Background

Several recent cohort studies have reported inverse linear associations of adiposity measures (e.g. body mass index, BMI) with risk of completed suicide (Magnusson et al. 2006; Mukamal et al. 2007; Bjerkeset et al. 2008; Batty et al. 2010; Mukamal et al. 2010). Stepwise associations are found across the range of BMI - low BMI is associated with increased suicide risk, and raised BMI is associated with reduced suicide risk. These associations are robust to controlling for a range of possible confounding factors, including mental illness, and are seen regardless of suicide method. In contrast, results from studies of non-fatal suicide attempts have been markedly less consistent (Jiang et al. 1999; Carpenter et al. 2000; Falkner et al. 2001; Eaton et al. 2005; Brunner et al. 2006; Dong et al. 2006; Crow et al. 2008; Osler et al. 2008; Dave et al. 2009; Batty et al. 2010). Likewise, studies report inconsistent findings for the association of obesity with depression (Lawlor et al. 2007; Atlantis et al. 2008; Bjerkeset et al. 2008; Luppino et al. 2010).

Body weight fluctuates over time, and distinct patterns of its change have been shown to be associated with particular psychiatric outcomes. For example, a prospective study of a sample of American children found associations of chronic obesity with defiant oppositional defiant disorder in boys and girls and with depressive disorders in boys (Mustillo et al. 2003). A recent birth cohort study finds associations of mood symptoms with change in body mass index over the course between age 15 to 53 years but the relationship varies by sex and age at onset of symptoms (Gaysina et al. 2010). However, there have been no studies investigating the association of the growth or adiposity trajectories with risk of suicidal behaviours.

One possible explanation for the conflicting findings of the association of adiposity with non-fatal suicide attempts is that the relationship is subject to confounding by other risk factors for suicide (e.g. smoking, alcohol use, physical and mental illness and socioeconomic position), which are associated with both body weight and suicide (Mukamal et al. 2010). One possible approach to examining the causality of the observed adiposity-suicide relationship is Mendelian randomisation (Batty et al. 2010), which uses genetic variations as non-confounded proxies for environmental exposures (Davey Smith et al. 2003; Lawlor et al. 2008). The recent discovery of common markers of obesity such as FTO (Frayling et al. 2007) and MC4R (Loos et al. 2008) makes it possible to investigate the association of adiposity with suicide using such an approach.

Aims

We aim to investigate whether self-harm/suicidal behaviour is associated with body weight or distinct growth/adiposity patterns within the ALSPAC cohort. We also aim to explore the potential of using the ALSPAC to investigate the causality of any observed relationship between adiposity and self-harm/suicidal behaviour using the Mendelian randomisation approach.