Childhood adversity, adult socioeconomic status and risk of work disability: a prospective cohort study.

OBJECTIVES: To examine the combined effects of childhood adversities and low adult socioeconomic status (SES) on the risk of future work disability.
METHODS: Included were 34 384 employed Finnish Public Sector study participants who responded to questions about childhood adversities (none vs any adversity, eg, parental divorce or financial difficulties) in 2008, and whose adult SES in 2008 was available. We categorised exposure into four groups: neither (reference), childhood adversity only, low SES only or both. Participants were followed from 2009 until the first period of register-based work disability (sickness absence >9 days or disability pension) due to any cause, musculoskeletal or mental disorders; retirement; death or end of follow-up (December 2011). We ran Cox proportional hazard models adjusted for behavioural, health-related and work-related covariates, and calculated synergy indices for the combined effects.
RESULTS: When compared with those with neither exposure, HR for work disability from any cause was increased among participants with childhood adversity, with low SES, and those with both exposures. The highest hazard was observed in those with both exposures: HR 2.53, 95% CI 2.29 to 2.79 for musculoskeletal disability, 1.55, 95% CI 1.36 to 1.78 for disability due to mental disorders and 1.29, 95% CI 1.20 to 1.39 for disability due to other reasons. The synergy indices did not indicate synergistic effects.
CONCLUSIONS: These findings indicate that childhood psychosocial adversity and low adult SES are additive risk factors for work disability. © Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2017. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

Both childhood adversities and low socioeconomic status in adulthood have been linked to increased work disability, but their combined effects are not known.

In a study of over 34 000 public sector employees, the highest risk of work disability was observed among those with both exposures.

The combined effects of childhood adversities and low adult socioeconomic status were additive rather than synergistic.

Introduction

Work disability is a major economic burden in the developed countries.1 Throughout the developed world, musculoskeletal disorders are the third leading cause of overall disease burden and the leading cause of work disability followed by mental disorders.1–5 At an individual level, the harmful effects of work disability include increased risks for lower income level, becoming excluded from the labour market, and increased risk of death.6–8

Employees’ health is dependent on individual and social factors operating from across the life course, including those related to workplace conditions.9 Studies have shown that negative life events during childhood may predict decreased labour market participation in young adults,10 as well as disability pension and long-term sickness absence in adult women.11 One study found that the risk of disability retirement increased in a dose–response manner with a growing number of childhood adversities.12

A major adverse exposure in adulthood is low socioeconomic status (SES). Conventional indicators of low SES, such as low education or low occupational status, as well as other measures including poor financial situation, and poor housing have been linked to increased risk of work disability.13–18 High SES, in turn, has been associated with lower risk of long-term work disability due to psychiatric disorders.19 A combination of childhood adversity and low SES in adulthood has been shown to associate with behavioural risk factors of disability retirement, such as obesity, smoking and physical inactivity.20

The childhood and adult exposures may have independent, additive or synergistic effects on adult health and work disability. Exposure to childhood adversities may trigger a long-term pattern of heightened psychological and physiological reactivity to stress, a biological pathway which may characterise important life course influence.21 Dysregulated or chronically activated stress responses have been linked to the progression of physical illnesses over time,22 and may explain why early adversities make individuals vulnerable to poor health.23 Furthermore, social pathways may contribute to adverse life course effects,21 because persons exposed to adverse experiences in childhood may be more likely to face further interpersonal problems later in life.20 However, the combined effects of exposure to childhood adversities and low SES in adulthood on work disability have not been examined.

To address this gap in evidence, we examined whether the associations for childhood adversity and low adult SES with work disability are independent, additive or synergistic. In addition to overall work disability, work disability due to musculoskeletal and mental disorders was analysed separately as these are the leading single causes of work disability.

Methods

Study participants

Participants of the Finnish Public Sector (FPS) study, an ongoing cohort study including employees of 10 towns and 6 hospital districts in Finland, formed the basis for the study population. The participants are from a wide range of occupations, from administrative personnel and professionals (eg, doctors) to semiskilled and unskilled workers (eg, cleaners). Surveys were sent to all permanent employees and those with a long-term contract with the target organisations at the time of the survey. Surveys have been repeated by 2–4 year intervals with responding always being voluntary.

Selection of the participants is presented in figure 1. Information on childhood adversities was collected in the 2008 survey that was sent to 50 284 eligible employees of the 10 towns (response rate 71%, n=35 498). Of the responders, 8 died and 599 retired before the start of follow-up, 366 had missing information on childhood adversities, 507 had no data on adult SES (ie, occupational status in 2008) and 2545 had no information for some of the covariates and they were all excluded. This resulted in an analytical sample of 31 473 participants. All register data were linked to the participants of the analytical sample using personal identification codes that were then recoded for confidentiality into research identification codes for the analyses. The linkages were complete for all participants. The analytic sample was similar to the sample with missing data on covariates in terms of proportion of men and low SES in adulthood, as well as according to mean age and disability rate (see online supplementary table S1). Compared with the eligible population, the analytic sample included slightly less men (22% vs 28%) and those with low adult SES (43% vs 48%). The mean age of the analytic sample was 47.1 (SD=9.6) years and that of the eligible 47.2 (SD=9.7) years.

Figure 1

Flow chart of the selection of the analytic sample. SES, socioeconomic status. 

Measurement of childhood adversity and low SES in adulthood

The questions regarding childhood adversities were modified from the Survey of Living Conditions developed by Statistics Finland.24 The participants were asked whether they had experienced any of the following six adversities in their childhood: divorce/separation of the parents, long-term financial difficulties in the family, serious conflicts in the family, frequent fear of a family member, serious or chronic illness of a family member or an alcohol problem of a family member (response categories for each item: no, yes or cannot say). Response category ‘cannot say’ was coded as a missing value (4% of all responses). The prevalence of the individual adversities ranged from 14.3% (fear of a family member) to 28.9% (serious or chronic illness of a family member) (see online supplementary table S2). For the analyses including calculation of synergy indices of the two exposures, we formed a binary variable for the childhood adversities (none vs any). In our preliminary analysis using a summary variable from the six questionnaire items (range from 0 to 6), even having one adversity was associated with a slightly increased risk of all-cause work disability (HR compared with having no adversities 1.05, 95% CI 1.00 to 1.11) and having all six adversities was associated with the greatest risk (HR 1.72, 95% CI 1.47 to 2.02). In a separate analysis of specific adversities, each individual adversity was associated with an increased disability risk (range of HRs from 1.14 (95% CI 1.09 to 1.19) for long-term illness to 1.31 (95% CI 1.24 to 1.38) for fear of a family member) (see online supplementary table S3).

Occupational position was used as an indicator of adult SES. For each participant, we obtained International Classification of Occupations (ISCO) codes for year 2008 from the employers’ registers. We dichotomised the occupational positions25 into ‘high’ (=ISCO classes 1–3; including managers, professionals and technicians) and ‘low’ (=ISCO classes 4–9; including clerical, service and manual workers) category like in a previous study.26

For the analyses, the participants were classified into four categories of the combined exposure: neither (no childhood adversity nor low adult SES, the reference group); childhood adversity only; low adult SES only and both (childhood adversity and low adult SES).

Ascertainment of work disability

We obtained data on all granted work disability periods from the registers of the Social Insurance Institution (SII) of Finland (sickness absence) and the Finnish Centre for Pensions (temporary and permanent work disability pensions). An episode of work disability could include a period of sickness absence beyond nine work days, or a period of temporary or permanent disability pension, or both, if sickness absence was followed by disability pension. Nine work days is the cut-off for sickness absence periods as the SII only reimburses and records sickness absence periods lasting more than nine work days. Disability pension can be granted for a person whose capacity for work is severely and long-term impaired by at least 60% by an illness diagnosed by a doctor.27

We examined work disability for any cause, and specifically due to musculoskeletal (International Classification of Diseases (IDC) 10th revision codes M00-M99) and mental (ICD–10 codes F00-F99) disorders and other causes. Most disability periods were due to sickness absence (93%). Of all participants, 2.5% were granted disability pension; of these musculoskeletal disorders covered 48% and mental disorders 20%. Due to low numbers, however, the disability pension was not analysed by diagnostic groups.

Assessment of covariates

Information about age and sex was derived from the employers’ registers. Age and sex were used as confounders in all analyses. With only one baseline assessment, we were not able to reliably separate whether the other covariates act as confounders or as mediators in the associations of childhood adversity and adulthood SES with work disability. From the questionnaires we obtained information about factors that have been linked to increased risk of work disability: marital status28 (married or cohabiting vs not); smoking16 (current smoker vs not); heavy alcohol use16 29 (consumption >24 units/week in men and >16 units/week in women (yes vs no)30; physical activity16 (measured as weekly Metabolic Equivalent Task (MET) hours31 and categorised as <14 MET hours/week= low and ≥14 MET hours/week= moderate-high)31; body mass index16 (BMI) (measured as weight/height squared in metres, kg/m2, categorised as ≤25: normal weight; 25–29.9: overweight; ≥30 obese)32; and psychological distress28 (measured by a psychological distress scale using the 12–item General Health Questionnaire33 with participants with a summary score of four or more were coded as cases of psychological distress).34 Chronic diseases (hypertension, cardiac failure, ischaemic heart disease, diabetes, asthma or other chronic obstructive lung disease, rheumatoid arthritis and mental disorders) were identified from the Drug Reimbursement Register maintained by the SII of Finland.35 Information on all cancers diagnosed in 2001 through 2008 was obtained from the Finnish Cancer Registry.36 Information about work-related covariates possibly related to work disability, job strain16 (ie, high job demands and low job control based on questions from the Job Content Questionnaire)37 and shift work (no shift work vs any type of shift work) was also obtained from the questionnaires. Type of job contract (fixed-term vs permanent) was from the employers’ registers.

Statistical analysis

We first examined the associations of the combined exposure variable with the health-related and work-related covariates. We used log-binomial regression models and present the results as age-adjusted and sex-adjusted prevalence ratios (PR) and their 95% CIs for each category of the exposure variable. The ‘neither’ category (no childhood psychosocial adversity and no low adult SES) served as the reference.

We then examined whether the health-related and work-related covariates predicted work disability using Cox proportional hazard models adjusting for age and sex. Participants employed at baseline (1 January 2009, that is, after the 2008 survey requesting childhood adversities) were followed up to the first period of work disability, statutory retirement, death or end of follow-up (ie, end of register data update on 31 December 2011). The results of this analysis are presented as HRs with 95% CI.

The main analysis examined the associations between the combined exposure to childhood adversity and low adult SES and work disability using the Cox proportional hazards models. We ran two models with different adjustments: model 1 was adjusted for age and sex that could be considered confounders. In model 2 we additionally adjusted for the possible mediators/confounders: marital status, smoking, heavy alcohol use, physical activity, BMI, chronic disease, psychological distress, job strain, shift work and type of job contract. We checked the proportionality assumption of the hazard models by calculating group*time interaction. p Value of 0.12 for the interaction indicated there was no evidence against the proportional hazards.

Finally, we calculated the synergistic effect of the two exposures using synergy index that indicates relative excess risk due to interaction between the two exposures.38–41 Synergy index values >1 indicate that the effects of adversity in childhood and SES in adulthood in combination are greater than one would expect from these factors in isolation.42 The formula for calculating synergy index for two dichotomised variables is: S = (RRA+B+−1) / [(RRA+B-−1) + (RRA-B+−1)]. Applied to the present study, RRA+B+ is the relative risk of work disability if both factors A (childhood adversity) and B (low adult SES) are present, RRA+B− is the relative risk of disability if A is present but B is absent and RRA-B+ is the relative risk of disability if B is present but A is absent.

As a sensitivity analysis, we used educational level as an additional indicator of adult SES. Information about education was obtained from the registers of Statistics Finland43 and was available for all participants. We categorised educational level as ‘high’ (=university degree) and 'intermediate/low' (=high school, or vocational school, or basic education). We ran models also for disability pension only, and using categorisation 0–1 vs 2–6 for the childhood adversities that was used in our previous study.20

Results

Mean follow-up time for all-cause work disability was 2.3 (SD=1.0) years. Of the participants, 78% were female and the mean age of the participants at baseline was 47.1 (range 19–68) years. Descriptive statistics of the study population are presented by the four exposure categories in table 1. The largest exposure group (33% of all participants) was the group of participants with at least one childhood adversity (and high SES) followed by the group with both exposures (28% of all participants).

Table 1

Descriptive statistics of the study population at baseline by childhood adversity and adult socioeconomic status (SES)

	All (n=31 473)	Neither (n=7342)	Childhood adversity only (n=10 613)	Low adult SES only (n=4868)	Both (n=8650)
Confounder/mediator	N (% of all participants)
Sex (p value<0.001)
Men	7 010	1855 (26)	2391 (34)	1097 (16)	1667 (24)
Women	24 463	5487 (22)	8222 (34)	3771 (15)	6983 (29)
Age group, years  (<0.001)
<35	3 884	1016 (26)	1303 (34)	555 (14)	1010 (26)
35–50	14 728	3608 (25)	5115 (35)	2113 (14)	3892 (26)
>50	12 861	2718 (21)	4195 (33)	2200 (17)	3748 (29)
Married or cohabiting (<0.001)
Yes	23 771	5903 (25)	8073 (34)	3606 (15)	6129 (26)
No	7 762	1439 (19)	2540 (33)	1262 (16)	2521 (32)
Smoking (<0.001)
No	26 582	6678 (25)	9412 (35)	3925 (15)	6567 (25)
Yes	4 891	664 (14)	1201 (24)	943 (19)	2083 (43)
Heavy alcohol use (<0.001)
No	28 340	6599 (23)	9404 (33)	4489 (16)	7848 (28)
Yes	3 133	743 (24)	1209 (39)	379 (12)	802 (26)
Physical activity (<0.001)
Moderate	23 328	5755 (25)	8097 (35)	3490 (15)	5986 (26)
Low	8 145	1587 (19)	2516 (31)	1378 (17)	2664 (33)
Body mass index (<0.001)
<25	15 286	4022 (26)	5439 (36)	2179 (14)	3646 (24)
25–30	10 889	2372 (22)	3581 (33)	1762 (16)	3174 (29)
>30	5 298	948 (18)	1593 (30)	927 (18)	1830 (34)
Chronic disease (<0.001)
No	26 688	6427 (24)	9085 (34)	4068 (15)	7108 (27)
Yes	4 785	915 (19)	1528 (32)	800 (17)	1542 (32)
Psychological distress (<0.001)
No	24 044	5904 (24)	7716 (32)	4027 (17)	6397 (27)
Yes	7 429	1438 (20)	2897 (39)	841 (11)	2253 (30)
Job strain (<0.001)
No	24 385	6265 (26)	8686 (36)	3500 (14)	5934 (24)
Yes	7 088	1077 (15)	1927 (27)	1368 (19)	2716 (39)
Shift work (<0.001)
No	25 012	6627 (26)	9230 (37)	3376 (14)	5779 (23)
Yes	6 461	715 (11)	1383 (21)	1492 (23)	2871 (45)
Job contract (<0.001)
Permanent	28 475	6568 (23)	9379 (33)	4551 (16)	7977 (28)
Fixed-term	2 998	774 (26)	1234 (41)	317 (11)	673 (22)

Childhood adversity, low adult SES and their combination were associated with an increased risk of smoking, low physical activity, obesity and having a chronic disease (table 2). Childhood adversity and low adult SES were synergistically associated with smoking. Exposure to childhood adversities only also increased the risk of heavy alcohol use and psychological distress. Both exposures alone and together were also associated with job strain and shift work, but only childhood adversity was associated with fixed-term job contract (table 2). Low adult SES was inversely associated with heavy alcohol consumption and fixed-term job contract. The exposures and the health-related and work-related covariates were also positively associated with the work disability outcomes, with the exception of the fixed-term job contract (see online supplementary table S4).

Table 2

Baseline associations (prevalence ratio, PR)* of exposure to childhood adversity and low adult socioeconomic status (SES) with risk factors

Exposure		Smoking		Heavy alcohol use		Low physical activity		Obesity		Chronic disease		Psychological distress
	N	PR	95% CI	PR	95% CI	PR	95% CI	PR	95% CI	PR	95% CI	PR	95% CI
Neither	7 342	1.00		1.00		1.00		1.00		1.00		1.00
Childhood adversity only	10 613	1.28	1.17 to 1.40	1.13	1.04 to 1.22	1.08	1.02 to 1.14	1.16	1.08 to 1.25	1.12	1.04 to 1.20	1.38	1.31 to 1.46
Low adult SES only	4 868	2.18	2.00 to 2.39	0.73	0.65 to 0.82	1.29	1.22 to 1.37	1.42	1.31 to 1.54	1.21	1.11 to 1.31	0.88	0.825 to 0.9
Both	8 650	2.76	2.55 to 2.99	0.89	0.81 to 0.97	1.41	1.33 to 1.48	1.61	1.50 to 1.72	1.37	1.28 to 1.47	1.32	1.25 to 1.40
Synergy index		1.20	1.02 to 1.41	0.77	0.29 to 2.00	1.10	0.86 to 1.42	1.04	0.80 to 1.34	1.15	0.77 to 1.72	1.22	0.84 to 1.77
		Job strain		Shift work		Fixed-term contract
Neither	7 342	1.00		1.00		1.00
Childhood adversity only	10 613	1.23	1.15 to 1.31	1.35	1.24 to 1.46	1.16	1.07 to 1.26
Low adult SES only	4 868	1.90	1.77 to 2.03	3.20	2.96 to 3.46	0.70	0.62 to 0.79
Both	8 650	2.08	1.95 to 2.21	3.43	3.19 to 3.68	0.78	0.71 to 0.86
Synergy index		0.96	0.84 to 1.10	0.95	0.86 to 1.06	1.51	0.61 to 3.77

*Adjusted for age and sex.

In figure 2 we show crude cumulative hazards of work disability by the four exposure categories of childhood adversity and adult SES. The highest cumulative probability during the follow-up was observed among those exposed to both (46%). Results of the main analysis are presented in table 3. The combined exposure was associated with a 1.95–fold (95% CI 1.85 to 2.06) hazard of work disability for any cause in the model adjusted for age and sex (model 1). The corresponding HR in a model including those with missing values for possible confounders/mediators was 1.95 (95% CI 1.85 to 2.05). The fully adjusted HR was 1.62 (95% CI 1.53 to 1.71). After adjustment for the possible confounders/mediators, the HR among those with both exposures was 2.53 (95% CI 2.29 to 2.79) for disability due to musculoskeletal disorders, 1.55 (95% CI 1.36 to 1.78) for disability due to mental disorders and 1.29 (95% CI 1.20 to 1.39) for disability due to other reasons (table 3). The synergy index was 1.1 in all baseline models and 0.9–1.0 in the fully adjusted models indicating little synergistic effects. Childhood adversity and low adult SES were independently associated with an increased risk of work disability. Adjustment for childhood adversity attenuated the association between low adulthood SES and work disability only slightly (3%) the adjusted HR for low adulthood SES being 1.69 (95% CI 1.62 to 1.75) (see online supplementary table S3).

Figure 2

Non-adjusted cumulative hazard of the occurrence of work disability by childhood adversity and adult socioeconomic status (SES).

Table 3

HRs for subsequent work disability* by childhood adversity and low adult socioeconomic status (SES)

Exposure		Model 1†		Model 2‡
	N/events	HR	95% CI	HR	95% CI
Work disability, any cause	31 473/11 820
Neither	7342/2035	1.00		1.00
Childhood adversity only	10 613/3361	1.15	1.09 to 1.22	1.08	1.02 to 1.14
Low adult SES only	4868/2188	1.71	1.60 to 1.82	1.52	1.42 to 1.62
Both	8650/4236	1.95	1.85 to 2.06	1.62	1.53 to 1.71
Synergy index		1.11	0.96 to 1.27	1.03	0.85 to 1.25
Musculoskeletal (ICD-10 M00-M99)	31 473/4889
Neither	7342/603	1.00		1.00
Childhood adversity only	10 613/1036	1.18	1.06 to 1.31	1.11	1.00 to 1.23
Low adult SES only	4868/1099	2.78	2.50 to 3.08	2.38	2.14 to 2.64
Both	8650/2151	3.15	2.87 to 3.46	2.53	2.29 to 2.79
Synergy index		1.10	1.01 to 1.20	1.03	0.93 to 1.15
Mental (ICD-10 F00-F99)	31 473/2357
Neither	7342/364	1.00		1.00
Childhood adversity only	10 613/813	1.57	1.38 to 1.78	1.40	1.23 to 1.59
Low adult SES only	4868/327	1.34	1.15 to 1.57	1.21	1.03 to 1.42
Both	8650/853	1.96	1.72 to 2.22	1.55	1.36 to 1.78
Synergy index		1.05	0.91 to 1.22	0.91	0.74 to 1.12
Other (ICD-10 not F or M)	31 473/7019
Neither	7342/1320	1.00		1.00
Childhood adversity only	10 613/2073	1.07	1.00 to 1.15	1.02	0.95 to 1.10
Low adult SES only	4868/1257	1.40	1.29 to 1.52	1.26	1.16 to 1.36
Both	8650/2369	1.52	1.42 to 1.63	1.29	1.20 to 1.39
Synergy index		1.09	0.88 to 1.36	1.04	0.73 to 1.48

*Sickness absence >9 days or work disability pension,

†Model 1 adjusted for sex, age,

‡Model 2 adjusted for sex, age, marital status, smoking, heavy alcohol use, physical activity, body mass index, chronic disease, psychological distress, job strain, shift work and type of job contract.

ICD-10, International Classification of Diseases 10th  revision diagnosis codes.

The sensitivity analysis using education as an indicator of adult SES resulted in largely similar findings (HR for all-cause disability 1.67, 95% CI 1.58 to 1.77 in model 2) (see online supplementary table S5). Analyses for disability pension only resulted in similar but slightly stronger associations for all-cause disability; the fully adjusted HR for disability pension was 2.79 (95% CI 2.14 to 3.65) among those exposed to childhood adversity and low adult SES. The corresponding HR was 2.96 (95% CI 2.24 to 3.91) among those exposed to childhood adversity and low education (see online supplementary table S6). Analyses using different categorisation of the childhood adversities also resulted in similar findings (HR for all-cause disability 1.69, 95% CI 1.60 to 1.79 in model 3) (see online supplementary table S7).

Discussion

We found that adversities in childhood and low SES in adulthood were independently associated with the risk of work disability. The combined effect of childhood adversities and low SES in adulthood on all-cause work disability as well as disability due to musculoskeletal and mental disorders or due to other reasons was additive rather than synergistic. These findings indicate that childhood adversities together with later exposure to low SES may lead to cumulative health effects such that each additional period of adversity in life adds to the health disadvantage. This interpretation is consistent with the accumulation model described in life course epidemiology.44

We are not aware of previous studies examining the possible synergistic effects of childhood adversity and low adult SES on work disability. Although no synergistic effect of these exposures was observed, both childhood adversity and low adult SES were associated with work disability, which is consistent with previous findings. For instance, in studies focusing only on childhood exposures, adversities have been linked with increased risk of work disability for any cause after controlling for adult conditions,11 12 45 and with lowered labour market participation.10 None of these studies examined cause-specific disability due to musculoskeletal or mental disorders separately. A more recent study suggested that childhood adversities are highly prevalent among young adults on disability pension due to mental disorders,46 which is also in agreement with our findings. A possible pathway from childhood adversity to mental problems is through changes in brain structure and function activated by adversities, which may further mediate the risk of psychopathology.47 48 Another potential pathway is via epigenetic regulation that is based on findings where maternal care in rats was found to affect the hypothalamic-pituitary-adrenal activity that is involved in stress responses also later in life.49

Associations between adult SES and work disability have been previously examined.14 16 45 In a study by Robroek and others a 1.84 times higher likelihood of receiving disability benefit was reported among those with low compared with high education.16 This is similar to our findings for any short-term or long-term work disability, but lower than the likelihood for disability pension among those with low education only. In their study by Robroek and others, the association was attenuated by 62% and did not remain significant after adjustment for health-related and work-related factors, whereas in our study the association remained robust after controlling for these pertinent risk factors. However, the findings are not directly comparable due to different definitions of disability; in the Netherlands, disability benefit could be granted only after 2 years of illness, whereas in Finland disability pension can be granted after 1 year of illness. Furthermore, our measure of any work disability included also sickness absence spells beyond nine work days.

Some previous studies have focused on the associations between adult SES and cause-specific work disability. Like in our study, indicators of low adult SES have been linked to an increased risk of disability due to depression,17 musculoskeletal and mental disorders.18 A possible reason for why low SES was, in particular, associated with work disability due to musculoskeletal disorders is physically demanding work50 that is more common in the low than high occupational groups.

The main strength of this investigation is the large study population and the comprehensive control for the risk factors as well as the high-coverage register data on the outcomes. However, there are also some limitations including the use of retrospectively assessed childhood adversity that is subject to reporting and recall biases. This may have both underestimated and overestimated the associations.51 The use of retrospective measures on childhood adversity also leaves room for differential misclassification error, which may limit the validity of the data. The reliability of the self-reported measure of childhood adversity may be good,52 but the validity of self-reported adversity can be assessed only by means of prospective studies beginning from childhood. There also remains possibility of unmeasured confounding as we had no information from childhood about possible confounders. In addition, persons with childhood adversities may be under-represented in our data because a larger proportion of these people tend to drop out of labour force prematurely or may fail to respond to surveys. Such under-representation may bias the results towards the null. As the study population consisted of the FPS employees that were predominantly female and virtually all of European origin, further studies in more diverse populations from various well-fare regimens and branches of industry are needed to confirm the generalisability of our findings.

Conclusions

Exposure to adversities in childhood and low SES in adulthood were independently associated with work disability. Exposure to both these risk factors was associated with the highest risk of work disability, although this was additive rather than synergistic association. Childhood adversity was associated with disability due to mental disorders in particular, whereas low adult SES was more strongly associated with disability due to musculoskeletal disorders. These findings suggest that psychosocial and socioeconomic exposures from across the life course have an impact on work disability.