This study offers an assessment of the cumulative incidence of all-cause sickness absence as well as the burden of chronic disease amongst the same individuals. The results of men and women differed in all age groups and chronic diseases were more prevalent amongst the sick listed than in the general population.
Information on incidence of sickness absence is important to grasp the size of the problem in terms of number of individuals involved, especially when following the development of sickness absence or planning preventive measures . Previous research has often focused on self-reported sickness absence, sickness absence duration or on certain groups of employees or diseases, thereby portraying different aspects of the problem, not always easy to compare to other results.
The one-year cumulative incidence in this study was 11.3%, meaning that 75,600 individuals in Västra Götaland region would be sick-listed for a period lasting ≥14 days at least once annually. The cumulative incidence of sickness absence was studied in Sweden and Norway in the 1980’s and the 1990’s [37, 38]. However, changes in both society and social insurance have taken place since then making comparisons difficult . A study by Lidwall and Marklund (2010) showed great variation in the number of ongoing sickness absence cases over the years 1992–2008, underscoring the inherent difficulty in comparing different periods . Eriksson et al. (2008) found that 40% of workers reported any sickness absence during the previous year, while only 4% had been absent over 29 days, a figure comparable to ours . Findings from Denmark show that the cumulative incidence of sickness absence (also of a minimum of two weeks) among employed people increased from 6.6% to 7.5% between 2000 and 2007, which is slightly lower than our findings .
Our results show that the cumulative incidence of sickness absence increased in the older age groups approaching retirement for both genders. This is not unexpected since medium and long term sickness absence is more common in older age group. Sweden has high employment rates overall and in older age groups , which means that more employees with health problems might be present in the workforce compared to other countries . This can be one reason for the higher sickness absence levels in Sweden. The ageing of the population and the trend towards a higher retirement age may also lead to increased numbers of individuals with chronic disease in the workforce .
With respect to the gender difference by age, the cumulative incidence among women increased stepwise between age groups, whereas amongst men the most prominent difference was between the middle and the oldest age groups. A Norwegian study found that women <50 years of age and sick-listed >8 weeks due to musculoskeletal disorders had a higher risk of chronicity than men in the same age group . The authors suggest that women may develop chronic musculoskeletal disorders at an earlier age than men. With respect to the current study, the substantial increase in cumulative incidence of sickness absence among women aged 31–50 years compared with men in the similar age group may be influenced by the phenomenon of earlier chronicity. Moreover, women in this particular age-group are often engaged in both work and family. According to the double burden hypothesis, women combining careers with responsibility for children and domestic work may face a higher risk of sickness absence . In a Swedish longitudinal study women taking on a parental role during follow up had increased odds for sickness absence compared to those not adding such a role . Other studies have also linked having young children to increases in sickness absence for women, however this has also been seen for men .
The labour market in Sweden is highly gender segregated. Women and men tend to work in different sectors (horizontal segregation) and men tend to have higher positions in the workplace and the occupational hierarchy (vertical segregation). Both horizontal and vertical segregation seem to influence sickness absence rates [13, 47]. A Finnish study found that the overall gender differences in sickness absence can be explained by the fact that it is more common that women have shorter absences . This does not seem to apply to our results as the gender difference was apparent despite exclusion of shorter absence periods.
In the current study, gender differences in relation to the different self-reported chronic disease groups were in line with what has been observed previously in Sweden (age group 35–64 years), where chronic musculoskeletal, gastrointestinal and psychiatric symptoms as well as rheumatoid arthritis, were more prevalent among women . Significantly more men reported having cardiovascular disease than women, correlating with the fact that women have been found to develop cardiovascular disease 7–10 years later in life than men . The importance of chronic disease as a contributory factor to gender differences has not been much explored. A Swedish study proposes a triple burden for women and men with chronic disease added to the demands experienced from paid and unpaid work . Women more often than men experience a double burden in daily life. The presence of chronic disease as a third burden may render women more vulnerable to the work-related burden and place them at increased risk for sickness absence.
In general, the newly sick-listed reported more chronic disease and were more likely to have co-morbid diseases than the random general population sample. The most commonly reported chronic diseases in the sick-listed group were musculoskeletal, gastrointestinal, psychiatric and cardiovascular diseases. Studies have suggested that complex symptoms such as musculoskeletal pain, digestive problems and mental problems are the leading causes of sickness absence . These conditions display different symptoms, may appear separately or in combination and may share a common mechanism of origin. In the current study, the high burden of musculoskeletal, gastrointestinal, psychiatric and co-morbid conditions among the sick-listed may be interpreted with this in mind. The higher disease burden when compared with the general population remained consistent throughout standardisation for age and socio-economic status. There are few studies examining the burden of sickness absence among individuals with chronic disease. A study on individuals with angina pectoris found an almost threefold higher rate of sickness absence compared with a sample not suffering from chronic disease . Similar results have also been reported for fibromyalgia and asthma [27, 28].
There seems to be an even higher risk when more than one chronic disease is present [24, 27, 28]. In the current study we found a substantially increased risk of having co-morbid diseases among the newly sick-listed compared with the general population. The reporting of neoplasia was nearly four times higher which might reflect that most individuals diagnosed with a malignancy become sick-listed. This might be due to severe physical symptoms and the difficult treatments together with the psychological and social impacts associated with this disease . Sick-listed men and women had an increased risk of reporting psychiatric disease, rheumatic disease and endocrine disease and sick-listed men had a substantially higher risk of reporting neurologic disease. These findings show that the burden of chronic diseases seemed to be significantly higher among newly sick-listed individuals compared with the general population, and that this difference was not explained by a different composition of the newly sick-listed population when it comes to sex, age or socio-economic status. It is also noteworthy that this result held true when standardized for socio-economic status in addition to age compared to standardizing for age alone indicating that those with a chronic disease are more likely to become sickness absent, regardless of socio-economic class.
Strengths and weaknesses
A major strength of the current study was the inclusion of all newly sick-listed individuals with employment providing a population based estimate of the cumulative incidence of sickness absence. Demographic information and socio-economic status were derived from highly reliable Statistics Sweden register data. A possible weakness in our methodology is an underestimation of the cumulative incidence if part of the working population was not at risk, possibly due to sickness absence.
The estimation of the one-year cumulative incidence might be subject to error due to seasonal variation. Sickness absence rates are lower during the second and third quarters and higher in the first and fourth quarters . The sickness absence cases included in this study were spread over the second part of the first quarter and the first half of the second quarter. It is therefore likely that the effects of seasonal variation in the current study were minimized. A study from 1974 addressed seasonal variation in all-cause sickness absence specifically and found that most seasonal variation was due to upper respiratory disease and bronchitis in addition to digestive disorders . This strengthens the assumption that the effects of seasonal variation in this study ought to have been minimal as sickness absence due to respiratory infections or gastroenteritis rarely exceeds two weeks  and would therefore not have affected our results.
Data in the second part of the study is based on self-reported data and there are several potential causes for bias. There was a drop-out in the first phase of participant inclusion. Not all individuals sick-listed between the 18th of February and the 15th of April 2008 were formally registered during the period and included in our study. The delay was primarily for administrative reasons, such as determination of income in order to calculate the appropriate sickness benefit. In our study we nevertheless include 49% of the sick-listed during the period, which is a large sample size. The group with later registration had a higher proportion of individuals with low income, of men (male/female ratio 1.32), of highly educated and first time sick-listed as well as a slight overrepresentation of immigrants (immigrant/non-immigrant ratio 1.08) according to our correspondence with the Department for Statistics and Analysis at the Swedish Social Insurance Agency. Unfortunately more detailed statistical information is unavailable at this time from the agency. Thus no information is available regarding any difference in the diagnostic groups dependent on the time of registration. Therefore there is a possibility for a systematic bias based on differences in registration date which we cannot completely exclude. This risk is somewhat offset by the large sample size but the smaller diagnostic groups (for example neoplastic disease) would be more sensitive to any such error.
Another possible source of bias is the relatively low response rate which may have led to a selection bias. The non-response rate was highest in the youngest age group, those with the lowest income and those born outside of the Nordic countries. The results are therefore less applicable to those groups. The results should also be interpreted in the light that the selection bias could also be affected by other variables not captured by the non-response analyses from Statistics Sweden. Information on chronic diseases and conditions was based on self-reported data so recall and reporting bias may therefore have occurred. Those who were sick-listed or had recently been sick-listed may have been more prone to remembering or reporting chronic diseases. This could have led to an overestimation of the standardized incidence ratios. A health response bias might also be present, with the healthiest individuals responding. Sensitivity and specificity of self-reported data on chronic disease has been found to vary from one disease group to another, with data for cardiovascular disease, diabetes and cancer being most reliable [57, 58] while data for musculoskeletal  and psychiatric  disorders is less reliable . Musculoskeletal disorders had the highest frequency in both samples and even if all of these might not be confirmable with clinical examination, these results give insight into the subjective experience of musculoskeletal problems among the sick-listed and the general population. The frequencies of psychiatric disorders in both samples are lower than the 15-25% 12 month prevalence of mental disorders reported by the WHO . This may be due to responders being less likely to disclose mental health problems due to stigma and the results should be interpreted with the possibility of underestimation of psychiatric problems in mind.
Finally, the newly sick-listed sample was composed solely of employed individuals while the general population sample was a random sample of the total population. While we have corrected for the differences in age and socioeconomic position between the groups with standardisation for these factors, we cannot however rule out the possibility of other variations between the groups, some of which might affect our results. This is a possible major limitation of our study, which is however difficult to avoid when seeking a comparison group for the sick-listed population. An example of this type of effect is that we may have underestimated the differences between the newly sick-listed employed population and the general population as the employed population was a healthier group than the general population, which included the unemployed, individuals on disability pension as well as individuals on recurrent or long-term sick-leave.