The effects of pressure biofeedback on hip and trunk muscle activity and lumbopelvic alignment during one-leg standing

Abstract

BACKGROUND:

During one-leg standing (OLS), optimum activity of the gluteus medius (Gmed), multifidus (MF), and quadratus lumborum (QL) muscles relies upon maintaining neutral lumbopelvic alignment. However, no studies have examined how using pressure biofeedback during OLS affects the activity of these muscles and the concomitant alignment of the pelvis and trunk.

OBJECTIVES:

The purpose of this study was to investigate the effect of pressure biofeedback on the activity of the Gmed, MF, and QL and the femoropelvic and trunk lean angles during OLS.

METHODS:

Twenty-four healthy males performed OLS with (PB $+$ ) and without (PB $-$ ) pressure biofeedback. For all OLS conditions, a pressure sensor was placed between the lateral surface of the humerus on the non-supporting side and the wall. Under the PB $-$ condition, participants performed preferred OLS while the examiner measured the maximum pressure caused by trunk lean. Under the PB $+$ condition, participants were asked to perform at a threshold of 50% of the maximal pressure (PB $+$ 1 condition) and with minimal change in pressure (PB $+$ 2 condition). Muscle activities of MF, QL, and Gmed as well as the femoropelvic and trunk lean angles were measured under various OLS conditions.

RESULTS:

The activity of the Gmed, MF, and QL was greater under both PB $+$ conditions than under the PB $-$ condition ( $p<$ 0.05). Also, both PB $+$ conditions resulted in a greater femoropelvic angle and reduced trunk lean angle. There were no significant differences in muscle activity, femoropelvic angle, or trunk lean angle between PB $+$ 1 and PB $+$ 2 ( $p>$ 0.05).

CONCLUSIONS:

These results suggest that pressure biofeedback is a useful modality for increasing the activity of the Gmed and trunk muscles, especially the MF muscle on the non-supporting leg side, and for preventing compensatory movements such as trunk deviation and pelvic lateral deviation during OLS.

1. Introduction

One-leg standing (OLS) requires balancing while standing on one leg [1]. It is not only used to assess postural stability and movement bias [2, 3], but also as a gluteus medius (Gmed) strengthening exercise [4]. The Gmed stabilizes the pelvis against the fixed femur during OLS [5, 6], producing a force equal to double the body weight to maintain the correct pelvic and trunk posture [6], and produces the greatest moment for its muscle length to meet functional needs such as OLS [6]. Lumbopelvic neutral alignment is required during OLS to increase Gmed activity because Gmed activity is affected by lumbopelvic alignment [7]. Lumbopelvic neutral alignment is maintained by the co-contraction of local and global muscles, including the multifidus (MF), which helps maintain an upright posture and regulates load transfer through the pelvis prior to movement [8, 9]. In addition, the quadratus lumborum (QL) plays an important role as a lumbar stabilizer [10, 11], and the bilateral activity of both QL muscles may control load transfer by ensuring pelvic stability.

During OLS, lateral pelvic deviations occur to maintain the center of mass on the supporting leg [12]. Because the pelvis and lumbar spine are connected by a kinematic link [6, 13], trunk deviation to the supported or unsupported side may occur during OLS.

A previous study reported that the activity of the Gmed was significantly lower when the pelvis was dropped or the trunk was deviated toward the support side [14, 15]. Elevation of the pelvis during OLS increases activation of the ipsilateral QL muscle. In addition, the MF maintains a neutral position to improve lumbopelvic stability [16], and its activation may be reduced if the trunk or pelvic alignment changes. Repeated performance of OLS without control of lumbopelvic alignment is related to musculoskeletal disorders in the low back or pelvis [9, 17]. Therefore, to increase the activity of the Gmed, MF, and QL during OLS, a method that can minimize lateral deviation of the pelvis is needed.

Biofeedback has been used as an effective method to induce correct postural alignment and improve muscle activity by providing real-time information on pressure at the contact site during exercise [18, 19]. Cynn et al. [18] reported that pressure biofeedback between the ground and lumbar side during side-lying hip abduction significantly decreased pelvic lateral tilt angle and muscle activity of the QL and significantly increased the activity of the Gmed. In another report, pressure biofeedback between the ground and abdominals during prone hip extension decreased the activity of the erector muscles and the angle of pelvic anterior tilt while simultaneously increasing gluteus maximus and hamstring activity [19].

Based on these previous findings, we believe that pressure biofeedback can be used for increasing the activity of the Gmed and correct the alignment of the pelvis and trunk during OLS. However, no studies have investigated how pressure-induced biofeedback affects pelvic and trunk alignment and muscle activity of the Gmed and trunk muscles during OLS. In addition, deviation of the pelvis and trunk during OLS occur together, but previous studies have investigated changes in Gmed activity by experimentally controlling pelvic deviation and trunk deviation separately [14, 15]. Thus, it is still necessary to identify changes in Gmed muscle activity together with changes in trunk and pelvic alignment caused by different OLS strategies.

Therefore, the aim of the present study was to investigate the effects of pressure biofeedback on pelvic and trunk alignment as well as Gmed, MF, and QL activity during OLS. We hypothesized that pressure biofeedback would reduce pelvic and trunk deviation and increase the activity of the Gmed, MF, and QL during OLS.

2. Methods

2.1 Participants

Twenty-four healthy male volunteers with right leg dominance (age, 32.46 $\pm$ 3.65 years; height, 177.59 $\pm$ 4.78 cm; weight 74.33 $\pm$ 6.24 kg) participated in this study, which took place from June to August of 2019. This cross-sectional study was included in the Clinical Research Information Service database, with reference number KCT0004059 (date of approval: June 13, 2019). The dominant leg was defined as the kicking leg Participants were included if they had no previous or current lower back pain, lower extremity injuries, or psychological problems and were able to maintain a 5-s OLS. We excluded volunteers who could not maintain the correct posture or who presented with pain in the back or lower extremities during OLS. All participants read and signed an informed consent form approved by the Pusan National University Yangsan Hospital Ethics Committee for Human Investigations (No. 05-2019-042).

Figure 1.

One-leg standing with and without pressure biofeedback. (A) Preferred one-leg standing without pressure biofeedback. (B) One-leg standing with pressure biofeedback. (C) Real-time pressure change display during one-leg standing.

The sample size was determined using the statistical package G*Power 3 based on previous findings [14]. The results of the power analysis showed that at least 22 participants would be required for a power of 0.80 at an $\alpha$ -level of 0.05. Owing to the possibility of withdrawals and loss to follow-up, we recruited 24 subjects.

2.2 Examination procedure

Prior to the experiment, all participants completed a questionnaire relating to demographic parameters. The examiners then measured lumbopelvic alignment and activity of the Gmed, MF, and QL under all OLS conditions with OLS performed at 30 ${}^{\circ }$ of hip flexion with the dominant leg supporting the body. Two examiners were involved in this study. Examiner 1 measured the lumbopelvic alignment, including the femoropelvic and trunk lean angles whereas examiner 2 measured the activity of the right Gmed, QL, and bilateral MF.

2.3 One-leg standing

Prior to performing the experiment, each participant practiced the OLS conditions under the supervision of examiner 1 for 10 min. All participants performed the following three OLS conditions: (1) without pressure biofeedback (PB $-$ ); (2) with pressure biofeedback corresponding to a threshold of 50% of the maximal pressure (PB $+$ 1); and (3) with pressure biofeedback with minimal change in pressure (PB $+$ 2). All participants performed under the PB $-$ condition first, followed by the PB $+$ 1 and PB $+$ 2 conditions

To perform under the PB $-$ condition, participants stood barefoot with their legs together and arms across their chests. The subjects were then instructed to lift the non-supporting (non-dominant) leg to 30 ${}^{\circ }$ of hip flexion [20]. The 30 ${}^{\circ }$ of hip flexion was measured using a goniometer before the experiment, and a target bar was used to identify 30 ${}^{\circ }$ of hip flexion while participants performed under the PB $-$ condition (Fig. 1A).

Before performing under the other PB conditions, a 4D-MT pressure sensor (ReLive, Seoul, Korea) was placed between the lateral surface of left (non-supporting) humerus and the wall with participants standing with their non-supported side next to the wall. The pressure sensor consisted of a cuff and device that receives pressure information. The pressure information was transferred in real-time via Bluetooth to the 4D-MT pressure analysis program (ReLive) on an android tablet PC. Real-time pressure changes were displayed on the tablet PC screen. Calibration was performed by placing the lateral surface of the humerus on the pressure sensor. We performed a pilot test with five subjects to determine the pressure threshold value for PB $+$ 1. To set the threshold value, the maximum wall-pressure value, caused only by trunk lean to the non-supporting side, was measured during the PB $-$ condition for all participants prior to the start of the experiment and without the subjects’ knowledge. In the pilot study, participants performed OLS with a threshold at 30%, 40%, and 50% of the maximum wall pressure value using pressure biofeedback. However, participants had difficulty in performing OLS while maintaining pressure under 40% of the maximum wall pressure. Thus, in the present study, the threshold value for PB $+$ 1 was set at 50% of the maximum wall pressure. To perform under the PB $+$ 1 condition, the participants were asked to perform OLS while maintaining pressure under 50% (Fig. 1B). For PB $+$ 2, participants were asked to perform OLS while maintaining pressure as close as possible to the pressure at their initial position (i.e., before performing hip flexion on the non-supporting side) (Fig. 1B). During both PB $+$ conditions, participants monitored the change in pressure caused by their trunk deviation on the tablet PC screen. A blue line displayed on the tablet PC screen represented the designated threshold, and a red bar indicated the pressure value in real time (Fig. 1C).

Data collection was discontinued if the position of the pressure sensor on the wall changed or the pressure sensor could not be maintained between the humerus and the wall due to postural sway with increased thoracic kyphosis, increased lumbar lordosis or kyphosis, trunk rotation, or knee flexion of the supporting leg. OLS posture was maintained for 5 seconds during each of the three conditions, which, in turn, were repeated three times. Each participant had a 30-s rest time between trials and a 20-min rest time between exercises to prevent muscle fatigue and learning effects.

2.4 Pelvic and trunk alignment

The femoropelvic angle and trunk lean angle were measured to analyze trunk alignment during all three OLS conditions [20]. Markers were attached bilaterally to the acromion processes, anterior superior iliac spines (ASIS), and femoral condyles, and a digital camera attached to a tripod was placed 2.5 m from the participants at the height of the ASIS. The femoropelvic angle was calculated as the angle between the line connecting the two ASISs and the line joining the ASIS and lateral femur of the supporting leg. A smaller femoropelvic angle indicates increased pelvic lateral deviation to the supporting leg side or increased hip adduction on the supporting leg side. The trunk lean angle was measured as the angle between the line connecting the midpoint of the two ASISs and the midpoint of both acromion processes and a vertical line (Fig. 2). Greater trunk lean angle indicates lateral trunk lean to the non-supporting leg side relative to the pelvis. The photographic images captured by the digital camera were transferred to a personal notebook and analyzed using ImageJ software (ImageJ 1.44, NIH). The mean of three measurements was used for statistical analyses.

2.5 Hip and trunk muscles activation

Wireless TeleMyo DTS (Noraxon, Inc., Scottsdale, AZ, USA) and Myo-Research Master Edition 1.06 XP software were used to collect electromyographic (EMG) data for the Gmed and trunk muscles. This device is an eight-channel portable microcomputer with an eight-channel A/D conversion (12-bit resolution). Before placing the electrodes, the attachment sites were shaved and cleaned using rubbing alcohol to minimize skin impedance. The surface electrode pairs (AG/AGCl) had a pre-gelled diameter of 10 mm, and the inter-electrode distance was 2 cm. First, the electrode on the right Gmed was attached on the proximal side one-third of the distance between the iliac crest and greater trochanter. Then, the electrodes on both QL muscles were attached approximately 4 cm lateral from the muscle belly of the erector spinae, at the mid-point between the iliac crest and 12th rib [21]. Finally, the electrodes on both MF muscles were attached above and below and aligned transversely to the posterior superior iliac spines [22], parallel to the direction of the muscle fibers in the middle of the muscle belly. The raw EMG signals were sampled at 1,000 Hz and calculated using the root mean square (RMS) with a 50-ms interval. A band pass filter of 20–450 Hz and notch filter at 60 Hz were also used.

Figure 2.

Measures of femoropelvic angle (A) and trunk lean angle (B).

Normalization of EMG data was performed using a maximal voluntary isometric contraction (MVIC) for 5 s. The Gmed was tested in the side-lying position with the examiner providing manual resistance at the ankle while the participant abducted his hip. The QL was also tested in the same position. When participants elevated the pelvis and ribs off the floor in trunk lateral flexion, the examiner applied manual resistance to the hip and pelvis [23]. The MF was tested in the prone position with the leg strapped to the table. The examiner applied manual resistance to the posterior scapula while the participant extended the trunk. All EMG data were measured for 5 s, and the first and last 1 s were excluded from the analysis. The MVIC was performed three times; participants were given a 3-min rest between trials.

All EMG data are expressed as percentage of MVIC (%MVIC). For each OLS condition, the EMG data for the Gmed and trunk muscles were measured three times for 5 s each, and the first and last 1 s were excluded from the analysis.

Table 1

Activity of the gluteus medius and trunk muscles under three one-leg standing conditions

	PB $-$ (95% CI)	PB $+$ 1 (95% CI)	PB $+$ 2 (95% CI)	$p$ -value	$F$ -value
Gmed (%MVIC)	19.09 $\pm$ 10.16 (14.80–23.38)	23.40 $\pm$ 10.05 (19.16–27.65)	24.77 $\pm$ 10.18 (20.47–27.65)	$<$ 0.001 ${}^{\ast }$	16.692
Right MF (%MVIC)	15.17 $\pm$ 7.66 (12.42–18.77)	18.47 $\pm$ 9.54 (15.24–22.99)	19.10 $\pm$ 9.97 (15.72–23.82)	0.02 ${}^{\ast }$	7.231
Left MF (%MVIC)	17.03 $\pm$ 8.75 (13.92–21.15)	25.33 $\pm$ 13.42 (20.82–31.69)	24.79 $\pm$ 13.59 (20.17–31.23)	$<$ 0.001 ${}^{\ast }$	22.02
Right QL (%MVIC)	11.82 $\pm$ 6.76 (9.74–15.32)	13.67 $\pm$ 8.40 (10.03–17.06)	13.33 $\pm$ 7.24 (10.21–16.25)	0.017	5.619
Left QL (%MVIC)	13.72 $\pm$ 8.86 (9.98–17.47)	17.24 $\pm$ 8.05 (13.84–20.64)	17.95 $\pm$ 9.59 (13.90–21.99)	0.002 ${}^{\ast }$	7.996

${}^{*}$ indicates statistically significant results. Abbreviations: CI, confidence interval; Gmed, gluteus medius; MF, multifidus; MVIC, maximal voluntary isometric contraction; OLS, one-leg standing; PB $+$ , one-leg standing with pressure biofeedback; PB $-$ , one-leg standing without pressure biofeedback; QL, quadratus lumborum.

Table 2

Effect size of the muscle activity between conditions

	PB $-$ versus PB $+$ 1	PB $-$ versus PB $+$ 2	PB $+$ 1 versus PB $\mathrm{+}$ 2
Gmed	0.43	0.56	0.14
Right MF	0.38	0.43	0.06
Left MF	0.70	0.65	0.04
Right QL	0.24	0.22	0.04
Left QL	0.41	0.46	0.08

Abbreviations: Gmed, gluteus medius; MF, multifidus; OLS, one-leg standing; PB $+$ , one-leg standing with pressure biofeedback; PB $-$ , one-leg standing without pressure biofeedback; QL, quadratus lumborum.

2.6 Statistical analysis

All data are presented as mean $\pm$ standard deviation. The Kolmogorov-Smirnov test was used to identify the normal distribution of the outcome measures. A one-way repeated-measures analysis of variance (ANOVA) was applied to identify differences in all variables during the three OLS conditions. If a significant difference was detected, the Bonferroni correction was performed for post hoc analysis. The effect size for EMG activity in the measured muscles was calculated using Cohen’s $d$ All statistical analyses were performed with SPSS software (ver. 20.0 SPSS, Inc., Chicago, IL, USA), and the level of statistical significance was set at $p<$ 0.05.

3. Results

The Kolmogorov-Smirnov test indicated normal distribution in all outcome variables. The muscle activity data and femoropelvic and trunk lean angles are shown in Tables 1 and 2.

The activity of the right Gmed, bilateral QL, and bilateral MF muscles was significantly higher under both PB $+$ conditions compared to the PB $-$ condition ( $p<$ 0.05) (Table 1). There were no significant differences in muscle activity between PB $+$ 1 and PB $+$ 2. The effect size values for muscle activity among conditions are shown in Table 2.

Table 3

Lumbopelvic alignment angles during the three one-leg standing conditions

	PB $-$ (95% CI)	PB $+$ 1 (95% CI)	PB $+$ 2 (95% CI)	$p$ -value	$F$ -value
Femoropelvic angle ( ${}^{\circ }$ )	86.02 $\pm$ 3.46 (84.56–87.48)	88.32 $\pm$ 3.21 (86.96–89.68)	88.60 $\pm$ 3.04 (87.39–89.96)	$<$ 0.001 ${}^{\ast }$	24.39
Trunk lean angle ( ${}^{\circ }$ )	2.85 $\pm$ 2.26 (1.90–3.81)	1.54 $\pm$ 0.96 (1.13–1.95)	1.42 $\pm$ 0.90 (1.04–1.80)	0.005 ${}^{\ast }$	10.142

${}^{*}$ indicates statistically significant results. Abbreviations: CI, confidence interval; OLS, one-leg standing; PB $+$ , one-leg standing with pressure biofeedback; PB $-$ , one-leg standing without pressure biofeedback.

There was a significant difference in both the femoropelvic and trunk lean angles between the OLS conditions with and without pressure biofeedback (Table 3). Compared to the PB $-$ condition, both PB $+$ 1 and PB $+$ 2 resulted in a significantly greater femoropelvic angle and significantly lower trunk lean angle ( $p<$ 0.05). There were no significant differences in the femoropelvic or trunk lean angles between PB $+$ 1 and PB $+$ 2.

4. Discussion

The present study showed that pressure biofeedback arising from an interface between the lateral arm on the non-supporting leg side and a wall during OLS led to increased activity of the Gmed on the supporting leg side and the bilateral MF and QL muscles, together with an increase in femoropelvic angle on the supporting leg side and decrease in trunk lean angle to the non-supporting leg side compared to OLS without pressure biofeedback.

Our findings show that Gmed, MF, and QL activity were significantly greater under both PB $+$ conditions compared to the PB $-$ condition. During the PB $-$ condition, the base of support was reduced to the supporting leg, resulting in compensatory trunk and pelvic deviation [12, 13, 24, 25]. In contrast, when trunk deviation is monitored through pressure in real time, trunk deviation and pelvic lateral deviation can be decreased. Decreasing pelvic deviation during OLS reduces the internal moment arm of the Gmed [7], which requires greater Gmed force to maintain balance, leading to increased Gmed activity [13]. Under the PB $-$ condition, there was greater pelvic deviation compared to both PB $+$ conditions, which explains the concomitant decrease in muscle activity. Previous findings showing increased EMG activity of the gluteal muscles together with a decreased pelvic tilt during hip movement when applying pressure biofeedback support our results [18, 19].

Load transfer occurs during OLS and is controlled by the co-activation of local and global muscles [8]. The self-bracing mechanism is generated by activation of the MF in the lumbopelvic neutral position to reduce load transfer [9]. Both PB $+$ conditions were closer to the lumbopelvic neutral position than the PB $-$ condition; thus, greater MF muscle activation was required to generate the self-bracing mechanism and maintain the lumbopelvic neutral position, resulting in increased EMG activity of MF muscles. In addition, QL muscle activity is reportedly increased as vertebral compression increases through symmetrically progressive axial loads [10]. Both PB $+$ conditions produced smaller changes in trunk and pelvic alignment than did the PB $-$ condition, resulting in an increased axial load. Thus, the PB $+$ conditions required greater activation of the QL muscles than the PB $-$ condition to maintain this posture, resulting in increased EMG activity of the QL muscles. Previous findings by Kang et al. [13] demonstrated that QL muscle activity increased with reduced lateral pelvic tilt during unilateral weight-bearing hip abduction, which supports our reasoning. Despite the significant differences in the EMG activity between the PB $-$ and PB $+$ conditions in both the QL muscles and MF muscle on the supporting leg side, the changes in EMG activity in these muscles were associated with small to medium effect sizes, whereas the effect sizes of changes in EMG activity in the Gmed on the supporting leg side and the MF muscle on the non-supporting leg side were medium to large under the PB $+$ conditions. In addition, considering the grading criteria for the interpretation of muscle activity suggested by a previous study (i.e., 0–10% MVIC: relatively inactive; 11–20% MVIC: minimally active; 21–35% MVIC: moderately active; 36–50% MVIC: moderately strongly active; greater than 50% MVIC: markedly active) [26], the EMG activity in the Gmed on the supporting leg side and the MF muscle on the non-supporting leg side changed from “minimally active” to “moderately active” under the PB $+$ conditions. Therefore, we believe that the changes in EMG activity in the Gmed on the supporting leg side and the MF muscle on the non-supporting leg side between the PB $-$ and PB $+$ conditions were clinically meaningful.

Our results demonstrated that the PB $+$ conditions led to a significantly greater femoropelvic angle and decreased trunk lean angle compared with the PB $-$ condition. That is, during the PB $-$ condition, the pelvis was deviated to the supporting side, and the trunk to the unsupported side, compared with both PB $+$ conditions. During the PB $-$ condition, pelvic deviation may have occurred to maintain the center of mass within the narrowed base of support, and the trunk deviation was likely a counterforce to the pelvic deviation to the supporting side. However, during the PB $+$ conditions, trunk and pelvic deviation may have been reduced by checking pressure changes in real time. Because the participants were made aware by the pressure increase that the trunk had deviated to the unsupported side, they were able to shift the trunk to the supported side to correct their lumbopelvic posture. This result seems to explain why the PB $-$ condition resulted in a greater trunk lean angle than both PB $+$ conditions. Also, the femoropelvic angle increased when the pelvic lateral deviation increased as the trunk leaned away from the supported side. Our results are in agreement with a previous study by Cynn et al. [18], who reported that pelvic lateral tilt was reduced when pressure changes were observed during side-lying hip abduction. In addition, Oh et al. [19] demonstrated that anterior tilt was decreased when pressure biofeedback was applied during prone hip extension. Therefore, we believe pressure biofeedback is a useful modality for correcting pelvic and trunk alignment during OLS.

We found no significant differences in muscle activity or lumbopelvic alignment between the PB $+$ 1 and PB $+$ 2 conditions. These findings may result from the similar amount of wall pressure under PB $+$ 1 and PB $+$ 2. During pilot testing to determine the pressure threshold values for PB $+$ 1, participants performed under PB $+$ conditions with a pressure threshold value set at 50% of the maximum wall pressure generated during the PB $-$ condition. However, they had difficulty performing under PB $+$ conditions with a threshold less than 40% of the maximum wall pressure value. Based on our present findings showing no significant differences in outcome measures between PB $+$ 1 and PB $+$ 2, combined with the findings of the pilot test, we can infer that about 50% of the maximum wall pressure would have occurred during PB $+$ 2 even if the participants tried to keep the wall pressure to a minimum. Thus, our findings indicate that at least 50% of the trunk deviation that occurred under the PB $-$ condition should be allowed to maintain balance during OLS tasks.

There are some limitations in this study. First, because only healthy men participated in this study, the results cannot be generalized to patients with low back pain or lower limb injury, to women, or to elderly people. Second, OLS conditions were performed only on the dominant side. To extend the clinical implications of this study, further studies will need to investigate OLS with biofeedback using both legs. Third, because this study investigated only the immediate effects of pressure biofeedback, no long-term effects are known. Finally, to clarify foot positions/movements and identify their effects on lumbopelvic alignment during OLS tasks, more information (e.g., center of pressure) is needed. In addition, a longer duration for OLS should be considered, as 5 s may be too short for participants to acquire balance during OLS.

5. Conclusions

We investigated the effects of pressure biofeedback during OLS on the activity of the Gmed and trunk muscles and lumbopelvic alignment. The use of real-time pressure biofeedback during one-leg standing increased the activity of the Gmed, MF, and QL and resulted in a greater femoropelvic angle and lower trunk lean angle compared to one-leg standing without biofeedback. Thus, we recommend using pressure biofeedback during the OLS exercise to increase activity of the Gmed and trunk muscles, especially the MF muscle on the non-supporting leg side, and to prevent trunk deviation and pelvic lateral deviation.

Author contributions

CONCEPTION: Soo-Yong Kim, Il-Young Yu and Min-Hyeok Kang.

PERFORMANCE OF WORK: Soo-Yong Kim, Il-Young Yu and Min-Hyeok Kang.

INTERPRETATION OR ANALYSIS OF DATA: Soo-Yong Kim and Min-Hyeok Kang.

PREPARATION OF THE MANUSCRIPT: Soo-Yong Kim, Il-Young Yu and Min-Hyeok Kang.

REVISION FOR IMPORTANT INTELLECTUAL CONTENT: Soo-Yong Kim and Min-Hyeok Kang.

SUPERVISION: Min-Hyeok Kang.

Ethical considerations

All participants read and signed an informed consent form approved by the Pusan National University Yangsan Hospital Ethics Committee for Human Investigations (No. 05-2019-042).

Funding

The authors report no funding.