American Journal of Epidemiology Copyright © 1992 by The Johns Hopkins University School of Hygiene and Put*; Health Al rights reserved

vol. 136, No 10 Printed in U.S.A.

Alcohol, Smoking, Coffee, and Cirrhosis

Arthur L. Klatsky1 and Mary Anne Armstrong2

Since most heavy drinkers do not develop alcoholic cirrhosis, other causes or predisposing factors are probable. The authors studied traits of 128,934 adults who underwent health examinations at the Oakland and San Francisco, California, facilities of the Kaiser Permanente Medical Care Program from January 1978 to December 1985 in relation to subsequent hospitalization or death from cirrhosis of the liver. In analyses adjusted for nine covariates, past and current alcohol drinking were strongly related to cirrhosis risk, but usual choice of alcoholic beverage had no independent relation. Cigarette smoking was independently related to risk of alcoholic cirrhosis, with cigarette smokers of a pack or more per day at trebled risk compared with lifelong nonsmokers. Coffee drinking, but not tea drinking, was inversely related to alcoholic cirrhosis risk, with persons who drank four or more cups per day at one-fifth the risk of noncoffee drinkers. This inverse relation between coffee consumption and risk of alcoholic cirrhosis was consistent in many subsets, including persons free of gastrointestinal disease and those with 5 or more years before hospitalization or death. Cigarette smoking and coffee consumption were not consistently related to risk of hospitalization or death for nonalcoholic cirrhosis. These data could mean that cigarette smoking promotes alcoholic cirrhosis and that coffee drinking might be protective. Am J Epidemiol 1992; 136:124857. alcohol drinking; caffeine; coffee; liver cirrhosis; liver cirrhosis, alcoholic; risk factors; tea; tobacco

Chronic ingestion of substantial amounts of alcohol is the leading cause of liver cirrhosis in developed countries, yet only a minority of heavier drinkers develop the condition (1-5). The hepatic toxicity of alReceived for publication September 16, 1991, and in final form June 9, 1992. Abbreviations: Q, confidence interval; ICO-8, International Classification of Diseases, Eighth Revision; ICD-9, International Classification of Diseases, Ninth Revision; RR, relative risk. 1 Department of Medicine, Kaiser Permanente Medical Care Program (Northern California Region), Oakland, CA. 2 Division of Research, Kaiser Permanente Medical Care Program (Northern California Region), Oakland, CA. Reprint requests to Dr. Arthur L. Klatsky, Department of Medicine, Kaiser Permanente Medical Care Program, 280 West MacArthur Boulevard, Oakland, CA 94611. Supported by the Alcoholic Beverage Medical Research Foundation, Inc., Baltimore, MD The authors thank Harald Kipp for computer programming, Cynthia Landy and Rita J. Coston for data collection, and Lyn Wender for technical assistance.

cohol, independent of nutritional deficiency, seems well established (4-6). While total lifetime alcohol dose is considered a factor in cirrhosis risk (3, 4), steady daily drinking may carry a higher risk than binge drinking (1,2,4). Although the matter is incompletely resolved, usual choice of alcoholic beverage (wine, beer, or hard liquor) has not been established as an independent factor (1,2, 4). Women are considered to be more susceptible to hepatotoxicity by alcohol (4-8). Differences in estimates of the safe limit of daily consumption probably reflect variations in individual susceptibility and problems related to accuracy of reported alcohol intake. Frequently cited limits range from 40 to 80 g per day (9, 10) in men and are as low as 20 g per day (9) in women. Thus, it is widely believed that alcoholic cirrhosis has multiple causes and that both

1248

Alcohol, Smoking, Coffee, and Cirrhosis

initiation and progression of the process may be dependent upon predisposing factors (2). In addition to alcohol dose, drinking pattern, sex, and, possibly, nutritional deficiency (7, II, 12), there is some evidence that a chronic hepatitis B virus carrier state may be a risk factor for alcoholic cirrhosis (13, 14). Other suspected but speculative risk factors include genetic predisposition, immune-mediated mechanisms, and interactions of alcohol with other unspecified toxic substances (15). Since there is a clear need for further study of other traits that might affect alcoholic cirrhosis risk (2, 5), we present here prospective data from a large, free-living population. This study examines several traits and habits in relation to subsequent hospitalization and death from cirrhosis.

MATERIALS AND METHODS

We studied 128,934 persons who underwent health examinations at the Oakland and San Francisco facilities of the Kaiser Permanente Medical Care Program, a prepaid health plan, from January 1978 through December 1985. The study population comprised 79.8 percent of all persons who underwent the health examination during the years of data collection. The remaining 20.2 percent included persons who were examined during absences of the research clerk, persons who declined, and those who failed to supply required inclusion data (age, sex, race, and alcohol use). Precise numbers for these subsets of excluded persons are not available. The multiphasic health examination, usually taken as a routine health examination, included a questionnaire which elicited information about sociodemographic traits, habits, and health history, as well as health measurements and tests. For persons who had more than one examination, data from the first one were used. On the basis of questionnaire items about alcohol use, we defined as lifelong abstainers nonusers who indicated that they had never or almost never drunk alcohol in the past. Ex-drinkers were those who had been non-

1249

drinkers in the previous year, but who indicated previous intake of alcohol. Drinkers were asked to categorize intake by usual number of daily drinks during the previous year, with the following options: nine or more per day, 6-8 per day, 3-5 per day, 12 per day, less than one per day but one or more per month, and less than one per month (special occasions only). Drinkers also indicated the number of days per week that they consumed wine, hard liquor, and beer. Drinkers were asked whether intake during the previous year was similar to that during the last 10 years, and if intake was reduced, the reasons for reduction. Subjects were asked whether they had ever smoked cigarettes, whether they had smoked during the previous year, and, if they were current smokers, the number of cigarettes smoked per day. Questionnaire items included "do you drink coffee?" and "do you drink tea?" with the following options for each: more than six, 4-6, and 1-3 cups per day, less than 1 cup per day, and "never or seldom." No information was available about prior changes in coffee or tea drinking habits. Partial correlations (controlled for age) were determined for the habits in the entire population. Hospitalizations at Northern California Kaiser Permanente facilities were ascertained through December 1988 or until subjects left the health plan. Hospitalization for cirrhosis was detected by computer search for a primary discharge diagnosis of International Classification ofDiseases, Adapted, Eighth Revision (ICDA-8), code 571. International Classification of Diseases, Ninth Revision (ICD-9) codes were converted to ICDA-8 codes after 1978. The first hospitalization for each person was used. Use of ICDA-8 codes enabled us to employ an existing data set for study of hospitalizations. Hospitalizations for cirrhosis were classified as alcoholic cirrhosis (ICDA-8 code 571.0) or as nonalcoholic cirrhosis (ICDA-8 codes 571.8 and 571.9). Death was ascertained by matching names of persons no longer enrolled in the Kaiser Permanente program to state vital statistics records by an automated matching system (16) and was validated independently by two persons. Subjects were

1250

Klatsky and Armstrong

followed for mortality until date of death or December 1988, whichever occurred earlier. Primary death certificate diagnoses of cirrhosis were classified by ICD-9 codes as alcoholic (ICD-9 codes 571.0-571.3) or as nonalcoholic (ICD-9 codes 571.4-571.9). Those deaths coded by ICDA-8 were reclas-. sified for analysis of deaths. We attempted to identify a subset free of history of liver disease or upper gastrointestinal history or symptoms which might have influenced habits. This was interpreted broadly. A person was considered to have upper gastrointestinal history if he or she answered yes to any of the 10 following medical history items: "Do you now have trouble swallowing"? ".. .abdominal or stomach pain"? ".. .liver trouble"? or ".. .unexplained weight loss"? "Has a doctor ever said you had gallstones or gallbladder problems"? ".. .liver disease (e.g., cirrhosis)"? ".. .stomach or duodenal ulcer"? ".. .hiatal (diaphragmatic) hernia"? ".. .operation on stomach"? or ".. .operation on gallbladder"? Subjects were excluded from analysis if basic information about age, sex, race, alcohol intake, or smoking was inadequate. Available hospitalization and outpatient records were reviewed to verify cirrhosis diagnosis. This review included 80 of 101 (79 percent) persons hospitalized for cirrhosis and 43 of 82 (52 percent) persons who died of cirrhosis. The review process resulted in reclassification of 10 hospitalized cases (eight from nonalcoholic to alcoholic and two from alcoholic to nonalcoholic). Six deaths were reclassified (four from nonalcoholic to alcoholic cirrhosis and two from alcoholic to nonalcoholic cirrhosis). For those unavailable for review, a diagnosis of alcoholic cirrhosis at hospitalization was used for death also, resulting in four more reclassifications of deaths from nonalcoholic to alcoholic cirrhosis. Drinking classifications were not changed on the basis of chart review. Multivariate analyses used the Cox proportional hazards model with the following covariates: age, sex, race, education, body

mass index (kilograms/meter2), marital status, and upper gastrointestinal history. Alcohol drinking (never-drinkers, ex-drinkers, and five categories of drinkers up to six drinks per day or more) and cigarette smoking (never-smokers, ex-smokers, less than one pack per day, and one or more packs per day) were studied as categorical variables. For nonalcoholic cirrhosis, the reference group for alcohol use was lifelong nondrinkers. For alcoholic cirrhosis, there were too few nondrinkers to use this category as the reference, so the reference group for alcohol use also included persons who reported current consumption of less than one drink per day. Definitions of beverage preference derived from the questionnaire items have been described previously (17). Coffee and tea consumption were studied as both categorical and continuous variables. For the study of coffee and tea as continuous variables, the following values were assigned to the categories: more than six cups = seven, 4-6 cups = five, 1-3 cups = two, less than one cup = 0.5, and never or seldom = 0. Because of missing values, the number of cases in analyses utilizing these continuous coffee and tea variables is less than the total number of cirrhosis cases. For economy, all analyses used a 10 percent sample of the entire population in addition to all cases. These analyses resulted in multiple comparisons. In this situation, a stricter definition of statistical significance (often p < 0.01) is required. The authors attempt to minimize the use of the word "significant" when describing findings. Assessment of the data should take into consideration the fact that repetitive patterns in multiple subsets add considerable strength to the credibility of individual findings of borderline "significance." RESULTS Traits of subjects and correlations of habits

Of 68 persons hospitalized for alcoholic cirrhosis, 22 died. Twenty-five persons died of alcoholic cirrhosis without hospitalization

Alcohol, Smoking, Coffee, and Cirrhosis

TABLE 1. Traits at initial examination ot persons hospitalized with cirrhosis and of noncases, Kaiser Permanente, 1978-1988 Entire study population

H capitalization

Trait

AfcohoSc cirmcws No.

Total Sex Male Female Race White Black Other UGI history • None 2:1 itemt Unknown Cigarette smoking (packs/day) Never Ex-smoker

Alcohol, smoking, coffee, and cirrhosis.

Since most heavy drinkers do not develop alcoholic cirrhosis, other causes or predisposing factors are probable. The authors studied traits of 128,934...
698KB Sizes 0 Downloads 0 Views