Precarious employment in Sweden : methodological development and health associations

Abstract

Precarious employment (PE), representing a multidimensional construct of (poor) employment conditions, has been found associated with several health outcomes. However, there is a lack of a recognized definition of PE and studies have varied greatly in their assessment of PE. In addition, previous studies have been limited in their design, the impact of labor market mobility among PEs has seldomly been explored, and a range of health outcomes are yet understudied. Further, in Sweden, response rates to national surveys have continuously decreased, calling for alternative sampling strategies. The aim of this thesis was hence two-fold: 1) to explore, develop and apply novel sampling- and exposure-assessment methods to study precarious employment in the Swedish labor market, and by its application, 2) study associations between precarious employment and health.

In Study I, the aim was to assess the performance of web-based respondent driven sampling among non-standard, precarious, employees in Stockholm County. The method was implemented between November 2016 and May 2017, and the resulting sample size (n=415), response rate (62%) and wave depth (1-15 waves) were considered acceptable. Equilibrium was reached for the variables assessed and estimations could be performed. In Study III, the aim was to operationalize PE with a typological and summative scale approach in register-data. PE was operationalized through five items and three dimensions in the Swedish working population of 2014, by the means of labor market data (LISA) (n=4,349,322). Through the typological approach, assessed by means of latent class analysis, one PE-type (“PER”) as well as five other employment types emerged. In the summative scale approach, scores ranged between -10 and +2 and >80% of individuals in the PER-type fell beneath the 25th percentile. The PER-type was generally associated with sociodemographic and occupational characteristics previously recognized as overrepresented in PE. In Study II, the aim was to study the associations between PE and self-reported health. Generalized linear models with Poisson distribution were applied for degree of PE, as assessed with the Swedish version of the Employment Precariousness Scale, and self-reported general health, mental health, and musculoskeletal pain. High PE was associated with poor general (a2PR 1.8, 95%CI: 1.2-2.6) and mental health (a2PR 1.7, 95%CI: 1.3-2.2). In Study IV, the aim was to estimate the risk of diagnosed common mental disorders, substance use disorders, and suicide attempt, as dependent on low-quality employment trajectories. Repeated measures latent class analysis was applied on LISA-data of the Swedish working population across 2005-2009 (n=2,743,764). The resulting six employment types, together with unemployment, were grouped in employment trajectories according to constant, fluctuating, and directional mobility patterns. Ten of the resulting 21 trajectories were considered as low quality. Diagnoses were retrieved from the national patient register between 2010-2017. Relative risk was estimated by cox proportional hazards regression models. Low-quality employment trajectories increased the risk of common mental disorders (a2HR range: 1.1-1.6), substance use disorders (a2HR range: 1.1- 2.2) and suicide attempt (a2HR range: 1.1-1.8). Gender differences emerged in the distribution of trajectories as well as in mental health risk.

In conclusion, this thesis provided alternative methods in recruiting, sampling and studying precarious employees in the Swedish labor market. Associations between PE, low-quality employment trajectories and poor health were found, particularly so in terms of mental health - results which emphasize the need to combat PE. Improving employment conditions is central, including stability of employment and protection from unemployment, decent wages, and union representation, i.e., decent employment conditions across a life-course perspective. The dynamic nature of the labor market constitutes a need for surveillance of the development of precarious conditions. Further knowledge gaps remain to be filled, including studies of physical health outcomes, as well as mental health outcomes currently understudied, such as substance use disorders and suicide. Moving forward, alignment of definition and assessment of PE is crucial to enhance cross-comparisons across studies and contexts. Longitudinal studies with life-course approaches, in which mechanisms and differentially affected sociodemographic groups are studied, would further serve to establish evidence on the negative health effects of PE, but also shed light on how employment precariousness and poor health could be alleviated through policy and labor market legislations.