The estimated effect of unemployment on depression may be biased by time-varying, intermediate, and time-constant confounding. One of the few methods that can account for these sources of bias is the parametric g-formula, but until now this method has required that all relevant confounders be measured.
We combine the g-formula with methods to adjust for unmeasured time-constant confounding. We use this method to estimate how antidepressant purchasing is affected by a hypothetical intervention that provides employment to the unemployed. The analyses are based on an 11% random sample of the Finnish population who were aged 30-35 in 1995 (n = 49,753) and followed until 2012. We compare estimates that adjust for measured baseline confounders and time-varying socio-economic covariates (confounders and mediators) with estimates that also include individual-level fixed-effect intercepts.
In the empirical data, around 10% of person–years are unemployed. Setting these person–years to employed, the g-formula without individual intercepts found a 5% (95% CI: 2.5 to 7.4%) reduction in antidepressant purchasing at the population level. However, when also adjusting for individual intercepts, we find no association (-0.1%, 95% CI: -1.8% to 1.5%).
The results indicate that the relationship between unemployment and antidepressants is confounded by residual time-constant confounding (selection). However, restrictions on the effective sample when using individual intercepts can compromise the validity of the results. Overall our approach highlights the potential importance of adjusting for unobserved time-constant confounding in epidemiologic studies, as well as demonstrating one way that this can be done.
This is an open-access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.
a Max Planck Institute for Demographic Research, Konrad-Zuse str. 1, 18057 Rostock, Germany
b Stockholm University Demography Department, Universitetsvägen 10 B, Stockholm, Sweden
c Department of Social Research, University of Helsinki, Unioninkatu 37 (P.O. Box 54), 00014 Helsinki, Finland
d London School of Economics & Political Science, Houghton St, London WC2A 2AE, United Kingdom
e Centre for Health Equity Studies (CHESS), Stockholm University and Karolinska Institutet, SE-106 91 Stockholm Sweden
Sources of funding: M.J. Bijlsma gratefully acknowledges support from the Max Planck Society. B. Wilson is funded in part by Vetenskapsrådet (Swedish Research Council), Swedish Initiative for Research on Microdata in the Social and Medical Sciences (#340–2013–5164) and by FORTE, Ageing well programme (#2016-07115). M. Myrskylä is supported by the European Research Council Starting Grant (COSTPOST) [#336475], and by the Max Planck Society within the framework of the project “On the edge of societies: New vulnerable populations, emerging challenges for social policies and future demands for social innovation. The experience of the Baltic Sea States” (2016-2021). P. Martikainen is funded by the Academy of Finland, Strategic Research Council PROMEQ project (#303615) and the Signe and Ane Gyllenberg Foundation.
Computing code is available in eAppendices 2 and 3.
Conflicts of interest statement: None declared.
* Corresponding author. E-mail address: firstname.lastname@example.org, Max Planck Institute for Demographic Research, Konrad-Zuse-Straße 1, Rostock, 18057, Germany. Tel. +49 (0)381 2081 211; Fax: +49 (0)381 2081-439.