Can uncertainty about your job make you mentally unwell enough to need a long sickness absence? A Swedish twin study found an association—but its strongest test raised a more complicated question.
When researchers compared twins who differed in employment uncertainty, the association became much smaller. The finding challenges simple causal headlines without proving that workplace conditions are harmless.
Quick answer
Does the twin study prove that insecure jobs cause mental-health sick leave?
No. Employment uncertainty was associated with later medically certified psychiatric sickness absence in the overall cohort, but the estimate was much smaller within exposure-discordant twin pairs. The study cannot firmly separate causal effects from shared familial and other factors.
Key takeaways
- The prospective Swedish STODS analysis included 9,326 employed twin individuals and register follow-up to 2018.
- Employment uncertainty concerned notice to quit, closure or reorganization—not every kind of job dissatisfaction.
- The outcome was medically certified sickness absence longer than 14 days for mood, anxiety or stress-related disorders.
- The overall adjusted odds ratio was about 1.46, versus 1.09 among 434 exposure-discordant twin pairs.
- Sensitivity analyses were mixed; neither causation nor a definitive genetic explanation was established.
When a job feels uncertain, what follows?
Restructuring, possible closure and the threat of dismissal can make ordinary working life feel unstable. It is tempting to draw a straight line from an insecure job to mental ill health and then to sick leave. Yet the people who encounter workplace uncertainty may also differ in health, family circumstances and earlier experiences. Observing a link does not tell us which of those explanations is correct.
A Swedish twin study published in Scientific Reports in September 2026 approached this question from an unusual angle. Instead of stopping at a comparison of workers with and without employment uncertainty, the researchers also compared twins who differed in their reported exposure. The contrast between those analyses is the heart of the investigation.
What exactly did the researchers study?
The researchers used the Swedish Twin project of Disability Pension and Sickness Absence, or STODS. Their main analysis included 9,326 twin individuals who were employed when they answered a survey in 2005 or 2006. Information from the Swedish Twin Registry was linked to national registers tracking medically certified sickness absence through the end of 2018. This was a prospective observational study, not an experiment assigning anyone to an insecure job.
Employment uncertainty was measured through three survey questions about insecurity associated with notice to quit, workplace closure and reorganization during the previous twelve months. A participant counted as exposed in the main analysis if they reported sometimes or often experiencing at least one of these. The measure describes specific work-related threats and events; it does not capture every possible feeling of job dissatisfaction or stress.
The outcome was not every episode of feeling unwell
The study examined register-recorded sickness absence spells lasting more than 14 days and certified for mood, anxiety or stress-related disorders. The relevant diagnostic categories were ICD-10 F30–F39 and F40–F48. This is narrower than asking whether somebody ever felt anxious, had trouble sleeping or took a few days off work. It also does not cover every reason for sickness absence.
People already on relevant sickness absence at baseline were excluded from the main analysis to improve the order of exposure and outcome. Nevertheless, symptoms before the survey might have influenced both experiences at work and later absence. A long follow-up is useful for studying subsequent events, but it does not by itself establish that the exposure caused the outcome.
The headline result: an association in the wider sample
In the overall cohort, people reporting employment uncertainty had higher odds of a subsequent qualifying sickness absence spell. The age- and sex-adjusted odds ratio was approximately 1.46 to 1.47, with a 95% confidence interval of roughly 1.30 to 1.65. The study abstract reports an adjusted odds ratio of 1.46 (1.30–1.64). That is evidence of an association in this population.
An odds ratio is not a percentage increase in an individual's probability of becoming ill. The paper also reports that 22% of the exposed group and 15% of the unexposed group had at least one qualifying absence during follow-up, but these are unadjusted group proportions. Neither number can predict what will happen to a particular employee.
What changed when twins were compared?
The investigators then focused on 434 same-sex twin pairs in which one twin reported employment uncertainty and the other did not. Such a comparison helps account for factors twins share, including aspects of family upbringing and, to differing degrees, inherited genetic variation. Among identical twins, inherited DNA is effectively shared; among fraternal twins, it is only partly shared.
In the combined discordant-pair analysis, the estimated odds ratio was 1.09 (95% confidence interval 0.75–1.58). This was much closer to no association than the overall estimate. It is compatible with the idea that shared familial factors contribute to the population-level relationship. But the confidence interval is wide, and the comparison cannot determine how much is attributable to genes, early environment or other explanations.
Why a weaker twin result is not the final answer
A smaller estimate in a co-twin comparison can be informative, but it is not proof that workplace uncertainty has no effect. Only pairs differing in the measured exposure contribute to that analysis, and there were relatively few such pairs. Random variation, measurement error and differences in experiences between twins can matter more when the sample is limited.
The authors also tested alternative ways to define employment uncertainty. Some sensitivity analyses supported attenuation in twin comparisons, whereas more restrictive definitions did not consistently show the same pattern. The researchers explicitly cautioned against firm conclusions about familial confounding. A scientifically careful summary must preserve that uncertainty instead of presenting the twin estimate as a definitive verdict.
Did identical and fraternal twins tell different stories?
Among 241 identical-twin pairs discordant for exposure, the reported odds ratio was 1.21 (95% confidence interval 0.73–2.01). Among 193 fraternal same-sex pairs, it was 0.96 (0.55–1.67). Both estimates were imprecise, with confidence intervals spanning values below and above one. They should not be interpreted as a reliable difference between genetic and environmental mechanisms.
Identical-twin comparisons are valuable because they hold inherited DNA and much of early family context constant. They cannot hold constant every later workplace experience, relationship, illness or reporting difference. Nor can a nonsignificant result establish that a real association is absent. The design strengthens causal scrutiny, but it does not turn an observational study into a randomized trial.
Personality was investigated, not used to label workers
The study considered neuroticism and extraversion as possible confounders and moderators. The researchers found no clear evidence that extraversion changed the relationship between employment uncertainty and qualifying sickness absence. Findings involving neuroticism were mixed across different measures of statistical interaction, and the overall conclusion did not support a clear personality moderation effect.
These results should not be turned into advice that particular personality types are destined to become ill after a workplace change. A personality questionnaire is not a diagnosis, and the study does not validate any individual prediction tool. Work organization, health history, social conditions and access to support are also relevant contexts that this analysis cannot fully resolve.
Strengths, blind spots and what remains unknown
Major strengths include a defined Swedish twin cohort, prospective register follow-up, medically certified absence records and a within-family comparison that ordinary workforce studies cannot perform. The authors also explored alternative exposure thresholds, a shorter three-year follow-up and other sensitivity analyses. These checks make the evidence more informative than a single cross-sectional correlation.
Important limits remain: employment uncertainty was self-reported, occupation and industry were not available, and short sickness absences of fourteen days or less were not captured. The study reflects Swedish workers and the Swedish insurance context. Residual confounding, reverse causation and uncertainty in the smaller twin analyses remain possible. It would be wrong to generalize its estimates to every country, workplace or employee.
What should readers take away?
The most defensible conclusion is two-part. In the broader Swedish sample, employment uncertainty was associated with later medically certified absence for mood, anxiety or stress-related disorders. In twins differing in exposure, the association was substantially weaker, raising the possibility that shared familial influences contribute. The study does not prove that job uncertainty causes psychiatric sickness absence, or that genetics explains away workplace conditions.
Employees experiencing distress deserve appropriate support regardless of what an observational study can establish about average causal effects. This article is research interpretation, not an individual medical assessment or legal guidance about employment. The broader TwinPare Research collection also examines screen use, depression and creativity through twin comparisons, but those studies address different exposures and outcomes.
Source notes
The source has been verified and editorially reviewed for this article. The limitations below show which level of conclusion the sources support.
- [wijkander-employment-uncertainty-sickness-absence-twins-2026] A twin study of employment uncertainty and sickness absence due to mental disorders accounting for familial factors and personality Maria Wijkander; Pia Svedberg; Jurgita Narusyte; Iman Alaie; Sandra Blomqvist; Petra Lindfors; Linda L Magnusson Hanson. Scientific Reports, 2026. Evidence type: Prospective observational Swedish STODS twin study of 9,326 employed individuals, register-based psychiatric sickness absence and 434 discordant twin pairs Limitation: Published 29 September 2026. Main cohort OR 1.46 (95% CI 1.30–1.64); within-pair OR 1.09 (0.75–1.58). The attenuated co-twin estimate does not establish causation or conclusively demonstrate familial confounding. PubMed DOI
Editorial source review
This section shows how the article's key factual claims are linked to the source.
Phrasings that require caution
- Odds ratios are not personal risk percentages.
- Do not equate reported employment uncertainty with a clinical psychiatric diagnosis.
- The outcome excludes sickness absences of fourteen days or less.
- Co-twin attenuation does not prove genetics is responsible or that workplace factors have no effect.
- No personalized mental-health or employment prediction is supported.
| ID | Claim | Source support | Caution |
|---|