Comparative diagnostic accuracy of panic provocation tests: A meta-analysis

Panic provocation tests have been widely used to investigate the biological underpinnings of panic disorder (PD). However, their diagnostic specificity remains uncertain, as several agents can also induce panic-like reactions in healthy individuals. This meta-analysis estimated and compared diagnostic odds ratios (DORs) across different provocation methods. Random-effects meta-analyses were conducted using the Hartung-Knapp adjustment. Based on 123 studies including 5693 participants, sodium lactate infusion showed the highest pooled DOR (24.77, 95% CI 19.03-32.25, GRADE certainty = High), followed by isoproterenol (18.85, 95% CI 12.39-28.66, GRADE certainty = Moderate), cholecystokinin tetrapeptide (CCK-4; 15.03, 95% CI 7.83-28.84, GRADE certainty = Moderate), meta-chlorophenyl piperazine (m-CPP; 14.51, 95% CI 4.67-45.07, GRADE certainty = Low), caffeine (14.11, 95% CI 4.79-41.54, GRADE certainty = Moderate), hyperventilation (11.89, 95% CI 6.51-21.73, GRADE certainty = High), carbon dioxide inhalation (CO₂; 11.12, 95% CI 8.63-14.32, GRADE certainty = High), and yohimbine (7.87, 95% CI 1.12-55.13, GRADE certainty = Very low). Breath holding, fenfluramine and flumazenil did not demonstrate statistically significant discriminative ability. Meta-regression analysis confirmed that sodium lactate yielded significantly higher DORs than m-CPP, hyperventilation, CO₂, fenfluramine, flumazenil, and yohimbine. No substantial between-study heterogeneity or publication bias was identified across the pooled analyses. These findings suggest that provocation methods targeting acid-base balance, especially sodium lactate infusion, are associated with superior diagnostic discrimination between individuals with PD and healthy controls.