Original Article

Prevalence of Fatty Liver in a General Population ofOkinawa, Japan

Hideyuki NOMURA, Seizaburo KASHIWAGI, Jun HAYASHI,

Wataru KAJIYAMA, Shunichi TANI* and Masashi GOTO*

A total of 2,574 residents in Yaeyama District of Okinawa, Japan, were investigated using real timeultrasonography to determine the real prevalence of fatty liver in the general population and to define itsassociated factors. Overall prevalence of fatty liver was 14.0%. Prevalence of fatty liver in persons under19 years old was only 1.2%, and increased with age to a maximumin persons 40-49 years of age and then

decreased. For persons over 20 years old, obesity index and serum levels of triglyceride and totalcholesterol were measured, and alcohol consumption was asked. Prevalence of fatty liver was significantlyhigher in drinkers than non-drinkers (p <0.01), and increased with alcohol consumption. Furthermore, inpersons not suffering from obesity prevalence of fatty liver was significantly higher in drinkers than innon-drinkers (p < 0.001). The results of logistic regression analysis indicated that obesity and elevatedserum triglyceride level in both sexes, and alcohol in males were significant predictors of fatty liver. Inconclusion, prevalence of fatty liver increased with age to a maximumin persons 40-49 years of age andoverall was 14.0%. Obesity was the strongest associated factor in both sexes and in males alcohol was alsoa strong factor.

Key Words: Alcohol, Fatty liver, Obesity, Triglyceride, Total cholesterol, Ultrasonography

In the past, fatty infiltration of the liver couldonly be diagnosed by liver biopsy. Therefore, thereare few extensive studies on the epidemiology of

f

atty liver in healthy residents.

With the advent of ultrasonography, fatty livercan be easily detected. When using ultrasono-graphy we can see the fat tissue has a moreincreased echogenicity than the normal liverparenchyma1"4^. In order to define real pre-

valence of fatty liver in the general population andits associated factors, using real time ultrasono-graphy, we carried out scanning of the fattychanges of the liver in healthy residents, in the

Yaeyama District of Okinawa, Japan.

MATERIALS AND METHODSPopulation studied

Residents of the Yaeyama District (IriomoteIsland, HaterumaIsland, Taketomi Island, Kohama

Island, Kuroshima Island) of Okinawa weresurveyed. Okinawa is located in the subtropical

zoneabout 1,000 km south of the main islands ofJapan, which lie in the temperate zone.A total of 2,574 healthy individuals who came

for free health examinations were subjected in1984. These examinations were announced by

d

istributing written notices to households.

The age- and sex-specific estimated population,numbers, and the percentages of individuals tested

are shown in table 1. Both the numbers andpercentages of females and males that were tested

F

rom the First Department of Internal Medicine, Faculty of Medicine,Kyushu University, Fukuoka and *Shionogi Kaiseki Center, Shionogi Co., Ltd. Osaka

Received for publication September 16, 1987

R

eprint request to: Hideyuki Nomura, MD,Yufuin Koseinenkin- Hospital, 252 Kawaminami, Yufuin-cho, Ooita-gun, Ooita, 879-5 1, Japan

142 JpnJMedVol27,No 2 (May, 1988)

Prevalence of Fatty liver in Okinawa

T

able 1. Age- and sex-specific estimated population and numbers and percentageof individuals tested, in the Yaeyama District, Okinawa, Japan.

Males Females Total

Tested Tested TestedAge Estimated Estimated Estimated

(years) population n % population n % population n %

0-9 267 264 98.9 202 197 97.5 469 4-61 98.3

10-19 145 132 91-0 179 U9 83-2 324 281 86.7

20-29 155 125 80.6 132 104 78.8 287 229 79.8

30-39 178 163 91.6 145 129 89.0 323 292 90.4

40-49 145 122 84-1 135 120 88.9 280 242 86.4

50-59 221 191 86.4 272 249 91.5 493 440 89.2

60-69 185 167 90.3 202 189 93.6 387 356 92.0

70+ 137 107 78.1 188 166 88.3 325 273 84.0

Total 1,433 1,271 88.7 1,455 1,303 89.6 2,888 2,574 89.1

weremuch the same. The percentages tested were

m

orethan 78.1 per cent in every age group.The field activities below were started in May,

1984and concluded in December, 1984. Physicalexaminations,ultrasound examination of the liver,

biliary tract, the spleen and blood samplecollections were carried out at public halls, schoolsand nurseries in each district.

Physical examinationsPhysical examinations were performed by a

member of our clinical staff. Weight (kg) andheight (cm) of all participants were measured. By

using modified Broca's index the obesity index(per cent) was calculated from the following

formula :The obesity index = 100 x (Weight/Standard

weight) where Standard weight = 0.9 x (Height -100).

Obesity was defined as more than 120 per centof the obesity index.

At that time, they were examined andinterviewed by a physician and had their chestX-rayed. Persons who were found to havetuberculosis, malignancy, protein malnutrition,severe hepatitis, and jejunoileal bypasses were

excluded from the study.

B

lood sample collectionBlood samples were drawn from the forearm by

venipuncture in the fasting condition and serawere stored at -20°C until tested. For participantsover 20 years old, serum levels of total cholesterol

and triglyceride were measured. More than 250mg/dl of serum total cholesterol level or more than170 mg/dl of serum triglyceride level was regarded

as hypercholesterolemia or hypertriglyceridemiarespectively.

UltrasonographyIn the fasting condition, real time ultrasono-

graphy of the liver, biliary tract, and spleen wasperformed by a physician with a scanner equipped

with a 3.5 MHz transducer (Aloka 21OF, AlokaCo.,Ltd., Japan). All ultrasonographies were

recorded by video tape and later the video tapeswere viewed for diagnosis. Fatty liver wasdiagnosed by ultrasonographic findings of Liver-Kidney contrast, vascular blurring, and deepattenuation5' 6^. Right intercostal scanning de-

monstrated the right lobe and the right kidney onthe same plane, and Liver-Kidney contrast was

obtained. Vascular blurring was the blurring of thehepatic and portal vein trunk and deep attenuation

was attenuation of the echo-beam in the deepportion of the right hepatic lobe. The right

JpnJMedVol27,No 2 (May, 1988) 143

Nomura et al

subcostal scanning demonstrated the right hepaticlobe and the hepatic and portal vein trunk. From

these findings we were able to evaluate vascularblurring and deep attenuation. Cirrhotic liver

which gives rise to bright liver echo patterns wasdeleted from fatty liver by physical examinations

and by ultrasonographic findings of the liver,biliary tract, and spleen7"9\

Q

ue stionnaire

For the collection of alcohol history, a standardquestionnaire was used. In order to assure thereliability of the answers, the same doctorinterviewed subjects over 20 years old forconsumption of alcohol. These interviews havebeen taken once a year since 1977. The answerswere recorded on private cards every year.Interviews were obtained from subjects of this

study in 1984. Those whose answers were negativefor alcohol consumption were considered non-

drinkers.The remainder were considered drinkers.

It was recognized that the duration of the alcoholintake in all drinkers over 25 years old was morethan five years. Almost all drinkers 20-24 years ofage began to consume alcoholic beverages from 20

years old.The frequency and amount of alcoholic

beverages currently being consumed were asked onthe drinkers questionnaire. They were asked howmany days they consumed alcoholic beverages in aweek and how much they drank beer, liquor, sake,and Awamori in a day. Awamori is a kind of liquormade from rice in Okinawa. The number of gramsof alcohol consumed per day was calculated from

the interviews.This information was recorded in standard unit

(633 ml bottle of beer, 43 mljigger of liquor, 180ml cup of sake, and 180 ml cup of Awamori).

Using information from the United StatesDepartment of Agriculture (USDA) Handbook

45610) and Medical consequences of alcholism1 1^ ,

we calculated the following ethanol contents:beer,22.3 g/bottle; liquor, 15.1 g/jigger; sake, 23.8

g/cup; and Awamori, 54.0 g/cup. With theseestimates, we calculated the total grams of ethanolconsumed in an average day.

AnalysisWe used 1,832 persons over 20 years old in the

following four analysises; ultrasonography, obesityindex, alcohol consumption, and serum levels oftotal cholesterol and triglyceride. We comparedprevalence of fatty liver between males andfemales and between drinkers and non-drinkers, ineach specified age group by chi-square test. Inorder to investigate the influence of the associatedfactors, i.e. obesity, alcohol (for only males), andserum levels of triglyceride and total cholesterol,on the prevalence of fatty liver, we used logisticregression analysis. Furthermore, to see theinteraction among the associated factors of fattyliver, we applied the Automatic InteractionDetector (AID) analysis12' 13) to the data.

Out of 957 females, only 34 drank alcoholbeverages. Therefore, in the analysis of logisticregression and AID for women, we countedalcoholic

consumption out.

RESULTS

Age- and sex-specific prevalence of fatty liver isshown in table 2. Overall prevalence was 14.0 percent. Prevalence increased with advancing age to amaximum in individuals 40-49, 50-59 years of ageand then decreased with advancing age. Prevalenceof fatty liver in individuals under 19 years old wasless than 2.8 per cent, considered to be very low. Prevalence of fatty liver was higher in malesthan in females in individuals younger than age 59years. However, the results of the comparisons ofprevalences of fatty liver by chi-square test,indicated that prevalence of fatty liver was notsignificantly higher at level 0.05 in males than infemales, in any of the age groups. Prevalence of fatty liver according to obesityindex and serum levels of triglyceride amongindividuals over 20 years old was shown in table 3and4. In males (table 3) and females (table 4),prevalence of fatty liver increased with obesityindex and increased with serum levels oftriglyceride. In every group of serum levels oftriglyceride, prevalence of fatty liver increasedwith obesity index. In every group of obesityindex, prevalence of fatty liver increased withserum levels of triglyceride. In individuals without obesity and withouthypertriglyceridemia, prevalence of fatty liver was

144 JpnJ MedVol27,No 2 (May, 1988)

Prevalence of Fatty liver in Okinawa

Table 2. Age- and sex-specific prevalence of fatty liver among 2574 individualsin the Yaeyama District, Okinawa, Japan.

Males Females Total

Age No. Prevalence No. Prevalence No. Prevalence(years) tested (%å ) tested (%) tested (%)

0- 9 264 0.4 197 0.0 461 0.2

10-19 132 3-0 U9 2.7 281 2.8

20-29 125 13.6 104 10.6 229 12.2

30-39 163 17.8 129 12.4 292 15.4

40-^9 122 27.9 120 23.3 242 25.6

50-59 191 27.8 249 23.7 440 25.5

60-69 167 15.6 189 23.8 356 19.9

70+ 107 8.4 166 14.5 273 12.1

Total 1.271 13.6 1.303 14.4 2.574 U.O

Table 3. Prevalence of fatty liver according to obesity index and serum levels oftriglycerids among 875 males over 20 years old in the Yaeyama District,Okinawa, Japan.

Serum levels of triglyceride (mg/dl)

< 170 170 - 24.9 > 250 Totalobesity å 

index No. of Prevalence No. of Prevalence No. of Prevalence No. of Prevalence(%) tested (%) tested (%) tested (%) tested (%)

<120 622 8.4 46 15.2 43 46.5 711 ll.1

120-129 57 24.6 21 52.4 21 76.2 99 41 -4

130-139 20 45.0 13 76.9 13 100.0 46 69.6

>140 9 66.7 5 100.0 5 100.0 19 84-2

Total 708 ll.4 85 38.8 82 65-9 875 19.2

Table 4. Prevalence of fatty liver according to obesity index and serum levels oftriglyceride among 957 females over 20 years old in the Yaeyama District,Okinawa, Japan.

Serum levels of triglyceride (mg/dl)

< 170 170 - 249 > 250 Totalobesity ;- -

index No. of Prevalence No. of Prevalence No. of Prevalence No. of Prevalence(%)' tested (%) tested (%) tested (%) tested (%)

<120 499 2.0 36 2.8 16 18.8 551 2.5

120-129 132 15.2 25 32.0 19 42.1 176 20.5

130-139 86 39.5 18 55.6 9 66.7 113 44.2

>140 86 68.6 10 70.0 21 81.0 117 70.9

Total 803 15.3 89 29.2 65 52.3 957 19.1

JpnJ MedVol27,No 2 (May, 1988) 145

Nomura et al

T

able 5. Prevalence of fatty liver according to alcohol consumption among 875males over 20 years old by obesity in the Yaeyama District, Okinawa,Japan.

Persons with obesity Persons without obesity TotalAlcohol ,

consumption No. of Prevalence No. of Prevalence No. of Prevalence(g/day) tested (%) tested (%) tested (%)

Non-drinkers 40 57. 5 195 3-6*" 235 1 2.8*

Drinkers '1-59 81 51.92926.837316.6

60-99 30 40.0 182 19.8 212 22.6

>100 13 92.3 42 38.1 55 50.9

Total 124 53-2 516 14-0** 640 21.6*

Total 164 54-3 711 ll.1 875 19.2

*: p < 0.01, **: P<C 0.001 Analysiswas done between non-drinkers and drinkers.

significantly higher in males (8.4 per cent) than infemales (2.0 per cent) (p <0.001).

Prevalence of fatty liver according to alcoholconsumption among 875 males over 20 years old

with obesity is shown in table 5. Prevalence offatty liver was significantly higher in 640 drinkers

(21.6 per cent) than 235 non-drinkers (12.8 percent) (p < 0.01). And prevalence of fatty liver

increased with alcohol consumption. In personswith obesity, in each group of alcohol consump-

tion, prevalence of fatty liver was more than 40.0per cent. Between 124 drinkers and 40 non-

drinkers difference in prevalence was not observed.In persons without obesity, only seven persons outof 195 non-drinkers were found to have fatty liver,and prevalence of fatty liver increased with alcoholconsumption. Prevalence of fatty liver was

significantly higher in 516 drinkers (14.0 per cent)

than in 195 non-drinkers (3.6 per cent) (p <0.001).

The result of logistic regression analysis offactors associated with fatty liver applied to 875

males over 20 years old is shown in table 6. Thisresult showed obesity to be the strongest predictorof fatty liver (p < 0.001). Regression coefficientsfor alcohol and serum triglyceride level were alsosignificant (both p < 0.001). However, serum total

cholesterol level was not significant (p=0.507).

The result of logistic regression analysis offactors associated with fatty liver applied to 957femalesis shown in table 7. The result showedobesity to be the strongest predictor of fatty liveras well (p < 0.001). Serum triglyceride level wasalso significant (p < 0.001). However, serum total

T

able 6. Logistic regression analysis of factors associated with the fatty liver in 875males over 20 years old, in the Yaeyama District, Okinawa, Japan.

Associated Factor Coefficients Standard Normal P valuefactors value or range errors deviation

Obesity Positive 0.0750 0.0075 9-94-2 < 0.001

Alcohol Drinkers 0.0162 0.0026 6.269 < 0.001

Triglyceride >170 0.0068 0.0010 6.769 < 0.001(mg/dl)

Total cholesterol? >250 -0.0021 0.0031 -0.66-4 0.507(mg/dl)

*: Serum triglyceride levels.f: Serum total cholesterol levels.

146JpnJMedVol27,No 2 (May, 1988)

Prevalence of Fatty liver in Okinawa

T

able 7. Logistic regression analysis of factors associated with the fatty liver in957 females over 20 years old in the Yaeyama District, Okinawa, Japan.

Associated Factor Coefficients Standard Normal P valuefactors value or range errors deviation

Obesity Positive 0.0980 0.0061 15-954 < 0.001

Triglyceride* >170 0.0066 0.0012 5.4-58 < 0.001

(mg/dl)Total cholesterol! >250 0.0006 0.0027 0.210 0.834

(mg/dl)

*: Serum triglyceride levels.f: Serum total cholesterol levels.

c

holesterol level was not significant (p=0.834).

The result of AID analysis of factors associatedwith fatty liver applied to 875 males over 20 yearsold is shown in figure 1. 875 males in cluster 1were classified into two more clusters; cluster 2,without obesity, ll.1 per cent and cluster 3, withobesity, 54.3 per cent, respectively. Cluster 2 was

then classified into cluster 4, without hypertri-glyceridemia, 8.4 per cent, and cluster 5, withhypertriglyceridemia, 30.3 per cent, respectively.Cluster 4 was then classified into cluster 6,

non-drinkers, 2.1 per cent, and cluster 7, drinkers,

ll.0 per cent, respectively. Cluster 7 was thenclassified into cluster 8, without hyper-

cholesterolemia, 1 1.0 per cent, and cluster 9, with

hypercholesterolemia, 12.5 per cent, respectively.Therefore it was suggested that in males obesitywas the strongest factor, followed by alcoholwhich was the strongest factor associated with

f

atty liver in males without obesity.The result of AID analysis of factors associated

with fatty liver applied to 957 females over 20years old is shown in figure 2. 957 females in

cluster 1 were classified into two more clusters;cluster2, without obesity, 2.5 per cent, andcluster 3, with obesity, 41.6 per cent, respectively.

Cluster 3 was then classified into cluster 4,without hypertriglyceridemia, 37.2 per cent and

cluster5, with hypertriglyceridemia, 54.9 per cent,respectively. Cluster 4 was then classified into

cluster6, without hypercholesterolemia, 36.6 percent, and cluster 7, with hypercholesterolemia,

42.9 per cent, respectively. Therefore, in femalesobesity was the strongest factor associated withfatty liver.

DISCUSSION

To our knowledge, this investigation representsthe first extensive cross-sectional study on theepidemiology of fatty liver among a generallyhealthy population using real time ultrasono-graphy. We have investigated epidemiological

Fig. 1. Automatic Interaction Detector analysis of factors associated with the fatty liver in 875 malesover 20 years old, in the Yaeyama District, Okinawa, Japan.

JpnJ MedVol27,No 2 (May, 1988) 147

Nomura et al

Fig. 2. Automatic Interaction Detector analysis of factors associated withthe fatty liver in 957 females over 20 years old,in the Yaeyama District,

Okinawa, Japan.

studies about hepatitis B virus and liver diseasessince 1970. Therefore, in this district we couldsurvey 90 per cent of the estimated population.Until recently it was possible to determine fatty

liver only by taking a liver biopsy, which requires astay in the hospital and many risk complica-tions14).

Prevalence of fatty liver in individuals under 19years was not fully investigated in the previous

study. Kinugawa et al.15^ reported that fatty liverwas found only in children with simple obesity. Inour study, prevalence of fatty liver in individuals

under 19 years was 2.8 per cent and considered tobe very low. Nine persons with fatty liver under 19years old were all obese and their serum levels of

triglyceride and total cholesterol were withinnormal ranges.

Since the report by Zelman et al. in 195216) ithas been well known that simple obesity in adultsis likely to be associated with fat deposition in theliver. Many other reports17"19^ also support this

hypothesis. The present study shows that pre-valence of fatty liver, both in males and in femalesover 20 years old, increased with the advancing

obesity index and prevalence of fatty liver inpersons with obesity were significantly higher thanthat in persons without obesity. These results weresimilar to the previous reports16~19\

Elevated serum triglyceride level was observedin persons with not only alcoholic fatty

liver20' 21) but also other types of fatty liver. The

present study shows that elevated serum tri-glyceride level was strong factor associated withfatty liver. Chronic ingestion of alcohol is known to causehepatic steatosis both in experimentalanimals22' 23) and humans24' 25). Also, LieberCS et al.26) reported that individuals with a

morphologically normal liver developed fatty liverwhen they were given alcohol even when under avariety of nondeficient diets. In the present study,prevalence of fatty liver in males without obesityincreased with consumption of alcohol. In personsboth without obesity and without hypertri-glyceridemia, prevalence of fatty liver wassignificantly higher in drinkers than in thenon-drinkers whose prevalence of fatty liver wasvery low. These results suggest that alcohol per sewas a strong factor associated with fatty liver. The result of logistic regression analysisindicated that in males obesity, alcohol, andelevated serum triglyceride level were significantpredictors of fatty liver, and in females, obesityand elevated serum triglyceride level weresignificant

predictors.

REFERENCES

1) Taylor KJW, Carpenter DA, Hill CR, et al. Gray scale ultrasound imaging: the anatomy and pathology of the liver. Radiology 119: 415, 1976. 2) Joseph AEA, Dewbury KC, MacGuire PG. Ultra- sound in the detection of chronic liver (the "bright

148 JpnJMedVol27,No 2 (May, 1988)

Prevalence of Fatty liver in Okinawa

liver"). Br J Radiol52: 184, 1979.3) Gosink BB, Lemon SK, Scheible W, et al. Accuracy

of ultrasonography in diagnosis of hepatocellulardisease. AJR 133: 19, 1979.

4) Foster KJ, Dewbury KC, Griffith AH, et al. Theaccuracy of ultrasound in the detection of fattyinfiltration of the liver. Br J Radiol 53: 440, 1980.

5) Yajima Y, Ohta K, Narui T, et al. Ultrasonographicaldiagnosis of fatty liver: significance of theLiver-Kidney contrast. Tohoku J Exp Med 139: 43,1983.

6) Saverymuttu SH, Joseph AEA, Maxwell JD.Ultrasound scanning in the detection of hepaticfibrosis and steatosis. Br Med J 292: 13, 1986.

7) Harbin WP, Robert NJ, Ferruci JT. Diagnosis ofcirrhosis based on regional changes in hepaticmorphology. Radiology 135: 273, 1980.

8) Sand ford NL, Walsh P, Matis C, et al. Isultrasonography useful in the assessment of diffuseparenchymal liver disease? Gastroenterology 89:186, 1985.

9) Debongnie JC, Pauls C, Fievez M, et al. Prospectiveevaluation of the diagnostic accuracy of liverultrasonography. Gut 22: 130, 1981.

10) Adams C. Agriculture Handbook No 456: Nutritivevalue of American foods in common units. USDepartment of Agriculture, Washington, DC, 1975.

ll) Ishii H. Medical consequences of alcoholism-Metabolic effects and medical complications causedby alcohol-. Igakushoin, Tokyo, 1981 (in Japanese).

12) Heald GI. The application of the automaticinteraction detector (AID) programme and multipleregression techniques to the assessment of storeperformance and site selection. Opl Res Q 23: 445,1972.

13) Ali MA, Hickman PJ, Clementson AT. Theapplication of automatic interaction detection (AID)

in operational research. Opl Res Q 26: 243, 1975.14) Sherlock S. Disease of the liver and biliary system.

7

th ed. Blackwell Scientific Publications, Oxford,1985, p28.

15) Kinugawa A, Tsunamoto K, Furukawa N, et al.Fatty liver and its fibrous changes found in simpleobesity of children. J Pediatric Gastroenterlogy and

Nutrition. 3: 408, 1984.

16) Zelman S. The liver in obesity. Arch Intern Med 90:141, 1952.

17) Adler M, Schaffner F. Fatty liver hepatitis andcirrhosis in obese patients. Am J Med 67: 811, 1979.

18) Galambos JT, Wills CE. Relationship between 505paired liver tests and biopsies in 242 obese patients.Gastroenterology 74: 1191, 1978.

19) Braillon A, Capron JP, Herve' MA, et al. Liver inobesity. Gut 26: 133, 1985.

20) Sherlock S. Disease of the liver and biliary system.7th ed. Blackwell Scientific Publications, Oxford,1985, p346.

2

1) Lieber CS, Savolainen M. Ethanol and lipids. AlcoholClin Exp Res 8: 409, 1984.

22) Porta EA, Hartroft WS, de la Iglesia FA. Hepaticchanges associated with chronic alcoholism in rats.Lab Invest 14: 1437, 1965.

23) DeCarli LM, Lieber CS. Fatty liver in the rat afterprolonged intake of ethanol with a nutritionallyadequate new liquid diet. J Nutr 91: 331, 1967.

2

4) Klatskin G. Alcohol and its relation to liver damage.Gastroenterology 41: 443, 1961.

25) Lebach WK. Cirrhosis in the alcoholic and itsrelation to the volume alcohol abuse. Ann NY AcadSci252: 85, 1975.

26) Lieber CS, Jones DP, DeCarli LM, et al. Effects ofprolonged ethanol intake: Production of fatty liverdespite adequate diets. J Clin Invest 44: 1009, 1965.

JpnJ MedVol27,No 2 (May, 1988) 149