Impact of social network on incidence of sick leave

The Swedish Institute for Labour Market Policy Evaluation recently published a report exploring how individuals are affected by sick leave in their social network. Through a randomised social experiment, it was found that if one group’s level of sick leave increases so will that of other individuals in the same network, due to social interaction and norms. These findings are significant for the future planning of sickness allowance systems.

Background

In Sweden, sickness allowance covers loss of income during temporary sickness. The level of compensation has varied somewhat over the years (see for example SE0712039I). Today, the allowance covers 80% of the person’s salary, up to a maximum level. During the first seven days of sick leave, it is up to the employee to decide whether they need to take time off for health reasons. However, when that time has passed, a doctor’s certificate is necessary to retain further compensation.

Previous research regarding sickness allowance systems has focused on direct effects, such as how the length and frequency of sick leave are affected by the level of compensation. However, a new report (in Swedish, 600Kb PDF), recently published by the Institute for Labour Market Policy Evaluation (Institutet för arbetsmarknadspolitisk utvärdering, IFAU), investigates whether individuals are affected by sick leave being taken by others in their surroundings. In medical sociological research, it is commonly known that perceived health and sickness also depend on the individual’s surroundings and existing social norms. For example, research carried out by Professor Phil Brown ‘explore(s) the effects of class, race, gender, language, technology, culture, the political economy, and institutional and professional structures and norms in shaping the knowledge base which produces our assumptions about the prevalence, incidence, treatment, and meaning of disease’ (Brown, 1995). The IFAU report confirms previous research findings by proving that social norms are likely to influence behaviour in relation to sick leave.

About the study

The report draws on the results of a previous study conducted in the second half of 1988 in the southwestern city of Gothenburg, which examined whether and how the duration of sick leave changed when the requirement for a doctor’s certificate was postponed. A randomly chosen case group living in the city of Gothenburg had their sick leave extended to two weeks without the requirement of a medical certificate, while a control group, also living in the city of Gothenburg, retained the old precondition of one week. As a result of this experiment, the average duration of sick leave among the case group increased by 0.18 days.

In the IFAU report, the authors added yet another dimension. They assumed that the postponed requirement of a doctor’s certificate for the case group should also have some effect on the control group, due to the hypothesis that the control group – observing that the case group individuals stayed at home from work longer during sickness – would follow the case group’s example. The social interaction between the two groups would, thus, make the experiment have an effect on the control group as well. The authors therefore decided to extend the study. Their hypothesis was that the control group (control group 1), living in the same city and hence likely to meet and interact with members of the case group, would subconsciously also change their sick leave behaviour in accordance with the case group, due to social norms.

Moreover, the hypothesis argued that, if social interactions affected sick leave behaviour, the effect should decline with the geographical distance from the case group. In other words, the assumption was that people living further away from the members of the case group would be less likely to interact with the latter and, hence, not be as affected by their increased level of sick leave. To prove this theory, another control group was added consisting of individuals living in the municipalities directly surrounding Gothenburg (control group 2) and yet another group from the municipalities surrounding those (control group 3).

The researchers measured sick leave for all groups and then estimated the relative difference between the case group and the three control groups. Some 428,730 individuals between the ages of 20 and 60 years participated in the experiment. Differences regarding men and women are not presented in the study.

Study results

The first hypothesis – changed behaviour in sick leave in control group 1 due to the postponed requirement of a medical certificate in the case group – was verified and the results can be divided into two aspects.

• Firstly, with regard to those in control group 1 who were returning from a period of sick leave, the marginal effect decreased by 0.035 percentage points as the case group’s level of sick leave increased. This means that, as the case group extended their days of absence due to health reasons, so did control group 1.

• Secondly, in relation to time taken off due to health reasons – that is, beginning a period of sick leave – the marginal effect turned out to be insignificant. In other words, while the case group was absent from work more often due to the increased sickness allowance, the control group did not show such an effect.

The second hypothesis, declaring that the effect on the control groups 2 and 3 should decline with the geographic distance from the case group, also proved to be true. In control group 2, the effect was 2.9% higher than in control group 3. The corresponding effect for control group 1 was 4.8%. Thus, the results showed that, the closer a control group individual was to the case group, the more their sick leave increased.

These results imply that social interactions make the control group act in accordance with the case group. Social interaction therefore has an effect on sick leave behaviour.

Commentary

Extensive debate has always surrounded the level and structure of the sickness allowance system. Against this background, the results of this study are significant for policy regarding sickness allowances. The research demonstrates that not only is the level of sickness allowance important in controlling how the allowance is used: social interaction and norms are equally significant factors. These results could have consequences for how sickness allowance systems, and maybe even allowance systems generally, are constructed.

Reference

Brown, P., ‘Naming and framing: The social construction of diagnosis and illness’, Journal of Health and Social Behavior, Vol. 35, 1995, pp. 34–52.

IFAU, Påverkas individen av omgivningens sjukfrånvaro? (in Swedish, 600Kb PDF), Report 2008:11, Uppsala, 2008.

Thomas Brunk and Lisa Olsson, Oxford Research