DOI: https://doi.org/10.21203/rs.3.rs-1827404/v1
Objective To characterize the distribution of paraspinal muscle in degenerative kyphosis (DK) and identify the impact of body mass index (BMI) and paraspinal muscle on sagittal alignment and balance status.
Summary of Background Data The relationship between sagittal alignment, BMI and paraspinal muscle in DK remains unidentified and the available methods for paraspinal muscle measurement are complex. It is assumed that the relationship between muscle and kyphotic severity can be characterized with a simplified procedure.
Methods Degenerative thoracolumbar kyphosis (TLK) or loss of lumbar lordosis (LL) was defined as DK group (81 cases), and 60 volunteers composed control group well-matched to DK group in terms of age and gender. DK was evaluated by TLK and LL; paraspinal muscle was depicted by lumbar crossing indentation value (LCIV) on T2-MRI; sagittal balance for DK was evaluated by pelvic incidence (PI) and pelvic tilt (PT). For DK patients, normal-weight, overweight and obesity were defined based on BMI. For balance status, lumbo-pelvic balance (SB1) and imbalance (SIB1) depicted by PI-LL and pelvic balance (SB2) and retroversion (SIB2) expressed by PT were both defined.
Results The mean LCIV (mLCIV) in DK group was smaller, and LCIV increased from T12-L1 to L4-L5 in DK group (P <0.01). In DK group, mLCIV in normal-weight subgroup was smaller than that in obesity subgroup (P <0.05). The mLCIV in DTLK only subgroup was larger compared to other 2 subgroups (P <0.05). In DK group, mLCIV was positively correlated with BMI and LL, and BMI and LL were influencing factors of mLCIV in DK group. The mLCIV was larger in SB1 group, while TLK and LCIV were predictors for PI-LL in DK patients. The mLCIV was also larger in SB2 group and LL was an independent predictor for PT in DK patients.
Conclusions: LCIV is smaller in DK group while larger in sagittal balance status. LCIV, but not BMI, is correlated with sagittal alignment. This concept simplifies the primay diagnosis of DK group and plays a substantial role in maintaining the sagittal balance.
Degenerative kyphosis (DK) is a common adult degenerative deformity caused by degeneration of the disc and facet joint [1, 2]. The typical changes in imaging were reported as sagittal imbalance or local deformity manifested by decreased lumbar lordosis (LL) or increased thoracolumbar kyphosis (TLK). Age, gender, and obesity have been reported to correlate with an increased incidence of spinal degeneration, which are associated with various musculoskeletal disorders, including impairment of spine [3, 4], while the relationship between obesity and DK still remains elusive. A previous study [5] demonstrated that obesity is not only a risk factor which accelerates disc degeneration, but also may promote the development of sagittal plane deformity. Recently, studies [6, 7] indicated that the progression of spinal deformity was not only related to body mass index (BMI), but also related to paraspinal muscles.
Paraspinal muscles play a significant role in stabilizing spine and maintaining sagittal alignment and balance, which has a close relationship with degeneration of spine[8, 9], which was mainly composed of psoas major, erector spinae and multifidius muscles. Studies [8–10] reported that the lower strength of a paraspinal muscle could increase the incidence of lower back pain and degenerative spinal deformity. Muscle cross-sectional area (CSA) and fat infiltration are frequently used to quantify paravertebral muscles for evaluating the relationship between sagittal parameters and muscle content [11–13]. However, this method was slightly complicated, which was inconvenient for rapid calculation of muscle content. Takayama et al. [14] proposed the concept of lumbar crossing indentation value (LCIV) to simplify measurement processes, and indicated that LCIV is highly correlated with CSA and fat infiltration, especially for patients who were aged > 50 years old. However, the relationship between sagittal parameters and LCIV in cases with DK remains to be further clarified.
In addition, Schwab et al. determined the threshold of pelvic incidence (PI) minus LL and pelvic tilt (PT) for evaluating sagittal balance for adult deformity [15], while the impact of LCIV was not determined. Based on these viewpoint, we hypothesized that the relationship between paraspinal muscle and kyphotic severity can be characterized in DK population and the content of paraspinal muscle may be measured with a simplified procedure according to Takayama et al. Therefore, the present study aimed to assess the effects of BMI and paraspinal muscle on sagittal alignment in patients with DK.
A single-center retrospective study was performed from January 2016 to December 2019. Patients who were diagnosed as degenerative TLK (DTLK) or loss of LL (LLL) were as allocated to DK group. Subjects who needed lumbar surgery without degenerative spinal deformity were classified into control group. The study was approved by the Ethics Committee of our hospital (ethics approval NO. 2018PHC076), and all participants signed the written informed consent form.
According to similar studies[1, 5, 6], the effect size |ρ| of all parameters ranged from 0.28 to 0.33 among patients with DS. We defined the α error possibility was 0.05 and the power (1- β error possibility) was 0.90, together with the estimation of exclusion rate was 10–20%, so the minimal sample of DK group was 90. Then the sample of control group was determined with 1:1 to 1:2 matching to DK group with propensity score matching methods. Therefore, a total of 90 patients in DK group and 67 cases with in control group were primarily screened in the protocol.
The inclusion criteria were as follows: (1) patients were diagnosed as DK group without coronal deformity; (2) control group involved healthy subjects without spinal deformity; (3) intact preoperative X-ray of spine and lumbar magnetic resonance imaging (MRI); (4) participants’ age > 50 years old and (5) all cases were in Grade N (< 4 cm) or P (4-9.5cm) of the global trunk balance by SRS-Schwab classification, which was evaluated by sagittal vertical axis (SVA) [15]. Exclusion criteria were (1) patients with coronal deformity; (2) with blurred radiography; (3) with adolescent idiopathic, congenital or traumatic deformity; (4) patients with spinal tumors, infections or spondylolisthesis; (5) patients who previously received lumbar spine surgery, (6) cases with SVA ˃9.5cm and (7) the patients were diagnosed with systemic disorders such as frailty syndrome[16]. Scoliosis was defined as a coronal Cobb angle of more than 10°. Coronal balance was defined as the distance between the central sacral line and the mid body of C7 less than 30 mm.
Patients underwent a 1.5 T Gyroscan ACS-NT scanner (Philips Medical Systems, Best, The Netherlands; matrix size, 512 × 512; field of view, 180 mm × 180 mm) in a supine position. A complete T2-weighted image (repetition time, 3,200 ms; echo time, 102 ms) from T12-L1 to L4-L5 was obtained, and then sagittal and axial images were uploaded onto the PACS workstation.
BMI was calculated as weight in kilograms divided by height in square meters. Patients were divided into normal weight (N), overweight (OW), and obesity (OB) according to BMI, where 18.5 ≤ N < 25.0 (kg/m2), 25.0 ≤ OW < 30.0 (kg/m2) and OB ≥ 30.0 (kg/m2).
Besides, TLK was measured from the T10 superior end plate to L2 inferior end plate and DTLK is TLK ≥ 15° caused by degeneration; LL was measured from the L1 superior end plate to S1 superior end plate and LLL was defined as LL < 25° [14]. According to Schwab et al.’ research, PI and PT are key parameters for quantifying sagittal balance. PI was defined as the angle between the line perpendicular to the sacral plate and the line connecting the midpoint of the sacral plate to the bicoxofemoral axis. PT was defined as the angle between a vertical line originated from the center of the bicoxofemoral axis and a line between the same point and the middle of the superior end plate of S1 (Fig. 1).
Segments were determined on the sagittal T2-weighted image and axial image for LCIV measurement. LCIV was vertical distance between the bulge of the muscle and the apex of spinous process (Fig. 1). The LCIV of each level from T12-L1 to L4-L5 was measured separately, while L5-S1 was ruled out due to axial-cutting gantry by the iliac crest and muscular anatomy different from upper levels.
All parameters were independently measured by two reviewers. Intra-observer reproducibility and inter-observer reliability of these measurements were both evaluated by the intraclass correlation coefficient (ICC). There, the ICC with 95% CI was identified, comparing the mean of all measurements from two observers. ICC <|0.40| indicated poor results; |0.40| to |0.75| was fair to good, and |0.75| to |1.00| was excellent reliability.
A subgroup analysis of gender and BMI (N, OW, and OB) was conducted. The LCIV was also compared among three subgroups (DTLK only, LLL only, and DTLK + LLL) in DK group.
Another subgroup analysis on LCIV was performed by PI and PT. DK group was divided into sagittal balance subgroup (SB-1 subgroup, PI-LL ≤ 10°, Grade A by SRS-Schwab classification) and sagittal imbalance subgroup (SIB-1 subgroup, PI-LL > 10°, Grade B and C ), as well as SB-2 subgroup (PT ≤ 20°, Grade L by SRS-Schwab classification) and SIB-2 subgroup (PT > 20°, Grade M and H).
The dichotomous between DK and control group was analyzed by χ2 test. Independent samples t-test was employed to compare the measured data between two groups, while analysis of variance (ANOVA) was used to compare the measured data among multiple groups. Pearson correlation analysis was utilized to assess relationship among TLK, LL, BMI, and LCIV. Multivariate regression analysis was applied to determine factors influencing LCIV, PI, and PT. SPSS 22.0 software (IBM, Armonk, NY, USA) was used to perform statistical analysis, and P < 0.05 was considered statistically significant.
Herein, 81 cases in DK group and 60 in control group were eventually enrolled, with 9 and 7 cases excluded, respectively. There was no significant difference in gender, age, and BMI between these two groups (Table 1). Intra-observer reproducibility and inter-observer reliability by ICC for all sagittal parameters and LCIV showed good to excellent agreement. (Table 2)
DK group | Contrl group | P | |
---|---|---|---|
Gender | 0.835 | ||
Male | 27 | 19 | |
Female | 54 | 41 | |
Age, y | 66.1 ± 8.9 | 66.3 ± 9.7 | 0.891 |
BMI, kg/m2 | 26.0 ± 3.4 | 26.2 ± 3.9 | 0.741 |
18.5–24.9 | 34 | 25 | |
25-29.9 | 32 | 26 | |
˃30 | 15 | 9 | |
Exclusion rate (%) | 10.0 | 11.6 | 0.591 |
TLK, ° | 22.0 ± 14.4 | 7.5 ± 4.3 | < 0.001 |
LL, ° | 34.4 ± 20.7 | 42.4 ± 6.6 | 0.002 |
Fotenote: DK: degenerative kyphosis; BMI: body mass index; TLK: thoracolumbar kyphosis; LL: lumbar lordosis |
Intra-observer reproducibility | Inter-observer reliability | |
---|---|---|
LCIV T12-L1, mm | 0.83 | 0.76 |
LCIV L1-L2, mm | 0.72 | 0.70 |
LCIV L2-L3, mm | 0.89 | 0.81 |
LCIV L3-L4, mm | 0.92 | 0.77 |
LCIV L4-L5, mm | 0.88 | 0.81 |
TLK, ° | 0.86 | 0.76 |
LL, ° | 0.81 | 0.75 |
PI, ° | 0.89 | 0.79 |
PT, ° | 0.78 | 0.78 |
Fotenote: ICC: intraclass correlation coefficient; LCIV: lumbar cross-sectional indentation value; TLK: thoracolumbar kyphosis; LL: lumbar lordosis; PI: pelvic incidence; PT: pelvic tilt |
The mean LCIV (mLCIV) in DK group was lower than that in control group (P < 0.001), and the same trend was noted from T12-L1 to L3-L4 in DK group (P < 0.05) except for L4-L5 (P = 0.190) (Table 3). The results showed that LCIV gradually increased from T12-L1 to L4-L5 in DK group (P < 0.001). In DK group, LCIV from T12-L1 to L2-L3 was lower than that of L3-L4 (P < 0.05) and L4-L5 (P < 0.001).
LCIV, mm | DK group | Control group | P |
---|---|---|---|
T12-L1 | 4.4 ± 5.3 | 9.7 ± 4.2 | < 0.001 |
L1-L2 | 5.0 ± 5.8 | 10.0 ± 4.2 | < 0.001 |
L2-L3 | 6.0 ± 5.9 | 10.1 ± 4.2 | < 0.001 |
L3-L4 | 8.1 ± 6.7 | 10.4 ± 4.2 | 0.037 |
L4-L5 | 12.4 ± 6.7 | 13.8 ± 4.1 | 0.190 |
mLCIV | 7.2 ± 5.4 | 10.8 ± 3.5 | < 0.001 |
Fotenote: LCIV: lumbar cross-sectional indentation value; DK: degenerative kyphosis |
By gender, mLCIV in female group (10.0 ± 3.0 mm) was lower than that in male group (12.8 ± 4.1 mm, P = 0.007) non-deformity subjects, while there were no statistical gender difference in DK group. In control group, there was a significant difference in mLCIV among N, OW, and OB subgroups (P = 0.030), and the difference was mainly related to T12-L1 and L1-L2 between N and OB subgroups (Figs. 2).
In control group, mLCIV was positively correlated with BMI and negatively correlated with TLK. However, in DK group, mLCIV was positively correlated with BMI and LL (P < 0.05). After integrating two groups, it was unveiled that mLCIV was positively correlated with BMI and LL, while negatively correlated with TLK (P < 0.01) (Table 4). BMI and LL were found as factors influencing mLCIV in DK group, while BMI and TLK were identified as factors influencing mLCIV in control group. For all patients, BMI, TLK, and LL were factors influencing mLCIV according to the following formula: LCIV = 0.3 × BMI − 0.4 × TLK + 0.6 × LL − 6.6 (Table 5).
BMI | TLK | LL | |||||
r | P | r | P | r | P | ||
DK group | TLK | 0.116 | 0.304 | ||||
LL | 0.039 | 0.727 | 0.315 | 0.004 | |||
mLCIV | 0.290 | 0.016 | -0.062 | 0.612 | 0.603 | < 0.001 | |
Control group | TLK | -0.034 | 0.794 | ||||
LL | 0.196 | 0.133 | 0.034 | 0.794 | |||
mLCIV | 0.412 | 0.002 | -0.304 | 0.028 | 0.093 | 0.510 | |
All cases | TLK | 0.046 | 0.591 | ||||
LL | 0.069 | 0.420 | 0.114 | 0.179 | |||
mLCIV | 0.297 | 0.001 | -0.271 | 0.003 | 0.544 | 0.000 | |
Fotenote: BMI: body mass index; TLK: thoracolumbar kyphosis; LL: lumbar lordosis; LCIV: lumbar cross-sectional indentation value; DK: degenerative kyphosis; r: correlation coefficient |
Coefficient | Unstandardized | Standardized | T | P | ||
B | SE | Beta | ||||
DK group | (constant) | -10.403 | 3.899 | -2.668 | 0.010 | |
BMI | 0.459 | 0.144 | 0.292 | 3.192 | 0.002 | |
LL | 0.156 | 0.024 | 0.603 | 6.600 | < 0.001 | |
Control group | (constant) | 3.551 | 3.012 | 1.179 | 0.244 | |
BMI | 0.348 | 0.108 | 0.398 | 3.218 | 0.002 | |
TLK | -0.243 | 0.105 | -0.285 | -2.306 | 0.025 | |
All cases | (constant) | -6.589 | 2.434 | -2.707 | 0.008 | |
BMI | 0.413 | 0.088 | 0.305 | 4.683 | < 0.001 | |
TLK | -0.133 | 0.024 | -0.362 | -5.518 | < 0.001 | |
LL | 0.174 | 0.020 | 0.581 | 8.874 | < 0.001 | |
Footnote: LCIV: lumbar cross-sectional indentation value; DK: degenerative kyphosis; BMI: body mass index; LL: lumbar lordosis; TLK: thoracolumbar kyphosis; SE: standard error |
There were 31 cases in SB-1 subgroup and 50 in SIB-1 subgroup with no significant difference in gender (P = 0.746), age (P = 0.321), and BMI (P = 0.149). TLK was found comparable between groups (25.0 ± 13.2° vs. 20.2 ± 14.9°, P = 0.141), while LL was higher in SB-1 subgroup (47.2 ± 12.8° vs. 26.5 ± 20.8°, P < 0.001).
There were significant differences in the variation from L2-L3 to L4-L5 between the two subgroups (P < 0.05), and mLCIV was greater in SB-1 subgroup (9.4 ± 5.2 mm vs. 5.8 ± 5.0 mm, P = 0.005) (Table 6). PI was negatively correlated with TLK (P = 0.050) and mLCIV (P < 0.001). TLK (β=-0.277, P = 0.006) and LCIV (β=-0.553, P < 0.001) were found as predictors for PI in DK group (Fig. 3).
LCIV, mm | PI-LL | PT | ||
---|---|---|---|---|
SB1 | SIB1 | SB2 | SIB2 | |
T12-L1 | 5.2 ± 6.1 | 3.9 ± 4.6 | 5.2 ± 5.3 | 3.5 ± 5.1 |
L1-L2 | 6.6 ± 6.1 | 4.0 ± 5.5 | 6.0 ± 6.0 | 3.8 ± 5.4 |
L2-L3 | 8.2 ± 6.1* | 4.6 ± 5.4 | 7.0 ± 5.8 | 4.9 ± 5.9 |
L3-L4 | 11.6 ± 6.4** | 5.9 ± 6.0 | 10.0 ± 6.2* | 6.0 ± 6.8 |
L4-L5 | 15.5 ± 5.4** | 10.5 ± 6.8 | 14.3 ± 5.9* | 10.3 ± 7.0 |
mLCIV | 9.4 ± 5.2** | 5.8 ± 5.0 | 8.5 ± 5.0* | 5.7 ± 5.4 |
Footnote: LCIV: lumbar cross-sectional indentation value; PI: pelvic incidence; LL: lumbar lordosis; PT: pelvic tilt | ||||
*: Signficant between variables in the same group (P < 0.05); **: Signficant between variables in the same group (P < 0.01) |
There were 44 patients in SB-2 subgroup and 37 in SIB-2 subgroup, with no significant difference in gender (P = 0.270), age (P = 0.113), and BMI (P = 0.183) between the two subgroups. It was uncovered that LL was higher in SB-2 subgroup than that in SIB-2 subgroup (39.2 ± 19.8° vs. 28.8 ± 20.5°, P = 0.022).
There were significant differences in L3-L4 (P = 0.013) and L4-L5 (P = 0.011) between these two subgroups, and mLCIV was greater in SB-2 subgroup than that in SIB-2 subgroup (8.5 ± 5.0 mm vs. 5.7 ± 5.4 mm, P = 0.028) (Table 6). PT was negatively correlated with LL (r=-0.351, P = 0.001) and mLCIV (r=-0.271, P = 0.024). Multivariate regression analysis showed that LL (β=-0.397, P = 0.006) was an independent predictor for PT in DK group (Fig. 3).
The spinal stability system consists of three subsystems: passive spinal column, active spinal muscles, and neural control unit [1, 8]. The passive musculoskeletal subsystem includes vertebrae, facet articulations, intervertebral discs, spinal ligaments, and joint capsules, as well as the passive mechanical properties of the muscles. The active musculoskeletal subsystem consists of the muscles and tendons surrounding the spinal column. The neural and feedback subsystem consists of the various force and motion transducers, located in ligaments, tendons, and muscles, and the neural control center [2, 8, 9, 11]. DK is a manifestation of sagittal malformation characterized by a decrease of LL or an increase of TLK. With the growth of the aging population, numerous scholars have concentrated on DK, in which the performance of spinal stability system mainly depends on the changes of body shape, including BMI and paraspinal muscles [6, 10].
A variety of methods are available to assess paraspinal muscles, while those methods are complex and time-consuming [17, 18]. Takayama et al. [14] proposed the concept of LCIV, which made the measurement of paraspinal muscles volume more reliable, simple, and efficient. They found a positive correlation between LCIV and CSA, while there was a negative correlation between LIV and age. They pointed out that the degeneration of paraspinal muscles may lead to the loss of LL, and the decreased tension of posterior extensor muscles plays a pivotal role in LLL [19, 20, 20].
There have been few studies on "body shape" and sagittal imbalance. Obesity may not only be a risk factor related to the natural degradation of spine, but also play a role in promoting occurrence of DLS [3]. The anterior vertebral bodies provide the primary load-bearing capacity of the thoracic spine [211]. Wang et al. [2] pointed out that BMI was closely associated with the formation and progression of DS. However, a number of scholars demonstrated that the loss of sagittal sequence was associated with absolute muscle loss, while higher BMI indicates higher degree of "obesity" with fat infiltration [1,222]. Compared with healthy adults, obese patients exhibited weaker paraspinal muscles and difficulty in upright posture, which would accelerate the degeneration of intervertebral discs and facet joints, resulting in spinal deformities. Kim et al. [5] confirmed that there was no significant correlation between BMI and sagittal imbalance. In the present study, the TLK and LL in both groups were not affected by BMI, which was in accordance with Kim et al.’ findings. A previous study demonstrated that multifidus and erector spinae have different roles in affecting the spinal-pelvic alignment and maintaining the sagittal balance. Multifidus at the lower lumbar spine level is critical for maintaining the curvature of the lumbar spine, whereas erector spinae at the lower lumbar level mainly affects the pelvic parameters [8]. Studies confirmed the correlation between sagittal alignment and cross-sectional area of paraspinal muscles in patients with lumbar spinal stenosis [9, 13]. Mannion et al. [233] believed that scoliosis is thought to progress during growth because spinal deformity produces asymmetrical spinal loading, generating asymmetrical growth, etc., in a vicious cycle.
In the present research, it was found that the LCIV gradually increased from top to bottom in DK group and LCIV in DK group was remarkably lower than that in control group except for L4-5. On the one hand, the apical vertebrae was mainly distributed on thoracolumbar and upper lumbar spine with DK, and there was much less impact on the bottom of the lumbar spine contrasted with the control group. On the other hand, the iliac crest and sacrum provided the attachment point of sacrospinous muscle and further thichened the paravertebral muscles in this area, regardless for DK and control group [14,244]. Additionally, it was unveiled that there was a significant correlation between LCIV and LL. It was shown that the loss of LL or even kyphosis may lead to a significant decrease in the content and strength of paraspinal muscles. Takemitsu et al. [14] found that the strength of posterior extensor muscle was significantly lower than that of flexor muscles in DK patients. Hongo et al. [255] indicated that decreased strength of paraspinal muscles was found to be associated with loss of LL and kyphosis in the elderly, and the LL could be recovered by exercise.
Although DK has been scarcely reported in Western countries, it is common in Asian countries because of different lifestyles and working posture [266]. Full of consensus, the SRS-Schwab classification for evaluating the sagittal deformity was then proposed [277] Hence, we, in the present study, divided patients with DK into balance group and imbalance group according to PI-LL and PT parameters. The results showed that the content of paraspinal muscles in the balance group was significantly higher than that in the imbalance group. Correlation analysis revealed that there was a significant correlation between paraspinal muscle content and PI-LL, PT, which further confirmed the close relationship between sagittal balance and paraspinal muscle content in patients with DK. Therefore, for the DK patients requiring to undergo deformity correction, it is highly essential to enhance rehabilitation training program, in order to simultaneously promote the recovery [28,299]. Besides, the relationship among TLK, BMI, and LCIV was assessed in the present study. For normal patients, muscle content can be roughly estimated by measuring height and weight. However, for patients with DK, the degree of kyphosis can be predicted by BMI and MRI findings. Therefore, we assessed the muscle mass by the degree of kyphosis and BMI, which was found reliable for surgeons to evaluate patients with DK.
There are some limitations in the current study. Firstly, LCIV was not as accurate as CSA. For muscle tissues infiltrated by massive fat, it was difficult to identify the muscle apex, which might cause bias. Secondly, since patients with DK were the main target in the current study, it remains unclear whether our findings can be helpful for other types of deformity, such as scoliosis. Thirdly, there was an obvious correlation between LCIV and sagittal alignment, while the causal relationship could not be determined. Finally, there was no simplified parameter, such as LCIV representing CSA, analogous to fat infiltration. Wecould propose the quantitative metric, combined with LCIV, to characterized the paraspinal muscle more accurately. .
LCIV gradually increased from upper to lower lumbar spine in DK patients, and it was remarkably lower in DK group than that in control group. For DK patients, LCIV was greater in balance group than that in imbalance group. With the increase of BMI, LCIV may increase in both groups, while BMI has neither obvious effect on sagittal balance nor correlation with LL and TLK. LCIV is positively correlated with LL, PI, and PT, indicating that LCIV plays a substantial role in maintaining the sagittal balance.
Ethics approval and consent to participate
This retrospective study was approved by the Institutional Review Board (IRB) of 301and Peking University People's hospitals. All patients involved in the study consent to participate in the study. And the written consent has been obtained from all the patients.
Consent for publication
All individual person’s data consent to publish.
Availability of data and materials
Please contact author for data requests.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions:
Conceptualization: Haiying Liu, Yan Liang; Data Curation: Haiying Liu, Shuai Xu, Chen Guo; Formal Analysis: Yan Liang, Shuai Xu; Investigation: Yan Liang, Chen Guo; Methodology: Shuai Xu; Yan Liang, Keya Mao; Project Administration: Haiying Liu; Resources: Shuai Xu; Software: Shuai Xu, Yan Liang; Validation: Yan Liang; Visualization: Haiying Liu; Writing & Editing: Haiying Liu, Shuai Xu, Yan Liang.
Acknowledgements:
We also acknowledge Yonggang Zhang who contributed towards the study by making substantial contributions to the design and the acquisition of data.
Funding:
Procurement of Government of National Health Commission of China [grant number 2127000218]
There is no conflict of interest.