Detecting failed WBC-reduction processes: a quality assurance program introduced in a blood center

Pragmatic yet statistically valid quality assurance (QA) programs are necessary so that blood centers can select, validate, and monitor their WBC-reduction processes. A QA system for WBC-reduction processes based on the practical application of statistical theory within a large blood center was developed. The system identifies parameters for procedure and component evaluation and provides sample size and formatting suggestions.

Analyses of both procedure and component performance were undertaken during the purchase, validation, and control of filtration and apheresis WBC-reduction processes at Blood Centers of the Pacific from 1997 through 1999. QA analysis was categorized on the basis of whether the process was new to the organization or was a modification of a previously validated system. The numbers of samples necessary to consistently detect failure in platelet yield, unit volume, pH, and WBC count was statistically determined by parametric and nonparametric techniques.

Parametric analysis (power analysis) of the mean +/- SD of smaller numbers of samples was highly sensitive to shifted distributions, but only if the shift was normally distributed. Nonparametric analysis, necessary when the nature of the underlying distribution is unknown, suggested a minimal sample of 40 was required to achieve high confidence that significant bimodal failure (a secondary population with WBCs 5% above the cutoff) would be detected.

A QA system, developed for the evaluation of new or revised WBC-reduction processes, was based on statistical analysis of normally and non-normally distributed process failure. The number of samples was determined that allowed the achievement of confidence and tolerance levels considered appropriate within the blood center. Suggestions for outlier evaluation and a format for performance documentation have also been developed. To better define blood center quality goals, further research is necessary on donor and component biologic variability and the most significant modes of WBC-reduction process failure.