Incidence and risk factors of linezolid-induced hyperlactatemia in ICU patients: development and validation of a nomogram

Blood lactate is a well-established prognostic indicator in critically ill patients, and hyperlactatemia significantly increases intensive care unit (ICU) mortality. However, studies on hyperlactatemia specifically attributable to linezolid (LZD) remain limited. Our study compared the incidence of hyperlactatemia induced by systemic therapy with LZD and vancomycin (VAN) in ICU patients. Furthermore, this study was dedicated to identifying independent risk factors for hyperlactatemia induced by LZD and constructing a new clinical prediction model. This single-center retrospective cohort study applied propensity score matching (PSM) analyses to balance baseline characteristics. Adult ICU patients treated with LZD or VAN for gram-positive infection between April 2020 and April 2022 were eligible. The primary outcome was the incidence of hyperlactatemia. The secondary outcomes comprised the 28-day survival, length of hospital stay, ICU stay, and mechanical ventilation duration. Within the LZD cohort, multivariable logistic regression was performed to identify independent risk factors for hyperlactatemia. Then, a nomogram prediction model was constructed based on these factors and internally validated by bootstrap resampling. A total of 647 patients were enrolled, including 293 (45.3%) in the LZD group and 354 (54.7%) in the VAN group. Subsequently, we successfully matched 175 patients treated with LZD and VAN by PSM. Compared with the VAN cohort, patients receiving LZD therapy exhibited markedly elevated prevalence of hyperlactatemia (18.29% vs. 5.71%; p < 0.001). The two groups did not have significant differences in terms of the 28-day survival, length of hospital stay, ICU stay, and mechanical ventilation duration. The old age, high nutrition risk in critically ill (NUTRIC) score, low platelet count, elevated blood glucose, and prolonged LZD duration were independent risk factors for LZD-induced hyperlactatemia, which were subsequently incorporated into the construction of a nomogram prediction model. The model demonstrated a substantial discrimination ability, as evidenced by an area under the curve (AUC) of 0.926, exhibited excellent calibration performance, and provided notable net clinical benefit. The incidence of hyperlactatemia was significantly higher in patients treated with LZD than with VAN. Therefore, routine arterial lactate monitoring is warranted during LZD therapy. The nomogram model can provide an insightful and applicable tool for predicting LZD-induced hyperlactatemia.