- Email: [email protected]
Alcohol consumption and physical symptoms in a Mexican American population
Drug and Alcohol Elsevier Scientific
Dependence, 16 (1986) Publishers Ireland Ltd.
ALCOHOL CONSUMPTION AMERICAN POPULATION
369-379
AND PHYSICAL
CARLOS F. MENDES DE LEON and KYRIAKOS Department of Preventive Medicine and Community Branch, Galveston, TX 77550 (U.S.A.) (Received (Received
369
SYMPTOMS IN A MEXICAN
S. MARKIDES Health,
University
of Texas Medical
June 24th, 1955) June 24th, 1985)
ABSTRACT
The negative health consequences of alcohol consumption are well established. The present study examines whether alcohol consumption is also associated with a series of self-reported physical symptoms. Analyses were performed on data from a three-generation study of Mexican Americans. Using four measures of alcohol consumption we found that alcohol consumption was related to physical symptoms only in the younger generation. While drinking levels in the older generation did not have sufficient variance to enable the assessment of their effects on symptoms, drinking levels in the middle generation were comparable to those in the younger generation. We hypothesized that the lack of association in the middle generation may reflect increased tolerance due to londer exposure to alcohol, to possible reduction of alcohol consumption by some middle-aged persons because of health problems, and to due possible removal from observation of certain middle-aged men whose heavy drinking led to health problems. Although alcohol consumption in the younger generation was not strongly associated with either psychosomatic or infectious symptoms, it was strongly related to upper respiratory symptoms among younger males. Among younger women, alcohol consumption was moderately related to physical symptoms with no symptom constellation having a particularly strong association with alcohol consumption. Key words: Mexican Americans - Alcohol Age differences - Physical symptoms
consumption
- Generations
-
INTRODUCTION
Extensive research has been done on the negative health consequences of alcohol. An overview of this can be found in The Fifth Special Report for 0376-8716/86/$03.50 o Elsevier Scientific Publishers Ireland Ltd. Printed and Published in Ireland
the U.S. Congress on Alcohol and Health, issued by the National Institute on Alcohol Abuse and Alcoholism [l] . From this report it can be learned that alcohol abuse is associated with increased mortality from such diseases as alcoholism, alcoholic psychosis, diseases of the digestive system, of the liver, and of the nervous system. Excessive alcohol consumption is also related with many adverse social consequences, such as traffic accidents, crime, homicide, suicide and family problems [l] . Moderate consumption, however, has been linked to less coronary heart disease [2,3], and is thought to serve as a buffer between stress and somatic depressive symptoms [ 4-61. Although there is a wealth of research on the effect of alcohol consumption on various diseases as well as on psychological symptoms, little is known about its effect on general health status. Belloc and Breslow [ 71 found that both men and women who averaged more than 5 drinks at a single occasion reported significantly poorer health status as measured by a composite index that included chronic conditions, disability, presence of physical symptoms and energy level. In a study of younger and older Navy men, it was found that problem drinkers had higher rates of morbidity as measured by number of hospital admissions and average length in a hospital, than non-alcoholic Navy controls. Older drinking Navy men, in particular, were more prone to accidents, violence and poisoning, mental disorders, diseases of the digestive system and infectious/parasitic diseases. They also had significantly more hospital admissions because of ‘symptoms and ill defined conditions’ [ 81. Among younger drinking Navy men a similar pattern was found. Infections of the respiratory system, accidents, mental disorders and gastrointestinal disorders were especially prevalent in this group. Younger drinking Navy men also had higher rates of hospital admissions because of ‘symptoms and ill-defined conditions’ [ 91 . Another study related physical health status to an index of alcohol use among three different age cohorts (18-29,30-49, and 50+ years) in the general population [lo]. The index of alcohol consumption divided the sample in a group at low risk for alcoholism and alcohol-related problems, and a high risk group. Physical health status was measured by one self-report item about present physical health. The high-risk groups in all three cohorts reported having poor or very bad health more often than the low risk groups, although this association reached statistical significance for only the youngest age group (18-29 years). The purpose of this study is to investigate the effect of alcohol consumption on general physical health. More specifically, do heavy alcohol consumers experience more physical symptoms than do light drinkers or non-drinkers? The three-generations study of Mexican Americans provided an opportunity to examine the relationship between alcohol consumption and physical symptoms at three different stages in the life cycle in an understudied population. Hispanics have been found to have a high percentage of abstainers (particularly among women) and a higher percentage of heavy
371
drinkers among men than the general population [ 11,121. The difference in alcohol consumption between men and women among Mexican Americans is wider than among other ethnic groups [13,14]. Therefore, analyses will be conducted separately for men and women. In general, we expect increasing levels of alcohol consumption to be associated with poorer general health status as measured by the number of self-reported physical symptoms. METHODS
The sample
The data employed in this study were drawn from interviews conducted in 1981 and 1982 with Mexican Americans in the San Antonio area. Area probability sampling involving selection of census tracts and blocks was employed to identify Mexican Americans aged 65-80 who had a child in the metropolitan area who, in turn, had an adult (18 years +) married or previously married child who also resided in the area [ 151. In the event that two or more such lineages were available, one was selected randomly. A total of 1125 persons were interviewed, 375 in each generation. About two-thirds of the sample was female, to a large extent due to the women’s greater longevity in the older generation and their tendency to marry and bear children at younger ages than the men. The mean age of the older generation was 74, for the middle 49, and for the younger generation 26. Measures
Alcohol consumption was assessed by two questions: ‘During the past 2 months, how often have you drunk wine, beer, or drinks containing alcohol? (coded l-6, from never to once a day or more); and, ‘when you drink wine, beer or drinks containing alcohol, about how many glasses do you usually have in one day?’ (the exact number was coded up to 7 or more glasses). These two measures were used separately in the analyses. A combination of them was also computed to form a frequency by volume measure by multiplying the two measures. In addition, a dummy variable was created to obtain a dichotomous variable of high levels of drinking: a value of one was assigned to persons consuming 6 or more drinks on a given day and zero to all others. Because of the small number of women in the high drinking category, analysis involving this variable was confined to men. The presence of physical symptoms was measured by a list of 20 commonly occurring complaints. Subjects were asked how often they had been bothered by any of the 20 symptoms in the past 2 or 3 months. Responses were coded on a 4-point scale: never, once in a while, often, most of the time (coded l-4). The list included the following symptoms: headaches, colds, asthma, difficulty in breathing, stomach problems, diarrhea, muscle ache, backache, shaky or trembly feeling, difficulty with eye sight not corrected with glasses, allergic reactions, chest pain, poor circulation, skin rash, fatigue, difficulty
372 hearing, flu, sore throat, infection and cough. The total symptom scale proved to have an alpha reliability-coefficient of 0.77. We performed factor-analysis using the maximum likelihood method with oblique rotation (Oblimin) of the factors, in order to examine the factor structure of the symptom-scale. The factor analysis yielded two interpretable factors. The first factor was labelled the psychosomatic symptom subscale, since it included symptoms usually associated with complaints of psychosomatic origin: headaches, stomach problems, muscle aches, backaches, shaky or trembly feelings, chest pains, poor circulation and fatigue. The second subscale consisted of symptoms that are usually associated with infectious ailments: colds, flu, sore throat, infection and cough. The alpha reliabilitycoefficients were 0.72 for the psychosomatic symptoms subscale and 0.76 for the infectious symptoms subscale. Four sociodemographic variable served as controls: age (measured in actual years of age), education (years of school completed), income (monthly household income, ranging from $0-199 to $2000 and over, and coded l---11), and marital status (married = 1, not married = 0). Analysis
We employed a series of multiple regression analyses that controlled for the sociodemographic variables in order to assess the independent association between the alcohol variables and the symptom scales. The analysis was conducted separately by generation and sex. Since number of drinks consumed and the frequency by volume measure were skewed for both men and women in all three generations, we employed a logarithmic transformation of these variables. RESULTS Table I presents data on frequency of alcohol consumption, the average number of drinks consumed on a given day by those who said they drank at least a few times a month, and percent of all subjects who consumed an average of 6 or more drinks on a given occasion, by generation and by gender. The table shows that in all generations, men report drinking considerably more than women, and that the younger two generations drink more than the older generation. Alcohol consumption levels look similar for older men, middle-aged women and younger women. Middle-aged men and younger men report drinking more frequently than the other groups, and their alcohol consumption levels appear to be about the same. We conducted zero-order correlation analysis between the alcohol consumption variables and the three symptom scales by generation and sex. No significant associations were found in the older generation. For the middleaged women, only 1 out of 12 correlations (between number of drinks and the total symptom-scale) reached statistical significance (r = 0.11, P < 0.05). For middle-aged men, also only one of the correlations between the alcohol
I
AND
VOLUME
12.4
39.0 27.6
13
41 29
33.3 29.4 5.9 20.6 5.9 2.9 2.9
100.0
11 10 2 7 2 1 1
34
Total
100.0
6.7 8.6
7 9
105
5.7
6
No. of glasses 1 2 3 4 5 6 7+
Total
Once a day or more Several times a week Once a week A few times a month Once a month or less frequently Never
Frequency
-
-
15
2
5 8
269
39 216
4
2 4
4
1.5
0.7 1.5
1.5
13.3
100.0
-
-
33.3 53.3
100.0
14.5 80.3
%
76
10 13 12 9 4 26 2
125
30 13
14
24 29
15
No.
No.
No.
%
Male
Female
24.0 10.4
11.2
19.2 23.2
100.0
13.2 17.1 15.8 11.8 5.3 34.2 2.6
100.0
AND SEX
65
l? 1; 13 6 6 6 2
125
84 94
39
7 17
4
No.
Female
2.9 6.9
1.6
100.0
20.0 29.2 20.0 9.2 9.2 9.2 3.1
100.0
34.3 38.4
15.9
%
87
7 16 18 8 6 24 8
143
38 10
31
32 28
4
No.
2.8
100.0
8.0 18.4 20.7 9.2 6.9 27.6 9.2
100.0
26.6 7.0
21.7
22.4 19.6
%
by persons
Male -
G3
of drinks at a single occasion
12.0
%
to the average number
BY GENERATION
Male
and volume
CONSUMPTION
G2
months
OF ALOCHOL
Gl
Frequency refers to previous.2 leaat ‘a few times a month’.
FREQUENCY
TABLE
93 79
37
3 19
60
12 16 17 6 2 5 2
231
-
No.
Female
40.3 34.2
16.0
1.3 8.2
100.0
20.0 26.7 28.3 10.0 3.3 8.3 3.3
100.0
-
96
who had drunk at
w
,”
314
variables and the symptoms was statistically significant: excessive drinking (6 drinks or more) at a single occasion was positively correlated to the infectious symptoms subscale (r = 0.22, P < 0.01). The correlation analysis proved more fruitful in the younger generation. All correlations for both sexes were positive, and almost all of them were statistically significant at the 0.05 level or beyond. Because of the lack of significant findings in the two older generations, we confined our regression analysis to younger men and women. The purpose of this analysis was to examine whether the significant correlations would persist after controlling for the effects of age, marital status, education and income. We also controlled for the effects of smoking (current smokers = 1; non-smokers = 0), because of its association with both alcohol consumption and physical symptoms. The correlations between smoking and the several measures of alcohol consumption were all highly significant (P < O.OOl), for both younger males and females, ranging from 0.20 to 0.43. Table II presents the results of the multiple regression analysis for the younger men. The results show that the significant correlations between alochol consumption and physical symptoms in general persisted after controlling for age, marital status, education, income and smoking. In particular the associations between the entire symptom scale and the psychosomatic symptoms scale and the several measures of alcohol consumption did not change very much after the controls. For the infectious symptoms scale, however, smoking entered the regression equation at a significant level for two of the four alcohol consumption variables, and approached statistical significance as a predictor variable in the equations for the remaining two variables. Table III displays the results of the regression analysis for the younger generation women. Again, the relationships between the physical symptoms scales and alcohol consumption were very similar to the results of the zeroorder correlations. Most of the correlations remained significant after the controls. We also examined the possiblility of a non-linear relationship between volume of alcohol consumption and physical symptoms. Such a non-linear relationship has been suggested by the hypothesis of the stress-buffering role of alcohol [ 4-61. We tested this possibility for 2nd and 3rd generation males separately by entering the square of the ‘number of drinks’ variable as a second step into the regression equation, after controlling for the linear effect of ‘number of drinks’ and the effects of the control variables. Only in the case of the psychosomatic symptoms subscale in younger men did we observe a significant non-linear (U-shaped) effect of number of drinks. However, the non-linear effect was considerably lower than the linear effect of number of drinks. It is interesting to note that Neff [6] found a similar relationship between alcohol consumption and a somatic symptoms scale, a scale very similar to our psychosomatic symptoms subscale. In Neff’s study, abstainers reported the most somatic symptoms, followed by the heavy drinkers and then the moderate drinkers.
0.235** 0.246** 0.037
0.26****
0.27****
0.08 < 0.001.
(0.307)
(1.459)
(2.375)
(0.557)
< 0.01; ***up
0.209**
beta
0.25****
r
Physicala symptoms
a0.05 < P < 0.10; **p < 0.05; ***p
Frequency of drinking Number of drinks Frequency x No. High drinking
Alcohol variables
MALES (/V = 142)
0.02
0.19**
0.17**
0.21***
r
-0.003
0.196*
0.174*
0.211x*
beta
Psychosomatic symptoms
(0.010)
(0.576)
(0.871)
(0.279)
0.247**
0.32****
0.082
0.243**
0.32****
0.16**
0.182**
beta
0.26****
r
Infectious symptoms
(0.328)
(0.700)
(1.175)
(0.231)
The regression analyses were performed with age, marital status,
FOR THIRD GENERATION
Numbers in parentheses are unstandardized regression coefficients. education and current smoking held constant.
RESULTS OF MULTIPLE REGRESSION ANALYSIS
TABLE II
0.198***
0.20****
(1.850)
(3.179)
< 0.01; ****p < 0.001
0.198***
(a) 0.175 (b) (0.747)
beta
0.20****
0.18***
r
Physica? symptoms
**0.05 < P < 0.10; **p < 0.05; ***p
Frequency of drinking Number of drinks Frequency x No.
Alcohol variables
FEMALES (N = 233)
0.16***
0.17***
0.14**
r
0.154**
0.161**
0.133*
beta
Psychosomatic symptoms
(0.762)
(1.371)
(0.302)
0.195*** 0.183*** 0.191***
0.17*** 0.18***
beta 0.17***
r
Infectious symptoms
(0.794)
(1.311)
(0.372)
The regression analyses were performed with age, marital status,
FOR THIRD GENERATION
Numbers in parentheses are unstandardized regression coefficients. education and current smoking held constant.
RESULTS OF MULTIPLE REGRESSION ANALYSIS
TABLE III
DISCUSSION
The results indicate that alcohol consumption was not related to general health status as measured by self-reported physical symptoms among older and middle-aged men and women in this Mexican American population. Since the older generation, in particular the older women, consume very low quantities of alcoholic beverages, a negative effect of alcohol consumption on health status could not be expected. As noted previously, middle-aged and younger men, and middle-aged and younger women in this population have approximately the same alcohol consumption levels. However, only the younger generation seems to experience adverse physical consequences associated with alcohol consumption. One explanation for this finding could be that middle-aged people have acquired a greater degree of tolerance for alcohol than younger people. Thus, the same amount of alcohol intake is less likely to produce symptomatic effects on middle-aged than on younger people. This cannot be inferred with certainty from cross-sectional data such as ours and we do not have information on previous drinking patterns. However, a longitudinal study by Glynn and colleagues [15] has found that drinking patterns among males remained relatively constant between 1973 and 1982. Furthermore, Caetano [ 171 reported that rates of heavy drinking and alcohol-related problems do not decrease significantly from the 20s to the 30s among Hispanics. It appears reasonable to assume that the majority of the middle-aged subjects in this study, in particular the men, have been consuming alcoholic beverages since their young adulthood. The presence of physical dependence on alcohol, particularly among the heavier drinkers, might also have influenced the relationship between alcohol consumption and physical symptoms. It is quite possible that a number of middle-aged persons drink in order to suppress physical symptoms. It is also possible that some middle-aged persons might have reduced their alcohol intake because of health problems, which would tend to lower the association between alcohol consumption and symptoms. This would probably be less likely among younger people, since they are less likely to stop drinking because of health problems. Finally, it must be acknowledged that generational differences in the association between alcohol consumption and physical symptoms might reflect the operation of differential selection. It is possible that certain middle-aged and older people whose heavy drinking led to poor health (symptoms), may have been removed from observation by death, institutionalization (hospital, prison, etc.), or some other factor. As noted above, the zero-order correlations between the drinking variables and the symptoms scales among younger men were the highest for the infectious symptom scale. In fact, analysis of the individual symptoms revealed that all four alcohol variables were significantly correlated with 4 of the 5 infectious symptoms: cold, flu, sore throat and cough. None of the alcohol consumption variables, however, was correlated with the occurrence of infections. It seems that the symptoms composing this scale are related to
318
disorders in the upper respiratory system. A similar finding was reported in a review of the adverse health effects of alcohol consumption, in which it was stated that severe alcohol use made individuals prone to respiratory tract infection [ 181. Kolb and Gunderson [9] reported that diseases of the respiratory system were by far the most common cause of hospital admissions among young (19-27 years old) Navy men during their first year of enlistment, and that they were more prevalent among alcohol abusers in comparison to non-alcoholic controls. Other individual symptoms that were associated with at least two of the alcohol variables among younger men were backaches, shaky or trembly feelings, and diarrhea. Among younger women, alcohol consumption was most strongly associated with cough and asthma followed by sore throat, difficulty breathing and backaches. In sum, only in the younger generation in this Mexican American population is alcohol consumption significantly related to physical symptoms. Although the middle generation had very similar drinking patterns by sex as the younger generation, they seem less affected by alcohol consumption. Alcohol consumption seems to affect in particular the upper respiratory tract. Possible mechanisms are neglect of nutrition, impairment of lung clearance, phagocytosis, and decreased immune responses [ 181. The upper respiratory symptoms may very well constitute the relatively short-term effects of alcohol consumption and, perhaps, people do develop a certain tolerance for such symptoms after continued exposure to alcohol. The present findings are confined to Mexican Americans, a population that contains a disproportionate number of heavy drinkers among men. Future research should attempt to replicate these findings with other groups and should do so using longtiudinal data in order to avoid the pitfalls of cross-sectional data. ACKNOWLEDGEMENTS
This research was supported by Grant ROl AGO417 from the National Institute on Aging and from a grant from the Hogg Foundation for Mental Health. We would like to thank an anonymous reviewer for useful comments and suggestions. REFERENCES 1 National Institute on Alcohol Abuse and Alcoholism, Fifth Special Report to the U.S. Congress on Alcohol and Health from the Secretary of Health and Human Services, Government Printing Office, Washington, DC, 1984. 2 D. Shurtleff, Some characteristics related to the incidence of cardiovascular disease and death, the Framingham Study, 16-year follow-up, The Framingham Study, Section 26, in: W.B. Kannel and T. Gordon (Eds.), Government Printing Office, 1970. 3 K. Yano, G.G. Rhoads and A. Kagan, New Engl. J. Med., Vol. 297,1977, pp. 405409.
319 4 R. Bell, K. Keeley and J. Buhl, in: Psychopathology and life events among alcohol users and non-users, in: F. Seixas (Ed.), Currents in Alcoholism, Vol. 3, Grune and Stratton, New York, 1977, pp. 103-123. 5 J.A. Neff and B.A. Husiani, J. Studies Alcohol, 34 (1982) 307. 6 J.A. Neff, J. Human Stress, 10 (1984) 35. 7 N.B. Belloc and L. Breslow, Prev. Med., 1 (1972) 409. 8 D. Kolb and E.K.E. Gunderson, Drug Alcohol Depend., 9 (1982) 181. 9 D. Kolb and E.K.E. Gunderson, J. Stud. Alcohol, 44 (1983) 84. 10 G.J. Warheit and J.B. Auth, The mental health and social correlates of alcohol use among differing life cycle groups, in: Nature and Extent of Alcohol Problems Among the Elderly, National Institute on Alcohol Abuse and Alcoholism, Research Monograph, 14, 1984. 11 R. Caetano, Drug Alcohol Depend., 12 (1983) 37. 12 KS. Markides and N. Krause, Alcohol consumption in three-generations of Mexican Americans: the influence of marital satisfaction, sex-role orientation and acculturation, in: B. Forster (Ed.), Alcohol and Drug Abuse in Special Populations, Wadsworth, Menlo Park, CA, in press. 13 R.L. Maril and A.N. Zavaleta, J. Studies Alcohol, 40 (1979) 480. 14 M.J. Paine, J. Studies Alcohol, 39 (1978) 894. 15 KS. Markides, S.K. Hoppe, H.W. Martin and D.M. Timbers, Marriage Fam., 45 (1983) 911. 16 R.J. Glynn, G.R. Bouchard, J.S. LoCastro and J.A. Hermos, Changes in Alcohol consumption behaviors among men in the normative aging study, in: Nature and Extent of Alcohol Problems Among the Elderly, National institute on Alcohol Abuse and Alcoholism, Research Monograph 14, 1984. 17 R. Caetano, Hispanic J. Behav. Sci., 6 (1984) 345. 18 M.J. Eckardt et al., J. Am. Med. Assoc., 246 (1981) 648.
Dependence, 16 (1986) Publishers Ireland Ltd.
ALCOHOL CONSUMPTION AMERICAN POPULATION
369-379
AND PHYSICAL
CARLOS F. MENDES DE LEON and KYRIAKOS Department of Preventive Medicine and Community Branch, Galveston, TX 77550 (U.S.A.) (Received (Received
369
SYMPTOMS IN A MEXICAN
S. MARKIDES Health,
University
of Texas Medical
June 24th, 1955) June 24th, 1985)
ABSTRACT
The negative health consequences of alcohol consumption are well established. The present study examines whether alcohol consumption is also associated with a series of self-reported physical symptoms. Analyses were performed on data from a three-generation study of Mexican Americans. Using four measures of alcohol consumption we found that alcohol consumption was related to physical symptoms only in the younger generation. While drinking levels in the older generation did not have sufficient variance to enable the assessment of their effects on symptoms, drinking levels in the middle generation were comparable to those in the younger generation. We hypothesized that the lack of association in the middle generation may reflect increased tolerance due to londer exposure to alcohol, to possible reduction of alcohol consumption by some middle-aged persons because of health problems, and to due possible removal from observation of certain middle-aged men whose heavy drinking led to health problems. Although alcohol consumption in the younger generation was not strongly associated with either psychosomatic or infectious symptoms, it was strongly related to upper respiratory symptoms among younger males. Among younger women, alcohol consumption was moderately related to physical symptoms with no symptom constellation having a particularly strong association with alcohol consumption. Key words: Mexican Americans - Alcohol Age differences - Physical symptoms
consumption
- Generations
-
INTRODUCTION
Extensive research has been done on the negative health consequences of alcohol. An overview of this can be found in The Fifth Special Report for 0376-8716/86/$03.50 o Elsevier Scientific Publishers Ireland Ltd. Printed and Published in Ireland
the U.S. Congress on Alcohol and Health, issued by the National Institute on Alcohol Abuse and Alcoholism [l] . From this report it can be learned that alcohol abuse is associated with increased mortality from such diseases as alcoholism, alcoholic psychosis, diseases of the digestive system, of the liver, and of the nervous system. Excessive alcohol consumption is also related with many adverse social consequences, such as traffic accidents, crime, homicide, suicide and family problems [l] . Moderate consumption, however, has been linked to less coronary heart disease [2,3], and is thought to serve as a buffer between stress and somatic depressive symptoms [ 4-61. Although there is a wealth of research on the effect of alcohol consumption on various diseases as well as on psychological symptoms, little is known about its effect on general health status. Belloc and Breslow [ 71 found that both men and women who averaged more than 5 drinks at a single occasion reported significantly poorer health status as measured by a composite index that included chronic conditions, disability, presence of physical symptoms and energy level. In a study of younger and older Navy men, it was found that problem drinkers had higher rates of morbidity as measured by number of hospital admissions and average length in a hospital, than non-alcoholic Navy controls. Older drinking Navy men, in particular, were more prone to accidents, violence and poisoning, mental disorders, diseases of the digestive system and infectious/parasitic diseases. They also had significantly more hospital admissions because of ‘symptoms and ill defined conditions’ [ 81. Among younger drinking Navy men a similar pattern was found. Infections of the respiratory system, accidents, mental disorders and gastrointestinal disorders were especially prevalent in this group. Younger drinking Navy men also had higher rates of hospital admissions because of ‘symptoms and ill-defined conditions’ [ 91 . Another study related physical health status to an index of alcohol use among three different age cohorts (18-29,30-49, and 50+ years) in the general population [lo]. The index of alcohol consumption divided the sample in a group at low risk for alcoholism and alcohol-related problems, and a high risk group. Physical health status was measured by one self-report item about present physical health. The high-risk groups in all three cohorts reported having poor or very bad health more often than the low risk groups, although this association reached statistical significance for only the youngest age group (18-29 years). The purpose of this study is to investigate the effect of alcohol consumption on general physical health. More specifically, do heavy alcohol consumers experience more physical symptoms than do light drinkers or non-drinkers? The three-generations study of Mexican Americans provided an opportunity to examine the relationship between alcohol consumption and physical symptoms at three different stages in the life cycle in an understudied population. Hispanics have been found to have a high percentage of abstainers (particularly among women) and a higher percentage of heavy
371
drinkers among men than the general population [ 11,121. The difference in alcohol consumption between men and women among Mexican Americans is wider than among other ethnic groups [13,14]. Therefore, analyses will be conducted separately for men and women. In general, we expect increasing levels of alcohol consumption to be associated with poorer general health status as measured by the number of self-reported physical symptoms. METHODS
The sample
The data employed in this study were drawn from interviews conducted in 1981 and 1982 with Mexican Americans in the San Antonio area. Area probability sampling involving selection of census tracts and blocks was employed to identify Mexican Americans aged 65-80 who had a child in the metropolitan area who, in turn, had an adult (18 years +) married or previously married child who also resided in the area [ 151. In the event that two or more such lineages were available, one was selected randomly. A total of 1125 persons were interviewed, 375 in each generation. About two-thirds of the sample was female, to a large extent due to the women’s greater longevity in the older generation and their tendency to marry and bear children at younger ages than the men. The mean age of the older generation was 74, for the middle 49, and for the younger generation 26. Measures
Alcohol consumption was assessed by two questions: ‘During the past 2 months, how often have you drunk wine, beer, or drinks containing alcohol? (coded l-6, from never to once a day or more); and, ‘when you drink wine, beer or drinks containing alcohol, about how many glasses do you usually have in one day?’ (the exact number was coded up to 7 or more glasses). These two measures were used separately in the analyses. A combination of them was also computed to form a frequency by volume measure by multiplying the two measures. In addition, a dummy variable was created to obtain a dichotomous variable of high levels of drinking: a value of one was assigned to persons consuming 6 or more drinks on a given day and zero to all others. Because of the small number of women in the high drinking category, analysis involving this variable was confined to men. The presence of physical symptoms was measured by a list of 20 commonly occurring complaints. Subjects were asked how often they had been bothered by any of the 20 symptoms in the past 2 or 3 months. Responses were coded on a 4-point scale: never, once in a while, often, most of the time (coded l-4). The list included the following symptoms: headaches, colds, asthma, difficulty in breathing, stomach problems, diarrhea, muscle ache, backache, shaky or trembly feeling, difficulty with eye sight not corrected with glasses, allergic reactions, chest pain, poor circulation, skin rash, fatigue, difficulty
372 hearing, flu, sore throat, infection and cough. The total symptom scale proved to have an alpha reliability-coefficient of 0.77. We performed factor-analysis using the maximum likelihood method with oblique rotation (Oblimin) of the factors, in order to examine the factor structure of the symptom-scale. The factor analysis yielded two interpretable factors. The first factor was labelled the psychosomatic symptom subscale, since it included symptoms usually associated with complaints of psychosomatic origin: headaches, stomach problems, muscle aches, backaches, shaky or trembly feelings, chest pains, poor circulation and fatigue. The second subscale consisted of symptoms that are usually associated with infectious ailments: colds, flu, sore throat, infection and cough. The alpha reliabilitycoefficients were 0.72 for the psychosomatic symptoms subscale and 0.76 for the infectious symptoms subscale. Four sociodemographic variable served as controls: age (measured in actual years of age), education (years of school completed), income (monthly household income, ranging from $0-199 to $2000 and over, and coded l---11), and marital status (married = 1, not married = 0). Analysis
We employed a series of multiple regression analyses that controlled for the sociodemographic variables in order to assess the independent association between the alcohol variables and the symptom scales. The analysis was conducted separately by generation and sex. Since number of drinks consumed and the frequency by volume measure were skewed for both men and women in all three generations, we employed a logarithmic transformation of these variables. RESULTS Table I presents data on frequency of alcohol consumption, the average number of drinks consumed on a given day by those who said they drank at least a few times a month, and percent of all subjects who consumed an average of 6 or more drinks on a given occasion, by generation and by gender. The table shows that in all generations, men report drinking considerably more than women, and that the younger two generations drink more than the older generation. Alcohol consumption levels look similar for older men, middle-aged women and younger women. Middle-aged men and younger men report drinking more frequently than the other groups, and their alcohol consumption levels appear to be about the same. We conducted zero-order correlation analysis between the alcohol consumption variables and the three symptom scales by generation and sex. No significant associations were found in the older generation. For the middleaged women, only 1 out of 12 correlations (between number of drinks and the total symptom-scale) reached statistical significance (r = 0.11, P < 0.05). For middle-aged men, also only one of the correlations between the alcohol
I
AND
VOLUME
12.4
39.0 27.6
13
41 29
33.3 29.4 5.9 20.6 5.9 2.9 2.9
100.0
11 10 2 7 2 1 1
34
Total
100.0
6.7 8.6
7 9
105
5.7
6
No. of glasses 1 2 3 4 5 6 7+
Total
Once a day or more Several times a week Once a week A few times a month Once a month or less frequently Never
Frequency
-
-
15
2
5 8
269
39 216
4
2 4
4
1.5
0.7 1.5
1.5
13.3
100.0
-
-
33.3 53.3
100.0
14.5 80.3
%
76
10 13 12 9 4 26 2
125
30 13
14
24 29
15
No.
No.
No.
%
Male
Female
24.0 10.4
11.2
19.2 23.2
100.0
13.2 17.1 15.8 11.8 5.3 34.2 2.6
100.0
AND SEX
65
l? 1; 13 6 6 6 2
125
84 94
39
7 17
4
No.
Female
2.9 6.9
1.6
100.0
20.0 29.2 20.0 9.2 9.2 9.2 3.1
100.0
34.3 38.4
15.9
%
87
7 16 18 8 6 24 8
143
38 10
31
32 28
4
No.
2.8
100.0
8.0 18.4 20.7 9.2 6.9 27.6 9.2
100.0
26.6 7.0
21.7
22.4 19.6
%
by persons
Male -
G3
of drinks at a single occasion
12.0
%
to the average number
BY GENERATION
Male
and volume
CONSUMPTION
G2
months
OF ALOCHOL
Gl
Frequency refers to previous.2 leaat ‘a few times a month’.
FREQUENCY
TABLE
93 79
37
3 19
60
12 16 17 6 2 5 2
231
-
No.
Female
40.3 34.2
16.0
1.3 8.2
100.0
20.0 26.7 28.3 10.0 3.3 8.3 3.3
100.0
-
96
who had drunk at
w
,”
314
variables and the symptoms was statistically significant: excessive drinking (6 drinks or more) at a single occasion was positively correlated to the infectious symptoms subscale (r = 0.22, P < 0.01). The correlation analysis proved more fruitful in the younger generation. All correlations for both sexes were positive, and almost all of them were statistically significant at the 0.05 level or beyond. Because of the lack of significant findings in the two older generations, we confined our regression analysis to younger men and women. The purpose of this analysis was to examine whether the significant correlations would persist after controlling for the effects of age, marital status, education and income. We also controlled for the effects of smoking (current smokers = 1; non-smokers = 0), because of its association with both alcohol consumption and physical symptoms. The correlations between smoking and the several measures of alcohol consumption were all highly significant (P < O.OOl), for both younger males and females, ranging from 0.20 to 0.43. Table II presents the results of the multiple regression analysis for the younger men. The results show that the significant correlations between alochol consumption and physical symptoms in general persisted after controlling for age, marital status, education, income and smoking. In particular the associations between the entire symptom scale and the psychosomatic symptoms scale and the several measures of alcohol consumption did not change very much after the controls. For the infectious symptoms scale, however, smoking entered the regression equation at a significant level for two of the four alcohol consumption variables, and approached statistical significance as a predictor variable in the equations for the remaining two variables. Table III displays the results of the regression analysis for the younger generation women. Again, the relationships between the physical symptoms scales and alcohol consumption were very similar to the results of the zeroorder correlations. Most of the correlations remained significant after the controls. We also examined the possiblility of a non-linear relationship between volume of alcohol consumption and physical symptoms. Such a non-linear relationship has been suggested by the hypothesis of the stress-buffering role of alcohol [ 4-61. We tested this possibility for 2nd and 3rd generation males separately by entering the square of the ‘number of drinks’ variable as a second step into the regression equation, after controlling for the linear effect of ‘number of drinks’ and the effects of the control variables. Only in the case of the psychosomatic symptoms subscale in younger men did we observe a significant non-linear (U-shaped) effect of number of drinks. However, the non-linear effect was considerably lower than the linear effect of number of drinks. It is interesting to note that Neff [6] found a similar relationship between alcohol consumption and a somatic symptoms scale, a scale very similar to our psychosomatic symptoms subscale. In Neff’s study, abstainers reported the most somatic symptoms, followed by the heavy drinkers and then the moderate drinkers.
0.235** 0.246** 0.037
0.26****
0.27****
0.08 < 0.001.
(0.307)
(1.459)
(2.375)
(0.557)
< 0.01; ***up
0.209**
beta
0.25****
r
Physicala symptoms
a0.05 < P < 0.10; **p < 0.05; ***p
Frequency of drinking Number of drinks Frequency x No. High drinking
Alcohol variables
MALES (/V = 142)
0.02
0.19**
0.17**
0.21***
r
-0.003
0.196*
0.174*
0.211x*
beta
Psychosomatic symptoms
(0.010)
(0.576)
(0.871)
(0.279)
0.247**
0.32****
0.082
0.243**
0.32****
0.16**
0.182**
beta
0.26****
r
Infectious symptoms
(0.328)
(0.700)
(1.175)
(0.231)
The regression analyses were performed with age, marital status,
FOR THIRD GENERATION
Numbers in parentheses are unstandardized regression coefficients. education and current smoking held constant.
RESULTS OF MULTIPLE REGRESSION ANALYSIS
TABLE II
0.198***
0.20****
(1.850)
(3.179)
< 0.01; ****p < 0.001
0.198***
(a) 0.175 (b) (0.747)
beta
0.20****
0.18***
r
Physica? symptoms
**0.05 < P < 0.10; **p < 0.05; ***p
Frequency of drinking Number of drinks Frequency x No.
Alcohol variables
FEMALES (N = 233)
0.16***
0.17***
0.14**
r
0.154**
0.161**
0.133*
beta
Psychosomatic symptoms
(0.762)
(1.371)
(0.302)
0.195*** 0.183*** 0.191***
0.17*** 0.18***
beta 0.17***
r
Infectious symptoms
(0.794)
(1.311)
(0.372)
The regression analyses were performed with age, marital status,
FOR THIRD GENERATION
Numbers in parentheses are unstandardized regression coefficients. education and current smoking held constant.
RESULTS OF MULTIPLE REGRESSION ANALYSIS
TABLE III
DISCUSSION
The results indicate that alcohol consumption was not related to general health status as measured by self-reported physical symptoms among older and middle-aged men and women in this Mexican American population. Since the older generation, in particular the older women, consume very low quantities of alcoholic beverages, a negative effect of alcohol consumption on health status could not be expected. As noted previously, middle-aged and younger men, and middle-aged and younger women in this population have approximately the same alcohol consumption levels. However, only the younger generation seems to experience adverse physical consequences associated with alcohol consumption. One explanation for this finding could be that middle-aged people have acquired a greater degree of tolerance for alcohol than younger people. Thus, the same amount of alcohol intake is less likely to produce symptomatic effects on middle-aged than on younger people. This cannot be inferred with certainty from cross-sectional data such as ours and we do not have information on previous drinking patterns. However, a longitudinal study by Glynn and colleagues [15] has found that drinking patterns among males remained relatively constant between 1973 and 1982. Furthermore, Caetano [ 171 reported that rates of heavy drinking and alcohol-related problems do not decrease significantly from the 20s to the 30s among Hispanics. It appears reasonable to assume that the majority of the middle-aged subjects in this study, in particular the men, have been consuming alcoholic beverages since their young adulthood. The presence of physical dependence on alcohol, particularly among the heavier drinkers, might also have influenced the relationship between alcohol consumption and physical symptoms. It is quite possible that a number of middle-aged persons drink in order to suppress physical symptoms. It is also possible that some middle-aged persons might have reduced their alcohol intake because of health problems, which would tend to lower the association between alcohol consumption and symptoms. This would probably be less likely among younger people, since they are less likely to stop drinking because of health problems. Finally, it must be acknowledged that generational differences in the association between alcohol consumption and physical symptoms might reflect the operation of differential selection. It is possible that certain middle-aged and older people whose heavy drinking led to poor health (symptoms), may have been removed from observation by death, institutionalization (hospital, prison, etc.), or some other factor. As noted above, the zero-order correlations between the drinking variables and the symptoms scales among younger men were the highest for the infectious symptom scale. In fact, analysis of the individual symptoms revealed that all four alcohol variables were significantly correlated with 4 of the 5 infectious symptoms: cold, flu, sore throat and cough. None of the alcohol consumption variables, however, was correlated with the occurrence of infections. It seems that the symptoms composing this scale are related to
318
disorders in the upper respiratory system. A similar finding was reported in a review of the adverse health effects of alcohol consumption, in which it was stated that severe alcohol use made individuals prone to respiratory tract infection [ 181. Kolb and Gunderson [9] reported that diseases of the respiratory system were by far the most common cause of hospital admissions among young (19-27 years old) Navy men during their first year of enlistment, and that they were more prevalent among alcohol abusers in comparison to non-alcoholic controls. Other individual symptoms that were associated with at least two of the alcohol variables among younger men were backaches, shaky or trembly feelings, and diarrhea. Among younger women, alcohol consumption was most strongly associated with cough and asthma followed by sore throat, difficulty breathing and backaches. In sum, only in the younger generation in this Mexican American population is alcohol consumption significantly related to physical symptoms. Although the middle generation had very similar drinking patterns by sex as the younger generation, they seem less affected by alcohol consumption. Alcohol consumption seems to affect in particular the upper respiratory tract. Possible mechanisms are neglect of nutrition, impairment of lung clearance, phagocytosis, and decreased immune responses [ 181. The upper respiratory symptoms may very well constitute the relatively short-term effects of alcohol consumption and, perhaps, people do develop a certain tolerance for such symptoms after continued exposure to alcohol. The present findings are confined to Mexican Americans, a population that contains a disproportionate number of heavy drinkers among men. Future research should attempt to replicate these findings with other groups and should do so using longtiudinal data in order to avoid the pitfalls of cross-sectional data. ACKNOWLEDGEMENTS
This research was supported by Grant ROl AGO417 from the National Institute on Aging and from a grant from the Hogg Foundation for Mental Health. We would like to thank an anonymous reviewer for useful comments and suggestions. REFERENCES 1 National Institute on Alcohol Abuse and Alcoholism, Fifth Special Report to the U.S. Congress on Alcohol and Health from the Secretary of Health and Human Services, Government Printing Office, Washington, DC, 1984. 2 D. Shurtleff, Some characteristics related to the incidence of cardiovascular disease and death, the Framingham Study, 16-year follow-up, The Framingham Study, Section 26, in: W.B. Kannel and T. Gordon (Eds.), Government Printing Office, 1970. 3 K. Yano, G.G. Rhoads and A. Kagan, New Engl. J. Med., Vol. 297,1977, pp. 405409.
319 4 R. Bell, K. Keeley and J. Buhl, in: Psychopathology and life events among alcohol users and non-users, in: F. Seixas (Ed.), Currents in Alcoholism, Vol. 3, Grune and Stratton, New York, 1977, pp. 103-123. 5 J.A. Neff and B.A. Husiani, J. Studies Alcohol, 34 (1982) 307. 6 J.A. Neff, J. Human Stress, 10 (1984) 35. 7 N.B. Belloc and L. Breslow, Prev. Med., 1 (1972) 409. 8 D. Kolb and E.K.E. Gunderson, Drug Alcohol Depend., 9 (1982) 181. 9 D. Kolb and E.K.E. Gunderson, J. Stud. Alcohol, 44 (1983) 84. 10 G.J. Warheit and J.B. Auth, The mental health and social correlates of alcohol use among differing life cycle groups, in: Nature and Extent of Alcohol Problems Among the Elderly, National Institute on Alcohol Abuse and Alcoholism, Research Monograph, 14, 1984. 11 R. Caetano, Drug Alcohol Depend., 12 (1983) 37. 12 KS. Markides and N. Krause, Alcohol consumption in three-generations of Mexican Americans: the influence of marital satisfaction, sex-role orientation and acculturation, in: B. Forster (Ed.), Alcohol and Drug Abuse in Special Populations, Wadsworth, Menlo Park, CA, in press. 13 R.L. Maril and A.N. Zavaleta, J. Studies Alcohol, 40 (1979) 480. 14 M.J. Paine, J. Studies Alcohol, 39 (1978) 894. 15 KS. Markides, S.K. Hoppe, H.W. Martin and D.M. Timbers, Marriage Fam., 45 (1983) 911. 16 R.J. Glynn, G.R. Bouchard, J.S. LoCastro and J.A. Hermos, Changes in Alcohol consumption behaviors among men in the normative aging study, in: Nature and Extent of Alcohol Problems Among the Elderly, National institute on Alcohol Abuse and Alcoholism, Research Monograph 14, 1984. 17 R. Caetano, Hispanic J. Behav. Sci., 6 (1984) 345. 18 M.J. Eckardt et al., J. Am. Med. Assoc., 246 (1981) 648.