The Statistical Monitoring by Adaptive RMSTD Tests: an efficient, informative, and customizable method for the complete internal quality control intended for low-frequent sampling of control measures

Two control mechanisms are relevant to perform an internal quality assurance: a permissible limit LSMC applied to single measures of control samples and a retrospective statistical analysis to detect increased imprecision and baseline drifts. A common statistical metric is the root mean square (total) deviation (RMSD/RMSTD). To focus on recent changes under low-frequent sampling conditions, the monitored amount of retrospective data is usually very small. Unfortunately, the calculated RMSTD of a small data set with n<50 samples has a significant statistical uncertainty that needs to be considered in adequate limit definitions. In particular, the minimum reasonable limit LRMSTD(n), applied to the RMSTD of a series of n samples, decreases from LSMC (e.g., 2.33*standard_deviation+bias) for n=1 towards Ltrue_RMSTD for n[-&gt;]{infty} (long-term statistics). Two mathematical approaches were derived to reliably estimate an optimal function to adjust LRMSTD(n) to small sample sizes. This knowledge led to the development of a new quality-control method: the Statistical Monitoring by Adaptive RMSTD Tests (SMART). SMART requires just one mandatory limit (either LSMC or Ltrue_RMSTD) per analyte. By definition of up to 7 possible alert levels, SMART can early recognize and evaluate both the significance of a single outlier and establishing critical trends or shifts in recent SMC data. SMART is intended to efficiently monitor and evaluate small amounts of control data.