Search for content and authors

Application of confidence intervals to bioanalytical method validation

Piotr J. Rudzki 1Andrzej Leś 1,2

1. Pharmaceutical Research Institute (IF), Rydygiera 8, Warszawa 01-793, Poland
2. Uniwersytet Warszawski, Wydział Chemii, Warszawa, Poland


Pharmacokinetic studies, e.g. innovative drug pharmacokinetic evaluation or generic drug bioequivalence testing, can be requested by the regulatory authorities during the registration process of medicinal products. The drug determination in biological fluids is performed by means of the instrumental analysis, including high-performance liquid chromatography (HPLC). To assure the reliability of results a selected bioanalytical method should be validated according to the specific requirements [1]. A detailed drug stability in biological matrix testing is one of the important steps within the validation. The FDA Guidance describes a standard approach of direct comparison of analytical results for the studied and reference samples, but also suggests that other statistical approaches may be used.

The presented method of statistical evaluation of stability is based on the application of confidence intervals. The idea, based on the assumption of equal variances for the studied and reference samples, was first reported by Timm et al. in 1985 [2]. To allow application of confidence intervals - when this condition is not met - a new method is proposed. Prior to statistical analysis the outliers in measurement results are identified and discarded. After logarithmic transformation of original data, F-Snedecor test is applied to compare variances in both groups of samples. The 90% confidence interval is calculated either using pooled variance (in the case of equal variances) or individual variances of the studied and reference samples [3,4].

As an example of confidence intervals application, freeze and thaw stability of gemcitabine in rat plasma was presented. After the 1st freeze-thaw cycle, variances in the studied and reference samples were equal and the 90% C.I was 102-108%. After the 3rd freeze-thaw cycle, the condition of equal variances was not met and the 90% C.I was 96-108%. Both confidence intervals for stability were within acceptance limits of 85-115%.

The described method of stability testing could be considered superior to the standard approach because confidence intervals include uncertainty of measurement. Application of confidence intervals during bioanalytical method validation can improve the quality of data obtained during pharmacokinetic studies.


[1] Guidance for Industry: Bioanalytical Method Validation. U.S. Department of Health and Human Services. FDA. CDER. CVM. May 2001.

[2] U. Timm, M. Wall, D. Dell. J. Pharm. Sci. 74 (1985) 972-977.

[3] S.Bolton: Pharmaceutical Statistcs, Practical and Clinical Applications, Second Edition, Marcel Dekker, Inc., New York and Basel 1990.

[4] W. Klonecki: Statystyka dla inżynierów, PWN, Warszawa i Wrocław 1999.


Legal notice
  • Legal notice:

Related papers

Presentation: Poster at VI Multidyscyplinarna Konferencja Nauki o Leku, by Piotr J. Rudzki
See On-line Journal of VI Multidyscyplinarna Konferencja Nauki o Leku

Submitted: 2008-03-18 14:11
Revised:   2009-06-07 00:44