Aims

To investigate the association between alcohol consumption and different causes of work cessation and estimate the loss of occupational activity among high consumers compared with low consumers.

Methods

From the prospective study of men employed in the French gas and electric company, 8442 men during a median follow-up of 8.4 years reported on their alcohol consumption. Information on work cessation was collected from the company administrative records. Hazard Ratios (HRs) by cause of work cessation (death, disability, retirement before or after age 55) were estimated using a competing risk method.

Results

An increasing quantity of daily alcohol consumption was associated with an increased risk of death, disability and retirement before age 55 (P trend ≤ 0.01, = 0.03 and ≤ 0.01, respectively), but not of retirement after age 55 (P trend = 0.56). Moreover, compared with low consumption, moderate, high or very high daily intakes were associated with an increased risk of early work cessation (combination of the three causes: death, disability and retirement before age 55) (HR = 1.14, 95% confidence interval (CI) = 1.05–1.25; HR = 1.23, 95% CI = 1.12–1.35 and HR = 1.49, 95% CI = 1.15–1.92 respectively). Between ages 50 and 60, we estimated that high or very high consumers could gain 6.04 months of occupational activity if they drank like low consumers.

Conclusions

Our results provide evidence of a dose-effect relationship between alcohol consumption and early work cessation.

INTRODUCTION

Alcohol consumption has been consistently associated with many diseases such as cancers, cardiovascular diseases (Corrao et al., 2004), digestive diseases (alcoholic hepatitis, pancreatitis, alcoholic cirrhosis) and mental or behavioural disorders (Boden and Fergusson, 2011). It is also associated with overall mortality through a J-shape curve (Rehm et al., 2001; Klatsky and Udaltsova, 2007). The World Health Organization (WHO) recently estimated that the proportion of alcohol-attributable deaths reaches 13.3% in Europe (WHO, 2014).

On the other hand, social and economic effects of alcohol in the workforce have been little investigated. Some studies reported that alcohol consumption is positively associated with loss of productivity (Thavorncharoensap et al., 2010), and with absenteeism in adults (Johansson et al., 2009; Salonsalmi et al., 2009; Jarl and Gerdtham, 2012) and adolescents (Mounteney et al., 2010; Alwan et al., 2011; Koutra et al., 2012). Others reported that a decrease in alcohol intake is associated with decreased absenteeism (Lenneman et al., 2011; Magnus et al., 2012). Nevertheless, little is known on the effects of alcohol consumption on work cessation. Some studies observed a positive association between high alcohol consumption and disability retirement (Upmark et al., 1997, 1999; Mansson et al., 1999; Salonsalmi et al., 2012; Skogen et al., 2012; Samuelsson et al., 2013), but they were mainly conducted in Northern European countries and none was conducted in Southern European countries.

The patterns of alcohol drinking (taking into account the quantity but also the social circumstances of alcohol drinking) of Northern Europe are different from those of Southern Europe. According to the WHO classification (scale from 1 (less risky pattern) to 5 (most risky pattern)), Northern European countries' consumptions are classified in the third category, namely because of high frequency of binge drinking, whereas Western/Southern European countries consumptions are mostly classified in the first category, characterized by a relatively high daily consumption during meals (WHO, 2014). However, in Southern European countries mean alcohol consumption is often higher than in Northern countries. For example, in 2010, the mean alcohol consumption was 10.9 l of pure alcohol/year in Europe, 9.2 l/year in Sweden and 12.2 l/year in France (WHO, 2014). In addition, WHO risk patterns primarily relate to morbidity and mortality rather than work-related outcomes. Studies of association between alcohol consumption and risk of work cessation in Southern European countries are thus warranted to better understand the association between alcohol consumption and work cessation. The large variability of quantity consumed could indeed allow the detection of a gradient in risk, which may not exist in countries with risky patterns of alcohol drinking highly driven by binge drinking.

We investigated the association between alcohol consumption and work cessation in a large prospective cohort of French workers of the French national electricity and gas company ‘EDF-GDF’. Because workers typically stayed with this company until retirement (Goldberg et al., 2015), we had a unique opportunity to simultaneously examine the different causes of labour market exit (death, disability and retirement). We also estimated the loss of occupational activity (in terms of time lost out of the workforce) of high consumers compared with low consumers.

MATERIAL AND METHODS

The GAZEL cohort study

The GAZEL cohort was established in 1989 among employees of the French national electricity and gas company ‘EDF-GDF’. In January 1989, all male employees, then aged 40–50 years, and all women employees, aged 35–50 years, were invited to participate in the cohort. Of the 44,922 eligible subjects (31,411 men and 13,511 women), 20,625 (45.9%) accepted: 15,011 men and 5614 women. Higher participation was associated with being male, being married, having children, holding managerial status and residing in some regions of France (Goldberg et al., 2007). EDF-GDF employees hold civil-servant-like status that entails job security and opportunities for occupational mobility. Typically, they were hired in their 20s and stayed with the company until retirement (Goldberg et al., 2015). Since enrolment in the cohort, they have been followed prospectively through a yearly self-administered questionnaire, including a large set of questions on health, lifestyle (including alcohol consumption), individual, familial, social and occupational factors (questionnaires available in French on http://www.gazel.inserm.fr/fr/documentation/questionnaires.html). Less than 1% of the participants have been lost to follow-up since 1989 (Goldberg et al., 2015).

The study protocol, including written consent from the cohort participants, was approved by the French authority for data confidentiality (Commission Nationale Informatique et Liberté # 105728) and by INSERM's Ethics Evaluation Committee (IRB0000388, FWA00005831).

Work cessation data

We obtained information on occupational status throughout the career in EDF-GDF (including retirement date, pension for disability or long-standing illness) as well as on mortality from the company administrative records. Because retirement pensions are paid by EDF-GDF, these records provide comprehensive and accurate information. Statutory age of retirement for EDF-GDF employees is between 55 and 60 years, depending mainly on the type of job: manual/unskilled workers are allowed to retire earlier than managers. Participants were considered in disability if they had been granted a pension for disability or longstanding illness, or if they had been on sick leave for >365 consecutive days.

Alcohol consumption

Data on health, lifestyle, alcohol consumption and other characteristics of the GAZEL participants were collected by annual surveys using a mailed self-administered questionnaire. Alcohol consumption was assessed in January 1989, January 1990 and January 1991 with questions on consumption frequency of wine, beer or cider, and spirits (‘Among these beverages, which do you never drink, which do you drink occasionally and which do you drink daily?’), and on the quantity consumed daily (‘For beverages daily consumed, indicate the number of drinks per day’). We focused on daily alcohol consumption. Indeed, we could not distinguish between real non-drinkers and former heavy drinkers among our non-drinker category, and the consumption frequency (weekly or monthly for example) and quantity were not recorded for occasional drinkers. The categories of non-consumers and occasional consumers can thus be composed of very mixed drinkers and were therefore not included in the present study.

In order to calculate the daily quantity consumed in g/day, we considered that each drink of alcohol contains 10 g of pure alcohol. For each questionnaire, we then classified the daily alcohol consumption as low, moderate, high or very high daily consumption according to the World Health Organization (WHO) classification (WHO, 2010), i.e. cut-offs of 20, 40 and 60 g/day for women and 40, 60 and 100 g/day for men. We then combined the answers to the three questionnaires to create a single variable categorized into four classes (low/moderate/high/very high consumption). First, we defined very high drinkers as people declaring a very high consumption at least at one of the three questionnaires. We then applied the same decision to define successively high, moderate and low consumers among the remaining population.

Study population

Follow-up started on 1 January 1 1991. Participants contributed person-years of follow-up until the date of work cessation or until 1 January 2014, whichever occurred first. We restricted our population to men because there were too few women with moderate or high alcohol consumption who ceased to work during follow-up. Among the 15,011 men included in the GAZEL cohort, we further excluded those who stopped working before 1991 (n = 193), as well as non-drinkers and occasionally alcohol consumers (n = 6376). Thus, 8442 men were available for analysis (Fig. 1).

Fig. 1.

Flowchart.

Fig. 1.

Flowchart.

Statistical analyses

Cox proportional hazards models with age as timescale were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) of transition from employment to work cessation according to alcohol categories. HRs by cause of work cessation (death, disability, retirement before or after age 55) were estimated using a competing risk method (Rosthøj et al., 2004; Putter et al., 2007). Missing data for alcohol (n = 61, 0.7%) were replaced after a single imputation from a conditional distribution using company records on sickness absences for alcoholism in addition to other covariates (Schafer and Graham, 2002). To test for a trend with alcohol consumption level (P trend), we considered daily drinking as a continuous variable (0: low consumers/1: moderate consumers/2: high consumers/3: very high consumers). We adjusted our models for potential confounders: socio-professional category at hiring (manual workers/clerical workers/intermediate/managers or executives) and smoking status in 1991 (<11 pack-years/≥11 pack-years, the median value in our population). We also tested the association with two other potential explaining variables: marital status and body mass index (BMI) in 1991 (categories presented in Table 1). Missing data for smoking status (n = 1248, 14.85%), marital status (n = 1195, 14.2%) and BMI (n = 1756, 20.8%) in 1991 were replaced with the value from the previous questionnaires (1990 or 1989). The remaining missing values (<1% for socio-professional categories, smoking and marital status and 13.5% for BMI) were included in our models as a separate category. In a sensitivity analysis, we excluded men with imputed data for alcohol consumption (n = 61).

Table 1.

Selected characteristics of participants, overall and according to the alcohol consumption level: GAZEL cohort, France (n = 8442)

 Total Alcohol consumption (g/day)
 
Low (1–40) Medium (41–60) High (61–100) Very high (>100) 
Number 8442 5559 1423 1342 118 
Age at baseline (years), mean (SD) 46.6 (2.9) 46.6 (2.9) 46.6 (2.9) 46.5 (2.9) 46.3 (2.9) 
Age at the end of follow-up (years), mean (SD) 55.0 (2.5) 55.2 (2.5) 54.7 (2.4) 54.7 (2.6) 53.9 (2.9) 
Work cessation for, n (%) 
 Death 188 (2.2) 97 (1.7) 36 (2.5) 47 (3.5) 8 (6.8) 
 Disability 75 (0.9) 47 (0.9) 6 (0.4) 17 (1.3) 5 (4.2) 
 Retirement 8176 (96.9) 5414 (97.4) 1381 (97.0) 1276 (95.1) 105 (89.0) 
Retirement before 55 years 3146 (38.5) 1953 (36.1) 581 (42.1) 564 (44.2) 48 (45.7) 
Retirement after 55 years 5030 (61.5) 3461 (63.9) 800 (57.9) 712 (55.8) 57 (54.3) 
Socio-professional categories at hiring, n (%) 
 Managers or executives 770 (9.1) 575 (10.3) 108 (7.6) 87 (6.5) 0 (0.0) 
 Intermediate 722 (8.6) 494 (8.9) 120 (8.4) 94 (7.0) 14 (11.9) 
 Clerical workers 1840 (21.8) 1233 (22.2) 270 (19.0) 313 (23.3) 24 (20.3) 
 Manual workers 5100 (60.4) 3252 (58.5) 924 (64.9) 847 (63.1) 77 (65.3) 
 Missing values 10 (0.1) 5 (0.1) 1 (0.1) 1 (0.1) 3 (2.5) 
Smoking status at baseline (1991), n (%) 
 <11 pack-years 4233 (50.1) 3222 (58.0) 558 (39.2) 409 (30.5) 44 (37.3) 
 ≥11 pack-years 4146 (49.1) 2311 (41.6) 846 (59.5) 922 (68.7) 67 (56.8) 
 Missing values 63 (0.8) 26 (0.5) 19 (1.3) 11 (0.8) 7 (5.9) 
Marital status at baseline (1991), n (%) 
 Married 7831 (92.8) 5180 (93.2) 1333 (93.7) 1217 (90.7) 101 (85.6) 
 Unmarried 175 (2.1) 109 (2.0) 23 (1.6) 39 (2.9) 4 (3.4) 
 Divorced 384 (4.6) 240 (4.3) 57 (4.0) 78 (5.8) 9 (7.6) 
 Widowed 38 (0.5) 24 (0.4) 7 (0.5) 7 (0.5) 0 (0.0) 
 Missing values 14 (0.2) 6 (0.1) 3 (0.2) 1 (0.1) 4 (3.4) 
BMI at baseline (1991), n (%) 
 <18.5 kg/m² 26 (0.3) 14 (0.3) 4 (0.3) 7 (0.5) 1 (0.9) 
 [18.5–25] kg/m² 3457 (41.0) 2433 (43.8) 539 (37.9) 453 (33.8) 32 (27.1) 
 [25–30] kg/m² 3383 (40.1) 2168 (39.0) 596 (41.9) 588 (43.8) 31 (26.3) 
 ≥30 kg/m² 440 (5.2) 248 (4.5) 80 (5.6) 105 (7.8) 7 (5.9) 
 Missing values 1136 (13.5) 696 (12.5) 204 (14.3) 189 (14.1) 47 (39.8) 
 Total Alcohol consumption (g/day)
 
Low (1–40) Medium (41–60) High (61–100) Very high (>100) 
Number 8442 5559 1423 1342 118 
Age at baseline (years), mean (SD) 46.6 (2.9) 46.6 (2.9) 46.6 (2.9) 46.5 (2.9) 46.3 (2.9) 
Age at the end of follow-up (years), mean (SD) 55.0 (2.5) 55.2 (2.5) 54.7 (2.4) 54.7 (2.6) 53.9 (2.9) 
Work cessation for, n (%) 
 Death 188 (2.2) 97 (1.7) 36 (2.5) 47 (3.5) 8 (6.8) 
 Disability 75 (0.9) 47 (0.9) 6 (0.4) 17 (1.3) 5 (4.2) 
 Retirement 8176 (96.9) 5414 (97.4) 1381 (97.0) 1276 (95.1) 105 (89.0) 
Retirement before 55 years 3146 (38.5) 1953 (36.1) 581 (42.1) 564 (44.2) 48 (45.7) 
Retirement after 55 years 5030 (61.5) 3461 (63.9) 800 (57.9) 712 (55.8) 57 (54.3) 
Socio-professional categories at hiring, n (%) 
 Managers or executives 770 (9.1) 575 (10.3) 108 (7.6) 87 (6.5) 0 (0.0) 
 Intermediate 722 (8.6) 494 (8.9) 120 (8.4) 94 (7.0) 14 (11.9) 
 Clerical workers 1840 (21.8) 1233 (22.2) 270 (19.0) 313 (23.3) 24 (20.3) 
 Manual workers 5100 (60.4) 3252 (58.5) 924 (64.9) 847 (63.1) 77 (65.3) 
 Missing values 10 (0.1) 5 (0.1) 1 (0.1) 1 (0.1) 3 (2.5) 
Smoking status at baseline (1991), n (%) 
 <11 pack-years 4233 (50.1) 3222 (58.0) 558 (39.2) 409 (30.5) 44 (37.3) 
 ≥11 pack-years 4146 (49.1) 2311 (41.6) 846 (59.5) 922 (68.7) 67 (56.8) 
 Missing values 63 (0.8) 26 (0.5) 19 (1.3) 11 (0.8) 7 (5.9) 
Marital status at baseline (1991), n (%) 
 Married 7831 (92.8) 5180 (93.2) 1333 (93.7) 1217 (90.7) 101 (85.6) 
 Unmarried 175 (2.1) 109 (2.0) 23 (1.6) 39 (2.9) 4 (3.4) 
 Divorced 384 (4.6) 240 (4.3) 57 (4.0) 78 (5.8) 9 (7.6) 
 Widowed 38 (0.5) 24 (0.4) 7 (0.5) 7 (0.5) 0 (0.0) 
 Missing values 14 (0.2) 6 (0.1) 3 (0.2) 1 (0.1) 4 (3.4) 
BMI at baseline (1991), n (%) 
 <18.5 kg/m² 26 (0.3) 14 (0.3) 4 (0.3) 7 (0.5) 1 (0.9) 
 [18.5–25] kg/m² 3457 (41.0) 2433 (43.8) 539 (37.9) 453 (33.8) 32 (27.1) 
 [25–30] kg/m² 3383 (40.1) 2168 (39.0) 596 (41.9) 588 (43.8) 31 (26.3) 
 ≥30 kg/m² 440 (5.2) 248 (4.5) 80 (5.6) 105 (7.8) 7 (5.9) 
 Missing values 1136 (13.5) 696 (12.5) 204 (14.3) 189 (14.1) 47 (39.8) 

BMI, body mass index; SD, standard deviation.

In order to examine loss of occupational activity associated with alcohol consumption, we then restricted our population to participants who ceased to work after 50 years (n = 8363), and grouped together high and very high consumers. Adjusted cumulative incidences of work cessation for death, disability, retirement before or after age 55 were estimated for manual workers and smokers of less than 11 pack-years (the reference categories in our analyses) at each point of the age range 50–60, and separately for participants with low, moderate and high or very high alcohol consumption. We chose the age range 50–60 years because this is the period during which most work cessations occur. Using these cumulative incidences, we then estimated the difference in the amount of time lost to occupational activity between 50 and 60 years according to the level of alcohol consumption. For example, occupational activity lost by high consumers compared with low consumers was calculated as the difference between the area under the curve of the cumulative incidence in high consumers and the area under the curve of the cumulative incidence in low consumers (Andersen, 2013). Confidence intervals for these differences were estimated using the bootstrap percentile method (Herquelot et al., 2011).

All tests were two-sided with P-value set at 0.05. All analyses were performed using the Statistical Analysis Systems (SAS) software, version 9.4, and the Cum Inc. (Rosthøj et al., 2004), boot and bootci macros (SAS Institute, Inc., Cary, North Carolina).

RESULTS

Among 8442 men (median duration of follow-up = 8.4 years, standard deviation (SD) = 3.7), 188 stopped working because of death, and 75 because of disability. Among the remaining 8179 men, 8173 had retired at the end of follow-up. Among them, 3146 (38.5%) retired before age 55. Characteristics of participants according to alcohol consumption are given in Table 1. The majority of the population (over 80%) had been hired at a low occupational position (manual or clerical workers), especially among high (86.4%) or very high (85.6%) alcohol consumers. Low consumers were less often heavy smokers in 1991 than high or very high consumers of alcohol.

The mean age at work cessation was 55.0 years (SD = 2.5), but low consumers ceased to work older (mean (SD) = 55.2 (2.5) years) than high or very high consumers (mean (SD) = 54.7 (2.6) and 53.9 (2.9) years respectively). The majority of work cessations due to death occurred before age 55 (n = 157, 83.5%, data not shown). However, only 79.4% of work cessations due to death occurred before age 55 among low consumers (n = 77), whereas the corresponding percentages among high and very high consumers were 87.2 and 87.5% respectively. Work cessation for disability also often occurred before age 55 in our population (58.0%), particularly in high (64.7%) or very high consumers (100.0%) but not in low consumers (42.6%). Retirement occurred more often after age 55 (61.5%) but a larger proportion of men retired before age 55 among high or very high consumers (44.2 and 45.7%, respectively) than among low consumers (36.1%).

Associations between alcohol intake and causes of work cessation are presented in Table 2. A very high alcohol intake was associated with an increased risk of work cessation for death and disability (hazard ratio (HR) adjusted for age, socio-professional categories at hiring and smoking status at baseline = 4.28, 95% confidence interval (CI) = 2.07–8.85 and HR = 5.96, 95% CI = 2.33–15.20 for death and disability, respectively). A high alcohol intake was also associated with an increased risk of work cessation for death (HR adjusted for age, socio-professional categories at hiring and smoking status at baseline = 1.81, 95% CI = 1.27–2.59). Moreover, an increasing quantity of daily alcohol consumption was associated with an increased risk of death, disability and retirement before age 55 (P trend ≤0.01, =0.03 and ≤0.01, respectively), but not of retirement after age 55 (P trend = 0.56).

Table 2.

Association between alcohol consumption and work cessation for death, disability and retirement: GAZEL cohort, France (n = 8442)

Alcohol consumption (g/day) N cases Person-years HR (95% CI)a P for trend HR (95% CI)b P for trend HR (95% CI)c P for trend 
Employment to death 
 Low (1–40) 97 47,796 1.00 (ref) <0.01 1.00 (ref) <0.01 1.00 (ref) <0.01 
 Medium (41–60) 36 11,552 1.57 (1.07–2.31) 1.58 (1.08–2.32) 1.43 (0.97–2.11) 
 High (61–100) 47 11,052 2.14 (1.51–3.03) 2.10 (1.48–2.97) 1.81 (1.27–2.59) 
 Very high (>100) 889 4.69 (2.28–9.66) 4.53 (2.20–9.34) 4.28 (2.07–8.85) 
Employment to disability 
 Low (1–40) 47 47,796 1.00 (ref) <0.01 1.00 (ref) 0.01 1.00 (ref) 0.03 
 Medium (41–60) 11,552 0.59 (0.25–1.39) 0.60 (0.26–1.42) 0.53 (0.22–1.25) 
 High (61–100) 17 11,052 1.74 (1.00–3.04) 1.67 (0.96–2.91) 1.44 (0.82–2.55) 
 Very high (>100) 889 7.28 (2.89–18.34) 6.75 (2.68–17.01) 5.96 (2.33–15.20) 
Employment to retirement <55 years 
 Low (1–40) 1953 47,796 1.00 (ref) <0.01 1.00 (ref) <0.01 1.00 (ref) <0.01 
 Medium (41–60) 581 11,552 1.23 (1.12–1.35) 1.18 (1.07–1.29) 1.14 (1.04–1.26) 
 High (61–100) 564 11,052 1.31 (1.19–1.44) 1.24 (1.13–1.37) 1.20 (1.09–1.32) 
 Very high (>100) 48 889 1.41 (1.06–1.88) 1.29 (0.97–1.72) 1.26 (0.95–1.68) 
Employment to retirement ≥55 years 
 Low (1–40) 3461 47,796 1.00 (ref) <0.01 1.00 (ref) 0.88 1.00 (ref) 0.56 
 Medium (41–60) 800 11,552 1.16 (1.07–1.25) 1.09 (1.01–1.17) 1.10 (1.02–1.19) 
 High (61–100) 712 11,052 1.06 (0.98–1.15) 0.97 (0.90–1.05) 0.99 (0.91–1.07) 
 Very high (>100) 57 889 1.36 (1.04–1.76) 1.05 (0.80–1.37) 1.05 (0.81–1.37) 
Alcohol consumption (g/day) N cases Person-years HR (95% CI)a P for trend HR (95% CI)b P for trend HR (95% CI)c P for trend 
Employment to death 
 Low (1–40) 97 47,796 1.00 (ref) <0.01 1.00 (ref) <0.01 1.00 (ref) <0.01 
 Medium (41–60) 36 11,552 1.57 (1.07–2.31) 1.58 (1.08–2.32) 1.43 (0.97–2.11) 
 High (61–100) 47 11,052 2.14 (1.51–3.03) 2.10 (1.48–2.97) 1.81 (1.27–2.59) 
 Very high (>100) 889 4.69 (2.28–9.66) 4.53 (2.20–9.34) 4.28 (2.07–8.85) 
Employment to disability 
 Low (1–40) 47 47,796 1.00 (ref) <0.01 1.00 (ref) 0.01 1.00 (ref) 0.03 
 Medium (41–60) 11,552 0.59 (0.25–1.39) 0.60 (0.26–1.42) 0.53 (0.22–1.25) 
 High (61–100) 17 11,052 1.74 (1.00–3.04) 1.67 (0.96–2.91) 1.44 (0.82–2.55) 
 Very high (>100) 889 7.28 (2.89–18.34) 6.75 (2.68–17.01) 5.96 (2.33–15.20) 
Employment to retirement <55 years 
 Low (1–40) 1953 47,796 1.00 (ref) <0.01 1.00 (ref) <0.01 1.00 (ref) <0.01 
 Medium (41–60) 581 11,552 1.23 (1.12–1.35) 1.18 (1.07–1.29) 1.14 (1.04–1.26) 
 High (61–100) 564 11,052 1.31 (1.19–1.44) 1.24 (1.13–1.37) 1.20 (1.09–1.32) 
 Very high (>100) 48 889 1.41 (1.06–1.88) 1.29 (0.97–1.72) 1.26 (0.95–1.68) 
Employment to retirement ≥55 years 
 Low (1–40) 3461 47,796 1.00 (ref) <0.01 1.00 (ref) 0.88 1.00 (ref) 0.56 
 Medium (41–60) 800 11,552 1.16 (1.07–1.25) 1.09 (1.01–1.17) 1.10 (1.02–1.19) 
 High (61–100) 712 11,052 1.06 (0.98–1.15) 0.97 (0.90–1.05) 0.99 (0.91–1.07) 
 Very high (>100) 57 889 1.36 (1.04–1.76) 1.05 (0.80–1.37) 1.05 (0.81–1.37) 

Cox proportional hazards model with competing risk method to estimate HRs by cause of work cessation (death, disability, retirement before or after age 55) and adjusted for confounders as noted in the footnote of the table.

HR, hazard ratio; CI, confidence interval.

aAdjusted on age.

bAdjusted on age and socio-professional categories at hiring.

cAdjusted on age, socio-professional categories at hiring and smoking status in 1991.

Moderate, high or very high daily intakes were associated with an increased risk of early work cessation (combination of the three causes: death, disability and retirement before age 55) (HR = 1.14, 95% CI = 1.05–1.25; HR = 1.23, 95% CI = 1.12–1.35 and HR = 1.49, 95% CI = 1.15–1.92 respectively; P trend <0.01, data not tabulated).

Figure 2 shows the adjusted cumulative incidences of work cessation for death, disability and retirement before age 55, for smokers of less than 11 pack-years and manual workers, by the level of alcohol intake in daily consumers, at each point of the age range 50–60 years. There was a steep increase for retirement at age 53, corresponding to an opportunity offered by the company to retire earlier than the legal age. The cumulative incidences of death, disability or retirement before age 55 increased more slowly over time in low consumers than in high or very high consumers.

Fig. 2.

Cumulative incidences of death, disability and retirement before age 55 between 50 and 60 years. GAZEL cohort, France (n = 8363). Cox-adjusted cumulative incidences of death (a), disability (b), retirement before age 55 (c) for smokers of less than 11 pack-years and manual workers, according to age between 50 and 60 years among participants with a low (continuous line), moderate (dashed line) and high or very high (dotted line) consumption of alcohol.

Fig. 2.

Cumulative incidences of death, disability and retirement before age 55 between 50 and 60 years. GAZEL cohort, France (n = 8363). Cox-adjusted cumulative incidences of death (a), disability (b), retirement before age 55 (c) for smokers of less than 11 pack-years and manual workers, according to age between 50 and 60 years among participants with a low (continuous line), moderate (dashed line) and high or very high (dotted line) consumption of alcohol.

Between the ages of 50 and 60 years, we estimated that each manual worker who smoke less than 11 pack-years with a high or a very high consumption of alcohol lost an average 6.04 months of occupational activity (95% CI = 3.48–8.88), compared with men who drank less than 40 g/day of alcohol (data not tabulated). Retirement before age 55 accounted for 5.03 months of this loss, while death accounted for 0.81 month and disability for 0.19 month. The equivalent figures for men drinking 41–60 g/day of alcohol compared with low drinkers were 4.02 months (95% CI = 1.20–6.84) of global loss from which 3.83 months were attributed to retirement before age 55, 0.36 month to death and a mean gain of 0.17 month to disability.

Additional adjustment for marital status or BMI, and exclusion of men with missing data on alcohol did not substantially modify the results (data not shown).

DISCUSSION

In this prospective cohort of workers, we reported a significant trend in the risk of work cessation because of death, disability, early retirement compared with the statutory age (55 years old) and early work cessation as the combination of the three causes with increasing daily alcohol consumption. Moreover, we estimated that compared with low consumers, each high or very high consumer lost on average the equivalent of 6.04 months out of the workforce between 50 and 60 years. Retirement before age 55 accounted for the major part of this loss (83%).

To our knowledge, this is the first study to simultaneously examine the association between daily alcohol consumption and the different causes of labour market exit. The few previous studies were mainly focused on disability retirement. In a study in the Finnish population, Salonsalmi et al. (2012) recently reported that those with the lowest risk of disability retirement were the moderate consumers, whereas non- and high consumptions (in terms of frequency or quantity) were associated with an increased risk of disability due to mental disorders. A Norwegian study (Skogen et al., 2012) observed that drinking problems measured by the CAGE questionnaire, but not mere alcohol consumption, were associated with disability retirement. Some older studies did not observe any association between alcohol consumption and disability retirement (Krause et al., 1997; Husemoen et al., 2004). In others, only non-consumers or former drinkers were consistently at higher risk of disability than low or moderate consumers (Samuelsson et al., 2013). Some studies also reported an increased risk of disability associated with increasing consumption levels (Upmark et al., 1997,  1999; Mansson et al., 1999), although the trend was not statistically tested. The comparison of these results with ours should be made cautiously. Indeed, the definition of high consumers was not consistent across studies and the patterns of alcohol consumption were probably different in these Northern European studies (WHO, 2014). In addition, laws regarding disability leave are country-specific (Council of Europe, 2002). More people are granted disability pension in Northern Europe than in France. Therefore, it is likely that some EDF-GDF employees who retired before age 55 in the present study would have been granted disability retirement if they had been working in Northern Europe. Our results extend thus the current knowledge about the dose-effect relationship between alcohol consumption and early retirement already reported in Northern European countries to Southern European countries in which alcohol drinking patterns are different and mostly based on daily consumption during meals.

The causes of disability were not available, but previous studies already investigated the potential differential effects of alcohol on various causes of disability. Recently, Salonsalmi et al. (2012) distinguished between mental and musculoskeletal causes of disability and reported that high alcohol consumption was more often associated with mental disability diagnoses, and non-consumption with all types of disability. These results suggest that alcohol and mental disorders are important factors behind disability retirement.

In addition, we estimated for the first time the burden of alcohol-related early retirement in the workforce of a Southern European country. We estimated that compared with low consumers, each high consumer could potentially gain an average 6.04 months of occupational activity if he drank the same as low drinkers. Although our adjusted estimations of time lost could also be attributable to other factors, our results strongly suggest that high but also moderate alcohol consumption has important deleterious consequences on the workforce. This is also the first analysis of the association between alcohol and early work cessation according to the different causes or early retirement. Even if HRs were higher for work cessation because of death or disability than retirement before age 55, death and disability are relatively rare and are a cause of retirement in only 3–10% of the cases. The predominant part of the loss of occupational activity is thus attributable to retirement before age 55. However, while we know that employees retired before age 55 had more often been hired at a low social position than other employees, we do not know why they stopped working before reaching the legal age. More studies simultaneously considering the different causes of early retirement are needed to better understand the impact of alcohol on work performances.

Strengths of our study include its prospective design from a large occupational cohort composed of white- and blue-collar employees from all regions of France, with long-term follow-up. Moreover, information on vital and occupational status throughout the whole career (mortality, disability and retirement) was comprehensive and accurate as it was directly obtained from the company administrative records. These records also enabled us to take into account early occupational characteristics of participants such as occupational status at hiring. In addition, we could perform adjustment for several potential confounders.

Our study has some limitations. First, analyses were restricted to men. However, studying the consequences of alcohol consumption in men is particularly interesting because the consumption range is generally larger in men than in women. Second, alcohol consumption was self-reported, so there might be some under-reporting of alcohol consumption levels. However, confidential questionnaires encourage fuller reports of drinking than face-to-face personal interviews (Gmel, 2000; Kraus and Augustin, 2001). Moreover, the questionnaires did not capture day-to-day variations of alcohol consumption, which could lead to a misestimation of daily alcohol consumption levels. However, the rate of heavy drinkers in the GAZEL cohort is close to the rate in the French general population measured by other surveys using different tools (Beck et al., 2007; Zins et al., 2011). Non-response may be a concern, since heavy drinkers are more likely to have missing data (Goldberg et al., 2001, 2006). In our population, the imputed values for alcohol were indeed mostly heavy drinkers. However, less than 1% of alcohol data were missing before imputation, and exclusion of men with missing data on alcohol did not substantially modify the results. In addition, our questionnaire could not distinguish between real non-drinkers and former heavy drinkers. That is why we chose to exclude these participants. Former studies described a J-curve relationship between non-consumption or abstinence, and disability retirement (Upmark et al., 1997, 1999), various disease incidences or mortality (Rehm et al., 2001; Corrao et al., 2004; Fillmore et al., 2007; Klatsky and Udaltsova, 2007). It has been hypothesized that this increased risk could arise from the combination of lifetime non-drinkers and abstainers (former heavy drinkers and subjects who abstain from drinking due to specific health problems) in the non-drinkers category (Fillmore et al., 2007). The elevated risk in non-drinkers could thus be the consequence of the ill health of former drinkers rather than a real protective effect of low or moderate alcohol consumption (taken as the reference category). Our questionnaire could also not specify the consumption frequency (weekly or monthly for example) and quantity in occasional drinkers. Moreover, the questionnaire did not contain a measure of binge drinking, which is thus likely to be included in the occasional consumption. The category of occasional consumers can thus be composed of very mixed drinkers. That is why we chose to exclude them. Nevertheless, binge drinking was probably not widespread in our population of French middle-aged men. In 2010, only 17% of French adults aged 45–54 years reported monthly heavy alcohol consumption compared with 29% of young adults (20–25 years) (Beck and Richard, 2013). Finally, our follow-up is relatively long and alcohol consumption may have changed over time. In the general French population, alcohol consumption has decreased over time (Beck et al., 2007; Spilka et al., 2011; WHO, 2014). Therefore, it could have been interesting to test the effect on work cessation of decreasing alcohol consumption among high consumers. However, our data on alcohol consumption were not suitable to perform these analyses.

In conclusion, our results provide evidence that increasing daily alcohol consumption is associated with increasing risk of early retirement for death, disability or retirement before age 55. These results suggest that high alcohol intake (but also moderate intake to a lesser extent) leads to an important social and economic prejudice for the workers, and presumably the company and society, namely in terms of lost occupational activity. Efforts should thus be made on diminution of alcohol consumption among daily drinkers, in order to limit the detrimental effect of this lifestyle. Further longitudinal studies, using repeated measures of alcohol consumption and taking its evolution into account, should be encouraged in order to evaluate the effects of decreasing alcohol consumption on work cessation.

ETHICAL APPROVAL

All procedures performed in studies involving human participants were in accordance with the ethical standards of the French authority for data confidentiality (Commission Nationale Informatique et Liberté # 105728) and by INSERM's Ethics Evaluation Committee (IRB0000388, FWA00005831). Informed consent was obtained from all individual participants included in the study.

AUTHORS' CONTRIBUTIONS

S.M., C.L. and M.Z. designed the study. S.M. and S.G. collated the data. S.M. and E.H. developed the statistical methods and undertook the analysis. S.M. made the first draft of the manuscript. All other authors contributed to interpretation of the data, provided background information and drafted and revised the manuscript.

FUNDING

The GAZEL Cohort Study was supported by CAMIEG, the French Institute of Health and Medical Research (INSERM), the French ‘Cohortes Santé TGIR Programme’, Agence nationale de la recherche (ANR), Agence française de sécurité sanitaire de l'environnement et du travail (AFSSET) and a grant from Lundbeck [grant number R13194LL to S.M.]. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

CONFLICT OF INTEREST STATEMENT

None declared.

ACKNOWLEDGEMENTS

The authors are indebted to all participants for providing the data used in the GAZEL Study. The author also thanks the GAZEL cohort study team responsible for overseeing data collection and EDF-GDF (Service des Etudes Médicales, Service Général de Médecine de Contrôle) and Diane Cyr for the English revision of the manuscript.

REFERENCES

Alwan
H
,
Viswanathan
B
,
Rousson
V
et al
. (
2011
)
Association between substance use and psychosocial characteristics among adolescents of the Seychelles
.
BMC Pediatr
 
11
:
85
.
Andersen
PK
. (
2013
)
Decomposition of number of life years lost according to causes of death
.
Stat Med
 
32
:
5278
85
.
Beck
F
,
Richard
J
. (
2013
)
Epidémiologie de l'alcoolisation
.
EMC - Endocrinologie-Nutrition
 
10
:
1
9
.
Beck
F
,
Guilbert
P
,
Gautier
A
. (
2007
)
Baromètre Santé 2005—Attitudes et Comportements de Santé.
INPES
.
Boden
JM
,
Fergusson
DM
. (
2011
)
Alcohol and depression
.
Addiction
 
106
:
906
14
.
Corrao
G
,
Bagnardi
V
,
Zambon
A
et al
. (
2004
)
A meta-analysis of alcohol consumption and the risk of 15 diseases
.
Prev Med
 
38
:
613
9
.
Council of Europe
. (
2002
)
Working Group on the Assessment of Person-Related Criteria for Allowances and Personal Assistance for People with Disabilities
.
Assessing Disabilities: Similarities and Differences. Integration of People with Disabilities
.
Fillmore
KM
,
Stockwell
T
,
Chikritzhs
T
et al
. (
2007
)
Moderate alcohol use and reduced mortality risk: systematic error in prospective studies and new hypotheses
.
Ann Epidemiol
 
17
:
S16
23
.
Gmel
G
. (
2000
)
The effect of mode of data collection and of non-response on reported alcohol consumption: a split-sample study in Switzerland
.
Addiction
 
95
:
123
34
.
Goldberg
M
,
Chastang
JF
,
Leclerc
A
et al
. (
2001
)
Socioeconomic, demographic, occupational, and health factors associated with participation in a long-term epidemiologic survey: a prospective study of the French GAZEL cohort and its target population
.
Am J Epidemiol
 
154
:
373
84
.
Goldberg
M
,
Chastang
JF
,
Zins
M
et al
. (
2006
)
Health problems were the strongest predictors of attrition during follow-up of the GAZEL cohort
.
J Clin Epidemiol
 
59
:
1213
21
.
Goldberg
M
,
Leclerc
A
,
Bonenfant
S
et al
. (
2007
)
Cohort profile: the GAZEL Cohort Study
.
Int J Epidemiol
 
36
:
32
9
.
Goldberg
M
,
Leclerc
A
,
Zins
M
. (
2015
)
Cohort profile update: The GAZEL Cohort Study
.
Int J Epidemiol
 
44
:
77
77g
.
Herquelot
E
,
Guéguen
A
,
Bonenfant
S
et al
. (
2011
)
Impact of diabetes on work cessation: data from the GAZEL cohort study
.
Diabetes Care
 
34
:
1344
9
.
Husemoen
LLN
,
Osler
M
,
Godtfredsen
NS
et al
. (
2004
)
Smoking and subsequent risk of early retirement due to permanent disability
.
Eur J Public Health
 
14
:
86
92
.
Jarl
J
,
Gerdtham
U-G
. (
2012
)
Does drinking affect long-term sickness absence? A sample selection approach correcting for employment and accounting for drinking history
.
Appl Econ
 
44
:
2811
25
.
Johansson
E
,
Böckerman
P
,
Uutela
A
. (
2009
)
Alcohol consumption and sickness absence: evidence from microdata
.
Eur J Public Health
 
19
:
19
22
.
Klatsky
AL
,
Udaltsova
N
. (
2007
)
Alcohol drinking and total mortality risk
.
Ann Epidemiol
 
17
:
S63
7
.
Koutra
K
,
Papadovassilaki
K
,
Kalpoutzaki
P
et al
. (
2012
)
Adolescent drinking, academic achievement and leisure time use by secondary education students in a rural area of Crete
.
Health Soc Care Community
 
20
:
61
9
.
Kraus
L
,
Augustin
R
. (
2001
)
Measuring alcohol consumption and alcohol-related problems: comparison of responses from self-administered questionnaires and telephone interviews
.
Addiction
 
96
:
459
71
.
Krause
N
,
Lynch
J
,
Kaplan
GA
et al
. (
1997
)
Predictors of disability retirement
.
Scand J Work Environ Health
 
23
:
403
13
.
Lenneman
J
,
Schwartz
S
,
Giuseffi
DL
et al
. (
2011
)
Productivity and health: an application of three perspectives to measuring productivity
.
J Occup Environ Med
 
53
:
55
61
.
Magnus
A
,
Cadilhac
D
,
Sheppard
L
et al
. (
2012
)
The economic gains of achieving reduced alcohol consumption targets for Australia
.
Am J Public Health
 
102
:
1313
9
.
Mansson
NO
,
Råstam
L
,
Eriksson
KF
et al
. (
1999
)
Alcohol consumption and disability pension among middle-aged men
.
Ann Epidemiol
 
9
:
341
8
.
Mounteney
J
,
Haugland
S
,
Skutle
A
. (
2010
)
Truancy, alcohol use and alcohol-related problems in secondary school pupils in Norway
.
Health Educ Res
 
25
:
945
54
.
Putter
H
,
Fiocco
M
,
Geskus
RB
. (
2007
)
Tutorial in biostatistics: competing risks and multi-state models
.
Stat Med
 
26
:
2389
430
.
Rehm
J
,
Greenfield
TK
,
Rogers
JD
. (
2001
)
Average volume of alcohol consumption, patterns of drinking, and all-cause mortality: results from the US National Alcohol Survey
.
Am J Epidemiol
 
153
:
64
71
.
Rosthøj
S
,
Andersen
PK
,
Abildstrom
SZ
. (
2004
)
SAS macros for estimation of the cumulative incidence functions based on a Cox regression model for competing risks survival data
.
Comput Methods Programs Biomed
 
74
:
69
75
.
Salonsalmi
A
,
Laaksonen
M
,
Lahelma
E
et al
. (
2009
)
Drinking habits and sickness absence: the contribution of working conditions
.
Scand J Public Health
 
37
:
846
54
.
Salonsalmi
A
,
Laaksonen
M
,
Lahelma
E
et al
. (
2012
)
Drinking habits and disability retirement
.
Addiction
 
107
:
2128
36
.
Samuelsson
Å
,
Ropponen
A
,
Alexanderson
K
et al
. (
2013
)
A prospective cohort study of disability pension due to mental diagnoses: the importance of health factors and behaviors
.
BMC Public Health
 
13
:
621
.
Schafer
JL
,
Graham
JW
. (
2002
)
Missing data: our view of the state of the art
.
Psychol Methods
 
7
:
147
77
.
Skogen
JC
,
Knudsen
AK
,
Mykletun
A
et al
. (
2012
)
Alcohol consumption, problem drinking, abstention and disability pension award. The Nord-Trøndelag Health Study (HUNT)
.
Addiction
 
107
:
98
108
.
Spilka
S
,
Tovar
M
,
Beck
F
et al
. (
2011
)
Les Niveaux D'usage Des Drogues En France En 2010, Exploitation Des Données Du Baromètre Santé
.
INPES
.
Thavorncharoensap
M
,
Teerawattananon
Y
,
Yothasamut
J
et al
. (
2010
)
The economic costs of alcohol consumption in Thailand, 2006
.
BMC Public Health
 
10
:
323
.
Upmark
M
,
Hemmingsson
T
,
Romelsjö
A
et al
. (
1997
)
Predictors of disability pension among young men
.
Eur J Public Health
 
7
:
20
8
.
Upmark
M
,
Möller
J
,
Romelsjö
A
. (
1999
)
Longitudinal, population-based study of self reported alcohol habits, high levels of sickness absence, and disability pensions
.
J Epidemiol Community Health
 
53
:
223
9
.
WHO
. (
2010
)
International Guide for Monitoring Alcohol Consumption and Related Harm
.
WHO
. (
2014
)
WHO | Global Status Report on Alcohol and Health 2014
.
Zins
M
,
Guéguen
A
,
Kivimaki
M
et al
. (
2011
)
Effect of retirement on alcohol consumption: longitudinal evidence from the French Gazel cohort study
.
PLoS One
 
6
:
e26531
.