Prevalence and Trends in Low Bone Density, Osteopenia and Osteoporosis in U.S. Adults With Non-Alcoholic Fatty Liver Disease, 2005–2014

Background & Aims Non-alcoholic fatty liver disease (NAFLD) is suggested to be associated with bone mineral density (BMD) alterations; however, this has not been ascertained. The current study aimed to investigate the changes in BMD and the prevalence of osteopenia/osteoporosis in US adults with or without NAFLD and to evaluate their association. Methods The study was conducted based on data collected from the U.S. National Health and Nutrition Examination Survey (NHANES) during the period 2005–2014. A total of 13 837 and 6 177 participants aged > 20 years were eligible for conducting the Hepatic Steatosis Index (HSI) and the US Fatty Liver Index (USFLI) analysis, respectively. Results From 2005–2014, a downward trend in femoral neck BMD was observed in subjects with NAFLD aged ≥ 40. After adjustment for potential confounders, an upward shift occurred in the prevalence of osteopenia/osteoporosis at the femoral neck in adults aged ≥ 40, particularly in women ≥ 60 years old and men below the age of 60. Moreover, a negative association was found between BMD and NAFLD markers (USFLI, HSI), whereas NAFLD with advanced fibrosis was positively associated with the prevalence of spine fractures. Conclusions There was a trend toward lower BMD and higher prevalence of osteopenia/osteoporosis at the femoral neck in US adults with NAFLD aged ≥ 40 years during the period of 2005–2014. NAFLD with advanced fibrosis was positively associated with a higher risk of spine fracture. More research is required to fully investigate the mechanism and consequence of poor bone health in NAFLD patients and consider optimum management of osteopenia/osteoporosis for this population.


INTRODUCTION
Osteoporosis is a systemic skeletal disease characterized by low bone mass and deterioration of architecture, associated with excessive bone fragility and fracture risk (1). It is estimated that 13 million individuals >50 years old are expected to suffer from osteoporosis by 2020 in US (2). During the period 1988-1994 to [2005][2006], a decline in the prevalence of osteoporosis was observed in the US population aged ≥ 50 years (3). According to the US Medicare data, a decrease in the rate of hip fractures was observed from 2002-2012, which plateaued during the period of 2013-2015 (4). In addition, by directly assessing the bone mineral density (BMD) data at various skeletal sites, more recent studies have confirmed a decline in femoral neck BMD in the US population from 2013-2014 (5).
Non-alcoholic fatty liver disease (NAFLD) has emerged as an noninfectious epidemic that threatens the health of~24% in the global population, paralleling the increase of other metabolic conditions, including type 2 diabetes (T2DM), obesity, hyperlipidemia, hypertension, and metabolic syndrome (6). Osteoporosis accounts for most cases of hepatic osteodystrophy, which is defined as a skeletal disorder of multifactorial origin caused by chronic liver disease (7). As a chronic liver disease, NAFLD and the advanced histological phenotype non-alcoholic steatohepatitis (NASH), may have a potential link with osteoporosis, which has piqued considerable scientific interest (8).
Several cross-sectional studies have attempted to elucidate the association between NAFLD and BMD, osteoporosis, and relevant fractures (9)(10)(11)(12)(13)(14)(15)(16)(17). Not surprisingly, there are many conflicting findings in these studies, given that various factors including age, gender, race, and the nutritional and menstrual status may affect bone metabolism. Some studies have identified a significant association of lower BMD or higher rates of osteoporotic fracture with NAFLD in both adolescents and adults (9)(10)(11)(12)(13)15), whereas other studies reported no significant associations and even opposite results (14,16,17). In addition, two recent meta-analyses also observed no significant association between fatty liver disease and decreased BMD (18,19). Moreover, there is no evidence showing temporal trends of BMD and prevalence of osteopenia/osteoporosis in the individuals with NAFLD. Therefore, in the present study, by accessing the nationally representative data from the US National Health and Nutrition Examination Survey (NHANES) in the period of 2005-2014, we aimed to evaluate the temporal trends of BMD, osteopenia/osteoporosis in the US adults with or without NAFLD, and their possible associations.

Study Population
The cross-sectional analysis was conducted based on the 2005-2014 period data of NHANES that includes the multistage, stratified, clustered probability samples and represent the noninstitutionalized US population. NHANES obtained written informed consent from all individuals, which was approved by the National Center for Health Statistics institutional review board. The socioeconomic, demographic, medical, and dietary data were collected through health-related interviews, laboratory and physical evaluations.

Study Design
Of all 41,209 participants in the 2005-2010 and 2013-2014 period in National Health and Nutrition Examination Survey (NHANES) 22 901 (55.9%) participants were > 20 years old. We excluded the participants with significant alcohol consumption (> 2 drinks/day for men or > 1 drink/day for women, n = 1,451), viral hepatitis (positive serum hepatitis B surface antigen and positive serum hepatitis C antibody, n = 421), pregnancy (n = 507), steatogenic medication (methotrexate, corticosteroids, valproate, amiodarone, and tamoxifen, n = 338) > 6 months, and individuals with missing data (BMD, Body mass index [BMI], serum aminotransferase, platelet count, n = 6,347). Additionally, 6,273 participants were lacking data regarding fasting blood glucose tests. Hence, the final cohort comprised 13,837 participants included in the subsequent analysis with Hepatic Steatosis Index (HSI), and 6,177 participants evaluated using the US Fatty Liver Index (USFLI) (Figure 1).
The data of BMD measurement of the proximal femur and lumbar spine are available from 2005-2014 except for 2011-2012. Additionally, only the participants aged ≥ 40 years have the BMD data for the 2013-2014 period. To reduce the variance of the estimates, the ages are categorized as 20-39, 40-59, and ≥ 60 years old. Therefore, we use the age of 40 as the limit in the main analysis and use ages 40 and 60 in the subgroup analysis stratified by age and sex. In the women aged ≥ 40 years, there are 1522 (64.6%) and 3373 (63.8%) postmenopausal women in USFLI and HSI analysis, respectively.

Clinical and Laboratory Characteristics
The data of each participant was collected using household interviews, physical measurements, and laboratory tests in mobile examination centers (MECs). The information of age, sex, race/ethnicity, education level, economic status, milk intake, and history of fracture were collected using a standardized questionnaire. Race/ethnicity included Hispanic (Mexican American, other-Hispanic), non-Hispanic white and black, and other. Education level was based on whether the individual graduated from high school. Marital status was defined as married or cohabiting with a partner versus other. We defined current smokers as individuals who have smoked ≥ 100 cigarettes in their lifetime and are still cigarette users. We defined individuals with hypertension as having systolic blood pressure (SBP) > 140 mmHg, or diastolic blood pressure (DBP) > 90 mmHg and/or currently taking antihypertensive drugs. We defined the individuals with diabetes mellitus as having fasting plasma glucose (FPG) ≥ 126 mg/dL and/or currently receiving insulin or other hypoglycemic agent treatment. The family income-to-poverty ratio was used to evaluate the individual's socioeconomic status: ≥ 1.00 = poverty, < 1.00 = below poverty level. Homoeostasis model assessment of insulin resistance (HOMA-IR) was used to reflect insulin resistance (IR) (20). Milk intake was defined as never, rarely (< once a week), sometimes (< once a day and > once a week), often (> once a day), and varied. BMI was calculated as weight (kg) divided by the square of height (m). The metabolic equivalent (MET) data were collected using the Global Physical Activity Questionnaire from 2007-2008 and was used to evaluate the energy expenditure of individuals in one week (21).
The radioimmunoassay (DiaSorin) method was used to measure total 25-

Definition of NAFLD and Advanced Fibrosis
For the definition of NAFLD, previously validated methods, including US Fatty Liver Index (USFLI)and Hepatic Steatosis Index (HSI), were used in epidemiologic studies in the absence of other causes of chronic liver disease and exposure to steatogenic medication. We used the published cut-off values of ≥ 30 and 36 for USFLI and HSI, respectively, to define the presence of NAFLD (22). Advanced fibrosis was evaluated according to a previous study (23), i.e., NAFLD fibrosis score (NFS) > 0.676 (high probability), -1.455 ≤ NFS ≤ 0.676 (intermediate probability), and NFS<-1.455 (low probability). We used the high probability of NFS score to define advanced fibrosis. The equations for calculation of the aforementioned markers are outlined in Supplementary Table 1.  period, the femur scans were analyzed with Hologic Discovery v12.4, and spine scans were analyzed with APEX v3.0. APEX v4.0 was used for the analysis of both femur and spine scans in 2013-2014. A cross-calibration procedure was conducted to standardize the newer system to the legacy system by the Hologic Service Team (21). After the assessment, no difference was found between mean BMD at five femur regions analyzed by Discovery v12.4 and APEX v4.0 (24).

BMD Measurement and Definition of Osteopenia/Osteoporosis
Osteopenia was defined as -2.5 < T-score ≤ -1.0, and osteoporosis was defined as T-score ≤ -2.5. T-score was calculated using the formula: T-score= (BMD repondent −mean BMD reference group )/SD reference group . As recommended by the World Health Organization (25), non-Hispanic white females aged 20-29 from the NHANES III report were included as the reference group for the T-score calculation of the femoral neck (26), and the reference group for the lumbar spine was obtained from the Vital and Health Statistics from the CDC (27).

Definition of NAFLD and Advanced Fibrosis
For the definition of NAFLD, previously validated methods, including US Fatty Liver Index (USFLI)and Hepatic Steatosis Index (HSI), were used in epidemiologic studies in the absence of other causes of chronic liver disease and exposure to steatogenic medication. We used the published cut-off values of ≥ 30 and 36 for USFLI and HSI, respectively, to define the presence of NAFLD (22,28). Advanced fibrosis was evaluated according to a previous study (23), i.e., NAFLD fibrosis score (NFS) > 0.676 (high probability), -1.455 ≤ NFS ≤ 0.676 (intermediate probability), and NFS<-1.455 (low probability). We used the high probability of NFS score to define advanced fibrosis. The equations for calculation of the aforementioned markers are outlined in Supplementary Table 1.

Statistical Analysis
NHANES adopted the complex sampling design and provided representative data for the civilian, noninstitutionalized US population, which requires appropriate sample weight to adjust for the potential bias of non-response and the unequal probability of inclusion in the survey. Based on the principle of using the smallest subsample weight, we adopted the mobile examination center (MEC) exam weight and the fasting subsample weight in the current analysis. We calculated frequencies (standard errors, SE) for categorical variables and the means ± SE for continuous variables. We investigated the temporal prevalence of osteopenia/osteoporosis through the 2005-2014 period stratified by sex and age as recommended by the Centers for Disease Control and Prevention (CDC). Test for trends was calculated through a linear regression model by including the midpoint of each survey cycle as a continuous variable. Relative SE > 0.30 indicated low precision of the estimate.
Using the NHANES cycle of 2005-2006 as reference, we calculated the odds ratios (ORs) of osteopenia/osteoporosis over time via logistic regression analysis. Model 1 was adjusted for the sociodemographic variables, including sex, age, race, BMI, waist circumference, smoking, educational, marital, and economic status. Model 2 was adjusted for the adjustments of model 1 in addition to nutritional status, 25(OH)D, and milk intake. Model 3 was adjusted for hypertension, diabetes, high-density lipoprotein cholesterol (HDL-C), triglyceride (TG), total cholesterol (TC), and low-density lipoprotein (LDL-C) in addition to model 2. Model 4 was adjusted for the adjustments of model 3 plus menopausal status. Model 5 was further adjusted for physical activity.
Further, we evaluated the bone quality difference between controls and individuals with NAFLD or that with advanced fibrosis. We calculated the prevalence of osteopenia/osteoporosis and fracture, and investigated the association between NAFLD (with advanced fibrosis) and the odds of osteopenia/osteoporosis and fracture using logistic regression after adjustment for the significant factors as described above. Additionally, the association between NAFLD (with advanced fibrosis) and BMD was assessed by multivariate regression analysis in the adjusted models as described above. The NALFD or fibrosis markers (USFLI, HSI, and NFS) were taken as the independent variables, and femoral neck or lumbar spine BMD taken as the dependent variables. All analyses were conducted using SPSS version 24.0 Complex Survey module (IBM Corporation, Armonk, NY, USA). P<0.05 was considered statistically significant.

Characteristics of Participants With or Without NAFLD From the 2005-2014 Period
Defined by USFLI, participants aged > 40 years with NAFLD exhibited a significant increasing trend with BMI, waist circumference, HOMA-IR, 25(OH)D, prevalence of diabetes and poverty, and a significant decreasing trend in the levels of TC, TG, LDL-C, prevalence of current smoker and marriage across the four NHANES cycle. Additionally, some variables, including age, sex, race, education level, proportions of hypertension, HDL-C level, and physical activity remained comparable from 2005-2014. In the participants aged ≥ 40 years old without NAFLD, compared with participants with NAFLD, the proportion of men and prevalence of hypertension significantly increased, whereas BMI, HOMA-IR, diabetes, and marriage were stable from 2005-2014 (Table 1). There were generally similar trends when NAFLD was defined using HSI (Supplementary Table 2).
Remarkably, a significant decreasing trend of BMD was observed at the femoral neck in participants with NAFLD aged ≥ 40 years old throughout the decade. The BMD at the lumbar spine did not change, nor there was a trend. These findings were consistent when NAFLD was defined using HSI ( Table 1 and  Supplementary Table 2).
Besides, in the 20-39 age group of participants with or without NAFLD, significantly decreasing trends of marriage and amount of physical activity were observed from 2005-2014. However, the trend concerning femoral neck and lumbar spine BMD remained stable across the three cycles (Supplementary Table 3).

Bone Status of Participants With or Without NAFLD Stratified by Age and Sex
Throughout the survey cycles, a significant declining trend of femoral neck BMD was observed in men aged 40-59 years old and women aged > 60 with NAFLD defined by both USFLI and HSI. Accordingly, the prevalence of osteopenia and osteoporosis at the femoral neck showed an increasing trend in the two groups ( Figure 2 and Supplementary Figure 1). Additionally, in the analysis of participants with NAFLD defined by HSI, we observed a significant declining trend of BMD and an increasing trend in the prevalence of osteopenia and osteoporosis at the femoral neck in men aged > 60 years old and women aged 40-59 across the cycles.
Among the participants without NAFLD from 2005-2014, there was a significantly elevated trend of the prevalence of femoral neck osteopenia and osteoporosis in men aged 40-59 defined by either USFLI or HSI (Figure 2 and Supplementary    Figure 1). Meanwhile, women aged over 60 without NAFLD defined by USFLI had a significantly elevated trend of the prevalence of osteopenia and osteoporosis in lumbar spine through the decade ( Figure 2). Furthermore, in the subgroup of participants from 20-39 years old, men without NAFLD defined by the two markers showed significantly growing trends in the prevalence of osteopenia and osteoporosis in both sites from 2005-2010, and the rising trend of this index was also observed in the femoral neck in women without NAFLD defined by USFLI (Figure 2 and Supplementary Figure 1).

Association Between the Prevalence of Osteopenia/Osteoporosis and the Survey Cycle in the Participants With or Without NAFLD
In participants aged ≥ 40 years with and without NAFLD defined by USFLI, after adjustment for sociodemographic characteristics, BMI, waist circumstance, nutritional and menopausal status, NAFLDrelated blood biochemical indices, and other metabolic disorders 1.59, 95% CI: 1.11-2.26; p = 0.004, respectively). This significance was not altered after further adjustment for physical activity (MET score) (model 5; p < 0.001 and p = 0.008, respectively) ( Table 2). As for the lumbar spine, no significant cycle trend for the prevalence of osteopenia and osteoporosis was observed in the final adjusted  Table 2).
Similar results were observed in the analysis when NAFLD was defined by HSI, as shown in Supplementary Table 4.

The Association Between Bone Status and NAFLD or NAFLD With Advanced Fibrosis
Compared with controls, the femoral neck and lumbar spine BMD were significantly greater in participants with NAFLD or NAFLD with advanced fibrosis (Supplementary Table 5). In participants aged over 40, BMD in both sites exhibited a significantly negative association with USFLI in the final adjusted model (model 5, b = -0.0004, 95% CI: -0.0008 to -0.0001; p = 0.028, and b = -0.0006, 95% CI: -0.0011 to -0.0001; p = 0.033). Similar results were observed in the analysis when NAFLD was defined by HSI. A significantly negative association was also observed between NFS and BMD of the femoral neck, but not lumbar spine (b = -0.0070, 95% CI -0.0123 to -0.0015, p = 0.012, and b = 0.0046, 95% CI -0.0034 to 0.0126, p = 0.262) ( Table 3).
Besides, although no significant association was observed between NAFLD and hip or spine fracture in the participants aged ≥ 40 years (model 5, OR: 0.66, 95% CI: 0.12-3.61; p = 0.629, and OR 0.65, 95% CI 0.23-1.79; p = 0.400), the association was  Table 4). Similar results were observed in the analysis when NAFLD was defined by HSI (Supplementary Table 6).   results further uncovered a declining trend of BMD in subjects with NAFLD in the US. This could have a significant clinical implication, given the fact that the prevalence of NAFLD reached 31.1% of NHANES participants in the 2015-2016 period. It is noted that the trend of bone loss in the femoral neck appears more obvious than that in lumbar spine. The inconsistent trend at the two bone sites may be attributable to the different bone loss patterns (29). Another possible reason is that the measurement of lumbar spine BMD could be confounded by certain factors, including osteophytes and aortic calcification (30). Apparently, in the subgroup analysis, there was a more significant downward trend of femoral neck BMD loss in men aged 40-59 years and women aged > 60 with NAFLD defined by USFLI. Accordingly, the two age groups had a growing trend of the prevalence of osteopenia and osteoporosis at the femoral neck. Hip fractures commonly cause more disability than fractures at other sites and are closely related to low BMD at the femoral neck (31). Indeed, existing evidence indicates that BMD at the femoral neck has the strongest ability to predict hip fracture (32). Another remarkable observation from the current study is that the prevalence of osteopenia/osteoporosis in younger men aged 20-39 is also increasing, not just in older adults. Nonetheless, there is the misconception that osteoporosis only occurs in the elderly or postmenopausal women. Currently, the majority of men at high risk of osteoporosis or fracture remain underappreciated and undertreated (33). Compared with women, there is a lack of long-term and powered studies of osteoporosis in men. Educational programs should be conducted not only in postmenopausal women but also in young and middle-aged men to prevent bone-related disorders, particularly in individuals with NAFLD.

DISCUSSION
In order to determine the possible factors that truly affect this upward shift of osteopenia/osteoporosis at the femoral neck, we investigated changes in sociodemographic characteristics, BMI, waist circumstance, nutritional status, NAFLD-related blood biochemical indices, lifestyle-related factors, and other metabolic disorders between survey cycles. After adjustment for the variables in model 4, we did not observe a marked alteration in the trend, which indicated that these variables perhaps do not play a primary role in the shift of bone status at the femoral neck. One important factor is the amount of physical activity, given that a sedentary lifestyle is associated with osteoporosis and decreased BMD(2004), and subjects with NAFLD are less likely to engage in physical activity and spend more sedentary time than controls. In our analysis, most notably, the MET scores decreased from 3716 and 3724 to 3183 min/week among individuals with NAFLD defined by USFLI and from 4583 to 4066 to 3645 min/week among those with NAFLD defined by HSI through the periods; adjustment for physical activity rendered the cycle trends of prevalence of lumbar spine osteopenia and osteoporosis into marginally insignificant in participants with NAFLD. Hence, the decreased MET scores across the survey periods may explain the decreasing BMD trend.
Over the past decades, there has been a growing awareness of the association between bone health and NAFLD.
Nonetheless, previous studies have reported conflicting results that NALFD is associated with lower, similar, or higher BMD compared with healthy controls. Various factors may interfere with the findings, such as age (children, adolescent, adult), sex (male, female), race (Hispanic, Non-Hispanic, Asian, others), menstrual status, measurement technology of BMD, diagnosis of NAFLD, sample size and other unknown confounding factors. In our current analysis, compared with participants without NAFLD, we found a higher BMD and a lower prevalence of osteopenia/osteoporosis both at the femoral neck and lumbar spine in individuals with NAFLD. It is possible that BMI has the most significant and strongest effect on the BMD difference (18). The literature has revealed that BMD is positively associated with BMI (34), and greater BMD was observed in obesity, regardless of the bone site and measurement methodology (35). The excessive body fat mass and lean mass lead to increased passive loading, and musclerelated strain, so enhanced formation of cortical bone in obesity may help explain these findings (36). On the other hand, the chronic inflammation caused by adiposity might be detrimental to bone health. The overproduction of several proinflammatory cytokines, including interleukin (IL)-1 (IL-1) and IL-6, encourage bone resorption and could be induced by excessive body fat mass because adipose tissue is a major source of these cytokines (37,38). Another possible mechanism is that obesity is often associated with a lower level of adiponectin, an endocrine factor that may cause lower BMD through inhibiting osteoprotegerin expression in osteoblasts (39). Interestingly, the observed negative association between NAFLD markers (USFLI, HSI) and BMD markedly challenges the conventional concept that NAFLD may prevent bone loss because of excessive body weight. In fact, accumulating evidence has suggested that the alterations in the production of several molecular coordinators caused by NAFLD may be detrimental to skeletal health, such as an excessive production of TNF-a (40), and deficiency of osteopontin (41), osteocalcin (42), and osteoprotegerin (43). Additionally, excess intrahepatic lipids could impair insulin sensitivity in NAFLD, which may account for the loss of bone mass (40).
A previous study has shown that with the increased incidence of NAFLD, the prevalence of its related advanced fibrosis increased steadily from 2015-2016 (44). NASH has been reported to accompany by a worse bone condition compared with simple steatosis (12). In the current study, we found that subjects with NAFLD-associated advanced fibrosis have a higher prevalence of fractures compared with controls, although the prevalence of osteopenia/osteoporosis was comparable between the groups. BMD was higher in subjects with NAFLD-related advanced fibrosis and increased with BMI or body weight, whereas the bone strength did not commensurately increase with BMI, and the load-to-strength ratio rose more quickly (45). In addition to bone factors, poorer muscle function caused by NAFLD may also be associated with increased falls and higher risk of fracture (46,47). Therefore, subjects with NAFLD-related advanced fibrosis may be more likely to fall and be more vulnerable to fracture.
Of significance, our findings suggest that earlier screening and therapeutic intervention for NAFLD could be benefits in improving bone health. It is well-recognized that lifestyle interventions, including diet and exercise, are an effective primary therapy for NAFLD. Reduction in liver fat and improvement of body composition and cardiac function were observed in NAFLD patients who performed appropriate exercise (48). Recent studies have also reported significant beneficial effects on both metabolic and skeletal health by football training in elder individuals with prediabetes (49). Likely, the benefits from exercise on bone status in patients with NAFLD would be highly expected, whereas the evidence is still lacking. Further evidence is also required to determine which form of exercise (resistance, aerobic, or high-intensity intermittent exercise) is suitable for NAFLD patients with poor bone conditions. Additionally, we observed that vitamin D levels appear to be lower in NAFLD patients compared with controls, which is consistent with previous studies (50). At present, the intake of vitamin D from food alone in adults is far from the Dietary Reference Intake (51). Interestingly, Jennings et al. recently reported that the Mediterranean diet, one of the most recommended dietary patterns for NAFLD, together with vitamin D supplements, could prevent bone loss at the femoral neck in patients with osteoporosis (52), suggesting a practicable way to the prevention of osteoporosis for NAFLD patients.
NHANES is a series of elaborately conducted surveys with comprehensive data coverage and consistent data acquisition methods over the period of 2005-2014. Such a set of timely and comprehensive data has ensured the reliability of the current study. However, some limitations exist. First, ultrasonography is most used generally for the diagnosis of NAFLD, which was not available in NHANES surveys. The actual prevalence of NAFLD and its related advanced fibrosis may be underestimated in the U.S. population by using noninvasive panels (USFLI, HSI, and NFS). Second, NHANES did not include longitudinal follow-up data, and thus the nature of our cross-sectional study is unable to explore causality. Third, although the subjects with viral hepatitis, significant consumption of alcohol, and long-term steatogenic medication were excluded from this study, we are unable to exclude other unknown causes, such as primary biliary cholangitis, hemochromatosis, autoimmune hepatitis, and primary sclerosing cholangitis. Fourth, the true prevalence of osteopenia/osteoporosis may be underestimated in the general U.S. population, since noninstitutionalized individuals tend to have higher BMD (53). Fifth, in our analysis, osteoporosis was diagnosed in subject younger than 50 years old owing to the NHANES recommendations about the age categories, which may differ from the ISCD's (The International Society For Clinical Densitometry) definition of osteoporosis. Finally, although we have adjusted for changes in a variety of demographic and clinical factors in the analysis, the possibility of residual confounders cannot be fully excluded.
In conclusion, the current study reveals the recent downward trend of BMD and the upward trend of osteopenia/osteoporosis in US adults, particularly in individuals with NAFLD, in the period of 2005-2014. It is noted that while the comorbidities of NAFLD including T2D, cardiovascular disease and chronic kidney disease have received considerable attention (54), the problems in bone health have been overlooked in NAFLD patients. Our study suggests more efforts urgently required to fully investigate the complex and interconnected relationship between NAFLD and bone status, and to devise feasible strategies for screening and therapeutic interventions for the individuals with both NAFLD and low BMD.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by National Center for Health Statistics. NHANES obtained written informed consent from all patients/participants, which was approved by the National Center for Health Statistics institutional review board.