Smoking as a confounder of the association of suicidality with serum lipid levels

We thank Berlin and colleagues for their comments on our study.1 The letter focused on smoking as a confounder of the association between suicidality and serum lipid levels. They commented that controlling for smoking would reduce the difference of serum lipid levels between suicidal and nonsuicidal patients or controls. This topic is very important, and we welcome this discussion.

We agree that smoking, as a demographic characteristic, plays an important role in suicidality. However, for this to have been accounted for in the findings, smoking must have been related to serum lipid levels and must have been associated with risk of suicidality. Regarding the former, nobody would disagree with the statement that smoking is associated with serum lipid levels; however, with respect to the latter, to be considered a confounder the association of smoking with suicidality must be independent from its association with serum lipid level. We agree that smoking may be related to suicidality, but as far as we have been able to find out, there is no evidence that smoking per se is an independent causal risk factor that is not linked to the change of serum lipid levels.

In addition, the associations among smoking, mental illness and suicide remain unknown. Some studies have found that smoking was not significantly associated with suicidal behaviour after adjusting for mental disorders.2,3 Based on these findings, the inference that smoking may be a confounder of the association between suicidality and serum lipid levels lacks sufficient evidence.

Our meta-analysis was based on a wide range of published articles (65 epidemiological studies), and the inverse associations between serum lipid levels and suicidality appeared to be consistent across most studies. Thus, the results of our study are sound and reliable.