Data material
The material of the present project consists of person-based data from municipal jobs and benefits offices that are linked to data from series of national registers.
Data registered at jobs and benefits offices
The public sickness benefit scheme in Denmark covers long-term sickness absence (> 21 days in 2010–2011, > 30 days in 2012) among employed, unemployed, self-employed and assisting spouses. The system is administered by municipal jobs and benefits offices, which according to the Sickness Benefits Act are committed to follow up and continuously evaluate each sick-listed person’s prognosis of return to the labour force [17]. The data to be used in the present project were collected from jobs and benefits offices in 21 (out of 98) Danish municipalities in connection with the above-mentioned Danish return-to-work program [17], which ran from 26 April 2010 to 30 September 2012. The obtained database contains inter alia reason for the sickness absence, the date of the first consultation with the jobs and benefits office and a personal identification number, which enable linkage to data in national registers [24].
Data from national registers
The following registers are used: The Central Person Register (CPR) [25], the Employment Classification Module (ECM) [26] and the Danish Register for Evaluation of Marginalisation (DREAM) [27].
The Central Person Register contains, inter alia, information on gender, family type, addresses and dates of birth, death and migrations for every person who is or has been an inhabitant of Denmark sometime between 1968 and present time. The Employment Classification Module contains annual person-information on, inter alia, the socio-economic status, occupation and industry of the inhabitants of Denmark. DREAM contains weekly person-based information on social transfer payments such as maternity/paternity benefits, sickness-absence benefits, unemployment benefits, social security cash benefits, and state educational grants. It has existed since 1991 and covers all inhabitants of Denmark. The weekly benefits data are recorded if the person has been on a benefit for one or more days of the week. However, since only one type of social transfer payment can be registered per week, types of benefits are prioritized in case of data overlap. The above-mentioned social transfer payments are prioritized in the order listed, i.e. maternity/paternity benefits have higher priority than sickness-absence benefits, which in turn have higher priority than unemployment benefits etc.
Study population and inclusion criteria
The study population consists of all 20–54 year old employed or unemployed people who (according to the jobs and benefits offices in the 21 municipalities of the Danish RTW program) were on long-term sickness-absence due to self-reported depression, anxiety, stress/burnout or mental ill health without further specification, sometime during the period 26 April 2010–30 September 2012. If a person was registered with more than one sickness-absence episode, of the above-mentioned kind during the above-mentioned period, then only the first of the episodes was included in the analyses. To be included in the present study, it was, moreover, required that, from two years prior to the concerned sickness absence episode until five years after the first visit to the jobs and benefits office, the person did not immigrate or emigrate. In total, 19,660 observations/persons fulfilled the inclusion criteria. A chart of inclusions/exclusions are given in Fig. 1.
Outcome variable
Welfare benefits at follow-up
This is a multinomial variable, which is divided into the following categories:
-
Did not receive any social transfer payments other than holiday allowance (DREAM-code: 121), state educational grants (DREAM-codes: 651, 652, 661) or maternity/paternity leave benefits (DREAM-code: 881)
-
Deceased or recipient of health related social transfer payments (DREAM-codes: 750–818, 890 − 818)
-
Recipient of other social transfer payments
The social transfer payments are based on registrations in DREAM. The social transfer payments of category two and three are considered adverse outcomes (welfare benefits received due to unfortunate circumstances). Since death is also an adverse (health-related) outcome, it is included category 2. Maternity benefits, state educational grants and holiday allowance are not considered adverse outcomes and are therefore included in category 1. In the text that follows, category 1 will be referred to as “self-reliant” while category 3 will be referred to as “recipient of non-health related welfare benefits”.
Primary predictors
Industrial sector (last recorded during a two year period preceding baseline)
A person’s main industry in a given calendar year is registered annually in the employment classification module. The industrial codes are based on the industrial classification DB07 [28]. In the present study, the industries are divided into the following sectors: Agriculture, forestry, hunting and fishing (DB07-codes: 01.11–03.22); Manufacturing, mining and quarrying (05.10–33.20); Construction (41.10–43.99); Wholesale and retail trade; repair of motor vehicles and motorcycles (45.11–47.99); Transporting and storage (49.10–53.20); Accommodation and food service (55.10–56.30); Public administration (84.11–84.13); Courts and prisons; Police; Fire Departments (84.23–84.25); Education (85.10–85.60); Human health and social work (86.10–88.99); Other; Unstated.
Job group skill level (last recorded during a two year period preceding baseline)
A person’s main occupation in a given calendar year is registered annually in the employment classification module. The occupations in the present study are divided into the following job group skill levels: Professionals; Technicians and associate professionals; Workers in occupations that require skills at a basic level; Workers in elementary occupations; Workers in occupations without skill requirements, in accordance with the Danish version of the International Standard Classification of Occupations (DISCO) [29].
Secondary predictors
The following secondary predictors were included: Self-reported reason for sickness absence (anxiety; depression; mental ill health not otherwise specified (NOS); stress/burnout), gender, age (10-year classes), family type (married or cohabitant with resident children; married or cohabitant without resident children; single with resident children; single without resident children) and employment status (employed; unemployed). Family type refers to the situation at the end of the calendar year preceding baseline. The other predictors refers to the situation at baseline.
Control variables
The following control variables were included: Unemployment insurance (Yes/No), Danish citizenship (Yes/No), calendar year at the start of the sickness absence episode (< 2012; 2012), time passed between the first day of sickness absence and the baseline visit at the jobs and benefits office (≤ 30; 31–60; >60 days), geographical region at baseline (Capital, Zealand; Southern Denmark; Central Jutland and Northern Jutland), assignment in the Danish RTW-program (intervention group; control group; not eligible for participation), weeks with health related social transfer payments during a two-year period prior to the baseline sickness absence episode (0; 1–26 ; >26) and ditto for non-health related social transfer payments (other than state educational grants and maternity/paternity leave benefits).
Statistical analyses
With outcome category 1 (self-reliant) as reference, multinomial logistic regression was used to estimate odds ratios (OR), with 99% confidence interval (CI), for being in outcome category 2 and 3 (“Deceased or recipient of health related welfare benefits” and “Recipient of non-health related welfare benefits”) at 1, 3 and 5 years after the baseline interview, as a function of job group skill level, industrial sector, reason for sickness absence, gender, age, family type and employment status. The effects of job group skill level, industrial sector, gender, age, family type and employment status was estimated in a model that included all of the variables in the sections entitled “Primary predictors”, “Secondary predictors” and “Control variables”. The effects of “reason for sickness absence” was estimated in a model that included all of the above-mentioned variables except for the variable named “assignment in the Danish RTW-program”. Likelihood ratio tests were used to test the null-hypotheses, which stated that the distribution of the outcome categories is independent of job group skill level and industrial sector, respectively. The hypotheses were tested for the status at 1, 3 and 5 years after the baseline interview, respectively. Sub-hypotheses, which stated that the odds-ratio for health and non-health related social transfer payments, respectively, is independent of job group skill level/industrial sector, would be tested if and only if the P-value of the parent null-hypothesis test was ≤ 0.01.
We hold that statistical significance in principle only can be declared in blinded statistical analyses, i.e. in analyses where the hypotheses are completely defined before the researchers have looked at any relation between the concerned exposure and outcome data that are to be used to test them. As mentioned above, the present study is not completely blinded. Hence, P > 0.01 would be regarded as ”not statistically significant” while P ≤ 0.01 would be regarded as “tentatively statistically significant”, where “tentatively” means “subject to further confirmation; not definitely”.