OUP user menu

STRESSORS AND ALCOHOL CONSUMPTION

BEGOÑA SAN JOSÉ, HANS A. M. VAN OERS, H. DIKE VAN DE MHEEN, HENK F. L. GARRETSEN, JOHAN P. MACKENBACH
DOI: http://dx.doi.org/10.1093/alcalc/35.3.307 307-312 First published online: 1 May 2000

Abstract

The objective of this study was to examine the relationship between negative life events and chronic stressors and drinking behaviour. Data suggested that some life events (getting divorced) and some chronic stressors (financial difficulties, unfavourable marital status, and unfavourable employment status) were positively related to abstinence among men and women. Furthermore, some life events (being a victim of a crime, decrease in financial position, divorce or reporting two or more life events) were positively associated with heavy drinking among men. Chronic stressors, such as unfavourable marital status and unfavourable employment status, were also related to heavy drinking among both men and women. Results presented here suggest that people under stressful conditions are more likely to either abstain or drink heavily rather than to drink lightly or moderately.

INTRODUCTION

Alcohol has been hypothesized to buffer and to serve as a coping mechanism against stress (Cappell and Greeley, 1987). It is commonly accepted that people drink alcohol in order to cope with the effects of stress (Allan and Cooke, 1985; Krause, 1991). Stressful life events and chronic stressors have been correlated with alcoholism (Linsky et al., 1985), alcohol abuse (Cole et al., 1990) heavy drinking (Wilsnack et al., 1991), and alcohol dependence and problems (Johnson and Pandina, 1993; Welte and Mirand, 1995) in different populations. However, while some studies have reported positive associations, others have found negative associations (e.g. Romelsjo et al., 1991; Temple et al., 1991). With regard to the relationship between stress and abstinence, some authors have reported that abstainers have more adverse life contexts (Mertens et al., 1996), but the opposite has also been reported (Neff and Husaini, 1982; Cole et al., 1990). After years of research, the relationship between stress and alcohol consumption remains inconclusive (Temple et al., 1991; Pohorecky, 1991; Pierce et al., 1994).

Although some researchers have found stronger relationships between stressful conditions and alcohol consumption among men than among women (e.g. Romelsjo et al., 1991) the opposite has also been reported (e.g. Allan and Cooke, 1985). In the present study, we have examined the relationship between stress and alcohol consumption separately for men and for women, and using different cut-off points for heavy drinking for each of the gender groups. Also, in order to assess the relationship between stressors and alcohol consumption, several types of events as well as the total number of events reported (Neff and Husaini, 1982) have been considered. Negative life events (and a sum of those events) and chronic stressors were assessed simultaneously in relation to alcohol consumption. We considered stressful situations as possible predictors of both abstinence and heavy drinking.

METHODS

This study was conducted within the framework of a large population study, the GLOBE study (Eindhoven, The Netherlands). For this study, a random sample of 27 14;000 non-institutionalized Dutch citizens in the age range 15–74 years was drawn from a region in the south of Holland (Eindhoven and surroundings). Those over 45 years of age, as well as those in the highest and lowest socio-economic strata (identified by postal code) were over-sampled. A questionnaire was mailed to these individuals and 18 14;973 responded. Personal interviews were conducted with some of the respondents to the postal questionnaire (n 14;= 14;3750) in order to collect more detailed data on health as well as on personality characteristics. The response rate was approximately 80% (n 14;= 14;2802). Differences in response rates between socio-demographic groups were small: women, elderly people, the wealthier, and the country dwellers responded more frequently than their counterparts, although the differences were not statistically significant (Mackenbach et al., 1994).

Respondents were asked about their experience of the following life events in the preceding 6 months: moving; important decrease in financial position; being a victim of a robbery, assault or rape; whether he/she had lost his/her own job; whether a partner or a family member had lost his/her job; whether a partner or another family member was seriously ill; whether their partner had died; whether a close relative or a close friend had died; or whether they had divorced or had broken up with their partner. Respondents answered yes or no to each of these questions. A variable was constructed to determine whether an accumulation of events was associated with drinking or drinking heavily. This variable had the following categories: none of the events listed; positive answer given to one of the events; positive answers given to two or more events.

Reporting having financial difficulties (difficulties affording food, housing, electricity, etc.), unfavourable marital status (being single, separated or divorced, and widowed), unfavourable employment status (being unemployed, receiving a disability pension, or having retired early or being retired), reporting housing problems (draught, cold, and damp), and neighbourhood problems (unpleasant smells, noise from neighbours, noise from traffic, and criminality) were considered as chronic stressors.

Alcohol consumption was measured as average alcohol intake (U/week). Questions about the amount (no. of glasses) of alcohol consumed/occasion as well as the frequency of consumption (drinking days/week) were combined to arrive at a measure of average alcohol intake. Abstainers (0 14;U/week) were compared to light or moderate drinkers (drinking ≤21 14;U/week for men; drinking ≤14 14;U/week for women); heavy drinkers (drinking above these limits) were also compared to light or moderate drinkers (reference category). These cut-off points (>21 14;U/week for men and >14 14;U/week for women) were chosen, because drinking below these limits has been repeatedly associated with health benefits (Jackson and Beaglehole, 1995). Furthermore, an even higher cut-off point was chosen (>29 14;U/ week for men and >21 14;U/week for women) in order to check for possible associations at higher levels of consumption.

Several variables that are known to be associated with both alcohol consumption and stress were included in the logistic regression models. These potential confounders were age, educational level, and level of urbanization.

Several logistic regression models were constructed to assess the relationship between stress and alcohol consumption. The outcome variable for all the models constructed was alcohol consumption. This variable had been dichotomized into abstainers vs light or moderate drinkers (reference category) and heavy drinkers vs light or moderate drinkers (reference category). These models also contained the confounders mentioned above. Analyses were done multivariately.

Some of the acute events have their chronic counterparts; these include losing a job in the previous 6 months and being unemployed; getting divorced in the previous 6 months and being divorced; death of a partner in the previous 6 months and being widowed. In the models looking at the acute event, we did not adjust for the corresponding chronic stressor (or vice versa). For example, in a model looking at the effect of having lost a job in the previous 6 months, employment status was excluded. Although decrease in financial position and having financial difficulties could be thought of as the acute and the chronic versions of the same thing, they were measured in very different ways, since having financial difficulties refers to having difficulties affording to pay for basic needs. A decrease in financial position may not lead to financial difficulties. The model looking at a decrease in financial position was adjusted for financial difficulties (or vice versa).

RESULTS

The associations of life events or chronic stressors with drinking or abstinence levels for men and women are shown in Table 1. Getting divorced was positively associated with abstinence among men and negatively among women. Among women, reporting having moved was positively related to abstinence from alcohol. A large number of chronic stressors proved to be statistically significantly related to abstinence among men and women. Financial difficulties, unfavourable marital status [being single (both men and women) and being separated or divorced, only among men] and unfavourable employment status (receiving a disability pension, or having retired early or being retired) were positively related to abstinence in men and women.

View this table:
Table 1.

Adjusted odds ratios (OR*) and 95% confidence intervals (CI) for the relationship between stress and abstinence

MalesFemales
StressOR*95% CIOR*95% CI
Males: >21 U/week vs ≤21 U/week (reference). Females: >14 U/week vs ≤14 U/week (reference).
*Adjusted for age, education and level of urbanization.
(A) In the model with divorce and death of a partner, marital status was excluded (or vice versa).
(B) In the model with lost job, employment status was excluded (or vice versa).
Life events
  Moved0.81(0.63–1.04)1.25(1.04–1.51)
  Victim of crime/assault0.82(0.60–1.10)0.86(0.68–1.09)
  Partner lost job0.91(0.60–1.38)1.15(0.91–1.47)
  Sick partner/family member0.99(0.80–1.21)1.01(0.87–1.17)
  Death of close relative/friend0.98(0.82–1.17)0.92(0.81–1.05)
  Worsened financial position1.03(0.82–1.29)0.94(0.78–1.13)
  Divorce/break-up (A)1.54(1.06–2.23)0.66(0.48–0.92)
  Death of partner (A)0.73(0.31–1.70)1.44(0.91–2.28)
  Lost job (B)1.01(0.76–1.35)0.95(0.72–1.26)
  Count of life events
    1 life event0.93(0.79–1.08)0.98(0.87–1.10)
    ≥2 life events0.91(0.75–1.09)0.98(0.85–1.12)
Chronic strains
  Financial difficulties1.28(1.08–1.52)1.30(1.14–1.48)
  Marital status (A)
    Single1.73(1.37–2.19)1.25(1.04–1.51)
    Separated/divorced1.52(1.13–2.03)1.00(0.81–1.25)
    Widowed1.25(0.83–1.87)1.18(0.96–1.43)
  Employment status (B)
    Unemployed1.31(0.91–1.89)0.97(0.66–1.41)
    Disability pension2.50(1.96–3.19)1.97(1.54–2.53)
    Retired early or retired1.93(1.50–2.49)1.91(1.49–2.45)
  Housing problems1.15(0.98–1.35)1.10(0.98–1.24)
  Neighbourhood problems0.99(0.85–1.14)0.94(0.84–1.05)

As shown in Table 2, respondents reporting life events and chronic stressors were also compared with regard to their likelihood of heavy drinking (>21 14;U/week for males; >14 14;U/week for females). Being a victim of a crime, getting divorced or breaking up, worsening of financial position, and reporting one or more life events were positively related to heavy drinking among men. Among men, being single, separated or divorced, and being unemployed were the chronic stressors positively related to heavy drinking. Among women, reporting the death of a close friend or relative was inversely related to heavy drinking. Being separated or divorced was positively related to heavy drinking, whereas having retired early or being retired was inversely related to heavy drinking.

View this table:
Table 2

Adjusted odds ratios (OR*) and 95% confidence intervals (CI) for the relationship between stress and heavy drinking

MalesFemales
StressOR*95% CIOR*95% CI
Males: >21 U/week vs ≤21U/week (reference). Females: >14 U/week vs ≤14 U/week (reference).
*Adjusted for age, education, and level of urbanization.
(A) In the model with divorce and death of a partner, marital status was excluded (or vice versa).
(B) In the model with lost job, employment status was excluded (or vice versa).
Life events
  Moved0.85(0.67–1.08)1.14(0.83–1.58)
  Victim of crime/assault1.46(1.15–1.87)1.15(0.81–1.63)
  Partner lost job0.83(0.55–1.25)1.24(0.81–1.89)
  Sick partner/family member0.94(0.77–1.16)1.23(0.97–1.57)
  Death of close relative/friend1.10(0.92–1.31)0.78(0.62–0.99)
  Decrease financial position1.30(1.04–1.64)0.96(0.69–1.33)
  Divorce/break up (A)1.91(1.40–2.62)1.19(0.75–1.89)
  Death of partner (A)0.64(0.22–1.86)1.65(0.74–3.67)
  Lost job (B)1.04(0.78–1.38)1.44(0.94–2.21)
  Count of life events
    1 life event1.18(1.02–1.38)1.20(0.98–1.46)
    ≥2 life events1.20(1.00–1.44)1.17(0.92–1.49)
Chronic strains
  Financial difficulties1.03(0.86–1.23)1.14(0.90–1.45)
  Marital status (A)
    Single1.39(1.12–1.73)0.77(0.56–1.05)
    Separated/divorced1.80(1.37–2.35)1.41(1.01–1.95)
    Widowed0.99(0.58–1.71)0.80(0.53–1.23)
  Employment status (B)
    Unemployed1.47(1.08–2.01)0.98(0.57–1.67)
    Disability pension1.03(0.78–1.36)0.62(0.37–1.05)
    Retired early or retired0.94(0.73–1.21)0.61(0.38–0.98)
  Housing problems1.01(0.86–1.19)1.05(0.85–1.30)
  Neighbourhood problems0.94(0.81–1.09)1.05(0.87–1.27)

A higher cut-off point (drinking >29 14;U/week for men; >21 14;U/week for women) was then chosen to check whether the relationships were more pronounced at higher alcohol consumption levels. The relationships observed when using this higher cut-off point were all in the same direction as those using the traditional cut-off points (results not shown).

DISCUSSION

Although previous studies have examined the relationship between stress and drinking behaviour, results presented in the literature are often contradictory and the need for further research has been suggested (e.g. Pohorecky, 1991).

Before interpreting our results, several limitations of our study should be noted. Firstly, information on stressors was self-reported. Although respondents reported life events, they may not have perceived them as a source of stress and as something to cope with. Although we lack information on the way the respondents experienced these events, several researchers have argued that, rather than talking about positive or negative events, one could talk about events that require adaptation for the individual (e.g. Rabkin and Streuning, 1976). Since information on life events was collected by means of a questionnaire, the limitations of the methodology should be considered (Brown, 1981; Paykel, 1983; Cooke, 1985). By using a checklist of events, the definition of the event and its interpretation are left to the individual and may vary (e.g. ‘being a victim of a crime’ may range from being raped to minor robbery). In both cases, respondents could have answered in the affirmative. It should be noted, however, that most of the life events considered in our study are objective and leave little room for personal interpretation. The differences in reports of life events become more relevant when they are related to the outcome under study, in this case alcohol consumption. If heavy alcohol users were more likely to respond in the affirmative even in the case of minor events (perhaps because alcohol use provokes a negative life perception) than light or moderate drinkers, the results presented here could be partly due to this differential reporting. We did not collect data on the magnitude of the event.

Secondly, information on alcohol consumption was also based on self-report. It is well known that respondents tend to under-report their alcohol intake. If that were the case in our study, the cut-off points of 21 or 29 and 14 or 21 14;U/week would have to be raised, since people would drink somewhat more than they report. The directionality and strength of the relationships would, however, remain as described here.

Our study is also limited by its cross-sectional nature. High alcohol consumption may increase the probability that the individual experiences stress and problems, rather than the individual drinking in order to cope with the problems (Allan and Cooke, 1985; Cole et al., 1990). In our study, for example, heavy drinking could be thought to increase the likelihood of the individual being a victim of a crime, rather than drinking heavily to cope with the stress of a crime or assault (Table 2). Future research should consider the topic from a longitudinal perspective, since some drinking patterns may lead to stressful situations (divorce, financial difficulties, unemployment, etc.), rather than the other way round.

Although the relationship between stress and alcohol consumption has been studied extensively, results are often contradictory (Temple et al., 1991; Pohorecky, 1991; Pierce et al., 1994). For example, while some authors report that unemployment is related to alcohol consumption in men and women (Janlert and Hammerstrom, 1992), others found no relationship between unemployment and quantity of alcohol consumed (Cooke and Allan, 1983; Seeman and Seeman, 1992). It has been suggested that acute, rather than chronic, financial difficulties may be related to alcohol consumption (Pierce et al., 1994). We were able to distinguish between acute changes in marital and employment status (recent divorce or death of a partner; job loss) and chronic status (being divorced or widowed; unemployment).

The relationship between marital status and alcohol consumption has also been reported. For example, Romelsjo et al. (1991) observed an increase in alcohol consumption among men and women who were divorced, compared to those who were not. Wilsnack et al. (1991), by contrast, found a negative association between being divorced and heavy drinking among women, and Hanna et al. (1993) suggested that changes in marital status were more important than marital status itself in relation to drinking behaviour. In our study, we were able to assess marital status as well as changes in it. Although getting divorced was positively related to abstinence among men, the opposite was observed among women. Being divorced, the chronic counterpart of getting divorced, was also positively related with abstinence among men, but not related to abstinence among women. With regard to heavy drinking, both getting divorced and being divorced were positively related to heavy drinking among men, whereas only the latter was associated with heavy drinking among women.

Divorce and death of a partner have been studied together in the literature, since they represent a change in marital status to ‘not married’ and a consequent change in the individuals' roles (Temple et al., 1991; Hajema and Knibbe, 1998). The authors of these studies found that becoming ‘unmarried’ was associated with heavy drinking. In our study, we distinguished between becoming separated or divorced and becoming widowed. Since we observed a relationship between being/getting divorced and alcohol consumption, but no relationship between the latter and death of a partner/being widowed, we suggest that these two ways of ‘becoming unmarried’ should be analysed separately in the future.

Financial stress and unemployment in relation to alcohol consumption have also been studied. Some authors found no association between financial stress and alcohol (Moos et al., 1989), whereas others found economic strain to be related to drinking problems, both positively and negatively (Seeman and Seeman, 1992; Pierce et al., 1994). The same holds for unemployment: Janlert and Hammerstrom (1992) found a positive association suggesting that unemployment causes financial stress and that people drink alcohol in order to cope with stress, although other authors did not (Cooke and Allan, 1983; Seeman and Seeman, 1992). The authors who found a negative association, on the contrary, attributed their findings to the fact that alcohol becomes less affordable with the decrease in income (Pierce et al., 1994). In our study, we had information on employment status as well as on financial stressors (acute and chronic), so we were able to assess simultaneously the relationship between unemployment, financial stressors, and alcohol consumption.

Chronic financial stress (financial difficulties) was positively associated with abstinence among men and women, whereas acute financial stress (worsening of financial position) was positively associated with heavy drinking among men. Being unemployed was positively associated with heavy drinking among men. These relationships were observed simultaneously, so our data suggest that the relationships described in earlier studies between unemployment and financial position are independent.

Furthermore, our data allowed us to distinguish between acute and chronic financial stressors. Whereas a decrease in financial position was associated with heavy drinking among men, reporting financial difficulties was associated with abstinence. A possible explanation could be that, as a result of financial difficulties, alcohol becomes unaffordable, whereas a decrease in financial position may not necessarily result in financial difficulties. In the case of chronic financial stressors, alcohol consumption may be perceived as an aggravating factor, and therefore, as contraindicated.

Contradictory results regarding the relationship between employment status and alcohol consumption have been reported in the literature. Although heavy drinking is thought to be incompatible with some jobs, being employed may create more drinking opportunities for the individual (drinking with colleagues, etc.). Hajema and Knibbe (1998) suggested that becoming unemployed could be interpreted as a loss of role in a person's life, and, according to their role theory, it is likely to be associated with high alcohol consumption. We found unemployment to be related to heavy drinking only among men. Losing one's own job in the preceding 6 months, on the contrary, did not seem to be associated with either abstaining or heavy drinking. Our study corroborates results by other researchers that long-term unemployment is related to heavy drinking. The fact that the situation has become chronic or that a role loss has resulted from the job loss could be alternative explanations for the finding.

In our study, abstinence was associated with getting divorced or breaking up with a partner, being single, separated or divorced, receiving a disability pension, and having retired early or being retired. Although these conditions could be thought to be stressful to the individual, they could also be circumstances that simply limit the individuals' drinking possibilities, resulting in an increased likelihood for abstinence.

We further observed that stressors that were related to abstinence were also related to heavy drinking. Personality characteristics (unavailable for our population) could explain why one individual facing a stressor will abstain while another individual facing the same stressor will drink heavily. Future research should explore these characteristics.

Several authors have hypothesized that accumulation of events may have a greater effect on alcohol consumption than each of the events considered separately (Neff and Husani, 1982; Neff, 1984; Romelsjo et al., 1991). In our study, reporting more than one life event was positively associated with heavy drinking among men, but since the low limits of the confidence intervals were very close to unity, we concluded that our study provided little support for this hypothesis.

Characteristics other than the ones studied here could also be related to both susceptibility to stress and alcohol consumption, for example type A personality behaviour (Pohorecky, 1991; Brown, 1998). Unfortunately we did not have information on personality for our study. Future studies should investigate this relationship.

The relationship between alcohol consumption and health measures, such as mortality, cardiovascular diseases or stroke (Marmot et al., 1981; Stampfer et al., 1988; Marques-Vidal et al., 1996), and even subjective health (Poikolainen et al., 1996; San José et al., 1999) has been reported to be U- or J-shaped. Light or moderate drinkers have been suggested to have better health than either abstainers or heavy drinkers. The underlying mechanisms of this curvilinear relationship are not yet well understood. Biological mechanisms, such as cholesterol levels or blood pressure changes, have been suggested. However, these mechanisms do not fully explain the relationship for objective health outcomes. The mechanisms underlying a curvilinear relationship for more subjective health outcomes are less clear. Characteristics of the abstainers and heavy drinkers, other than their drinking habits, have been mentioned. Results presented here suggest that people under stressful conditions are more likely to either abstain or drink heavily. These results support the hypothesis that more than their drinking patterns, other characteristics of abstainers and heavy drinkers put them at increased risk for developing health burdens. Future research should combine these three elements: stress, health, and alcohol consumption.

Acknowledgments

Researchers' activities were conducted under a fellowship for the education of research abroad granted to B.S.J. by the Basque Government, Department of Education, Universities and Investigation, Spain.

Footnotes

  • * Author to whom correspondence should be addressed.

REFERENCES

View Abstract