The retrospective cohort study by Ahrens et al. examined the dose-dependent relationship between intra-operative dexmedetomidine and postoperative delirium after non-cardiac surgery 1. The authors should be commended for assembling a large cohort, employing a careful adjustment strategy and using multiple non-linear modelling approaches. We wish, however, to highlight several methodological considerations that temper the conclusion that an optimal dose lies between 0.25–0.35 μg.kg-1. This degree of dosing precision is difficult to justify from a single-centre observational study in which dexmedetomidine was administered to only 4.2% of patients, indicating strong treatment selection. The low-dose group, from which the putative benefit is inferred, contained only 39 delirium events. Furthermore, groups differed substantially in: surgical duration; work relative value units; fluid administration; opioid and propofol doses; regional anaesthesia use; and neuromuscular blockade reversal strategy. These imbalances likely reflect not only dexmedetomidine dose but also case-mix, clinician preference and broader anaesthetic strategy. This is a concern amplified by the 16-year study period during which peri-operative practice evolved considerably. Multivariable adjustment, propensity methods and the reported E-value of 2.7 mitigate but do not eliminate residual confounding by indication 2. The exposure construct warrants scrutiny. The primary exposure was cumulative dose in μg.kg-1, dichotomised at the cohort median, and the proposed optimal range was derived from flexible models fitted to the same dataset. Cumulative dose is inherently entangled with procedure duration, timing of administration and delivery mode (bolus, infusion or both). The same total dose given rapidly is not pharmacodynamically equivalent to the same dose infused over several hours. The narrow 0.10 μg.kg-1 window is therefore likely model-dependent and should be regarded as hypothesis-generating rather than prescriptive. Delirium was ascertained from discharge documentation, keyword searches, diagnostic codes and Confusion Assessment Method scores in patients in ICU. Although pragmatic, this composite strategy may under-detect hypoactive delirium and identify cases preferentially in patients monitored more intensively or those with longer admissions 3, 4. The study provides valuable evidence that the association between dexmedetomidine and postoperative delirium may be non-linear. However, a modelled nadir in retrospective data should not be equated with a clinically validated best dose. The findings support a testable dose–response hypothesis that requires prospective confirmation before informing dosing recommendations.
Guo et al. (Thu,) studied this question.