The length and frequency of previous sickness absence (SA) spells have been shown to be associated with future SA. The aim was to examine if this pattern persisted during the COVID-19 pandemic among workers in retail and sales.

We used pseudonymised, individual-level data from three nationwide Swedish administrative registers to conduct a prospective cohort study.

Sweden.

All 306 933 blue-collar workers in retail and wholesale, aged 18–67 in Sweden in 2019.

Likelihood and length of SA.

We used a Negative Binomial Hurdle model to estimate incidence rate ratios (IRR) and odds ratios (ORs) to determine if SA patterns differed in 2020–2021 compared with 2018–2019. We examined how these patterns varied according to the length and frequency of SA in the preceding year. Only SA spells >14 days were included.

54 993 (18.5%) workers had SA during 2020–2021, an increase from 46 024 (15.6%) in 2018–2019. We observed a dose-response association between the number of prior SA days and the likelihood and length of future SA days, both before and during the pandemic. The likelihood of subsequent SA was higher in 2020–2021 compared with 2018–2019 among individuals with up to 180 prior SA days. Individuals with no prior SA had a lower average number of subsequent SA days during the pandemic (IRR (95% CI) 0.96 (0.94–0.98)) than in 2018–2019, while those with 1–30, 31–90 or 181–365 prior SA days had a higher average number of SA days during 2020–2021.

Individuals with many SA days prior to the pandemic were at particularly high risk of lengthy SA during the pandemic years.