Short Sleep Duration Is Associated With Abnormal Serum Aminotransferase Activities and Nonalcoholic Fatty Liver Disease

The proportion of adults in the United States who report sleeping 6 or fewer hours per night increased from 22.3% in 1985 to 29.2% in 2012 despite recommendations that adults maintain sleep duration of 7 hours or more for optimal health.1Ford E.S. Cunningham T.J. Croft J.B. Trends in self-reported sleep duration among US adults from 1985 to 2012.Sleep. 2015; 38: 829-832Google Scholar, 2Consensus Conference P. Watson N.F. Badr M.S. et al.Joint Consensus Statement of the American Academy of Sleep Medicine and Sleep Research Society on the recommended amount of sleep for a healthy adult: methodology and discussion.J Clin Sleep Med. 2015; 11: 931-952Google Scholar Sleep deprivation increases metabolic risks known to be associated with nonalcoholic fatty liver disease (NAFLD). We studied the association between duration and quality of sleep and indicators of NAFLD in the US general population. Of the 22,692 adult (>20 y) participants in the National Health and Nutrition Examination Survey (NHANES) for 2005 to 2012, subjects with significant alcohol consumption, viral hepatitis, pregnancy, or malignancy, or with missing data (n = 1654) on sleep duration/quality, serum alanine aminotransferase (ALT), or viral hepatitis were excluded. The final study sample consisted of 17,245 adults. NHANES divided sleep duration into 5 or fewer, 6, 7, 8, and 9 or more hours per night. Sleep quality, available only in NHANES 2005 to 2006 and 2007 to 2008, was categorized into good, moderate, and poor. The primary outcome variable was a serum ALT level greater than 30 IU/L for men and greater than 19 IU/L for women.3Ruhl C.E. Everhart J.E. Elevated serum alanine aminotransferase and gamma-glutamyltransferase and mortality in the United States population.Gastroenterology. 2009; 136: 477-485 e11Google Scholar The secondary outcome variable was the US fatty liver index, which consisted of age, race/ethnicity, waist circumference, γ-glutamyltransferase level, fasting glucose level, and insulin level, with a threshold of 30 to diagnose NAFLD.4Ruhl C.E. Everhart J.E. Fatty liver indices in the multiethnic United States National Health and Nutrition Examination Survey.Aliment Pharmacol Ther. 2015; 41: 65-76Google Scholar Given the sampling design of NHANES, appropriate sample weights (primary sampling unit, strata, and weighting) were used to reconstitute population level data for the entire United States. The association between sleep duration/quality and abnormal ALT level was examined using the multivariable logistic regression analyses using SAS 9.2 (SAS institute, Cary, NC) and STATA 13.0 (StataCorp, College Station, TX). The NHANES data projected that 15.9% of US adults slept for 5 or fewer hours per night, 23.8% slept for 6 hours, 27.1% slept for 7 hours, 26.3% slept for 8 hours, and 7.0% slept for 9 or more hours. In descriptive analyses (Table 1), certain variables (eg, diabetes, hypertension, blood levels of glucose and triglycerides, education, and socioeconomic status) had a U- or inverse U-shaped association with sleep duration, whereas other variables (eg, body mass index, waist circumference) showed a linear relationship, with progressively shorter sleep duration associated with poor outcomes. Serum ALT activities belonged in the latter group (P for trend < .001).Table 1Association of Various Characteristics of NHANES Participants With Sleep Duration (n = 17,245)Sleep duration, h≤5 (n = 2739)6 (n = 4104)7 (n = 4669)8 (n = 4530)≥9 (n = 1203)Descriptive dataaData are expressed as the means ± SD or proportion ± SE. Age,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. y45.5 ± 17.045.1 ± 15.545.2 ± 14.746.3 ± 17.147.9 ± 21.1 Male sex,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. %48.5 ± 1.1452.3 ± 0.9850.4 ± 0.8346.2 ± 0.8039.2 ± 1.47 Hypertension,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. %32.6 ± 0.9728.5 ± 0.8825.5 ± 0.7628.9 ± 1.0332.0 ± 1.66 Diabetes,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. %13.3 ± 0.739.7 ± 0.657.0 ± 0.519.4 ± 0.5211.5 ± 1.00 Ethnicity,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. %Mexican American7.10 ± 0.808.48 ± 0.948.31 ± 0.9110.29 ± 1.009.74 ± 1.39Other Hispanic6.73 ± 1.005.35 ± 0.725.09 ± 0.675.16 ± 0.654.72 ± 0.91Non-Hispanic white57.42 ± 2.4265.92 ± 1.9672.87 ± 1.6769.43 ± 1.8469.93 ± 2.31Non-Hispanic black20.82 ± 1.6913.23 ± 1.187.11 ± 0.688.73 ± 0.739.80 ± 1.18Asian/other7.93 ± 0.887.01 ± 0.626.63 ± 0.636.39 ± 0.755.82 ± 0.79 Smoking,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. %Never48.0 ± 1.3954.4 ± 1.1558.4 ± 1.0457.9 ± 1.1354.4 ± 1.93Ex-smoker30.8 ± 1.2523.4 ± 1.0216.4 ± 0.8217.9 ± 0.8224.3 ± 1.76Current smoker21.2 ± 0.8922.2 ± 0.7125.2 ± 0.8224.2 ± 0.9721.2 ± 1.62 Body mass index,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. kg/m229.7 ± 8.6928.8 ± 7.2228.1 ± 6.1527.9 ± 7.1927.7 ± 8.47 Waist circumference,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. cm96.0 ± 28.995.2 ± 24.794.3 ± 21.193.4 ± 24.090.9 ± 29.7 Higher education,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. %76.7 ± 0.9583.2 ± 0.9785.1 ± 0.9481.2 ± 0.9575.5 ± 1.69 Marital status,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. %57.1 ± 1.3464.1 ± 1.3269.2 ± 0.9364.1 ± 1.1755.8 ± 2.04 Poverty,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. %19.1 ± 1.0714.0 ± 0.8310.5 ± 0.6714.9 ± 0.9018.9 ± 1.69 Total cholesterol level,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. mg/dL196.5 ± 47.2196.3 ± 40.2197.2 ± 37.9196.7 ± 41.5193.5 ± 43.7 Triglyceride level,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. mg/dL147.0 ± 151.9133.4 ± 103.9128.0 ± 93.5132.2 ± 98.2136.4 ± 93.3 HDL cholesterol level,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. mg/dL51.3 ± 17.352.1 ± 15.453.2 ± 14.453.4 ± 16.252.9 ± 16.4 Fasting glucose level, mg/dL106.6 ± 37.0104.9 ± 32.3103.2 ± 25.8105.0 ± 31.7105.6 ± 34.7 HbA1c level,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. %5.70 ± 1.235.57 ± 0.915.48 ± 0.765.56 ± 0.945.58 ± 1.00 Aspartate aminotransferase level,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. IU/L26.0 ± 20.125.3 ± 10.825.2 ± 13.325.1 ± 10.824.9 ± 10.5 ALT level,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. IU/L26.1 ± 18.625.8 ± 16.325.5 ± 18.824.8 ± 14.923.8 ± 18.6 γ-glutamyltransferase level,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. IU/L32.6 ± 66.127.2 ± 31.024.9 ± 25.426.5 ± 32.825.0 ± 26.5 Fasting insulin level,bP < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate. pmol/L86.4 ± 80.778.5 ± 68.375.2 ± 67.874.3 ± 67.680.5 ± 96.5 CRP level, mg/dL (n = 13,027)0.465 ± 0.8970.400 ± 0.7630.339 ± 0.6740.364 ± 0.6310.512 ± 0.977Multivariable analyses, OR (95% CI) Association with abnormal ALT levelcThe multivariable model included age, sex, ethnicity, education level, marital status, economic status, body mass index, waist circumference, smoking status, diabetes, hypertension, and total cholesterol.1.35 (1.11–1.65)1.24 (1.06–1.46)1.17 (0.97–1.40)1.15 (0.95–1.39)Reference Association with NAFLDdThe multivariable model included age, sex, education level, marital status, economic status, smoking status, diabetes, hypertension, cholesterol level, and triglyceride level.1.45 (1.08–1.95)1.33 (1.04–1.70)1.27 (0.96–1.67)1.00 (0.76–1.31)ReferenceSensitivity analyses (association with abnormal ALT level), OR (95% CI) Incorporate insulin resistance and triglyceridescThe multivariable model included age, sex, ethnicity, education level, marital status, economic status, body mass index, waist circumference, smoking status, diabetes, hypertension, and total cholesterol.1.35 (1.01–1.80)1.22 (0.96–1.54)1.15 (0.91–1.45)1.22 (0.94–1.58)Reference Incorporate serum CRPcThe multivariable model included age, sex, ethnicity, education level, marital status, economic status, body mass index, waist circumference, smoking status, diabetes, hypertension, and total cholesterol.1.30 (1.05–1.60)1.20 (1.01–1.44)1.12 (0.92–1.38)1.11 (0.91–1.35)Reference Use higher NHANES-defined ALT cut-off value (males > 40; females > 31)cThe multivariable model included age, sex, ethnicity, education level, marital status, economic status, body mass index, waist circumference, smoking status, diabetes, hypertension, and total cholesterol.1.16 (0.84–1.62)0.93 (0.69–1.25)0.86 (0.64–1.16)0.88 (0.66–1.18)ReferenceNOTE. There were 1654 subjects with missing data who were excluded from the analysis. Excluded subjects were similar in age, sex, and smoking status to those included. However, they were more likely to be non-white, less well educated, single, and poor compared with those in the analysis.NAFLD is defined by the US fatty liver index, which is calculated as follows: (e-0.8073 × non-Hispanic black + 0.3458 × Mexican American + 0.0093 × age + 0.6151× loge (γ-glutamyltransferase) + 0.0249 × waist circumference + 1.1792 × loge (insulin) + 0.8242 × loge (glucose) – 14.7812)/ (1+ e-0.8073 × non-Hispanic black + 0.3458 × Mexican American + 0.0093 × age + 0.6151× loge (γ-glutamyltransferase) + 0.0249 × waist circumference + 1.1792 × loge (insulin) + 0.8242 × loge (glucose) – 14.7812) × 100.Analyses for fasting glucose, triglyceride, and insulin concentrations for the US fatty liver index and for insulin resistance were limited to 8419 participants who were examined after an overnight fast of a minimum of 8 hours.CI, confidence interval; CRP, C-reactive protein; HbA1c, hemoglobin A1c; HDL, high density lipoprotein; OR, odds ratio.a Data are expressed as the means ± SD or proportion ± SE.b P < .05 (at least 1 of the 5 groups being significantly different from the rest) by the chi-square test or the analysis of variance test as appropriate.c The multivariable model included age, sex, ethnicity, education level, marital status, economic status, body mass index, waist circumference, smoking status, diabetes, hypertension, and total cholesterol.d The multivariable model included age, sex, education level, marital status, economic status, smoking status, diabetes, hypertension, cholesterol level, and triglyceride level. Open table in a new tab NOTE. There were 1654 subjects with missing data who were excluded from the analysis. Excluded subjects were similar in age, sex, and smoking status to those included. However, they were more likely to be non-white, less well educated, single, and poor compared with those in the analysis. NAFLD is defined by the US fatty liver index, which is calculated as follows: (e-0.8073 × non-Hispanic black + 0.3458 × Mexican American + 0.0093 × age + 0.6151× loge (γ-glutamyltransferase) + 0.0249 × waist circumference + 1.1792 × loge (insulin) + 0.8242 × loge (glucose) – 14.7812)/ (1+ e-0.8073 × non-Hispanic black + 0.3458 × Mexican American + 0.0093 × age + 0.6151× loge (γ-glutamyltransferase) + 0.0249 × waist circumference + 1.1792 × loge (insulin) + 0.8242 × loge (glucose) – 14.7812) × 100. Analyses for fasting glucose, triglyceride, and insulin concentrations for the US fatty liver index and for insulin resistance were limited to 8419 participants who were examined after an overnight fast of a minimum of 8 hours. CI, confidence interval; CRP, C-reactive protein; HbA1c, hemoglobin A1c; HDL, high density lipoprotein; OR, odds ratio. In multivariable analyses, shorter sleep duration had a progressively higher odds of abnormal ALT level. For example, compared with the reference group with sleep duration of 9 or more hours, subjects with a sleep duration of 5 or fewer hours were 35% more likely to have an abnormal ALT level. In contrast, sleep quality showed no association. In sensitivity analyses incorporating insulin resistance and fasting triglyceride levels, using different ALT cut-off values, or incorporating serum C-reactive protein levels, the effects of sleep duration on ALT level remained unchanged. Finally, sleep duration was associated significantly with NAFLD as defined by the US fatty liver index: adults who slept 5 or fewer hours had a 45% higher odds of having NAFLD compared with those who slept 9 or more hours (P for trend = .001). Approximately 70.1 million US adults slept for 6 or fewer hours in 2012, compared with 38.6 million in 1985.1Ford E.S. Cunningham T.J. Croft J.B. Trends in self-reported sleep duration among US adults from 1985 to 2012.Sleep. 2015; 38: 829-832Google Scholar Based on these data, we estimate 32 million sleep-deprived Americans have abnormal ALT levels and 28.6 million have NAFLD. Because we excluded major etiologies of chronic liver disease, abnormal ALT levels seen in these data likely are linked to NAFLD. Despite limitations related to missing data, we believe this is a robust, US population–based analysis about sleep health and NAFLD. Prior studies, including a recent meta-analysis, reported inconsistent or nonsignificant associations.5Shen N. Wang P. Yan W. Sleep duration and the risk of fatty liver disease: a systematic review and meta-analysis.Sci Rep. 2016; 6: 31956Google Scholar These associations may be driven by obesity-related insulin resistance and proinflammatory milieu associated with poor sleep.6Beccuti G. Pannain S. Sleep and obesity.Curr Opin Clin Nutr Metab Care. 2011; 14: 402-412Google Scholar, 7Liu R. Liu X. Zee P.C. et al.Association between sleep quality and C-reactive protein: results from national health and nutrition examination survey, 2005-2008.PLoS One. 2014; 9: e92607Google Scholar In addition, sleep deprivation has been hypothesized to disrupt the hypothalamic–pituitary–adrenal axis, which may contribute to NAFLD.8Balbo M. Leproult R. Van Cauter E. Impact of sleep and its disturbances on hypothalamo-pituitary-adrenal axis activity.Int J Endocrinol. 2010; 2010: 759234Google Scholar However, it also is possible that sleep duration has an independent association with serum ALT level in subjects with or without NAFLD or other chronic liver disease. For example, some of the metabolic abnormalities commonly associated with NAFLD have a U-shaped relation with sleep (ie, longer-duration sleep being unhealthy), whereas the relationship between sleep duration and ALT level was linear. The association between long sleep duration and metabolic disorders may represent residual confounding or reverse causation, highlighting the limitation of this observational study; firm causal relationships may not be inferred and prospective interventional studies are needed to confirm the association.2Consensus Conference P. Watson N.F. Badr M.S. et al.Joint Consensus Statement of the American Academy of Sleep Medicine and Sleep Research Society on the recommended amount of sleep for a healthy adult: methodology and discussion.J Clin Sleep Med. 2015; 11: 931-952Google Scholar In summary, optimal sleep duration (≥7 h) is associated with a lower likelihood of abnormal ALT levels and NAFLD, independent of metabolic risk factors. Improving sleep health may provide a novel opportunity for intervention in patients with abnormal ALT levels and/or NAFLD.