Alimentary Pharmacology and Therapeutics
Fatty liver indices in the multiethnic United States National Health and Nutrition Examination Survey C. E. Ruhl* & J. E. Everhart†
*Social & Scientific Systems, Inc., Silver Spring, MD, USA. † Department of Health and Human Services, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD, USA.
Correspondence to: Dr C. E. Ruhl, Social & Scientific Systems, Inc., 8757 Georgia Avenue, 12th floor, Silver Spring, MD 20910, USA. E-mail: [email protected]
Publication data Submitted 19 August 2014 First decision 9 September 2014 Resubmitted 10 October 2014 Accepted 11 October 2014 EV Pub Online 6 November 2014 This article was accepted for publication after full peer-review.
SUMMARY Background Validated non-invasive measures of fatty liver are needed that can be applied across populations and over time. A fatty liver index (FLI) including body mass index, waist circumference, triglycerides and gamma glutamyltransferase (GGT) activity was developed in an Italian municipality, but has not been validated widely or examined in a multiethnic population. Aims We evaluated this FLI in the multiethnic U.S. National Health and Nutrition Examination Survey (NHANES) and also to explore whether an improved index for the U.S. population (US FLI) could be derived. The US FLI would then used to examine U.S. time trends in fatty liver prevalence. Methods We studied 5869 fasted, viral hepatitis negative adult participants with abdominal ultrasound data on fatty liver in the 1988–1994 NHANES. Time trend analyses included 21 712 NHANES 1988–1994 and 1999–2012 participants. Results The prevalence of fatty liver was 20%. For the FLI, the area under the receiver operating characteristic curve [AUC; 95% confidence interval (CI)] was 0.78 (0.74–0.81). The US FLI included age, race-ethnicity, waist circumference, GGT activity, fasting insulin and fasting glucose and had an AUC (95% CI) of 0.80 (0.77–0.83). Defining fatty liver as a US FLI ≥ 30, the prevalence increased from 18% in 1988–1991 to 29% in 1999–2000 to 31% in 2011–2012. Conclusions For predicting fatty liver, the US FLI was a modest improvement over the FLI in the multiethnic U.S. population. Using this measure, the fatty liver prevalence in the U.S. population increased substantially over two decades. Aliment Pharmacol Ther 2015; 41: 65–76
ª 2014 John Wiley & Sons Ltd doi:10.1111/apt.13012
65
C. E. Ruhl and J. E. Everhart INTRODUCTION Fatty liver is perhaps the most common liver disease in the United States and other Western countries.1, 2 However, uniform measurements and definitions have not been applied across populations to determine the prevalence of fatty liver or its risk factors. The criterion standard for diagnosis of hepatic steatosis requires a liver biopsy for histological evaluation. Non-invasive radiological measures including magnetic resonance imaging (MRI) and ultrasonography are also useful for identification of fatty liver, but are expensive, require expert interpretation and are not easily performed in large populations. Therefore, validated non-invasive, easily ascertained measurements are needed to identify patients at high risk for fatty liver in whom additional screening may be indicated, and for use in epidemiologic studies of the general healthy population. The fatty liver index (FLI) was derived from a 2002–2003 study of a single municipality of fewer than 8000 total population in North Central Italy.2, 3 Among 216 participants with suspected liver disease based on elevated alanine aminotransferase (ALT) or gamma glutamyltransferase (GGT) activities and 280 without suspected liver disease,4 the best predictors of ultrasound-diagnosed fatty liver were fasting insulin and triglyceride concentrations, body mass index (BMI), sex and GGT activity. The final set of predictors excluded insulin which was not thought to be a widely available measurement and included waist circumference and not sex.4 The FLI predicted hepatic steatosis with an area under the receiver operating characteristic (ROC) curve (AUC; 95% confidence interval (CI)] of 0.84 (0.81–0.87). A FLI of 2 drinks per day for men or >1 drink per day for women; n = 481) and persons missing data on alcohol intake (n = 192), leaving 5196 persons for analysis. For analyses evaluating the FLI, we additionally excluded persons with missing data on one or more components of the FLI, resulting in a sample of 4415 for analyses of all hepatic steatosis and of 3955 for analyses excluding persons with significant alcohol intake. For time trend analyses, NHANES III and NHANES 1999–2012 were utilized. From NHANES III we included 6911 morning-examined, fasted, viral hepatitis negative participants aged 20 years and older. Of 51 678 sampled persons aged 20 years and over in NHANES 1999–2012, 36 071 (70%) attended an examination at a mobile examination centre. We studied 15 315 persons, examined in the morning after an overnight fast. Participants were excluded if they were positive for serum hepatitis B surface antigen or hepatitis C antibody (n = 354) or missing data on hepatitis serology (n = 160), resulting in 14 801 persons available for analysis. Data were collected on factors known or thought to be related to hepatic steatosis:15, 16 age (years), sex, race-ethnicity (non-Hispanic white, non-Hispanic black, Mexican American), alcohol consumption (drinks/day; 0, 2), cigarette smoking (never, former, 0.02, beginning with the highest P-value. Continuous factors whose distributions were skewed to the right were log transformed before being added to regression models. Quadratic terms for continuous factors in the final model were considered for inclusion by adding them individually to the model.
FLI ¼ ðe0:953 loge ðtriglyceridesÞ þ 0:139 BMI þ 0:718 loge ðGGTÞ þ 0:053 waist circumference15:745 Þ= ð1 þ e0:953 loge ðtriglyceridesÞ þ 0:139 BMI þ 0:718 loge ðGGTÞ þ 0:053 waist circumference15:745 Þ 100 Aliment Pharmacol Ther 2015; 41: 65-76 ª 2014 John Wiley & Sons Ltd
67
C. E. Ruhl and J. E. Everhart Table 1 | Characteristics of participants by hepatic steatosis status in the third U.S. National Health and Nutrition Examination Survey, 1988–1994
Characteristic Prevalence of hepatic steatosis (%) Age [years; mean (s.d.)] Women (%) Race-ethnicity (%) Non-Hispanic white Non-Hispanic black Mexican American Drinking (drinks/day; %) 0 >0 to 2 Cigarette smoking (%) Never Former 0.3 (mg/dL; %)
No hepatic steatosis (N = 4585)
Hepatic steatosis* (N = 1284)
P-value†
Unadjusted R 2‡
80.4 41.1 (15.0) 53.4
19.6 48.0 (14.4) 45.8
–
Fatty liver indices in the multiethnic United States National Health and Nutrition Examination Survey C. E. Ruhl* & J. E. Everhart†
*Social & Scientific Systems, Inc., Silver Spring, MD, USA. † Department of Health and Human Services, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD, USA.
Correspondence to: Dr C. E. Ruhl, Social & Scientific Systems, Inc., 8757 Georgia Avenue, 12th floor, Silver Spring, MD 20910, USA. E-mail: [email protected]
Publication data Submitted 19 August 2014 First decision 9 September 2014 Resubmitted 10 October 2014 Accepted 11 October 2014 EV Pub Online 6 November 2014 This article was accepted for publication after full peer-review.
SUMMARY Background Validated non-invasive measures of fatty liver are needed that can be applied across populations and over time. A fatty liver index (FLI) including body mass index, waist circumference, triglycerides and gamma glutamyltransferase (GGT) activity was developed in an Italian municipality, but has not been validated widely or examined in a multiethnic population. Aims We evaluated this FLI in the multiethnic U.S. National Health and Nutrition Examination Survey (NHANES) and also to explore whether an improved index for the U.S. population (US FLI) could be derived. The US FLI would then used to examine U.S. time trends in fatty liver prevalence. Methods We studied 5869 fasted, viral hepatitis negative adult participants with abdominal ultrasound data on fatty liver in the 1988–1994 NHANES. Time trend analyses included 21 712 NHANES 1988–1994 and 1999–2012 participants. Results The prevalence of fatty liver was 20%. For the FLI, the area under the receiver operating characteristic curve [AUC; 95% confidence interval (CI)] was 0.78 (0.74–0.81). The US FLI included age, race-ethnicity, waist circumference, GGT activity, fasting insulin and fasting glucose and had an AUC (95% CI) of 0.80 (0.77–0.83). Defining fatty liver as a US FLI ≥ 30, the prevalence increased from 18% in 1988–1991 to 29% in 1999–2000 to 31% in 2011–2012. Conclusions For predicting fatty liver, the US FLI was a modest improvement over the FLI in the multiethnic U.S. population. Using this measure, the fatty liver prevalence in the U.S. population increased substantially over two decades. Aliment Pharmacol Ther 2015; 41: 65–76
ª 2014 John Wiley & Sons Ltd doi:10.1111/apt.13012
65
C. E. Ruhl and J. E. Everhart INTRODUCTION Fatty liver is perhaps the most common liver disease in the United States and other Western countries.1, 2 However, uniform measurements and definitions have not been applied across populations to determine the prevalence of fatty liver or its risk factors. The criterion standard for diagnosis of hepatic steatosis requires a liver biopsy for histological evaluation. Non-invasive radiological measures including magnetic resonance imaging (MRI) and ultrasonography are also useful for identification of fatty liver, but are expensive, require expert interpretation and are not easily performed in large populations. Therefore, validated non-invasive, easily ascertained measurements are needed to identify patients at high risk for fatty liver in whom additional screening may be indicated, and for use in epidemiologic studies of the general healthy population. The fatty liver index (FLI) was derived from a 2002–2003 study of a single municipality of fewer than 8000 total population in North Central Italy.2, 3 Among 216 participants with suspected liver disease based on elevated alanine aminotransferase (ALT) or gamma glutamyltransferase (GGT) activities and 280 without suspected liver disease,4 the best predictors of ultrasound-diagnosed fatty liver were fasting insulin and triglyceride concentrations, body mass index (BMI), sex and GGT activity. The final set of predictors excluded insulin which was not thought to be a widely available measurement and included waist circumference and not sex.4 The FLI predicted hepatic steatosis with an area under the receiver operating characteristic (ROC) curve (AUC; 95% confidence interval (CI)] of 0.84 (0.81–0.87). A FLI of 2 drinks per day for men or >1 drink per day for women; n = 481) and persons missing data on alcohol intake (n = 192), leaving 5196 persons for analysis. For analyses evaluating the FLI, we additionally excluded persons with missing data on one or more components of the FLI, resulting in a sample of 4415 for analyses of all hepatic steatosis and of 3955 for analyses excluding persons with significant alcohol intake. For time trend analyses, NHANES III and NHANES 1999–2012 were utilized. From NHANES III we included 6911 morning-examined, fasted, viral hepatitis negative participants aged 20 years and older. Of 51 678 sampled persons aged 20 years and over in NHANES 1999–2012, 36 071 (70%) attended an examination at a mobile examination centre. We studied 15 315 persons, examined in the morning after an overnight fast. Participants were excluded if they were positive for serum hepatitis B surface antigen or hepatitis C antibody (n = 354) or missing data on hepatitis serology (n = 160), resulting in 14 801 persons available for analysis. Data were collected on factors known or thought to be related to hepatic steatosis:15, 16 age (years), sex, race-ethnicity (non-Hispanic white, non-Hispanic black, Mexican American), alcohol consumption (drinks/day; 0, 2), cigarette smoking (never, former, 0.02, beginning with the highest P-value. Continuous factors whose distributions were skewed to the right were log transformed before being added to regression models. Quadratic terms for continuous factors in the final model were considered for inclusion by adding them individually to the model.
FLI ¼ ðe0:953 loge ðtriglyceridesÞ þ 0:139 BMI þ 0:718 loge ðGGTÞ þ 0:053 waist circumference15:745 Þ= ð1 þ e0:953 loge ðtriglyceridesÞ þ 0:139 BMI þ 0:718 loge ðGGTÞ þ 0:053 waist circumference15:745 Þ 100 Aliment Pharmacol Ther 2015; 41: 65-76 ª 2014 John Wiley & Sons Ltd
67
C. E. Ruhl and J. E. Everhart Table 1 | Characteristics of participants by hepatic steatosis status in the third U.S. National Health and Nutrition Examination Survey, 1988–1994
Characteristic Prevalence of hepatic steatosis (%) Age [years; mean (s.d.)] Women (%) Race-ethnicity (%) Non-Hispanic white Non-Hispanic black Mexican American Drinking (drinks/day; %) 0 >0 to 2 Cigarette smoking (%) Never Former 0.3 (mg/dL; %)
No hepatic steatosis (N = 4585)
Hepatic steatosis* (N = 1284)
P-value†
Unadjusted R 2‡
80.4 41.1 (15.0) 53.4
19.6 48.0 (14.4) 45.8
–
