Background & Aims: Nonalcoholic fatty liver disease (NAFLD) is a common cause of liver disease in children and adolescents. The majority of studies of NAFLD in children have been in select populations of the clinically obese. Study aims were to estimate the prevalence of elevated alanine aminotransferase (ALT, as a marker of NAFLD) in a general contemporary adolescent population and to identify leading risk factors for ALT elevation (>30 U/L). Methods: We analyzed data of adolescent participants (aged 12–19 years; N = 5586) in the National Health and Nutrition Examination Survey 1999–2004, a representative sample of the civilian noninstitutionalized US population. Results: The prevalence of elevated ALT levels (>30 U/L) was 7.4% among white adolescents, 11.5% among Mexican American adolescents, and 6.0% among black adolescents. Elevated ALT levels were prevalent in 12.4% of male subjects compared with 3.5% of female subjects. Multivariable associations with elevated ALT levels were found for sex (odds ratio [OR] male vs female, 7.7; 95% confidence interval [CI], 3.9–15.1), ethnicity (OR black vs white, 0.6; 95% CI, 0.3–1.3; OR Mexican American vs white, 1.6; 95% CI, 1.0–2.6), waist circumference (OR per 1 SD, 1.4; 95% CI, 1.0–2.0), and fasting insulin level (OR per 1 SD, 1.6; 95% CI, 1.2–2.1). Age, C-reactive protein levels, and triglyceride levels were also positively and socioeconomic position inversely associated with elevated ALT levels. The magnitude of associations were similar across ethnic groups. Conclusions: ALT level is associated with waist circumference and insulin resistance even in a young population. These characteristics could be utilized to identify adolescents who may benefit from screening for NAFLD, offering an opportunity to prevent disease progression at an early age. Background & Aims: Nonalcoholic fatty liver disease (NAFLD) is a common cause of liver disease in children and adolescents. The majority of studies of NAFLD in children have been in select populations of the clinically obese. Study aims were to estimate the prevalence of elevated alanine aminotransferase (ALT, as a marker of NAFLD) in a general contemporary adolescent population and to identify leading risk factors for ALT elevation (>30 U/L). Methods: We analyzed data of adolescent participants (aged 12–19 years; N = 5586) in the National Health and Nutrition Examination Survey 1999–2004, a representative sample of the civilian noninstitutionalized US population. Results: The prevalence of elevated ALT levels (>30 U/L) was 7.4% among white adolescents, 11.5% among Mexican American adolescents, and 6.0% among black adolescents. Elevated ALT levels were prevalent in 12.4% of male subjects compared with 3.5% of female subjects. Multivariable associations with elevated ALT levels were found for sex (odds ratio [OR] male vs female, 7.7; 95% confidence interval [CI], 3.9–15.1), ethnicity (OR black vs white, 0.6; 95% CI, 0.3–1.3; OR Mexican American vs white, 1.6; 95% CI, 1.0–2.6), waist circumference (OR per 1 SD, 1.4; 95% CI, 1.0–2.0), and fasting insulin level (OR per 1 SD, 1.6; 95% CI, 1.2–2.1). Age, C-reactive protein levels, and triglyceride levels were also positively and socioeconomic position inversely associated with elevated ALT levels. The magnitude of associations were similar across ethnic groups. Conclusions: ALT level is associated with waist circumference and insulin resistance even in a young population. These characteristics could be utilized to identify adolescents who may benefit from screening for NAFLD, offering an opportunity to prevent disease progression at an early age. Nonalcoholic fatty liver disease (NAFLD) consists of a range of liver abnormalities in the absence of other established causes of liver damage and is the commonest liver disorder in developed countries. In its mildest form (steatosis), it is characterized by an accumulation of triglycerides in hepatocytes; its more advanced form, nonalcoholic steatohepatitis, is characterized by liver cell injury.1Day C.P. Liver disease Non-alcoholic fatty liver disease: current concepts and management strategies.Clin Med. 2006; 6: 19-26Crossref PubMed Scopus (118) Google Scholar, 2Matteoni C.A. Younossi Z.M. Gramlich T. et al.Nonalcoholic fatty liver disease: a spectrum of clinical and pathological severity.Gastroenterology. 1999; 116: 1413-1419Abstract Full Text Full Text PDF PubMed Scopus (2750) Google ScholarWith the epidemic of childhood obesity and the recognition that the antecedents of obesity, insulin resistance, and NAFLD begin in early life, and that duration of NAFLD probably affects the likelihood of progression to more severe disease, concerns about NAFLD in children and adolescents have become a focus of public health researchers and practitioners.3Diehl A.M. Hepatic complications of obesity.Gastroenterol Clin North Am. 2005; 34: 45-61Abstract Full Text Full Text PDF PubMed Scopus (28) Google Scholar, 4Sathya P. Martin S. Alvarez F. Nonalcoholic fatty liver disease (NAFLD) in children.Curr Opin Pediatr. 2002; 14: 593-600Crossref PubMed Scopus (36) Google Scholar, 5Sokol R.J. The chronic disease of childhood obesity: the sleeping giant has awakened.J Pediatr. 2000; 136: 711-713Abstract Full Text PDF Scopus (81) Google Scholar The whole spectrum of NAFLD has been described in children,4Sathya P. Martin S. Alvarez F. Nonalcoholic fatty liver disease (NAFLD) in children.Curr Opin Pediatr. 2002; 14: 593-600Crossref PubMed Scopus (36) Google Scholar with nonalcoholic steatohepatitis first described in a pediatric population in the literature in the 1980s.6Moran J.R. Ghishan F.K. Halter S.A. et al.Steatohepatitis in obese children: a cause of chronic liver dysfunction.Am J Gastroenterol. 1983; 78: 374-377PubMed Google Scholar Presentation with and progression to cirrhosis in children have also been described.4Sathya P. Martin S. Alvarez F. Nonalcoholic fatty liver disease (NAFLD) in children.Curr Opin Pediatr. 2002; 14: 593-600Crossref PubMed Scopus (36) Google Scholar, 7Schwimmer J.B. Deutsch R. Rauch J.B. et al.Obesity, insulin resistance, and other clinicopathological correlates of pediatric nonalcoholic fatty liver disease.J Pediatr. 2003; 143: 500-505Abstract Full Text Full Text PDF PubMed Scopus (416) Google Scholar, 8Molleston J.P. White F. Teckman J. et al.Obese children with steatohepatitis can develop cirrhosis in childhood.Am J Gastroenterol. 2002; 97: 2460-2462Crossref PubMed Google ScholarThe diagnosis of NAFLD should be suspected when there is an increased level of alanine aminotransferase (ALT) and/or echogenic liver detected by ultrasonography together with the absence of any established causes of liver disease.9American Gastroenterological Association medical position statement: nonalcoholic fatty liver disease.Gastroenterology. 2002; 123: 1702-1704Abstract Full Text Full Text PDF PubMed Scopus (204) Google Scholar A definitive diagnosis, and determination of severity, requires a liver biopsy, which is not feasible in large-scale epidemiologic studies. In epidemiologic studies, ALT is the most useful biomarker of NAFLD because it is the liver enzyme most closely correlated to liver fat accumulation.10Westerbacka J. Corner A. Tiikkainen M. et al.Women and men have similar amounts of liver and intra-abdominal fat, despite more subcutaneous fat in women: implications for sex differences in markers of cardiovascular risk.Diabetologia. 2004; 47: 1360-1369Crossref PubMed Scopus (258) Google ScholarMost studies of the associations among obesity, insulin resistance, and NAFLD in children and adolescents have been in select populations of the clinically obese.4Sathya P. Martin S. Alvarez F. Nonalcoholic fatty liver disease (NAFLD) in children.Curr Opin Pediatr. 2002; 14: 593-600Crossref PubMed Scopus (36) Google Scholar, 11Burgert T.S. Taksali S.E. Dziura J. et al.Alanine aminotransferase levels and fatty liver in childhood obesity: associations with insulin resistance, adiponectin, and visceral fat.J Clin Endocrinol Metab. 2006; 91: 4287-4294Crossref PubMed Scopus (291) Google Scholar Studies of these associations in general population samples of children/adolescents are important because clinical populations may not be representative of associations in obese children who are not referred for medical care. For example, in the US National Health and Nutrition Examination Survey (NHANES) III survey, overweight and obesity were associated with elevated ALT levels in 12–18 year olds, but the prevalence of elevated levels in those who were obese (10%) was considerably lower than reported levels in clinically based studies (commonly between 25% and 30%).12Strauss R.S. Barlow S.E. Dietz W.H. Prevalence of abnormal serum aminotransferase values in overweight and obese adolescents.J Pediatr. 2000; 136: 727-733Abstract Full Text PDF PubMed Scopus (337) Google ScholarIn addition to associations with obesity and components of the metabolic syndrome,13Park H.S. Han J.H. Choi K.M. et al.Relation between elevated serum alanine aminotransferase and metabolic syndrome in Korean adolescents.Am J Clin Nutr. 2005; 82: 1046-1051PubMed Google Scholar other factors may be related to variation in liver enzyme levels in early life. Limited evidence of early life determinants of NAFLD suggests that NAFLD is more common in male subjects compared with female subjects and among adolescents of Hispanic origin (compared with white and black adolescents) and white compared with black adolescents.14Schwimmer J.B. McGreal N. Deutsch R. et al.Influence of gender, race, and ethnicity on suspected fatty liver in obese adolescents.Pediatrics. 2005; 115: e561-e565Crossref PubMed Scopus (204) Google Scholar Similar differences in the prevalence of NAFLD and elevated ALT levels by ethnic group have been consistently documented in adults.15Ioannou G.N. Boyko E.J. Lee S.P. The prevalence and predictors of elevated serum aminotransferase activity in the United States in 1999-2002.Am J Gastroenterol. 2006; 101: 76-82Crossref PubMed Scopus (252) Google Scholar, 16Lonardo A. Carani C. Carulli N. et al.“Endocrine NAFLD” a hormonocentric perspective of nonalcoholic fatty liver disease pathogenesis.J Hepatol. 2006; 44: 1196-1207Abstract Full Text Full Text PDF PubMed Scopus (101) Google Scholar, 17Browning J.D. Szczepaniak L.S. Dobbins R. et al.Prevalence of hepatic steatosis in an urban population in the United States: impact of ethnicity.Hepatology. 2004; 40: 1387-1395Crossref PubMed Scopus (2880) Google Scholar, 18Clark J.M. Brancati F.L. Diehl A.M. The prevalence and etiology of elevated aminotransferase levels in the United States.Am J Gastroenterol. 2003; 98: 960-967Crossref PubMed Scopus (1076) Google ScholarIntrauterine factors are important determinants of components of the metabolic syndrome,19Lawlor D.A. Davey Smith G. Ebrahim S. Life course influences on insulin resistance: findings from the British Women’s Heart and Health Study.Diabetes Care. 2003; 26: 97-103Crossref PubMed Scopus (64) Google Scholar and there is evidence that intrauterine factors are also important in liver development and function. According to the fetal programming hypothesis, fetal undernutrition in mid and late gestation results in maintaining the brain at the expense of the growth of the trunk, including the liver,20Barker D.J. Fetal origins of coronary heart disease.BMJ. 1995; 311: 171-174Crossref PubMed Scopus (2503) Google Scholar, 21Barker D.J. Martyn C.N. Osmond C. et al.Abnormal liver growth in utero and death from coronary heart disease.Br Med J. 1995; 310: 703-704Crossref PubMed Scopus (80) Google Scholar, 22Barker D.J. Fetal programming of coronary heart disease.Trends Endocrinol Metab. 2002; 13: 364-368Abstract Full Text Full Text PDF PubMed Scopus (522) Google Scholar which may result in permanent alterations to liver function.23Gruenwald P. Chronic fetal distress and placental insufficiency.Biol Neonat. 1963; 5: 215-265Crossref PubMed Google Scholar In a cohort of older British women, inverse associations between birth weight and levels of ALT, γ-glutamyltransferase (GGT), and alkaline phosphatase were found even when controlling for potential confounders.24Fraser A, Ebrahim S, Davey Smith G, et al. The association between birthweight and adult markers of liver damage and function. Paediatr Perinat Epidemiol (in press).Google ScholarThe aims of this study were to estimate the prevalence of ALT elevation in a general adolescent population, to examine whether a range of early life characteristics are associated with elevated ALT levels in adolescence, and to determine the magnitude and strength of these associations in adolescent participants (defined as age 12–19 years) in NHANES 1999–2004.Materials and MethodsNHANES is a complex, multistage probability sample of the civilian noninstitutionalized population of the United States.25Centers for Disease Control and Prevention (CDC)National Center for Health Statistics (NCHS)National Health and Nutrition Examination Survey. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 2001–2004, Hyattsville, MD2007Google Scholar Since 1999, NHANES has become a continuous survey. Data from 3 cross-sectional surveys, 1999–2000, 2001–2002, and 2003–2004, are included here. For our main analyses, we included all participants aged 12–19 years. Blood samples were available for participants aged 12 years and older, and information on birth weight and maternal smoking during pregnancy was available for those aged up to 15 years. Information on ethnicity and family income were obtained from a family member. A poverty-income ratio (PIR) was calculated for each family based on self-reported family income in relation to poverty threshold, family size, and calendar year. Values <1 are below the official poverty threshold, while PIR values ≥1 indicate income above the poverty level (range, 0–5). Data on age, birth weight, and maternal smoking during pregnancy were obtained from an in-person administered questionnaire. The questionnaire was completed by a family member at least 18 years old who was the most knowledgeable about the child (usually the mother or father).Participants underwent a medical examination in which weight, standing height, and waist circumference were measured in a standardized fashion. Seated resting blood pressure was measured by a physician, calculated as the average of available measurements (minimum, 1; maximum, 4), excluding the first measurement if more than one was available. In addition, subjects aged 12 years and older were asked to provide a blood sample that was analyzed for liver enzyme levels.ALT, aspartate aminotransferase (AST), and GGT levels were measured using an enzymatic rate method. High-density lipoprotein cholesterol (HDL-C) level was measured by direct immunoassay. A high-sensitivity C-reactive protein assay was conducted using latex-enhanced nephelometry. Fasting insulin, glucose, and triglyceride levels were available for a random subgroup of adolescents who attended the morning clinics (n = 2655). Fasting insulin level was measured in 1999–2000 using a radioimmunoassay and in subsequent surveys using a 2-site immunoenzymometric assay (Pharmacia method in 2001–2002 and a Tosoh method in 2003–2004). A linear regression model was used to adjust 2001–2002 values to 2003–2004 values. The distribution of insulin values in all 3 surveys was comparable. Plasma glucose level was measured using an enzyme hexokinase method. The homeostasis model assessment of insulin resistance (HOMA-R) was calculated as the product of fasting glucose (mmol/L) and insulin (μU/mL) divided by the constant 22.5.26Matthews D.R. Hosker J.P. Rudenski A.S. et al.Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man.Diabetologia. 1985; 28: 412-419Crossref PubMed Scopus (25088) Google Scholar Triglyceride levels were measured using a timed end point method. Tests for hepatitis B surface antigen were performed using a sandwich radioimmunoassay (Abbott Laboratories, Abbott Park, IL), and specimens were tested for antibody to hepatitis C using direct solid-phase enzyme immunoassay with the anti–hepatitis C virus screening enzyme-linked immunosorbent assay. Positive specimens were repeated in duplicate according to the same procedure. Repeatedly positive specimens were tested supplementally using the Chiron RIBA Processor System (Chiron Corp, Inc, Emeryville, CA).27Centers for Disease Control and Prevention (CDC)National Center for Health Statistics (NCHS)National Health and Nutrition Examination Laboratory Protocols. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 2001–2002, Hyattsville, MD2007Google Scholar, 28Centers for Disease Control and Prevention (CDC)National Center for Health Statistics (NCHS)National Health and Nutrition Examination Laboratory Protocols. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 2003–2004, Hyattsville, MD2007Google ScholarStatistical AnalysisIn all analyses, both the sampling probability (via weights) and cluster effects due to the correlation of observations among subjects from a given area were accounted for by using the “svy” procedures in Stata version 9 (Stata Corp, College Station, TX).ALT, AST, GGT, C-reactive protein, fasting insulin and HOMA-R, and triglyceride levels were natural log transformed to normalize distributions. Fasting insulin and HOMA-R are nearly perfectly correlated (Pearson’s r = 0.99). We report results for fasting insulin; however, all results are the same if fasting insulin is replaced with HOMA-R. Logistic regression was used to assess associations of characteristics with elevated ALT level, defined as ALT level >30 U/L, the value previously used to define elevated ALT level in adolescents in NHANES III (1988–1994).12Strauss R.S. Barlow S.E. Dietz W.H. Prevalence of abnormal serum aminotransferase values in overweight and obese adolescents.J Pediatr. 2000; 136: 727-733Abstract Full Text PDF PubMed Scopus (337) Google Scholar This threshold had a sensitivity of 0.92 for detecting the fatty-fibrotic pattern proven by ultrasonography among obese children.29Tazawa Y. Noguchi H. Nishinomiya F. et al.Serum alanine aminotransferase activity in obese children.Acta Paediatr. 1997; 86: 238-241Crossref PubMed Scopus (129) Google ScholarWe conducted a secondary analysis using 40 U/L as a threshold defining elevated ALT level for the sake of comparison, because this and similar thresholds have been previously used in studies in adolescents13Park H.S. Han J.H. Choi K.M. et al.Relation between elevated serum alanine aminotransferase and metabolic syndrome in Korean adolescents.Am J Clin Nutr. 2005; 82: 1046-1051PubMed Google Scholar, 30Schwimmer J.B. McGreal N. Deutsch R. et al.Influence of gender, race, and ethnicity on suspected fatty liver in obese adolescents.Pediatrics. 2005; 115: e561-e565Crossref PubMed Scopus (204) Google Scholar and to indicate NAFLD in adults.15Ioannou G.N. Boyko E.J. Lee S.P. The prevalence and predictors of elevated serum aminotransferase activity in the United States in 1999-2002.Am J Gastroenterol. 2006; 101: 76-82Crossref PubMed Scopus (252) Google Scholar, 18Clark J.M. Brancati F.L. Diehl A.M. The prevalence and etiology of elevated aminotransferase levels in the United States.Am J Gastroenterol. 2003; 98: 960-967Crossref PubMed Scopus (1076) Google Scholar, 31Prati D. Taioli E. Zanella A. et al.Updated definitions of healthy ranges for serum alanine aminotransferase levels.Ann Intern Med. 2002; 137: 1-10Crossref PubMed Scopus (1104) Google Scholar In addition, linear regression was used to assess associations of characteristics with ALT as a continuous variable; because in general population studies of adults, associations of liver enzyme levels with body mass index (BMI) and components of the metabolic syndrome are linear and do not demonstrate threshold effects.32Sattar N. Scherbakova O. Ford I. et al.Elevated alanine aminotransferase predicts new-onset type 2 diabetes independently of classical risk factors, metabolic syndrome, and C-reactive protein in the West of Scotland Coronary Prevention Study.Diabetes. 2004; 53: 2855-2860Crossref PubMed Scopus (286) Google Scholar, 33Perry I.J. Wannamethee S.G. Shaper A.G. Prospective study of serum gamma-glutamyltransferase and risk of NIDDM.Diabetes Care. 1998; 21: 732-737Crossref PubMed Scopus (281) Google ScholarOf 7205 12 to 19-year-old adolescents in the 1999–2004 NHANES, 6339 (88%) had data on ALT levels. Of these, 5 adolescents were positive for hepatitis B surface antigen and 12 were positive or indeterminate for hepatitis C antibody and were subsequently excluded from all analyses. Complete data on all measurements that did not require fasting, except for birth weight and maternal smoking during pregnancy, were available for 5586 adolescents (78% of all adolescents). Data on birth weight and maternal smoking during pregnancy were available only for adolescents 12–15 years old (n = 2686; 95% of 12–15 year olds with complete data on all other variables). Of the 5586 adolescents included in this study, 2655 attended the morning clinics and also had data on fasting insulin, fasting glucose, and triglyceride levels. All analyses included adolescents for whom complete data on all variables were available.Because fasting measurements were only available for a random subsample of adolescents, multivariable analyses excluding terms for fasting measurements were constructed for (1) all adolescents and (2) adolescents for whom fasting measurements were available. This allowed us to compare associations between the whole sample and the subgroup of adolescents who attended the morning clinics. Fasting measurements were then added into the model for the fasting subsample, and all 3 models are displayed.ResultsElevated ALT level (defined as ALT level >30 U/L) was present in 8.0% of the study population. Mean ALT level was 19.3 U/L (SD, 15.6; interquartile range, 5–196; range, 5–526). Characteristics of the study population by ALT elevation are presented in Table 1. Among white adolescents the prevalence was 7.4%, among Mexican Americans it was 11.5%, among black adolescents it was 6.0%, and among adolescents classified as “other” (other race, including multiracial, other Hispanic subjects) it was 10.6% (Figure 1). Elevated ALT level was prevalent among 12.4% of male subjects compared with 3.5% of female subjects. Adolescents with an elevated ALT level were older and had higher waist circumference, BMI, systolic blood pressure, C-reactive protein level, fasting glucose level, insulin level, triglyceride level, AST level, and GGT level. They were also more likely to be poor and had lower HDL-C levels.Table 1Characteristics by ALT Levels (n = 5586, Weighted Sample N = 25,991,934)CharacteristicALT ≤30 U/L (n = 5095)ALT >30 U/L (n = 491)P valueContinuous variablesaMeans (SE). Age (y)15.4 (0.1)16.3 (0.2)< .01 Birth weightbGeometric mean. (kg)3.4 (0.0)3.4 (0.1).64 Waist circumference (cm)79.7 (0.4)95.0 (1.1)< .01 BMI (kg/m2)22.8 (0.14)28.3 (0.4)< .01 Systolic blood pressure (mm Hg)108.1 (0.28)115.9 (1.0)< .01 HDL-C level (mmol/L)1.3 (0.3)1.2 (0.3)< .01 C-reactive protein levelbGeometric mean. (mg/dL)0.05 (1.0)0.13 (1.1)< .01 Fasting glucose levelcFasting subsample (n = 2655). (mmol/L)5.1 (0.0)5.4 (0.1).03 Fasting insulin levelbGeometric mean.cFasting subsample (n = 2655). (μU/L)8.1 (1.0)14.2 (1.1)< .01 Triglyceride levelbGeometric mean.cFasting subsample (n = 2655). (mmol/L)0.9 (1.0)1.3 (1.1)< .01 AST level (U/L)bGeometric mean.21.7 (1.0)34.3 (1.0)< .01 GGT level (U/L)bGeometric mean.13.1 (1.0)25.8 (1.0)< .01Categorical variables Sex (%)< .01 Male87.612.4 Female96.53.5 Ethnicity (%)< .01 White92.67.4 Black94.06.0 Mexican American88.511.5 Other89.410.6 Maternal smoking in pregnancydAvailable for adolescents 12–15 years old (n = 2686). (%).20 Yes93.36.7 No94.85.2 Beneath poverty threshold (%).03 Yes91.88.2 No90.010.0a Means (SE).b Geometric mean.c Fasting subsample (n = 2655).d Available for adolescents 12–15 years old (n = 2686). Open table in a new tab The multivariable associations of characteristics and nonfasting measurements with elevated ALT level were comparable for the full sample and the fasting subsample, although confidence intervals (CIs) were wider in the smaller subsample (Table 2). Elevated ALT level was strongly associated with being older, male, and of Mexican American origin. Elevated ALT level was also associated with greater waist circumference, C-reactive protein level, and systolic blood pressure and more weakly with being poor. When fasting measurements were added, evidence was found for positive associations of fasting insulin and triglyceride levels with elevated ALT level, but not of fasting glucose level. Because waist circumference was more strongly associated with ALT level than BMI, this measure was used in the multivariable models as a marker of adiposity and BMI was not simultaneously included in models containing waist circumference. Replacing waist circumference with BMI did not alter results, except that there was stronger evidence that black ethnic origin was associated with less ALT elevation compared with white ethnic origin (odds ratio [OR], 0.5; 95% CI, 0.3–1.0; P = .06 in the model including BMI compared with OR, 0.7; 95% CI, 0.3–1.3; P = .20 in the model including waist circumference). When birth weight and maternal smoking during pregnancy were included in the models, there was no evidence of associations with elevated ALT level. Furthermore, the associations of other characteristics with elevated ALT level were not substantially altered. Therefore, models including birth weight and maternal smoking (available only for a subgroup of adolescents 12–15 years old) are not presented.Table 2Multivariable Analysis of Associations With Elevated ALT Level (>30 U/L)VariableComplete data sample (N = 5586)Fasting subsample, model 1 (n = 2655)Fasting subsample, model 2 (n = 2655)OR (95% CI)P valueOdds ratio (95% CI)P valueOdds ratio (95% CI)P valueAge (y)1.1 (1.0–1.2).031.2 (1.1–1.3)< .011.3 (1.1–1.4)< .01Sex (male vs female)3.8 (2.6–5.4)< .016.7 (3.3–13.6)< .017.7 (3.9–15.1)< .01Ethnicity (vs white) Black0.9 (0.5–1.2).230.7 (0.3–1.3).210.6 (0.3–1.3).20 Mexican1.5 (1.1–2.0)< .011.7 (1.0–2.8).031.6 (1.0–2.6).24 Other1.7 (1.0–3.0).072.1 (0.8–5.9).152.0 (0.8–5.1).53Waist circumference (cm)1.05 (1.0–1.1)< .011.05 (1.04–1.11)< .011.05 (1.02–1.09).04Systolic blood pressure (mm Hg)1.02 (1.00–1.04).041.02 (0.99–1.05).271.01 (0.98–1.04).58HDL-C level (mmol/L)0.7 (0.4–1.3).770.4 (0.1–1.3).120.6 (0.2–1.7).33C-reactive protein levelaOn a log scale. (mg/dL)1.2 (1.1–1.3)< .011.2 (1.0–1.4).051.2 (1.0–1.4).05PIR (over vs under poverty threshold)0.7 (0.5–1.0).080.8 (0.5–1.6).100.8 (0.5–1.5).07Fasting glucose level (mmol/L)1.0 (0.8–1.2).53Fasting insulin levelaOn a log scale. (μU/L)1.9 (1.3–2.8).09Triglyceride levelaOn a log scale. (mmol/L)1.5 (0.9–2.4).10a On a log scale. Open table in a new tab The multivariable model including the fasting measures (fasting subsample, model 2, Table 2) was rerun entering standardized continuous variables (ie, each value of a given characteristic is divided by the standard deviation of that characteristic). This was performed to compare the strength of the association of each characteristic with elevated ALT level. The strongest associations with elevated ALT level were for sex (OR male vs female, 7.7), ethnicity (OR black vs white adolescents, 0.6; OR Mexican American vs white adolescents, 1.6), age (OR per 1 SD change, 1.7; SD, 2.3 years), fasting insulin level (OR per 1 SD change, 1.6; SD, 2.0 μU/L), waist circumference (OR per 1 SD change, 1.4; SD, 14.5 cm), C-reactive protein level (OR per 1 SD change, 1.3; SD, 4.2 mg/dL), and triglyceride level (OR per 1 SD change, 1.2; SD, 1.6 mmol/L).We also repeated the main analysis defining elevated ALT level as >40 U/L. The prevalence of ALT level >40 U/L in the whole study population was 3.6%. Among male subjects it was 5.6%, and among female subjects it was 1.6%. Among white adolescents it was 3.1%, among Mexican Americans it was 6.1%, and among black adolescents it was 2.3%. Table 3 presents the multivariable model for the fasting subsample (equivalent to fasting subsample 2; Table 2). Results were similar to those obtained for ALT level >30 U/L; male sex, Mexican ethnic origin, greater waist circumference, fasting insulin level, and triglyceride level were all associated with ALT level >40 U/L. C-reactive protein level and poverty were no longer associated with the outcome using this higher threshold.Table 3Multivariable Analysis of Associations With ALT Level >40 U/LVariableFasting subsample (n = 2655)OR (95% CI)P valueAge (y)1.3 (1.1–1.5)< .01Sex (male vs female)5.0 (1.6–15.4)< .01Ethnicity (vs white) Black0.8 (0.3–2.0).58 Mexican2.9 (1.5–5.7)< .01 Other3.0 (1.0–8.9).05Waist circumference (cm)1.03 (1.01–1.05).01Systolic blood pressure (mm Hg)1.0 (1.0–1.0).90HDL-C level (mmol/L)0.6 (0.1–2.6).50C-reactive protein levelaOn a log scale. (mg/dL)1.2 (0.9–1.7).17PIR (over vs under poverty threshold)1.5 (0.7–3.2).31Fasting glucose level (mmol/L)1.0 (0.8–1.3).70Fasting insulin levelaOn a log scale. (μU/L)2.2 (1.2–4.2).01Triglyceride levelaOn a log scale. (mmol/L)1.8 (1.0–3.4).07a On a log scale. Open table in a new tab We also repeated the analysis including fasting measures treating ALT as a continuous variable, as opposed to a binary, elevated versus not elevated variable. In this analysis, black ethnic origin was associated with lower mean ALT level (compared