Predictive value of the kinetics of procalcitonin and C-reactive protein for early clinical stability in patients with bloodstream infections due to Gram-negative bacteria.

Diagnostic microbiology and infectious disease(2018)

引用 9|浏览23
暂无评分
摘要
OBJECTIVE:To investigate whether the magnitude of the change in procalcitonin (PCT) and C-reactive protein (CRP) levels between day 1 and day 2 after the blood culture date is associated with early clinical stability (ECS) on day 3 in patients with bacteremia due to Gram-negative bacteria (GNB). MATERIALS/METHODS:A prospective cohort study carried out in a 950-bed tertiary hospital in Spain between March 2013 and May 2014. Patients with GNB bacteremia were included. Changes in PCT and CRP kinetics from day 1 to day 2 (∆%PCT, ∆%CRP) were expressed as percentage of decline in blood levels. Logistic regression was used to identify predictors of ECS. Classification and regression tree analysis was performed to identify breakpoints. The discriminatory power of ∆%CRP and ∆%PCT as predictors of ECS was assessed by the area under the ROC (AUROC). RESULTS:71 patients were included, and 53 (74.56%) reached ECS. Multivariate analyses showed that SOFA score on day 1, ∆%PCT, and ∆%CRP were associated with ECS after controlling for confounders. ∆%PCT ≥ 30% (decline) and ∆%CRP ≥ 10% (decline) predicted ECS only among patients with SOFA≤3 on day 1 (n = 54; 43 reached ECS). In these patients, the AUROCs for the prediction of ECS were 0.96 (95% CI: 0.90-1) for ∆%CRP and 0.96 (95% CI: 0.90-1) for ∆%PCT, respectively. CONCLUSIONS:In the subgroup of patients with a SOFA score on day 1 ≤3, a ≥30% decline in PCT or a ≥10% decline in CRP between day 1 and day 2 was a very good predictor of ECS (which in turn was associated with a lower 30-day mortality and a greater clinical cure on day 14). Patients who do not achieve this decrease may need more intensive workup. In this subgroup (with a SOFA on day 1 ≤3), CRP may be preferred due to its lower cost.
更多
查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要