Differential effects of systemic immune inflammation indices on hepatic steatosis and hepatic fibrosis: evidence from NHANES 1999–2018 Authors: Shuyin Duan, Zhanwen Tu, Lijuan Duan, Runqi Tu Published: 2024 ABSTRACT AbstractBackgroundSeveral studies have demonstrated that systemic immune inflammation index (SII) has a positive relationship with hepatic steatosis. However, it is lack of system evidence for the correlation between SII and hepatic fibrosis. The objective of this study was to evaluate the relationships between SII and hepatic steatosis or hepatic fibrosis.MethodsA cross-sectional analysis was performed from the National Health and Nutrition Examination Survey (NHANES). Fibrosis-4 index (FIB-4), NAFLD fibrosis score (NFS) and hepamet fibrosis score (HFS) were the indicators for hepatic fibrosis; fatty liver index (FLI), NAFLD liver fat score (LFS) and Framingham steatosis index (FSI) were the indicators for hepatic steatosis. Pearson’s test, generalized linear model (GLM) and restricted cubic splines (RCS) were used to analyze associations of SII with hepatic fibrosis and hepatic steatosis.ResultsA total of 21,833 participants were enrolled in the study. Pearson’s test and GLM revealed that there were negative relationships between SII and hepatic fibrosis (FIB-4, NFS and HFS), while positive relationships between SII and hepatic steatosis (FLI, LFS and FSI). The corresponding β (95%CI) of SII and hepatic fibrosis were − 0.35(-0.46, -0.24), -0.67(-0.71, -0.63) and − 0.10(-0.12, -0.09), respectively. The corresponding β (95%CI) of SII and hepatic steatosis were 6.12(4.75, 7.50), 0.22(0.12, 0.31) and 0.27(0.20, 0.34), respectively. Statistically significant non-linear association were found in SII with hepatic fibrosis and hepatic steatosis in RCS model (all P < 0.001).ConclusionThere was a negative significant association between SII and hepatic fibrosis, while a positive significant association between SII and hepatic steatosis.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12876-024-03557-5.Keywords: Systemic immune inflammation index, Hepatic steatosis, Hepatic fibrosis Introduction On June 24, 2023, the American Association for the Study of Liver Diseases (AASLD), the European Association for the Study of the Liver (EASL) and the Latin American Society for the Study of the Liver (ALEH), Led by 53 experts from around the world, published the Newly named Multi-Society Delphi Consensus on Fatty Liver Disease. Proposed to rename non-alcoholic fatty liver disease (NAFLD) to Metabolically associated Fatty degeneration Liver Disease (MASLD) [1]. MASLD is a main cause of liver disease worldwide, which warrants the attention of related medical workers and health policy makers [2]. With the rapid societal development and lifestyle changes, MASLD has been gradually increased. Several studies showed that prevalence exceeding 20% in developed countries [2] as well as a global prevalence of 25% [3]. Additionally, health care utilization and expenditure are particularly high among MASLD patients those requiring hospital admission [4]. One study suggests that overnutrition is the primary driver of MASLD, which could induce expansion of fat reservoirs and accumulation of ectopic fat [3]. Another study also indicates that fructose could induce to ATP depletion and suppression of mitochondrial fatty acid oxidation, leading to increased production of reactive oxygen species [5]. Numerous studies have also indicated that inflammatory probably play a crucial role in disease progression [6, 7]. Systemic immune inflammation index (SII) is constructed based on platelet counts, lymphocyte and neutrophil, which is a comprehensive biomarker and reflects local immune response and systemic inflammation of the whole body [8]. Several studies have demonstrated that SII is an independent predictor of patients with hepatocellular carcinoma, sepsis and colorectal cancer [8–11]. SII can objectively reflects the severity of inflammation as well as immune pathways in the host [9, 12]. More and more biomarkers about the MASLD have been studied. On the one hand, these are hepatic fibrosis related biomarkers, such as fibrosis-4 index (FIB-4) [13], NAFLD fibrosis score (NFS) [14] and hepamet fibrosis score (HFS) [15]. On the other hand, these are hepatic steatosis related biomarkers, such as fatty liver index (FLI) [16], NAFLD liver fat score (LFS) [17] and Framingham steatosis index (FSI) [18]. Numerous studies have proved that inflammation is an important factor in hepatic steatosis [19–21]. In addition, one study has suggested that SII is not association with hepatic fibrosis [22]. Furthermore, there is still a lack of a systematic investigation to explore the association of SII with hepatic steatosis and hepatic fibrosis. Thus, in this study, we aim to assess the association of SII with hepatic steatosis and hepatic fibrosis, using a large sample of adults from the National Health and Nutrition Examination Survey (NHANES). Materials and methods Data and sample source NHANES was a large population-based cross-sectional survey conducted by the National Center for Health Statistics (NCHS) in the United States. The Details of the NHANES design and data publicly available had been described at https://www.cdc.gov/nchs/nhanes/. Briefly, NHANES was a stratified, multistage probability approach, which surveyed about 5000 participants from across the country every year. The data included demographic data, questionnaire data, examination data, laboratory data and limited access data. The survey had been conducted over 10 cycles (1999–2018). Subjects were excluded if they met any of the following criteria: (1) missing data on SII, FIB-4, NFS, HFS, FLI, FSI and LFS; (2) age < 18 years; (3) with HBV or HCV. Systemic immune-inflammation biomarkers detection The SII was derived from complete blood count. The standardized protocols for measuring blood count were provided by Laboratory Procedures Manual of NHANES. The details can be found at https://www.cdc.gov/nchs/nhanes/biospecimens/serum_plasma_urine.htm. In this study, the following equation was utilized to calculate SII = platelet counts × neutrophil counts/lymphocyte counts [8]. The SII optimal cutoff point was 330. Hepatic fibrosis-related biomarkers The FIB-4 was an index, which can accurately predict fibrosis in patients coinfected with hepatitis C virus or human immunodeficiency virus. The overall accuracy was 87% [13]. The equation for FIB-4 was: NFS was constructed from laboratory variables and routine clinical and could accurately predict fibrosis in MASLD. The formulas included age, hyperglycemia, BMI, platelet count, albumin and AST/ALT ratio. The area under the receiver operating characteristic curve (ROC curve) was 0.82 [14]. The formulas for NFS were: HFS was constructed based on age, sex, diabetes, homeostatic model assessment (HOMA), AST, albumin and platelets. The area under the ROC curve was 0.85 [15]. The equation for HFS was: Hepatic steatosis biomarker detection FLI was used to predict liver fat and was simple to perform by body mass index (BMI), waist circumference (WC), triglycerides (TG) and gamma-glutamyl-transferase (GGT), which had an accuracy of 0.84 (95%CI 0.81–0.87) [16]. The formulas for FLI were: LFS was used to identify hepatic steatosis based on metabolic syndrome (MS), type 2 diabetes (T2DM), insulin, aspartate aminotransferase (AST) and the ratio of AST/alanine aminotransferase (ALT). The area under the receiver operating characteristic curve of LFS was 0.86 with proton magnetic resonance spectroscopy as the standard for diagnosing fatty liver [17]. The equation for LFS was: FSI was used to estimate patients with liver fat with a c-statistic value of 0.845. The FSI formulas were based on age, gender, BMI, TG, hypertension (HTN), hyperglycemia and ALT/AST [18]. The formulas for FSI were: Covariates The baseline characteristics included demographic factors (age, gender, education, marital status, ratio of family income to poverty (PIR) and race), behavior factors (drug, smoke, drink and physical activity (PA)) and health risk factors (HTN and T2DM). Statistical analysis All the participants were divided into healthy control group and SII group. Continuous and categorical baseline characters were presented as mean ± standard deviation (SD) and number (percentage), respectively. The distribution differences of variables were compared by weighted student’s t test for continuous data and weighted chi-square test for categorical data, respectively. The SII data was unevenly distribution and skewed to right. Based on this, the SII data need to be ln-transformed. The Pearson’s test was used to analyze the correlation between ln-SII and MASLD related biomarkers values. Three models were developed to assess associations of ln-SII and MASLD related biomarkers values by using weighted generalized linear model (GLM): crude model was not adjusted; model 1 adjusted for age, gender, education, marital status, PIR and race. model 2 adjusted for age, gender, education, marital status, PIR, race, drug, smoke, drink, physical activity, HTN and T2DM. Furthermore, the weighted restricted cubic splines (RCS) were used to explore the non-linear relationships between ln-SII and MASLD related biomarkers values. Finally, we performed sensitivity analyses to further validate the robustness of our results. the GLM were conducted to assess association of traditional inflammation marker ln-WBC and hepatic fibrosis biomarkers and hepatic steatosis biomarkers. All data were analyzed by using R version 4.2.2 and all the statistical significance was set P-value < 0.05 and all statistical test was two tails. Results Study population characteristics A total of 21,833 participants were included in this study (Fig. 1). After weighted, there were 2,015,554,456 subjects. Overall, a total of 1,568,281,782 subjects had a high level of SII. The age of mean ± SD for individuals with SII and healthy control were 46.83 ± 16.66 and 45.09 ± 16.94, respectively. Except for PIR and Drug, different distributions in the other selected variables were found between individuals with SII and healthy control (All P < 0.05). Whether the subjects with HTN was distributed differently between SII and healthy control (P < 0.001). Participants weighted and no weighted demographic factors, behavior factors and health risk factors were shown in Table 1 and Table S1, respectively. Flowchart of subjects selected from the NHANES 1999–2018 Table 1Weighted demographic characteristics of participants (n = 20451) in the NHANES 1999–2018VariablesTotal (N = 2015554456)Healthy control (N = 447272674)SII (N = 1568281782)t/c2 P Age (mean ± SD)46.45 ± 16.7445.09 ± 16.9446.83 ± 16.664.658<0.001Gender (%)66.248<0.001 Men971,664,879(48.21)245,800,566(54.96)725,864,313(46.28) Women1,043,889,577(51.79)201,472,108(45.04)842,417,469(53.72)Education (%)6.3310.013 High school or below828,596,365(41.11)174,284,177(38.97)654,312,187(41.72) College graduate or above1,186,958,091(58.89)272,988,497(61.03)913,969,594(58.28)Marital status (%)12.374<0.001 Married/Living with partner1,303,333,702(64.66)286,995,069(64.17)1,016,338,633(64.81) Never married343,833,355(17.06)66,148,055(14.79)277,685,301(17.71) Widowed/Divorced/Separated345,234,038(17.13)89,851,139(20.09)255,382,900(16.28)PIR (%)1.3840.254 <2637,661,336(31.64)146,611,896(32.78)491,049,440(31.31) 2–4566,662,080(28.11)122,935,507(27.49)443,726,573(28.29) >4678,909,954(33.68)147,994,835(33.09)530,915,119(33.85)Race (%)65.301<0.001 Mexican American166,457,513(8.26)38,484,880(8.6)127,972,632(8.16) Other Hispanic113,572,019(5.63)23,613,121(5.28)89,958,898(5.74) Non-Hispanic White1,384,000,248(68.67)263,652,878(58.95)1,120,347,370(71.44) Non-Hispanic Black215,326,154(10.68)82,549,318(18.46)132,776,837(8.47)Other Race136,198,522(6.76)38,972,477(8.71)97,226,045(6.2)Drug (%)0.8260.365 No999,466,032(49.59)224,618,703(50.22)774,847,329(49.41) Yes713,136,318(35.38)166,331,840(37.19)546,804,478(34.87)PA (%)42.535<0.001 Inactive842,368,918(41.79)164,863,472(36.86)677,505,446(43.2) Moderate494,552,596(24.54)121,788,888(27.23)372,763,708(23.77) Active416,970,957(20.69)119,662,167(26.75)297,308,790(18.96)Smoke (%)11.834<0.001 No1,087,542,971(53.96)263,001,982(58.8)824,540,988(52.58) Yes926,643,515(45.97)183,648,904(41.06)742,994,610(47.38)Drink (%)1.298<0.001 No429,467,843(21.31)93,436,135(20.89)336,031,707(21.43) Yes1,488,221,947(73.84)328,000,781(73.33)1,160,221,165(73.98)HTN (%)19.835<0.001 No1,197,775,212(59.43)283,864,142(63.47)913,911,070(58.27) Yes778,903,608(38.64)155,724,339(34.82)623,179,269(39.74)T2DM (%)4.170.043 No1,081,461,358(53.66)247,829,497(55.41)833,631,860(53.16) Yes933,963,420(46.34)199,443,177(44.59)734,520,243(46.84)ln-transformed SII (mean ± SD)6.17 ± 0.525.49 ± 0.336.36 ± 0.39//Hepatic fibrosis FIB-4 (mean ± SD)1.03 ± 0.751.27 ± 1.130.96 ± 0.58-12.183<0.001 NFS (mean ± SD)-0.51 ± 1.29-0.07 ± 1.22-0.63 ± 1.28-20.028<0.001 HFS (mean ± SD)0.77 ± 0.400.85 ± 0.340.75 ± 0.44-14.692<0.001Hepatic steatosis LFS (mean ± SD)-0.18 ± 2.65-0.38 ± 2.35-0.12 ± 2.735.649<0.001 FLI (mean ± SD)50.03 ± 32.7045.32 ± 32.6052.37 ± 32.619.283<0.001 FSI (mean ± SD)-0.92 ± 1.89-1.18 ± 1.87-0.84 ± 1.898.429<0.001 Weighted demographic characteristics of participants (n = 20451) in the NHANES 1999–2018 Associations of SII with MASLD related biomarkers values The Pearson’s correlation analysis was constructed to explore the correlation between ln-transformed SII with MASLD related biomarkers values. As shown in Fig. 2, there were negative correlation between ln-transformed SII and hepatic fibrosis biomarkers value (FIB-4 (r=-0.25), NFS (r=-0.26) and HFS (r=-0.09), all P < 0.001); there were positive correlation between ln-transformed SII and hepatic steatosis biomarkers value (FLI (r = 0.08), LFS (r = 0.04) and FSI (r = 0.07), all P < 0.001). Correlation analysis Pearson correlation coefficients between ln-transformed SII with MASLD related biomarkers values. The correlation coefficient was represented by numbers and colors. White was a positive correlation and black was a negative correlation. Circle bigger implied a stronger correlation. **indicated p < 0.01; ***indicated p < 0.001 Figure 3 suggested that ln-transformed SII were negatively related to hepatic fibrosis biomarkers value (FIB-4, NFS and HFS) in the three models (all P < 0.001). After adjusting for the potential confounding factors, the corresponding β (95%CI) between ln-transformed SII and hepatic fibrosis biomarkers value were − 0.35(-0.46, -0.24), -0.67(-0.71, -0.63) and − 0.10(-0.12, -0.09), respectively. Associations of ln-transformed SII and hepatic fibrosis biomarkers value were analyzed by using weighted generalized linear models. Crude model was not adjusted; Model 1 was adjusted for age, gender, education, marital status, PIR and race. Model 2 was adjusted for age, gender, education, marital status, PIR, race, drug, drink, smoke, PA, HTN and T2DM. The dots and lines were exhibited the β and corresponding 95% confidence interval of associations of SII with hepatic fibrosis Figure 4 exhibited that ln-transformed SII were positively related to hepatic steatosis biomarkers value (FLI, LFS and FSI) in the three models (all P < 0.001). After adjusting for the potential confounding factors, the corresponding β (95%CI) between ln-transformed SII and hepatic steatosis biomarkers value were 6.12(4.75, 7.50), 0.22(0.12, 0.31) and 0.27(0.20, 0.34), respectively. Associations of ln-transformed SII and hepatic steatosis biomarkers value were analyzed by using weighted generalized linear models. Crude model was not adjusted; Model 1 was adjusted for age, gender, education, marital status, PIR and race. Model 2 was adjusted for age, gender, education, marital status, PIR, race, drug, drink, smoke, PA, HTN and T2DM. The dots and lines were exhibited the β and corresponding 95% confidence interval of associations of SII with hepatic steatosis The non-linear association of SII with MASLD related biomarkers values The nonlinear relationship was explored between ln-transformed SII and β (the association of ln-transformed SII value with MASLD related biomarkers value) by RCS. Figures 5 and 6 showed statistically significant non-linear association was found in ln-transformed SII with β (all P < 0.001). With the ln-transformed SII increasing, the β (association of ln-transformed SII value with hepatic fibrosis biomarkers value) gradually decreases. However, with the ln-transformed SII increasing, the β (association of ln-transformed SII value with hepatic steatosis biomarkers value) gradually increases. RCS results of ln-transformed SII and β (the association of ln-transformed SII value with hepatic fibrosis biomarkers value). The result was adjusted for age, gender, education, marital status, PIR, race, drug, drink, smoke, PA, HTN and T2DM RCS results of ln-transformed SII and β (the association of ln-transformed SII value with steatosis fibrosis biomarkers value). The result was adjusted for age, gender, education, marital status, PIR, race, drug, drink, smoke, PA, HTN and T2DM Associations of WBC with MASLD related biomarkers values To test the robustness of our findings, we conducted sensitivity analyses (Table S2). The results suggested that ln-transformed WBC were negatively related to hepatic fibrosis biomarkers value (FIB-4, NFS and HFS) in the three models. After adjusting for the potential confounding factors, the corresponding β (95%CI) between ln-transformed WBC and hepatic fibrosis biomarkers value were − 0.33(-0.40,-0.26), -0.37(-0.45,-0.29) and − 0.02(-0.04,0.01), respectively. ln-transformed WBC were positively related to hepatic steatosis biomarkers value (FLI, LFS and FSI) in the three models (all P < 0.001). After adjusting for the potential confounding factors, the corresponding β (95%CI) between ln-transformed WBC and hepatic steatosis biomarkers value were 24.33(22.38,26.28), 1.24(1.08,1.40) and 1.14(1.04,1.24), respectively. Discussion In this study, Pearson’s correlation analysis, GLM and RCS were constructed to reveal the associations between SII and MASLD related biomarkers values. There was negative significant association between SII and hepatic fibrosis biomarkers value (FIB-4, NFS and HFS). In contrast, there was positive significant association between SII and hepatic steatosis biomarkers value (FLI, LFS and FSI). SII, as a novel biomarker, can be easily obtained from routine complete blood count tests (platelet counts, neutrophil, and lymphocyte) and reflect the overall status of the immune response and systemic inflammation of the subjects. In the study, SII was constructed to predict the recurrence and survival for hepatocellular carcinoma patients [8]. In the recent years, more and more studies use SII to explore the association of systemic inflammation with the subjects’ healthy status (lupus nephritis [23], psoriasis [24], arthritis [25] and MASLD [21, 26, 27]). One study, based on NHANES, revealed that ln-transformed SII was an independent risk factor for MASLD risk (defined by a US Fatty Liver Index (USFLI) score exceeding 30), and the corresponding OR and 95%CI was 1.46 (1.27, 1.69) [26]. Another study also found that there was a significant positive association between SII and hepatic steatosis index (HSI) and OR (95%CI) was 1.30 (1.10, 1.52) [21]. Although above studies used the different biomarkers to detect the hepatic steatosis, the result also indicted that there was a positive association between SII and hepatic steatosis, which were similar with our study. Besides the association between SII and hepatic steatosis, two studies also explored the relationship between SII and hepatic fibrosis [22, 28]. The first study implied that there was no link between SII and liver fibrosis (based on liver stiffness measurement), and the corresponding β (95%CI) and P value were 0.000 (-0.000, 0.001) and 0.263 [22]. While the second research suggested that there was a negative association between SII and advanced liver fibrosis (according to FibroScan model) in the individuals with diabetes mellitus, and the OR (95%CI) was 0.33 (0.15, 0.70) [28]. In this study, there was negative significant association between SII and hepatic fibrosis (FIB-4, NFS and HFS). The inconsistencies in the results of these studies may stem from differences in the included population and the biological markers of liver fibrosis. The underling mechanism of relationship between SII and hepatic steatosis was not yet completed understood. Evidences suggested that immune-inflammation response played an important role in hepatic steatosis. Cell death caused by inflammasome pathways played a key role in the pathogenesis of MASLD, and interleukin (IL)-1 and IL-18 released after cell death can cause a strong inflammatory process and induce the recruitment of inflammatory cells [29, 30]. Another theory was that nutrient overload was the primary cause of hepatic steatosis. The excess visceral fat could cause macrophage infiltration and result in a pro-inflammatory state, which induced insulin resistance. Inappropriate lipolysis resulted in aberrant fatty acid transport to the hepatic cell in the state of insulin resistance, which could induce a decrease in metabolic capacity. Lipotoxic lipids were formed as a result of lipid metabolic imbalances that lead to cellular stress and inflammasome activation [6, 31]. The mechanism underlying the relationship between SII and hepatic fibrosis remained unclear. Based on previous describe, we postulated that SII may play a pivotal role in both the initiation and progression of hepatic steatosis. However, as the disease progresses toward hepatic fibrosis, there was a discernible decline in complete blood count, predominantly lymphocytes. This could be attributed to a series of pathophysiological evolutions, such as hypersplenism, bone marrow haematopoietic inhibition, and endocrine shifts within the organism. Taken together, these observations suggested that lymphocytes, as a component of the SII formula, have different developmental trends in hepatic steatosis and hepatic fibrosis stages, and may explain the phenomenon of SII different trends in hepatic steatosis and hepatic fibrosis [28]. It’s worth noting that these biomarkers relied on non-invasive and could be easily obtained in clinical examination. Although, Liver biopsy was the gold standard for diagnosis, its sampling errors, inherent risks of complications, and invasive nature restrict its use in clinical practice [32]. these biomarkers offer possibilities for screening or monitoring MASLD at a non-invasive and low financial burden level. It was important to emphasize some limitations of this study. Firstly, the NHANES study was a cross-sectional design, which limits its ability to make established cau sal conclusions. Secondly, even with consideration of several influencing factors, the potential impact of hidden variables was a concern. Thirdly, due to limited information in the data set, the exclusion of subjects with incomplete data may introduce bias into the analysis. Fourthly, the current findings may not be persuasive for other racial groups, as non-Hispanic whites predominate among NHANES. Fifthly, this study relies on non-invasive biomarkers. While these indices are valuable, the lack of data from more direct diagnostic tools such as FibroScan for fibrosis evaluation and abdominal ultrasound or transient elastography for steatosis evaluation limits the robustness of the findings. Sixthly, in the exclusion criteria, we just excluded subjects infected with HBV and HCV. Other liver disease, rheumatoid arthritis and systemic lupus erythematosus weren’t excluded in this study, which might influence the robustness of our results. In addition, utilizing the NHANES database inevitably introduces the possibility of imprecise data capture and recall bias. Although these limitations existed in this study, to some extent, the relatively large epidemiological study also could reflect the associations between SII and hepatic fibrosis or hepatic steatosis. Conclusion The results reveal there was a negative significant association between SII and hepatic fibrosis, while a positive significant association between SII and hepatic steatosis. SII was a cost-effective, credible, and easily obtained from routine complete blood count tests for discerning the presence of hepatic steatosis or hepatic fibrosis. Moreover, more extensive prospective studies are required for comprehensive validation. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements The authors thank all the staff and participants in NHANES for their contribution of donation, collection and sharing data. Abbreviations Alanine aminotransferase Aspartate aminotransferase Body mass index Fibrosis-4 index Fatty liver index Framingham steatosis index Hepamet fibrosis score Homeostatic model assessment Hypertension Gamma-glutamyl-transferase Generalized linear model Liver fat score Syndrome Nonalcoholic fatty liver disease NAFLD fibrosis score National Health and Nutrition Examination Survey National Center for Health Statistics Physical activity Ratio of family income to poverty Restricted cubic splines Receiver operating characteristic curve Standard deviation Systemic immune inflammation index Triglycerides Type 2 diabetes Waist circumference Author contributions Shunyin Duan: Writing – review & editing, Writing – original draft, Data curation, Formal analysis, Conceptualization. Zhanwen Tu: Writing – review & editing, Formal analysis, Data curation. Lijuan Duan: Writing – review & editing. Runqi Tu: Writing – review & editing, Data curation, Formal analysis, Conceptualization. Funding This study was not supported by any funding. Data availability The National Center for Health Statistics Research Ethics Evaluation Committee assessed and approved the National Health and Nutrition Examination Survey (NHANES) data. Additionally, each participant gave their informed permission. The survey data are openly accessible online, providing a valuable resource for data users and researchers globally (www.cdc.gov/nchs/nhanes/). Declarations Ethics approval and consent to participate The National Center for Health Statistics (NCHS) and the Centers for Disease Control and Prevention (CDC) conducted the NHANES survey. The National Center for Health Statistics’ Research Ethics Assessment Board evaluated and approved the NHANES study protocol. This study used a de-identified publicly available dataset that did not involve direct human experimentation or further interventions, and no additional ethical approvals were required for this type of research. For further confirmation, please refer to the link to the NCHS ethics approval document for the NHANES data (www.cdc.gov/nchs/nhanes/irba98.htm). Conflict for Publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Shuyin Duan and Zhanwen Tu contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The National Center for Health Statistics Research Ethics Evaluation Committee assessed and approved the National Health and Nutrition Examination Survey (NHANES) data. Additionally, each participant gave their informed permission. The survey data are openly accessible online, providing a valuable resource for data users and researchers globally (www.cdc.gov/nchs/nhanes/).