Workers drinking patterns: the impact on absenteeism in the Australian work-place

Transcription

1 RESEARCH REPORT doi: /j x Workers drinking patterns: the impact on absenteeism in the Australian work-place Ann M. Roche 1, Ken Pidd 1, Jesia G. Berry 2 & James E. Harrison 2 National Centre for Education and Training on Addiction, Flinders University, Adelaide, Australia 1 and Research Centre for Injury Studies, Flinders University, Adelaide, Australia 2 ABSTRACT Aims To examine the relationship between Australian workers patterns of alcohol consumption and absenteeism. Design A secondary analysis of the 2001 National Drug Strategy Household Survey data. Setting Australia Participants A total of workers aged 14 years. Measurements Alcohol consumption levels associated with National Health and Medical Research Council (NHMRC) guidelines for short- and long-term harm were identified and compared with self-reported measures of absenteeism due to alcohol use and due to any illness/injury. Findings More than 40% of the Australian work-force consumed alcohol at risky or high-risk levels at least occasionally. High-risk drinkers were up to 22 times more likely to be absent from work due to their alcohol use compared to low-risk drinkers. Short-term high-risk drinkers were also significantly more likely to be absent from work due to any illness or injury than employed low-risk drinkers. Young employees and males were more likely to report alcohol-related absenteeism compared to older workers and females. Conclusions The relationship between workers alcohol consumption patterns and absenteeism is more substantial than previously recognized or documented. Alcohol-related absenteeism is not restricted to small numbers of chronic heavy drinkers, but also involves the much larger number of risky non-dependent drinkers who drink less frequently at risky levels. To improve workers health and wellbeing and enhance productivity and economic prosperity, appropriate education, prevention and policy strategies are warranted and necessitate revision of previously narrow approaches undertaken with work-place programmes. Keywords Absenteeism, alchohol, Australia, drinking patterns, survey, workforce, work-place. Correspondence to: Ken Pidd, National Centre for Education and Training on Addiction, Flinders University, GPO Box 2001, Adelaide, SA 5001, Australia. ken.pidd@flinders.edu.au Submitted 30 March 2007; initial review completed 4 July 2007; final version accepted 10 January 2008 INTRODUCTION Much of the large body of international research literature on alcohol and the work-place concerns prevalence of use in the work-place and factors contributing to such use. Less research concerns workers overall patterns of drinking and work-place outcomes. While a consistent relationship between alcohol consumption and absenteeism has been established [1 5], the precise nature of this relationship remains unclear. In particular, virtually nothing is known about the profiles of risky drinkers and their levels of absenteeism. Moreover, few investigations have utilized large nationally representative samples from which generalizable findings can be extrapolated. Australian research on the impact of workers consumption patterns on absenteeism is scarce [6,7]. Two previous studies examined alcohol-related absenteeism in specific industries [8,9]. Apart from our recent work on the cost of alcohol-related absenteeism [10], only two studies have examined alcohol-related absenteeism in the whole Australian work-force [11,12]. Hagen et al. [12] used Australian Bureau of Statistics (ABS) National Health Survey (NHS) data to examine the relationship between risk levels of alcohol consumption, as defined by 1992 National Health and Medical Research Council (NHMRC) recommendations [13] (consistent with long-term harm risk levels shown in Table 1), and self-reported absenteeism. They found little difference in absenteeism between low-risk and hazardous/medium-risk-level drinkers (8.4% and 8.1%, respectively), but a larger percentage of harmful/highlevel drinkers (10.3%) reported being absent. They neither reported statistical significance nor controlled for covariates (e.g. age and gender). Bush & Wooden [11],

2 Alchohol and absenteeism in Australia 739 Table 1 National Health and Medical Research Council Australian alcohol guidelines [14]. Low risk (standard drinks*) Risky (standard drinks*) High risk (standard drinks*) Risk of short-term (acute) harm Males Up to or more (on any 1 day, no more than 3 days per week) (on any 1 day) (on any 1 day) Females Up to or more (on any 1 day, no more than 3 days per week) (on any 1 day) (on any 1 day) Risk of long-term (chronic) harm Males Up to or more (on an average day) (on an average day) (on an average day) Up to or more (overall weekly level) (overall weekly level) (overall weekly level) Females Up to or more (on an average day) (on an average day) (on an average day) Up to or more (overall weekly level) (overall weekly level) (overall weekly level) *A standard drink equals 10 g (12.5 ml) of alcohol. using the same ABS data set and alcohol guidelines, found that employees drinking at harmful/high levels (11.2%) were significantly (P < 0.05) more likely to report being absent, compared to low-risk (9.0%) and hazardous/medium-risk (8.4%) drinkers. After adjusting for age and gender, harmful/high-risk drinkers were approximately 1.2 times more likely than other employees to be absent from work. Current Australian alcohol guidelines [14] focus on patterns of consumption associated with risk of harm in the long term (chronic harm) and in the short term (acute harm) (see Table 1). Guidelines for risk of harm in the long term are similar to international alcohol consumption guidelines, such as those determined by the World Health Organization [15]. They focus on mean levels of consumption associated with chronic harms that result from regular heavy use over extended periods of time (e.g. heart disease, liver cirrhosis, dementia, etc.). Risk of chronic harm rises with regular average consumption. Guidelines for risk of harm in the short term focus on levels of consumption during single drinking occasions associated with acute harms (e.g. injury, accidents, raised blood pressure, stroke, etc.). As consumption on a single drinking occasion increases, so does risk of acute harms. Drinking at, or above, the short-term risky or high-risk levels outlined in Table 1 increases the probability of acute health and social problems, including injury or death [14,16]. Measures of both short- and long-term risk of harm may provide a more accurate indication of the relationship between alcohol use and absenteeism than indicators of long-term risk alone. Most research to date has employed an indirect extrapolation method to establish an inferred association between consumption levels and absenteeism derived by asking respondents to report days absent from work due to any illness or injury, categorizing respondents according to consumption risk level (i.e. abstainer, low risk, hazardous/risky, harmful/high risk) and comparing overall absenteeism rates for each risk category. Except for Pidd et al. [10], no other studies identify absences associated directly with alcohol use. This paper presents a secondary analysis of selected data from the 2001 National Drug Strategy Household Survey (NDSHS) and is part of a suite of papers that provide the most comprehensive data available to date on a nationally representative sample of Australian workers alcohol consumption patterns [10,17]. Measures included in this survey allowed for direct assessment of alcohol-related absenteeism, as well as absenteeism due to any illness/injury. The NDSHS data set also provided details of respondents short- and long-term patterns of alcohol consumption. METHODS Design The 2001 NDSHS collected data on awareness, attitudes and behaviour relating to alcohol, tobacco and illicit drug use from Australians aged 14 years or over. Data were collected from July to December 2001 using three methods: the drop-and-collect method (n = ); faceto-face method (n = 2055); and computer-assisted telephone interviews (CATI) (n = 2040) [18]. Response rates were 51%, 39% and 49%, respectively. A multi-stage stratified sampling methodology was utilized and data were weighted by age, sex and geographical region to be representative of the total population of Australia [19].

3 740 Ann M. Roche et al. Measures of alcohol consumption Workers alcohol consumption was classified according to short- and long-term risk levels determined by the 2001 NHMRC guidelines (Table 1), using a method identical to the original analysis of the NDSHS [20]. An abstainer was defined as a person who had never had a full serve of alcohol or a person who had consumed a full serve of alcohol but not in the 12 months prior to the survey. A recent drinker was defined as a person who had consumed a full serve of alcohol in the past 12 months. Recent drinkers were classified into short- and long-term risk categories of alcohol consumption outlined in Table 1 (i.e. low-risk, risky and high-risk) utilizing a graduated-frequency (GF) method. The GF method asked respondents to record the frequency and quantity of standard drinks consumed over the past 12 months. The GF method is described elsewhere [17,21]. For short-term risk levels, respondents were classified into mutually exclusive groups according to frequent (at least weekly), infrequent (at least monthly) or occasional (at least yearly) short-term risky or high-risk consumption. Absenteeism The 2001 NDSHS asked respondents to report the number of days missed from work, school, Technical and Further Education (TAFE) or university due to (i) their personal use of alcohol in the 3 months prior to the survey; or (ii) any illness or injury in the 3 months prior to the survey. As the NDSHS absenteeism questions did not allow for the determination of consecutive or nonconsecutive days missed from work, absenteeism was categorized as no days missed or 1 or more days missed. Employment status Respondents indicated their employment status by selecting one of the following categories: working full-time for pay; working part-time for pay; full-time student; parttime student; unemployed looking for work; home duties; retired or on a pension. Only respondents who selected working full- or part-time for pay were included in the current analyses. Industry groups were determined using Australian and New Zealand Standard Industrial Classification (ANZSIC) codes [22]. Occupational groups were determined using Australian Standard Classification of Occupation (ASCO) codes [23]. Occupation was categorized further as: professionals (managers, administrators, professionals and associate professionals); white-collar (advanced, intermediate and elementary clerical and service workers); and blue-collar (tradespeople, production and transport workers and labourers). Statistical analysis Data were analysed using the svy suite of commands in STATA version 9.2 statistical software to enable analyses to take account of the complex sample design [24]. Descriptive analyses were used to determine the proportion of the work-force that consumed alcohol at shortterm and long-term risk levels. Logistic regression was used to examine the association between alcohol consumption at short-term and long-term risk levels and (i) alcohol-related absenteeism, and (ii) any illness or injury absenteeism. The logistic regression models used in this report were restricted to workers who provided complete data on all relevant variables (alcohol-related absenteeism: 9879 workers, illness/injury absenteeism: 9249 workers). We first examined the univariate associations of each independent socio-demographic variable (age, gender, marital status, education, annual household income, country of birth, jurisdictional location and regional location) with absenteeism due to (i) alcohol; or (ii) any illness or injury. Socio-demographic covariates that reached significance of P < 0.05 were added into the multivariate model and deleted subsequently from the model if significance reduced to P > Odds ratios (ORs) and associated t statistics and P-values are reported. For all reported percentages and ORs, 95% confidence intervals (CIs) are provided. RESULTS Just over half the respondents (51%) to the 2001 NDSHS were employed either full- or part-time (survey respondents n = , estimated population n = ) and 10.6% of these were abstainers (95% CI, %). More than 40% (42.8%) of employed respondents reported drinking at risky or high-risk levels at least occasionally. Nearly 8% (7.8%, 95% CI, %) of the work-force drank at short-term risky or high-risk levels frequently (at least weekly), 16.7% (95% CI, %) drank at short-term risky or high-risk levels infrequently (at least monthly) and 18.3% (95% CI, %) drank at short-term risky or high-risk levels occasionally (at least yearly). For long-term risk, 8% (95% CI, %) of the work-force drank at long-term risky levels and 3.0% (95% CI, %) drank at longterm high-risk levels. The short- and long-term categories overlap almost entirely; 43.9% (95% CI, %) of the work-force drank at short- and/or long-term risk levels. The proportion drinking at these levels varied by age and gender (Fig. 1). Table 2 classifies respondents according to alcohol consumption risk category and the proportion in each category who reported absenteeism due to personal

4 Alchohol and absenteeism in Australia 741 Male workers Female workers All workers Figure 1 Proportion of the work-force aged 14 years and over who drank at either short- or long-term risk levels in the previous 12 months, by age group and gender, Australia 2001 Percentage (%) Total Age group (years) alcohol use and absenteeism due to any illness or injury in the 3 months prior to the survey. The proportion of workers who were absent for 1 or more days in the previous 3 months due to their personal alcohol use increased with higher levels of risky or high-risk consumption for both short- and long-term risk categories (F 7,6011 = 75.0, P < and F 3,2641 = 87.6, P < 0.001). A similar pattern is apparent for absences from any illness/ injury for both short- and long-term risk categories (F 7,6194 = 19.7, P < and F 3,2691 = 9.5, P < 0.001). However, in terms of the total sample population, fewer workers who frequently (at least weekly) drank at risky or high-risk levels reported absenteeism compared to workers who drank at low-risk levels or drank infrequently (at least monthly or yearly) at risky or high-risk levels (Table 2). Alcohol-related absenteeism Of the workers (estimated population n = ) who responded to the question concerning alcohol-related absenteeism, 3.5% (survey respondents n = 415, estimated population n = ) reported missing at least 1 work day due to their alcohol use (Table 2). A significantly larger percentage (F 1,905 = 16.0, P < 0.001) of males (4.2%, 95% CI, %) compared to females (2.5%, 95% CI, %) reported missing at least 1 work day due to their alcohol use (Table 3). Significant differences were observed between age groups. The likelihood of missing a work day due to alcohol use declined with age, with the highest proportion of workers reporting alcohol-related absenteeism being females aged years (11.0%) (Table 3). Table 4 displays the percentage of workers by industry and occupational groups reporting 1 or more work days missed in the past 3 months due to their alcohol use. There was a significant difference by industry grouping (F 9,8218 = 2.6, P = 0.006), with the hospitality industry having the largest proportion of workers (7.2%) reporting alcohol-related absenteeism. There were no significant differences in alcohol-related absenteeism between occupational groups (Table 4). Significant differences in alcohol-related absenteeism were observed according to respondents marital status (F 2,2187 = 71.9, P < 0.001). A larger proportion of never married workers (8.5%, 95% CI, %) and divorced or separated workers (4.3%, 95% CI, %) reported alcohol-related absenteeism compared to married workers or those in a de facto relationship (1.7%, 95% CI, %). A significantly larger proportion (F 3,2610 = 3.1, P < 0.026) of workers with a high school education or less (4.2%, 95% CI, %) reported alcohol-related absenteeism compared to workers with postgraduate qualifications (1.9%, 95% CI, %). A significantly larger proportion (F 2,1792 = 4.6, P < 0.010) of Australian-born workers (3.8%, 95% CI, %) reported alcohol-related absenteeism, compared to workers born in non-english-speaking countries (1.0%, 95% CI, %). No significant differences in alcoholrelated absenteeism were observed according to respondents location or household income. The alcohol-related absenteeism multivariate model Table 5 shows the final model for alcohol-related absenteeism associated with risky or high-risk alcohol consumption in the short term (F 11,895 = 24.8, P < 0.001) and long term (F 7,899 = 38.2, P < 0.001). Three covariates that were significant in the univariate analyses (i.e. industry, education level and country of birth) were nonsignificant when entered into the multivariate models and omitted from the final alcohol-related absenteeism model.

5 742 Ann M. Roche et al. Table 2 Self-reported alcohol-related and illness or injury absenteeism in the previous 3 months by alcohol consumption category, Australia, 2001.* Alcohol-related absenteeism Illness/injury absenteeism Survey n work-force (millions) Absent for 1 day Workers taking sick leave (millions) Survey n work-force (millions) Absent for 1 day Workers taking sick leave (millions) Short-term risk Abstainer % % ( ) ( ) ( %) ( ) Low risk % % ( ) ( %) ( ) ( ) ( %) ( ) At least yearly risky % % ( ) ( %) ( ) ( ) ( %) ( ) At least yearly high risk % % ( ) ( %) ( ) ( ) ( %) ( ) At least monthly risky % % ( ) ( %) ( ) ( ) ( %) ( ) At least monthly high risk % % ( ) ( %) ( ) ( ) ( %) ( ) At least weekly risky % % ( ) ( %) ( ) ( ) ( %) ( ) At least weekly high risk % % ( ) ( %) ( ) ( ) ( %) ( ) Total* % % ( ) ( %) ( ) ( ) ( %) ( ) Long-term risk Abstainer % % ( ) ( ) ( %) ( ) Low risk % % ( ) ( %) ( ) ( ) ( %) ( ) Risky % % ( ) ( %) ( ) ( ) ( %) ( ) High risk % % ( ) ( %) ( ) ( ) ( %) ( ) Total* % % ( ) ( %) ( ) ( ) ( %) ( ) *The totals for short-term and long-term risk are the same, as they involve the same population grouped according to different risk (i.e. short- or long-term) categories of drinking. Risk levels within each category (i.e. short- or long-term) are mutually exclusive; employees were surveyed, 841 were non-respondents for the question concerning alcohol-related absenteeism and 1663 were non-respondents for the question concerning illness/injury absenteeism. CI: confidence interval.

6 Alchohol and absenteeism in Australia 743 Table 3 Self-reported alcohol-related and illness or injury absenteeism in the previous 3 months by age and gender, Australia, Male workers Female workers All workers Age (years) work-force (millions) Proportion absent for 1 day work-force (millions) Proportion absent for 1 day work-force (millions) Proportion absent for 1 day Alcohol-related absenteeism* % % % ( ) ( %) ( ) ( %) ( ) ( %) % % % ( ) ( %) ( ) ( %) ( ) ( %) % % % ( ) ( %) ( ) ( %) ( ) ( %) % % % ( ) ( %) ( ) ( %) ( ) ( %) % % % ( ) ( %) ( ) ( %) ( ) ( %) % % % ( ) ( %) ( ) ( ) ( %) Total % % % ( ) ( %) ( ) ( %) ( ) ( %) Illness/injury absenteeism ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) Total ( ) ( ) ( ) ( ) ( ) ( ) *Significant differences observed between age groups for males (n = 5877, F 5,4127 = 16.6, P < 0.001), females (n = 5731, F 4,3960 = 23.6, P < 0.001) and total males + females (n = 11,608, F 5,4165 = 29.8, P < 0.001). Significant differences observed between age groups for males (n = 5467, F 5,4298 = 26.2, P < 0.001), females (n = 5319, F 5,4377 = 36.0, P < 0.001) and total males + females (n = 10,786, F 5,4385 = 57.6, P < 0.001). CI: confidence interval. After adjusting for age, gender and marital status, the alcohol-related absenteeism ORs were larger for workers who drank at risky or high-risk levels compared to workers who were low-risk drinkers (Table 5). For both short- and long-term risk levels, as consumption increased so did the likelihood of alcohol-related absenteeism. Compared to low-risk drinkers, workers drinking at short-term high-risk levels at least yearly, at least monthly or at least weekly were 3.1, 8.7 and 21.9 times (respectively) more likely to report alcohol-related absenteeism. Workers drinking at long-term risky or high-risk levels were 4.3 and 7.3 times (respectively) more likely to report alcohol-related absenteeism, compared to low-risk drinkers. Absenteeism due to any illness or injury Of the workers (estimated population n = ) who responded to the question concerning illness/injury-related absenteeism, 39.7% (survey respondents n = 4328, estimated population n = ) reported missing at least 1 work day in the 3 months prior to the survey (Table 2). A significantly larger percentage of females (F 1,902 = 19.2, P < 0.001) (42.6%, 95% CI, %) reported missing at least 1 work day compared to males (37.6%, 95% CI, %). Significant differences were observed between age groups, with the likelihood of missing a work day due to illness or injury declining with age (Table 3). There

7 744 Ann M. Roche et al. Table 4 Self-reported alcohol-related and illness or injury absenteeism in the previous 3 months by industry of employment and occupation, Australia Alcohol-related absenteeism* Illness/injury absenteeism Respondents work-force (millions) Proportion absent for 1 day Respondents work-force (millions) Proportion absent for 1 day Industry Education % % ( ) ( %) ( ) ( %) Hospitality % % ( ) ( %) ( ) ( %) Agriculture % % ( ) ( %) ( ) ( %) Manufacturing % % ( ) ( %) ( ) ( %) Retail % % ( ) ( %) ( ) ( %) Construction % % ( ) ( %) ( ) ( %) Transport % % ( ) ( %) ( ) ( %) Mining % % ( ) ( %) ( ) ( %) Wholesale % % ( ) ( %) ( ) ( %) Financial % % ( ) ( %) ( ) ( %) Administration and defence % % ( ) ( %) ( ) ( %) Services % % ( ) ( %) ( ) ( %) Total % % ( ) ( %) ( ) ( %) Occupation Professional ( ) ( ) ( ) ( ) White-collar ( ) ( ) ( ) ( ) Blue-collar ( ) ( ) ( ) ( ) Total ( ) ( ) ( ) ( ) *Significant differences were observed for alcohol-related absenteeism between industry groups (F 9,8218 = 2.6, P = 0.006) but not for occupation groups (F 2,1772 = 2.7, P = 0.071). Significant differences were observed for any illness or injury absenteeism between industry groups (F 11,9537 = 6.8, P < 0.001) and occupation groups (F 2,1772 = 6.1, P = 0.003). Of the respondents to question concerning alcohol-related absenteeism, 579 were missing data on industry and 656 were missing data on occupation. Of the respondents to question concerning any illness or injury absenteeism, 509 were missing data on industry and 567 were missing data on occupation. Totals differ due to differences in the number of respondents that provided information concerning their occupation and employment industry. CI: confidence interval. was a significant difference by industry grouping (F 11,9537 = 6.8, P < 0.001), with the administration and defence industry having the highest proportion (50.6%) of workers reporting illness/injury absenteeism (Table 4). Significant differences in illness/injury absenteeism were also observed between occupational groups (F 2,1772 = 6.1, P = 0.003), with the smallest proportion of workers reporting illness/injury absenteeism being blue-collar workers (36.6%) (Table 4). Significant differences in illness/injury absenteeism were also observed according to respondents marital status (F 3,2639 = 50.6, P < 0.001) and country of birth (F 2,1775 = 14.5, P < 0.010). A larger proportion of never married workers (51.6%, 95% CI, %) reported illness/injury absenteeism

8 Alchohol and absenteeism in Australia 745 Table 5 Adjusted odds ratios (ORs) for absenteeism in the previous 3 months by alcohol consumption category; Australia Alcohol-related absenteeism* Illness/injury absenteeism Odds ratio 95% CI t P Odds ratio 95% CI t P Short-term risk Abstainers Low risk 1.00 At least yearly risky At least yearly high risk < At least monthly risky < At least monthly high risk < At least weekly risky < At least weekly high risk < Long-term risk Abstainers Low 1.0 Risky < High risk < The categories of short- and long-term overlap as they involve the same population grouped according to different risk (i.e. short- or long-term). However, risk levels with each risk category (i.e. short- or long-term) are mutually exclusive. *Adjusted for age, gender and marital status. Adjusted for age, gender, marital status, industry of employment, country of birth and regional location (urban/non-urban). CI: confidence interval. compared to married workers or those in a de facto relationship (35.6%, 95% CI, %) and a smaller proportion of workers born overseas in non-englishspeaking countries (29.1%, 95% CI, %) and English-speaking countries (35.8%, 95% CI, %) reported illness/injury absenteeism compared to Australian-born workers (41.3%, 95% CI, %). There were also significant differences in illness/injury absenteeism according to respondents jurisdictional (F 5,4556 = 2.6, P = 0.025) and regional location (F 1,902 = 8.1, P = 0.005). Compared to other jurisdictions, the largest proportion of workers reporting illness/ injury absenteeism were South Australians (45.1%, 95% CI, %), while a smaller proportion of workers living in non-urban areas (37.1%, 95% CI, %) reported illness/injury absenteeism compared to urbandwelling workers (41.0%, 95% CI, %). Absenteeism due to any illness or injury was less common among workers with an annual household income of less than AUD$ (33.6%, 95% CI, %) compared to workers with an annual household income of AUD$ or more (42.1%, 95% CI, %) (F 3,2583 = 6.8, P < 0.001). No significant differences in illness/injury absenteeism were observed according to respondents level of education. The illness/injury absenteeism multivariate model Table 5 also shows the final model for illness/injury absenteeism associated with risky or high-risk alcohol consumption in the short term (F 26,875 = 17.1, P < 0.001) and long term (F 22,879 = 19.6, P < 0.001). Occupational group and jurisdictional location were omitted from the models due to multicollinearity with industry of employment and regional location, respectively. Annual household income, which was significant in univariate analyses, became non-significant in the multivariate models and was omitted from the final illness/injury model. After adjusting for age, gender, marital status, industry, country of birth and regional location, the ORs for illness/injury absenteeism were larger for workers who drank at short-term risky or high-risk levels than for workers who were low-risk drinkers (Table 5). Compared to workers who were low-risk drinkers, the odds of illness/injury absenteeism in the previous 12 months were 1.3 times larger for workers who drank at least yearly at short-term high-risk levels and 1.5 times larger for workers who drank at least weekly at short-term highrisk levels. The odds of illness/injury sick leave in the previous 3 months were not significantly larger for workers who drank at long-term risky or high-risk levels compared to workers who were low-risk drinkers (Table 5). DISCUSSION This paper reports a secondary analysis of the 2001 National Drug Strategy Household Survey (NDSHS), and is part of a suite of papers that provide the most comprehensive data available to date on a nationally representative sample of Australian workers alcohol consumption patterns [10,17]. The findings detail the extent to which

9 746 Ann M. Roche et al. risky drinking by workers has a negative impact on the work-place in terms of absenteeism and have important implications for work-place policies and prevention strategies. The findings presented here are unique in at least three respects. First, they provide a direct measure of the association between absenteeism and alcohol consumption based on respondents own reports of days off work attributed to their drinking. Secondly, they provide an examination of drinking patterns for both acute and chronic harms; whereas previous studies have focused only on chronic harms. In addition, they provide an assessment of the relationship between the frequency of short-term risky or high risk drinking (i.e. at least weekly, monthly or yearly) and absenteeism. The combined application of these three unique methodological features provides particularly sensitive measures of the contribution of workers alcohol use to absenteeism. Alcohol-related absenteeism More than 40% of the Australian work-force consumed alcohol at risky or high-risk levels for short-term harm on a frequent, infrequent or occasional basis. This pattern of drinking had significant negative consequences for individual workers and their work-places. Compared to lowrisk drinkers, workers who drank at risky and high-risk levels were significantly more likely to be absent from work due to their alcohol use. While the total proportion of the study population reporting a day off in the last 3 months due to their drinking was relatively small (3.5%), the probability of such an absence occurring among workers who drank weekly at risky/high-risk levels was extremely high. Workers who drank weekly at short-term risky or high-risk levels were, respectively, 12 and 22 times more likely to report alcohol-related absenteeism than short-term low-risk drinkers. This is of concern, as nearly one in 10 employees drink at these levels. Even workers who only drank at short-term risky or high-risk levels infrequently (at least monthly) or occasionally (at least yearly), were significantly more likely ( times) than low-risk drinkers to report alcoholrelated absenteeism. Again, this is of concern because 35% of the work-force drank at these levels. Compared to long-term low-risk drinkers, workers who drank at longterm risky levels were four times more likely to report alcohol-related absenteeism, while those who drank at long-term high-risk levels were seven times more likely. Alcohol-related absenteeism was also heavily skewed demographically, with males, younger workers and workers in certain industry groups more likely to take an alcohol-related sick day. This finding is of particular relevance to industries and occupational groups with high concentrations of young male and female workers, and to those industries identified as having large proportions of heavy drinkers. While only a relatively small proportion of the drinking population (i.e. 3.5%) reported alcohol-related absenteeism, this translates none the less into very large numbers of workers and days off. The current data indicate that in a 3-month period approximately Australians (weighted sample) took at least 1 day off work due to their drinking. Using the same NDSHS data set, we previously estimated the extent of alcohol-related absenteeism in 2001 to be nearly 2.7 million days lost at a cost of AUD $437 million [10]. While these figures may seem relatively small, given the size of the national economy and work-force, our findings have important practical implications for at least two reasons. First, the cost of alcohol-related absenteeism, whether borne by the private or public sector, remains a substantial preventable impost on the economy. Moreover, our estimate exceeds, by a factor of 12, previous estimates of alcohol-related absenteeism [25]. Secondly, and perhaps more importantly, our finding that alcohol-related absenteeism was heavily skewed among specific work-force subpopulations indicates scope for cost-effective targeted interventions. Illness/injury-related absenteeism Forty per cent of workers surveyed had taken a day off in the past 3 months due to any illness or injury. Workers who regularly (at least weekly) drank at risky or highrisk levels were significantly more likely than low-risk drinkers to take a day off work due to illness/injury. This is consistent with previous research that has found heavy alcohol consumption to be associated with increased sickness absence [26]. Importantly, workers who drank at high-risk levels only infrequently (i.e. at least monthly) or very occasionally (i.e. at least yearly) were more likely than low-risk drinkers to report illness/injury absenteeism. Again, this finding is consistent with previous research that highlights the elevated risk of harm associated with irregular heavy drinking bouts [16]. Implications for work-place policy and prevention strategies The findings of this study have important implications for work-place policy and preventive strategies. Employees patterns of risky alcohol consumption have a wide range of negative impacts for the work-place, not least of which is absenteeism. Most of this drinking is by non-dependent risky drinkers whose patterns of drinking are readily amenable to intervention and modification. That is, much of the loss to industry and the economy through risky drinking can be prevented. Widespread work-place education programmes are needed that focus on what constitutes at-risk levels of consumption and that are tailored to specific industry

10 Alchohol and absenteeism in Australia 747 settings and groups of workers. Such programmes need to highlight the potential implications of workers risky consumption levels for individual employees and the work-place as a whole. There is growing evidence to support the efficacy of such interventions to reduce the incidence of alcohol-related absenteeism and alcoholrelated risk to safety [27 30]. Educational programmes need to target young workers in particular. There is evidence, albeit limited, to indicate alcohol-related health and safety education can influence young employees attitudes and behaviours regarding alcohol use [31]. Such interventions warrant closer investigation via rigorous randomized controlled trials. These are strongly encouraged as the next logical step. The present findings also highlight the need to take a whole of work-place approach when designing and implementing prevention strategies. Traditionally, workplace interventions have tended to focus on employees who are chronic heavy drinkers or have been identified as problem drinkers. While these employees are of concern, the results of the current study show that workers who drink at short-term risk levels, even infrequently, are at elevated risk of alcohol-related work-place absenteeism. Thus, negative work-place outcomes associated with alcohol use are not restricted to the relatively small number of heavy drinkers who drink at chronic long-term high-risk levels, but also involve the much larger number of workers who drink infrequently at risky levels [10]. Hence, targeting all workers through education, prevention and policy strategies is an appropriate approach. This necessitates a substantial revision of the narrow approaches often taken with work-place programmes. Methodological issues The findings reported here are not without qualification. First, the alcohol consumption measures utilized in the 2001 NDSHS underestimate the total volume of alcohol consumed in Australia in 2001 [21]. Secondly, the overall response rate for the 2001 NDSHS was less than 50% [20] and high-intake drinkers may have been less likely than other drinkers to participate. Thirdly, confounders may have also played a role. Drinkers, for instance, are more likely to be smokers than abstainers [32] and smokers have higher levels of absenteeism than nonsmokers [33]. Similarly, depression is associated with both absenteeism [34] and heavy alcohol consumption [35]. A range of other health and life-style differences between drinkers and abstainers may also account for differences in illness or injury absenteeism. Finally, the inherent limitations of self-report measures aside, the reliability and validity of the absenteeism measures used in the NDSHS can be questioned. Direct self-report measures may be more valid and reliable indicators of alcohol-related absenteeism compared to inferred associations based on self-reported illness/injury absenteeism. However, individuals may not always recognize that alcohol contributed to a work absence, or they may wrongly attribute an absence to alcohol. The absenteeism measures used did not allow an examination of leave duration. Thus, the degree to which consumption patterns impact brief or extended periods of absenteeism could not be determined. Implications for future research Previous research on alcohol and absenteeism has used consumption measures that focused only on long-term (chronic) harm. Such measures are of limited value. Consumption patterns associated with short-term (acute) harms also need to be examined. Similarly, the use of direct rather than indirect measures of alcohol-related absenteeism produced a more precise indication of the relationship between alcohol consumption patterns and absenteeism and extent of alcohol-related absenteeism. Future research should incorporate accurate direct measures which distinguish between brief and extended periods of alcohol-related absenteeism. Finally, future research should acknowledge the range of health and life-style factors with potential to confound the relationship between alcohol consumption and absenteeism by using randomized controlled study designs. CONCLUSION The risky alcohol consumption patterns of the workforce are associated with substantial negative outcomes for individual employees and the work-place at large. Hitherto, these have not been well measured. There are various strategies that employers can utilize to address the social and economic costs associated with workers risky alcohol consumption. Relatively simple, effective and inexpensive interventions have the potential to improve the health and wellbeing of workers and contribute to overall levels of work-place safety and productivity. Acknowledgements This project was funded by the Australian Government Department of Health and Ageing in support of the National Alcohol Strategy. Mark Cooper-Stanbury provided valuable technical advice enabling comparability between this analysis of survey data and previous work based on this source. References 1. Blum T. C., Roman P. M., Martin J. K. Alcohol consumption and work performance. J Stud Alcohol 1993; 54:

11 748 Ann M. Roche et al. 2. Cunradi C. B., Greiner B. A., Ragland D. R., Fisher J. Alcohol, stress-related factors, and short-term absenteeism among urban transit operators. J Urban Health 2005; 82: Jones S., Casswell S., Zhang J. F. The economic costs of alcohol-related absenteeism and reduced productivity among the working population of New Zealand. Addiction 1995; 90: McFarlin S. K., Fals-Stewart W. Workplace absenteeism and alcohol use: a sequential analysis. Psychol Addict Behav 2002; 16: Upmark M., Moller J., Romelsjo A. Longitudinal, population-based study of self reported alcohol habits, high levels of sickness absence, and disability pensions. J Epidemiol Commun Health 1999; 53: Allsop S., Bush R., Phillips M., Midford R., Vincent N., Ask A. et al. Alcohol and Other Drugs in the Australian Workplace: A Critical Literature Review. Adelaide: National Centre for Education and Training on Addiction (NCETA), Flinders University of South Australia; Allsop S., Phillips M., Calogero C., editors. Drugs and Work: Responding to Alcohol and Other Drug Problems in Australian Workplaces. Melbourne: IP Communications; Hocking B., Grain H., Gordon I. Cost to industry of illnesses related to alcohol and smoking. A study of Telecom Australia employees. Med J Aust 1994; 161: Webb G., Redman S., Hennrikus D., Kelman G. R., Gibberd R., Sanson-Fisher R. The relationship between high-risk drinking and problem drinking and the occurrence of work injuries and related absences. J Stud Alcohol 1994; 55: Pidd K. J., Berry J. G., Roche A. M., Harrison J. E. Estimating the cost of alcohol-related absenteeism in the Australian workforce: the importance of consumption patterns. Med J Aust 2006; 185: Bush R., Wooden M. Smoking, Alcohol and Absence in the Workplace: An Analysis of the 1989/1990. National Health Survey. Adelaide: National Institute of Labour Studies, Flinders University; Hagen R., Egan D., Eltringham A. Work, Drugs and Alcohol. Melbourne: Occupational Health and Safety Commission; Pols R. G., Hawks D. V. Is There a Safe Level of Daily Consumption of Alcohol for Men and Women?, 2nd edn. Canberra: Australian Government Publishing Services; National Health and Medical Research Council (NHMRC). Australian Alcohol Guidelines: Health Risks and Benefits. Canberra: NHMRC; World Health Organization (WHO). WHO Guide to Mental and Neurological Health in Primary Care, 2nd edn. London: WHO Collaborating Centre for Research and Training for Mental Health; Rehm J., Room R., Graham K., Monteiro M., Gmel G., Sempos C. T. The relationship of average volume of alcohol consumption and patterns of drinking to burden of disease: an overview. Addiction 2003; 98: Berry J. G., Pidd K., Roche A. M., Harrison J. E. Prevalence and patterns of alcohol use in the Australian work-force: findings from the 2001 National Drug Strategy Household Survey. Addiction 2007; 102: Australian Institute of Health and Welfare. National Drug Strategy Household Survey, 2001 [computer file].canberra: Social Sciences Data Archives, The Australian National University; Roy Morgan Research. National Drug Strategy Household Survey Technical Report. Melbourne: Australian Institute of Health and Welfare (AIHW); Australian Institute of Health and Welfare (AIHW) National Drug Strategy Household Survey: First Results. Canberra: AIHW; Stockwell T., Donath S., Cooper-Stanbury M., Chikritzhs T., Catalano P., Mateo C. Under-reporting of alcohol consumption in household surveys: a comparison of quantity frequency, graduated-frequency and recent recall. Addiction 2004; 99: Australian Bureau of Statistics (ABS). Australian and New Zealand Standard Industrial Classification (ANZSIC). ABS Cat. no Canberra: ABS; Australian Bureau of Statistics (ABS). Australian Standard Classification of Occupation (ASCO), 2nd edn. ABS Cat. no Canberra: ABS; StataCorp. Stata Statistical Software: Release 9.2. College Station, TX: Stata Corporation; Collins D. J., Lapsley H. M. The Social Costs of Drug Abuse in Australia in 1988 and Canberra: Australian Government Publishing Service; Norstrom T. Per capita alcohol consumption and sickness absence. Addiction 2006; 101: Bennett J. B., Patterson C. R., Reynolds G. S., Wiitala W. L., Lehman W. E. K. Team awareness, problem drinking, and drinking climate: workplace social health promotion in a policy context. Am J Health Prom 2004; 19: Cook R., Back A., Trudeau J., McPherson T. Integrating substance abuse prevention into health promotion programs in the workplace: a social cognitive intervention targeting the mainstream user. In: Bennett J. B., Lehman W. E. K., editors. Preventing Workplace Substance Abuse: Beyond Drug Testing to Wellness. Washington, DC: American Psychological Association; 2003, p Heirich M., Sieck C. J. Helping at-risk drinkers reduce their drinking: cardiovascular wellness outreach at work. In: Bennett J. B., Lehman W. E. K., editors. Preventing Workplace SubstanceAbuse:BeyondDrug TestingtoWellness. Washington, DC: American Psychological Association; 2003, p Spicer R. S., Miller T. R. Impact of a workplace peer-focused substance abuse prevention and early intervention program. Alcohol Clin Exp Res 2005; 29: Pidd K. The impact of workplace support and workplace identity on training transfer: an Australian case study. Int J Training Dev 2004; 8: Australian Institute of Health and Welfare (AIHW) National Drug Strategy Household Survey: Detailed Findings. AIHW cat. no. PHE 66, Drug Statistics Series no. 16. Canberra: AIHW; Bush R., Wooden M. Smoking and absence from work: Australian evidence. Soc Sci Med 1995; 41: Lim D., Sanderson K., Andrews G. Lost productivity among full-time workers with mental disorders. J Ment Health Policy Econ 2000; 3: Goldstein B. I., Velyvis V. P., Parikh S. V. The association between moderate alcohol use and illness severity in bipolar disorder: a preliminary report. J Clin Psychiatry 2006; 67:

12