Effect of alcohol consumption on estimated glomerular filtration rate and creatinine clearance rate

Abstract

Background. Moderate alcohol consumption is widely recognized as beneficial in the prevention of cardiovascular disease, yet the renal effects of alcohol intake are still controversial. The present study is designed to investigate the influence of alcohol consumption on calculated creatinine clearance rate (CCr) and glomerular filtration rate (GFR) in a Southern Taiwan Pai-Wan aboriginal community with a high prevalence of alcohol consumption.

Methods. This is a cross-sectional community-based study. The 1466 aboriginal subjects, 40–95 years of age, are a stratified random subpopulation identified during an integrative health care programme. They were sampled for drinking patterns. The main outcome measurements were serum creatinine, estimated CCr and GFR.

Results. Subjects with alcohol consumption had significantly higher levels of serum triglycerides, high-density lipoprotein cholesterol, uric acid, estimated CCr and GFR values than non-drinkers. Their blood pressure was also significantly higher. They had lower total cholesterol and low-density lipoprotein cholesterol concentrations. Increasing alcohol consumption was independently and significantly associated with a higher level of estimated CCr and GFR when analysed as both a categorical and continuous variable.

Conclusions. The present study shows that chronic alcohol consumption has a negative effect on blood pressure and lipid profile and stimulates the estimated GFR.

alcohol, blood pressure, creatinine clearance rate, glomerular filtration rate

Introduction

Moderate alcohol consumption has well-known beneficial effects on the prevention of coronary heart disease morbidity and mortality [1]. This is primarily attributed to alcohol intervening in the atherogenetic process and regulation of vasoactive peptide secretion [2].

The association between alcohol consumption and renal function, however, is poorly understood. There is evidence that chronic alcohol consumption may cause direct damage to the kidneys [3,4]. It may also indirectly alter renal function by elevating blood pressure [5], inducing electrolyte imbalances [4] and inducing hyperuricaemia [6]. In one study, alcohol intake had no affect on glomerular filtration rate (GFR) and renal plasma flow [7]; however, in another study with rats, chronic ingestion led to impairment of GFR associated with renal hypertrophy [8]. In 1984, Savdie and colleagues demonstrated that drinkers had significantly higher blood pressure that non-drinkers, and drinkers of two or fewer drinks per day were found to have higher serum creatinine than non-drinking control subjects. This changed, however, to decreased serum creatinine at the level of three or more drinks per day [9]. In a prospective study of 1658 nurses, Knight et al. [10] showed that moderate alcohol intake has no long- term adverse effect on renal function as assessed by calculated creatinine clearance rate (CCr) and GFR, and may in fact have a renoprotective effect in women with hypertension.

It is apparent that chronic alcohol consumption regulates renal function. To describe this relationship more fully, we examined the renal effects of alcohol consumption using calculated CCr and GFR, modified through the Cockcroft–Gault calculator and Modification of Diet in Renal Disease (MDRD) in a southern Taiwan aboriginal community with a high prevalence of alcohol consumption.

Subjects and methods

Data collection

Between January 2000 and December 2003, the Taiwan Bureau of National Health Insurance conducted the Taiwan Aboriginal Health Interview and Integrated Health Care Programme in rural indigenous people >30 years old. Using population registries, this study selected a stratified random sample of 1466 subjects, representative for age, gender and aboriginal community. The overall participation rate was 59.0%, according to the 2002 Official Household Registry.

Under the supervision of trained personnel, participants completed a comprehensive questionnaire concerning socio-economic status, physical activity, smoking status and eating habits, including alcohol consumption. In addition, a physician conducted interviews for medical history and drug use. Questionnaires were checked for proper completion and plausibility. Field work quality and standardization were ascertained regularly. All study subjects lived in the same region at the time of the study and were of Pai-Wan ancestry without known ancestry from other ethnicities. This study was approved by the Human Research Ethics Committee of our hospital, and informed consent was obtained from each participant.

For the purposes of the present study, we excluded participants with elevated serum creatinine >1.5 mg/dl, proteinuria (urine protein ≥300 mg/dl), urolithiasis or other known renal disease, and/or other reasons, including age (<40 years old), residence outside aboriginal communities or those with missing information. The ex-users may include some subjects who had given up drinking because of adverse effects on their health [11]. Hence, ex-users (who had stopped drinking for >1 year before the interviews) were excluded from the present study. We also excluded subjects with diabetes mellitus, because valid GFR estimates cannot be obtained for diabetic patients with normal renal function by estimation of creatinine clearance, nor from serum creatinine concentration measurements and rate of change in GFR [12].

Venous blood samples were drawn for biochemical analysis after a fasting period of at least 8 h. All biochemical analyses were carried out within 2 h of blood sampling at the Pingtung Christian Hospital laboratory. Serum creatinine was analysed by means of the Kinetic Jaffé's method on a SYNCHRON CX System analyser (SYNCHRON, Los Angeles, CA) using reagents from Beckman (Beckman Coulter Diagnostic, Los Angeles, CA). Serum triglycerides, total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, uric acid, albumin, serum urea nitrogen and glucose were determined by standard commercial methods on a parallel-multichannel analyser (SYNCHRON, Los Angeles, CA). The inter-assay coefficients of variation were 2.1% for creatinine, 4.1% for triglycerides, 2.2% for total cholesterol, 5.5% for LDL cholesterol, 4.9% for HDL cholesterol, 1.7% for glucose, 3.1% for blood urea nitrogen and 3.2% for albumin. The blood biochemistry analyses were under internal and external quality control at the laboratory according to the College of American Pathologists’ surveys.

We used three primary estimates of renal function. These were serum creatinine, estimated CCr calculated by the modification of the Cockcroft–Gault formula [13] and estimated GFR calculated with the MDRD-extended version. These two formulae are freely available from the Internet (http://www.kidney.org/professionals/kdoqi/index.cfm).

The habitual alcohol consumption of study subjects was reviewed with a structured questionnaire. Subjects who had consumed one serving of an alcoholic beverage (including beer, liquor and wine) or more per month for at least 1 year were defined as ever drinkers. Among them, current drinkers were those with these habits in the past year, and ex-users were those who had stopped such behaviour for at least 1 year before the interviews. The ever drinkers who had stopped drinking <1 year before the interviews and current drinkers were defined as current alcohol users in the present study. For all of the ever drinkers, a detailed history of their drinking habits was recorded, including daily consumption, primary alcohol type, frequency, amount per consumption, age of commencement, duration of practice, whether additional beverage types were consumed (yes/no) and if (yes) what beverages and the amount consumed. Average alcohol consumption (in g per month) was calculated by multiplying the frequency and amount of alcohol intake, using a standard ethanol content of 20 vol% in Taiwan michiu, 4.5 vol% in Taiwan beer, 8 vol% in Taiwan whisbih, 8 vol% in Taiwan paulid-B and 57 vol% in Kinmen kao liang liquor. If the questionnaire was answered with a range of values, the midpoint of the range of frequency or amount was used. Valid questionnaires were examined further.

In order to assess the relationship between dosage and research variables, the subjects were categorized into four groups: non-drinkers, light drinkers (≤150 g alcohol/month), moderate drinkers (151–1500 g alcohol/month) and heavy drinkers (>1500 g alcohol/month). A socio-economic status index (low, middle and high) was created from education level, occupation and household income. Smokers were subjects who had smoked one cigarette or more per day for at least 1 year. Similar to our classification of drinkers, current smokers were those who had practised these habits within the past year, and ex-users were those who had stopped the practice for at least 1 year before the interviews. Those who had smoked <100 cigarettes in their entire lives were categorized as non-smokers. In this study, a smoker who had stopped the practice <1 year before the interviews was defined as a current smoker.

Blood pressure was measured after the subject had rested for 5 min in a sitting position. In this study, hypertension was defined as a systolic blood pressure (SBP) ≥140 mmHg, a diastolic blood pressure (DBP) ≥90 mmHg, or under antihypertensive treatment. Diabetes mellitus was defined according to the 1997 World Health Organization criteria, or having received treatment for diabetes mellitus. Height, body weight, waist and hip circumferences were measured, and body mass index (BMI) and waist to hip ratio (WHR) were calculated.

Statistical analysis

Descriptive data were examined for all variables. For continuous variables, results are presented as the mean±SD. Statistical differences in variables were compared using one-way analysis of variance (ANOVA) and unpaired Student's t-test for normally distributed variables. Before statistical testing, serum triglycerides were logarithmically transformed to achieve a normal distribution. Categorical variables were recorded as frequency counts, and intergroup comparisons were analysed by a χ² test. General linear modelling (GLM) function analysis was used to control for potential confounders other than age (e.g. gender, BMI and smoking status).

Associations between estimated CCr and GFR and other parameters were first analysed by simple linear regression and then by multiple linear stepwise regression. Variables used in this analysis were age, gender, BMI, smoking, drinking status, SBP, DBP, fasting sugar, total cholesterol, triglycerides, HDL cholesterol, LDL cholesterol and uric acid levels, but old age, sex and BMI for estimated CCr, and old age and sex for estimated GFR in multiple linear stepwise regression analysis.

Multiple-adjusted least-square means of estimated CCr and GFR, lipid profiles, blood pressure and WHR as well as the SEM within the four alcohol consumption groups were tested for trends using information on maximum monthly alcohol amount, compared with non-drinkers as a reference and calculated using analysis of variance with proc GLM using the Statistical Analysis System (SAS). The pairwise correlations method was used to calculate the relationship among estimated CCr, GFR and amount of alcohol consumption. Statistical significance was accepted if P<0.05. All statistical analysis was performed using SAS statistical software, version 8.0 (SAS Institute Inc., Cary, NC).

Results

Data were collected from a total of 1466 rural indigenous people; 50 subjects with known renal disease and 160 for other reasons listed in Subjects and methods were excluded. None of the drinking subjects had clinical evidence of alcohol abuse [14].

Table 1 shows the 1256 subjects grouped to probe for differences in clinical and laboratory characteristics in non-alcohol users, alcohol users and ex-users. The values of age, gender, BMI, smoking, hypertension, diabetes mellitus rates, blood pressure, fasting sugar, total cholesterol, HDL cholesterol, LDL cholesterol and uric acid levels differ among the three types of alcohol consumption. Interestingly, serum creatinine levels are not statistically different among non-alcohol users, alcohol users and ex-users, while the estimated CCr and GFR are significantly higher in ex-users and alcohol users than in non-alcohol users. Table 1 also presents the clinical and biochemical information of 146 diabetic subjects, showing that those with an alcohol habit had a significantly higher estimated CCr and GFR. As stated in Subjects and methods, those with diabetes and ex-users were excluded from further study.

Table 1.

Clinical and biochemical characteristics of aborigines studied

Variable	All subjects included							Diabetes subjects
	Non-alcohol user	Alcohol user	Ex-user	Pa	Non-alcohol user	Alcohol user	Pb
No.	693	436	127		129	17
Age (years)	66.5±11.6	64.0±11.8	59.4±13.9	<0.0001	67.0±9.8	63.0±10.9	0.126
Gender (M/F)	229/464	215/221	84/43	<0.0001	50/79	10/7	0.114
BMI (kg/m2)	25.5±5.0	25.2±4.2	27.0±4.5	0.0004	26.1±4.0	28.0±3.5	0.084
Current smoker (%)	52 (7.3)	150 (33.0)	59 (46.8)	<0.0001	20 (15.5)	5 (29.4)	0.153
% of hypertension	321 (46.9)	225 (51.4)	69 (58.5)	0.001	87 (68.5)	12 (70.6)	0.862
% of diabetes mellitus	129 (18.6)	17 (3.9)	27 (21.8)	<0.0001	–	–	–
Systolic BP (mmHg)	134±21	137±21	140±23	0.006	138±19	143±19	0.385
Diastolic BP (mmHg)	81±12	84±14	82±13	0.021	83±12	82±16	0.842
Fasting sugar (mmol/l)	5.4±2.3	5.4±2.5	6.0±2.5	0.049	10.0±4.2	10.3±4.6	0.802
Total cholesterol (mmol/l)	4.94±0.98	4.73±1.01	5.04±1.09	0.0001	5.25±1.17	4.71±0.97	0.070
Triglycerides (mmol/l)c	1.44±1.07	1.59±1.30	1.59±1.32	0.138	2.17±1.73	2.39±1.32	0.361
HDL cholesterol (mmol/l)	1.1±0.3	1.2±0.3	1.1±0.3	0.001	1.1±0.2	1.1±0.4	0.457
LDL cholesterol (mmol/l)	3.01±0.83	2.65±0.83	3.17±0.93	<0.0001	3.13±0.87	2.27±0.57	0.014
Uric acid (mmol/l)	0.38±0.10	0.41±0.10	0.44±0.10	<0.0001	0.39±0.11	0.40±0.12	0.841
Creat (µmol/l)	88.4±26.5	88.4±17.7	88.4±26.5	0.310	99.9±50.4	86.6±26.5	0.261
CGCr (ml/min)	57.8±22.6	63.0±24.4	74.1±31.8	<0.0001	58.1±24.2	75.1± 26.0	0.009
MDRDGFR-E (ml/min/1.73 m2)	69.4±16.6	75.6±19.3	76.7±20.7	<0.0001	67.4±22.7	84.3±22.5	0.005

Data are expressed as mean±SD.

^aBy one-way ANOVA test for continuous variables and by χ² test for the categorical variables.

^bBy unpaired t-test or χ² test.

^cSignificant difference was tested using log-transformed data.

CGCr = estimated creatinine clearance calculated by the Modification of the Cockcroft–Gault Calculator [13] (adjusted by age, sex, weight and creatinine); MDRDGFR-E = estimated glomerular filtration rate calculated by the Modification of Diet in Renal Disease (MDRD) calculator-extended version (adjusted by age, sex, race, serum urea nitrogen, creatinine and albumin); BP = blood pressure; BMI = body mass index; HDL = high-density lipoprotein; LDL = low-density lipoprotein; Creat = creatinine.

A total of 419 non-diabetic, renal disease-free, alcohol users were collected for analysis. After frequency matching for age and sex, 419 non-diabetics, renal disease-free, non-alcohol user subjects were randomly selected as controls. Table 2 shows that those with an alcohol habit had a significantly higher smoking rate, blood pressure, serum triglycerides, HDL cholesterol, uric acid, estimated CCr and GFR values, and lower total cholesterol and LDL cholesterol concentrations than non-drinkers. When adjusted for age, gender, BMI and smoking status, drinkers had significantly higher blood pressure, serum triglycerides, HDL cholesterol and uric acid levels, and lower total cholesterol and LDL cholesterol concentrations. Again, higher estimated CCr and GFR values were observed when adjusting for smoking status for estimated CCr values and BMI and smoking status for estimated GFR values. Serum creatinine levels, socio-economic status, physical activity (data not shown), BMI, proportion of hypertensive subjects and fasting sugar level were the same between the two groups.

Table 2.

Clinical and biochemical characteristics of non-diabetes non-renal disease

Variable	Non-alcohol user	Alcohol user	P	Pa
No.	419	419
Age (years)	64.1±8.6	63.6±11.6	0.517
Gender (M/F)	212/207	212/207	1.000
BMI (kg/m2)	25.8±4.3	25.2±4.2	0.053
Current smoker (%)	28 (6.7)	129 (30.8)	<0.0001
% of hypertension	201 (49.9)	206 (50.9)	0.779
Systolic BP (mmHg)	132±15	137±21	<0.0001	<0.0001
Diastolic BP (mmHg)	81±10	84±14	0.002	0.001
Fasting sugar (mmol/l)	5.5±2.3	5.4±2.5	0.840
Total cholesterol (mmol/l)	4.92±0.95	4.71±0.99	0.002	0.019
Triglycerides (mmol/l)b	1.33±0.94	1.48±1.15	0.021	0.028
HDL cholesterol (mmol/l)	1.1±0.3	1.2±0.3	0.003	0.006
LDL cholesterol (mmol/l)	3.06±0.85	2.62±0.83	<0.0001	0.001
Uric acid (mmol/l)	0.38±0.10	0.41±0.10	<0.0001	<0.0001
Creat (µmol/l)	88.4±17.7	88.4±17.7	0.741
CGCr (ml/min)	60.5±18.1	64.1±24.3	0.020	<0.0001c
MDRDGFR-E (ml/min/1.73 m2)	71.0±15.0	76.3±18.9	<0.0001	<0.0001d

Data are expressed as mean±SD. Comparisons between non-drinkers and drinkers are performed by unpaired t-test or χ² test.

^aData are adjusted by general linear models for age, gender, BMI and smoking.

^bSignificant difference was tested using log-transformed data.

^cData are adjusted by general linear models for smoking.

^dData are adjusted by general linear models for BMI and smoking.

CGCr = estimated creatinine clearance calculated by the Modification of the Cockcroft–Gault Calculator [13] (adjusted by age, sex, weight and creatinine); MDRDGFR-E = estimated glomerular filtration rate calculated by the Modification of Diet in Renal Disease (MDRD) calculator-extended version (adjusted by age, sex, race, serum urea nitrogen, creatinine and albumin); BP = blood pressure; BMI = body mass index; HDL = high-density lipoprotein; LDL = low-density lipoprotein; Creat = creatinine.

Simple linear regression analysis revealed that age, BMI, DBP, fasting plasma sugar, serum total cholesterol, triglycerides, HDL cholesterol and alcohol drinking status were associated with estimated CCr. Age, BMI, smoking status, alcohol drinking status, DBP, fasting sugar, triglycerides and uric acid were associated with estimated GFR. Multiple linear stepwise regression analysis confirmed that alcohol drinking status, DBP, triglycerides and uric acid levels are independent contributing factors for estimated CCr, and that alcohol drinking status, triglyceride and uric acid levels are independent contributing factors for estimated GFR (Table 3).

Table 3.

Linear regression analysis of variables associated with estimated CCr and GFR

Variable	CCr													GFR
	Simple					Multiplea							Simple					Multipleb
	Estimate	SE	P	Estimate	SE	P	95% CI	Estimate	SE	P	Estimate	SE	P	95% CI
Age	−1.45	0.05	<0.001					−0.58	0.06	<0.001
Sex	−1.46	1.52	0.336					1.81	1.23	0.140
BMI	2.91	0.13	<0.001					0.24	0.11	0.041
Smoking	3.60	2.17	0.097					3.87	1.49	0.010
Alcohol drinking	4.42	1.58	0.005	7.68	1.56	<0.001	4.62–10.74	5.66	1.08	<0.001	6.11	1.07	<0.001	4.02–8.20
Systolic BP	0.01	0.04	0.820					0.02	0.03	0.448
Diastolic BP	0.27	0.06	<0.001	0.24	0.06	<0.001	0.12–0.35	0.12	0.04	0.018
Fasting sugar	0.13	0.06	0.033					0.10	0.04	0.035
Total cholesterol	0.05	0.02	0.019					−0.03	0.01	0.122
Triglyceridesc	0.05	0.01	<0.001	0.05	0.01	<0.001	0.03–0.06	0.03	0.01	0.006	0.02	0.01	0.004	0.01–0.03
HDL cholesterol	−0.26	0.09	0.006					−0.06	0.08	0.367
LDL cholesterol	0.06	0.04	0.130					−0.03	0.03	0.348
Uric acid	−0.48	0.45	0.287	−1.23	0.43	0.005	(−2.07)–(−0.38)	−0.68	0.31	0.029	−1.17	0.31	0.0001	(−1.77)–(−0.57)

^aIn multiple linear stepwise regression analysis, all covariates (holds age, sex and BMI) were used for analysis.

^bIn multiple linear stepwise regression analysis, all covariates (holds age and sex) were used for analysis, n = 838.

^cSignificant differences tested using log-transformed data.

CCr = estimated creatinine clearance calculated by the modification of the Cockcroft–Gault Calculator [13] (adjusted by age, sex, weight and creatinine); GFR = glomerular filtration rate based on the Modification of Diet in Renal Disease calculator-extended version (adjusted by age, sex, race, serum urea nitrogen, creatinine and albumin); BMI = body mass index; BP = blood pressure; HDL = high-density lipoprotein; LDL = low-density lipoprotein.

Table 4 shows the adjusted mean values of estimated CCr and GFR, serum total cholesterol, HDL cholesterol, LDL cholesterol, triglycerides, SBP, DBP and WHR in subjects categorized based on the amount of alcohol consumption. There are significant trends in the association of alcohol consumption doses with estimated CCr and GFR, HDL cholesterol, LDL cholesterol, triglyceride values, SBP and DBP, even after adjusting for age, gender, smoking status and BMI. Figure 1 shows the correlations between estimated CCr, GFR values and amount of alcohol consumption. The amount of alcohol consumption showed a significantly positive relationship with estimated GFR values (r = 0.116, P = 0.018) and estimated CCr values (r = 0.097, P = 0.049).

Fig. 1.

Correlation between amount of alcohol consumption and estimated creatinine clearance (CCr) was calculated by the Modification of the Cockcroft–Gault Calculator [13] (adjusted by age, sex, weight and creatinine) and estimated glomerular filtration rate (GFR) calculated by the Modification of Diet in Renal Disease (MDRD) Calculator-extended version (adjusted by age, sex, race, serum urea nitrogen, creatinine and albumin). The amount of alcohol consumption had a significant and positive association with Cockcroft–Gault CCr and MDRD GFR.

Table 4.

Mean drinking duration, estimated CCr, GFR, lipid profiles, blood pressure and WHR according to quantities of alcohol consumed

Variable	Alcohol consumption group (g/month)							P-value for trend
	A (0)	B (≤150)	C (151–1500)	D (>1500)
No. (%)	419 (50.0)	297 (35.4)	103 (12.3)	19 (2.3)
Drinking duration (years)a	–	24.6±14.9	29.1±15.1	32.2±12.2	0.002
CGCr (ml/min)	68.5±0.8	70.5±1.4	71.3±2.4	74.7±5.8	0.039
MDRDGFR (ml/min/1.73 m2)	76.4±0.8	77.7±0.9	79.6±1.7	88.7±4.6	0.023
Total cholesterol (mmol/l)	5.01±0.05	4.88±0.05	4.78±0.10	4.94±0.23	0.113
HDL cholesterol (mmol/l)	1.1±0.0	1.3±0.0	1.3±0.0	1.2±0.1	0.001
LDL cholesterol (mmol/l)	3.09±0.05	2.78±0.05	2.57±0.10	2.18±0.20	<0.0001
Triglycerides (mmol/l)b	1.14±0.01	1.28±0.01	1.47±0.01	2.12±0.01	<0.0001
Systolic BP (mmHg)	138±1	146±2	146±2	131±6	<0.0001
Diastolic BP (mmHg)	82±1	86±1	87±1	82±3	<0.0001
WHR (cm)	0.88±0.01	0.88±0.01	0.89±0.01	0.85±0.02	0.612

Data are expressed as mean±SEM adjusted for the following confounders, if appropriate: age, gender, smoking and body mass index for lipids, blood pressure; adjusted age, gender and smoking for WHR; adjusted smoking for CGCr; adjusted smoking and body mass index for MDRDGFR.

^aData are expressed as mean±SD.

^bSignificant differences tested using log-transformed data.

CGCr = estimated creatinine clearance calculated by the Modification of the Cockcroft–Gault Calculator [13] (adjusted by age, sex, weight and creatinine); MDRDGFR = glomerular filtration rate based on the Modification of Diet in Renal Disease calculator-extended version (adjusted by age, sex, race, serum urea nitrogen, creatinine and albumin); HDL = high-density lipoprotein; LDL = low-density lipoprotein; BP = blood pressure; WHR = waist-to-hip ratio.

Discussion

The principal finding of the present study is that subjects with chronic alcohol consumption have significantly higher estimated GFR and CCr values than non-drinkers. Our study also confirms the effects of alcohol ingestion on blood pressure and lipid profiles in previous studies [2,5]. This is the first epidemiological study to use estimated CCr and GFR calculations to demonstrate that alcohol consumption may have an effect on renal function.

In the present study, we found that the mean serum creatinine level in subjects with alcohol consumption was not statistically different from that of non-drinkers. Serum creatinine concentration is widely used as an index of renal function, but creatinine concentration measurement may be affected by factors other than GFR. GFR measurement has been well known as the gold standard of renal function assessment [15] and probably the best variable for diagnosing and monitoring kidney disease. However, GFR measurement is expensive and labour-intensive and therefore not widely used in routine measurement of renal function in large-scale epidemiological studies. Cockcroft and Gault [13] and Levey [15] proposed formulae for CCr and GFR calculation, which were found reasonably to estimate renal function. However, possible factors influencing the results should be mentioned to avoid bias [16]. In our work, the study design and laboratory analysis were arranged in order to avoid this possible inconsistency. Serum creatinine measurement was rigorously supervised for quality and centralized in one laboratory. Jaffé's method was used to reduce possible bias by multipoint calibration against authenticated serum-based standards to give the serum creatinine through true creatinine values. Study populations were also ethnically consistent.

Alcohol may have both detrimental and salutatory effects on renal function. The adverse effects may include glomerular and tubulo-interstitial damage, especially in subjects with pre-existing renal disease or with alcohol overconsumption. In micropuncture experiments, high concentrations of alcohol do not interfere with tubular cell function [17], while in vitro studies show that ethanol interferes with the Na⁺K⁺-ATPase of rabbit renal brush border membrane vesicles [18]. In the present study, the estimated CCr and GFR were calculated, although it is possible that alcohol consumption had an effect on creatinine tubular secretion. The mechanism relating alcohol consumption to renal function regulation is not clearly known, but most plausible is the connection of alcohol-related effects on blood pressure, lipid profile and vasoactive peptides regulation.

Ethanol intake may induce both hypertensive and antihypertensive effects. Parekh and Klag recently considered the main relationship between alcohol consumption and renal function regulation to be increased blood pressure with alcohol consumption [19]. In general, it is accepted that ethanol alters the activity of a number of neurotransmitters and hormonal systems that in turn mediate a number of its effects. Ethanol intake may inhibit the activity of substances having a suppressor effect and/or stimulate the activity of substances having antihypertensive effects [20]. Thus, alcohol consumption may affect regulation of vasoactive substances, consequently affecting renal haemodynamics and GFR.

There are several limitations in the present study. As in all studies that rely on participants’ report, recall bias and social desirability bias may have produced a spurious association. Furthermore, our definition of non-drinkers (i.e. who never consumed >12 drinks per year in their lives) most probably includes some people who were infrequent drinkers. The inclusion of infrequent drinkers in the non-drinkers’ category will cause the odds ratios (ORs) for moderate alcohol intake to be underestimated. Subjects with alcohol abuse may not have responded to the invitation to participate in the present study and may have had severe interference of renal function not included in the present study. However, the effects of alcohol abuse on renal function are another issue for study. We studied only aborigines of Taiwan, so caution is needed in generalizing our conditions in other populations. The MDRD formula is not tested for accuracy in renal function evaluation of the present study population, and thus may be used for relative measurements of GFR and comparisons between alcohol and non-alcohol users.

In summary, the present data indicate that alcohol consumption may have an effect on GFR and CCr, and have diverse systemic effects on blood pressure and lipid profiles. However, the present analysis should not be construed as endorsing heavy alcohol consumption for longevity. Instead, our goal is to evaluate potential pathways for the observed effects between alcohol and renal function in alcohol intake using epidemiological data. Moderate alcohol intake consistent with public health recommendations should be encouraged as part of a global health strategy. The results of our study raise an important public health issue and need to be confirmed by large-scale and cohort studies in other populations.

We are grateful to the staff of the Aboriginal Health Promotion Center and the health bureaus in aboriginal communities for their assistance in measurements and other organizational aspects of this study. This work was supported by grants from the National Science Council (93-2314-B-475-002) and the Department of Health of Taiwan (DOH92-HP-1111).

Conflict of interest statement. None declared.

Author notes

1Department of Clinical Research, Pingtung Christian Hospital, Pingtung 900, 2Graduate Institute of Dental Sciences, 3Graduate Institute of Oral Health Sciences and 4Graduate Institute of Medicine, Kaohsiung Medical University, Kaohsiung 807, Taiwan