Multivariate statistical process control of a pharmaceutical tablet manufacturing process using near-infrared spectroscopy

O'Neil, A.J., Hailey, P.A., Jee, R.D. and Moffat, A.C. (2001) Multivariate statistical process control of a pharmaceutical tablet manufacturing process using near-infrared spectroscopy. AAPS Pharmaceutical Sciences, 3 (Supplement 463).

Full text not available from this repository.


Purpose. To determine the ability of near-infrared spectroscopy for multivariate statistical control of a pharmaceutical tablet manufacturing process. Methods. Near-infrared (NIR) spectroscopic measurements were made of batches of pharmaceutical blends and their tablets produced over a five year manufacturing period (n = 205 production batches). These measurements included diffuse reflectance measurements of the blend samples and reflectance and transmission measurements of the tablets. These measurements were subjected to multivariate analysis and statistical process control techniques. These were principal components analysis (PCA) and partial least squares regression (PLSR). PCA models of each process stage (blending and tabletting) and of the whole process (combined NIR data) were produced from just the NIR measurements. PLSR models of each process stage and the whole process required regression of NIR measurements with reference analytical laboratory data (e.g. drug content, tablet hardness, tablet thickness, friability, content uniformity, moisture content, ultra-violet molar absorption coefficient). In each case, each spectrum was represented as a single point in reduced dimensional space and optimisation procedures were employed to determine the region of the reduced dimensional space encompassing highest quality batches. These batches were enclosed within Hotelling’s T2 control ellipsoids (95% and 99% confidence). Results. Both multivariate approaches were able to detect highest quality batches at each process stage and for models of the entire process and detect trends in the process performance over time. Of those batches found to have Hotelling’s T2outside the 99% control ellipsoids (n = 13 batches), these had analytical results within the limits of the specification but with some values close to the limits. Conclusions. NIR spectroscopy was found to be an appropriate analytical technique for the statistical process control of this tablet manufacturing process and for identification of trends in process operating performance and prediction of final tablet quality from NIR measurements of blends.

Item Type:Article
Departments, units and centres:Department of Pharmaceutical and Biological Chemistry > Centre for Cancer Medicine
ID Code:282
Journal or Publication Title:AAPS Pharmaceutical Sciences
Deposited By:Library Staff
Deposited On:19 Oct 2006
Last Modified:04 Feb 2011 16:09

Repository Staff Only: Item control page

School of Pharmacy Staff Only: Edit a copy to replace this item