Attentional bias to pain-related information: a meta-analysis of dot-probe studies