Sage Journals: Discover world-class research

Abstract

Objective

Epidemiological studies on the association of serum oestradiol levels and inflammatory markers have reported inconsistent and conflicting results. Therefore, we investigated the association between serum oestradiol and high-sensitivity C-reactive protein (CRP) levels in women on the basis of their menopausal status.

Methods

This cross-sectional study examined the association between serum oestradiol and CRP levels on the basis of menopausal status in 151 premenopausal women aged 42.7 ± 6.7 years and 394 postmenopausal women aged 58.1 ± 6.7 years who participated in a health examination program. Multiple linear regression analysis was conducted using CRP levels as the dependent variable.

Results

Multiple linear regression analysis showed that serum oestradiol levels were inversely associated with CRP levels in premenopausal women (β coefficient = −0.298) after adjusting for age, body mass index, smoking, mean arterial pressure, and levels of fasting plasma glucose, triglycerides, high-density lipoprotein cholesterol, aspartate aminotransferase, and alanine aminotransferase. However, this association was not found in postmenopausal women after adjusting for the same confounding factors.

Conclusions

Serum oestradiol levels are inversely associated with CRP levels in premenopausal women, but not in postmenopausal women. Lower oestrogenic activity may at least partly contribute to the pathogenesis of chronic inflammation, particularly in premenopausal women.

Keywords

Menopause oestradiol C-reactive protein chronic inflammation cardiovascular disease sex hormone

Introduction

Cardiovascular disease (CVD) is caused by multifactorial interrelated mechanisms, among which chronic low-grade inflammation plays a major role.¹ In women, menopause is an important contributing factor to CVD in addition to traditional risk factors, including obesity, hypertension, type 2 diabetes, and dyslipidaemia.² Postmenopausal women are more susceptible to weight gain and visceral fat accumulation accompanied by a marked decrease in oestrogen levels.³ These changes influence lipid metabolism and the vascular endothelium,⁴ increasing the risk of atherosclerotic CVD in postmenopausal women.⁵

Current evidence supports the existence of sex-specific relationships between sex hormones and inflammatory markers. There is a consistent inverse relationship between serum testosterone levels and inflammatory markers in men.^6,7 However, epidemiological studies that investigated the association between serum oestradiol levels and inflammatory markers in women, particularly when categorized by menopausal status, have shown inconsistent and even conflicting results. Some observational studies found a positive association of serum oestradiol levels and inflammatory markers in postmenopausal women^8,9 and an inverse association in premenopausal women,^10,11 whereas others did not show significant relationships between these variables.¹² Although the reasons for this discrepancy among studies are not clear, the effects of endogenous oestrogen on chronic low-grade inflammation may differ by menopausal status.

C-reactive protein (CRP) has traditionally been considered a non-specific marker of inflammation, but recent epidemiological evidence has highlighted the significance of high CRP levels in patients with CVD through cross-sectional and longitudinal studies.¹³ We hypothesized that serum oestrogen levels play differing roles in inflammation according to menopausal status. Therefore, this study aimed to investigate the association between serum oestradiol and CRP levels in women while taking into account their menopausal status.

Methods

Study participants

We reviewed the medical records of 689 women who voluntarily visited the Health Promotion Center of Gangnam Severance Hospital, for a routine health check-up, between November 2013 and July 2015. Natural menopause was defined as the absence of menstrual periods for 12 consecutive months. The exclusion criteria were as follows: current intake of oral contraceptives; exogenous oestrogen replacement; history of tamoxifen therapy; induced menopause, including bilateral oophorectomy-, radiation-, or drug-induced menopause; history of ischaemic heart disease or cerebrovascular, thyroid, respiratory, renal, liver, or rheumatological disease; missing data; and no fasting for 12 hours before testing (n = 81). Of the remaining participants, those with a CRP level ≥10 mg/L were excluded to rule out the possibility of infection or an inflammatory disorder (n = 9). We also excluded women who reported irregular menstruation over the previous year to rule out the possibility of polycystic ovarian syndrome or perimenopause (n = 54). Written informed consent was obtained from each participant. This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the institutional review board of Yonsei University Gangnam Severance Hospital (institutional review board number: 3-2019-0267).

Data collection

Each participant completed a self-reported questionnaire on lifestyle habits and menstrual and medical history. Cigarette smoking, alcohol consumption, and physical activity were determined on the basis of responses. Cigarette smoking was defined as a current habit of smoking. Alcohol drinking was defined as alcohol consumption on 2 or more days per week. Regular exercise was defined as engaging in purposeful physical activity three or more times per week. Menstrual history was determined by the response to the following question: “Has there been menstruation for 1 or more years?” with three response options. These response options were as follows: “yes,” “no, but there has been intermittent menstruation over the last year,” and “no, there has been no menstruation.” Medical examinations were performed by trained medical staff using a standardized procedure. Body weight and height were measured to the nearest 0.1 kg and 0.1 cm, respectively, and participants wore light indoor clothing without shoes. Body mass index (BMI) was calculated as the weight in kilograms divided by the square of the height in meters (kg/m²). Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured using the patient’s right arm with a standard mercury sphygmomanometer (Baumanometer; W.A. Baum Co Inc., Copiague, NY, USA). Mean arterial pressure was calculated using the following equation: SBP + 2 × DBP)/3.

All blood samples were obtained from the antecubital vein after fasting for 12 hours. Fasting plasma glucose, triglycerides, high-density lipoprotein (HDL) cholesterol, aspartate aminotransferase (AST), and alanine aminotransferase (ALT) levels were measured via enzymatic methods using the AU5800 automated chemistry analyser (Beckman Coulter, Fullerton, CA, USA). Serum oestradiol levels were measured by electrochemiluminescence immunoassay using the Cobas e601 immunoanalyser (Roche Diagnostics, Basel, Switzerland). In the present study, serum oestradiol levels were measured because there is strong agreement between serum and plasma oestrogen levels.¹⁴ High-sensitivity CRP levels were measured using the Roche/Hitachi 912 System (Roche Diagnostics, Indianapolis, IN, USA) with a latex-enhanced immunoturbidimetric method with a lower limit of detection of 0.02 mg/L.

Statistical analysis

Normal distribution of data was evaluated with determination of skewness using the Kolmogorov–Smirnov test. Triglyceride, AST, ALT, oestradiol, and CRP levels had a skewed distribution. The clinical characteristics of the cohort were compared by menopausal status using the independent two-sample test or the Wilcoxon rank-sum test for continuous variables taking into account normality of distribution, and the chi-squared test was used for categorical variables. Continuous data are presented as the mean ± standard deviation or median (interquartile range), whereas categorical data are presented as frequency. Pearson’s correlation was used to examine bivariate correlations between log-transformed CRP levels and clinical variables. Multiple linear regression analysis was conducted with the log-transformed CRP level as the dependent variable to examine the independent association between oestrogen and CRP levels. In this analysis, model 1 was adjusted for age and BMI. Model 2 was adjusted for age, BMI, cigarette smoking, mean arterial pressure, and levels of fasting plasma glucose, log-transformed triglyceride, and high-density lipoprotein-cholesterol. Model 3 was adjusted for age, BMI, cigarette smoking, mean arterial pressure, and levels of fasting plasma glucose, log-transformed triglycerides, high-density lipoprotein cholesterol, log-transformed AST, and log-transformed ALT. All analyses were conducted using SAS statistical software (version 9.4; SAS Institute Inc., Cary, NC, USA). All statistical tests were two-sided and a P value < 0.05 was considered statistically significant.

Results

Table 1 shows the clinical and biochemical characteristics of 545 participants who were included in the final analysis, including 151 premenopausal women and 394 postmenopausal women, by menopausal status. The mean ± standard deviation age of the premenopausal women was 42.7 ± 6.7 years (range: 21–52 years), whereas that of the postmenopausal women was 58.1 ± 6.7 years (range: 46–75 years). The mean age, BMI, blood pressure, and fasting plasma glucose levels, and the median triglyceride, AST, ALT, and CRP levels were significantly higher in postmenopausal women than in premenopausal women (all P < 0.001). However, mean HDL cholesterol and median oestradiol levels were significantly lower in postmenopausal women than in premenopausal women (both P < 0.001).

Table 1.

Characteristics of the study population.

Variables	Premenopausal women(n = 151)	Postmenopausal women(n = 394)	P value
Age (years)	42.7 ± 6.7	58.1 ± 6.7	<0.001
BMI (kg/m²)	21.5 ± 3.1	23.1 ± 3.0	<0.001
SBP (mmHg)	114.6 ± 17.8	124.8 ± 19.1	<0.001
DBP (mmHg)	70.5 ± 11.1	76.0 ± 11.0	<0.001
MAP (mmHg)	85.2 ± 12.9	92.3 ± 13.2	<0.001
FPG (mmol/L)	4.83 ± 0.63	5.30 ± 1.01	<0.001
Triglycerides (mmol/L)	0.76 (0.58–1.09)	0.90 (0.74–1.38)	<0.001
HDL cholesterol (mmol/L)	1.53 ± 0.32	1.41 ± 0.35	<0.001
AST (U/L)	17 (15–20)	21 (18–25)	<0.001
ALT (U/L)	14 (11–18)	19 (15–26)	<0.001
C-reactive protein (mg/L)	0.49 (0.30–1.00)	0.70 (0.32–1.50)	<0.001
Oestradiol (pmol/L)	442.4 (205.6–742.6)	29.4 (18.4–45.2)	<0.001
Current smoking (%)	6.9	2.7	0.061
Alcohol drinking (%)	20.4	12.3	0.017
Regular exercise (%)	30.4	40.0	0.044

Data are expressed as mean ± standard deviation, median (interquartile range), or %.

P values were calculated using the independent two sample t-test or the Wilcoxon rank-sum test for continuous variables on the basis of normality of distribution, or the chi-squared test for categorical variables.

Alcohol drinking refers to consuming alcohol on 2 or more days per week. Regular exercise refers to activity three or more times per week.

BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; MAP, mean arterial blood pressure; FPG, fasting plasma glucose; HDL, high-density lipoprotein; AST, aspartate aminotransferase; ALT, alanine aminotransferase.

Table 2 shows the results of Pearson’s correlation analysis between log-transformed CRP levels and clinical variables. Log-transformed CRP levels were significantly correlated with BMI, SBP, log-transformed triglyceride levels, log-transformed AST levels, and log-transformed oestradiol levels in premenopausal women (all P < 0.05). In postmenopausal women, log-transformed CRP levels were significantly correlated with age, BMI, SBP, DBP, mean arterial pressure, and levels of fasting plasma glucose, log-transformed triglycerides, HDL cholesterol, log-transformed AST, and log-transformed ALT (all P < 0.001). However, log-transformed CRP levels were not correlated with log-transformed oestradiol levels.

Table 2.

Correlation between log-transformed CRP levels and clinical variables.

	Premenopausal women		Postmenopausal women
	r	P value	r	P value
Age (years)	0.075	0.372	0.186	<0.001
BMI (kg/m²)	0.294	<0.001	0.388	<0.001
SBP (mmHg)	0.167	0.045	0.187	<0.001
DBP (mmHg)	0.115	0.189	0.178	<0.001
MAP (mmHg)	0.140	0.094	0.188	<0.001
FPG (mmol/L)	0.058	0.490	0.169	<0.001
Triglycerides (mmol/L)*	0.313	<0.001	0.168	<0.001
HDL cholesterol (mmol/L)	−0.144	0.085	−0.234	<0.001
AST (U/L)*	0.107	0.020	0.190	<0.001
ALT (U/L)*	0.149	0.074	0.198	<0.001
Oestradiol (pmol/L)*	−0.303	<0.001	−0.008	0.872

P values were calculated using Pearson’s correlation analysis.

*Indicates log-transformed values.

CRP, C-reactive protein; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; MAP, mean arterial blood pressure; FPG, fasting plasma glucose; HDL, high-density lipoprotein; AST, aspartate aminotransferase; ALT, alanine aminotransferase.

Table 3 shows the results of multiple linear regression analysis of the independent relationship between oestradiol and CRP levels. Log-transformed oestradiol levels were inversely associated with log-transformed CRP levels only in premenopausal women (P < 0.01 for all three models). However, log-transformed oestradiol levels were not associated with log-transformed CRP levels in postmenopausal women after adjusting for age, BMI, cigarette smoking, mean arterial pressure, and levels of fasting plasma glucose, log-transformed triglycerides, HDL cholesterol log-transformed AST, and log-transformed ALT.

Table 3.

Multiple linear regression analysis of the independent relationship between log-transformed oestradiol levels and log-transformed C-reactive protein levels.

	β coefficient	Standard error	P value
Premenopausal women
Model 1	−0.301	0.088	<0.001
Model 2	−0.295	0.087	<0.001
Model 3	−0.298	0.084	0.001
Postmenopausal women
Model 1	0.004	0.066	0.949
Model 2	−0.016	0.070	0.818
Model 3	−0.018	0.070	0.799

β refers to the standardized beta coefficient.

Model 1: adjusted for age and body mass index.

Model 2: adjusted for age, body mass index, cigarette smoking, mean arterial pressure, and levels of fasting plasma glucose, log-transformed triglycerides, and high-density lipoprotein cholesterol.

Model 3: adjusted for age, body mass index, cigarette smoking, mean arterial pressure, and levels of fasting plasma glucose, log-transformed triglycerides, high-density lipoprotein cholesterol, log-transformed aspartate aminotransferase, and log-transformed alanine aminotransferase.

Discussion

We found that serum oestradiol levels were inversely and independently associated with CRP levels in premenopausal women after adjusting for potential confounding variables. However, there was no such inverse relationship in postmenopausal women. Our findings are consistent with the results of previous studies, which showed that oestradiol levels were inversely associated with inflammatory markers in premenopausal women.^10,11

Previous studies on the association between oestradiol levels and inflammatory markers in women were inconclusive. Using the BioCycle Study dataset, Gaskins et al.¹⁰ reported an inverse association between oestradiol and CRP levels in 259 reproductive-aged women in the United States, which is consistent with the findings of our study. In contrast, serum oestradiol levels were positively associated with CRP levels in 513 Italian postmenopausal women who were investigated in the Invecchiare in Chianti study.⁹ Therefore, additional longitudinal studies are warranted to establish the relationship between serum oestradiol and CRP levels on the basis of menopausal status.

The mechanism underlying the menopause-specific relationship between serum oestradiol and CRP levels remains unclear. Oestrogen has an inhibitory effect on inflammatory cytokine production and inflammatory cell migration in non-reproductive tissues,^15,16 and oestrogen receptors are highly expressed in vascular smooth muscle and endothelial cells throughout the human body.^16,17 Oestrogen’s anti-inflammatory effects are, in part, mediated by nitric oxide production and cytokine suppression¹⁵ because nitric oxide is a key vasodilator.^18,19 Nitric oxide also plays an anti-inflammatory role in the endothelium owing to its role as a reactive oxygen species scavenger and an inhibitor of leukocyte recruitment.^20–22 Moreover, oestrogen decreases levels of tumour necrosis factor-α, which is a major pro-inflammatory cytokine.²³ These cascades potentially inhibit the synthesis and release of platelet-activating factors and chemokines, including interleukin-6, and may also down-regulate leukocyte-recruiting adhesion molecules, including intercellular adhesion molecule-1, vascular cell adhesion molecule-1, and E-selectin.^24–26 Oestrogen could also decrease production of reactive oxygen species in mitochondria, and this reduces chronic low-grade inflammation.^27,28 However, these effects of oestradiol may differ with the menopausal status. The lack of an inverse relationship between oestradiol and CRP levels in postmenopausal women may be explained by the observation that 80% of circulating oestradiol in such individuals originates from aromatization of testosterone, especially in adipose tissue.^9,29 Aromatase, which is an enzyme that converts testosterone into oestradiol, is stimulated by inflammatory cytokines and CRP.

There are several limitations of this study. First, this study had a cross-sectional design. Therefore, the results should be interpreted while taking into account causal and temporal considerations. Additional longitudinal studies are required to establish causality between serum oestradiol and CRP levels based on menopausal status. Second, the study participants included only women who underwent routine health screening at a single hospital. Therefore, these women may have comprised a relatively healthier proportion of the community and our cohort may not be representative of the general population. Third, oestradiol levels were measured by electrochemiluminescence immunoassay and not by liquid chromatography-mass spectrometry, even though the latter is considered the best method for measuring serum oestradiol levels. However, the electrochemiluminescence immunoassay method has several advantages, including a rapid turnaround time, high-throughput, and full automation. Fourth, serum oestradiol and CRP levels may fluctuate during the menstrual cycle.³⁰ We did not measure serum oestradiol levels and CRP levels at a common point during the cycle in our premenopausal participants. Fifth, other sex hormones, such as testosterone, were not measured in the present study. Further studies are required to determine the relationship between other sex hormones and inflammatory markers while taking into account menopausal status. Finally, although CRP is currently considered a reliable biomarker of chronic inflammation,¹³ caution is required when assuming anti- or pro-inflammatory activity based only on this biomarker because we did not directly quantify other serum inflammatory markers, such as tumour necrosis factor-α and interleukin-6, to support our findings.

Despite these limitations, this study has several strengths. We assessed the association between serum oestradiol and CRP levels in premenopausal and postmenopausal women separately. This may provide useful insight into the association between serum oestradiol and CRP levels on the basis of menopausal status. Moreover, a wide range of confounding factors closely related to chronic inflammation, including BMI, smoking status, blood pressure, fasting plasma glucose levels, triglyceride levels, HDL-cholesterol levels, and hepatic enzymes, were considered when performing multiple linear regression analyses. Additionally, to determine the true nature of the relationship between oestradiol and CRP levels in pre- and postmenopausal women, our study excluded participants with induced or secondary menopause, as well as women who used oral contraceptives or who were undergoing oestrogen replacement therapy. Induced menopause causes a sudden onset of obesity and metabolic disturbance followed by an abrupt decline in ovarian hormones, which could cause chronic inflammation.³¹

In conclusion, our data show that serum oestradiol levels are inversely associated with CRP levels in premenopausal women, but not in postmenopausal women. Our findings suggest that lower oestrogenic activity may at least partly contribute to the pathogenesis of chronic inflammation, particularly in premenopausal women.

Footnotes

Declaration of conflicting interest

The authors declare that there is no conflict of interest.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

ORCID iD

Yong-Jae Lee

References

Golia

Limongelli

Natale

, et al. Inflammation and cardiovascular disease: from pathogenesis to therapeutic target. Curr Atheroscler Rep 2014; 16: 435.

Rosano

Vitale

Marazzi

, et al. Menopause and cardiovascular disease: the evidence. Climacteric 2007; 10: 19–24.

Davis

Castelo-Branco

Chedraui

, et al. Understanding weight gain at menopause. Climacteric 2012; 15: 419–429.

Izumi

Muano

Mori

, et al. Common carotid artery stiffness, cardiovascular function and lipid metabolism after menopause. Life Sci 2006; 78: 1696–1701.

Newson

Menopause and cardiovascular disease.

Post Reprod Health 2018; 24: 44–49.

Carrero

Qureshi

Parini

, et al. Low serum testosterone increases mortality risk among male dialysis patients. J Am Soc Nephrol 2009; 20: 613–620.

Bobjer

Katrinaki

Tsatsanis

, et al. Negative association between testosterone concentration and inflammatory markers in young men: a nested cross-sectional study. PLoS One 2012; 8: e61466.

Störk

Bots

Grobbee

, et al. Endogenous sex hormones and C-reactive protein in healthy postmenopausal women. J Intern Med 2008; 264: 245–253.

Maggio

Ceda

Lauretani

, et al. SHBG, sex hormones, and inflammatory markers in older women. J Clin Endocrinol Metab 2001; 96: 1053–1059.

10.

Gaskins

Wilchesky

Mumford

, et al. Endogenous reproductive hormones and C-reactive protein across the menstrual cycle: the BioCycle Study. Am J Epidemiol 2012; 175: 423–431.

11.

Blum

Müller

Huber

, et al. Low-grade inflammation and estimates of insulin resistance during the menstrual cycle in lean and overweight women. J Clin Endocrinol Metab 2005; 90: 3230–3235.

12.

Wunder

Yared

Bersinger

, et al. Serum leptin and C-reactive protein levels in the physiological spontaneous menstrual cycle in reproductive age women. Eur J Endocrinol 2006; 155: 137–142.

13.

Ridker

Rifai

Cook

, et al. Non-HDL cholesterol, apolipoproteins A-I and B100, standard lipid measures, lipid ratios, and CRP as risk factors for cardiovascular disease in women. JAMA 2005; 294: 326–333.

14.

Coburn

Stanczyk

Falk

, et al. Comparability of serum, plasma, and urinary estrogen and estrogen metabolite measurements by sex and menopausal status. Cancer Causes Control 2019; 30: 75–86.

15.

Chakrabarti

Lekontseva

Davidge

ST.

Estrogen is a modulator of vascular inflammation.

IUBMB Life 2008; 60: 376–382.

16.

Gilliver

SC.

Sex steroids as inflammatory regulators.

J Steroid Biochem Mol Biol 2010; 120: 105–115.

17.

Gray

Sharif

Webb

, et al. Oestrogen and the cardiovascular system: the good, the bad and the puzzling. Trends Pharmacol Sci 2001; 22: 152–156.

18.

MacRitchie

Jun

Chen

, et al. Estrogen upregulates endothelial nitric oxide synthase gene expression in fetal pulmonary artery endothelium. Circ Res 1997; 81: 355–362.

19.

Haynes

Sinha

, et al. Src kinase mediates phosphatidylinositol 3-kinase/Akt-dependent rapid endothelial nitric-oxide synthase activation by estrogen. J Biol Chem 2003; 278: 2118–2123.

20.

Kubes

Kurose

Granger

DN.

NO donors prevent integrin-induced leukocyte adhesion but not P-selectin-dependent rolling in postischemic venules.

Am J Physiol 1994; 267: H931–H937.

21.

Wink

Hanbauer

Krishna

, et al. Nitric oxide protects against cellular damage and cytotoxicity from reactive oxygen species. Proc Natl Acad Sci USA 1993; 90: 9813–9817.

22.

Niu

Smith

Kubes

Intracellular oxidative stress induced by nitric oxide synthesis inhibition increases endothelial cell adhesion to neutrophils.

Circ Res 1994; 74: 1133–1140.

23.

Srivastava

Weitzmann

Cenci

, et al. Estrogen decreases TNF gene expression by blocking JNK activity and the resulting production of c-Jun and JunD. J Clin Invest 1999; 104: 503–513.

24.

Korpelainen

Gamble

Smith

, et al. The receptor for interleukin 3 is selectively induced in human endothelial cells by tumor necrosis factor alpha and potentiates interleukin 8 secretion and neutrophil transmigration. Proc Natl Acad Sci USA 1993; 90: 11137–11141.

25.

Camussi

Albano

Tetta

, et al. The molecular action of tumor necrosis factor-alpha. Eur J Biochem 1991; 202: 3–14.

26.

Kim

Moon

Kim

, et al. Vascular endothelial growth factor expression of intercellular adhesion molecule 1 (ICAM-1), vascular cell adhesion molecule 1 (VCAM-1), and E-selectin through nuclear factor-kappa B activation in endothelial cells. J Biol Chem 2001; 276: 7614–7620.

27.

Stirone

Duckles

Krause

, et al. Estrogen increases mitochondrial efficiency and reduces oxidative stress in cerebral blood vessels. Mol Pharmacol 2005; 68: 959–965.

28.

Reuter

Gupta

Chaturvedi

, et al. Oxidative stress, inflammation, and cancer: how are they linked? Free Radic Biol Med 2010; 49: c1603–c1616.

29.

Zhao

Nichols

Valdez

, et al. Tumor necrosis factor-alpha stimulates aromatase gene expression in human adipose stromal cells through use of an activating protein-1 binding site upstream of promoter 1.4. Mol Endocrinol 1996; 10: 1350–1357.

30.

Wander

Brindle

O'Connor

KA.

C-reactive protein across the menstrual cycle.

Am J Phys Anthropol 2008; 136: 138–146.

31.

Romero-Aleshire

Diamond-Stanic

Hasty

, et al. Loss of ovarian function in the VCD mouse-model of menopause leads to insulin resistance and a rapid progression into the metabolic syndrome. Am J Physiol Regul Integr Comp Physiol 2009; 297: R587–R592.

Serum oestradiol levels are inversely associated with C-reactive protein levels in premenopausal women,but not postmenopausal women

Abstract

Objective

Methods

Results

Conclusions

Keywords

Introduction

Methods

Study participants

Data collection

Statistical analysis

Results

Discussion

Footnotes

Declaration of conflicting interest

Funding

ORCID iD

References