Correlates of and relationship among lifetime alcohol abuse and alcohol dependence in older community residents in Brazil.

Misinterpretation in major surveys of alcohol use disorder as described by DSM–IV (Hasin et al, 2007) has raised serious questions regarding the extent of alcohol use disorder, and the relationship between alcohol abuse and alcohol dependence. While the adverse social, physical and mental effects of alcohol misuse are well known (Council on Scientific Affairs, 1996), there is little information on the determinants of alcohol abuse (societal impact) and alcohol dependence (physiological impact). We therefore examined their separate and combined associations with demographic, social and health characteristics in a representative community-resident sample aged 60 years and over. We hypothesised that, while for each of the three groups (those with alcohol abuse, those with alcohol dependence, and those with both) there would be associated demographic characteristics, abuse would be more closely associated with social characteristics, dependence with health characteristics, and the combined presence of abuse and dependence with both social and health characteristics.

Methods

Data were gathered by carefully trained and monitored interviewers in 1995 using face-to-face structured household surveys of 7920 representative community residents aged 60 years and over, in nine regions covering the southernmost Brazilian state of Rio Grande do Sul (a wine-producing area). Information from one region was problematic and was dropped, resulting in a sample of 7040 persons, of whom 79 (1.1%) declined to participate, yielding an analysis sample of 6961 (Conselho Estadual do Idoso, 1997). The ethics commit-. The ethics committee of the Federal University of São Paulo approved the study.

Dependent variable: evaluation of alcohol use

Use of alcohol was evaluated according to participants’ yes/no responses to each of the following questions (asked in Portuguese):

Has a family member, friend, physician, priest ever commented or suggested that you were drinking too much?

Have you ever tried to stop drinking but been unable to do so?

Have you ever had trouble at work or school because of alcohol, such as drinking or missing work?

Have you ever been involved in fights or arrested for being drunk?

Has it ever seemed to you that you were drinking too much?

While not constituting an established measure, the individual items are comparable to those commonly used in similar surveys. A positive response to question 1, 3 or 4 was accepted as indicating lifetime abuse of alcohol. A positive response to question 2 or 5 indicated lifetime dependence (Hasin et al, 2007). Participants were not asked when the problem occurred. The sample was classified into four mutually exclusive groups: abuse only, dependence only, both abuse and dependence, neither abuse nor dependence.

Independent variables

The independent variables are listed in Table 1.

Table 1

Basic descriptive characteristics of the sample by report of alcohol abuse only, dependence only, and abuse and dependence: numbers (%) of participants

	Total sample (n = 6961)		Abuse only (n = 103)		Dependence only (n = 244)		Abuse and dependence (n = 387)
Demographic characteristics
Gender
Male	2368	(34.0)	74	(71.8)***	193	(79.1)***	332	(85.8)***
Female	4593	(66.0)	29	(28.2)	51	(20.9)	55	(14.2)
Age (years)
60–64	1866	(26.8)	32	(31.1)	84	(34.4)***	147	(38.0)***
65–69	2085	(30.0)	23	(22.3)	82	(33.6)	111	(28.7)
70–74	1067	(15.3)	19	(18.4)	38	(15.6)	55	(14.2)
75–79	1216	(17.5)	23	(22.3)	24	(9.8)	51	(13.2)
80+	727	(10.4)	6	(5.8)	16	(6.6)	23	(5.9)
Education
<4 years	4594	(66.0)	73	(70.9)	162	(66.4)	293	(75.7)***
>4 years	2344	(34.0)	30	(29.1)	81	(33.2)	93	(24.0)
Income
Low income (<US$200)	4323	(62.1)	61	(59.2)	147	(60.2)	238	(61.5)
Higher income (>US$200)	2414	(34.7)	40	(40.8)	93	(38.1)	134	(34.6)
Race
White	5862	(84.2)	84	(81.6)	194	(79.5)**	275	(71.1)***
Afro-Brazilian	473	(6.8)	8	(7.8)	14	(5.7)	58	(15.0)
Other	625	(9.0)	11	(10.7)	36	(14.8)	54	(14.0)
Religion
Catholic	5245	(75.3)	83	(80.6)	184	(75.4)	290	(74.9)
Evangelical	1077	(15.5)	10	(9.7)	38	(15.6)	51	(13.2)
Other	609	(8.7)	10	(9.7)	21	(8.6)	45	(11.6)
Place of birth
Urban	2363	(33.9)	31	(30.1)	76	(31.1)	121	(31.3)
Rural	4529	(65.1)	69	(67.0)	167	(68.4)	264	(68.2)
Social characteristics
Physical activity
No	4316 (62.0)	61	(59.2)	149	(61.1)	247	(63.8)
Yes	2608	(37.5)	42	(40.8)	95	(38.9)	140	(36.2)
Use tobacco
Yes	1302	(18.7)	37	(35.9)***	102	(41.8)***	182	(47.0)***
No	5632	(80.9)	66	(64.1)	142	(58.2)	205	(53.0)
Marital status
Married	3161	(45.4)	63	(61.2)***	138	(56.6)***	219	(56.6)***
Never married	471	(6.8)	10	(9.7)	20	(8.2)	38	(9.8)
No longer married	3328	(47.8)	30	(29.1)	86	(35.2)	130	(33.6)
Children
Yes	6492	(93.3)	101	(98.1)	224	(91.8)	364	(94.1)
No	445	(6.4)	2	(1.9)	20	(8.2)	21	(5.4)
Living arrangements
Live with someone	5893	(84.7)	93	(90.3)	207	(81.9)	328	(84.8)
Live alone	1056	(15.2)	10	(9.7)	37	(18.1)	59	(15.2)
Employed
No	5992	(86.1)	92	(89.3)	196	(80.3)**	290	(74.9)***
Yes	940	(13.5)	11	(10.7)	46	(19.7)	96	(24.8)
Participate in social activities
No	4221	(60.6)	70	(68.0)	156	(63.9)	260	(67.2)***
Yes	2736	(39.3)	33	(32.0)	88	(36.1)	127	(32.8)
Participate in religion-affiliated activities
No	1977	(28.4)	40	(38.8)*	80	(32.8)	161	(41.6)***
Yes	4964	(71.3)	63	(61.2)	164	(67.2)	225	(58.1)
Health characteristics
Activities of daily living problems
0	4238	(60.9)	62	(60.2)	151	(61.9)	234	(60.5)**
1 or 2	2195	(31.5)	30	(29.1)	79	(32.4)	108	(27.9)
3 or more	526	(7.6)	11	(10.7)	14	(5.7)	45	(11.6)
Vascular conditions
Yes	4390	(63.1)	60	(58.3)	124	(50.8)***	190	(49.1)***
No	2542	(36.5)	43	(41.7)	120	(49.2)	197	(50.9)
Respiratory conditions
Yes	2059	(29.6)	31	(30.1)	95	(38.9)***	166	(42.9)***
No	4902	(70.4)	72	(69.9)	149	(61.1)	221	(57.1)
Kidney problems
Yes	897	(12.9)	16	(15.5)	27	(11.1)	77	(19.9)***
No	6066	(87.1)	87	(84.5)	217	(88.9)	310	(80.1)
Osteoporosis
Yes	1047	(15.0)	6	(5.8)**	25	(10.2)*	42	(10.9)**
No	5916	(85.0)	97	(94.2)	219	(89.8)	345	(89.1)
Psychiatric problem
Yes	2722	(39.1)	42	(40.8)	110	(45.1)	192	(49.6)***
No	4241	(60.9)	61	(59.2)	134	(54.9)	195	(50.4)

Values may not total to n and percentages may not sum to 100 because of missing data.

Chi square test:

P < 0.05;

P < 0.01;

P < 0.001 (‘abuse only’, ‘dependence only’, ‘abuse and dependence’ each compared with ‘no abuse or dependence’).

Physical activity (i.e. exercise) was assessed by asking: ‘In the last 3 months have you practised regular physical activity?’ Responses were recorded as ‘yes’ (once a week or more) or ‘no’. Employment status was recorded as ‘employed’ if the participant was still working (the type of work was immaterial) or ‘not employed’ if not working or did not know the answer. Problems with activities of daily living (ADL) were assessed using a five-item unidimensional scale. The number of impaired activities was recoded as 0, 1 or 2, or 3 or more.

Preliminary analyses of 18 self-reported physical health conditions indicated that only vascular conditions, respiratory problems, kidney problems and osteoporosis were relevant. Vascular conditions include any mention of heart disease, hypertension, diabetes, stroke or varicosities. Respiratory problems include any mention of bronchitis or pneumonia.

The presence of a psychiatric condition was assessed by a validated Brazilian modification of the Short Psychiatric Evaluation Schedule (Blay et al, 1988).

Statistical analysis

Percentages were used to describe the sample, and χ2 to compare each of the three alcohol use groups with the group recording neither lifetime abuse nor lifetime dependence. Because of the small size of the abuse-only group, separate blockwise logistic regression analyses were first run to identify the significant variables within each block (demographic, social, health characteristics – see Table 1). These significant variables were then entered into an initial multivariable polytomous logistic regression, and a final model was run using only the variables found to be significant. Analyses were performed using SPSS 13.0.

Results

Lifetime alcohol abuse was recorded for 734 participants (10.6%), of whom 103 (1.5%) reported abuse only, 244 (3.5%) dependence only, and 387 (5.6%) both abuse and dependence. Two-thirds of the sample were female, the majority were aged 60–69, of low education and low income, rural birth, White (84%) and Catholic (75%) (Table 1). In univariate analyses (Table 1) the ‘abuse only’ group differed from those with neither abuse nor dependence on 5 of the 21 characteristics examined (male, use tobacco, married, less likely to participate in religion-affiliated activities, and less likely to have osteoporosis). In addition to all these characteristics except religion-affiliated activities, ‘dependence only’ participants were more likely to be younger, of ‘other’ race/ethnicity, employed and to have a respiratory condition, but less likely to report a vascular condition. Participants reporting abuse and dependence additionally had little education, were less likely to participate in social activities, and were more likely to have ADL problems, kidney problems and psychiatric problems.

The final controlled analysis (Table 2) yielded a more restricted set of significant variables, but showed a similar increase in number and type of associations, going from ‘abuse only’ to ‘dependence only’ to ‘abuse and dependence’. The significant associates of ‘abuse only’ were male gender and tobacco use. ‘Dependence only’ was additionally associated with ‘other’ race/ethnicity (as compared with White), increased likelihood of respiratory and psychiatric problems, and decreased odds of vascular conditions. The same associations held for ‘abuse and dependence’, but with more marked odds ratios.

Table 2

Polytomous logistic regression comparing alcohol abuse only, alcohol dependence only, and both abuse and dependence with no reported alcohol abuse

	Abuse only		Dependence only		Abuse and dependence
	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P
Demographic characteristics
Gender
Male	6.32 (4.04, 9.18)	0.001	9.08 (6.57, 12.54)	0.001	15.44 (11.40, 20.89)	0.001
Race
White	0.86 (0.45, 1.64)	0.644	0.64 (0.44, 0.95)	0.025	0.63 (0.45, 0.89)	0.007
Afro-Brazilian	1.25 (0.49, 3.17)	0.639	0.73 (0.38, 1.40)	0.340	2.16 (1.38, 3.36)	0.001
Other	Reference		Reference		Reference
Social characteristics
Use tobacco	2.21 (1.45, 3.37)	0.001	2.44 (1.85, 3.22)	0.001	2.77 (2.20, 3.49)	0.001
Health characteristics
Vascular condition present	0.99 (0.65, 1.50)	0.944	0.71 (0.54, 0.94)	0.017	0.65 (0.51, 0.82)	0.001
Respiratory condition present	0.94 (0.61, 1.47)	0.797	1.35 (1.02, 1.78)	0.038	1.51 (1.20, 1.91)	0.001
Psychiatric problem present	1.50 (0.98, 2.30)	0.063	1.91 (1.43, 2.54)	0.001	2.36 (1.85, 3.00)	0.001

The variables considered are the demographic, social and health conditions listed in Table 1. The significant variables identified in a blockwise analysis were entered into a penultimate polytomous multivariable logistic regression model, from which the resulting significant variables were then selected for the current, final model.

Discussion

The data come from a large community-resident sample aged 60 years and over, representative of the state of Rio Grande do Sul in Brazil, who provided information on multiple aspects of ageing. The five-item questionnaire permitted a rough assessment of lifetime alcohol abuse (three items) and alcohol dependence (two items). According to the responses to these items, 10.6% of the sample (men, 25.4%; women, 2.9%) reported alcohol-related problems, with 1.5% reporting abuse only, 3.5% dependence only, and 5.6% abuse and dependence.

Comparison with other studies in Brazil is difficult. Focus on lifetime use is infrequent; the samples rarely include participants aged 60 and over, are often small, resulting in questionable findings, and have used different measures to assess alcohol use. Our finding of a 10.6% prevalence rate compares with reports ranging from 2.7% in Campinas, south-eastern Brazil (based on 93 participants aged 60 and over) (Barros et al, 2007) to 12% for frequent or heavy drinkers in a sub-sample aged 60 years and over in a national survey (Laranjeira et al, 2007).

The demographic and health associations of alcohol misuse we found are comparable to those reported elsewhere, which implicate male gender, younger age, tobacco use, adverse physical health conditions except for vascular status, and adverse psychiatric status (National Institute on Alcohol Abuse and Alcoholism, 2002), providing confidence that the five alcohol questions have both content and criterion validity.

We hypothesised that ‘abuse only’ would be associated with demographic and social variables, and ‘dependence only’ with health-related characteristics. In controlled analyses, our hypothesis regarding ‘abuse only’ held, but characteristics associated with dependence included both health conditions (as hypothesised) and characteristics encompassed by ‘abuse only’. The same ‘dependence only’ characteristics held for ‘abuse and dependence’, but the associations were stronger.

Our findings also address two key questions: whether ‘abuse only’, ‘dependence only’ and ‘abuse and dependence’ are hierarchically associated; and whether they represent unique, non-progressive, manifestations of alcohol misuse. We argue for hierarchical association based on the finding that, in these older persons, dependence is uniquely associated with health effects in addition to the effects associated with alcohol abuse, and that these associations are intensified among those reporting abuse and dependence. We argue for the possibility of non-progression by noting that nearly half of those reporting lifetime alcohol misuse report only lifetime abuse, or only lifetime dependence. Since lifetime alcohol abuse has consistently been reported to decline with age (a finding further confirmed even in this older sample), we assume (but cannot confirm) that alcohol abuse occurred at an earlier time and may not necessarily progress. In support, Hasin et al (1990) found that, over 4 years, only 30% progressed from alcohol abuse only to alcohol dependence, while 39% with alcohol dependence had remitted.

Our data have significant limitations. Our measure does not meet diagnostic criteria, so our findings must be interpreted cautiously. Information is self-reported; however, self-report has been found to be valid for alcohol use and problems (Bongers & Van Oers, 1998) and for various health conditions (Beckett et al, 2000). We have no information on when problems with alcohol use occurred and whether they were still present. This cross-sectional design cannot distinguish between cause and effect.

Nevertheless, these data suggest that, while there is a gradient of associations and of adverse effects going from ‘abuse’ to ‘dependence’ to ‘abuse and dependence’, progression to a more serious stage need not necessarily occur.