Previous studies using a modified Stroop test suggested that suicide attempters, in contrast to depressed patients with no suicidal history, display a particular attentional bias toward suicide-related cues. However, negative results have also been reported. In the present study, we collected new data and pooled them as part of a meta-analysis intended to shed further light on this question.

We conducted:

– a cross-sectional study comparing performance on the modified Stroop task for suicide-related, positively-valenced and negatively-valenced words in 33 suicide attempters and 46 patient controls with a history of mood disorders;

– a systematic review and a meta-analysis of studies comparing performance on the modified Stroop task among patients with vs. without a history of suicidal acts in mood disorders.

The cross-sectional study showed no significant difference in interference scores for any type of words between suicide attempters and patient controls. A meta-analysis of four studies, including 233 suicide attempters and 768 patient controls, showed a significant but small attentional bias toward suicide-related words (Hedges’g = 0.22; 95% CI [0.06 to 0.38]; Z = 2.73; P = 0.006), but not negatively-valenced words (Hedges’g = 0.06; 95% CI [−0.09 to 0.22]; Z = 0.77; P = 0.4) in suicide attempters compared to patient controls.

Positively-valenced words and healthy controls could not be assessed in the meta-analysis.

Our data support a selective information-processing bias among suicide attempters. Indirect evidence suggests that this effect would be state-related and may be a cognitive component of the suicidal crisis. However, we could not conclude about the clinical utility of this Stroop version at this stage.

The authors have not supplied their declaration of competing interest.