Background

Raising the retirement age and getting more people in work at higher ages are concerns in many countries [1, 2]. The shift toward staying in paid work at an older age is highly needed in many Western countries due to the growing proportion of elderly [3, 4]. This is progressively possible due to increased life expectancy [5], higher number of healthy life years [6], and somewhat improved self-rated health among individuals above age 64 [7, 8].

Women tend to retire earlier than men [9] and more men than women work past age 65 [10]. This gender difference seems to become larger with age, as there are over twice as many men as women in paid work after the age of 71 [10]. Paid work past the age of 65 is also associated with factors such as country of birth [11], type of living area [12], educational level [11, 13], and family situation [12, 14,15,16].

In Sweden, a substantial dropout in work participation occurs at the age of 65 [17]. However, the number and proportion of people working beyond the standard retirement age of 65 has increased rapidly over the past decades [10]. The percentage of 66–71-year-olds in paid work increased from 9.8% in 1995 to 24.0% in 2010 [10]. Work participation is lower at higher ages, although a small increase has also been noted among people above age 71 [10].

Mental health and working past age 65

To the best of our knowledge, there are no studies that have examined the association between sickness absence (SA) or disability pension (DP) due to mental diagnoses in the years leading up to the standard retirement age of 65 and being in paid work after 65. However, there are a few longitudinal studies, mostly based on self-reports, on different health conditions and paid work at an older age. In a European-wide study, 65–80-year-olds in paid work had better mental health compared to those not in paid work [18]. Moreover, not being diagnosed with a mental condition was associated with extended employment among Finnish public sector employees who had reached standard retirement age [19]. Further, people with on-going depression have been found to have a higher probability of retirement [20]. An Australian study found that poor mental health predicted overall retirement in men aged 45–75, especially early retirement [21]. However, this association was not found in women [21]. In contrast, a study from the Netherlands found no significant association between mental health and staying in paid work past the retirement age [22].

Somatic health and working past age 65

Similar to results regarding mental disorders, there are no studies to the best of our knowledge that have investigated the association between SA and/or DP due to somatic diagnoses in the years leading up to the standard retirement age and staying in paid work after that. However, good somatic health [22] and absence of chronic disease [23] have been seen to predict working beyond the standard retirement age in the Netherlands. Indeed, physical health is shown to be better among working 65–80-year-olds compared to those not in paid work in this age group [18], a phenomenon often called the healthy worker effect. By the age of 60, age-related losses of functional capacity, such as deteriorated vision, hearing, and physical capacity, as well as non-communicable diseases (e.g., stroke, chronic respiratory disorders, and heart disease) give rise to the majority of disabilities [24]. Poor self-rated health increases the likelihood of exiting from paid work through DP before age 65 [25, 26]. In addition, a higher mean number of SA days when aged 60–64 was found in individuals exiting the workforce at the age of 65 compared to those staying in paid work past 65 [17]. A study based on survey data from 16 European countries concluded that people who still work beyond age 65 have relatively good health [18]. However, more knowledge is needed on how specific aspects of morbidity are associated with not being in paid work.

Altogether, current research shows that health is an important factor for staying in work at older age [27]. Globally, women have higher rates of morbidity [28]. In addition, SA rates are higher among women than men in many European countries [29, 30]. DP is also more common in women, at least in Nordic countries [31,32,33].

Sickness absence, disability pension, and working past age 65

Some, but not most, morbidity affects people’s work capacity and can thus lead to the need for SA, or even DP if the impaired work capacity is permanent. Thus, SA and DP are not measures of morbidity per se but rather the consequences of morbidity, in terms of work incapacity. Most people with different types of diagnoses are not sickness absent due to them, as their morbidity does not affect their work capacity regarding their actual work tasks to the extent that SA/DP is needed [34]. It is, therefore, of interest to explore to what extent people are in any paid work after age 65, despite such previous (temporary or permanent) reduction in work capacity – i.e., SA or DP.

Currently, there are only a few studies on the associations of SA and DP with working past 65. Only one of these studies included information on SA diagnoses, and found that SA due to mental diagnoses was much less common after the age of 65 than in the ages 60–64, whereas the differences for somatic diagnoses were smaller [17]. Given this finding, and previous findings that show that SA spells due to mental diagnoses tend to have longer durations compared to most other SA spells [35, 36], this raises the question of whether SA or DP due to mental diagnoses could be a greater risk factor for not being in paid work after the standard retirement age than SA or DP due to somatic diagnoses. Currently, knowledge is lacking in the area.

The aim of this study was to examine the associations between having been on SA or DP due to mental, somatic, or both mental and somatic diagnoses when aged 60–64 with being in paid work at some point when aged 66–71, among women and men, adjusting for sociodemographic factors.

Methods

Study design

This prospective population-based cohort study examined microdata, linked at individual level, from three nationwide registers: Statistics Sweden’s LISA register (covering sociodemographic and income variables), the Social Insurance Agency’s MiDAS register (information on SA and DP, including main diagnoses according to the International Classification of Diseases (ICD-10) [37]), and the National Board of Health and Welfare’s Cause of Death register. Data for the years 2005 to 2016 were used, with 2010 as a baseline; 2005–2009 was regarded as the exposure interval and 2011–2016 as the outcome interval. The base population was all the 123,738 individuals (50.0% women) who turned 65 in 2010 and lived in Sweden in December 2010.

From this group we included everyone who had an income from work and/or received work-related benefits (SA benefits, parental leave allowance and/or work injury compensation from the National Social Insurance Agency) high enough to qualify for SA benefits in any of the five years prior to 2010 (2005–2009). In order to be eligible for SA benefits, a minimum annual work income of 24% of the price base amount (a government set amount that adjusts to annual inflation and is used to calculate different types of public benefits) was required. In order to not exclude people on long-term SA or DP, we adjusted the minimum income requirement to 75% of the minimum annual work income for SA benefits, as the SA benefit in most cases covers 75% of the lost income. This limit was an annual income of 7092 SEK in 2005 and 7704 SEK in 2009 (≈787–711 Euros/yearFootnote 1). After the applied inclusion criteria, the study cohort included 98,551 individuals (48.2% women).

Social insurance in Sweden

In Sweden, all individuals with income from work or unemployment benefits are covered by the public sickness insurance system and can receive SA benefits if they have reduced work capacity due to disease or injury. Among employees, employers provide sick pay for the first 14 days of a SA spell, thereafter the Social Insurance Agency provides benefits. After the first seven days of a SA spell, a physician certificate, including information on diagnosis, is required. There is no time limit for the duration of SA spells, and some can last for several years. After the age of 65, some restrictions apply. Individuals aged 65–69 years can get SA benefits for up to a total of 180 days during those years, after which the Social Insurance Agency may restrict further claims if the reduced work capacity is assessed as permanent. From the age of 70, people may not claim SA benefits for more than 180 days total. All people aged 19–64 with permanently reduced work capacity due to disease or injury can be granted DP. SA benefits and DP can be granted for full- or part-time: 100, 75, 50, or 25% of ordinary work hours [38]. SA benefits cover 80% and DP about 65% of lost income, up to a certain level [39].

In Sweden, the standard retirement age was 65 years in the study period. However, the age of retirement was flexible and could start as early as 61. Most employees had the right to keep their permanent position until age 67, after which the choice to retain or terminate their employment was up to the employer. Despite the flexible pension system being designed to enable and encourage people to extend their working lives, a considerable number of people in Sweden retire before or at the age of 65 [40]. Still, many continue with some type of work.

Exposures

The exposures were SA or DP due to mental and/or somatic diagnoses when aged 60–64 years (in the years 2005–2009). Only SA spells exceeding 14 days were included. SA and DP diagnoses were categorized in the following way:

  • mental diagnosis (ICD-10: F00-F99 and Z73),

  • somatic diagnosis (ICD-10: all other diagnoses, including missing diagnoses),

  • both mental and somatic diagnoses,

  • no SA or DP (reference category).

As SA and DP spells in many instances overlap (due to the possibility of partial SA and DP) or follow one another successively (in the working population those granted DP have first been on SA), we also conducted analyses where we used a combined variable of SA and DP. The categories were then as shown above but SA and DP were combined (SA and/or DP due to mental, somatic, or both mental and somatic diagnoses, as well as neither SA nor DP).

Outcome

The outcome variable was whether or not an individual was in paid work at some point in 2011–2016, (i.e., when aged 66–71), defined as having an income from work or work-related benefits high enough to qualify for SA as previously described (75% of 24% of the price base amount, i.e., 7704–7974 SEK/year ≈861–873 Euros/yearFootnote 2).

Covariates

The following sociodemographic factors were adjusted for in the analyses: sex (woman or man), country of birth (Sweden, other Nordic countries, EU-27 countries (except Denmark, Finland, and Sweden), or rest of the world (all other countries, including missing)), type of living area (large cities (Stockholm, Gothenburg, Malmö), medium-sized cities (> 90,000 inhabitants within 30 km of municipal centre), or small towns/rural areas (< 90,000 inhabitants within 30 km of municipal centre)), level of education (elementary (0–9 years, including missing), high school (10–12 years), or university/college (> 12 years)), and family situation (married or cohabiting without children under 18 years living at home, married or cohabiting with children under 18 years living at home, single without children under 18 years living at home, or single with children under 18 years living at home). Information on covariates was obtained from LISA and applies to 2010.

Statistical analysis

Descriptive analyses were conducted. Logistic regressions were performed to calculate crude and mutually adjusted odds ratios (OR) and 95% confidence intervals (95% CI), for associations between previous SA and/or DP (when aged 60–64) in specific diagnosis groups and being in paid work when aged 66–71, crude and adjusting for sociodemographic factors. First SA and DP were considered as separate, and then as a combined variable. All analyses were conducted for all and also stratified by sex. Moreover, Nagelkerke R Squares and the Omnibus Tests of Model Coefficients were calculated. All statistical analyses were performed with IBM SPSS Statistics version 27.

Results

Characteristics of the study cohort are presented in Table 1. About one third of all individuals had been on SA at some point when aged 60–64; 40.8% of the women and 31.3% of the men. Few individuals had been on DP (7.9%) at some point during this time period. Somatic diagnoses were the most common cause for DP (78.6% of all DP) and the clear leading cause for SA (84.6% of all SA).

Table 1 Characteristics of the cohort of people in Sweden in paid work when aged 60–64
Full size table

The percentage in paid work when aged 66–71 (48.2% of all in the cohort) in each sociodemographic or diagnosis group are presented in Table 2. A larger proportion of those who had not been on SA (50.6% compared to 40.5–44.3%, depending on diagnosis group), respectively not on DP (50.2% compared to 20.9–27.2%) when aged 60–64, were in paid work after age 65.

Table 2 Percentages in paid work when aged 66–71 among those who were in paid work when aged 60–64
Full size table

Results from the logistic regression treating SA and DP as separate predictors conducted among all individuals are presented in Table 3. In this analysis, men were more likely to be in paid work compared to women. Results from the corresponding sex-stratified analyses are presented in Table 4.

Table 3 Odds ratios for different previous factors for being in paid work at 66–71
Full size table
Table 4 Odds ratios for different previous factors for paid work at 66–71; sex-stratified
Full size table

Sickness absence

Generally, having been on SA when aged 60–64 was associated with a lower likelihood of being in paid work when aged 66–71. SA due to mental diagnoses was associated with the lowest likelihood of being in paid work compared to no SA among all individuals (OR: 0.77; 95% CI: 0.72–0.83). Having been on SA due to both somatic and mental diagnoses showed no significant association with the likelihood of being in paid work past the age of 65 among all (0.95; 0.87–1.04) nor among women (0.98; 0.88–1.09), yet there was a negative association among men (0.77; 0.65–0.91).

Disability pension

Having been on full- or part-time DP at some point when aged 60–64 was associated with a lower likelihood of working after age 65. The results from non-stratified and sex-stratified analyses were strongly in line with each other. DP due to mental diagnoses was associated with a much lower OR (0.38; 95% CI: 0.34–0.43) of being in paid work after 65 compared to those without DP. The corresponding ORs regarding DP due to somatic diagnoses were comparable. People with DP due to both mental and somatic diagnoses were the least likely to be in paid work after the age of 65, compared to those without DP.

Sickness absence and/or disability pension

Results with the combined variable of SA and/or DP among all are shown in Table 5, and the corresponding sex-stratified analyses are shown in Table 6. The combined variable of SA and DP demonstrated ORs that largely lay in-between those of the diagnosis groups of SA and DP alone, especially for men.

Table 5 Odds ratios for SA and DP combined and being in paid work at 66–71
Full size table
Table 6 Odds ratios for SA and DP combined and being in paid work at 66–71; sex-stratified
Full size table

Discussion

In this prospective population-based cohort study of all 98,551 individuals in Sweden who turned 65 in 2010 and had been in paid work at some point when aged 60–64, nearly half were in paid work at some point when aged 66–71. Prior SA was negatively associated with being in paid work. However, SA demonstrated somewhat different strengths of associations depending on sex and SA diagnosis group, whereas DP had similar associations with paid work irrespective of sex and DP diagnosis group (e.g., somatic DP among all 0.36; 0.34–0.39). When combining SA and DP, the ORs largely lay in-between those of the respective diagnosis groups of SA and DP alone. To the best of our knowledge, this study is the first to explore the association between specific SA and DP diagnosis groups and being in paid work after the age of 65.

Sickness absence

Both women and men with previous SA due to mental diagnoses were less likely to be in paid work after age 65 than those with previous SA due to somatic diagnoses, which was also found when conducting analyses on the combined variable of SA and/or DP. More knowledge is warranted on possible reasons for this result, and whether this differs by type of occupation or type of mental diagnosis. It might be that the current stigma of mental disorders [41] decreases the likelihood of being in paid work, e.g., through hesitations to ask for relevant work adjustments. The result might also be affected by the fact that, at least in Sweden, SA spells due to mental diagnoses can last for a substantially longer time compared to SA due to most other conditions [35], which in itself might be a hinder for returning to work [42]. Thus, future studies should include also information on SA duration.

SA due to somatic diagnoses when aged 60–64 was also associated with not being in paid work, however, this association was less strong than that for SA due to mental diagnoses. In Sweden, like in many other countries, mental and musculoskeletal diagnoses are the most common reason for SA [43]. Among the musculoskeletal diagnoses, low back and neck pain are especially common [44], which might be a result of physically straining work conditions. Physically straining conditions, in turn, might hinder older people with somatic diagnoses to be in paid work, which could possibly be one reason for the decrease in work participation after 65. This hypothesis is supported by the result that people with lower education, who are more likely to work under physically straining conditions [45, 46], were less likely to be in paid work after 65.

At the same time, the absence of somatic SA could indicate a comparatively good work capacity among those in the older working force. Having a good work capacity might possibly imply a relatively good health status among these individuals, which in turn has been seen to predict working beyond the standard retirement age [23]. Still, it is important to note that absence of SA does not equal absence of morbidity [34]. Most morbidity does not affect people’s work capacity to the extent that they require SA or DP. Altogether, the results from this study indicate that somatic SA in the years leading up to the standard retirement age is associated with a lower likelihood of paid work after 65, although the association is perhaps not especially strong. These results are in line with prior research using other, broader measures of morbidity.

Surprisingly, prior SA due to both somatic and mental diagnoses did not show a significant association with being in paid work in the non-stratified analyses nor among women, whereas it was associated with a lower likelihood among men. Notably, the number of men was almost a one-third of the number of women within the group of SA due to both somatic and mental diagnoses. This surprising finding requires further research to explain. Future research could investigate to what degree the people who had previous SA in both somatic and mental diagnoses remained in paid work, whether they are in paid work for a shorter duration of time or work part-time to a greater extent, etc. Future research could also examine the duration and timing of the respective SA spells, and examine this in relation to the morbidities that underlie the SA spells.

Disability pension

Unsurprisingly, prior DP was a risk factor for not being in paid work after 65 among both women and men, irrespective of diagnosis group. Nevertheless, nearly 2000 individuals with DP when aged 60–64 were in paid work after 65, i.e., approximately 25% of those with prior DP. This can seem like a very high proportion, however, it must be seen in the light of that partial DP is possible in Sweden, unlike in many other countries. Thus, some of these individuals might have been on DP for, e.g., 25% of ordinary work hours, and in part-time work the rest of the time.

Especially at older ages, when different types of morbidity are more common, work adaptations can be crucial to facilitate a prolonged working life. A recent Swedish governmental inquiry on how to promote extended working lives suggested flexible work hours, reduced work hours, and the possibility to adapt work tasks based on one’s capacity [47]. However, for those who have previously been granted DP, possibilities for work adjustments have already been assessed, and been judged as not possible.

Strengths and limitations

Strengths of this study are the prospective cohort study design, that microdata, linked at individual level from several high-quality administrative registers [48,49,50] could be used for all the people in a country who fulfilled the inclusion criteria (i.e., not a sample), the large cohort allowing for sub-group analyses, that data were not self-reported (i.e., not hampered by recall bias), that there was no non-response or drop-outs, and the long study period (12 years). Another strength is that we did not include shorter SA spells (≤14 days), which means that we only include those SA spells that led to a substantial reduction of work capacity for at least more than two weeks. Consequently, shorter, often self-certified SA spells due to, e.g., the common cold, stomach flu, or a sprained ankle were not included. Also, the SA and DP diagnoses had been assessed by the patient’s treating physician and approved by the Social Insurance Agency. Additionally, we included income from work irrespective of type of employment. This also includes self-employment, which is important as many who work beyond 65 are self-employed [18].

Like every study, this study has some limitations that should be taken into consideration. We had information on income from paid work, however, not on hours worked nor on type of work. There are multiple definitions of being in paid work in the literature, based on e.g., self-reported employment status, hours worked, income, registered employment [51,52,53]. Our definition was based on having had registered income from work, in order to include most individuals with some degree of paid work, as many of those who work past age 65 work part-time [18, 54]. Consequently, this study’s definition of paid work was very inclusive, which could both pose as a strength and a limitation. It does mean that several heterogeneous phenomena are captured using this outcome measure. In addition, future studies might want to differentiate analyses by levels of income or number of years with earnings when aged 66–71, and differentiate between those who remain continuously in work and those who stop working and then return.

By also including those who had income from paid work at any point when aged 60–64, we avoided the health selection that can arise from excluding those who leave work via full or part-time old-age pension before the standard retirement age of 65. Also, many continue in paid work after having taken early old-age pension and can then apply for SA. However, those who retired early were also less likely to have SA or DP during the exposure window, since some of them were at risk of SA and DP for a shorter time than those who were in paid work throughout the exposure period. However, they may still have had SA before the exposure window. This means that we might have underestimated the association between SA before age 65 and not working after age 65. Since DP is granted on a permanent basis until a person turns 65, this is less likely to be an issue regarding DP.

In addition, it should be noted that the Nagelkerke R Squares were quite low (see Tables 3 and 4), indicating that besides prior SA, DP, and the sociodemographic covariates, there were other factors that can explain work participation past the age of 65 than those accounted for in this model. Such factors could, e.g., be influence by significant others [15, 55], the employer [11], societal and group norms regarding retirement [56], work environment [22, 51], motivation to continue in paid work [22], personality [57], type of morbidity, as well as duration, rate (part- or full-time), and number of SA and DP spells. Moreover, one limitation is the possible overlap between SA and DP, which might affect the results of the analysis in which SA and DP were separate exposure variables. This is also the reason why we conducted analyses with a combined variable of SA and DP.

Further, the measures of SA and DP are also relatively crude, by bundling diagnoses together into mental respectively somatic diagnoses, and not distinguishing between part-time and full-time SA and DP or having SA or DP early or late in the study period. As a result, as with work, heterogeneous phenomena may be included in each of the groups, which may lead to an over- or underestimation of the association between SA/DP and paid work after age 65. Future studies should investigate if results vary with more specific SA and DP diagnoses, with SA/DP timing, length, and rate, as well as with e.g., specific morbidities and in specific occupations.

Still, since all individuals in Sweden fulfilling the inclusion criteria were included in this study, the results can be assumed to at least be generalizable to countries with a similar public SA, DP, and old-age pension system.

Conclusions

This study showed that previous SA due to mental diagnoses was negatively associated with being in paid work after the age of 65; the same was the case for previous SA due to somatic diagnoses, although that association was less strong. The likelihood of being in paid work was the lowest among individuals with prior DP, regardless of diagnosis group and sex. The findings fill a clear knowledge gap and indicate that mental disorders among older workers are an important aspect to consider when striving to extend people’s working lives. More knowledge of factors that affect the likelihood of being in paid work provide opportunities to target important diagnosis groups, further increasing participation in paid work in the growing group of older persons.