==== Front Ann Saudi MedAnn Saudi MedAnnals of Saudi Medicine0256-49470975-4466King Faisal Specialist Hospital and Research Centre 2258844110.5144/0256-4947.2012.288asm-3-288Original ArticleEpidemiological, clinical, and biochemical characteristics of Saudi patients with nonalcoholic fatty liver disease: a hospital-based study Al-hamoudi Waleed aEl-Sabbah Mohamed aAli Safiyya aAltuwaijri Mansour aBedewi Mohamed bAdam Mustafa cAlhammad Alwaleed cSanai Faisal dAlswat Khalid aAbdo Ayman a a Gastroenterology Unit, Department of Medicine, College of Medicine, King Saud University, Saudi Arabia b Radiology Department, College of Medicine, King Saud University, Saudi Arabia c Laboratory Department, College of Medicine, King Saud University, Saudi Arabia d Hepatobiliary Sciences, King Abdulaziz Medical City, Saudi ArabiaCorrespondence: Dr. Waleed K. Al-hamoudi, Gastroenterology Unit (59), Department of Medicine, College of Medicine, King Saud University, PO Box 2925, Riyadh 11461, Saudi Arabia., T: +966-1-467-1217, F: +966-1-467-1217, walhamoudi@gmail.comMay-Jun 2012 32 3 288 292 Copyright © 2012, Annals of Saudi Medicine2012This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.BACKGROUND The estimated prevalence of nonalcoholic fatty liver disease (NAFLD) in Saudi Arabia is 7% to 10%. Despite the high prevalence of risk factors including diabetes, obesity, and hyperlipidemia, no recent epidemiological studies have measured the disease burden. We aimed to determine the characteristics of Saudi NAFLD patients attending a university hospital, and study factors affecting alanine aminotransferase (ALT) levels. DESIGN AND SETTING A prospective study among patients referred for ultrasonography in King Khalid University Hospital in Riyadh, Saudi Arabia from February to May 2009. PATIENTS AND METHODS NAFLD was defined as an appearance of fatty liver on routine abdominal ultrasound in the absence of coexisting liver disease and alcohol consumption. Patients were classified into normal and high ALT (ALT >60 U/L) level groups for analysis. RESULTS The prevalence of NAFLD was 16.6% (218/1312). Patients with normal ALT had the mean (SD) age of 45.9 (10.6) years and the mean body mass index of 34.5 (7.9) kg/m2. Forty percent of the 151 patients with normal ALT had diabetes, 66.2% were obese, and 29.1% had hypertension. Forty-three patients (23%) had high ALT levels. These patients had significantly lower age (P=.003) and fasting blood sugar (P=.03) than the normal ALT group. Non-diabetic patients (odds ratio 0.30, 95% CI 0.1–0.8), men (female OR 0.23, 95% CI 0.1–0.5), lower cholesterol (P=.001), high-density lipoprotein (P=.006), and low-density lipoprotein (P=.008) levels were more likely to be observed among patients with high ALT levels. In a multivariate analysis, younger age (OR 0.96, 95% CI 0.93–0.99), being male (OR 0.23, 95% CI 0.09–0.57), and a lower cholesterol level (OR 0.55, 95% CI 0.37–0.82) were significant predictors of high ALT levels. CONCLUSION Based on the high prevalence of obesity and diabetes, the prevalence of NAFLD will continue to be high, unless awareness is inculcated among the local population. ==== Body Nonalcoholic fatty liver disease (NAFLD) consists of a broad spectrum of fatty liver changes, ranging from mild steatosis to nonalcoholic steatohepatitis (NASH) and cirrhosis. It is one of the most common liver diseases encountered worldwide and is the most common cause of abnormal liver enzymes in many developed countries.1 In Saudi Arabia, the prevalence has been reported to be 7% to 10%.2–4 Despite its benign course in the majority of cases, around 10% to 20% of patients may go on to develop advanced fibrosis and cirrhosis.1 Major risk factors include diabetes mellitus, obesity, and hyperlipidemia. These risk factors are extremely common in Saudi Arabia. Recent data suggests that the overall prevalence of these risk factors are 23.7%, 35.5%, and 54%, respectively. 5,6 To design future studies on the diagnosis and management of this disease, elucidating the basic epidemiological characteristics is mandatory. Unfortunately, only one large epidemiological study has been performed on NAFLD in Saudi Arabia, and this did not study contributing risk factors.4 We, therefore, undertook this study to determine the clinical, epidemiological, and biochemical characteristics of Saudi patients with NAFLD. Since the presence of an elevated alanine aminotransferase (ALT) level is often the first clue in the diagnosis of NAFLD,1 we evaluated ALT levels as well as levels of other biochemical liver markers. Further, we explored potential predictors of high ALT levels in patients whom we categorized as having NASH. PATIENTS AND METHODS For the purpose of this study, NAFLD was defined as an appearance of fatty liver on routine abdominal ultrasound in the absence of coexisting liver disease and alcohol consumption. All patients having an ultrasound in King Khalid University Hospital (KKUH), Riyadh, Saudi Arabia, from February to May 2009 were screened for fatty liver as part of the study, regardless of the primary ultrasound indication. Any patient with an appearance of fatty liver on ultrasound was referred from the ultrasound department to the hepatology clinic for recruitment into the study. Patients who were found to have normal liver enzymes and no other evidence of liver disease were given lifestyle modification advice, published studies on weight loss, and healthy diet and were discharged. Patients who were found to have high liver enzymes were followed routinely in the hepatology clinic with standard conventional medical care. All ultrasonography images were read by a single experienced radiologist. Patients were included if they were above 18 years old, had a fatty liver detected by routine ultrasound, and were willing to give an informed consent. Patients were excluded if they had other liver diseases (manifested by hepatitis B surface antigen [HBsAg] or hepatitis C virus antibody [HCV Ab] positivity, antinuclear antibodies [ANA], antismooth muscle antibodies [ASMA] titer equal or above 1/160, abnormal iron studies, or a low serum ceruloplasmin), consumed any alcohol, were on medications that may cause steatosis, or if the patient refused to be enrolled. Other exclusion criteria included secondary causes of NAFLD such as bowel bypass or recent weight loss of more than 10 kg in 6 months. All patients had laboratory investigations including complete blood count, ALT, aspartate aminotransferase (AST), alkaline phosphatase (ALP), gamma glutamyl transferase (GGT), bilirubin, albumin, ANA, ASMA, anti-mitochondrial antibodies, serum iron (Fe), transferrin saturation, serum ceruloplasmin, lipid profile, and fasting blood sugar. Direct enzyme-linked immunosorbent assay (ELISA) was used for detection of HBsAg and indirect ELISA for detection of HCV Ab. The recombinant immunoblot assay was used for confirmation of doubtful results. Details of demographic data, including alcohol consumption, the presence of diabetes, and hyperlipidemia were obtained from medical files. Informed consent was obtained from all patients prior to the start of the study. The study was approved by the Medical Ethics Committee in the College of Medicine, King Saud University. All variables were checked for normality. Data was analyzed with and without outliers. Influential points that greatly affected the distribution of certain variables were removed from the final analysis (8 patients in total). Descriptive statistics were summarized as mean (standard deviation), median (range), or frequency (percentage) as appropriate. The OR with 95% CI was used to assess group differences for categorical variables and the t test was used to assess difference between continuous variables. Multivariate logistic regression was performed to evaluate the predictors of high ALT levels. All tests were 2-sided with a 5% level of significance. All analyses were performed with Stata version 10 (Stata Corp, Texas, USA). RESULTS A total of 1312 upper abdominal ultrasounds were performed within the 4-month-study period. A total of 218 patients were labeled as having fatty liver on imaging, giving a prevalence rate of 16.6%. Of these, 6 refused to give blood samples, 6 had hepatitis B, and 4 had hepatits C, leaving a remaining of 202 patients in the study. The mean age for the 202 patients was 44.7 (11.5) years, including 102 (51.0%) men and 100 (49.0%) women. The mean body mass index (BMI) and waist circumference were 34.5 (8.1) kg/m2 and 112.9 (15.4) cm, respectively. Sixty-nine patients (34.2%) had diabetes, 137 (67.8%) were obese (BMI ≥ 30 kg/m2), and 54 (26.7%) had hypertension. Details of patient demographics and basic laboratory results are summarized in Table 1. Of the 202 patients, 43 (21.3%) had abnormally high ALT levels (defined as any reading above our laboratory’s upper limit of normal, i.e., >60 U/L) and were labeled as having NASH. This group was analyzed separately in comparison to the group with normal ALT levels. After removing outlying data points, we were left with a final sample of 151 patients in the normal ALT group, including 68 (45.0%) men and 83 (55.0%) women (Table 2). The mean age was 45.9 (10.6) years, while the mean BMI and waist circumference were 34.5 (7.9) kg/m2 and 113.5 (15.5) cm, respectively. Sixty patients (40.0%) had diabetes, 100 (66.2%) were obese (BMI ≥30 kg/m2), and 44 (29.1%) had hypertension. Women were more obese (OR 2.8, 95% CI 1.5–5.2). When we looked at the family history for risk factors of NAFLD, we found that 95 (62.9%) patients had a family history of diabetes, 83 (55.0%) had a family history of hypertension, 48 (31.8%) had a family history of obesity, and 24 (15.9%) had a family history of liver disease. Thirteen (8.6%) patients smoked. Physical activity, which was defined as walking for 30 minutes at least 5 days per week, was found in 32 (21.2%) of the patients. Forty-eight (32.8%) patients had hyperlipidemia, and mean cholesterol, triglycerides, and albumin levels were 4.9 (1.0) mmol/L, 1.7 (0.9) mmol/L, and 39.1 (3.7) g/L, respectively. The mean ALT level was 39.2 (9.5) U/L. Mean AST, ALP, and GGT levels were 19.8 (7.2) U/L, 95.2 (27.0) U/L, and GGT 44.4 (25.6) U/L, respectively. The mean serum Fe level was 13.1 (5.8) μmol/L, while mean fasting blood sugar level was 6.7 (3.4) mmol/L. We compared various parameters between those with normal ALT levels and high ALT levels (Table 2). The high ALT group had significantly lower age (P=.003) and fasting blood sugar (P=.03) than the normal ALT group. Non-diabetic patients (OR 0.30, 95% CI 0.1–0.8) and men (female OR 0.23, 95% CI 0.1–0.5) were more likely to be observed among the high ALT level group though no differences were seen with regard to triglycerides level (P=.54), obesity (OR 1.16, 95% CI 0.5–2.5), BMI (P=.96), or hypertension (OR 95% CI 0.94, 0.4–2.1). Slightly lower cholesterol (P=.001), HDL (P=.006), and LDL (P=.008) levels were found in those with high ALT levels. In the multivariate analysis (Table 3), younger age (OR 0.96, 95% CI 0.93–0.99), being male (OR 0.23, 95% CI 0.09–0.57), and having a lover cholesterol level (OR 0.55, 95% CI 0.37–0.82) were significant predictors of high ALT levels. Keeping other variables constant, for each 1 year increase in age, there is about a 4% decrease in the odds of being in the high ALT level group. Similarly, the odds of being in the high ALT level group decreases by 77% for women and every 1 unit increase in cholesterol level decreases the odds of being in the high ALT level group by 45%. DISCUSSION The estimated prevalence of NAFLD in our study was found to be 16.6%, which is slightly higher than the rates found in previous studies in Saudi populations.4 A retrospective study conducted by El-Hassan et al in 1992 looked at the radiological prevalence of NAFLD. In their study, 138 (9.7%) of 1425 computed tomography (CT) scans revealed fatty infiltration of the liver. Of these, 39% had abnormal liver enzymes.4 Another study by Al-Quorain et al examined the liver biopsies of 544 patients, and 7.2% of these patients were labeled as having fatty liver disease.3 The difference in prevalence of NAFLD in our study compared to Al-Quorain and colleagues may be attributed to differences in diagnostic methods and case definitions. Our study used ultrasound to diagnose fatty livers, whereas their study used a more specific one, i.e., liver biopsy. In a study by Akbar and Kawther aiming to find the prevalence and characteristics of NAFLD in Saudi type 2 diabetic patients, NAFLD was diagnosed in 55% of the 116 patients. 2 However, this study was targeted at diabetic patients and hence is not comparable to our population. Our study found prevalence rates closer to those found in Europe and Japan where the prevalence has been reported to be around 14% to 21%.7,8 Similarly, the prevalence in the United States from the National Health and Nutrition Examination Study (NHANES III) was 24%.9 Our study is not an accurate representation of the Saudi population since patients were recruited from a population of people who were sent for upper abdominal ultrasounds and as such selection bias is very probable. However, we believe that our figure is close to that of the actual population because ultrasounds were requested by general practitioners in patients not known to have primary liver disease. Moreover, the true prevalence of NAFLD is difficult to study because many people with early stages of the disease are relatively asymptomatic and the gold standard for diagnosing and staging NAFLD is via a liver biopsy.10 Our study population had more women than men with NAFLD (55% vs. 45%, respectively), which is in contrast to that found in the NHANES III study in which men had higher prevalence rates than women (33% vs. 17%). NAFLD was previously thought to be more common in women.11,12 A significant relationship was also found between the presence of NAFLD and female gender in the study by Akbar and Kawther.2 However, these series of patients were obtained from gastrointestinal clinics, and more recent population-based studies have concluded that NAFLD could be more common in men.8,13,14 The higher number of women in our study could be due to female gender being a surrogate factor for obesity, since women were more likely to be obese compared to men (OR 2.8, 95% CI 1.5–5.2), a finding that echoes results found by Al-Nozha et al., in the Saudi population previously.6 Unfortunately, this study has the major limitation of not having a control group; thus we are unable to conclude that females are more likely to have NAFLD. Nonetheless, obesity has been shown in many studies to be a determinant in causing NAFLD.13–16 The mean BMI in our population was alarmingly higher at 34.5 (7.9) kg/m2 compared to that found in the NHANES III study that was only 25.6 kg/m2 (10th percentile 20.5 kg/m2, 90th percentile 33.5 kg/m2). This is a worrying trend especially considering that the lack of physical activity is a common phenomenon in Saudi Arabia, being as high as 79% in the population we studied and 96% in the Saudi Coronary Artery Disease registry.17 NAFLD has been shown in various studies to be independently associated with type 2 diabetes, low HDL cholesterol, high triglycerides, and insulin resistance. 9,13,15,16 These factors were also reflected in our study. Forty percent of patients had diabetes and 32.8% had hyperlipidemia. Comparing our NASH patients with those with normal ALT levels, lower age (P=.003), being male (female OR 0.23, 95% CI 0.1–0.5), and lower fasting blood sugar [6.7 (3.4) vs. 5.4 (1.6) mmol/L, P=.003] were independently associated with elevated ALT levels. This is similar to the NHANES III study where lower age and male gender were associated with elevated ALT levels. However, that study had a higher proportion of people with high fasting blood sugar in the group with elevated ALT levels unlike ours. The higher fasting blood sugar in the normal ALT group could reflect the impaired blood glucose in the high number of diabetic patients found in this group. Non-diabetic patients (OR 0.30, 95% CI 0.1–0.8) were more likely to be observed among the high ALT-level group in our population, similar to that in the NHANES III study. Lower cholesterol (P=.001), HDL (P=.006), and LDL (P=.008) levels were also found in those with high ALT levels. Unlike the NHANES III study, we did not find any significant correlation between elevated ALT levels and higher BMI or triglycerides. Our multivariate analysis for predictors of high ALT levels revealed that younger age, male gender, and low cholesterol levels were associated with elevated ALT levels. This is similar to previous studies that showed association between elevated ALT levels and younger age and male gender.9,14 Our study used ultrasound in the diagnosis of NAFLD even though liver biopsy is the gold standard for diagnosing NAFLD after exclusion of other liver diseases. However, liver biopsy is costly and involves a moderate risk of complications. Apart from helping to exclude other possible liver disease etiologies and staging the progression of NAFLD, there are some arguments against liver biopsy given the absence of effective treatment approaches. The histologic interpretation of possible uneven lipid distribution varies from one pathologist to another, increasing the likelihood of sampling variability of liver biopsies.18 A population-based cohort study found that NAFLD patients who did not undergo a liver biopsy had a significantly higher survival than patients selected for biopsy, thus making it difficult to justify the benefit of biopsy in the general population with NAFLD.19 Ultrasound is more sensitive but less specific than CT in fatty liver detection. However with a sensitivity greater than 80% and specificity greater than 90%, ultrasound represents the most convenient and least expensive modality for diagnosing fatty liver. 20–23 For these reasons, ultrasound is widely used in surveying the prevalence of NAFLD. In conclusion, to our knowledge, this is currently the most up-to-date report on the prevalence of NAFLD in a hospital-based study in Saudi. Among this cohort of patients seen at the ultrasound unit in KKUH, the prevalence was found to be 16.6%. Based on the high prevalence of obesity and Type 2 diabetes, it is safe to assume that the prevalence of NAFLD will continue to be high unless awareness is inculcated among the local population. Further studies should be conducted to study the change in prevalence of NAFLD within the population over time to closely monitor the epidemiology of this disease. Table 1 General characteristics of nonalcoholic fatty liver disease patient population of Saudi Arabia under study (n=202). Variable Mean (SD), median (range), or frequency (%) Age (years) 44.7 (11.5) Gender  Male 102 (51.0)  Female 100 (49.0) BMI (kg/m2) 34.5 (8.1) ALT (U/L) 42 (16–672) ALP (U/L) 93 (15–352) Triglycerides (mg/dL) 1.7 (1.0) Cholesterol (mg/dL) 4.8 (1.1) Serum ferritin (μmol/L) 13.7 (6.2) Fasting blood sugar (mmol/L) 6.5 (3.2) Ferritin (ng/mL) 95.8 (1.5–5 238) TIBC (μmol/L) 72.6 (13.8) Diabetes  No 133 (65.8)  Yes 69 (34.2) Hypertension  No 148 (73.3)  Yes 54 (26.7) Hyperlipidemia  No 142 (70.3)  Yes 60 (29.7) SD: Standard deviation, BMI: body mass index, ALT: alanine aminotransferase, ALP: Alkaline phosphatase, TIBC: total iron–binding capacity. Table 2 Univariate analysis of predictors of high alanine aminotransferase levels and different parameters. Variable Normal alanine aminotransferase (n=151) (mean (SD) or requency [%]) High alanine aminotransferase (n=43) (mean (SD) or frequency [%]) P value or OR, 95% CI Age (years) 45.9 (10.6) 39.6 (13.5) .003b Gender  Male 68 (45.0) 31 (72.1) .23, .1–.5a  Female 83 (55.0) 12 (27.9) BMI (kg/m2) 34.5 (7.9) 34.4 (8.7) .96b Obese  No 51 (33.8) 14 (32.6) 1.16, .5–2.5a  Yes 100 (66.2) 29 (67.4) Diabetes  No 91 (60.3) 34 (79.1) .30, .1–.8a  Yes 60 (39.7) 9 (20.9) Fasting blood sugar (mmol/L) 6.7 (3.4) 5.4 (1.6) .03b Cholesterol (mg/dL) 4.9 (1.0) 4.2 (1.2) .001b HDL (mmol/L) 1.2 (0.3) 1.1 (0.3) .006b LDL (mmol/L) 3.0 (0.9) 2.6 (0.8) .008b Triglycerides (mg/dL) 1.7 (0.9) 1.8 (1.2) .54b Hypertension  No 107 (70.9) 33 (76.7) .94, .4–2.1a  Yes 44 (29.1) 10 (23.3) a OR: Odds ratio (OR) and 95% confidence interval (CI), b t test. SD: standard deviation, BMI: body mass index, HDL: high-density lipoprotein, LDL: low-density lipoprotein, Table 3 Multivariate analysis of showing statistically significant predictors of high alanine aminotransferase levels. Predictor Multivariate analysis OR (95% CI) Age (years) 0.96 (0.93–0.99) Gender (female) 0.23 (0.09–0.57) Cholesterol (mg/dL) 0.55 (0.37–0.82) Logistic regression with odds ratio (OR) and 95% confidence interval (CI). ==== Refs REFERENCES 1 Rector RS Thyfault JP Wei Y Ibdah JA Nonalcoholic fatty liver disease and the metabolic syndrome An update World J Gastroenterol 2008 14 185 92 18186553 2 Akbar DH Kawther AH Nonalcoholic Fatty Liver Disease in Saudi Type 2 Diabetic Subjects Attending a Medical Outpatient Clinic Diabetes Care 2003 263 351 2 3 al-Quorain A Satti MB al-Hamdan AR al-Gindan Y Ibrahim E Khatib R Pattern of chronic liver disease in the eastern province of Saudi Arabia. A hospital-based clinicopathological study Trop Geogr Med 1994 46 358 60 7892703 4 El-Hassan AY Ibrahim EM al-Mulhim FA Nabhan AA Chammas MY Fatty infiltration of the liver Analysis of prevalence, radiological and clinical features and influence on patient management Br J Radiol 1992 65 774 8 1393413 5 Al-Nozha MM Al-Maatouq MA Al-Mazrou YY Al-Harthi SS Arafah MR Khalil MZ Diabetes mellitus in Saudi Arabia Saudi Med J 2004 25 1603 10 15573186 6 Al-Nozha MM Al-Mazrou YY Al-Maatouq MA Arafah MR Khalil MZ Khan NB Obesity in Saudi Arabia Saudi Med J 2005 26 824 9 15951877 7 Bellentani S Saccoccio G Masutti F Crocè LS Brandi G Sasso F Prevalence of and risk factors for hepatic steatosis in Northern Italy Ann Intern Med 2000 132 112 7 10644271 8 Nomura H Kashiwagi S Hayashi J Kajiyama W Tani S Goto M Prevalence of fatty liver in a general population of Okinawa, Japan Jpn J Med 1988 27 142 9 3047469 9 Clark JM Brancati FL Diehl AM Nonalcoholic fatty liver disease Gastroenterology 2002 1221649 57 10 Clark JM Diehl AM Defining nonalcoholic fatty liver disease Implications for epidemiologic studies Gastroenterology 2003 124 248 50 12512048 11 Sheth SG Gordon FD Chopra S Nonalcoholic steatohepatitis Ann Intern Med 1997 126137 45 12 Falck-Ytter Y Younossi ZM Marchesini G Mc-Cullough AJ Clinical features and natural history of nonalcoholic steatosis syndromes Semin Liver Dis 2001 21 17 26 11296693 13 Clark JM Brancati FL Diehl AM The prevalence and etiology of elevated aminotransferase levels in the United States Am J Gastroenterol 2003 98 960 7 12809815 14 Ruhl CE Everhart JE Determinants of the association of overweight with elevated serum alanine aminotransferase activity in the United States Gastroenterology 2003 12 471 9 15 Cortez-Pinto H Camilo ME Baptista A De Oliveira AG De Moura MC Non-alcoholic fatty liver Another feature of the metabolic syndrome? Clin Nutr 1999 18 353 8 10634920 16 Lainé F Bendavid C Moirand R Tessier S Perrin M Guillygomarc’h A Prediction of liver fibrosis in patients with features of the metabolic syndrome regardless of alcohol consumption Hepatology 2004 391 639 46 17 Al-Nozha MM Al-Hazzaa HM Arafah MR Al-Khadra A Al-Mazrou YY Al-Maatouq MA Prevalence of physical activity and inactivity among Saudis aged 30–70 years. A population-based cross-sectional study Saudi Med J 2007 28 559 68 17457478 18 Ratziu V Charlotte F Heurtier A Gombert S Giral P Bruckert E LIDO Study Group Sampling variability of liver biopsy in nonalcoholic fatty liver disease Gastroenterology 2005 128 1898 906 15940625 19 Adams LA Lymp JF St Sauver J Sanderson SO Lindor KD Feldstein A The natural history of nonalcoholic fatty liver disease A population-based cohort study Gastroenterology 2005 129 113 21 16012941 20 Joseph AE Saverymuttu SH al-Sam S Cook MG Maxwell JD Comparison of liver histology with ultrasonography in assessing diffuse parenchymal liver disease Clin Radiol 1991 43 26 31 1999069 21 Mendler MH Bouillet P Le Sidaner A Lavoine E Labrousse F Sautereau D Dual-energy CT in the diagnosis and quantification of fatty liver Limited clinical value in comparison to ultrasound scan and single-energy CT, with special reference to iron overload J Hepatol 1998 28 785 94 9625313 22 Mathiesen UL Franzén LE Aselius H Resjö M Jacobsson L Foberg U Increased liver echogenicity at ultrasound examination reflects degree of steatosis but not of fibrosis in asymptomatic patients with mild/moderate abnormalities of liver transaminases Dig Liver Dis 2002 34 516 22 12236486 23 Hultcrantz R Gabrielsson N Patients with persistent elevation of aminotransferases Investigation with ultrasonography, radionuclide imaging and liver biopsy J Intern Med 1993 23 37 12