Prediction of precipitation-induced phlebitis: a statistical validation of an in vitro model.

Abstract

To avoid phlebitis, new intravenous (IV) parenterals are often screened by injection into animals. This method is not only expensive and time consuming, it is also detrimental to the animals. An alternate method, focusing on precipitation as the cause, uses an in vitro dynamic injection model that requires less money and time and reduces the need for live models. Validation of the dynamic injection apparatus, for predicting mechanical phlebitis, is established. Twenty-one currently marketed IV products were injected into isotonic Sorenson's phosphate buffer flowing at 5 mL/min. The resulting opacities, produced by precipitation, are measured in an ultraviolet flow cell. These opacity data, coupled with literature reports on phlebitis occurrence, were used to generate a logistic regression that indicates the probability of phlebitis given an opacity value measured by the apparatus. Regression results are supported by a receiver operator characteristic curve that establishes the most ideal cut-off opacity value. This opacity value provides the highest combined sensitivity (statistical power) and specificity while minimizing false-positive and false-negative results. Both analyses show that an opacity value of 0.003 best delineates phlebitic and nonphlebitic products. Measures of sensitivity (0.83), specificity (0.93), positive predictive value (0.93), and negative predictive value (0.78) indicate the model's predictive accuracy and reliability. These results support the use of the dynamic model in place of animals for preliminary phlebitis testing of new IV injectables.

• Authors: Jennifer L H Johnson, Yan He, Samuel H Yalkowsky
• Journal: Journal of pharmaceutical sciences (J Pharm Sci) Vol. 92 Issue 8 Pg. 1574-81 (Aug 2003) ISSN: 0022-3549 [Print] United States
• PMID: 12884244 (Publication Type: Journal Article, Validation Study)
• Copyright: Copyright 2003 Wiley-Liss, Inc.

Topics

• Chemical Precipitation
• Drug Evaluation, Preclinical (methods, statistics & numerical data)
• In Vitro Techniques
• Infusions, Parenteral (adverse effects)
• Models, Statistical
• Phlebitis (chemically induced)
• Predictive Value of Tests