Development of Pediatric Morphine Clearance Prediction Model and Comparison of Different Prediction Methods

• Main Authors: Shih, Yi-Yun, 施懿芸
• Other Authors: Hu, Teh-Min
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/69671421813207667309

碩士 === 國防醫學院 === 藥學研究所 === 102 === Background: Before twenty centuries, there were few pediatric clinical trials, so doctors decided the pediatric dose by extrapolating from the adult dose based on literature or previous experiences. However, because of the difference of physiology and organ maturation between adults and pediatrics, it often caused medication errors like giving the wrong dose to pediatrics. In order to improve this situation, besides that U.S. Food and Drug Administration promoted and encouraged pharmaceutical companies to conduct pediatric clinical trials, there were many researchers going into the prediction of pediatric doses. The researchers tried to use factors which may change with the development, such as age, weight, body surface area, or organ function, to predict pediatric clearance more accurately and further to predict pediatric doses. Objectives: (1) Current published pediatric morphine population pharmacokinetic model only included patients up to 3 years old, and this study is focus on whole age of pediatric population. (2) Most of the current published pediatric morphine population pharmacokinetic model have age-related maturation model to predict pediatric morphine clearance, and this study attempts to find out the relationship between pediatric weight and morphine clearance and to develop pediatric morphine clearance prediction model by weight as the only variable. (3) Compare the prediction results with common published prediction models. Methods: The present study develops pediatric morphine clearance prediction models by retrospective analysis. Clinical morphine clearance reports were collected from PubMed database. After data processing and data analysis, we used EXCEL to random assigned individual pediatric morphine clearance data into two groups with the ratio of 3 to 2, the former is the training group and the latter is the testing group. We used simple linear regression with weight and clearance data of the training group to develop single-phase and two-phase,5 or 10 kg cut point, prediction model. After internal validation, we used the data of the testing group to validate externally. Besides, we compared prediction results of our model with common published morphine prediction methods with the testing group data. Results: A total of 161 individual pediatric morphine clearance data and 51 aggregate pediatric morphine clearance data were analyzed. The pediatric age range is from 0 to 17 years old, weight range is from 0.68 to 54 kg, and the clearance range is from 0.52 to 75.6 ml/min/kg. From the results of data analysis, we found that the relationship between the pediatric weight and morphine clearance can be described by an exponential function and there is a cut point around 5 or 10 kg. The parameters of the model we developed as following: the weight exponent of single-phase prediction model is 1.7,the weight exponents of two-phase prediction model under 5 or 10 kg are 1.08 and 1.42, and the weight exponents above 5 or 10 kg are 1.14 and 0.57. No matter in internal or external validation, the ratio of accurate prediction of single-phase and two-phase prediction model, 5 or 10 kg cut point, is about 70%, and there is no statistically significant difference. From the results of prediction methods comparison, we found that (1) the predicted values of Anderson model and Anand model are almost within 0.33 to 3 fold observed values, and the predicted values of neonates and infants of the allometric 3/4 power model are all without 5 fold observed values. (2) the ratios of the accurate prediction are no significant differences between the single-phase, two-phase of 5 or 10 kg cut point, and Anderson model, but are significantly higher than the allometric 3/4 power model. (3) there seems to be a cut point around 10 kg in Anderson model and Anand model which use age and weight as variables. Conclusions: There seems to be a cut point of 5 or 10 kg in the exponential relationship between pediatric weight and morphine clearance, and the exponent of under 5 or 10 kg should be larger than 1. The maturation model of Anand model appears to be useful as a factor to correct the substantial error introduced by the allometric 3/4 power model especially in neonates and infants. Keywords: allometry, pediatric morphine clearance, pediatric morphine clearance prediction model