Impressive. Timely. These two words capture the essence of a study by Lundin et al published in this issue (see page 22).1 The authors examine a cohort of Swedish men born in 1949–51 and the effects of unemployment (during the 1992–4 recession) on subsequent mortality (1995–2003). The data assembled for the analysis are impressive. The findings are timely in light of the worldwide surges in unemployment.

As can only happen in Scandinavia, the authors have assembled an extraordinarily comprehensive dataset. The study is based on a cohort of 49 321 males tested for compulsory military subscription in 1969–70, restricted to those 37 798 males alive in 1995 who had been in full-time employment in 1990–1 with complete data on all covariates. Figure 1 attempts to convey the richness and temporal ordering of confounding, exposure and outcome data.

Bartley and Ferrie comment more thoroughly on this impressive study in an accompanying commentary (see page 5).2 They particularly draw attention to the select nature of this Swedish …