• Michael J Parker, Mark A Lovich, Amy C Tsao, Hao Deng, Timothy Houle, and Robert A Peterfreund.
• From the *Division of Pulmonary, Critical Care and Sleep Medicine, Department of Medicine Beth Israel Deaconess Medical Center, Boston, Massachusetts; †Department of Anesthesiology and Pain Medicine, Steward-St. Elizabeth's Medical Center, Boston, Massachusetts; and ‡Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital, Boston, Massachusetts.
• Anesth. Analg. 2017 Apr 1; 124 (4): 1129-1134.

BackgroundLaboratory data suggest that newly initiated drug infusions reach steady-state delivery after a significant time lag. Depending on drug and carrier flow rates and the infusion system's common volume, lag times may exceed 20 or more minutes, especially in the neonatal/pediatric critical care environment. This study tested the hypothesis that a computer-executed algorithm controlling infusion pumps in a coordinated fashion predictably hastens the achievement of the intended steady-state drug delivery in a model of neonatal/pediatric drug infusion.MethodsWe constructed an in vitro model of neonatal/pediatric drug infusions through a pediatric 4-Fr central venous catheter at total system flows of 2 mL/h or 12 mL/h, representing a clinically relevant infusion range. Methylene blue served as the model infused drug for quantitative analysis. A novel algorithm, based on Taylor Dispersion Theory of fluid flow through tubes and executed by a computer, generated flow patterns that controlled and coordinated drug and carrier delivery by syringe pumps. We measured the time to achieve the intended steady-state drug delivery by conventional initiation of the drug infusion ("turning on the drug pump") and by algorithm-controlled infusion initiation.ResultsAt 2 mL/h total system flow, application of the algorithm reduced the time to achieve half of the intended drug delivery rate (T50) from 17 minutes [17, 18] to 3 minutes [3, 3] (median, interquartile range). At 12 mL/h total system flow, application of the algorithm reduced T50 from 6 minutes [6, 7] to 3 minutes [3, 3] The bootstrapped median difference is -14 (95% confidence interval [CI], -16 to -12, adjusted P=.00192) for 2 mL/h flow and -3 (95% CI, -4 to -3, adjusted P=.02061) for 12 mL/h flow. Compared with conventional initiation, the additional fluid required by the algorithm-directed infusion was 0.43 and 1.03 mL for the low- and high-infusion rates, respectively.ConclusionsThe output of infusion pumps can be predictably controlled and coordinated by a computer-executed algorithm in a model of neonatal/pediatric drug infusions. Application of an algorithm can reduce the time to achieve the intended rate of infused drug delivery with minimal incremental volume administration.

18 keywords...

hide…