Healthy community-dwelling older (≥65 years) women who could independently walk more than 100 m were consecutively enrolled in a single center from July 2018 to December 2018. Participants who had experienced the following were excluded: (1) low back pain with moderate severity (numeric rating scale 5 and over); (2) history of any type of lumbar spine surgery; (3) history of hip fracture surgery and arthroplasty of the hip or knee; (4) disorders of the central nervous system such as stroke, parkinsonism, or spinal cord injury; (5) communication disorder such as severe hearing loss; (6) musculoskeletal conditions affecting physical function such as limb amputation; (7) long-term use of corticosteroids due to inflammatory disease; (8) malignancy requiring treatment within 5 years; and (9) other medical conditions requiring active treatment; additionally, individuals who refused to participate in the study were also excluded.

BIA (InBody 720; InBody, Seoul, Korea) was used to analyze body composition including lean body and fat masses. ASM was calculated as the sum of the lean mass in the bilateral upper and lower extremities⁹⁾ and standardized by dividing by the squared height (ASM/Ht², kg/m²). HGS was measured using a hand-grip dynamometer (T.K.K.5401; Takei Scientific Instruments, Tokyo, Japan),¹⁰⁾ as described previously.¹¹⁾ Briefly, participants were asked to perform the following while sitting in a straight-backed chair with their feet flat on the floor: adduct and neutrally rotate the shoulder, flex the elbow to 90°, and place the forearm in a neutral position with the wrist between 0° and 30° extension and between 0° and 15° ulnar deviation. Participants were instructed to squeeze the handle as hard as possible for 3 seconds, and the maximum contraction force (kg) was recorded. Gait speed was measured using a 6-meter usual gait speed (m/s) as recommended by the Asian Working Group for Sarcopenia. The definition and cutoffs for sarcopenia were also adopted from these guidelines.¹²⁾

Functional examination using a short physical performance battery (SPPB) was derived from three objective physical function tests (i.e., time taken to cover 4 m at a comfortable walking speed, time taken to stand from sitting in a chair five times without stopping, and ability to maintain balance for 10 seconds in three different foot positions at progressively more challenging levels).¹³⁾ A score from 0 to 4 was assigned to the performance of each task, with higher scores indicating better lower body function. The Timed Up and Go test (TUG) has shown excellent test–retest reliability in older adults.¹⁴⁾ The participants were provided with verbal instructions to stand up from an armchair, walk 3 m as fast as possible, turn back at a cone set out by the researchers, walk back, and sit down in the chair. They were allowed to wear their regular footwear and use a walking aid if needed. A stopwatch was started on the word ‘go’ and stopped when the participant was completely seated with their back against the backrest. The time to complete the test was recorded in three consecutive trials using the first trial to familiarize the participants with the test. The best time from the three trials was analyzed.¹⁵⁾ The Oswestry Disability Index (ODI) is one of the most commonly used instruments for measuring disability in spinal disorders. It consists of 10 items that assess the level of pain and interference with several physical activities. We used the Korean version of the ODI.¹⁶⁾ The Back Performance Scale (BPS) consists of five tests: sock test, pick-up test, roll-up test, fingertip-to-floor test, and lift test. The five tests comprising the BPS demonstrate associations with each other and each test contributes to the high internal consistency, implying that the tests share a common characteristic in measuring physical performance.¹⁷⁾ The BPS sum score (0–15) is calculated by adding the individual scores for the five tests.

We measured the isometric back extensor strength using a hand-held dynamometer (PowerTrack II; JTECH Medical, Salt Lake City, UT, USA).¹⁸⁾ Briefly, the participants stood in full extension with their backs to a wall and feet flat on the floor with heels touching the wall. An inelastic belt was looped through the anchor rails and secured firmly 1 cm below the anterior superior iliac spine to restrain movement and maintain participant contact with the wall during the test. The participants were instructed to flex forward approximately 15° at the hips so that the dynamometer could be positioned posterior to the spinous process of the 7th thoracic vertebrae. In this way, counter pressure was provided by the fixed wall behind the participants’ back to avoid tester-induced variations in resistance. Although this method is novel, it showed a strong positive relationship with back extensor strength measured using the gold-standard isokinetic dynamometry and high inter-instrument validity and reliability.¹⁹⁾

For each patient, one lateral radiograph of the whole spine was obtained and digitized. All measurements were performed using imaging software (INFINITT PACS M6; INFINITT Healthcare, Seoul, Korea), as previously described.^20,21) The following spinopelvic radiographic parameters were analyzed: sacral slope (SS), pelvic incidence (PI), pelvic tilt (PT), lumbar lordosis (LL), thoracic kyphosis (TK), and sagittal vertical axis (SVA).

Lumbar spine CT scans (Ingenuity CT; Philips Healthcare, Cleveland, OH, USA) were performed to measure the cross-sectional area (CSA) and mean density (in Hounsfield unit [HU]) of the LPM (multifidus [MF] and erector spinae [ES]).²²⁾ The mean density reflected the degree of intramuscular fat content because the values decreased as the fat content increased. Before each CT scan, a calibration was performed using air as the standard. CT scanning was performed with the participant in the supine position with a 120-kV and 140-mA protocol. Using 1-mm thin-section axial CT scan images, the axial images were reformatted with each lumbar intervertebral disc level (T12/L1-L5/S1) parallel to the adjacent vertebral endplates. These axial images at each intervertebral disc level were reconstructed at 2.5-mm intervals, which included cross-sectional images of LPM. The measurement of ES and MF was performed from the level of L1/L2 to L5/S1 using a specially designed radiological workstation (MEDIP; MEDICALIP, Seoul, Korea) (Fig. 1). The CSA was measured by manually constructing points around the outer margins of the individual muscles using a touchscreen LCD monitor (XPS 15 9570; Dell Inc., Round Rock, TX, USA) and digital touchscreen pen (PN556W Dell Active Pen; Dell Inc.). After the CSA and mean density of paraspinal muscles were separately measured on the bilateral sides, the mean and sum values at all levels were calculated.²³⁾

All demographic and clinical data, including CT scan images, were obtained with the approval of the Institutional Review Board of SMG-SNU Boramae Medical Center (No. 16-2017-45). Written informed consent was obtained from all participants.

The relationships between sarcopenic indices and functional outcomes with SSB parameters were measured by Pearson correlation coefficients. The independent effects on the sum of the total LPM CSA were determined using multivariate regression models adjusted for six key factors: age,⁷⁾ ASM/Ht², HGS, gait speed,¹⁾ back extensor strength,²⁴⁾ and PT angle. ASM/Ht² and HGS were included because they are the basic variables for diagnosing sarcopenia and PT was included in the regression analysis as the most relevant variable of SSB. An adjusted model was developed through backward elimination with a significance level of 0.2 to enter and 0.05 to retain. We also evaluated possible multiple collinearities between covariates by correlation analysis and collinearity statistical tests (tolerance and variance inflation factor tests) during regression analysis. IBM SPSS Statistics version 21.0 for Windows (IBM Corp., Armonk, NY, USA) was used for all analyses. p-values of less than 0.05 were considered statistically significant.