Optimizing therapeutics for hospitalized patients with impaired renal function

描述（由申请人提供）： 三分之一的住院患者在住院期间有潜在的肾功能损害或发生急性肾功能衰竭。这些患者中的大多数被处方了对肾功能敏感的药物，并且应该调整剂量以避免药物蓄积和毒性。三十多年的研究已经产生了各种技术来估计肾清除率和其他药代动力学参数，这些参数可以指导处方医生如何调整给药剂量或频率。然而，这些技术很少被医生或其他提供者可靠地应用，并且在肾功能不全的情况下不准确的给药已被确定为可预防的药物不良事件的主要原因之一。 为处方者提供自动指导的计算机化处方最终将提供预防给药错误所需的保障和实时帮助。然而，现有的实施方式可能不够复杂以提供准确的剂量推荐。例如，广泛使用的Cockcroft-Gault肾清除率估计值在体重和年龄的极端情况下以及肾功能的快速变化下是不准确的。所有这些警告都经常出现在住院患者中。目前还不清楚，如果采用更有针对性的个性化方法，有多少由“单一模型”方法产生的药物推荐会改变。我们假设肾脏剂量决策支持系统的3项主要变更将显著改变指南：1）使用针对患者特征定制的模型估计药物清除率2）通知药物清除率变更，这将改变现有药物治疗3）将药物特异性药代动力学模型应用于治疗范围较窄的剂量治疗。我们将通过模拟每个重新设计的组件在患者和处方数据的大型数据集中的性能来测试我们的假设。 与公共卫生的相关性：肾功能受损的住院患者需要特殊护理和监测，以避免药物蓄积和直接肾毒性。需要计算机化的护理系统来最佳地协助药物给药，并在健康状况的突然变化使先前的药物给药过时时协助医生保持警惕。

DESCRIPTION (provided by applicant): A third of patients admitted to the hospital have underlying renal impairment or develop acute renal failure during their hospital course. The majority of these patients are prescribed a drug that is sensitive to renal function, and should be dose adjusted to avoid drug accumulation and toxicity. Over three decades of research has produced a variety of techniques to estimate renal clearance and other pharmacokinetic parameters which can instruct prescribers on how to adjust doses or frequencies of administration. However, few of these techniques are reliably applied by physicians or other providers, and inaccurate dosing in the setting of renal insufficiency has been identified as one of the major causes of preventable adverse drug events. Computerized prescriptions with automated guidance for prescribers will ultimately provide the safeguards and real-time assistance that is needed to prevent dosing errors. However, the existing implementations may not be sophisticated enough to offer accurate dosing recommendations. For example, the widely used Cockcroft-Gault estimate for renal clearance is inaccurate at the extremes of weight and age and with rapid changes in kidney function. All of these caveats are frequently present in hospitalized patients. It is unknown how many of the drug recommendations generated by a 'single-model1 approach would be altered if a more tailored, individualized approach was applied. We hypothesize that 3 major changes to a renal dose decision support system will substantially alter the guidance: 1) estimate drug clearance with a model tailored to patient characteristics 2) notification of changes to drug clearance which should alter existing drug therapy 3) apply medication-specific pharmacokinetic models to dose therapies with a narrow therapeutic range. We will test our hypothesis by simulating the performance of each redesigned component in a large dataset of patient and prescription data. Relevance to public health: Hospitalized patients with impaired kidney function need special care and monitoring to avoid drug accumulation and direct kidney toxicity. Computerized systems of care are needed to optimally assist with drug dosing and to assist physicians with maintaining vigilance when sudden changes in health status make prior drug dosing obsolete.