Prevalence of Nonalcoholic Fatty Liver Disease in the US
Prevalence of Nonalcoholic Fatty Liver Disease in the US
In this cross-sectional, nationally representative study, we found that hepatic steatosis detected by ultrasonography was present in almost 32.5 million adults or 21.4% of the US noninstitutionalized population aged 20–74 years between 1988 and 1994. The vast majority of the cases could be classified as NAFLD, with an overall prevalence of 19.0% (approximately 28.8 million people). Considerable variations in the prevalence of NAFLD by sex and by race/ethnicity were observed, with men and Mexican-American adults being disproportionally affected by all of these conditions and non-Hispanic blacks having a significantly lower burden of disease. In addition, we found that there were strong independent associations between diabetes or insulin resistance, dyslipidemia, and obesity with NAFLD.
Our findings regarding the magnitude of NAFLD and nonalcoholic steatohepatitis in the general US population have important implications. The rates of diabetes and obesity have been increasing steadily over the last 2 decades, and on the basis of our results, we might expect a parallel increase in the number of people with NAFLD. In 2007–2008, the US Centers for Disease Control and Prevention reported that approximately 72.5 million adults in the United States were obese; based on our observed association between obesity and NAFLD, approximately 55 million adults currently might have NAFLD. Furthermore, because as many as 15% of people with NAFLD can progress to more advanced forms of liver disease such as fibrosis and cirrhosis, monitoring the prevalence of chronic liver disease would be prudent. Similarly, the association between NAFLD and hepatocellular carcinoma also highlights the relevance.
Over the last 2 decades there has been an increase in the number of people with end-stage liver disease requiring transplantation and in the incidence of hepatocellular carcinoma. Others have suggested that these increases were due to chronic viral hepatitis. However, our data indicate that NAFLD is a more common condition, and its potential role as an underlying factor in these increases deserves attention, especially given recent supporting research.
A number of prior studies have estimated the prevalence of NAFLD in the United States, including 1 study by our group. Most of these studies were based on the same national cross-sectional survey but were limited to examination of liver enzymes, which, in conjunction with other laboratory tests and self-reported data, allowed for an estimation of "explained" (e.g., elevated alcohol consumption, viral hepatitis) and "unexplained" elevation of liver enzymes in the general population, the latter of which was assumed to be NAFLD. Liver enzymes are surrogate markers of liver disease with known limited sensitivity and specificity and intraindividual variation. In fact, compared with the prevalence among the general population obtained in these studies, our study found a 3-fold higher prevalence of NAFLD depending on the definition used.
Another important population-based study based on the Dallas Heart Study used proton magnetic resonance spectroscopy to quantify hepatic triglyceride content among 2,287 participants from Dallas County, Texas. The investigators defined hepatic steatosis as 5.5% or more hepatic fat as measured by proton magnetic resonance spectroscopy. By using that definition and tool, they found a prevalence of hepatic steatosis of 31%. There are several reasons why we may have found a lower prevalence. First, the Dallas Heart Study was conducted from 2000 to 2002 and included a representative sample of the population of Dallas County, which may not be representative of the US population; NHANES III occurred from 1988 to 1994 and included a nationally representative sample. Indeed, 43% of the study population in the Dallas Heart Study had obesity, whereas in the NHANES III population, the nationwide prevalence of obesity was 23.5%. Second, different imaging techniques were used. Our results are based on ultrasonography, which has a lower sensitivity for mild disease; therefore, our estimates may be conservative. More recently, Williams et al. used ultrasonography to detect hepatic steatosis and found a 46% prevalence of NAFLD among a sample of 328 middle-aged adults who were seen at the Brooke Army Medical Center, including active duty military personnel, their dependents, and military retirees. Similar to the differences observed with the Dallas Heart Study, our results may differ from these because of geographical differences and time trends in the prevalence of risk factors, as well as a different population studied (NHANES does not include most military personnel).
We found significant differences in the prevalence of NAFLD by racial/ethnic groups and by sex independent of age, income, and metabolic and anthropometric characteristics. Although there is the potential of residual confounding by these factors, our findings are consistent with those of a number of previous studies and have important public health implications for the design and implementation of prevention and treatment strategies. A number of social, environmental, and genetic factors may account for this variation and merit future research.
Our study has several limitations. First, these data were collected more than 20 years ago; however, this is the only nationally representative US study with ultrasonography data available to determine the prevalence of this condition in the general population by key sociodemographic characteristics. In addition, the correlates of NAFLD found in our study should remain consistent. Although the overall sensitivity and specificity of ultrasonography compared with biopsy are approximately 85% and 94%, respectively, ultrasonography is relatively insensitive to smaller amounts of hepatic steatosis. In addition, ultrasonography cannot detect inflammation and fibrosis, which are hallmarks of more advanced stages of NAFLD. In the absence of a standard definition, we defined NAFLD by excluding people with hepatic steatosis in the presence of elevated alcohol intake or the use of antiretroviral medications, which were factors significantly associated with ultrasonography-detected hepatic steatosis. Our study was cross-sectional and therefore has some limitations and biases inherent to the design; we cannot establish causality or temporality between the risk factors studied and NAFLD. We describe the correlates of prevalent disease that may well be different from the correlates of incident cases. In addition, some selection bias may also be present because individuals with more severe forms of disease may have been less likely to participate in NHANES III; however, this bias would result in the underestimation of the observed association between the risk factors and NAFLD.
In conclusion, in 1988–1994 in the United States, the prevalence of NAFLD was already very high (19%). Furthermore, NAFLD disproportionally affects Mexican Americans, older adults, and people with diabetes and obesity. The aging of the population, along with the increasing prevalence of diabetes and obesity, is expected to contribute to an increase in the prevalence of these conditions and in the overall burden of liver disease in the United States. Randomized clinical trials have shown promising results by using lifestyle modification, vitamin E and pioglitazone, and rosiglitazone, but there is no US Food and Drug Administration–approved drug for treating NAFLD. More research is needed to develop effective treatment options for people with NAFLD. In addition, given the prevalence of this condition, the development of improved assessment tools that can detect and stage NAFLD among the general population would be beneficial for future research.
Discussion
In this cross-sectional, nationally representative study, we found that hepatic steatosis detected by ultrasonography was present in almost 32.5 million adults or 21.4% of the US noninstitutionalized population aged 20–74 years between 1988 and 1994. The vast majority of the cases could be classified as NAFLD, with an overall prevalence of 19.0% (approximately 28.8 million people). Considerable variations in the prevalence of NAFLD by sex and by race/ethnicity were observed, with men and Mexican-American adults being disproportionally affected by all of these conditions and non-Hispanic blacks having a significantly lower burden of disease. In addition, we found that there were strong independent associations between diabetes or insulin resistance, dyslipidemia, and obesity with NAFLD.
Our findings regarding the magnitude of NAFLD and nonalcoholic steatohepatitis in the general US population have important implications. The rates of diabetes and obesity have been increasing steadily over the last 2 decades, and on the basis of our results, we might expect a parallel increase in the number of people with NAFLD. In 2007–2008, the US Centers for Disease Control and Prevention reported that approximately 72.5 million adults in the United States were obese; based on our observed association between obesity and NAFLD, approximately 55 million adults currently might have NAFLD. Furthermore, because as many as 15% of people with NAFLD can progress to more advanced forms of liver disease such as fibrosis and cirrhosis, monitoring the prevalence of chronic liver disease would be prudent. Similarly, the association between NAFLD and hepatocellular carcinoma also highlights the relevance.
Over the last 2 decades there has been an increase in the number of people with end-stage liver disease requiring transplantation and in the incidence of hepatocellular carcinoma. Others have suggested that these increases were due to chronic viral hepatitis. However, our data indicate that NAFLD is a more common condition, and its potential role as an underlying factor in these increases deserves attention, especially given recent supporting research.
A number of prior studies have estimated the prevalence of NAFLD in the United States, including 1 study by our group. Most of these studies were based on the same national cross-sectional survey but were limited to examination of liver enzymes, which, in conjunction with other laboratory tests and self-reported data, allowed for an estimation of "explained" (e.g., elevated alcohol consumption, viral hepatitis) and "unexplained" elevation of liver enzymes in the general population, the latter of which was assumed to be NAFLD. Liver enzymes are surrogate markers of liver disease with known limited sensitivity and specificity and intraindividual variation. In fact, compared with the prevalence among the general population obtained in these studies, our study found a 3-fold higher prevalence of NAFLD depending on the definition used.
Another important population-based study based on the Dallas Heart Study used proton magnetic resonance spectroscopy to quantify hepatic triglyceride content among 2,287 participants from Dallas County, Texas. The investigators defined hepatic steatosis as 5.5% or more hepatic fat as measured by proton magnetic resonance spectroscopy. By using that definition and tool, they found a prevalence of hepatic steatosis of 31%. There are several reasons why we may have found a lower prevalence. First, the Dallas Heart Study was conducted from 2000 to 2002 and included a representative sample of the population of Dallas County, which may not be representative of the US population; NHANES III occurred from 1988 to 1994 and included a nationally representative sample. Indeed, 43% of the study population in the Dallas Heart Study had obesity, whereas in the NHANES III population, the nationwide prevalence of obesity was 23.5%. Second, different imaging techniques were used. Our results are based on ultrasonography, which has a lower sensitivity for mild disease; therefore, our estimates may be conservative. More recently, Williams et al. used ultrasonography to detect hepatic steatosis and found a 46% prevalence of NAFLD among a sample of 328 middle-aged adults who were seen at the Brooke Army Medical Center, including active duty military personnel, their dependents, and military retirees. Similar to the differences observed with the Dallas Heart Study, our results may differ from these because of geographical differences and time trends in the prevalence of risk factors, as well as a different population studied (NHANES does not include most military personnel).
We found significant differences in the prevalence of NAFLD by racial/ethnic groups and by sex independent of age, income, and metabolic and anthropometric characteristics. Although there is the potential of residual confounding by these factors, our findings are consistent with those of a number of previous studies and have important public health implications for the design and implementation of prevention and treatment strategies. A number of social, environmental, and genetic factors may account for this variation and merit future research.
Our study has several limitations. First, these data were collected more than 20 years ago; however, this is the only nationally representative US study with ultrasonography data available to determine the prevalence of this condition in the general population by key sociodemographic characteristics. In addition, the correlates of NAFLD found in our study should remain consistent. Although the overall sensitivity and specificity of ultrasonography compared with biopsy are approximately 85% and 94%, respectively, ultrasonography is relatively insensitive to smaller amounts of hepatic steatosis. In addition, ultrasonography cannot detect inflammation and fibrosis, which are hallmarks of more advanced stages of NAFLD. In the absence of a standard definition, we defined NAFLD by excluding people with hepatic steatosis in the presence of elevated alcohol intake or the use of antiretroviral medications, which were factors significantly associated with ultrasonography-detected hepatic steatosis. Our study was cross-sectional and therefore has some limitations and biases inherent to the design; we cannot establish causality or temporality between the risk factors studied and NAFLD. We describe the correlates of prevalent disease that may well be different from the correlates of incident cases. In addition, some selection bias may also be present because individuals with more severe forms of disease may have been less likely to participate in NHANES III; however, this bias would result in the underestimation of the observed association between the risk factors and NAFLD.
In conclusion, in 1988–1994 in the United States, the prevalence of NAFLD was already very high (19%). Furthermore, NAFLD disproportionally affects Mexican Americans, older adults, and people with diabetes and obesity. The aging of the population, along with the increasing prevalence of diabetes and obesity, is expected to contribute to an increase in the prevalence of these conditions and in the overall burden of liver disease in the United States. Randomized clinical trials have shown promising results by using lifestyle modification, vitamin E and pioglitazone, and rosiglitazone, but there is no US Food and Drug Administration–approved drug for treating NAFLD. More research is needed to develop effective treatment options for people with NAFLD. In addition, given the prevalence of this condition, the development of improved assessment tools that can detect and stage NAFLD among the general population would be beneficial for future research.