Working life sequences over the life course among 9269 women and men in Sweden; a prospective cohort study

Katalin Gémes; Katriina Heikkilä; Kristina Alexanderson; Kristin Farrants; Ellenor Mittendorfer-Rutz; Marianna Virtanen

doi:10.1371/journal.pone.0281056

. 2023 Feb 15;18(2):e0281056. doi: 10.1371/journal.pone.0281056

Working life sequences over the life course among 9269 women and men in Sweden; a prospective cohort study

Katalin Gémes ^1,^*, Katriina Heikkilä ^1,², Kristina Alexanderson ¹, Kristin Farrants ¹, Ellenor Mittendorfer-Rutz ¹, Marianna Virtanen ^1,³

Editor: Sreeram V Ramagopalan⁴

PMCID: PMC9931102 PMID: 36791056

Abstract

Objectives

To investigate working life courses in women and men and possible associations with socioeconomic, health-, and work-related factors.

Methods

A 15-year prospective cohort study of individuals aged 18–50 in paid work at baseline and answering the Swedish Living Conditions Surveys (2000–2003, N = 9269) and their annual economic activity, using nationwide registers. We used sequence and cluster analyses to identify and group similar working life sequences. Multinomial logistic regression was used to examine associations of sex, socioeconomic, health-, and work-related factors with sequence cluster memberships.

Results

We identified 1284 working life sequences, of which 65% represented continuous active (in paid work/studying) states. We then identified five sequence clusters, the largest one with individuals who were continuously active (n = 6034, 65% of the participants; 54% of women and 76% of men) and smaller ones with interruptions of the active state by long-term parental-leave, unemployment, and/or sickness absence/disability pension (SA/DP), or retirement. Women were more likely than men to belong to the “Parental-leave periods” (odds ratio [OR]: 33.2; 95% confidence intervals [CI]: 25.6, 43.1) and the “SA/DP periods” sequence clusters (OR: 1.8; 95% CI: 1.4, 2.1), also after adjustment for covariates. In both sexes, low education and poor health were the strongest predictors of belonging to the sequence cluster “Unemployment & SA/DP periods”. Predictors of the “Parental-leave periods” sequence cluster differed between women and men.

Conclusions

In a cohort of individuals in paid work at baseline, the majority of women and men worked most of each year although women were more likely to have some interruptions characterized by long-term parental-leave or SA/DP periods than men, independently of socioeconomic, health-, and work-related factors.

Introduction

In high-income countries, increased life expectancy and low birth rates have led to an increase in the old-age dependency ratio, i.e., there are fewer workers in the labour market to cover the increasing healthcare and social security costs [1]. Many governments have started to reform public pension systems in order to extend working lives and avoid ‘the pensions crisis’, e.g., by raising the retirement age. Extending working life, and ensuring good health and active participation in the labour market for all are crucial to succeeding with such policy changes [2]. However, there is a gap between women and men in the duration of working life; men seem to work more years than women and women seem to have lower working life participation rates throughout the life course [2, 3] This can be related to various reasons, including that women often spend longer time with unpaid work, such as caring for children or other family members, have higher morbidity during working age, or take old-age pension earlier than men [2, 4]. Furthermore, more women work in temporary or part-time jobs, often jobs characterised by unstable and insecure labour market positions [2, 5, 6] compared to men. Furthermore, women’s jobs are more likely to be characterised by poorer career prospects, higher psychosocial workload [7, 8], poorer control over working hours [9], and higher rate of workplace bullying and harassment [10, 11]. On the other hand, men more often work longer hours, report higher effort-reward imbalance, lower social support, and higher physical demands at work [7]. Many of these work characteristics have been shown to be associated with interruptions in working-life participation or early exit from the labour market by means of long-term sickness absence (SA) and disability pension (DP) [10–15]. However, studies mapping the relationship between socioeconomic, health-, work- related factors and diverse working life courses, for example, those including longer health-, work-, and family-related career interruptions, are scarce. A major reason for this is the limited availability of good quality data which can combine information from representative surveys and administrative registers. The assessment of perceived physical and psychosocial work-characteristics, self-rated health, and lifestyle requires survey-based methods while administrative registers are important sources of information on annual states regarding employment-, family and health related benefits, without the problem of drop out during the follow-up.

Some of the previous research focused on the examination of the inter- and intra-individual diversity of work-family life courses [3, 16–20]. McMunn at al., [16] examined the changes in work-family life courses over time and showed that women’s work-family life courses converged to men’s while the diversity of the individual work-family life courses increased. They also described that those work-family life courses which indicated a weaker link to paid work, e.g., having early childbirth without employment, which was more common in women than men, was found to be a predictor of poor mental health [17], lower subjective well-being [18], lower quality of life [19], worse body mass index trajectories [20], and also a lower probability of later employment [3]. However, as the focus was on the consequences of work-family life courses, these studies did not include health-related breaks in their assessment, such as long-term SA and DP. According to the integrated life-course perspective proposed by Amick et al. [21], both labour market and health transitions create and shape an individual’s working life course. Health-, work-, and family-related states and transitions are interrelated and the individual’s working life course takes place within the labour market and social contexts.

Sequence analysis is an advanced analytical tool to capture the complexity of working life courses [22, 23] and it has been used in previous research in relation to work-family life courses [3, 17–20], Therefore, we will use this analytical approach in our study to examine sequences of family-, work-, and health-related life courses simultaneously. In line with the integrated life course perspective, we aimed to map 15 years of working life courses among individuals aged 18 to 50 and in paid work at inclusion, and among them, to identify groups with similar working life sequences, and to examine whether sex was associated with working life sequences, independently of socioeconomic, health-, and work-related conditions. Furthermore, we aimed to examine which of these factors were associated with the membership of specific working-life sequence groups separately among women and men. The specific research questions were to explore: 1) which types of overarching future working life sequence groups can be identified during a 15-year period in a cohort of individuals in paid work at baseline? 2) whether women have more often than men working life courses characterized by several interruptions, such as parental leave, SA/DP, or unemployment? 3) whether adjusting for baseline sociodemographic, health- and work-related factors weaken the possible associations between sex and the specific sequence cluster membership? 4) whether these factors are independently associated with the specific sequence cluster membership in a separate analysis among women and men?

Materials and methods

The research was approved by the Regional Ethical Review Board, Stockholm, Sweden, who also waived the need for informed consent [24, 25].

A prospective cohort study was conducted using microdata from several sources. First, the Living Conditions Surveys (ULF/later SILC: Undersökningarna av levnadsförhållanden/EU Statistics on Income and Living Conditions) [26] was used. This is an annual, mainly phone-based survey by Statistics Sweden on a random sample of individuals aged 16 to 84 and living in Sweden. We used data from the surveys of 2000, 2001, 2002, and 2003. Overall, 5678 responded in 2000 (76% of invited), 5805 in 2001 (78% of invited), 6322 in 2002 (75% of invited), and 6363 in 2003 (75% of invited). In our study, we included respondents who were aged 18–50 years at survey date, who were in paid work (i.e., full- or part-time employed or self-employed), and had been living in Sweden for at least two years, i.e., the survey year and the preceding year, and had no missing information on variables we were interested in (S1 Fig).

Statistics Sweden linked the survey data to microdata from nationwide registers, using the unique identification numbers assigned to all registered residents in Sweden [27], and anonymized the dataset. Register data were obtained for the year before the survey and annually for up to 15 years after. Follow-up began at the beginning of the year following the survey year. Individuals who emigrated from Sweden or died during the follow-up period (n = 252) were excluded (S1 Fig).

Survey data

Economic hardship (having difficulties in making ends meet during the last year: yes; no), working conditions such as number of weekly working hours (≤35; >35 and ≤45; >45 hours per week), exposure to noise at work (yes, to medium-strong noise for substantial amount of time; no or only light noise), hectic work schedule (yes; no), mentally strenuous job (yes; no), physically strenuous job (yes; no), possibility to learn new things at work (yes; no), and having had a work accident in the past 12 months (yes; no). Health and health-related behaviors included self-rated health (good/average; poor), having long-term illness or long-term health problem (yes; no), smoking (daily smoker; not a daily smoker), overweight/obesity (body mass index, BMI ≤25 kg/m²; >25 kg/m²).

Register data

Baseline sociodemographic variables, annual labor market status, and year of emigration were obtained from the Longitudinal Integrated Database for Health Insurance and Labor Market Studies (LISA) held by Statistic Sweden [28]. Information on sickness absence/disability pension (SA/DP) spells (start and end dates and yearly net days) were obtained from the Micro-Data for Analysis of the Social Insurance System (MiDAS) held by the Social Insurance Agency [29]. Year of death from the Cause of Death Register held by the National Board of Health and Welfare.

We considered the following register-based sociodemographic factors at baseline: sex (woman; man), age (<31; 31–40; >40 years), nationality (Swedish; not Swedish), both parents/participants themselves being born outside Sweden (yes/no); educational level (compulsory; secondary; university/college), type of living area (Stockholm; Gothenburg/Malmö; other large cities; middle size cities, small towns and rural areas), family situation (single without children aged <18 years; married or cohabitating without children aged <18 years; single with children aged <18 years; married or cohabiting and living with children aged <18 years), and a history of long-term SA/DP (>183 net days of SA and/or DP during the survey year or the preceding year: yes/no).

The following mutually exclusive annual labor market states were defined: “active” if the participant received income from work or student benefit and was not categorized to the other states;”unemployed” if the participant had unemployment benefit >183 days/year; “on parental leave” if the participant had parental-leave benefits >183 days/year; “on SA/DP” if the participant had SA and/or DP benefits >183 net days/year; and “retired” if >50% of the annual income was from old-age pension.

Social security in Sweden

In Sweden, people aged at least 16 years and having income from work, unemployment benefits, or parental-leave benefits are eligible for SA benefit if morbidity leads to work incapacity. The first SA day is a waiting day without compensation; from the eighth day of a SA spell, a medical certificate from the treating physician is required [30]. For employed people, the employer pays the SA benefit for days 2–14, after that, the Swedish Social Insurance Agency pays the benefit. For the unemployed, this happens from the second day of the SA spell. In order to not to introduce bias, we used information on SA spells >14 days. All people in Sweden aged 19–64 can be granted DP if morbidity has led to long-term or permanent reduction of work capacity. Both SA and DP can be granted for full- or part-time of ordinary work hours. Everyone over the age of 16, capable of working, and registered as a jobseeker is eligible for an unemployment benefit [31]. Sweden, similar to other Nordic countries, follows the dual-earner family policy model where both parents are supposed to be involved in both labour market activity and domestic work, and both parents are entitled to extensive number of days with parental-leave benefits [32]. In the years studied, parental-leave benefits could be obtained for 480 days per child until the child was ten years. Some of these days were earmarked for each respective parent, for the rest of the days, the parents could decide how to share [32]. Tax-financed child care is available for children aged 1 to 10 years for a small, income-based fee. The median age at the birth of the first child in Sweden was 28.2 for women and 30.7 for men in 2000, during the first survey year and it was 29.2 and 32.5, respectively, in 2016, the last follow-up year. In 2000, the fertility rate in Sweden was 1.5 children and increased to 1.8 by 2016 [33]. There is no fixed retirement age in Sweden although in the years studied, one could apply for old-age pension from the age of 61; the usual retirement age was 65 years.

Statistical analysis

Differences between women and men regarding socioeconomic, work, and health-related factors were described by calculating proportions and chi² statistics. We then conducted sequence analyses to identify temporal sequences of annual labour market activity states and to group them into empirically distinct sequence clusters. A sequence was defined as an ordered list of elements (here, several possible labour market activity states on a yearly basis) [22] over a period of 15 years. The relative proportion of each of the five states for each year was displayed in state proportion plots. The mean duration of years within a given labour market activity state and the mean number of episodes of different states were also calculated. To group similar sequences, we conducted weighted cluster analyses. First, we calculated dissimilarity measures between the individual sequences by using optimal matching algorithms defined by the cost of transforming one sequence into another [21, 22, 34]. For the dissimilarity measures, a cost of one was assigned to insertion and deletion of states, and for substitution, the cost of transition was based on values of the calculated transition matrix. Rare transitions contributed to a higher cost, and frequent transitions to a lower cost. Second, we used hierarchical agglomerative cluster algorithm with Ward’s linkage to grouped similar sequences to clusters [21, 22, 34]. Third, we evaluated the resulting 2–10 cluster solutions by using several partition quality measures listed in S1 Table and by looking at the hierarchical cluster tree diagrams of the resulting numbers and membership sizes of the clusters and the suggested sequence patterns. We also tested the constant substitution cost and partition around medoids (PAM) clustering algorithms, to see whether they resulted in similar cluster solutions.

After deciding the optimal sequence cluster solution, assigning all individuals to one of the clusters, and creating sequence index plots by clusters, we used multinomial logistic regression analysis to examine the associations between sex and the sequence cluster membership. To examine whether the associations between sex and cluster memberships attenuated, several sets of covariates were gradually included in the models. Model 1 was adjusted for age only, model 2 additionally included all socioeconomics; model 3, included model 2 plus the health-related covariates, and model 4 included model 3 plus the work-related covariates. We calculated odds ratios (OR) with 95% confidence intervals (CI) to compare sequence cluster memberships, using men and the cluster with the largest number of individuals as the reference categories. Thereafter, we investigated the associations between covariates and sequence cluster memberships separately among women and men. In sensitivity analyses we showed other possible cluster solutions a solution for one cluster less and one cluster more compared to the chosen one. To explore whether these cluster solutions were similar among women and men, we presented the results from cluster analysis stratified by sex. Furthermore, we ran the multivariable adjusted analysis between sex and cluster memberships with these alternative cluster solutions. We also reran the main analysis by adjusting the models for those variables at baseline that can be assumed to be stable over time for the majority of individuals, such as education, baseline age, birth country, nationality, living area and SA/DP history.

Data management was performed with SAS (version: 9.4) and statistical analysis with R (version 4.0.5) using the TraMineR [35] and WeightedCluster [34] packages.

Results

Descriptive statistics of the study variables are provided in Table 1. Compared with men, women, on average, had higher education, were more often single with children, worked shorter hours, were less likely to have physically strenuous work and more likely to have mentally strenuous work.

Table 1. Baseline characteristic of the study participants.

	All (n = 9269)	Women(n = 4516)	Men(n = 4753)	p ^*
Sociodemographic, socioeconomic (n, (%))
Age
<31 years	2760 (29.8)	1293 (28.6)	1467 (30.9)	0.01
31–40 years	3232 (34.9	1562 (34.6)	1670 (35.1)
>40 years	3277 (35.4)	1661 (36.8)	1616 (34.0)
Educational level
Compulsory	1058 (11.4)	409 (9.1)	649 (13.7)	<0.01
Secondary	4997 (53.9)	2347 (52.0)	2650 (55.8)
University, college	3214 (34.7)	1760 (39.0)	1454 (30.6)
Type of living area
Stockholm	1900 (20.5)	946 (20.9)	954 (20.1)	0.08
Gothenburg/Malmö	1468 (15.8)	729 (16.1)	739 (15.5)
Other larger cities	3335 (36.0)	1647 (36.5)	1688 (35.5)
Small, middle-sized cities, rural	2566 (27.7)	1194 (26.4)	1372 (28.9)
Family situation
Living alone without children	1865 (20.1)	664 (14.7)	1201 (25.3)	<0.01
Married/cohabiting with children <18 years	4503 (48.6)	2275 (50.4)	2228 (46.9)
Married/cohabiting without children	2314 (25.0)	1151 (25.5)	1163 (24.5)
Living alone with children <18 years	587 (6.3)	426 (9.4)	161 (3.4)
Not Swedish citizen	387 (4.2)	199 (4.4)	188 (4.0)	0.30
Both parents/participants themselves being born outside Sweden	1201 (13.0)	608 (13.5)	593 (12.5)	0.17
Experiencing economic hardship	1484 (16.0)	757 (16.8)	727 (15.3)	0.06
Health-related factors (n (%))
Poor self-rated health	232 (2.5)	122 (2.7)	110 (2.3)	0.26
Previous SA/DP >183 net days	191 (2.1)	118 (2.6)	73 (1.5)	<0.01
Long-term illness or health-problem	3145 (33.9)	1598 (35.4)	1547 (32.5)	<0.01
Daily smoker	1609 (17.4)	913 (20.2)	696 (14.6)	<0.01
Overweight/obese (BMI >25 kg/m²)	3608 (38.9)	1284 (28.4)	2324 (48.9)	<0.01
Working conditions (n (%))
Weekly working hours
36–44	6130 (66.1)	2643 (58.5)	3487 (73.4)	<0.01
≤35	2067 (22.3)	1634 (36.2)	433 (9.1)
≥45	1072 (11.6)	239 (5.3)	833 (17.5)
Physically strenuous work	2512 (27.1)	1001 (22.2)	1511 (31.8)	<0.01
Exposed to noise	2614 (28.2)	870 (19.3)	1744 (36.7)	<0.01
Mentally strenuous job	4268 (46.0)	2310 (51.2)	1958 (41.2)	<0.01
Monotonous job	3852 (41.6)	1940 (43.0)	1912 (40.2)	0.01
Hectic work	6667 (71.9)	3317 (73.4)	3350 (70.5)	<0.01
Little/no opportunity to learn new things at work	6650 (71.7)	3154 (69.8)	3496 (73.6)	<0.01
Work-related accident	776 (8.4)	298 (6.6)	478 (10.1)	<0.01

Open in a new tab

*P-value for difference in χ²-test for categorical variables and two-tailed Student’s t-test for age.SA: Sickness absence; DP: disability pension

We identified 1284 unique future labour market sequences among the participants (Figs 1, 2). The sequence with the largest number of participants included “active” states during the whole follow-up period (65.1%). Other common sequences were “active” states alternating with “on parental leave”, “unemployed”, and “on SA/DP states (Figs 1, 2 and S2, S3 Figs). With regard to transitions to the “active” state, the most likely transitions were staying “active” (“active”-“active”; 97%) and returning to an “active” state after “on parental leave” state (“on parental leave”-“active”; 83%) (S2 Table). The lowest probability of transition was returning to the “active” state from the “retired” state (0.1%) (S2 Table).

The cluster partition quality measures showed the best fits for a five- and a seven-cluster solution (lowest point biseral correlation for five clusters and highest Hubert’s Gamma, Average Silhouette Width, and Calinszki-Habarasz indexes for the seven-clusters solution) but only with a minimal difference in indexes between the five- and seven-cluster solutions (S1 Table, S4 and S5 Figs). To consider the feasibility and readability of the multinomial regression analyses, we chose the five-sequence cluster solutions for further analysis (Fig 3). In the five-cluster solution, the mean time spent in each state is presented in S6 Fig. Four-, six-, and seven-cluster solutions are presented in S7–S9 Figs. Most participants (n = 6034; 65.1%; 53.9% of women, 75.8% of men) belonged to the “Active” cluster, which comprised exclusively “active” state sequences during the entire follow-up. Sequences which combined mostly “active” states with some “on parental leave” states were also identified as a cluster (n = 1179; 12.7%; 24.4% of women and 1.6% of men), and this was named as “Parental-leave periods”. The third largest cluster included a mix of sequences, combining mostly “active” states with “unemployed”, and “on SA/DP” states and was named “Unemployment & SA/DP periods” cluster (n = 1018; 11.0%; 10.0% of women and 12.5% of men). Finally, we identified two relatively small clusters, one with sequences including “on SA/DP” states besides “active” states (n = 710; 7.7%; 9.1% of women and 6.3% of men), therefore, named “SA/DP periods” cluster and one with sequences including “active” and “retired” sequence states (n = 328; 3.5%; 3.2% of women and 3.8% of men), and named as “Retirement”. Using other types of distance measures and cluster algorithms resulted in similar cluster distributions as was found with a five-cluster solution that showed a good cluster partition quality and identified similar clusters.

When the cluster analysis was stratified by sex, we found a four-cluster solution to have the best fit for women and a four- and five-clusters solutions to have the best fit for men (S3 Table, S10 Fig). We present the four clusters solutions for both women and men with mean time spent in each state (S11–S14 Figs).

The results regarding the association between sex and cluster membership are presented in Table 2. The “Active” cluster was used as the reference category and the ORs and 95% CIs for the membership of another cluster were calculated for women, with men as the reference category. Adjustment for different sets of variables did not change the estimates considerably. After adjusting for all covariates, women were still 33.2 times more likely to belong to the “Parental-leave periods” cluster (95% CI: 25.6 to 43.1) and 1.8 times more likely to belong to the “SA/DP periods” cluster than men (95% CI: 1.4 to 2.1). There was no sex difference regarding the memberships of the “Unemployment & SA/DP periods” (adjusted OR: 1.0, 95% CI: 0.9 to 1.2) and the “Retirement” cluster (adjusted OR: 0.9, 95% CI: 0.7 to 1.1). When the models were adjusted for covariates that were time-fixed, we got similar estimates (adjusted OR for “Parental leave cluster”: 32.1, 95% CI:25.0 to 41.2; adjusted OR for “SA/DP periods”: 1.8, 95% CI: 1.5 to 2.1; adjusted OR for “Unemployment & SA/DP periods”: 1.1, 95% CI: 1.0 to 1.3; and adjusted OR for “Retirement” was 1.0, 95% CI: 0.8 to 1.2). The results of the baseline and multivariable-adjusted sensitivity analyses for the four and six clusters solutions are presented in S4 Table.

Table 2. Multinomial regression analysis for the associations between sex and different working life sequence clusters.

Cluster name	Number of individuals	Model 1	Model 2	Model 3	Model 4
Cluster name	Number of individuals	OR (95% CI) for women compared with men	OR (95%CI) for women compared with men	OR (95%CI) for women compared with men	OR (95%CI) for women compared with men
Active	6034	Ref.	Ref.	Ref.	Ref.
Unemployment & SA/DP periods	1018	1.1 (0.9, 1.3)	1.2 (1.0, 1.4)	1.2 (1.0, 1.3)	1.0 (0.9, 1.2)
Parental leave periods	1179	33.3 (26.0, 42.7)	30.7 (23.9, 39.4)	32.1 (24.9, 41.4)	33.2 (25.6, 43.1)
SA/DP periods	710	1.8 (1.5, 2.1)	1.9 (1.6, 2.2)	1.9 (1.6, 2.2)	1.8 (1.4, 2.1)
Retirement	328	0.9 (0.8, 1.2)	1.0 (0.8, 1.2)	1.0 (0.8, 1.2)	0.9 (0.7, 1.1)

Open in a new tab

SA/DP: sickness absence, disability pension; OR: odds ratio; CI: confidence interval

Model 1: adjusted for age.

Model 2: model 1 + education, both parents/participants themselves was born outside Sweden, nationality, family situation, living area, economic hardship.

Model 3: model 2 + previous SA/DP, long-term illness or health-problem, daily smoking and overweight/obesity.

Model 4: model 3 + working hours, physically strenuous work, monotonicity, noise exposure, mentally strenuous job, hectic work schedule, opportunity to learn new things and job accidents.

We then stratified our analysis by sex and investigated which variables were associated with different cluster memberships among women and men separately, again using the “Active” sequence cluster as the reference category. Lower education, living outside of the Stockholm area, living alone without children, economic hardship, both parents/participants themselves being born outside of Sweden, and variables indicating poor health and health-behaviours were associated with the “Unemployment & SA/DP periods” clusters membership in both sexes (S5 and S6 Tables). Among women, younger age (S5 Table) and among men (S6 Table), older age, working <35 hours per week (compared with ≥35 and <45 hours) and some work-related factors were associated with the “Unemployment & SA/DP periods” cluster membership. Younger age was in both sexes associated with membership of the “Parental-leave periods” cluster. Higher education, economic hardship, some health-related factors, working <35 hours per week (compared with ≥35 and <45 hours), and other work-related factors were associated with membership of the “Parental-leave periods” cluster among women, while among men, living in Stockholm, cohabiting with a partner and children, and being on SA in the previous year were associated with the “Parental-leave periods” cluster membership. Belonging to the “SA/DP periods” cluster was strongly associated with a history of SA/DP among both sexes as well as with older age, lower education, poor health and health related factors. Living outside the Stockholm area, living alone with children, either participants themselves or their parents being born outside of Sweden, and some work-related factors were associated with the “SA/DP periods” cluster membership among women, and either participants themselves or their parents being born outside of Sweden and living alone among men. Older age and history of SA/DP were associated with belonging to the “Retirement” cluster among women while older age, being married/cohabiting without children, and poor health were associated with the “Retirement” cluster membership among men.

Discussion

In this prospective cohort study of a random sample of individuals initially in paid work in Sweden, we identified a total of five clusters of working life sequences over the 15-year follow-up. The vast majority (65%; 54% of women and 76% of men) belonged to the “Active” sequence cluster in which employment or studying was the exclusive labour market activity state. Other clusters combined mostly “active” states with “on parental leave”, “unemployed”, “on SA/DP”, and “retired” states. Women were 33.2 times more likely than men to belong to the “Parental-leave periods” and 1.8 times more likely to the “SA/DP periods” clusters than men, and these associations were only to a small degree affected by adjustments for sociodemographic factors, living conditions, health, and work-related factors.

Our finding that women were more likely than men to have working life patterns characterised by long parental-leave periods is in line with the studies from the U.S. and Finland [36–38]. Finland has a similar welfare state model as Sweden [39], encouraging equal contribution to childcare activities among the parents, providing good quality, public, low-fee childcare, and promoting equal participation in the labour market. However, similar sex differences as in our study were found in Finland in the early career trajectories even among those with a stable labour market attachment [37]. A U.S. study from the National Longitudinal Survey of Youth showed that women were more likely to have unstable employment trajectories and lower level of employment over their life course than men [36]. In Sweden, long parental leave among women has been associated with a lower likelihood of upward occupational mobility [40].

In this cohort of people in paid work, women were not more likely than men to end up in working life sequences characterized by long-term unemployment or retirement. Instead, we found that women were slightly more likely to belong to a sequence with longer SA/DP periods. While several work- and non-work-related factors at baseline were adjusted for, it is still possible that some of these factors changed over time, which we could not consider, e.g., stress from work and domestic activities, which might lead to higher SA/DP rates among women. Sex differences in disease susceptibility and incidence for diseases such as common mental disorders with a higher incidence among women than men, may also have contributed to the observed differences. A Finnish register-based study with a 20-year follow-up found no sex difference in the long-term risk of DP following the first SA spell [41]. However, the outcome of interest was DP, while we measured both SA and DP with different lengths and sequences over the working life course, and therefore, we were also able to capture individuals with longer, but still temporary SA, with the possibility of later returning to the labour market.

When we analysed the association between baseline characteristics and cluster membership among women and men separately, we found that overall, the same characteristics predicted membership in the “Unemployment & SA/DP periods” sequence clusters in both sexes. Lower education, either participants themselves or their parents born outside of Sweden, economic hardship, living outside of Stockholm, and poor health were associated with this cluster membership. This is in line with the findings of a recent review in Europe which concluded that self-reported good health was the major facilitator of long working life among both women and men [2]. Other studies found that poor health and several psychosocial work characteristics, such as less possibility to develop skills and knowledge and low appreciation at work were associated with worse labour market attachment [42] as well as with higher risk of long SA and DP [10, 11, 13, 14].

Low education has also been shown to be associated with unstable working life trajectories [36]. However, in our study, higher education and adverse work environment were associated with the membership of a sequence cluster where active states were interrupted with longer parental-leave periods among women, which somewhat contradicts the findings of a previous Finnish study where lower education, lower attachment to the labour market, and immigrant background were the strongest predictors of long parental leave among women [37]. Another Swedish study also found that the use of parental-leave benefits was higher among immigrant women compared with native women [43] but this difference decreased along with time spent in Sweden, suggesting that lower attachment to the labour market was the main reason of longer parental-leave use among immigrant women [43]. The reasons for our different results might be related to a different baseline population (in our study, those in paid work) and a different outcome definition, as our study focused on complex, long-term working life patterns including unemployment and SA/DP instead of a single outcome, such as the number of parental-leave days. Being born in Sweden and living in Stockholm was a predictor of membership in the “Parental-leave periods” cluster among men, which suggests that while the Swedish parental-leave benefit policy aims to motivate more equal sex distribution in taking parental leave, there might still be educational, cultural, financial, or other policy-related factors which act against this [5, 39]. In sum, our study fits in the integrated life-course perspective proposed by Amick et al. [21], which provides a more dynamic approach of describing labour market and health experiences in working-age individuals [21].

Strengths and limitations

The key strengths of our study are that we linked population-based survey data with annual and comprehensive nationwide register microdata [27, 44] and the long follow-up with minimal attrition, thus minimizing selection biases that can impact the findings of many other types of observational studies. Compared with other methods, sequence analysis allowed us to deal with the complexity of the future course of the working life.

One limitation of our study is that self-reporting in ULF/SILC survey might introduce some degree of bias due to misclassification. When available, we used both survey-based and register-based information but some information, such as health behaviors and work-related variables were only self-reported. Even though we conducted sensitivity analyses with different analytical parameters and methods which resulted in similar cluster structures, certain individuals might have ended up in different clusters when using different analytical decisions. However, the “Active”, “Parental-leave periods”, and “Retirement” clusters were consistently built up from the same or similar number of individuals and it was mostly the “Unemployment & SA/DP periods” and the “SA/DP periods” clusters that varied through the different clustering algorithms. Furthermore, the definition of states is also a simplified representation of working life sequences and could not fully capture the complexity of people’s working life.

Our findings are generalizable to Sweden and, to some extent, to similar welfare and economic settings during the investigated time-period. Nordic countries have provided flexible and generous parental-leave benefits and child care for a long time now, and sex differences in employment are less pronounced than in other countries that provide more limited access to those. Furthermore, to be able to assess the contribution of working conditions, we limited our study sample to those who reported being in paid work at baseline and therefore, our results are not generalizable to individuals marginalized from the labour market or to those who had not entered the labour market.

Conclusions

In a cohort of people initially in paid work in Sweden, we found that during the 15-year follow-up, the absolute majority (65%) of all individuals were uninterruptedly economically active (working or studying). About a third of them belonged to working life sequence clusters with active states combined with some interruptions. Women had a much greater likelihood than men of having a working life course characterized by employment interrupted with parental-leave and SA/DP periods but no sex difference were observed in a working life course characterized by a mixture of SA/DP and unemployment periods. These findings do not indicate that men did not take parental leave at all, but rather that during the follow-up of 15 years, fewer of them spent more than 183 days a year on parental leave benefits. However, the findings still suggest that there is a gender gap in the working life course even in Sweden where the Nordic welfare state model provides extensive and equal support to the labour market participation for women and men and promotes equal sharing of childcare.

Supporting information

S1 Fig. Selection of the study participants.

(PDF)

Click here for additional data file.^{(40.8KB, pdf)}

S2 Fig. Sequence index plots of different states of future yearly working life sequences over a 15-year period.

(TIF)

Click here for additional data file.^{(894.6KB, tif)}

S3 Fig. Sequence index plots of different states of future yearly working life sequences over a 15-year period by sex.

(TIF)

Click here for additional data file.^{(830.4KB, tif)}

S4 Fig. Hierarchical cluster tree of sequences of future yearly working life sequences over a 15-year period.

(TIF)

Click here for additional data file.^{(979KB, tif)}

S5 Fig. Measures of cluster partitions quality for different numbers of sequence clusters.

ASW: Average Silhouette Width HG: Hubert’s Gamma, PBC: Point Biserial Correlation HC: Hubert’s C.

(TIF)

Click here for additional data file.^{(932.2KB, tif)}

S6 Fig. Mean time spent in each sequence states by sequence clusters.

(TIF)

Click here for additional data file.^{(1.1MB, tif)}

S7 Fig. Sequence index plots of four clusters of working life sequences.

(TIF)

Click here for additional data file.^{(820.2KB, tif)}

S8 Fig. Sequence index plots of six clusters of working life sequences.

(TIF)

Click here for additional data file.^{(802.8KB, tif)}

S9 Fig. Sequence index plots of seven clusters of working life sequences.

(TIF)

Click here for additional data file.^{(998.6KB, tif)}

S10 Fig. Measures of cluster partitions quality for different numbers of sequence clusters by women and men.

(TIF)

Click here for additional data file.^{(1.2MB, tif)}

S11 Fig. Sequence index plots of four clusters of working life sequences for women.

(TIF)

Click here for additional data file.^{(930.6KB, tif)}

S12 Fig. Mean time spent in each sequence states by sequence clusters for women.

(TIF)

Click here for additional data file.^{(897.3KB, tif)}

S13 Fig. Sequence index plots of four clusters of working life sequences for men.

(TIF)

Click here for additional data file.^{(920KB, tif)}

S14 Fig. Mean time spent in each sequence states by sequence clusters for men.

(TIF)

Click here for additional data file.^{(883KB, tif)}

S1 Table. Transition probabilities between different activity.

(DOCX)

Click here for additional data file.^{(36.6KB, docx)}

S2 Table. Measures of cluster partitions quality for different numbers of clusters.

(DOCX)

Click here for additional data file.^{(26.1KB, docx)}

S3 Table. Measures of cluster partitions quality for different numbers of clusters for women and men.

(DOCX)

Click here for additional data file.^{(23.6KB, docx)}

S4 Table. Multinomial regression analysis for the associations between sex and different working life sequence clusters for alternative 6 clusters solutions.

(DOCX)

Click here for additional data file.^{(22.1KB, docx)}

S5 Table. Associations between baseline factors and cluster membership among women.

(DOCX)

Click here for additional data file.^{(52.7KB, docx)}

S6 Table. Associations between predictors and cluster membership among men.

(DOCX)

Click here for additional data file.^{(52.3KB, docx)}

Acknowledgments

We thank to Annika Evolahti (Department of Clinical Neuroscience, Karolinska Institutet) for her editing support.

Data Availability

These data cannot be made publicly available due to privacy regulations. According to the General Data Protection Regulation, the Swedish law SFS 2018:218, the Swedish Data Protection Act, the Swedish Ethical Review Act, and the Public Access to Information and Secrecy Act, these types of sensitive data can only be made available for specific purposes, including research, that meets the criteria for access to this type of sensitive and confidential data as determined by a legal review. Readers may contact Professor Kristina Alexanderson (kristina.alexanderson@ki.se) and Mira Muller (mira.muller@ki.se) regarding the data.

Funding Statement

This work was financially supported by a research grant from the Swedish Research Council for Health, Working Life and Welfare, FORTE [grant number 2018-00547, receiver: MV]. We utilised data from the REWHARD consortium supported by the Swedish Research Council (grant number 2017-00624). There was no additional external funding received for this study. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Rechel B, Grundy E, Robine J-M, Cylus J, Mackenbach JP, Knai C, et al. Ageing in the European Union. Lancet Lond Engl. 2013;381: 1312–1322. doi: 10.1016/S0140-6736(12)62087-X [DOI] [PubMed] [Google Scholar]
2.Edge CE, Cooper AM, Coffey M. Barriers and facilitators to extended working lives in Europe: a gender focus. Public Health Rev. 2017;38: 2. doi: 10.1186/s40985-017-0053-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Stafford M., Lacey R., Murray E., Carr E, Fleischmann M, Stansfeld S, et al. Work–family life course patterns and work participation in later life. Eur J Ageing 2019; 16, 83–94. doi: 10.1007/s10433-018-0470-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Worts D., Corna L, Sacker A, McMunn A, McDonough P. Understanding older adults’ labour market trajectories: A comparative gendered life course perspective. Longitudinal and Life Course Studies,2016; 7,347–367. doi: 10.14301/llcs.v7i4.389 [DOI] [Google Scholar]
5.Gonäs L, Wikman A, Vaez M, Alexanderson K, Gustafsson K. Gender segregation of occupations and sustainable employment: A prospective population-based cohort study. Scand J Public Health. 2019;47: 348–356. doi: 10.1177/1403494818785255 [DOI] [PubMed] [Google Scholar]
6.Virtanen M, Oksanen T, Pentti J, Ervasti J, Head J, Stenholm S, et al. Occupational class and working beyond the retirement age: a cohort study. Scand J Work Environ Health. 2017;43: 426–435. doi: 10.5271/sjweh.3645 [DOI] [PubMed] [Google Scholar]
7.Campos-Serna J, Ronda-Pérez E, Artazcoz L, Moen BE, Benavides FG. Gender inequalities in occupational health related to the unequal distribution of working and employment conditions: a systematic review. Int J Equity Health. 2013;12: 57. doi: 10.1186/1475-9276-12-57 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Löve J, Hensing G, Holmgren K, Torén K. Explaining the social gradient in sickness absence: a study of a general working population in Sweden. BMC Public Health. 2013;13: 545. doi: 10.1186/1471-2458-13-545 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Virtanen M, Oksanen T, Batty GD, Ala-Mursula L, Salo P, Elovainio M, et al. Extending employment beyond the pensionable age: a cohort study of the influence of chronic diseases, health risk factors, and working conditions. PloS One. 2014;9: e88695. doi: 10.1371/journal.pone.0088695 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Kivimäki M, Elovainio M, Vahtera J. Workplace bullying and sickness absence in hospital staff. Occup Environ Med. 2000;57: 656–660. doi: 10.1136/oem.57.10.656 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Nielsen MB, Indregard A-MR, Øverland S. Workplace bullying and sickness absence: a systematic review and meta-analysis of the research literature. Scand J Work Environ Health. 2016;42: 359–370. doi: 10.5271/sjweh.3579 [DOI] [PubMed] [Google Scholar]
12.Waenerlund A-K, Gustafsson PE, Hammarström A, Virtanen P. History of labour market attachment as a determinant of health status: a 12-year follow-up of the Northern Swedish Cohort. BMJ Open. 2014;4: e004053. doi: 10.1136/bmjopen-2013-004053 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Duchaine CS, Aubé K, Gilbert-Ouimet M, Vézina M, Ndjaboué R, Massamba V, et al. Psychosocial Stressors at Work and the Risk of Sickness Absence Due to a Diagnosed Mental Disorder: A Systematic Review and Meta-analysis. JAMA Psychiatry. 2020;77: 842–851. doi: 10.1001/jamapsychiatry.2020.0322 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Knardahl S, Johannessen HA, Sterud T, Härmä M, Rugulies R, Seitsamo J, et al. The contribution from psychological, social, and organizational work factors to risk of disability retirement: a systematic review with meta-analyses. BMC Public Health. 2017;17: 176. doi: 10.1186/s12889-017-4059-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Gilbert-Ouimet M, Trudel X, Aubé K, Ndjaboue R, Duchaine C, Blanchette C, et al. Differences between women and men in the relationship between psychosocial stressors at work and work absence due to mental health problem. Occup Environ Med. 2020;77: 603. doi: 10.1136/oemed-2019-106242 [DOI] [PubMed] [Google Scholar]
16.McMunn A, Lacey R, Worts D, McDonough P, Stafford M, Booker C, et al. De-standardization and gender convergence in work–family life courses in Great Britain: A multi-channel sequence analysis. Advances in Life Course Research. 2015;26: 60–75. doi: 10.1016/j.alcr.2015.06.002 [DOI] [Google Scholar]
17.McMunn A, Lacey R, Worts D, Kuh D, McDonough P, Sacker A.Work-family life courses and psychological distress: Evidence from three British birth cohort studies, Advances in Life Course Research,2021;50100429, ISSN 1040-2608, doi: 10.1016/j.alcr.2021.100429 [DOI] [PubMed] [Google Scholar]
18.Lacey R., Stafford M., Sacker A. et al. Work-Family Life Courses and Subjective Wellbeing in the MRC National Survey of Health and Development (the 1946 British birth cohort study). Population Ageing 9, 69–89 (2016). doi: 10.1007/s12062-015-9126-y [DOI] [PMC free article] [PubMed] [Google Scholar]
19.WAHRENDORF M. (2015). Previous employment histories and quality of life in older ages: Sequence analyses using SHARELIFE. Ageing and Society, 35(9), 1928–1959. doi: 10.1017/S0144686X14000713 [DOI] [Google Scholar]
20.Lacey R., Sacker A., Bell S. et al. Work-family life courses and BMI trajectories in three British birth cohorts. Int J Obes 41, 332–339 (2017). doi: 10.1038/ijo.2016.197 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Amick BC, McLeod CB, Bültmann U. Labor markets and health: an integrated life course perspective. Scand J Work Environ Health. 2016;42: 346–353. doi: 10.5271/sjweh.3567 [DOI] [PubMed] [Google Scholar]
22.Abbott A. Sequence Analysis: New Methods for Old Ideas. Annu Rev Sociol. 1995;21: 93–113. doi: 10.1146/annurev.so.21.080195.000521 [DOI] [Google Scholar]
23.Macidoe H, Abbott A. Sequence analysis and optimal matching techniques for social science data. Handbook of Data Analysis. SAGE Publications, Ltd; 2004. pp. 387–406. doi: 10.4135/9781848608184 [DOI] [Google Scholar]
24.World Medical Association Declaration of Helsinki; Ethical Principles for Medical Research Involving Human Subjects. Bulletin of the World Health Organization. 2001;79(4):373–4. [PMC free article] [PubMed] [Google Scholar]
25.Ludvigsson J, Håberg S, Knudsen G, Lafolie P, Zoega H, Sarkkola C, et al. Ethical aspects of registry-based research in the Nordic countries. Clinical Epidemiology. 2015;7:491–508. doi: 10.2147/CLEP.S90589 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Statistics Sweden. Living Conditions Surveys (ULF/SILC. Available: https://www.scb.se/en/finding-statistics/statistics-by-subject-area/living-conditions/living-conditions/living-conditions-surveys-ulfsilc/ [Google Scholar]
27.Ludvigsson JF, Otterblad-Olausson P, Pettersson BU, Ekbom A. The Swedish personal identity number: possibilities and pitfalls in healthcare and medical research. Eur J Epidemiol. 2009;24: 659–667. doi: 10.1007/s10654-009-9350-y [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Ludvigsson JF, Svedberg P, Olén O, Bruze G, Neovius M. The longitudinal integrated database for health insurance and labour market studies (LISA) and its use in medical research. Eur J Epidemiol. 2019;34: 423–437. doi: 10.1007/s10654-019-00511-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Niklas Österlund. MiDAS—sjukpenning och rehabiliteringspenning (MiDAS—sickness benefit and rehabilitation allowance). Swedish Social Insurance Agency.; 2011. [Google Scholar]
30.Social Insurance in Figures 2018. Stockholm: The Swedish Social Insturance Agency; 2018. [Google Scholar]
31.Lundgren B. Recent development in unemployment insurance in Sweden Unemployment Insurance and Employment Maintenance. Brussels: International Social Security Association; 2006. [Google Scholar]
32.National Agency for Education. Descriptive Data on child care and schools in Sweden in 2000. Report No. 192. Stockholm. [Google Scholar]
33.Statistic Sweden, Population Statistic (ww.scb.se) [Google Scholar]
34.Studer M. WeightedCluster Library Manual A practical guide to creating typologies of trajectories in the social sciences with R. LIVES Working Papers, 24.; 2013. Available: 10.12682/lives.2296-1658. 2013.24. [DOI] [Google Scholar]
35.Gabadinho A, Ritschard G, Müller NS, Studer M. Analyzing and Visualizing State Sequences in R with TraMineR. J Stat Softw. 2011;40: 1–37. doi: 10.18637/jss.v040.i04 [DOI] [Google Scholar]
36.Weisshaar K, Cabello-Hutt T. Labor Force Participation Over the Life Course: The Long-Term Effects of Employment Trajectories on Wages and the Gendered Payoff to Employment. Demography. 2020;57: 33–60. doi: 10.1007/s13524-019-00845-8 [DOI] [PubMed] [Google Scholar]
37.Kuitto K, Salonen J, Helmdag J. Gender Inequalities in Early Career Trajectories and Parental Leaves: Evidence from a Nordic Welfare State. Soc Sci. 2019;8. doi: 10.3390/socsci8090253 [DOI] [Google Scholar]
38.Erosa A, Fuster L, Restuccia D. Fertility Decisions and Gender Differences in Labor Turnover, Employment, and Wages. Rev Econ Dyn. 2002;5: 856–891. doi: 10.1006/redy.2002.0195 [DOI] [Google Scholar]
39.Duvander A-Z, Ferrarini T, Thalberg S. Swedish parental leave and gender equality Achievements and reform challenges in a European perspective. Institutet för Framtidsstudier; 2005. Report No.: 91-89655-69–9. Available: https://www.iffs.se/media/1118/20051201134956filu8yijlraac7u4fv7gumy.pd [Google Scholar]
40.Evertsson M, Duvander A-Z. Parental Leave—Possibility or Trap? Does Family Leave Length Effect Swedish Women’s Labour Market Opportunities? Eur Sociol Rev. 2011;27: 435–450. [Google Scholar]
41.Klein J, Reini K, Saarela J. Sickness Absence and Disability Pension in the Very Long Term: A Finnish Register-Based Study With 20 Years Follow-Up. Front Public Health. 2021;9: 142. doi: 10.3389/fpubh.2021.556648 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.de Wind A, Geuskens GA, Ybema JF, Blatter BM, Burdorf A, Bongers PM, et al. Health, job characteristics, skills, and social and financial factors in relation to early retirement–results from a longitudinal study in the Netherlands. Scand J Work Environ Health. 2014; 186–194. doi: 10.5271/sjweh.3393 [DOI] [PubMed] [Google Scholar]
43.Mussino E, Duvander A-Z. Use It or Save It? Migration Background and Parental Leave Uptake in Sweden. Eur J Popul. 2016;32: 189–210. doi: 10.1007/s10680-015-9365-x [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Ludvigsson JF, Almqvist C, Bonamy A-KE, Ljung R, Michaëlsson K, Neovius M, et al. Registers of the Swedish total population and their use in medical research. Eur J Epidemiol. 2016;31: 125–136. doi: 10.1007/s10654-016-0117-y [DOI] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0281056.r001

Decision Letter 0

Renuka Sane

30 Aug 2022

PONE-D-22-14059Working life sequences over the life course among 9269 women and men in Sweden; a prospective cohort studyPLOS ONE

Dear Dr. Gemes,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Oct 14 2022 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Renuka Sane

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please provide additional details regarding participant consent. In the ethics statement in the Methods and online submission information, please ensure that you have specified what type you obtained (for instance, written or verbal, and if verbal, how it was documented and witnessed). If your study included minors, state whether you obtained consent from parents or guardians. If the need for consent was waived by the ethics committee, please include this information.

3. We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match.

When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section.

4. Thank you for stating the following financial disclosure:

“This work was financially supported by the Swedish Research Council for Health, Working Life and Welfare, FORTE (2018-00547) and utilised data from the REWHARD consortium, supported by the Swedish Research Council (VR grant number. 2017-00624).”

Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

If this statement is not correct you must amend it as needed.

Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf.

5. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

6. Please include your full ethics statement in the ‘Methods’ section of your manuscript file. In your statement, please include the full name of the IRB or ethics committee who approved or waived your study, as well as whether or not you obtained informed written or verbal consent. If consent was waived for your study, please include this information in your statement as well.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This study uses sequence analysis on a Swedish cohort created from registry data to create employment sequences for men and women. Sequence analysis is a useful tool for characterising life courses and is often used to characterise employment life courses. However, there are several major issues with the paper as it currently stands.

First, it is not clear what the rationale and motivation, or even the research question is here. The introduction begins with a focus on extended working lives but the sequences included are truncated at age 50 so it is not possible to contribute to this policy debate with this analysis. Perhaps the aim of the analysis is to describe diversity in employment life courses, but this hasn't really been achieved with only five groups capturing employment sequences for two genders. A secondary aim is to examine whether a range of factors are associated with employment sequences but there are no hypothesised directions of association nor any particular attempt to explain why these factors are important to investigate. Later in the statistical analysis section the authors state the regression models are to examine whether these covariates explain gender differences, but this is not developed in the introduction and would need careful thought to conceptually distinguish between confounders and mediators.

I suspect one reason for the lack of rationale and clear objectives is a lack of engagement with the literature. For example, the authors state that 'the contribution of working, living and health conditions to working-life participation among women and men has rarely been examined from a life course perspective. This could hardly be further from the truth. See studies by Lacey, McMunn, Berkman's group, Wahrendorf, Stone, Engels, Stafford, McDonough and others. There is a relatively large life course literature on this topic.

Which leads to my third comment which is the lack of engagement with key life course concepts such as timing of transitions, duration, accumulation, destandardisation, diversity etc (for one eg. see McMunn et al ALCR 2015).

In relation to the analysis conducted, the main issue is the lack of diversity in five groups which cover both men and women. There is nothing much of interest to report with these five groups. The purpose of sequence analysis is to illuminate life course patterns for different groups. The current analysis simply show that women have weaker ties to the labour market than men (which we know, even in Sweden) and results could have just as easily been conducted with count variables for the number of years in each state.

The authors go to lengths to justify their selection of five groups, but I was unfamiliar with many of the acronyms included in ST2 and they are not explained in the text. Neither are we given a clear description of the clusters in terms of Ns and gender % in each group, nor the duration of states and timing of transitions within each cluster. If the authors are interested in gender differences, they could consider running the cluster analysis separately for men and women and seeing how different the group solutions are between the two.

Small points:

What is SA/DP, this is never described.

Why have you mentioned cause of death data when you are not looking at mortality?

Some tables are lacking titles and a clear description of what numbers in them represent.

Having a similar number of people in each cluster is not a pre-requisite, nor even particularly useful, for determining the number of clusters.

ST4 is referred to in the text but I could not find it in the supplementary material.

Reviewer #2: This paper aims to analyze the gender gap in interruptions in working life sequences for individuals aged 18-50 in Sweden. The paper is very clearly and concisely written, authors have done a thorough exercise in terms of setting up the background, conducted a sound empirical analysis and provided a helpful discussion of their results. The research question assessed in the paper is a relevant one as Sweden is one of the developed nations with relatively smaller sex differences in employment/wage rates and progressive parental-leave options, these are social norms that many other nations wish to emulate. Hence a lot can be learned from the results of this paper, to be specific in spite of a generous parental-leave option which encourages equal sharing of parental responsibilities it is women who face higher number of interruptions in their working life sequences as represented by higher likelihood of women (vs men) belonging to “Parental-leave periods” and “Sickness Absence/Disability Pension” categories.

Below are my comments for the authors.

The analysis in the paper relies on a cohort dataset where 9269 individuals are followed up over a course of 15 years. The empirical analysis comprises of two parts. The first part comprises of cluster analysis (which groups similar working life sequences into 5 mutually exclusive categories) and the second part uses the information from part one to analyze the gender difference in the working life sequences (5 mutually categories from part one).

Major comments:

1) The cluster analysis results in 2-10 cluster solutions. And here I think researcher’s degree of freedom creep in. Authors chose a 5 cluster solution and provided the following short description “The measure of cluster partition quality (S2 Table, S4 and S5 Fig), in combination with the decision to keep the number of individuals within clusters on a reasonable level were used to decide the number of clusters to be the above-listed five.”.

I would recommend strongly to do two things:

a) Provide some more details about the choice being made i.e. a 5 cluster solution (instead of say a 4 or 6 cluster solution). Essentially the authors need to substantiate the critical choice they made with few objective arguments. For example, which measures of partition quality were used in this critical choice? To be precise, mention the values being higher or lower in comparison to other competing choices.

b) Since the analysis in part two, i.e. multinomial logit regressions (to assess gender gap) uses the cluster solution (mentioned above) as the dependent variable so the authors should ideally show that their results are robust to alternate choices of cluster solution. This would basically mean that Table 2 needs to be replicated for alternate cluster solutions. Looking at Table S2 Table it could mean using 3 cluster or 4 cluster or 6 cluster solution (instead of the original choice i.e. 5). This can be supplementary table but please at least 2 more cluster solutions (ideally one with higher of cluster solutions and another with lower value).

2) The detailed results related to separate results for men and women (Table S4 and S5) use sociodemographic variables from the baseline survey (the individuals from this survey were then followed up for next 15 years to create a trajectory of their working states). However not all of these sociodemographic characteristics are time invariant. For example, living alone with children variable (many respondents might not have had children at the time of survey but subsequently this status might have changed). Similar arguments hold true hours-per-week (as nature of work might have changed during the 15 year follow up period), higher education, health-related factors etc. Can the authors comment on this? Also if indeed these variables are time variant then the analysis should only be restricted to those variables which are time-invariant (and mention this as a limitation as well).

3) Provide more details about how the parental-leave policy is structured to provide equal parental benefits. This would go a long way in terms of explaining the huge gender difference which is observed for the parental-leave category (OR 33.2). This could also provide a helpful discussion which steers towards better policy design.

Minor comments:

1) Why does figure S2 have the highest value on Y axis as 8933, shouldn’t it be 9269?

2) All figures are pixelated, that is their appearance is not the best. It would be highly recommended to provide higher resolution images.

3) The manuscript refers to Table S4 in text but this table was missing from the manuscript (I could not locate a hyperlink for this)

4) Can authors provide very short (1-2 line) description about average at which males and females have their first child in Sweden, mean number of children, within country migration statistics (for example migration to and from Stockholm area. Would be helpful to know these details as some variables being used in the analysis are related to these).

5) The citation for reference number 2 has curly brackets (in the introduction) while all other citations have been put inside square brackets.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2023 Feb 15;18(2):e0281056. doi: 10.1371/journal.pone.0281056.r002

Author response to Decision Letter 0

13 Dec 2022

Point-by-point responses to the reviewers’ comments

Reviewer #1:

1. (1)”This study uses sequence analysis on a Swedish cohort created from registry data to create employment sequences for men and women. Sequence analysis is a useful tool for characterising life courses and is often used to characterise employment life courses. However, there are several major issues with the paper as it currently stands. First, it is not clear what the rationale and motivation, or even the research question is here.”

Authors’ response:

We thank the reviewer for the thorough review. We agree that the introduction was not clear enough and have now revised it to better reflect the study’s aims. Moreover, in line with your suggestions, we clarified the aim and rationale of the study and the hypotheses as following: “The specific research questions were to explore: 1) which types of overarching future working life sequence groups can be identified during a 15-year period in a cohort of individuals in paid work at baseline? 2) whether women have more often than men working life courses characterized by several interruptions, such as parental leave, SA/DP, or unemployment? 3) whether adjusting for baseline sociodemographic, health- and work-related factors weaken the possible associations between sex and the specific sequence cluster membership? 4) whether these factors are independently associated with the specific sequence cluster membership in a separate analysis among women and men?” (page 6)

2. “The introduction begins with a focus on extended working lives but the sequences included are truncated at age 50 so it is not possible to contribute to this policy debate with this analysis.”

Authors’ response:

We might have been unclear, however, to clarify: the sequences were not truncated at the age of 50. As we wanted to follow all people for 15 years, we included those who were 50 years or younger at inclusion, as the traditional age for old-age pension is 65 in Sweden. We now rephrased the first paragraph in the Introduction section, so the focus now is less on the old-pension age itself, as this paper was not intended to contribute to the policy debate on pension age, but more to map working life sequences in a population of those in paid work at baseline, and to investigate whether there are any important sex differences in working life sequences; and in that case, what they might be.

3. “Perhaps the aim of the analysis is to describe diversity in employment life courses, but this hasn't really been achieved with only five groups capturing employment sequences for two genders.”

Authors’ response:

Our research aim was to map the overarching future working-life sequences in a cohort of individuals in paid work when included. This is now made clear in the aim of the revised manuscript.

That we ‘only’ included individuals in paid work by default limited the observable complexity of working life sequences. Therefore, while we expected diversity in working-life sequences we did not anticipate a high level of complexity given our study population and the annual resolution of sequences. As our aim was to use the integrated life-course approach we defined labour market states on an annual basis to be able to integrate both family-, work-, and health-related interruptions of working life. Register data on sickness absence and disability pension have rarely been used in other types of studies.

Actually, we regard the relative lack of complexity in the working-life sequences as one of the important findings of our study - as it suggests that the majority of both women and men have a working life course that is not interrupted by long breaks due to longer sickness absence, unemployment, or parental-leave periods. The mean time spent in an economically active state was close to 15 years, while the mean time spent in other states was less the 2 years– something that also reflects the limited diversity of working life sequences in this population. With this limited diversity in this population, a relatively low number of clusters could be expected. The chosen 5-cluster solution was a compromise between the best cluster partitioning, the highest possible complexity, and having sufficient power to be able to run the multinomial regression analysis. In the revised manuscript, following your comment, we now also present other cluster solutions in the supplement material (Fig S7, S8 and S9, Table S3, Table S3 and S4). Improvement in cluster partitioning measures over 5 clusters was minimal and using over 7 clusters they did not improve further at all.

4. “A secondary aim is to examine whether a range of factors are associated with employment sequences but there are no hypothesised directions of association nor any particular attempt to explain why these factors are important to investigate.

Authors’ response:

You are right, we had not hypothesised about the directions of associations. This was an exploratory observational study aiming to gain knowledge on the associations between employment sequence clusters and socioeconomic, health- and work-related factors such aspects. The choice of these factors was based on two considerations. One was based on which factors a large number of studies have shown are of importance for people’s work situation, sickness absence, etcetera. The other was more pragmatic: we could only include factors that were available, through the surveys and registers. In this study, our focus was to use survey data on the detailed work environment, including physical and psychosocial work environment, and information on subjective health, health behaviours, and chronic conditions. We have elaborated in the Introduction (on pages 4-5) about the selection of these factors.

5. Later in the statistical analysis section the authors state the regression models are to examine whether these covariates explain gender differences, but this is not developed in the introduction and would need careful thought to conceptually distinguish between confounders and mediators.”

Authors’ response:

Thank you for this comment, you are right and we have now modified this sentence in the Introduction as follows: “. To examine whether the associations between sex and cluster memberships attenuated, several sets of covariates were gradually included in the models.”.

Our study design did not allow us to investigate any causal mechanism, or distinguish between confounders and mediators, but only to map the most common patterns examining health-, work-, and family-related interruption in working life course and their possible association with sex and to map the main factors that were associated with different working life sequence clusters in women and men. As we measured all such factors only at baseline, causal inference would not be possible with this study design. However, we agree with the reviewer that conceptualizing and quantifying the effect of confounders and mediators on working life courses in women and men is an important research question, and we would like to pursue it in future research.

6. (2)”I suspect one reason for the lack of rationale and clear objectives is a lack of engagement with the literature. For example, the authors state that 'the contribution of working, living and health conditions to working-life participation among women and men has rarely been examined from a life course perspective. This could hardly be further from the truth. See studies by Lacey, McMunn, Berkman's group, Wahrendorf, Stone, Engels, Stafford, McDonough and others. There is a relatively large life course literature on this topic.”

Author’s response:

Thank you for alerting us on not having referred to all the relevant literature on this topic. We are aware that these aspects have been researched by many important researchers in many different projects, and we should of course have mentioned these. What we aimed to state is that information on sickness absence and disability pension has seldom been included in studies using sequence analysis, the focus has rather been on work-family life courses and work participation of women in relation to different work- and health-related outcomes. We have now revised the Introduction and included the most relevant literature and clearer pointed out that sickness absence and disability pension were not included in the previous literature (page 5).

7. (3). ”Which leads to my third comment which is the lack of engagement with key life course concepts such as timing of transitions, duration, accumulation, destandardisation, diversity etc (for one eg. see McMunn et al ALCR 2015).”

Author’s response:

Thank you for referring to this important paper. We clearly need to be more explicit when using the concepts. Therefore, we now refer to the concepts of “duration” and “diversity” in the Introduction as follows: “Some of the previous research focused on the examination of the inter- and intra-individual diversity of work-family life courses [16-21]. McMunn at al., [16] examined the changes in work-family life courses over time and showed that women’s work-family life courses converged to men’s while the diversity of the individual work-family life courses increased” referring to the Mc Munn et al ALCR 2015 paper. However, other concepts such as de-standardization and differentiation are not relevant to our study questions as we did not compare different time periods, but we mapped the sequences within the same cohort. Nor did we aim to focus on specific transitions or accumulation of exposure in this study. The aim of this study was descriptive, to map the most common working-life sequences and describe their diversity by forming groups of similar sequences. We agree with the reviewer; these concepts are important from the life-course perspective and can be studied specifically after this first step of explorative research.

8. (4.) “In relation to the analysis conducted, the main issue is the lack of diversity in five groups which cover both men and women. There is nothing much of interest to report with these five groups. The purpose of sequence analysis is to illuminate life course patterns for different groups. The current analysis simply show that women have weaker ties to the labour market than men (which we know, even in Sweden) and results could have just as easily been conducted with count variables for the number of years in each state.”

Author’s response:

We agree that in this cohort of individuals in paid work at baseline, the diversity of life course patterns is limited. However, we regard this by itself as an important finding, showing that most both women and men in Sweden who are attached to the labour market actually have uninterrupted working life courses. This finding is important and has not been shown before in a large cohort with register-based information on working life states. As we have described above in Point 4, we chose sequence analysis to be able to capture different working life sequences with both health-, family-, and work-related interruptions using register microdata, which is one of the novel aspects in our study and required a method that can handle mutually exclusive labour market states.

Therefore, we argue that methods based on count variables could not be able to capture this complexity as they could not handle several types of outcomes in one analysis. In addition, we prepared the study protocol, including the analyses, a priori, without knowing how much the sequences would differ from each other. We agree that several analytical approaches are important in studies of working life courses, and we see our study as one contribution in this area.

9. “(5.) The authors go to lengths to justify their selection of five groups, but I was unfamiliar with many of the acronyms included in ST2 and they are not explained in the text. Neither are we given a clear description of the clusters in terms of Ns and gender % in each group, nor the duration of states and timing of transitions within each cluster. If the authors are interested in gender differences, they could consider running the cluster analysis separately for men and women and seeing how different the group solutions are between the two.”

Author’s response:

We have now added the description of the acronyms to the tables. We also added to the Results section information on the different statistical measures we considered when deciding on the 5-cluster solutions (“The cluster partition quality measures showed the best fits for a five- and a seven-clusters solution (lowest point biseral correlation for five clusters and highest Hubert’s Gamma, Average Silhouette Width, and Calinszki-Habarasz indexes for the seven-clusters solution) but only with a minimal difference in indexes between the five- and seven-clusters solutions (Table S2, Fig S4, and Fig S5).”). Moreover, we now added the % of women in each sequence group beside the number of individuals, which was already presented in the Results section (page 12) . We also added a figure for each cluster for the mean time spent in each state (now Fig S6). This study did not aim to focus on the timing of specific transitions – that would have required other analytical methods in a study with other aims. Moreover, we added the suggested sex-stratified analyses as a supplement (now Fig S10, S11, S12, S13, S14, and Table S3) and added their description in the Method section: “To explore whether these cluster solutions were similar among women and men, we presented the results from cluster analysis stratified by sex.” and Results section “When the cluster analysis was stratified by sex, we found a four-clusters solution to have the best fit for women and a four- and five-clusters solutions to have the best fit for men (Table S3, Fig S10). We also present the four clusters for both women and men with mean time spent in each state in Supplement Material (Fig S11, Fig S12, Fig S13, and Fig S14).”

10. Small points:

10.1 “What is SA/DP, this is never described.”

Authors’ response:

Thank you for alerting us on this! We have now modified and expanded the Abstract and Methods section to clarify that SA/DP meant sickness absence/disability pension.

10.2 “Why have you mentioned cause of death data when you are not looking at mortality?”

Author’s response:

Mortality was not an outcome of interest; however, we used the Cause of Death Register for information when defining the study cohort and identifying those who died during the follow-up and were censored. This is why we mention the use of this data source under the Methods section. However, we have now rewritten this paragraph to explicitly state which information we obtained from this register: “To identify those who died during the follow-up, we obtained the year of death from the Cause of Death Register held by the National Board of Health and Welfare.”

10.3 “Some tables are lacking titles and a clear description of what numbers in them represent.”

Authors’ response:

Thank you, we have now added titles and descriptions to all tables.

10.4 “Having a similar number of people in each cluster is not a pre-requisite, nor even particularly useful, for determining the number of clusters.”

Authors’ response:

We agree with the Reviewer about this, and we did not use this as a criterion to determine the number of clusters. Instead, we used different measures of partition quality to decide on the best fit. The five- and seven-cluster solutions had very similar cluster partition measures, so we decided to present the five clusters and continue with the multinomial regression considering the feasibility of further analysis and the readability of the results. Now we added the following description of our decision steps under the Results: “The cluster partition quality measures showed the best fits for a five- and a seven-clusters solution (lowest point biseral correlation for five clusters and highest Hubert’s Gamma, Average Silhouette Width, and Calinszki-Habarasz indexes for the seven-clusters solution) but only with a minimal difference in indexes between the five- and seven-clusters solutions (Table S2, Fig S4, and Fig S5). To consider the feasibility and readability of the multinomial regression analyses we chose the five-sequence cluster solutions for further analysis (Fig 3). In the five-cluster solution, the mean time spent in each state is presented in Fig S6. Four-, six-, and seven-cluster solutions are presented in Fig S6, Fig S7, and Fig S8.”

10.5“ST4 is referred to in the text but I could not find it in the supplementary material.”

Authors’ response:

We have now uploaded the missing table which now is referred to as Table S6.

PLoS One. doi: 10.1371/journal.pone.0281056.r003

Decision Letter 1

Sreeram V Ramagopalan

17 Jan 2023

Working life sequences over the life course among 9269 women and men in Sweden; a prospective cohort study

PONE-D-22-14059R1

Dear Dr. Gémes,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Sreeram V. Ramagopalan

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

PLoS One. doi: 10.1371/journal.pone.0281056.r004

Acceptance letter

Sreeram V Ramagopalan

23 Jan 2023

PONE-D-22-14059R1

Working life sequences over the life course among 9269 women and men in Sweden; a prospective cohort study

Dear Dr. Gémes:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Sreeram V. Ramagopalan

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 Fig. Selection of the study participants.

(PDF)

Click here for additional data file.^{(40.8KB, pdf)}

S2 Fig. Sequence index plots of different states of future yearly working life sequences over a 15-year period.

(TIF)

Click here for additional data file.^{(894.6KB, tif)}

S3 Fig. Sequence index plots of different states of future yearly working life sequences over a 15-year period by sex.

(TIF)

Click here for additional data file.^{(830.4KB, tif)}

S4 Fig. Hierarchical cluster tree of sequences of future yearly working life sequences over a 15-year period.

(TIF)

Click here for additional data file.^{(979KB, tif)}

S5 Fig. Measures of cluster partitions quality for different numbers of sequence clusters.

ASW: Average Silhouette Width HG: Hubert’s Gamma, PBC: Point Biserial Correlation HC: Hubert’s C.

(TIF)

Click here for additional data file.^{(932.2KB, tif)}

S6 Fig. Mean time spent in each sequence states by sequence clusters.

(TIF)

Click here for additional data file.^{(1.1MB, tif)}

S7 Fig. Sequence index plots of four clusters of working life sequences.

(TIF)

Click here for additional data file.^{(820.2KB, tif)}

S8 Fig. Sequence index plots of six clusters of working life sequences.

(TIF)

Click here for additional data file.^{(802.8KB, tif)}

S9 Fig. Sequence index plots of seven clusters of working life sequences.

(TIF)

Click here for additional data file.^{(998.6KB, tif)}

S10 Fig. Measures of cluster partitions quality for different numbers of sequence clusters by women and men.

(TIF)

Click here for additional data file.^{(1.2MB, tif)}

S11 Fig. Sequence index plots of four clusters of working life sequences for women.

(TIF)

Click here for additional data file.^{(930.6KB, tif)}

S12 Fig. Mean time spent in each sequence states by sequence clusters for women.

(TIF)

Click here for additional data file.^{(897.3KB, tif)}

S13 Fig. Sequence index plots of four clusters of working life sequences for men.

(TIF)

Click here for additional data file.^{(920KB, tif)}

S14 Fig. Mean time spent in each sequence states by sequence clusters for men.

(TIF)

Click here for additional data file.^{(883KB, tif)}

S1 Table. Transition probabilities between different activity.

(DOCX)

Click here for additional data file.^{(36.6KB, docx)}

S2 Table. Measures of cluster partitions quality for different numbers of clusters.

(DOCX)

Click here for additional data file.^{(26.1KB, docx)}

S3 Table. Measures of cluster partitions quality for different numbers of clusters for women and men.

(DOCX)

Click here for additional data file.^{(23.6KB, docx)}

S4 Table. Multinomial regression analysis for the associations between sex and different working life sequence clusters for alternative 6 clusters solutions.

(DOCX)

Click here for additional data file.^{(22.1KB, docx)}

S5 Table. Associations between baseline factors and cluster membership among women.

(DOCX)

Click here for additional data file.^{(52.7KB, docx)}

S6 Table. Associations between predictors and cluster membership among men.

(DOCX)

Click here for additional data file.^{(52.3KB, docx)}

Data Availability Statement

[pone.0281056.ref001] 1.Rechel B, Grundy E, Robine J-M, Cylus J, Mackenbach JP, Knai C, et al. Ageing in the European Union. Lancet Lond Engl. 2013;381: 1312–1322. doi: 10.1016/S0140-6736(12)62087-X [DOI] [PubMed] [Google Scholar]

[pone.0281056.ref002] 2.Edge CE, Cooper AM, Coffey M. Barriers and facilitators to extended working lives in Europe: a gender focus. Public Health Rev. 2017;38: 2. doi: 10.1186/s40985-017-0053-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref003] 3.Stafford M., Lacey R., Murray E., Carr E, Fleischmann M, Stansfeld S, et al. Work–family life course patterns and work participation in later life. Eur J Ageing 2019; 16, 83–94. doi: 10.1007/s10433-018-0470-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref004] 4.Worts D., Corna L, Sacker A, McMunn A, McDonough P. Understanding older adults’ labour market trajectories: A comparative gendered life course perspective. Longitudinal and Life Course Studies,2016; 7,347–367. doi: 10.14301/llcs.v7i4.389 [DOI] [Google Scholar]

[pone.0281056.ref005] 5.Gonäs L, Wikman A, Vaez M, Alexanderson K, Gustafsson K. Gender segregation of occupations and sustainable employment: A prospective population-based cohort study. Scand J Public Health. 2019;47: 348–356. doi: 10.1177/1403494818785255 [DOI] [PubMed] [Google Scholar]

[pone.0281056.ref006] 6.Virtanen M, Oksanen T, Pentti J, Ervasti J, Head J, Stenholm S, et al. Occupational class and working beyond the retirement age: a cohort study. Scand J Work Environ Health. 2017;43: 426–435. doi: 10.5271/sjweh.3645 [DOI] [PubMed] [Google Scholar]

[pone.0281056.ref007] 7.Campos-Serna J, Ronda-Pérez E, Artazcoz L, Moen BE, Benavides FG. Gender inequalities in occupational health related to the unequal distribution of working and employment conditions: a systematic review. Int J Equity Health. 2013;12: 57. doi: 10.1186/1475-9276-12-57 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref008] 8.Löve J, Hensing G, Holmgren K, Torén K. Explaining the social gradient in sickness absence: a study of a general working population in Sweden. BMC Public Health. 2013;13: 545. doi: 10.1186/1471-2458-13-545 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref009] 9.Virtanen M, Oksanen T, Batty GD, Ala-Mursula L, Salo P, Elovainio M, et al. Extending employment beyond the pensionable age: a cohort study of the influence of chronic diseases, health risk factors, and working conditions. PloS One. 2014;9: e88695. doi: 10.1371/journal.pone.0088695 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref010] 10.Kivimäki M, Elovainio M, Vahtera J. Workplace bullying and sickness absence in hospital staff. Occup Environ Med. 2000;57: 656–660. doi: 10.1136/oem.57.10.656 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref011] 11.Nielsen MB, Indregard A-MR, Øverland S. Workplace bullying and sickness absence: a systematic review and meta-analysis of the research literature. Scand J Work Environ Health. 2016;42: 359–370. doi: 10.5271/sjweh.3579 [DOI] [PubMed] [Google Scholar]

[pone.0281056.ref012] 12.Waenerlund A-K, Gustafsson PE, Hammarström A, Virtanen P. History of labour market attachment as a determinant of health status: a 12-year follow-up of the Northern Swedish Cohort. BMJ Open. 2014;4: e004053. doi: 10.1136/bmjopen-2013-004053 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref013] 13.Duchaine CS, Aubé K, Gilbert-Ouimet M, Vézina M, Ndjaboué R, Massamba V, et al. Psychosocial Stressors at Work and the Risk of Sickness Absence Due to a Diagnosed Mental Disorder: A Systematic Review and Meta-analysis. JAMA Psychiatry. 2020;77: 842–851. doi: 10.1001/jamapsychiatry.2020.0322 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref014] 14.Knardahl S, Johannessen HA, Sterud T, Härmä M, Rugulies R, Seitsamo J, et al. The contribution from psychological, social, and organizational work factors to risk of disability retirement: a systematic review with meta-analyses. BMC Public Health. 2017;17: 176. doi: 10.1186/s12889-017-4059-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref015] 15.Gilbert-Ouimet M, Trudel X, Aubé K, Ndjaboue R, Duchaine C, Blanchette C, et al. Differences between women and men in the relationship between psychosocial stressors at work and work absence due to mental health problem. Occup Environ Med. 2020;77: 603. doi: 10.1136/oemed-2019-106242 [DOI] [PubMed] [Google Scholar]

[pone.0281056.ref016] 16.McMunn A, Lacey R, Worts D, McDonough P, Stafford M, Booker C, et al. De-standardization and gender convergence in work–family life courses in Great Britain: A multi-channel sequence analysis. Advances in Life Course Research. 2015;26: 60–75. doi: 10.1016/j.alcr.2015.06.002 [DOI] [Google Scholar]

[pone.0281056.ref017] 17.McMunn A, Lacey R, Worts D, Kuh D, McDonough P, Sacker A.Work-family life courses and psychological distress: Evidence from three British birth cohort studies, Advances in Life Course Research,2021;50100429, ISSN 1040-2608, doi: 10.1016/j.alcr.2021.100429 [DOI] [PubMed] [Google Scholar]

[pone.0281056.ref018] 18.Lacey R., Stafford M., Sacker A. et al. Work-Family Life Courses and Subjective Wellbeing in the MRC National Survey of Health and Development (the 1946 British birth cohort study). Population Ageing 9, 69–89 (2016). doi: 10.1007/s12062-015-9126-y [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref019] 19.WAHRENDORF M. (2015). Previous employment histories and quality of life in older ages: Sequence analyses using SHARELIFE. Ageing and Society, 35(9), 1928–1959. doi: 10.1017/S0144686X14000713 [DOI] [Google Scholar]

[pone.0281056.ref020] 20.Lacey R., Sacker A., Bell S. et al. Work-family life courses and BMI trajectories in three British birth cohorts. Int J Obes 41, 332–339 (2017). doi: 10.1038/ijo.2016.197 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref021] 21.Amick BC, McLeod CB, Bültmann U. Labor markets and health: an integrated life course perspective. Scand J Work Environ Health. 2016;42: 346–353. doi: 10.5271/sjweh.3567 [DOI] [PubMed] [Google Scholar]

[pone.0281056.ref022] 22.Abbott A. Sequence Analysis: New Methods for Old Ideas. Annu Rev Sociol. 1995;21: 93–113. doi: 10.1146/annurev.so.21.080195.000521 [DOI] [Google Scholar]

[pone.0281056.ref023] 23.Macidoe H, Abbott A. Sequence analysis and optimal matching techniques for social science data. Handbook of Data Analysis. SAGE Publications, Ltd; 2004. pp. 387–406. doi: 10.4135/9781848608184 [DOI] [Google Scholar]

[pone.0281056.ref024] 24.World Medical Association Declaration of Helsinki; Ethical Principles for Medical Research Involving Human Subjects. Bulletin of the World Health Organization. 2001;79(4):373–4. [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref025] 25.Ludvigsson J, Håberg S, Knudsen G, Lafolie P, Zoega H, Sarkkola C, et al. Ethical aspects of registry-based research in the Nordic countries. Clinical Epidemiology. 2015;7:491–508. doi: 10.2147/CLEP.S90589 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref026] 26.Statistics Sweden. Living Conditions Surveys (ULF/SILC. Available: https://www.scb.se/en/finding-statistics/statistics-by-subject-area/living-conditions/living-conditions/living-conditions-surveys-ulfsilc/ [Google Scholar]

[pone.0281056.ref027] 27.Ludvigsson JF, Otterblad-Olausson P, Pettersson BU, Ekbom A. The Swedish personal identity number: possibilities and pitfalls in healthcare and medical research. Eur J Epidemiol. 2009;24: 659–667. doi: 10.1007/s10654-009-9350-y [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref028] 28.Ludvigsson JF, Svedberg P, Olén O, Bruze G, Neovius M. The longitudinal integrated database for health insurance and labour market studies (LISA) and its use in medical research. Eur J Epidemiol. 2019;34: 423–437. doi: 10.1007/s10654-019-00511-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref029] 29.Niklas Österlund. MiDAS—sjukpenning och rehabiliteringspenning (MiDAS—sickness benefit and rehabilitation allowance). Swedish Social Insurance Agency.; 2011. [Google Scholar]

[pone.0281056.ref030] 30.Social Insurance in Figures 2018. Stockholm: The Swedish Social Insturance Agency; 2018. [Google Scholar]

[pone.0281056.ref031] 31.Lundgren B. Recent development in unemployment insurance in Sweden Unemployment Insurance and Employment Maintenance. Brussels: International Social Security Association; 2006. [Google Scholar]

[pone.0281056.ref032] 32.National Agency for Education. Descriptive Data on child care and schools in Sweden in 2000. Report No. 192. Stockholm. [Google Scholar]

[pone.0281056.ref033] 33.Statistic Sweden, Population Statistic (ww.scb.se) [Google Scholar]

[pone.0281056.ref034] 34.Studer M. WeightedCluster Library Manual A practical guide to creating typologies of trajectories in the social sciences with R. LIVES Working Papers, 24.; 2013. Available: 10.12682/lives.2296-1658. 2013.24. [DOI] [Google Scholar]

[pone.0281056.ref035] 35.Gabadinho A, Ritschard G, Müller NS, Studer M. Analyzing and Visualizing State Sequences in R with TraMineR. J Stat Softw. 2011;40: 1–37. doi: 10.18637/jss.v040.i04 [DOI] [Google Scholar]

[pone.0281056.ref036] 36.Weisshaar K, Cabello-Hutt T. Labor Force Participation Over the Life Course: The Long-Term Effects of Employment Trajectories on Wages and the Gendered Payoff to Employment. Demography. 2020;57: 33–60. doi: 10.1007/s13524-019-00845-8 [DOI] [PubMed] [Google Scholar]

[pone.0281056.ref037] 37.Kuitto K, Salonen J, Helmdag J. Gender Inequalities in Early Career Trajectories and Parental Leaves: Evidence from a Nordic Welfare State. Soc Sci. 2019;8. doi: 10.3390/socsci8090253 [DOI] [Google Scholar]

[pone.0281056.ref038] 38.Erosa A, Fuster L, Restuccia D. Fertility Decisions and Gender Differences in Labor Turnover, Employment, and Wages. Rev Econ Dyn. 2002;5: 856–891. doi: 10.1006/redy.2002.0195 [DOI] [Google Scholar]

[pone.0281056.ref039] 39.Duvander A-Z, Ferrarini T, Thalberg S. Swedish parental leave and gender equality Achievements and reform challenges in a European perspective. Institutet för Framtidsstudier; 2005. Report No.: 91-89655-69–9. Available: https://www.iffs.se/media/1118/20051201134956filu8yijlraac7u4fv7gumy.pd [Google Scholar]

[pone.0281056.ref040] 40.Evertsson M, Duvander A-Z. Parental Leave—Possibility or Trap? Does Family Leave Length Effect Swedish Women’s Labour Market Opportunities? Eur Sociol Rev. 2011;27: 435–450. [Google Scholar]

[pone.0281056.ref041] 41.Klein J, Reini K, Saarela J. Sickness Absence and Disability Pension in the Very Long Term: A Finnish Register-Based Study With 20 Years Follow-Up. Front Public Health. 2021;9: 142. doi: 10.3389/fpubh.2021.556648 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref042] 42.de Wind A, Geuskens GA, Ybema JF, Blatter BM, Burdorf A, Bongers PM, et al. Health, job characteristics, skills, and social and financial factors in relation to early retirement–results from a longitudinal study in the Netherlands. Scand J Work Environ Health. 2014; 186–194. doi: 10.5271/sjweh.3393 [DOI] [PubMed] [Google Scholar]

[pone.0281056.ref043] 43.Mussino E, Duvander A-Z. Use It or Save It? Migration Background and Parental Leave Uptake in Sweden. Eur J Popul. 2016;32: 189–210. doi: 10.1007/s10680-015-9365-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0281056.ref044] 44.Ludvigsson JF, Almqvist C, Bonamy A-KE, Ljung R, Michaëlsson K, Neovius M, et al. Registers of the Swedish total population and their use in medical research. Eur J Epidemiol. 2016;31: 125–136. doi: 10.1007/s10654-016-0117-y [DOI] [PubMed] [Google Scholar]

PERMALINK

Working life sequences over the life course among 9269 women and men in Sweden; a prospective cohort study

Katalin Gémes

Katriina Heikkilä

Kristina Alexanderson

Kristin Farrants

Ellenor Mittendorfer-Rutz

Marianna Virtanen

Roles

Abstract

Objectives

Methods

Results

Conclusions

Introduction

Materials and methods

Survey data

Register data

Social security in Sweden

Statistical analysis

Results

Table 1. Baseline characteristic of the study participants.

Fig 1. State distribution plots and mean time in different states of future yearly working life sequences over a 15-year period.

Fig 2. State distribution plots of future yearly working life sequences over a 15-year period, among women and men, respectively.

Fig 3. Sequence index plots of the five clusters of working life sequences.

Table 2. Multinomial regression analysis for the associations between sex and different working life sequence clusters.

Discussion

Strengths and limitations

Conclusions

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Renuka Sane

Roles

Author response to Decision Letter 0

Decision Letter 1

Sreeram V Ramagopalan

Roles

Acceptance letter

Sreeram V Ramagopalan

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases