This study was conducted in compliance with the principles of the Declaration of Helsinki and approved by the Research Ethics Committee of the Graduate School of Biomedical Sciences at Nagasaki University (approval number: 21101406-2). All participants received written and oral information about the study’s purpose and procedures, had the opportunity to ask questions, were given sufficient time to decide on participation, and provided written consent.

This longitudinal cohort study involved a multidimensional assessment within 2 weeks of admission, followed by a pain survey via telephone at 12 weeks post-admission. Participants were patients diagnosed with acute VFs based on MRI or lateral radiographs, and who received conservative treatment at the Department of Orthopaedic Surgery, Nagasaki Memorial Hospital, Japan, between December 2021 and December 2023. Patients with VF classified as A1 to A4 according to the AOSpine Thoracolumbar Injury Classification System [16], and without neurological symptoms such as paralysis, were treated conservatively. Japanese-speaking patients aged ≥65 years who consented to participate were included. Exclusion criteria were injury from high-impact trauma, cognitive impairment (Mini-Mental State Examination [MMSE] score <23 after 2 weeks), comorbidities hindering rehabilitation, acute illness, or new VFs within 12 weeks of admission.

Previous studies have reported that patients with VF who wear rigid or soft orthoses experience greater pain relief than those without orthoses [2]. Therefore, based on the attending physician’s judgment, each patient was prescribed either a rigid or soft orthosis. A previous study reported the achievement of pain relief with up to 6 months of orthosis use [2]. Another study has demonstrated similar analgesic effects with orthosis use for 6 to 12 weeks [17]. Additionally, a prior study showed suppressed vertebral body collapse following 12 weeks of orthosis application [18]. Based on these findings, the duration of orthosis use in this study was set at 12 weeks from hospital admission. Additional medical treatment was prescribed at the discretion of the attending physician. NSAIDs, acetaminophen, and weak opioids, which are effective in managing pain associated with VF [3], were used as part of the pain management strategy.

The rehabilitation program started immediately after admission, following conservative treatment protocols described in previous studies [19]. Initially, patients rested until orthosis completion, with bed rest or assisted toilet use. The orthosis was adjusted according to injury status. Once the orthosis was fitted, patients could mobilize while wearing it. Rehabilitation began upon admission or the next day, starting with limb movements in bed to prevent inactivity. After orthosis completion, patients progressed to standing and walking exercises. The program included individualized muscle strengthening, ADL, and instrumental ADL (IADL) exercises to prepare patients for discharge home.

Data collected included age, sex, height, weight, VF site, number of existing VFs, comorbidities before admission, ADL and IADL before admission, medication use (pain and osteoporosis medications) after admission, and cognitive function at 2 weeks post-admission. Body mass index (BMI) was calculated using height and weight. The Charlson comorbidity index (CCI) was calculated based on pre-admission comorbidities. ADLs before admission were assessed using the Barthel index (BI), IADL using the Tokyo Metropolitan Institute of Gerontology Index of Competence (TMIG-IC), and cognitive function using the MMSE.

The primary outcome was back pain intensity during movement, assessed at admission and 2 and 12 weeks post-admission using a 5-point Verbal Rating Scale (VRS) (0=no pain, 1=mild, 2=moderate, 3=severe, 4=intolerable). Patients reported their most intense pain. Assessments were conducted in person at admission and 2 weeks, and via telephone interview at 12 weeks. Pain sensitivity was assessed using Pressure Pain Threshold (PPT) and Conditioned Pain Modulation (CPM). PPT was measured at the affected site (affected) and remote sites (remote) using a digital force gauge (RZE-100; Aikoh Engineering Co., Osaka, Japan), referring to previous studies [20,21]. The affected PPT was measured bilaterally at the erector spinae muscles 2.5 cm lateral to the spinous process of the injured vertebra, while the remote PPT was measured at the central part of the biceps brachii muscle belly on the non-dominant hand side (Fig. 1). Pressure stimuli were progressively increased at an acceleration rate of 3 N per second, and participants were instructed to say “stop” when the pressure caused sharp pain. Three measurements were taken at 60-second intervals at each site, and the mean value was used. For CPM, ischemic compression of the arm served as the conditioned stimulus, and the affected PPT was used as the test stimulus [22]. PPT was first measured at the affected site before applying the conditioned stimulus. Next, ischemic compression of the arm was applied to elicit diffuse noxious inhibitory control. Finally, the affected PPT was measured after the conditioned stimulus reached a VRS score of 3 (Fig. 2).

Imaging evaluation included MRI at admission and lateral radiography at 2 weeks post-admission. Vertebral compression was assessed on lateral radiographs based on a previous study [19]. The heights of the anterior, middle, and posterior vertebral walls were expressed as percentages relative to the posterior wall height of the superior vertebra, with lower values indicating more severe compression.

Signal changes in the injured vertebrae were classified using MRI T2-weighted images (T2WI) under radiologist (N.A.) supervision, based on previous studies [23]. Sagittal T2WI were analyzed in three slices (midline, bilateral vertebral arch root medial margins) and classified as confined low-signal change, diffuse low-signal change, diffuse high-signal change, or fluid sign (Fig. 3A–D). The diffuse type was defined as signal changes affecting ≥50% of the vertebral body in at least two slices. Paraspinal muscle cross-sectional area (CSA) was evaluated using MRI at admission. The multifidus and erector spinae muscles between Th12–L1 and L4–L5 were collectively analyzed as paraspinal muscles. Measurements included CSA, functional CSA (fCSA, excluding fatty infiltration), and fat infiltration percentage (FI%), as described elsewhere [24]. The mean CSA of the left and right paraspinal muscles at the vertebral body level was used (Fig. 3E). MRI images were converted to 8-bit grayscale (0–255), with areas >120 classified as muscle and areas ≤120 as fat (Fig. 3F). fCSA was calculated by subtracting the fat-infiltrated area from the CSA, and FI% was calculated as the percentage of fat within the CSA. Image processing was conducted using ImageJ ver. 1.47 (National Institutes of Health, Bethesda, MD, USA).

The Pain Catastrophizing Scale-6 (PCS-6) was used to assess catastrophic thinking, the Geriatric Depression Scale-5 (GDS-5) to evaluate depression, and the Tampa Scale for Kinesiophobia-11 (TSK-11) to measure fear of movement. Higher scores on each scale indicated greater symptom severity.

Physical function was assessed using hand grip strength, isometric knee extension strength, the Timed Up and Go test, the 5-Times Sit-to-Stand Test (5SST), and the 6-Minute Walking Distance (6MWD). ADL ability was assessed using the motor-Functional Independence Measure (motor-FIM).

Physical activity was assessed using the Active-style Pro (HJA-750C; Omron, Kyoto, Japan). The device recorded sedentary time (1.0–1.5 metabolic equivalents of task [METs]), low-intensity activity time (1.6–2.9 METs), and moderate-to-vigorous intensity activity time (≥3.0 METs). The analysis used mean values of each activity time during the first week following 2 weeks of hospitalization.

Patients were categorized into two groups based on motion pain severity at 12 weeks post-admission: non-chronic pain (NCP) (VRS 0–1) and CP (VRS 2–4) groups, following our previous study [19]. Group comparisons were conducted using t-tests, Mann-Whitney U test, and chi-square test. Binomial logistic regression with likelihood ratio-based variable reduction was used to assess the predictors of CP at 12 weeks, adjusting for age and BMI. Variables with significant group differences were included as independent variables. Model fit was evaluated using the Hosmer-Lemeshow test, with p>0.05 indicating adequacy. Statistical significance was set at p<0.05. Analyses were performed using IBM SPSS Statistics ver. 27.0 (IBM Corp., Armonk, NY, USA).

During the study period, 172 patients were admitted to the hospital. Of these, 71 patients met the inclusion criteria after excluding patients with cognitive impairment (MMSE ≤23) (n=50), those transferred to another hospital within 2 weeks of admission (n=13), those injured by high-energy trauma (n=10), those whose rehabilitation was hindered by comorbidities or complications (n=23), and those who did not provide consent (n=5). Of the 71 eligible patients, 66 were included in the final analysis after further excluding those who developed acute illnesses (n=2) and those who were re-injured (n=3) within 12 weeks post-admission. Of the analyzed participants, 39 (59.1%) were in the NCP group (VRS <2 at 12 weeks), while 27 (40.9%) were in the CP group (VRS ≥2) (Fig. 4).

There were no significant between-group differences with respect to age at admission, sex, BMI, CCI, BI, and TMIG before injury, number and site of acute or previous VFs, pain medications, osteoporosis medications, MMSE scores, VRS at admission, or the type of orthosis used (Table 1).

The imaging assessment results are summarized in Table 2. There were no significant between-group differences regarding vertebral height loss at 2 weeks post-admission or MRI findings at hospitalization.

Table 3 summarizes the comparison of outcome measures between the two groups at 2 weeks after admission. The CP group had significantly lower remote PPT and low-intensity activity time than the NCP group (p=0.010 and p<0.001, respectively). The CP group also had significantly higher PCS-6 rumination scores and 5SST times than the NCP group (p=0.012 and p=0.020, respectively). There were no significant between-group differences regarding the other outcomes.

Table 4 shows the binomial logistic regression analysis results, identifying two significant early risk factors for transitioning to CP 12 weeks post-hospitalization. The 5SST score was associated with an increased likelihood of developing CP (odds ratio [OR], 1.11; 95% confidence interval [CI], 1.01–1.21; p=0.041), while low-intensity activity time was associated with a decreased likelihood (OR, 0.98; 95% CI, 0.96–0.99; p=0.007). The logistic regression model demonstrated an adequate fit according to the Hosmer-Lemeshow test (χ²=3.05, degrees of freedom=7, p=0.880).