The Evidence Clash Between Statins and Delirium: Should They Stay, Should They Go, or Should They Be Started?*
Statin use outside of the ICU is common; 25% of Americans adults older than 40 currently take one (5). Among very old adults (> 80 yr old), a group making up an ever-increasing proportion of ICU admissions and at increased risk for delirium (17, 18), the rate of statin use, at 33%, is even higher (6). As the evidence evaluating the association between statin use in critically ill adults and delirium occurrence increases, ICU clinicians are faced with two key questions: 1) Should a critically ill adult taking a statin prior to ICU admission be continued on this therapy? 2) Should a statin-naïve adult who becomes critically ill be initiated on a statin?
In this issue of Critical Care Medicine, Mather et al (19) publish a retrospective, cohort study that helps inform the first question. In their analysis, 1,475 medical ICU patients receiving statin therapy at the time of ICU admission and continued on it throughout hospitalization were compared with a propensity-matched control group of 1,475 medical ICU patients not administered statin therapy. The propensity score for receiving statin therapy was calculated from a logistic regression analysis that considered age, race, gender, and 24 different comorbidities based on their association with either delirium occurrence (e.g., dementia) or statin use (e.g., ischemic heart disease). Using a Cox proportional hazards model, the odds of delirium occurrence over the course of hospitalization was halved in those patients administered a statin (odds ratio [OR], 0.47; 95% CI, 0.38–0.56]. An important limitation of this analysis is that it failed to consider competing risks for delirium (e.g., coma) nor factors, that often change on a daily basis, with the potential to influence delirium occurrence (e.g., severity of illness, benzodiazepine exposure) (20, 21).
The conclusions by Mather et al (19) are consistent with two prior prospective cohort studies that have evaluated the association between statin use and delirium occurrence in critically ill medical and surgical adults (12, 13). Like the Mather analysis, each incorporated a propensity score analysis to reduce confounding related to nonrandom, patient-related factors that could potentiate statin use. However, unlike Mather, each evaluated the daily risk for delirium and included daily severity of illness as a covariate in the time-dependent, random-effects multivariable logistic regression models used.
The first study revealed that ICU statin use the previous evening was associated with a greater odds of being delirium-free (OR, 2.28; 95% CI, 101–5.13) and having a lower serum C-reactive protein (CRP) (β, –0.52; p < 0.01) the next day (12). A lack of association between statin use and being delirium-free the next day was observed after serum CRP values were controlled for in the model (OR, 1.56; 95% CI, 0.64–3.79; p= 0.32). This suggests that statin-mediated declines in inflammation accounted for the reduction in delirium observed.