Association of low-dose ketamine with hallucinations in critically ill patients: a target trial emulation.
Intensive care medicine – May 05, 2025
Source: PubMed
Summary
Low-dose ketamine, commonly used for pain relief and sedation in intensive care, increases hallucination risk by over 6 times in critically ill patients. Among 7,500+ patients studied, those receiving ketamine for analgesia experienced hallucinations much earlier and more frequently (26% vs 7%) than those who didn't. This important finding helps doctors better balance ketamine's benefits against its potential to trigger delirium and other side effects.
Abstract
Ketamine use is a potentially modifiable risk factor for hallucinations. We aimed to use target trial emulation to investigate the association between low-dose ketamine and development of hallucinations in critically ill patients in the intensive care unit (ICU). Retrospective study using data from a university affiliated ICU in Melbourne, Australia. Application of marginal structural models and parametric g-formulas to assess the impact of low-dose ketamine on the development of hallucinations. We studied 7514 patients from June 2016 to April 2021. Of these, 625 patients (8%) received low-dose ketamine, beginning at a median of 0 (0-1) days from ICU admission and at a mean daily dose of 0.11 (0.08-0.15) mg/kg/h. Low-dose ketamine treated patients had a higher rate of hallucinations within 30 days of ICU admission (26% vs. 7%; p < 0.001) and the first episode of hallucination occurred earlier than in unexposed patient (2 [1-3] vs. 3 [1-7] days from ICU admission; p < 0.001). After adjustment for baseline and time-dependent confounders, low-dose ketamine was associated with a higher risk of hallucinations within 30 days (OR, 6.46 [95% CI 5.17-8.07]; p < 0.001). These findings were confirmed with parametric g-formulas. In ICU patients, low-dose ketamine was strongly associated with an increased risk of hallucinations. However, these findings should be interpreted with caution due to the observational nature of the study and the risk of residual confounding.