *p* < 0.2 in the univariate analysis were included in the subsequent multivariate analysis with a stepwise algorithm. All statistical analyses were performed using EZR software (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a modified graphical user interface of R commander (The Foundation for Statistical Computing, Vienna, Austria) designed to add statistical functions frequently used in biostatistics. The level of significance was set at *p* < 0.05.
