Control Charts for Quality Control of Medicines (164 Tablet Desintegration Times)

Abstract

A consistent quality of a process or product, like the manufacturing of tablets is, traditionally, tested by 1 sample chi-square tests of their weights, diameters, desintegration times. E.g., tablets may only be approved, if the standard deviation of their diameters is less than 0.7 mm. E.g., a 50 tablet sample with a standard deviation of 0.9 mm is significantly different from 0.7 (^ = symbol of power term).

$$ \begin{array}{l}\mathrm{Chi}-\mathrm{square}=\left(50-1\right)0.9\hat{\mkern6mu} 2/0.7\hat{\mkern6mu} 2=81\hfill \\ {}50-1\;\mathrm{degrees}\;\mathrm{of}\;\mathrm{freedom}\hfill \\ {}\mathrm{p}<0.01\left(\mathrm{one}\;\mathrm{sided}\right)\hfill \end{array} $$

The example is from the Chap. 44, entitled “Clinical data where variability is more important than averages”, pp 487–497, in: Statistics applied to clinical studies 5th edition, Springer Heidelberg Germany, 2012, from the same authors). Nowadays, we live in an era of machine learning, and ongoing quality control, instead of testing now and then, has become more easy, and, in addition, provides information of process variations over time and process performance. Control charts available in SPSS and other data mining software is helpful to that aim.