Does Exposure Timing Matter? Glycemic Indices and External Ventricular Drain–Associated Infection

<title>Abstract</title> <p> <bold>Objective:</bold> To examine whether glycemic exposure is associated with external ventricular drain (EVD)-associated infection using analyses that distinguish exposures preceding infection from those following it. <bold>Design:</bold> Retrospective cohort study. <bold>Setting:</bold> Single tertiary academic hospital, January 2018 to July 2023. <bold>Patients:</bold> One hundred seven consecutive patients undergoing EVD insertion for intracerebral or subarachnoid hemorrhage. <bold>Interventions:</bold> None (observational study). <bold>Measurements and Main Results</bold> : Infection was defined by 2015 CDC/NHSN criteria. Seven hyperglycemic indices were evaluated with conventional and equivalence analyses. Hypoglycemia (&lt;70 mg/dL) was examined over three windows—whole admission, before infection diagnosis, and the first 72 hr—and as a time-varying covariate. Infection occurred in 21 patients (19.6%). No hyperglycemic index was associated with infection (odds ratios, 0.68–1.05 per 1 SD), and equivalence could not be established. Whole-admission hypoglycemia was more frequent in infected patients (47.6% vs 24.4%; OR, 2.74 [95% CI, 0.90–8.30]), but the association attenuated over the first 72 hr (OR, 1.14 [95% CI, 0.11–6.68]); the difference between exposure windows was not significant (ratio of odds ratios, 0.43; p = 0.285). Infected patients underwent approximately twice as many glucose measurements (median, 84 vs 40; p = 0.009). Hypoglycemia was associated with in-hospital mortality (adjusted OR, 2.68; 95% CI, 1.10–6.54). Among nine patients with hypoglycemia within 72 hr, before any infection, eight (88.9%) died versus 35.8% of the rest (p = 0.003). <bold>Conclusions:</bold> No association was observed between glycemic indices and EVD-associated infection after accounting for exposure timing and differential glucose measurement, although the small number of infection events limits the range of effects that can be excluded. Hypoglycemia was more closely associated with mortality than infection, suggesting that it reflects illness severity rather than a cause of EVD-associated infection. </p>