Firefighters Burn Center, Clinical Pharmacist, Department of Pharmacy, Regional One Health, 877 Jefferson Avenue, Memphis, TN 38103, USA; Department of Clinical Pharmacy, College of Pharmacy, University of Tennessee Health Science Center, 881 Madison Ave, Memphis, TN 38163, USA.
Department of Plastic Surgery, Firefighters Regional Burn Center, Regional One Health, 877 Jefferson Avenue, Memphis, TN 38103, USA; Department of Plastic Surgery, College of Medicine, University of Tennessee Health Science Center, Memphis, TN 38103, USA.
Burns. 2019 Mar;45(2):423-432. doi: 10.1016/j.burns.2018.09.025. Epub 2018 Oct 16.
Vancomycin empirical dosing studies in thermally injured patients have netted low successful target attainment and most excluded renal dysfunction, limiting applicability. In a previous study, the authors performed a retrospective analysis of 124 patients' measured pharmacokinetic parameters to calculate optimal dose and interval for intermittent infusion regimens and find predictors of clearance and total daily dose. The objective of this study was to improve the accuracy of attaining goal therapeutic targets with initial vancomycin regimens in patients with thermal injury through retrospective modeling.
In this phase 2 study, variables collected and calculated regimens in phase 1 were utilized to try and create an improved empiric vancomycin dosing algorithm in patients with thermal injury. Logistic regression was utilized to determine best predictors of dosing vancomycin every 6 and 8h. The strongest models were built as individual algorithms and tested for accuracy of target attainment. Each algorithm produced a regimen for each patient that was then tested utilizing each patient's actual measured pharmacokinetic parameters.
Univariable logistic regression of 41 variables identified 27 and 23 to be predictive of dosing every 8 or 6h, respectively. The most predictive multivariable model for dosing every 8h consisted of creatinine clearance (CrCl)≥80ml/min, Acute Kidney Injury Network classification <1, and total body surface area burned≥10 percent. For dosing every 6h, CrCl≥80ml/min, age≤40years old, days since injury≤6, and serum creatinine (SCr)≤0.8 were most predictive. Based on the top 5 multivariable models for each dosing interval, 7 algorithms were built to produce recommended regimens. The highest performing algorithm resulted in trough concentrations of <10mg/L (23%), 10-20mg/L (65%), 15-20mg/L (26%), and >20mg/L (11%); area under the concentration curve (AUC)>400mghr/L (83%); and AUC >400mghr/L without having a trough >20mg/L (72%).
The algorithm that resulted in the highest target attainment without overdosing recommended 15mg/kg dosed every 24h for CrCl≥30, every 12h for CrCl 31-79, every 8h for patients with CrCl≥80ml/min, and every 6h only if the patient with a CrCl≥80ml/min is also≤40 years old and has a SCr≤0.8. Caution is warranted for groups underrepresented in this study, such as those with very low CrCl, a low BMI, or receiving renal replacement therapy. This algorithm should be validated in other centers for patients with thermal injuries.
对烧伤患者进行万古霉素经验性给药研究的结果显示，大多数患者的目标药物浓度达标率较低，且这些研究大多排除了肾功能障碍患者，这限制了其适用性。在之前的一项研究中，作者对 124 名患者的药代动力学参数进行了回顾性分析，以计算间歇输注方案的最佳剂量和间隔，并发现了清除率和总日剂量的预测因子。本研究的目的是通过回顾性建模来提高烧伤患者初始万古霉素治疗方案达到目标治疗效果的准确性。
在这项 2 期研究中，利用第 1 阶段收集和计算的方案，尝试为烧伤患者建立一种改进的经验性万古霉素给药算法。利用逻辑回归确定每 6 小时和 8 小时给药的最佳预测因子。将最强的模型构建为单独的算法，并测试其目标药物浓度达标率。为每个患者生成一个方案，然后利用每个患者的实际药代动力学参数对该方案进行测试。
对 41 个变量进行单变量逻辑回归分析，分别有 27 个和 23 个变量可预测每 8 小时或 6 小时给药。预测每 8 小时给药的最佳多变量模型包括肌酐清除率（CrCl）≥80ml/min、急性肾损伤网络分类<1 和总体表烧伤面积≥10%。预测每 6 小时给药的最佳模型为 CrCl≥80ml/min、年龄≤40 岁、伤后时间≤6 天和血清肌酐（SCr）≤0.8。基于每个给药间隔的前 5 个多变量模型，构建了 7 种算法以生成推荐方案。性能最高的算法可使谷浓度<10mg/L（23%）、10-20mg/L（65%）、15-20mg/L（26%）和>20mg/L（11%）；浓度-时间曲线下面积（AUC）>400mghr/L（83%）；且 AUC>400mghr/L 而谷浓度不超过 20mg/L（72%）。
该算法的目标药物浓度达标率最高，且不会过量用药，建议 CrCl≥30 时每 24 小时 15mg/kg 给药，CrCl 为 31-79 时每 12 小时给药，CrCl≥80ml/min 时每 8 小时给药，CrCl≥80ml/min 且年龄≤40 岁且 SCr≤0.8 时每 6 小时给药。对于本研究中代表性不足的人群，如 CrCl 非常低、BMI 较低或接受肾脏替代治疗的患者，应谨慎使用。该算法应在其他中心的烧伤患者中进行验证。
