Effect on suicide rate of having reduced unemployment is uncertain
BMJ 1999; 318 doi: https://doi.org/10.1136/bmj.318.7188.941 (Published 03 April 1999) Cite this as: BMJ 1999;318:941
- Ilkka Henrik Mäkinen, Researcher (Ilkka.Makinen{at}sh.se)
- Centre on Health of Societies in Transition, University College of South Stockholm, Box 4101, 141 04 Huddinge, Sweden
EDITOR—Lewis and Sloggett recommend policies to reduce unemployment in order to prevent suicide.1 They show that those unemployed in 1981 were overrepresented among suicides 2-11 years later. This seems to imply that the suicides are caused by the unemployment. Their exclusion of cases up to two years after registration to “reduce selection effects” shows that they think that the consequences are long term. Selection problems can hardly be avoided or reduced here by exclusion of the cases nearest in time to registration; it just blurs the hypothesis by excluding the most obvious examples (becoming unemployed as a suicidal crisis).
Causal hypotheses may be supported by studies at the collective level since these avoid confounders at the individual level2—for example, changes in the Swedish unemployment rate and male suicide rate correlated significantly over 1920-41.3 The effect appeared in the first year of periods of mass unemployment.
As unemployment re-emerged dramatically in Sweden in 1990, rising suicide rates were expected, but they didn't occur. These “natural experiments” show that the relation between unemployment and suicide is far from clear. The only consistent finding in Sainsbury et al's study4 and my replication of it5 was that the unemployment rate correlated significantly negatively with that of male suicide rates in both 1961-3 and 1977-9 in 18 European countries.
Recent British trends in unemployment and suicide have seemed to coincide: the male rates, especially in the younger age groups (15-24, 25-34, 35-44), correlate significantly with national unemployment rates (r=3D0.46-0.83, all P<0.025). When annual changes in unemployment and suicide rates are cross correlated, however, the relations disappear completely even if we allow for lagged effects (figure). There is no indication of unemployment “causing” suicide so that increases in its level would be followed by later increases in suicide rates.
Annual changes in male suicide rate and unemployment rate, 1970-94. Adjustment compensates for change in unemployment registration in 1982 (Denman J, McDonald P. Unemployment statistics from 1881 to the present day. Labour Market Trends 1996 Jan:5-18). r=correlation between change in unemployment rate and change in male suicide rate in same year; r(lag3)=correlation between change in unemployment rate in one year and change in male suicide rate three years later
The risks in recommending reduced unemployment as a remedy for suicide are twofold: inefficiency (lacking the causal link, one might invest large resources without reducing suicide rates) and lack of focus (by paying attention to phenomena only marginally connected with the problem, we might lose the focus from more credible forms of prevention).The effects of having more jobs are uncertain, but making unemployment financially easier to bear and reducing extreme situations could be justified even from the point of view of preventing suicide.