The Association among Smoking, Heavy Drinking, and Chronic Kidney Disease

Abstract

End-stage renal disease is an important public health problem. There were estimated to be 440,000 patients with end-stage renal disease in the United States as of 2003 (1), and based on earlier data an estimated additional 8 million US adults have chronic kidney disease (CKD), defined as a glomerular filtration rate of less than 60 ml/minute per 1.73 m² (2, 3), who are at risk of progression to end-stage renal disease and its attendant complications (4). Although a number of risk factors for CKD, including diabetes (4, 5), hypertension (3, 4, 6), and obesity (4, 7, 8), have been described consistently in the literature, studies looking at modifiable lifestyle factors such as smoking and consumption of alcohol are sparse and sometimes contradictory (8–12). In the current paper, we examined the factors associated with CKD in a population-based study in Wisconsin, with particular emphasis on the role of smoking and consumption of alcohol.

MATERIALS AND METHODS

Study population

The methods used to identify and describe the population have appeared in previous reports (13–15). In brief, a private census of the population of Beaver Dam, Wisconsin, was performed from September 1987 to May 1988 to identify all residents in the city or township of Beaver Dam who were 43–84 years of age. Of the 5,924 eligible persons (98 percent Caucasians), 4,926 (83.1 percent) participated in the baseline examination between March 1, 1988, and September 14, 1990. A total of 3,684 persons (81.1 percent) participated in the 5-year follow-up examination between March 1, 1993, and June 14, 1995. Comparisons between participants and nonparticipants at the time of the baseline and 5-year follow-up examinations have appeared elsewhere (13, 14). Both the baseline and follow-up examinations followed a similar protocol. Written, informed consent was obtained from each subject at each examination. The study was approved by the Human Subjects Committee of the University of Wisconsin School of Medicine and Public Health, Madison, Wisconsin.

The current paper presents two sets of analyses: 1) cross-sectional analysis with prevalent CKD as the outcome of interest and 2) longitudinal analysis with 5-year incident CKD as the outcome of interest. Of the 4,926 persons who participated in the baseline examination, 4,898 persons with serum creatinine measurements and complete covariate information form the study population for the cross-sectional analysis. Of the 3,984 persons who participated in the baseline and 5-year follow-up examinations, 3,392 persons with serum creatinine measurements taken at both of these examinations form the study population for the longitudinal analysis.

Exposure ascertainment

The baseline examination and the follow-up examination included measuring weight, height, and systolic and diastolic blood pressure by a trained observer and administering a standardized questionnaire that collected information regarding participants' demographic characteristics, details regarding cigarette smoking, consumption of alcohol, medical histories and medications taken, including the diagnosis of diabetes or hypertension by a physician, and use of nonsteroidal antiinflammatory agents (NSAIDs). Casual blood specimens were obtained for measurement of plasma glucose and serum creatinine.

Age was defined as the participants' age at the baseline examination. Education was categorized as below high school, high school, or beyond high school. Body mass index was defined as participants' weight (kg)/height (m)². Hypertension was defined as systolic blood pressure of 140 mmHg or higher, diastolic blood pressure of 90 mmHg or higher, or the combination of a self-reported hypertension diagnosis by a physician and use of antihypertensive medications. Persons were defined as having diabetes on the basis of a casual blood sugar measurement of higher than 200 mg/dl (11.1 mmol/liter), or if they had a history of diabetes diagnosis by a physician and were treated with insulin, oral hypoglycemic agents, or diet. History of cardiovascular disease was defined as the presence or absence of a physician-diagnosed episode of myocardial infarction, angina, or stroke.

Cigarette smoking status at the time of the baseline examination was determined as follows (15). A subject was classified as a nonsmoker if he/she had smoked fewer than 100 cigarettes in his/her lifetime, as a former smoker if he/she had smoked this number of cigarettes or more in his/her lifetime but had stopped smoking at least 1 year before the examination, and as a current smoker if he/she had not stopped smoking. The total pack-years smoked was defined as the number of cigarettes smoked per day divided by 20, multiplied by the number of years smoked. Former smokers were queried regarding the years since they stopped smoking. Years since stopped smoking among former smokers were categorized as 15 or more, 5–14, and less than 5.

The examination questionnaire contained questions regarding alcohol consumption (15). Subjects were asked about their average weekly use of beer, wine, and liquor in the previous year. In addition, they were asked about past periods of drinking, including whether or not they ever consumed four or more drinks daily. From these data, a current drinker was defined as a person who had consumed alcoholic beverages within the past year, a former drinker was a person who had consumed alcoholic beverages in the past but not within the previous year, and a nondrinker had never consumed alcoholic beverages. The distribution of alcohol consumption was highly skewed; 51.2 percent of the population consumes alcohol less than weekly or none at all. Alcohol consumption frequency was categorized as none/less than or equal to one serving per week, 2–4 servings per week, 5–6 servings per week, 1–3 servings per day, and four or more servings per day. To further evaluate heavy drinking, we created the following categories. A current heavy drinker was defined as a person consuming four or more servings of alcoholic beverages daily, a former heavy drinker had consumed four or more servings daily in the past but not in the previous year, and a non-heavy drinker had never consumed four or more servings daily on a regular basis.

Outcome of interest—chronic kidney disease

Chronic kidney disease was defined as an estimated GFR of less than 60 ml/minute per 1.73 m² on the basis of the National Kidney Foundation's Kidney Disease Outcome Quality Initiative working group definition (2). Prevalent CKD (n = 324) was defined as the presence of CKD among study participants at the baseline examination. Incident CKD (n = 114) was defined as the development of new CKD at the 5-year follow-up examination among study participants who did not have CKD at the baseline examination.

Statistical methods

We performed two sets of analyses: 1) cross-sectional analysis with prevalent CKD as the outcome of interest and 2) longitudinal analysis with 5-year incident CKD as the outcome of interest. In cross-sectional analysis, we examined the relation between various exposures, including age, gender, education, body mass index, current use of NSAIDs, hypertension, diabetes, history of cardiovascular disease, smoking, and alcohol consumption, and prevalent CKD. We examined the relation between smoking and prevalent CKD in detail by first examining smoking status (never smoker, former smoker, current smoker) and then the dose-response relation as pack-years of smoking (0, <15, 15–34, ≥35 pack-years) and years since stopped smoking among former smokers (≥15, 5–14, <5 years). We examined the relation between alcohol and prevalent CKD by examining alcohol consumption status (never drinker, former drinker, current drinker), alcohol consumption frequency (0–<1 serving per week, 2–4 servings per week, 5–6 servings per week, 1–3 servings per day, ≥4 servings per day), and the effect of heavy drinking, that is, four or more servings per day (never heavy drinker, former heavy drinker, current heavy drinker). In longitudinal analysis, we examined selected smoking and alcohol consumption variables found to be strongly associated in the cross-sectional analysis and their relation to incident CKD. We examined the effect of joint exposure to current smoking (absent, present) and current heavy drinking (absent, present) by creating four corresponding categories. Effect modification was formally evaluated by including cross-product interaction terms in the corresponding multivariable models. We were limited by the number of incident CKD cases to do more detailed analyses within different subgroups. For consistency in reporting, for both the cross-sectional and longitudinal analyses, we used multivariable logistic regression models to calculate odds ratios and 95 percent confidence intervals. We used two models: the age- and sex-adjusted model and the multivariable-adjusted model additionally adjusting for education, body mass index, current NSAID use, hypertension status, diabetes status, history of cardiovascular disease, smoking status, and heavy drinking. We calculated the population attributable risk of CKD associated with current smoking and heavy drinking using Levin's formula as described by Hanley (18). We repeated the analysis for smoking and alcohol consumption in relation to kidney disease using the Cockcroft–Gault formula to estimate creatinine clearance with similar results (19). SAS, version 9.2, statistical software (SAS Institute, Inc., Cary, North Carolina) was used for all analyses.

RESULTS

The mean age of study participants at the baseline examination was 62.3 years (range: 43–86 years). The mean estimated GFR at the baseline examination of prevalent CKD cases compared with noncases was 51.8 and 98 ml/minute per 1.73 m², respectively. The mean baseline estimated GFR of those who developed incident CKD compared with those who did not develop CKD was 90.2 and 114 ml/minute per 1.73 m², respectively. The mean estimated GFR decline was 12 (interquartile range: 16.2) ml/minute per 1.73 m². In multivariable models (table 1), increasing categories of age, male gender, obese persons, use of NSAIDs, presence of hypertension, diabetes, and history of cardiovascular disease were associated with prevalent CKD. Lower education was associated with prevalent CKD in the age- and sex-adjusted model but failed to reach statistical significance (alpha = 0.05) in the multivariable model. Current smoking was associated with CKD; current alcohol consumption was not associated.

We examined the relation between smoking and prevalent CKD in more detail in table 2. Compared with those who never smoked, current smokers had a multivariable odds ratio of 2.2. The analysis of pack-years of smoking showed a strong gradient of association with increasing cumulative dose of smoking. Further, among former smokers, the association with CKD was higher among those with fewer years since stopped smoking compared with 15 or more years since stopping smoking.

We examined the relation between alcohol consumption and prevalent CKD in table 3. Taken together, only heavy drinking was associated with CKD. In comparing alcohol consumption frequency, heavy drinking (four or more servings of alcohol per day) was associated with an odds ratio of 1.6. Compared with persons who were never heavy drinkers, both former heavy drinkers (odds ratio (OR) = 1.3) and current heavy drinkers (OR = 1.8) were more likely to have CKD. Lower levels of alcohol consumption do not appear to be harmful.

In subgroup analyses (table 4) by gender, the association between current smoking and CKD appears to be stronger among men (OR = 3.3) than among women (OR = 1.4). The association between heavy drinking and CKD was present among both women and men but failed to reach statistical significance (alpha = 0.05) among women.

In longitudinal analysis (table 5), current smoking and current heavy drinking at baseline were significantly associated with incident CKD. Furthermore, joint exposure to both current smoking and current heavy drinking (table 6) was associated with a significantly higher odds ratio of incident CKD than were their respective individual associations. The population attributable risk of CKD associated with current smoking was 10 percent and with heavy drinking was 4.8 percent.

We performed several sets of supplementary analyses. First, we repeated the longitudinal analysis (table 5) with proportional hazards models. Compared with that of never smokers (referent group), the multivariable hazards ratio of incident CKD for former smokers was 1.12 (95 percent confidence interval (CI): 0.63, 1.98) and for current smokers was 1.93 (95 percent CI: 1.15, 3.25). For heavy drinkers, compared with never heavy drinkers (referent group), the multivariable hazards ratio for former heavy drinkers was 1.29 (95 percent CI: 0.65, 2.56) and for current heavy drinkers was 1.80 (95 percent CI: 0.89, 3.66). Second, we performed the longitudinal analysis after excluding 42 additional persons with dipstick proteinuria (defined as urinary protein of ≥0.3 g/liter) at the baseline examination, as we suspected that such persons either had CKD or were at risk of CKD. Compared with that for never smokers, the multivariable odds ratio of incident CKD for former smokers was 1.12 (95 percent CI: 0.60, 2.07) and for current smokers was 1.89 (95 percent CI: 1.06, 3.39); compared with that for never heavy drinkers, the multivariable odds ratio for former heavy drinkers was 1.10 (95 percent CI: 0.47, 2.56) and for current heavy drinkers was 1.86 (95 percent CI: 0.79, 4.37). Third, we repeated the analysis with an alternate definition of incident kidney disease: reduction in estimated GFR of 50 percent or more over 5 years (n = 105); the overall results were similar. Compared with that for never smokers, the multivariable odds ratio of incident CKD for former smokers was 1.08 (95 percent CI: 0.58, 1.99) and for current smokers was 2.04 (95 percent CI: 1.18, 3.54). Similarly, compared with that for never heavy drinkers, the multivariable odds ratio for former heavy drinkers was 1.19 (95 percent CI: 0.57, 2.48) and for current heavy drinkers was 1.89 (95 percent CI: 0.90, 3.98). Finally, the result for the association between smoking and alcohol consumption with incident CKD was similar when the Cockcroft–Gault formula was used to estimate creatinine clearance.

DISCUSSION

In a population-based sample consisting predominantly of older adults, smoking was found to be associated with chronic kidney disease independent of body mass index, NSAID use, alcohol consumption, hypertension, diabetes, and other confounders. The association between smoking and CKD was supported by evidence of a dose-response trend. We also found an independent association between heavy drinking (≥4 servings of alcohol per day) and CKD. Further, joint exposure to smoking and heavy alcohol consumption was associated with almost fivefold odds of developing CKD than was their absence.

Smoking has been shown to be associated with end-stage renal disease previously (5). The association between smoking and stages of kidney disease earlier in the continuum is increasingly being recognized. Several previous studies have identified smoking as a potential risk factor for CKD among those with diabetes (20, 21). However, similar studies in the general population have not been consistent. Several case-control studies failed to detect an association between smoking and CKD (8, 10). In contrast, other case-control (12), cross-sectional (9, 22), and recent longitudinal (4, 23) data support the hypothesis of an association between smoking and CKD.

In the current study, the association between smoking and CKD was present in both cross-sectional and longitudinal analyses. The findings of a higher odds ratio of CKD associated with increasing pack-years of smoking and of the inverse association with years since stopped smoking, taken together, are supportive of a dose-response trend. The magnitude of association between smoking and CKD, its independence from traditional CKD risk factors, evidence of dose-response trend, and the consistency within subgroup analysis by gender all suggest that these findings are less likely to be due to chance. Further, a number of biologic mechanisms by which smoking can result in kidney damage have been identified, including the promotion of renal atherosclerosis (24), alterations in systemic and renal hemodynamics (25), and effects on endothelial function (26).

In the current study, consumption of four or more servings of alcohol per day was found to be independently associated with CKD in both cross-sectional and longitudinal analyses. This finding is consistent with results from previous case-control studies (10, 27). In another case-control study, CKD was associated only with moonshine consumption, not other alcoholic beverages (8). In a recent prospective study, Schaeffner et al. (11) reported a protective association between moderate alcohol consumption and incident CKD among US male physicians; compared with men who consumed no more than one drink per week, men who consumed at least seven drinks per week had an odds ratio of 0.71 (95 percent CI: 0.55, 0.92). Heavy alcohol consumption was not studied. In our study, moderate alcohol consumption was not harmfully associated with CKD; however, we failed to observe a statistically significant (alpha = 0.05) protective effect.

In the current study, joint exposure to both current smoking and heavy drinking was associated with a higher odds ratio of CKD than were their individual effects. It is possible that the biologic mechanisms by which smoking can result in kidney damage may be accentuated by the effect of heavy drinking on the kidney, including alcohol-induced hypertension, rhabdomyolysis, or the direct toxic effects of alcohol (28–30). Almost 60 percent of heavy drinkers in our sample were also current smokers. These data indicate that, regarding kidney disease, individual-level interventions aimed at addressing both smoking and heavy alcohol consumption simultaneously may be more beneficial. In contrast, at the population level, 10 percent of CKD in the Beaver Dam Township appeared to be related to smoking, and 5 percent appeared to be related to heavy drinking, suggesting population-level interventions targeting smoking to be more beneficial.

Several study limitations need to be considered when interpreting our results. The relatively homogeneous nature of our cohort (98 percent Caucasians) limits generalizability, particularly to high-risk groups for kidney disease, such as African Americans; however, it reduces confounding. Our creatinine measurements were not directly calibrated to Cleveland Clinic standards; studies have shown that the measure of serum creatinine can vary across laboratories and that calibration differences can account for differences in GFR, particularly at higher values (16). However, the incidence and prevalence of kidney disease in our sample are similar to published estimates in other Caucasian studies, including the Framingham Offspring Study for incidence (4) and the non-Hispanic White subgroup of the Third National Health and Nutrition Examination Survey for prevalence (3), suggesting that the CKD estimates from our study are comparable with those from other general population samples. The definition of CKD in the current study differs from the National Kidney Foundation's definition (2) in that it excludes kidney damage (no data on albuminuria) and is limited to one visit (no measure of chronicity). Stages 1 and 2 of CKD were not considered in our study as data on albuminuria were not available. A small change in GFR, for example, from 62 to 58 ml/minute per 1.73 m², could have led to individuals being classified as incident cases. However, because of the prospective nature of the study, the misclassification is likely to be nondifferential and to underestimate the true association. Moreover, the main study results remained essentially similar in a supplementary analysis examining 50 percent or more reduction in estimated GFR as the outcome of interest, suggesting these findings to be relatively robust. Potential laboratory drift in the measurement of serum creatinine could result in misclassification of incident CKD cases in the longitudinal analysis; this misclassification is likely to be nondifferential and to underestimate the true association. We were limited by our sample size to analyze incident kidney disease in greater detail. Smoking and drinking habits can change with time. Questionnaire-based data collection has the chance of misclassification. However, other reports, such as that by Klein et al. (15), related to smoking and alcohol consumption from our cohort are similar to or have been validated by other studies, suggesting that this population provides reliable reports of smoking and alcohol use. In addition, the prospective method of exposure measurement in our study most likely would lead to random misclassification; resultant bias is potential underestimation of the association among smoking, alcohol consumption, and kidney disease.

In summary, in this population-based study, smoking was found to be associated with chronic kidney disease, independent of several important confounders. Similarly, heavy drinking, defined as consumption of four or more servings of alcohol per day, was also associated with chronic kidney disease. Individuals who were both current smokers and heavy drinkers had substantially higher odds of developing kidney disease, suggesting a greater potential beneficial effect for simultaneously addressing both of these lifestyle risk factors in the clinical setting.

Supported by National Institutes of Health grant EY06594 (R. K., B. E. K. K.) and, in part, by Research to Prevent Blindness, New York, New York (R. K., B. E. K. K., Senior Scientific Investigator Awards).

Conflict of interest: none declared.

References