%0 Journal Article %A K. Rao Gurijala %A Andrew Barnett %T Statistical Method for Trending of Excursions in Clean Room Microbiological Monitoring Data %D 2021 %R 10.5731/pdajpst.2020.011791 %J PDA Journal of Pharmaceutical Science and Technology %P pdajpst.2020.011791 %X A statistically robust set of rules is proposed for trending excursions in environmental monitoring data. These rules were designed to minimize false alarms when the process is in control, but signal quickly when the process goes out of control. An adverse trend is an early warning that the system is drifting from normal operating conditions. Prompt action may prevent further deterioration and avoid costly out-of-specification events. Adverse trends are defined as an alert level excursion rate of greater than 2.5% and an action level excursion rate of greater than 0.15%. These definitions were derived from setting action levels at 99.85th percentile and alert levels at 97.5th percentile. These percentiles were chosen because they are functional equivalents of control limits and warning limits used in statistical process control charting, which are set at three and two standard deviations above the mean, respectively. In addition, the USP recommended microbial recovery rates should also be implemented as trend metrics for microbial environmental monitoring of aseptic processing facilities. Occasional isolated alert level excursions may occur even if the process remains in a state of control. However, repeated alert level excursions occurring at a rate greater than 2.5% indicate the process is changing and the system is drifting from normal operating conditions. An adverse trend of alert level excursions should be investigated for root cause. It is critical to determine if an alert level excursion, at its onset, triggers an adverse trend. A total of 24 rules at various sample sizes were tested for their ability to detect an alert level excursion adverse trend at the onset of an excursion using data obtained over a period of one year. Rationale for choosing these rules is described. %U https://journal.pda.org/content/pdajpst/early/2021/02/19/pdajpst.2020.011791.full.pdf