This register-based cohort study examined sick leave trajectories among working-age adults (≤66 years) treated for transient ischemic attack (TIA) in Gothenburg, Sweden, from November 2014 to December 2019, and compared them with age- and sex-matched general population controls. The study quantified sick-leave prevalence at multiple time points and measured changes in sick-leave days during the year before and after the index event.
All patients aged ≤66 years diagnosed with TIA at the three stroke units of Sahlgrenska University Hospital were identified through the regional Väststroke quality register. Controls were sampled from the Total Population register (Statistics Sweden) and matched 4:1 by exact age and sex; each control was assigned the same index date as the corresponding TIA case. Additional data were drawn from the national Riksstroke register, the Longitudinal Integration Database for Health Insurance and Labour Market Studies (LISA), and the Swedish Social Insurance Agency. Data were pseudonymized prior to analysis.
Participants were classified as being on sick leave if they had an active benefit episode at any of five time points: 30 days before the index date, at the index date, 14 days, 2 months, or 1 year after the index date. All degrees of sick leave (25%–100%) were included. Diagnostic ICD-10 codes reported by certifying physicians were available for each sick-leave episode and grouped into prespecified diagnostic categories including TIA, stroke, mental health, cardiovascular, neurological, musculoskeletal, respiratory, cancer, other, and unknown.
Disability pension status was recorded before and during the first post-index year. Employment status was derived from LISA using the official November classification in the calendar year preceding the index date. Yearly disposable income was divided into tertiles for the combined study population, and education was categorized by highest completed level (<10, 10–12, >12 years).
The study included 347 individuals with TIA and 1,388 matched controls (median age 59 years, range 26–66; 40.6% women). The TIA cohort had a slightly higher proportion with post-secondary education (>12 years: 39.0% vs. 37.4%) and a higher proportion in the highest income tertile (39.3% vs. 32.4%). Non-employment was more common among individuals with TIA (20.2% vs. 17.0%). Disability pension prevalence before the index date was 14.7% in the TIA cohort and 11.2% in controls.
There was a moderate correlation between sick-leave days before and after TIA (Spearman ρ = 0.56, p < 0.001). Individuals with TIA had more sick-leave days during the year before the event than controls (mean 26 [SD 78] vs. 14 [SD 55], p < 0.001). Within-person, mean sick-leave days increased by 39 days in the year after TIA compared with the year before (95% CI 28–50 days, p < 0.001).
Prevalence of ongoing sick leave in the TIA cohort rose from 8.1% at 30 days before the index date to 18.4% at the index date, peaking at 14 days after the event (35.7%). By comparison, 3.9% of controls had sick leave at 30 days before and 3.4% at 14 days after their index dates. The TIA cohort prevalence declined to 23.1% at 2 months and 11.5% at 1 year, whereas controls remained near 3–4% at those time points.
At 14 days after TIA, the most common registered sick-leave diagnosis among those with TIA was TIA itself (53 of 124; 42.7%), followed by mental health diagnoses (16 of 124; 12.9%) and stroke (15 of 124; 12.1%). At 2 months, TIA remained the most frequent diagnosis (24 of 80; 30.0%), followed by mental health and stroke (each 15 of 80; 18.8%). At 1 year, only 4 of 40 (10.0%) individuals on sick leave had TIA as the recorded cause, while stroke accounted for 9 of 40 (22.5%) and mental health for 14 of 40 (35.0%). In the control cohort, mental health diagnoses were the most common cause of sick leave across time points.
In multivariable binary logistic regression excluding individuals with prior disability pension and cases with missing covariate data, higher disposable income and being employed were independently associated with higher odds of ongoing sick leave 14 days after TIA. Compared with the lowest income tertile, the middle tertile had an adjusted OR of 2.54 (95% CI 1.22–5.29) and the highest tertile an adjusted OR of 2.11 (95% CI 1.03–4.34). Employed individuals had an adjusted OR of 3.11 (95% CI 1.23–7.88) versus non-employed. Education and the other included variables were not significantly associated. Model diagnostics reported acceptable calibration and no problematic multicollinearity.
Although TIA is defined by rapid symptom resolution without infarction, this cohort demonstrates that a substantial proportion of working-age adults experience immediate work incapacity: over one-third were on sick leave 14 days after the event. While most individuals returned to work over subsequent months, about one in ten remained on sick leave at 1 year, indicating persistent functional or psychological difficulties for a subset. The prominence of mental health and stroke diagnoses among those still on sick leave at 1 year suggests varied underlying causes for prolonged absence. The authors note that increased sick-leave days before TIA may reflect pre-existing morbidity that is not captured by the TIA diagnosis alone and that these patterns point to potential unmet needs for rehabilitation and support in return-to-work processes.
This study used national and regional registers with pseudonymized data; data cannot be shared publicly because of Swedish regulations, but access requests can be submitted to the Sahlgrenska Academy and are subject to independent review. The authors report that all statistical analyses were performed using R and that ethical approval was granted. Specific limitations discussed in the source text were those inherent to register-based designs; further details beyond what is reported in the source were not provided in the article excerpt.