The result of this study suggests that SEP modifies the association between average volume of alcohol consumption and sickness absence, such that at any given level of alcohol use individuals with lower SEP had higher rates of sickness absence. Differential distribution of HED and work conditions explain a large part but not all of the differential vulnerability, subsequently the effect modification of alcohol use by SEP remained.
The current results extend and further support previous research on the social gradient in alcohol-related harm [
13,
15,
16]. In contrast to a recently published study from Norway, we found that the same levels of drinking were associated with higher rates of sickness absence in groups with lower compared to higher SEP [
32]. Specifically, all patterns of drinking among the manual workers were associated with an increased risk of sickness absence when compared to non-manual employees with moderate drinking. It is notable that the association between average volume and sickness absence, foremost appears to apply to manual workers, considering that heavy drinking was not associated with higher rates of sickness absence compared to moderate drinkers among non-manual employees. The highest IRR was found for heavy drinking manual workers who had approximately five times higher rates of sickness absence compared to the reference group. Calculations of RERI revealed that approximately 60% of the excess risk among heavy drinking manual workers was attributable to the interaction between alcohol use and SEP.
To address our second aim, we adjusted the main association for several measures of other lifestyle factors, health, and factors related to sickness absence such as physical and psychosocial work conditions [
5,
33]. In line with previous research on mortality and alcohol-related morbidity and mortality we found that the joint effect of alcohol consumption and SEP could not be attributed to differences in HED, other lifestyle factors, and health [
13,
15,
16]. Work conditions appear to explain a substantial but not all the joint effect, subsequently SEP remained a modifying factor. In contrast to previous findings for more alcohol specific outcomes, such as alcohol-related morbidity and mortality, the differential vulnerability was explained by other factors to a smaller extent [
13,
15,
16]. A potential reason for this is that sickness absence is a much more complex outcome, relating to reduced work ability usually linked to ill health as well as loss of work-related income. Consequently, there could be other underlying mechanisms with regard to sickness absence that might explain that the social gradient such as differential consequences [
34]. This mechanism suggests that individuals with higher SEP might be more capable to deal with the direct cost of health care and income forgone, as well as being an occupation that can better hide alcohol-related absenteeism, and adapt the current work conditions to prevent new spells of sickness absence [
8]. In other words, although having the same level of alcohol-related health problems, it might be more difficult to manage one’s job within a low SEP occupation. Moreover, there could potentially be a healthy worker effect as individuals with consistent harmful alcohol use which causes health problems might be selected out of the labour market to a higher degree, implying that the role of alcohol for the social gradient in alcohol-related absenteeism probably is larger than estimated by our current analyses [
35]. Furthermore, the results of the current study demonstrated that differences in physical and psychosocial work explained a large of the social gradient found in sickness absence, which seems reasonable as some occupations are more mentally and physically challenging than others [
8]. Finally, it is notable that abstainers had similar rates of sickness absence among both SEP-groups in the fully adjusted model, implying that the mechanisms underlying the increased sickness absence rates among this drinking group probably is less socially patterned.
To better tackle health inequalities in health in general and sickness absence in particular, an increased understanding of how an individual’s social position influence their risk of ill health is vital [
36]. In line with previous research, the current study found that the social gradient in sickness absence is not only related to differential exposure of alcohol but also due to the differential vulnerability to the negative effects of alcohol [
15,
36]. Consequently, universal alcohol policy measures that effectively help reduce the total alcohol consumption will most likely have a larger impact on sickness absence among groups with low SEP, as they are the most vulnerable - subsequently reducing the social gradient of this outcome [
36,
37]. Furthermore, previous research on an aggregated level suggests that deceased total alcohol consumption can lead to a reduction in sickness absence costs [
38,
39].
Strengths and limitations
Major strengths of the current study include being able to prospectively follow a large cohort, detailed information on exposure before the outcome, and several known risk factors. Unfortunately, we were however only able to include one measurement point of alcohol consumption which is a limitation as alcohol consumption tend to change during the life course [
40]. Furthermore, there could be a difference in adjusting harmful alcohol consumption prior to sickness absence, where higher SEP groups a more likely to adjust their consumption downward [
41]. Furthermore, including register-based information on the outcome of sickness absence ensures nationwide coverage and complete follow-up information. Several reviews have, however, shown that the effects of alcohol on sickness absence are found when sickness absence is based on register data and not to the same extent when the information is self-reported [
9,
42].
Further strengths include being able to include extensive information on other risk-factors that could potentially explain the differential vulnerability to alcohol in relation to sickness absence among SEP groups as this has not been extensively done in previous research [
8].
Due to power issues, we were unable to stratify the analyses by sex which could be seen as a limitation as drinking behaviour and sickness absence differs between males and females, where males consume large quantities of alcohol and females tend to be overrepresented among sickness absence beneficiaries [
43,
44]. Previous research has found that increased alcohol use increase the risk of sickness absence among both men and women, where some evidence suggests this association could be stronger for females compared to males [
8,
9].
There is most likely an under-representation of heavy and problem drinkers in the sample, suggesting that the association between alcohol use and sickness absence might be underestimated to some extent. It should also be acknowledged that the threshold for HED in this study, set at 120 g on a single occasion, is higher than the standard cut-off. It is possible that using a lower threshold, such as 60 g, could lead to a larger attenuation of the main association. Finally, alcohol use and other explanatory factors were only assessed at baseline. Previous research indicates that accounting for time-varying changes in health behaviors tend to yield larger estimates of their role for socioeconomic inequalities in health [
41,
45].