| |
Does the liver accelerate ageing: Talking muscles and liver?
|
| |
| |
Download the PDF here
Download the PDF here
From Jules: inflammation from the liver & to the liver I think contributes to the accelerated fibrosis we see in HIV but this I don't think was ever studied.
The paper from Koo and colleagues provides a unique insight into the relationship between age-related loss of muscle mass, metabolic control and liver disease. We have known the relationship between metabolism, abdominal adiposity and liver health for some time [15]. Although previous reports have shown that sarcopenia relates to NAFLD, it is the extension of this to NASH, independently of insulin resistance, which is of most interest. There are several important questions that remain unanswered; which comes first - sarcopenia or NASH? Are sarcopenia and NASH both organ-level presentations of a more systematic factor (transplanted NASH patients develop NASH again, suggesting that there is something beyond the liver influencing fibrosis). Indeed, obesity affects liver by increasing inflammation and tumourgenesis via elevated IL-6 and TNFα [16].
Pro-inflammatory cytokines can drive cellular senescence and limit regenerative capacity in regions around it. This cellular cross-talk could be a 'butterfly effect' of cellular senescence. In this context, NASH could be viewed as the livers response to accelerated ageing. Finally, loss of muscle mass has been known for decades to be common in advanced liver disease, namely in cirrhosis [17], with a negative effect on disease progression [18], and sarcopenia improves the prediction of mortality independent of etiology [19]. In the study of Koo et al., sarcopenia was associated with severe fibrosis (≥F2) [6], which raises the possibility of an independent effect of poor liver function. However, cirrhosis was rare in the population (only 7% of cases), and sarcopenia was present throughout the histological spectrum of NAFLD. It is very likely that the altered metabolic milieu of NAFLD (obesity and diabetes) may significantly contribute to accelerate the loss of muscle mass.
Jnl of Hepatologt 2017 January
Despite what the general media would like us to believe, we live in a time of unparalleled health. People are living longer; the increase in longevity has been driven by a decline in early life mortality as a result of improved hygiene and nutrition in the early 20th century, with the advent of new drugs delivering a decline in late life mortality in the 21st century [1]. As a consequence of our success in ageing, we are now facing a new problem, how to live well as well as longer.
Ageing is not programmed but results from a gradual, lifelong accumulation of damage to cells and tissues of the body over time. Consequently, there is no single cause for ageing, but spans (not exclusively) the accumulation of mitochondrial DNA mutations, aberrant epigenetic markings, nuclear genome instability and telomere erosion caused by chronic inflammation, metabolic stress and oxidative stress/redox imbalance amongst others [1]. This accumulation of cellular damage drives the ageing process and results in the development of frailty and chronic age-related diseases. Crucially, however, we know that many of the factors involved in the ageing process can be attenuated by lifestyle factors and accelerated with poor lifestyle choices [2] and also with some classes of drugs (such as chemotherapy [3] or anti-retroviral drugs [4]). This creates a disconnect between an individual's biological age and their chronological age - some people age quicker and some people age slower.
Ageing is characterized by loss of muscle, which can ultimately result in a loss of physical function and is then termed sarcopenia. Despite our best efforts the loss of muscle mass with age cannot be prevented, but loss of muscle is accelerated with loss of insulin sensitivity and poor glucose control, which is not surprising as insulin itself has a powerful effect on muscle growth. Our most recent understanding of muscle suggests that muscle peptides and cytokines released from muscle also have endocrine and paracrine effects [5]. As such, a decline in muscle mass not only has an effect on physical function, the loss of muscle mass can directly affect other physiological systems in the body (and indeed, vice versa).
What is new?
The paper from Koo and colleagues in the Journal of Hepatology describes the prevalence of sarcopenia in 309 people with biopsy proven healthy liver, non-alcoholic fatty liver disease (NAFLD) or non-alcoholic steatohepatitis (NASH) [6]. They conclude that as liver disease worsens, the prevalence of sarcopenia rises, with 1 in 4 people with NASH having sarcopenia vs. 1 in 10 for people without liver disease. Multivariate analyses controlling for age, gender, body mass index (BMI), hypertension, diabetes, and smoking status reveal an odds ratio of 2.8 (95% CI, 1.21-4.30). Odds ratio for having sarcopenia was even elevated after controlling for insulin resistance (OR, 2.05; 95% CI, 1.01-4.16). Although other cohorts have examined links between liver and sarcopenia, the current study applied improved methods for the detection of sarcopenia and also liver health.
Sarcopenia per se is a pathological disorder characterized by a generalized loss of skeletal muscle mass; low muscle mass is associated with weakness (low muscle strength, also referred to as "dynapenia"), but both measures should be related to body shape, i.e., BMI. The Sarcopenia Project of the Foundation for the National Institutes of Health (NIHSP) defined the cut-points for low lean mass and muscle strength on the basis of 9 large community-dwelling elderly cohorts, based on appendicular lean mass adjusted for BMI (ALMBMI) and maximal grip strength adjusted for BMI (GSMaxBMI) in men and women [6] (Table 1). The whole analysis was based on lean mass measured from total body scans using dual-energy X-ray absorptiometry (DXA), where appendicular lean mass was the sum of lean mass from both arms and legs, and grip strength was measured by a handheld dynamometer. The cut-points of weakness and low appendicular mass defined by NIHSP proved useful for epidemiological and clinical purposes [[7], [8]], both predicting increased likelihood for 10-year incident mobility impairment, independently of each other, not 10-year mortality risks [9].
What this means?
The paper from Koo and colleagues provides a unique insight into the relationship between age-related loss of muscle mass, metabolic control and liver disease. We have known the relationship between metabolism, abdominal adiposity and liver health for some time [15]. Although previous reports have shown that sarcopenia relates to NAFLD, it is the extension of this to NASH, independently of insulin resistance, which is of most interest. There are several important questions that remain unanswered; which comes first - sarcopenia or NASH? Are sarcopenia and NASH both organ-level presentations of a more systematic factor (transplanted NASH patients develop NASH again, suggesting that there is something beyond the liver influencing fibrosis). Indeed, obesity affects liver by increasing inflammation and tumourgenesis via elevated IL-6 and TNFα [16]. Pro-inflammatory cytokines can drive cellular senescence and limit regenerative capacity in regions around it. This cellular cross-talk could be a 'butterfly effect' of cellular senescence. In this context, NASH could be viewed as the livers response to accelerated ageing. Finally, loss of muscle mass has been known for decades to be common in advanced liver disease, namely in cirrhosis [17], with a negative effect on disease progression [18], and sarcopenia improves the prediction of mortality independent of etiology [19]. In the study of Koo et al., sarcopenia was associated with severe fibrosis (≥F2) [6], which raises the possibility of an independent effect of poor liver function. However, cirrhosis was rare in the population (only 7% of cases), and sarcopenia was present throughout the histological spectrum of NAFLD. It is very likely that the altered metabolic milieu of NAFLD (obesity and diabetes) may significantly contribute to accelerate the loss of muscle mass.
Moving forward, prospective cohorts will teach us more about which comes first, sarcopenia or NASH, and the mechanisms underpinning this. However, there are important implications for clinical care today. Sarcopenia carries with it significant individual burden, whether or not a patient has NASH. The ability to move is essential for the maintenance of independence and also for the prevention of secondary disease. The excess risk of falls and frailty alongside negative impact on quality of life with sarcopenia [20] mean that care teams should be actively considering functional wellbeing whilst clinically managing NASH. These data highlight the pressing need for multi-disciplinary teams to support patients. Whilst effective therapies for NASH are limited, or in early development, helping patients maintain muscle function through diet and exercise (and possibly pharmacotherapies), may be the most potent therapy available to us. Whilst we are reaping the benefits of living longer, the paper by Koo and colleagues is a stark reminder that we need to focus on ageing well, and treating the patient as a whole, not just the liver.
------------------
In conclusion, sarcopenia was associated with a roughly twofold increased risk of NASH and significant fibrosis in NAFLD patients. In particular, sarcopenia was a risk factor for biopsy-proven NASH and significant fibrosis, independent of obesity and insulin resistance. Given the significant association between low skeletal muscle mass and both NASH and significant fibrosis, increasing muscle mass, especially ASM, might be a new central strategy for the prevention and management of NASH and significant fibrosis. Nevertheless, additional prospective longitudinal studies are still warranted to confirm our findings and to further elucidate the causal relationship between sarcopenia and the development of NASH.
Sarcopenia is an independent risk factor for non-alcoholic steatohepatitis and significant fibrosis
Background & Aims
We explored whether sarcopenia is associated with the histological severity of non-alcoholic fatty liver disease (NAFLD), especially non-alcoholic steatohepatitis (NASH) and significant fibrosis.
Methods
In a biopsy-proven NAFLD cohort, the appendicular skeletal muscle mass (ASM) was measured. Sarcopenia was defined as a ASM/body weight (ASM%) value beyond two standard deviations below the mean for healthy young adults.
Results
Among the entire set of 309 subjects, the prevalence of sarcopenia in subjects without NAFLD, with non-alcoholic fatty liver (NAFL), and with NASH were 8.7%, 17.9%, and 35.0%, respectively (p <0.001). ASM% was inversely correlated with the severity of fibrosis (p <0.001), and the prevalence of significant fibrosis (≥F2) was higher in subjects with sarcopenia than in those without (45.7% vs. 24.7%; p <0.001). A crude analysis revealed that sarcopenia was associated with NAFLD (odds ratio [OR], 3.82; 95% confidence interval [CI], 1.58-9.25), which became insignificant after adjustment for body mass index (BMI), diabetes, and hypertension. Among NAFLD subjects, subjects with sarcopenia were more likely to have NASH than those without sarcopenia through a multivariate analysis adjusted for age, gender, BMI, hypertension, diabetes, and smoking status (OR, 2.28; 95% CI, 1.21-4.30), and this finding was obtained even after adjustment for insulin resistance (OR, 2.30; 95% CI, 1.08-4.93). Sarcopenia was also associated with significant fibrosis independent of BMI and insulin resistance (OR, 2.05; 95% CI, 1.01-4.16).
Conclusions
In this large biopsy-proven NAFLD cohort, sarcopenia was significantly associated with NASH and significant fibrosis.
Lay summary
Low muscle mass was found to be associated with histological severity in non-alcoholic fatty liver disease, and sarcopenia was significantly associated with non-alcoholic steatohepatitis and significant fibrosis, independent of obesity, inflammation, and insulin resistance.
Introduction
Non-alcoholic fatty liver disease (NAFLD), defined as hepatic steatosis that is not caused by significant alcohol consumption or other causes of liver disease, is currently the most prevalent liver disease worldwide [1]. Lipid accumulation and peroxidation and associated inflammation can induce hepatocellular damage and subsequent hepatic fibrosis [2], which results in non-alcoholic steatohepatitis (NASH) [1]. NAFLD may progress to NASH, advanced fibrosis, cirrhosis, or hepatocellular carcinoma [[3], [4]]. Because patients with NASH or advanced fibrosis have higher rates of liver-related [[5], [6]] and non-liver-related mortality [[5], [6], [7]] than those with non-alcoholic fatty liver (NAFL), early identification and intervention of this high-risk group may reduce the burden associated with these diseases.
Insulin resistance is one of the main pathophysiological mechanisms underlying the development of NAFLD [[1], [8]]. Because the skeletal muscle is the primary tissue responsible for insulin-mediated glucose disposal [[9], [10], [11]], low skeletal muscle mass reduces insulin-mediated glucose disposal, independent of obesity, and might explain the association between NAFLD and insulin resistance, which cannot be explained by fat mass [12]. Recent epidemiological studies have shown that sarcopenia is also associated with NAFLD and advanced fibrosis based on the detection of non-invasive markers in an Asian population [[13], [14], [15]]. These previous studies defined NAFLD using validated non-invasive serum panels [[13], [14]] or liver attenuation indices as measured by computed tomography (CT) [15]. However, liver biopsy remains the gold standard for characterizing liver histology in subjects with NAFLD [1]. A considerable number of patients with significant steatosis on biopsy are not recognized by imaging [16], and non-invasive markers for the diagnosis of NAFLD may also result in the misclassification of NASH [17] or advanced fibrosis [18].
In this prospective cohort study, we aimed to determine the association between sarcopenia and the histological severity of NAFLD. Specifically, we investigated whether the presence of sarcopenia might be associated with the risk of NASH and significant fibrosis, independent of obesity, metabolic risk factors, and insulin resistance.
Discussion
In this study, the prevalence of sarcopenia was associated with biopsy-proven NASH and significant fibrosis in subjects with NAFLD. These associations persisted after further adjustment for obesity, metabolic risk factors, and insulin resistance. Skeletal muscle mass was associated with not only the histological grades of steatosis and hepatocellular ballooning but also the stage of fibrosis. Patients with sarcopenia exhibited approximately twofold increased odds of suffering from NASH or significant fibrosis.
Recent epidemiological studies have shown that low skeletal muscle mass is associated with NAFLD [[13], [14], [15]]. Advanced fibrosis, as estimated using non-invasive fibrosis markers, is also associated with low skeletal muscle mass, independent of obesity or metabolic control [[13], [14]]. However, these aforementioned studies defined NAFLD using non-invasive markers [[13], [14]] or liver attenuation indices as measured by CT [15]. Non-contrast CT exhibits high performance for qualitatively diagnosing steatosis of 30% or greater; however, the diagnostic performance of non-contrast CT is not clinically acceptable for quantitative assessment of steatosis [34]. The use of non-invasive markers for the diagnosis of NAFLD may also result in the misclassification of NASH [17] or advanced fibrosis [18]: a recent study reported that the majority of patients with advanced fibrosis or cirrhosis were misclassified using the FIB-4 score [18]. Currently, no clinical tools are available to reliably distinguish NASH from NAFL. Thus, liver biopsy remains necessary for accurate diagnosis of NASH and fibrosis [1]. A relatively small-sized study with NASH patients showed that a stepwise significant decrease in muscle area from the control group to the NASH group to the cirrhosis group [35]. We confirmed this finding with a larger sample size and showed an independent association between NASH or significant fibrosis and sarcopenia after adopting different sarcopenia criteria. In the current study, among the various histological parameters associated with NASH, steatosis and hepatocellular ballooning, but not lobular inflammation, were significantly associated with ASM%. This result was in line with the finding from a previous study, which found that hepatocellular ballooning was most closely associated with insulin resistance [[36], [37]].
Even though the data regarding NAFLD and cardiovascular disease are still debated [38], several studies have shown that NAFLD with advanced fibrosis is a significant predictor of mortality from cardiovascular diseases [[6], [7]], as well as of liver-related events [5]. Although several pharmacotherapies including vitamin E [39], obeticholic acid [39], and glucagon-like peptide-1 analogue [40] show promising results, no approved pharmacotherapies for NASH and advanced fibrosis are currently available. The current treatment of choice is lifestyle modification, including weight reduction [1]. Increasing the skeletal muscle mass may be a promising potential treatment option for NAFLD [[41], [42], [43]]. Furthermore, resistance training is effective for reducing steatosis [[41], [43]], independent of weight reduction, in NAFLD patients.
The importance of visceral obesity has been implicated in the pathogenesis and prognosis of NAFLD and NASH [[44], [45], [46]]. However, the finding that patients without visceral obesity may also develop NASH and significant fibrosis suggests that other mechanisms beyond visceral obesity might account for advanced liver damage [47]. The skeletal muscle is the primary tissue responsible for insulin-mediated glucose disposal, and the prominent role of skeletal muscle in insulin resistance has been confirmed by epidemiological [[9], [10]] and experimental [11] studies. The finding that the association between NASH and sarcopenia was attenuated but remained significant after adjustment for HOMA-IR suggests that the association between sarcopenia and NASH might be mediated, in part, by insulin resistance rather than obesity.
Inflammation may also serve as an important link between sarcopenia and NASH. In the current study, ASM% showed a significant inverse correlation with the hsCRP level. Low muscle mass is closely associated with chronic inflammation [48]. Furthermore, subclinical inflammation and oxidative stress mediated by pro-inflammatory cytokines promote the catabolic stimulation of muscle [49], which might result in the loss of muscle mass [[50], [51]]. Oxidative stress and chronic inflammation are also important in the development of NASH [[2], [52]]. Our finding that the association between sarcopenia and NASH was slightly attenuated after adjustment for the hsCRP level indicates that inflammation might also partially serve as a mediator between sarcopenia and NASH.
It is acknowledged that a progressive decrease in muscle mass and an increase in fat mass, particularly that of the visceral component, are common body compositional changes associated with aging [[53], [54]]. In the elderly population, sarcopenia has been closely related to many clinical consequences, including metabolic impairment [9], increased cardiovascular risk [26], and mortality [55]. Furthermore, compared with sarcopenia in young adults, sarcopenia in the elderly population is more strongly associated with a susceptibility to diabetes [56]. Because sarcopenia is closely interconnected with diabetes and obesity, our study sheds light on the impact of sarcopenia on significant fibrosis and NASH independently of diabetes and obesity. Given that people are living longer in most developed nations, it is of paramount importance to manage age-related sarcopenia because we may expect an increase in the prevalence of NASH as well as of diabetes accompanying increased aging.
The first strength of this study was that we investigated the association between sarcopenia and the histological severity of NAFLD in a prospective cohort. The second strength is that the histological diagnoses of NASH and significant fibrosis were reviewed and established by a single pathologist who specialized in liver pathology. The third strength is the confirmation of the association between sarcopenia and the histological severity of NAFLD with adjustment for a variety of clinical confounders including insulin resistance and inflammatory markers. Furthermore, we successfully supported the association between the histological severity of fibrosis and ASM% through a measurement of liver stiffness.
The main limitation of this study is that the results regarding the causality of the observed relationships should be interpreted with caution due to the cross-sectional nature of the study design. Second, we estimated the skeletal muscle mass using BIA. Imaging using CT, magnetic resonance imaging and dual-energy X-ray absorptiometry (DXA) have been used for in vivo measurements of the skeletal muscle mass in humans. Although BIA has been reported to provide an accurate estimate of DXA-derived ASM in a population with various clinical disorders [[24], [57]], there might be a difference between fat-free mass measured by BIA and muscle mass according to the body water content [58]. Furthermore, BIA cannot provide information on muscle quality. Third, HOMA-IR was used to determine insulin resistance in the current study. Although HOMA-IR is an indirect measure of insulin resistance, this index has been validated using the hyperinsulinemic-euglycemic clamp technique [[59], [60]] and is thus currently used in various epidemiologic studies. Finally, because this study included only subjects of East Asian ethnicity, the conclusions might not be generalizable to other ethnic populations.
In conclusion, sarcopenia was associated with a roughly twofold increased risk of NASH and significant fibrosis in NAFLD patients. In particular, sarcopenia was a risk factor for biopsy-proven NASH and significant fibrosis, independent of obesity and insulin resistance. Given the significant association between low skeletal muscle mass and both NASH and significant fibrosis, increasing muscle mass, especially ASM, might be a new central strategy for the prevention and management of NASH and significant fibrosis. Nevertheless, additional prospective longitudinal studies are still warranted to confirm our findings and to further elucidate the causal relationship between sarcopenia and the development of NASH.
Results
Clinical characteristics according to the spectrum of NAFLD
Among the total of 309 subjects (mean age, 53 ± 14 years; men, 46.9%), 123 (men, 42.3%) and 117 subjects (men, 55.6%) were classified as biopsy-proven NASH and NAFL, respectively; thus, 51.3% of all NAFLD patients were diagnosed with NASH. The baseline characteristics of the study population are shown in Table 1. There were noticeable differences in clinical and anthropometric characteristics according to the spectrum of NAFLD: BMI, waist circumference, the prevalence of diabetes mellitus and hypertension, serum ALT, aspartate aminotransferase (AST), and gamma-glutamyl transferase (GGT) levels, and HOMA-IR displayed a linear correlation with the severity of NAFLD, with the direction indicating that NASH reflects a poorer health status (Table 1). As NAFLD severity increased, the high sensitivity C-reactive protein (hsCRP) level also increased significantly (p <0.001).
Although there was no significant difference in ASM according to NAFLD severity, significant differences in ASM/BMI and ASM% according to NAFLD severity were observed (p = 0.008 and p <0.001, respectively; Table 1). Subjects with NASH showed a significantly lower ASM% compared to those without NAFLD or those with NAFL (p = 0.001 and p = 0.004, respectively; Fig. 1A).
Clinical characteristics according to sarcopenia status
The prevalences of sarcopenia_Wt were 8.7%, 17.9% and 35.0% in the no NAFLD, NAFL, and NASH groups, respectively (p <0.001; Table 1 and Fig. 1B). This trend was consistent with the prevalence of sarcopenia using an alternative definition, sarcopenia_BMI (p = 0.003; Table 1 and Fig. 1C). As expected, subjects with sarcopenia were more metabolically unfavorable than those without sarcopenia. Subjects with sarcopenia_Wt showed a higher BMI, waist circumference, serum AST, GGT, and hsCRP levels, and HOMA-IR than those without sarcopenia (Table 2). The ASM% showed a significant inverse correlation with the hsCRP level (Spearman's ρ = -0.260, p <0.001; Supplementary Fig. 1).
Liver histology according to ASM
Subjects with sarcopenia_Wt had more severe histological grades of steatosis and hepatocellular ballooning (p for trend = 0.005 and <0.001, respectively) and a higher fibrosis stage (p for trend <0.001) than those without sarcopenia_Wt (Table 3). The prevalence of significant fibrosis (≥F2) was significantly higher among those with sarcopenia (45.7%) than among those without sarcopenia (24.7%) (p <0.001). ASM% inversely correlated with the severity of lobular inflammation (Spearman's ρ = -0.147, p = 0.010), hepatocellular ballooning (Spearman's ρ = -0.163, p = 0.004), and fibrosis (Spearman's ρ = -0.223, p <0.001) but not steatosis (Spearman's ρ = -0.096, p = 0.093; Figs. 2A-D). Subjects with significant fibrosis (≥F2) displayed a significantly lower ASM% than those without significant fibrosis (< F2) (p = 0.001; Fig. 2E).
Sarcopenia and NASH
As shown in Table 4, in the age- and gender-adjusted analysis, sarcopenia_Wt was significantly associated with NAFLD (odds ratio [OR], 3.81; 95% confidence interval [CI], 1.57-9.25). However, this trend was attenuated and remained non-significant after further adjustment. Given that subjects with NASH have a higher risk of progression to cirrhosis and of mortality [3], we investigated whether sarcopenia could have a crucial effect on NASH independent of NAFL. Among patients with NAFLD, the presence of sarcopenia_Wt was associated with a 2.5-fold increase in the risk of NASH (OR, 2.46; 95% CI, 1.35-4.48). This association persisted after adjustment for age, gender, BMI, hypertension, diabetes, and smoking status (Model 1). The addition of the total cholesterol, triglyceride, HDL-cholesterol, and ALT levels did not significantly reduce the OR for this association (Model 2; OR, 2.59; 95% CI, 1.22-5.48). Because inflammation was closely associated with the sarcopenic index, a multivariate analysis adjusted for the hsCRP level showed that the significant association was slightly attenuated but remained (Model 3; OR, 2.58; 95% CI, 1.21-5.48). As insulin resistance plays an important role in NASH, we performed an additional adjustment for HOMA-IR, and a similar statistically significant association was observed (Model 4; OR, 2.30; 95% CI, 1.08-4.93). We then conducted a sensitivity analysis to examine the robustness of our findings. Analysis using a different cut-off parameter (sarcopenia_BMI) also revealed a significant association: the ORs for NASH were 2.37 (95% CI, 1.21-4.64), 2.15 (95% CI, 1.08-4.30), and 2.66 (95% CI, 1.18-5.98) in the age- and gender-adjusted model, Model 1, and Model 2, respectively. After further adjustment for insulin resistance, this association was attenuated and remained significant (Model 4; OR, 2.33; 95% CI, 1.02-5.34).
Sarcopenia and significant fibrosis
Among the patients with NAFLD, those with significant fibrosis were older, had a higher prevalence of female, diabetes, and hypertension, and displayed lower serum albumin levels and platelet counts than those without significant fibrosis (Supplementary Table 1). The presence of sarcopenia_Wt was associated with the presence of significant fibrosis (OR, 2.01; 95% CI, 1.12-3.61; Table 5). This association persisted after adjustment for age, gender, BMI, hypertension, diabetes, and smoking status (Model 1; OR, 2.12; 95% CI, 1.09-4.10) and after additional adjustment for total cholesterol, triglyceride, and HDL-cholesterol levels (Model 2; OR, 2.21; 95% CI, 1.10-4.44). The significant association between sarcopenia_Wt and significant fibrosis was maintained even after the addition of HOMA-IR to the analysis (Model 3; OR, 2.05; 95% CI, 1.01-4.16). In the sensitivity analysis using sarcopenia_BMI, sarcopenia_BMI was independently associated with an increased risk of significant fibrosis based on Model 1 (OR, 2.59; 95% CI, 1.28-5.23), Model 2 (OR, 2.48; 95% CI, 1.19-5.17), and Model 3 (OR, 2.24; 95% CI, 1.06-4.73; Table 5).
|
|
| |
| |
|
|
|
