Document Server@UHasselt >
Education >
School for Information Technology >
Master theses >

Please use this identifier to cite or link to this item: http://hdl.handle.net/1942/19401

Title: Statistical models for stability studies
Authors: Uwimpuhwe, Germaine
Advisors: BRAEKERS, Roel
DIYA, Luwis
Issue Date: 2015
Publisher: tUL
Abstract: Stability studies are conducted at all phases of the drug development cycle, with the main objective of having a stable product on market. In this project we aimed at evaluating if the shelf life could be extended from 24 (current shelf life) to 36 months, quantifying pharmaceutical stability such as shelf life, release limit, degradation rate ( annually and at the end of both shelf lives) and consumer/producer risk. The assay data are longitudinal from 50 different batches, which were put in stability chamber within two storage conditions and monitored up to 24 months. Linear mixed model in frequentist(classical and quantile regression) and bayesian approach were fitted. From exploratory data analysis and model reduction, random intercept model with linear mean structure was selected. The highest degradation rate after one, two and three years were from API B at storage condition 25°C /60%RH. Given a shelf life of 24 (36) months, the release limit for API A were 96.94 (97.37) for batches mean , and 97.37 (97.37) for individual tablets. The similar analysis was done for API B. The estimated producer and consumer risk were around zero for a shelf life of 24 month and bit high (9%) for shelf life of 36 months at storage condition 30°C/75%RH. In summary the shelf life for the drug stored at condition 25°C/60%RH can be extended to 36 months which was not the case for condition 30°C/75%RH.
Notes: Master of Statistics-Biostatistics
URI: http://hdl.handle.net/1942/19401
Category: T2
Type: Theses and Dissertations
Appears in Collections: Master theses

Files in This Item:

Description SizeFormat
N/A1.84 MBAdobe PDF

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.