The present study demonstrates that deprivation is common in the population considered, and that it has a strong association with tobacco, excessive alcohol intake, and psychotropic drug use. It highlights that risk patterns vary with the substance concerned, sex, and between active and retired people. The risk of substance use differed between deprivation dimensions. The material and social deprivation index used here was defined from seven criteria (low educational level, manual worker, unemployed, living alone, nationality other than western European, low income, and not being a home-owner) generally used in the literature [28, 30]. The nationality criterion was included in the deprivation index considered because it may be associated with cultural disadvantages, poor work/living conditions, poor living environment, poor health and access to care. It should be noted that racial composition has been included in deprivation indexes by several authors . The seven eigenvalues found with the principal component analysis were close enough (between 0.73 and 1.33) to suggest that all components contributed almost equally to the D value. Cronbach's alpha was modest. A similar observation was highlighted in the literature after elimination of redundant items in each domain. The deprivation index described here reflects the multidimensional character of community socioeconomic status .
These findings indicate that material and social conditions are potential risk factors for harmful health-related behaviours during both working life and retirement, and that the presence of several dimensions of deprivation is associated with a very high risk. This is consistent with the results of other studies, although, to our knowledge, they did not focus on all three substances studied here [19, 23, 28–30, 32]. Tobacco, alcohol and psychotropic drugs are strong contributors to social disparities in health [1, 3–5, 8, 23]. Two studies in France showed a strong relationship of the cumulative number of deprivation dimensions with tobacco, cannabis, psychotropic, tranquillizer and antidepressant use, as well as with physical and mental health status, obesity, underweight, diabetes and hypertension [5, 24].
Any selection bias here would be small: 96% of households had telephones at the time of the study, and only 16% had confidential addresses. Discussions before the survey, for example with associations of people with disabilities, suggested that this list would not be biased with regard to health status or living conditions. The participation rate was rather modest but similar to that achieved in similar surveys in France [1, 49]. The age and sex distributions of the sample reflect those of the general population of Lorraine . The quality of the completed questionnaires was very good. As mentioned above, all the factors studied had been validated and used in other studies [5, 8, 13, 18, 43].
Although the study was conducted on a large sample, the results should be interpreted with caution due to possible selection bias. The self-administered occupational health history questionnaire is considered reliable and valid . A study analysing non-response bias in a mailed health survey showed that respondents and non-respondents were of similar sex and age distributions, and close in terms of health care expenditure . Similar observations were reported by the Maastricht Cohort Study . The prevalences of various variables in the present sample were similar to the directly standardized adjusted rates computed in reference to the Lorraine population . This is due, as noted above, to similar age and sex distributions in the sample and the Lorraine population as a whole. It should be noted that our study would underestimate the differences, as the most economically deprived (i.e. those with no home and therefore no telephone) were not included in the sample.
Our study found that 30% of men and 22% of women were current smokers. These figures differ slightly from those reported among French people aged 12–75 in 1999 (32% and 26% respectively) . Alcohol abuse (in terms of the Deta index) was similar at 13.3% in men and 4% in women . The higher prevalence of alcohol abuse among men seen here was also found in the ESEMeD study . Psychotropic drug use was common (23.8% in men and 41.0% in women). Comparison with other studies is difficult because of variations in the study populations, the psychotropic drugs considered, and the methodological approaches adopted. In France, one-third of workers use drugs for work-related reasons, 20% to feel better, 12% to control an awkward symptom, and 18% to relax after a difficult day's work [41, 54]. The ESEMeD study found that the prevalence of antidepressant, anxiolytic, and antipsychotic or mood-stabilizing drug use over a 12-month period was 19.2% in France, 15.5% in Spain, 13.7% in Italy, 13.2% in Belgium, 7.4% in the Netherlands, and 5.9% in Germany . In Belgium, Bruffaerts et al.  reported that about 19% of people aged 18 years or more said they had used a psychotropic drug over the previous 12 months.
The present study reveals a strong association between deprivation and tobacco and psychotropic drug use, for the two age groups (<40 and ≥40 years), among working/other non-retired men and women. In other words, the likelihood of tobacco and psychotropic drug use increases with the deprivation score, and the association appears early and persists throughout the working lives of both sexes. These findings are consistent with those of other studies. It is reported that one-third of the French working population use medications or other legal psychoactive substances in order to cope with work-related difficulties, and that such use is more common in manual workers . Manual workers have poorer working conditions that may lead to physical and mental disturbances  and consequently to psychotropic drug use. Manual workers also have a higher prevalence of tobacco use [2, 5] and of disabilities than other workers in both age groups <40 and ≥40 years . Physical job demands lead to fatigue  and the development of work-related stress reactions, psychological overload, and health problems . Cumulative job stress is common and is associated with increased risk of mental health disorders and psychotropic drug use . The volume of services provided and job dissatisfaction are associated with hypnotic and tranquillizer use . Interesting findings here were that among the seven deprivation domains studied: (1) low educational level, being a manual worker and low income were associated with psychotropic drug use among men as well as women; (2) tobacco use was related to low educational level, being a manual worker, low income and not being a home-owner in men, and to low income, living alone, unemployment and not being a home-owner in women. Other studies have stated that unemployment, low educational level, being a manual worker, being divorced or widowed, and living alone were associated with an increased risk of tobacco and psychotropic drug use in France and in Europe [1, 2, 5, 35, 36]. No relationship between nationality and any substance use was observed in our study – this was also true for psychotropic drug use in Europe (when controlling for co-factors) .
With regard to alcohol abuse, our study found a strong relationship with deprivation in male, but not female, working/other non-retired people. Alcohol abuse was more common in men than in women (12.5% vs. 3.3%, p < 0.001). This sex difference has also been reported by other studies in France [25, 42] which related increasing deprivation to alcohol abuse, but not daily alcohol use . It should be noted that of the seven deprivation criteria considered, only low income, living alone and not being a home-owner were related to alcohol abuse in men, and no association was observed among women. The differences between the sexes may be explained in part by the fact that alcohol abuse is three-fold lower in women, who are more likely to take psychotropic drugs [1, 17]. Excessive alcohol use appeared here to be more associated with poverty and poor living conditions  than with working issues. Other investigations have found a relationship between living alone and unemployment, and partner-relationship disruptions are strongly associated with suicidal behaviour among individuals with alcoholism [35, 36].
Another important finding is that risk patterns for substance use clearly differ between active and retired men, and between retired men and retired women. Indeed, among the active and retired men, there was a gradient in the relationship between psychotropic drug useand deprivation score, whereas smoking and alcohol abuse were associated only with D ≥ 3. The lower overall associations between deprivation and smoking and alcohol abuse among retired men (except D ≥ 3 group for alcohol abuse) may be explained as follows: (1) smoking and alcohol abuse are less frequent in retired people (compared with other generations/age groups); (2) premature mortality (before 65/70 years) is higher among subjects most at risk, particularly manual workers, smokers and alcohol users [37, 38] (this was also observed in the 9-year prospective premature mortality (<70 years) analysis of the sample studied: manual workers had an increased adjusted risk ratio of 1.84, 95% CI 1.00–3.37 compared with upper class; risk ratios for current smokers and ex-smokers were 1.76, 95% CI 1.08–2.88 and 1.52, 95% CI 0.96–2.40 respectively; and subjects with alcohol abuse had a risk ratio of 2.07, 95% CI 1.31–3.26 – as yet unpublished data); and (3) the absence of stressful working conditions and a lesser perception of complaints without the physical and mental demands of a job. The strong relationship between psychotropic drug use and deprivation in retired men may be explained by a higher prevalence of altered health status and disability due to aging  that could increase psychotropic drug use [1, 60, 61]. Regarding retired women, only subjects aged less than 70 years with D ≥ 3 had a significant risk for psychotropic drug use (that would be due to manual workers who had a higher consumption, close to significance). It should be noted that only three of the seven deprivation domains were associated with excessive alcohol and psychotropic drug use in retired men: having been a manual worker, low income, and not being a home-owner. Among retired women, only not being a home-owner was related to alcohol abuse. Therefore substance use in retired people relates to previous work conditions and poverty.
The present survey demonstrates that an accumulation of several deprivation dimensions is associated with marked deterioration in health-related behaviours [25, 28–32, 35, 36, 39] to cope with very difficult living and working conditions. In many industrialized countries, people start smoking at an increasingly younger age, putting themselves at greater, and earlier, risk of avoidable tobacco-related illnesses [35, 62]. Alcohol abuse and psychotropic drug use are very common and often begin at a young age [18, 43, 62]. Findings over recent decades have shown that multiple deprivations affect many people from childhood onwards, and that the accumulation of various aspects of disadvantages leads to marked deteriorations in living conditions, health-related behaviours and health status [2, 8, 23, 25–32, 34–36, 39, 41]. Observed geographical differences in heath outcomes are attributed to the individual characteristics of members of the populations concerned and their living conditions and lifestyle [63, 64]. Some adult health problems and premature mortality may be influenced by the childhood circumstances of the person concerned [2, 65, 66]. Interventions should be designed and evaluated to address these issues, and the most promising should be implemented on a large scale.
In the present study, women were different from men in that the association between deprivation and substance use was less evident, particularly among retired people. This may be attributable to less job demands, and to differences in the substances consumed in order to deal with everyday life. There were clear differences between age groups in the patterns of both deprivation and substance use. Low educational level was more common among older people, but younger people were more likely to report low income and not being a home-owner. Among men, the proportion of manual workers was higher in the 30–59 age group than in the group aged 60+ years. This tendency can be explained by premature death: for the sample studied between 1996 and 2004, there were 85 deaths in people under 65, and 41 between 65 and 70; premature mortality (<70 years) rates per 1,000 person-years were 1.32, 4.33 and 12.6 for the age groups ≤49, 50–59 and ≥60 years, respectively (data not yet published). Compared with other age groups, men aged 40–69 years were less likely to be living alone, as were women under 50. This phenomenon, which was more marked in women, can in part be explained by the premature death of the spouse. Such changes reflect the deprivation many people face, and the different ways in which it manifests itself over a lifetime.
The results of our study highlight the associations between accumulation of deprivation dimensions and tobacco, alcohol, and psychotropic drug use at various ages. The role of cumulative advantages/disadvantages during the life-time or the mechanism for inequalities across a temporal process are investigated by few authors [67, 68]. Ross and Wu  showed that among the subjects aged 20 to 64 in the United States, the gap in self-reported health, physical functioning, and physical well-being among people with high and low educational attainment increases with age, and that household income does not explain education's effect. In our study, a stronger relationship in the working and other non-retired people aged 40+ than in those aged 39 or less was found between the deprivation score and psychotropic use only. DiPrete and Eirich  examined the different theoretical and empirical cumulative advantages models proposed by sociologists, sociobiologists, social psychologists, and economists as a mechanism for inequality. However, these interesting issues could not be treated in our cross-sectional study. We rather examined the relationships between the accumulation of several deprivation dimensions and tobacco, alcohol, and psychotropic drug use at the time of the survey. Many factors play a role in creating or reducing inequality over time, including: mortality of subjects most at risk, cultural changes, movements of the population (for example north-south for older people, rural migration, etc.), and changes in society as a whole (education levels, living conditions, air pollution, working procedures and hazards, life style, family structure, etc.).