*3.1. Demographics*

A total of 40 patients with a mean follow-up of 5.3 years were included in this study. The mean age was 73.9 years. Figure 1 shows the time course of VAS for low back pain and EQ-5D after OVF. Time had a significant main effect on VAS for low back pain and EQ-5D (*p* < 0.001 and *p* < 0.001, respectively). Comparison of VAS scores for low back pain between consecutive time points showed a significant difference between 0 and 12 weeks (*p* < 0.001), but not between 12 and 48 weeks (*p* = 0.97), or between 48 weeks and final follow-up (*p* = 0.99) (Figure 1). Comparison of EQ-5D scores between consecutive time points showed a significant difference between 0 and 12 weeks (*p* < 0.001), but not between 12 and 48 weeks (*p* = 0.82), or between 48 weeks and final follow-up (*p* = 0.99) (Figure 1).

**Figure 1.** *Cont*.

**Figure 1.** Temporal trends in outcome measures. The visual analog scale (VAS) for low back pain (0–10, with higher scores indicating severe pain) and the European Quality of Life-5 Dimensions Questionnaire (EQ-5D, −0.111 to 1, with higher scores indicating better quality of life). Means with standard deviations at baseline and each follow-up are shown. \* *p* < 0.05, NS not significant.

*3.2. Characteristics of Patients with Residual Low Back Pain at 5 Years after OVF*

We then divided the patients into two groups according to their VAS score at the last follow-up: the residual low back pain group and the no low back pain group. Of the 40 patients analyzed in this study, 10 (25.0%) reported residual low back pain at a mean 5.3 years after OVFs. The baseline characteristics of the patients are shown in Table 1. In the residual low back pain group, the patients were older, and the percentage of patients with a history of pre-existing vertebral fracture was higher. No significant differences were observed in the other background variables between the two groups.

**Table 1.** Baseline characteristics of the patients.


Data are presented as mean ± standard deviation or *n* (%). \* *p* < 0.05. VAS, visual analog scale.

Table 2 shows the differences in PROMs between the groups with VAS scores < 3.5 and ≥3.5 at the final follow-up. VAS scores for low back pain did differ not significantly between the two groups at 0 and 12 weeks, but were significantly worse in the residual low back pain group at 48 weeks and final follow-up (*p* < 0.001 and <0.001, respectively). The EQ-5D score was not significantly different between the two groups at 0 and 12 weeks, but was significantly worse in the residual low back pain group at 48 weeks and final follow-up (*p* < 0.001 and *p* = 0.001, respectively). We then examined the trends in the VAS score for low back pain in the residual low back pain group and no low back pain group. In the no low back pain group, a significant difference was found between VAS scores at 0 and 12 weeks (*p* < 0.001), but no significant difference was noted in the VAS scores between 12 and 48 weeks or between 12 weeks and final follow-up (*p* = 0.41 and 0.30, respectively). In the residual low back pain group, no significant difference was found in the VAS scores between 12 and 48 weeks (*p* = 0.08), but a significant difference was noted in the VAS scores between 0 and 12 weeks and between 12 weeks and final follow-up (*p* < 0.001 and *p* = 0.02, respectively).


**Table 2.** Patient-reported outcome measures.

\* *p* < 0.05. VAS, visual analog scale; EQ-5D, European Quality of Life-5 Dimensions.

Table 3 shows the differences in the radiographic assessment between the groups with VAS scores of <3.5 and ≥3.5. No significant difference was observed in the anterior vertebral body compression percentage between the two groups, although there was a trend toward lower anterior vertebral body compression percentage in the residual low back pain group throughout the period from 0 to 48 weeks (*p* = 0.17, 0.11, and 0.09, respectively).

**Table 3.** Radiographic assessment.


Data are presented as mean ± standard deviation or *n* (%). *p* < 0.05. VAS, visual analog scale.

Lastly, the predictors at 12 weeks after OVFs for residual low back pain at the final follow-up were evaluated using a stepwise multiple logistic regression analysis (Table 4). Based on the univariate analysis, the dependent variable was defined as the presence of residual low back pain at the final follow-up, and the independent variables were age, previous vertebral fracture, and EQ-5D score at 12 weeks after OVF. As a result, the independent risk factors at 12 weeks were identified as age (OR = 1.19; 95% CI, 1.01–1.46; *p* = 0.04) and previous vertebral fracture (OR = 6.28; 95% CI, 1.24–39.83; *p* = 0.03).


**Table 4.** Multiple logistic regression analysis: independent risk factors of residual low back pain (VAS ≥ 3.5 at final follow-up).

\* *p* < 0.05. VAS, visual analog scale.

## **4. Discussion**

This study investigated the course of acute vertebral fracture in terms of pain and QOL. When comparing VAS for low back pain and EQ-5D scores between consecutive time points, a significant difference was observed between 0 and 12 weeks, but not between 12 and 48 weeks or between 48 weeks and final follow-up. Twenty-five percent of patients had residual low back pain at the final follow-up. The patients with residual low back pain after OVF had a higher percentage of pre-existing vertebral fractures and were older than those who did not have residual low back pain. A stepwise logistic regression analysis identified age and previous vertebral fracture as predictors of residual low back pain at the final follow-up.

This study showed that when comparing VAS scores for low back pain and EQ-5D scores between consecutive time points, a significant difference was found between 0 and 12 weeks, but not between 12 and 48 weeks or between 48 weeks and final follow-up. This result is consistent with previous reports that pain improved by 3 months after the fracture and did not change significantly until 12 months thereafter [12]. Collectively, these results sugges<sup>t</sup> that if severe pain remains after the acute phase, it might be unlikely that the pain will improve spontaneously.

Among patients with acute vertebral fractures, 25% had mild or severe low back pain for an average of 5.3 years after injury based on VAS for low back pain. Patients with mild or severe low back pain were older and had a higher percentage of patients with pre-existing vertebral fractures than those with moderate or no pain. The results were partially consistent with a previous report stating that chronic pain after acute spine fractures was only maintained in patients with multiple compression fractures, reduced height, and low bone density [13].

In this study, the VAS score for low back pain was not significantly different between the residual low back pain group and no low back pain group at 0 and 12 weeks, but was significantly worse in the residual low back pain group at 48 weeks and final follow-up. Regarding the transition of pain within the group, although not significant, the low back pain tended to improve after 12 weeks in the no low back pain group. By contrast, back pain deteriorated after 12 weeks in the residual low back pain group. In a randomized controlled trial comparing vertebroplasty and conservative treatment for patients with vertebral fractures who reported severe pain for more than 3 months, vertebroplasty was associated with better pain relief and improved functional outcomes at 1 year compared with conservative treatment [14]. Therefore, taking into account the improvement of pain in patients who report severe low back pain 3 months after a vertebral fracture, vertebroplasty should be considered rather than conservative treatment. This treatment strategy should be tested in the future.

Since OVF-induced pain significantly improves by 12 weeks, we decided to investigate predictors for residual low back pain at 12 weeks after OVF. A stepwise logistic regression analysis identified age and previous vertebral fracture as predictors for residual low back pain at a mean of 5.3 years after OVFs. Therefore, when a new vertebral fracture occurs in an older patient with a pre-existing vertebral fracture, the patient is likely to have residual low back pain in the future. Furthermore, risk factors for OVFs include older age, low bone mineral density, and pre-existing vertebral fractures [15]. Therefore, elderly patients with new OVF and pre-existing vertebral fractures are at risk of further subsequent

fractures. In this study, we do not know whether subsequent vertebral fractures occurred in this group of patients after 48 weeks, because imaging evaluation was not performed in the final follow-up. According to a post-hoc analysis of the original prospective study, patients with subsequent vertebral fractures at Week 48 had significantly more severe low back pain than those without subsequent fractures at Week 48 [16]. Therefore, in this study, it cannot be ruled out that the presence of subsequent vertebral fractures at the time of the final follow-up may be associated with residual low back pain. However, if a new OVF occurs in an older patient with a pre-existing vertebral fracture, it may be desirable to provide intensive osteoporosis treatment to prevent subsequent fractures. Further research is needed to determine which osteoporosis drugs are the most effective in reducing subsequent fractures in elderly patients with pre-existing vertebral fractures.

This study had some limitations. First, several patients were excluded after enrollment which might have led to a slight decrease in the sample size. Accordingly, attrition bias may limit the internal validity of this study. Second, we did not investigate the bone mineral density in this study. Although it is undeniable that the severity of osteoporosis may affect back pain, a decrease in bone mineral density does not necessarily lead to an increase in low back pain. In fact, the authors of several studies concluded that there is no evidence supporting a relationship between low back pain and bone mineral density [17,18]. Third, given the small percentage of patients who continued to attend the hospital, no radiographic evaluation was performed at the last follow-up. This prevented us from assessing the relationship between residual low back pain and non-union, subsequent fractures, and spinal alignment at the final follow-up. Lastly, the results of the multiple logistic regression analysis showed that there were two independent variables. Accordingly, the event per variable (EPV) was five in this model. However, the rule of thumb of 10 or more EPV in logistic models is not a well-defined bright line [19]. A simulation study showed that statistical problems are uncommon with 5–9 EPV, and still observed with 10–16 EPV [19]. Further studies are required to address these limitations and to validate our findings.
