Dose-escalation strategy and electrolyte-cardiac safety of amphotericin B colloidal dispersion: a preregistered emulated target trial with physician-preference instrumentation
Year: 2026
Type: study-registration
Abstract: This preregistration describes the analysis plan for a single-center retrospective cohort study using routinely collected electronic health records (Guangxi Zhuang Autonomous Region People's Hospital, China; index courses 2022-01-21 to 2026-10-15; administrative cutoff 2026-10-21; ethics approval LL2026LBR). In 598 adults receiving a first intravenous course of amphotericin B colloidal dispersion (ABCD), we emulate a target trial comparing rapid dose attainment (day-0 prescribed dose >=150 mg) versus gradual dose escalation, using leave-one-out physician and ward prescription preference as instruments. Primary objective: the causal dose-response of cumulative ABCD exposure (mg/kg) on the 14-day risk of severe hypokalemia (<3.0 mmol/L) or severe hypomagnesemia (<0.7 mmol/L), estimated with marginal structural models. Secondary objectives: (1) strategy-contrast cumulative incidence functions with discharge/death as competing events; (2) potassium/magnesium trajectory subphenotypes (latent class mixed models) and their association with AKI and 30-day mortality; (3) case-crossover analysis of potassium/magnesium nadirs and incident ECG repolarization abnormalities/arrhythmias (text-defined outcome; numeric QTc is unavailable in the exported reports); (4) potassium-replacement burden and refractory hypokalemia; (5) PK/PD simulation of prescribed doses against the Aspergillus ECOFF. Expected outcomes: a well-powered dose-response estimate (63 severe hypokalemia events among 279 baseline-normokalemic patients with post-index measurements) and a strategy contrast (478 rapid vs 120 gradual) with approximately 80% power for a 40% vs 25% 14-day risk difference. Physician and ward preference instruments show first-stage F statistics of 15.9 and 10.9. Negative controls (platelet trajectory, 24-h sodium change) and E-values will quantify residual confounding. This registration is completed before any outcome analysis. Deviations from the registered plan will be logged and reported in the manuscript.
Source: Open Science Framework
Author: Zhongwen Feng
Language: English
Subfield: Infectious Diseases
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7223702061