Low organisational justice and heavy drinking: a prospective cohort study

Objectives: To investigate whether low perceived organisational injustice predicts heavy drinking among employees. Methods: Data from a prospective occupational cohort study, the 10-Town Study, on 15 290 Finnish public sector local government employees nested in 2432 work units, were used. Non-drinkers were excluded. Procedural, interactional and total organisational justice, heavy drinking (⩾210 g of absolute alcohol per week) and other psychosocial factors were determined by means of questionnaire in 2000–2001 (phase 1) and 2004 (phase 2). Multilevel logistic regression analyses taking into account the hierarchical structure of the data were conducted and adjustments were made for sex, age, socio-economic status, marital status, baseline heavy drinking, psychological distress and other psychosocial risk factors such as job strain and effort/reward imbalance. Results: After adjustments, participants who reported low procedural justice at phase 1 were approximately 1.2 times more likely to be heavy drinkers at phase 2 compared with their counterparts reporting high justice. Low perceived justice in interpersonal treatment and low perceived total organisational justice were associated with increased prevalence of heavy drinking only in the model adjusted for sociodemographics. Conclusions: This is the first longitudinal study to show that low procedural justice is weakly associated with an increased likelihood of heavy drinking.

An increasing number of empirical studies suggest that the extent to which employees are treated justly in the workplace may affect their health. Organisational justice involves a procedural component and an interactional component. Procedural justice relates to whether decisionmaking procedures are consistently applied, suppress bias, and are accurate, correctable and ethical. 1 The interactional component of organisational justice is concerned with whether people believe that their supervisor considers their viewpoints, shares information concerning decisionmaking, and treats individuals fairly and in a truthful manner. 2 The health effects of organisational justice have been examined in several studies based on large well-established British and Finnish employee cohorts. A lower level of organisational justice has been associated with increased mental health problems, greater medically-certified absence from work, incident coronary heart disease and higher self-reported morbidity, irrespective of other workrelated psychosocial factors. [3][4][5][6][7][8][9][10] The mechanisms underlying these associations are not well known, although evidence has begun to emerge. In principle, organisational justice, like other psychosocial factors, could be related to ill health directly through physiological stress mechanisms and indirectly through behavioural risk factors such as heavy drinking. 11 According to Elovainio et al, 12 a low level of justice was related to increased lowfrequency band systolic arterial pressure variability and reduced high-frequency heart rate variability, suggesting that cardiac dysregulation is one of the stress mechanisms through which low perceived justice increases the risk of health problems.
Evidence of the relationship between organisational justice and behavioural risk factors is scarce. Low procedural justice was cross-sectionally associated with high alcohol consumption in male hospital employees. 3 However, the temporal aspects of this association were unclear and it is not known whether the association between a low level of justice and heavy drinking was independent of other psychosocial work factors such as job strain 13 14 or effort/reward imbalance (ERI), 14 15 which may also increase alcohol consumption.
Both procedural and interactional justice deal with how supervisors relate to their employees. This relationship may be crucial for mental health. 4 Research on depression suggests that life events in combination with humiliation and devaluation are more likely to lead to depression than life events alone. 16 Humiliation of employees may well be a component of low interactional justice. It is also possible that poor relationships between supervisors and employees are a result of the supervisors being treated badly themselves by their superiors. Such a hierarchy of poor relationships may reflect a malfunctioning organisation in which low procedural justice may be either a cause or a consequence. A low procedural component also refers to a failure of the procedures used in the workplace to create clarity and consistency for the employees around decisions at work. 17 Procedural justice can be related to heavy drinking through depression and other mental health problems, but evidence supporting this is indirect. Procedural justice predicted doctor-diagnosed depression in a longitudinal study 4 and has also been linked with minor psychiatric morbidity both cross-sectionally 7 and longitudinally. 4 In the Whitehall II study, a favourable change in justice reduced the immediate risk of psychiatric morbidity, while an adverse change increased the immediate and longer-term risk. 10 Mental health problems, depression in particular, can increase substance use. 18 19 Data from the 10-Town Study, an ongoing large-scale prospective occupational cohort study, have been used to examine the longitudinal association between perceived organisational justice and heavy drinking. The purpose of our study was to examine whether levels of perceived procedural, interactional and total organisational justice (a combined index of procedural and interactional justice) were associated with subsequent heavy drinking independently of other psychosocial characteristics of the work environment. We used multilevel modelling to take account of the fact that individual employees are nested in social units comprised of workplaces. 20

Study design and study population
The ongoing prospective Finnish 10-Town cohort study focuses on all personnel employed by ten local governments. The baseline data on organisational justice, alcohol consumption and other factors were collected in 2000 and 2001 through selfadministered questionnaires from 32 299 women and men. The response rate was 67%. Of these employees, 24 196 were still working for the same employers at the time of the follow-up survey in 2004. The response rate at phase 2 for individuals who remained employed by the target organisations at the phase 2 survey and who had responded at phase 1 was 79% (n = 19 077). The average length of follow-up was 3.6 years.
This study focused on those 15 290 participants (11 745 women and 3545 men) with data on organisational justice at baseline, who reported alcohol use at follow-up, who worked in work units of at least three employees, and who had no missing values for any of the covariates. Non-drinkers (n = 2437) had been excluded because they represent the two very different populations of teetotallers and ex-drinkers. Ex-drinkers include recovering alcoholics and people with diseases that are worsened by drinking, whereas teetotallers may have health problems, or religious or moral reasons for abstaining. 21 The final dataset included individuals (employees) nested within 2432 work units in 10 towns. The work unit of each respondent was identified from the employers' records based on a five-level organisational hierarchy classification. Work unit was the lowest level in the organisational hierarchy.
Any differences from the eligible population were small. The figures for the final cohort (baseline) compared with the eligible population (n = 47 351) were mean age 44.8 years compared with 44.5 years and 77% women compared with 72%.
Approval from the ethics committee of the Finnish Institute of Occupational Health was obtained for the study.

Assessment of organisational justice
Organisational justice was measured using the scale developed by Moorman. 2 The procedural justice scale (seven items, Cronbach's a = 0.91 at baseline; 4-year test-retest correlation coefficient r = 0.53) measured the degree to which respondents considered the procedures used in the workplace were designed to collect accurate information necessary for making decisions, to provide opportunities for appealing or challenging the decision, to generate standards so that decisions could be made with consistency, and to hear the concerns of all those affected by the decision.
The interactional justice scale 2 (six items, Cronbach's a = 0.92 at baseline; 4-year test-retest correlation coefficient r = 0.44) measured whether respondents thought that their supervisors were able to suppress personal biases, to treat subordinates with kindness and consideration, and to take steps to deal with subordinates in a truthful manner.
In both scales responses were given on a five-point scale ranging from 1 = strongly disagree to 5 = strongly agree.
In addition to the scales of procedural and interactional organisational justice, we constructed a scale of total organisational justice by calculating a mean score of these two scales (13 items, Cronbach's a = 0.92 at baseline; 4-year test-retest correlation coefficient r = 0.53). The correlation between the two subscales was 0.49 (p,0.001). The intra-class correlation (ICC) of total justice was 15%. ICC evaluates between-group variance relative to total (between and within) variance. 22 The within-group (inter-rater) agreement index (r wg ) can be used to measure the agreement between workers in the same work unit regarding their perception of organisational justice. An r wg index value of 0.70 or higher is perceived as acceptable agreement. 23 In our sample, the average deviation of an individual's perception of total organisational justice from the mean level of her/his work unit was 0.89, indicating a significant homogeneity in the perceptions of organisational justice within a work unit.
The participants were divided into three groups based on the distribution of the organisational justice scores. The bottom third indicated a low level, the middle third a medium level, and the top third a high level of justice.

Assessment of heavy drinking
Heavy drinking was assessed at baseline and at follow-up. The questions concerning drinking habits were framed as follows: ''Have you ever consumed at least one glass of some alcoholic drink?'' and ''How much do you consume of the following alcoholic drinks on average (beer per week; wine or other mild drinks per week; spirits per month)?''. The volume of beer, wine and spirits consumed was calculated as grams of absolute alcohol per week. 24 One unit of pure alcohol (12 g) is equal to a 12 cl glass of wine, a single 4 cl measure of spirits or a 33 cl bottle of beer. A dichotomous variable was created to represent heavy drinking, with a cut-off point corresponding to the weekly consumption of 210 g or more of absolute alcohol. 25 All other respondents reporting alcohol intake below these limits (light-moderate drinkers) were placed in the reference category.

Assessment of covariates at baseline
The demographic baseline characteristics used in the analysis included sex, age, socio-economic status (SES) based on ISCO-88 classification (managers and professionals, associate professionals, clerks, service workers and manual workers) and marital status (married or cohabiting vs single, divorced or widowed). Sex, age and occupational status were obtained from the employers' registers and age was grouped into three categories (19-34, 35-50 and 51-62 years).
Psychological distress was measured by the 12-item version of the General Health Questionnaire (Cronbach's a = 0.89). 26 Individuals scoring >4 were estimated to have high mental distress.
Assessment of job strain was based on a modified Job Content Questionnaire 27 comprised of the job demand scale (three items, Cronbach's a = 0.75 at baseline) and job control scale (nine items, Cronbach's a = 0.82 at baseline). The responses were given on a Likert scale of 1 = very little to 5 = very much. The total scores for each of the two constructs were computed. To construct the job strain measure, the means of job demand scores were subtracted from the means of job control scores. 6 28-30 All participants were divided into three groups based on the distribution of this score.
Effort in work was measured by the following question: ''How much do you feel you invest in your job in terms of skill and energy?''. Rewards in return for work were assessed by three questions on (1) income and job benefits, (2) recognition and prestige, and (3) personal satisfaction (Cronbach's a = 0.63 at baseline). 31 The response format for all questions was a fivepoint Likert scale ranging from 1 = very little to 5 = very much. The indicator of ERI was obtained by calculating the ratio between the response score in the effort scale and the mean response score in the reward scale. The resulting quotient was divided into thirds to indicate low, medium and high ERI.

RESULTS
The baseline study characteristics are shown in table 1. Most participants were 35-50 years old and 76.9% were married. The most common SES group was managers and professionals (36.2%), followed by service workers (23.4%).

Statistical analysis
Because the participants were nested within work units, we applied multilevel logistic regression analysis to study the associations between perceived organisational justice at baseline and heavy drinking at follow-up among current drinkers. This method accounts for the non-independence of observations within groups. The results of the modelling are presented as odds ratios (ORs) and their 95% confidence intervals (CIs). The hypothetically most favourable condition (highest level of justice) was selected as a reference category in each indicator of perceived organisational justice. The analysis was conducted in three steps. In the first model, only sex, age, SES and marital status were controlled for. The second model also included baseline heavy drinking and psychological distress. The third model was additionally adjusted for job strain and ERI.
The statistical significance of interactions between justice variables and sex were tested by including interaction terms.
All models were fitted using the SAS version 9.1.3 program package (SAS Institute, Cary, NC, USA). The multilevel analyses were performed by using the Glimmix procedure.
In the follow-up survey, 11.9% of the participants were heavy drinkers who consumed 210 g or more of absolute alcohol a week. Heavy drinking was more frequent among men (25.4%) than among women (7.9%). The percentage of heavy drinkers was higher in the two oldest than in the youngest age group. Moreover, heavy drinking was more frequent among managers and professionals (14.1%) and manual workers (15.9%) than among other SES groups (p value for differences in all cases ,0.001 except with age where p = 0.011; data not shown).
The level of perceived organisational justice in the final cohort corresponded well with that among the respondents of the first survey. The mean scores for total organisational justice were 3.38 (SD 0.77, range 1-5) for the first and 3.38 (SD 0.79, range 1-5) for the latter group. Therefore bias due to selection procedures or sample attrition seems unlikely.
Women reported high interactional justice slightly more often than men. Workers in the oldest and the youngest age groups experienced higher levels of procedural and total justice more often than workers in the middle age group. High justice was more frequent in managers and professionals as well as in service workers than in other SES groups. As expected, high job strain and high ERI were connected with low organisational justice (p value for difference ,0.001 in all cases in x 2 two-tailed test; data not shown).
Only 4.4% of participants were exposed to high procedural but low interactional justice, whereas 6.6% were exposed to high interactional but low procedural justice. However, 19.4% of the participants were exposed to high procedural and high interactional justice.
There were no significant differences in the means of perceived procedural, interactional and total justice between the baseline and follow-up surveys. At baseline, the means (SD) of procedural, interactional and total organisational justice were 3.15 (0.86), 3.64 (0.92) and 3.38 (0.77), respectively. In the follow-up survey, the corresponding figures were 3.11 (0.88), 3.66 (0.93) and 3.37 (0.79), respectively (p,0.001). The range was 1-5 in all cases.
A total of 734 participants were drinkers at baseline but nondrinkers at follow-up, and 687 participants were heavy drinkers at baseline but non-heavy drinkers at follow-up. The mean of perceived total organisational justice in participants who were drinkers at baseline but teetotallers at follow-up was 3.45 (SD 0.78; range 1.08-5.00).
There were differences in heavy drinking between work units. In the multilevel model, the component variance of work unit was significant (var 0.039, SE 0.005, p,0.001). In 48% of the work units there were no heavy drinkers, while in 10% of work unit at least a third of the participants were heavy drinkers. Table 2 shows the associations between perceived organisational justice variables at baseline and heavy drinking at followup. After adjustment for sex, age, SES, marital status, heavy drinking at baseline, psychological distress, job strain and ERI, low perceived procedural justice was associated with approximately 16% higher odds of heavy drinking. Low perceived interactional justice (OR 1.19, 95% CI 1.05 to 1.34) and low perceived total justice (OR 1.23, 95% CI 1.09 to 1.40) were associated with a higher likelihood of heavy drinking in the model adjusted for sociodemographics but became non-significant after further adjustments. We also conducted the analyses using the higher (250 g) and lower (190 g) cut-off points of heavy drinking. 14 The results were in the same direction as those with the chosen cut-off point. In the fully adjusted model, the OR for heavy drinking in the lowest tertile of perceived procedural justice was 1.15 (95% CI 0.93 to 1.41) with the cut-off point of 250 g and 1.21 (95% CI 1.05 to 1.38) with the cut-off point of 190 g (data not shown).
We also conducted SES-stratified analyses for perceived procedural justice. The results were in the same direction in all SES groups, although low perceived procedural justice was significantly associated with a higher likelihood of heavy drinking only among managers and professionals (OR 1.28, 95% CI 1.01 to 1.62) (data not shown).
The effect of an interaction between each indicator of perceived organisational justice and sex on heavy drinking at follow-up was tested by including the corresponding interaction term in regression models that already included the main effects. No sex interactions were found (p values for interaction: procedural justice p = 0.184, interactional justice p = 0.802 and total justice p = 0.772).
To test the possibility of reversed causality, we examined whether baseline heavy drinking predicted perceived procedural justice at follow-up after adjustment for sex, age and baseline procedural justice. We did not find significant association. Thus, reverse causation is unlikely to explain the observed association between perceived procedural justice and subsequent heavy drinking (data not shown).

DISCUSSION
Research on organisational justice provides insight into an often forgotten aspect of the impact of work organisation on health. So far, only a few studies have examined the association between organisational justice and health-risk behaviours. This prospective study showed that low levels of perceived procedural organisational justice predicted heavy drinking in women and men in a large well characterised cohort of Finnish local government employees. The associations were not accounted for by baseline characteristics such as sex, age, marital status, SES, baseline heavy drinking, psychological distress, job strain and effort/reward imbalance. The odds ratios were rather small but still potentially important from a public health point of view as heavy alcohol consumption appears to increase morbidity and mortality risk, whereas moderate alcohol consumption has been shown to be associated with a decreased risk. [32][33][34] Conditions of low control and low organisational justice can occur simultaneously in the same work environment. 35 This was also the case in the present study, as 45.8% of participants with high job strain and 47.2% of participants with high ERI were also in the lowest tertile of procedural justice (p,0.001 in both cases). However, employees who have control over their jobs may still experience that their supervisor is unreliable, inconsiderate and untruthful. Moreover, favourable procedural justice evaluation may act as a mediating mechanism through which a high degree of job control protects employees against the development of strain symptoms. 35 In our study, heavy drinking was predicted by the procedural rather than the interactional component of perceived justice. This is in accordance with earlier evidence on other health outcomes 3 7 ; previous research reports less consistent health effects for interactional justice than for procedural justice. This suggests that a low level of justice in the work environment characterised by unjust organisational policies, practices and procedures is a greater risk to health and health behaviours than unfair treatment from an immediate supervisor.
Our findings suggest that the concept of organisational justice adds to prior research based on the established theoretical models of psychosocial risk factors. This view is also supported by studies on other outcomes, such as psychiatric disorders, 4 10 self-rated health, 5 coronary heart disease, 6 cardiac dysregulation 9 and smoking. 36 Earlier evidence on the association between psychosocial work-related factors and alcohol use is mixed, is mostly crosssectional, and mostly relates to the job strain model. The job strain model postulates that job strain emerges from the combination of high job demands and low job control generating increased risk of ill health and adverse health behaviours. 37 Some studies have found job strain variables (high job strain, high demands, low control or passive jobs) to be associated with alcohol consumption, 38 alcohol abuse-dependence 39 or alcohol use. 13 However, these findings are contradicted by several studies in which either no association was Table 2 Associations of perceived procedural, interactional and total organisational justice at baseline with heavy drinking at follow-up Adjusted odds ratios and their 95% confidence intervals (95% CIs) from multilevel logistic regression models are shown (n = 15 290). ERI, effort/reward imbalance; SES, socio-economic status.
found between alcohol use and job strain variables, [40][41][42][43] or the directionality of the associations was partly contrary to what was expected. 14 When also taking into account work stressors other than job strain, research in general supports a weak association of work-related stressors with elevated alcohol consumption and problem drinking. 44 45 A more recent model of occupational stress, the ERI model, is based on the assumption that the mismatch between high effort spent and low rewards (financial, self-esteem and social) gained from work, can result in an increased risk of ill health. 46 The evidence of the association between ERI and alcohol use is limited, but the results from the Whitehall II cohort indicate an increased risk of alcohol dependence in men with high ERI. 15 Low effort in both women and men and low rewards in men have also been associated with heavy drinking. 14 The primary focus of the job strain model is on task level characteristics. 5 It is possible that organisational justice captures more basic elements of social structure. Moreover, unlike the models of job strain and ERI, the justice approach is directly focused on managerial treatment and managerial procedures. Therefore, it covers all types of unfairness, not only that arising from disproportionate demands in relation to job control 6 or the specific exchange process between efforts and rewards. Employees may be affected not only by rewards as such but also by procedures used to determine how these rewards will be distributed. 7 Low organisational justice has been associated with occupational strain 34 and with negative emotional reactions. 47 Both of these have been associated with unhealthy patterns of cardiovascular and immunological response as well as with health problems. 48 The workplace stress paradigm suggests that employee alcohol use may represent a strategy to cope with negative emotions resulting from exposure to adverse work conditions. 44 Individuals have been suggested to use alcohol to reduce negative mood states. 49 Therefore, low organisational justice can represent a source of stress and negative emotional reactions at work that can potentially contribute to health problems 34 and health-risk behaviours such as heavy drinking. In their longitudinal study, Elovainio et al 8 found that the effect of low organisational justice on minor psychiatric morbidity, such as depression and self-reported health, was mediated by sleeping problems. Both depression and sleeping problems have been related to elevated alcohol consumption in previous studies. [50][51][52][53] However, the observed association between low organisational and heavy drinking was weak. This is in line with the suggestion by Frone 44 that although most adults consume alcohol, it is unlikely that most workers use alcohol to cope with adverse work conditions. Many other coping mechanisms relieve the negative emotions resulting from work stressors more effectively and have fewer negative side effects. On the other hand, people have numerous non-work-related reasons for their heavy drinking. The features of procedural justice may also reflect social integration and social support. 35 Even if strictly speaking it is not known where heavy drinking occurs, our measure assessed most likely heavy drinking outside of work. A recent national survey in the US showed that only 7% of workers were estimated to use alcohol during the workday. 54 If our results reflect a causal association and the impact that the work environment may have on out-of-work behaviours, this is of practical importance.

Study strengths and limitations
Our evidence was based on a well-characterised cohort, prospective design and repeated measurements. The response rate was satisfactory and the data were fairly representative of both manual and non-manual occupations. The large sample size allowed the detection of weak effects. Moreover, multilevel modelling was used and we adjusted for multiple covariates, thereby minimising confounding bias. Although the possibility of confounding by an unknown factor can never be excluded, a major bias in our study is unlikely. The contribution of procedural justice to heavy drinking largely persisted after controlling for sociodemographic factors, minor psychological distress and psychosocial factors. The longitudinal design allowed us to address the issue of reverse causality. No evidence was found to support the hypothesis that a perception of low procedural justice would be a consequence of being a heavy drinker.
However, interpretations of these findings should take into consideration study limitations. First, this study assessed heavy drinking and organisational justice with self-reports. It is well know that self-report data on alcohol use can be subject to under-reporting. 55 This can lead to underestimation of the real association between organisational justice and heavy drinking. Moreover, under-reporting of alcohol use may be associated with organisational justice. If exposure to low organisational justice at baseline increases under-reporting of alcohol use more at baseline than at follow-up, it might lead to an apparent increase of drinking among those exposed to low organisational justice and inflate the association between organisational justice and subsequent alcohol use.
As justice was self-reported, the assessment may reflect both the respondent and the work environment. The individual perception is assumed to be essential for organisational justice to affect health. 20 However, previous studies have suggested that self-reported justice levels may accurately reflect organisational reality 56 57 and both individual-level justice scores and more objective ecological scores have been shown to be predictive of mental health. 8 Indeed, people may also take into account the experiences of others when they form justice judgements. 58 Second, our alcohol consumption measure did not differentiate between escapist and social drinking. Heavy drinking can be problematic, but it may also reflect a more socially oriented non-escapist and not necessarily problematic type of behaviour. 59 Third, we cannot rule out the possibility that heavy drinking and low perceived organisational justice could both be products of an unmeasured third factor. For example, personality characteristics such as hostility as well as other factors such as mood, depression and non-work-related risks, may affect both the perception of justice and heavy drinking. However, previous studies on health-related personality traits, such as hostility and anxiety, do not indicate a strong relationship between personality and justice perceptions. 60 Fourth, the response rates of our surveys were satisfactory, but attrition was slightly higher among heavy drinkers. Such selection may have caused a healthy worker effect and thus attenuated the association between organisational justice and heavy drinking.
Fifth, the standard measure of ERI was not available in this study. However, studies using both original and proxy measures have found support for the ERI model regardless of the measure. 61 Nevertheless, there is a possibility that our crude measure did not fully capture the ERI model and also the Cronbach a for the rewards scale was rather moderate.
Sixth, we could not assess the effect of low organisational justice exposure duration, since the length of the period in which the worker had been exposed to low organisational justice by baseline was unknown to us.
Finally, as our evidence was based on a female-dominated cohort of public sector workers, further research is needed on other occupational groups and in other countries. Further research is also needed to directly investigate the mechanisms that may underlie the effects of low procedural justice on health-risk behaviours such as heavy drinking.

CONCLUSIONS
In this study, we found a 16% excess likelihood of heavy drinking in employees with low procedural justice perceptions after 4 years of follow-up. Our results on heavy drinking offer further support for the previously reported findings that low organisational justice can pose health risk for employees not attributable to other work-related psychosocial factors. bmjupdates+ bmjupdates+ is a unique and free alerting service, designed to keep you up to date with the medical literature that is truly important to your practice. bmjupdates+ will alert you to important new research and will provide you with the best new evidence concerning important advances in health care, tailored to your medical interests and time demands.
Where does the information come from? bmjupdates+ applies an expert critical appraisal filter to over 100 top medical journals A panel of over 2000 physicians find the few 'must read' studies for each area of clinical interest Sign up to receive your tailored email alerts, searching access and more… www.bmjupdates.com