Moderators of nocebo effects in controlled experiments: A multi-level meta-analysis

Abstract

Nocebo effects are a heterogenous phenomenon in which contextual cues trigger or exacerbate symptoms independently of active interventions. Suggestion, conditioning, and social observation are widely recognised as hallmark methods for inducing nocebo effects, but the extent to which nocebo effects are differentially influenced by suggestion type (e.g., direct or indirect suggestion) and mode of administration (e.g., verbal, textual, visual, etc.) across symptom domains remains unknown. We conducted a pre-registered meta-analysis (PROSPERO registration number CRD42023402097) to quantitatively synthesize available research on the factors that moderate effects in controlled nocebo experiments. Of 8469 search results, 105 experiments comprising 5017 participants and 391 effect sizes were analyzed. A multi-level meta-analysis revealed an overall moderate effect size for nocebo effects, g=0.50, [0.39, 0.62]. The magnitude of symptom expectancy effects was a significant moderator of nocebo effects. Verbal suggestion and social observation yielded moderate and comparable nocebo effects whereas technological devices, sham stimulation, and conditioning were independently associated with the induction of large nocebo effects. Greater specificity in the reporting of nocebo induction methods is required to elucidate the efficacy of different types of suggestions in inducing nocebo effects.