Maintaining a Validated State – PV, PM and Statistics associated with Current Regulation 2014

Maintaining a Validated State PV, PM and Statistics associated with Current Regulation 2014 is a webinar that covers topics such as:

• Appropriate Application of Simple Statistical Tools, the Scientific Methods, facts, theories, proposals, functional requirements (FRS), acceptance criteria, formation of a hypothesis that is logical with sound scientific rationale by being scientifically based making it defendable
• Continued monitoring and/or sampling at the levels established during the development and qualification stages until sufficient data is available to generate statistically significant variability estimates. Once the variability is known, sampling and/or monitoring should be adjusted to a statistically significant level. Variation is to be used to detect the potential for defect complaints, OOS including OOT and OOL results, including deviation reports, process yield variations, BPR deficiencies, incoming raw material variances, adverse events and many other issues that may be found to enhance a validated (cost effective with minimal patient risk) means of operation. Hence change control becomes a critical component using SSR (sound scientific rationale) to manage an on-going validated state
• The arithmetic, together with certain numeric tables, yields the material on which to base the inference and measure the level of uncertainty associated with the process variables. The arithmetic is often routine, requiring no special mathematical training for the user, however the choice of the statistical model requires significant comprehension of the process in order to select the appropriate statistical model, then make sure the sampling fits the model so the arithmetic can be done to determine how the ANOVA results can be interpreted to demonstrate a maintained validated state
• Objective Evaluation - To most scientists, statistics is logic or common sense with a strong admixture of arithmetic procedures. The logic supplies the method by which data are to be collected and determined how extensive they are to be. Critical thinking as opposed to assumed awareness creates significant issues during the evaluation phase of any process
• Examples of Objective Data: Yield, pressure, pH, flow rate, time and lapse time, amounts, number of defects, length, duration, and many, many more depending on the validated equipment and associated process
• Process observations and data retrieval are the raw materials with which quality and statistical workers deal which means the results need to be in the form of numbers (objective evidence) as result of a measurement, not subjective results from check boxes. Tracking the pass/fail results is only an "appetizer" to the next phase of the assessment of the validated process including equipment and operator performance. Numbers are what we us to the constitute and address the situation to determine if the data is a variability or variation
• ANOVA - Analysis of Variance
• Our Responsibility: We are to manage the collection based on the statistical model. Presentation, characterization and, summarization of data including potential elements such as the results from the regression analysis, z-score, randomized block and a multitude of other means of assessing the consistency and identified variables of the validated process
• Experimental Design - By now we should begin to understand the need to be able know how we want to analyze the data before we begin to collect individual observations
• Not used to test hypotheses about variances used to test hypotheses about means (bar x and double bar x). Lifetimes of two different types of light bulbs. Effectiveness of two different toothpastes. Now we calculate the sum of squares and variances to measure variation between and within groups. Why? - Because we are trying to test a hypothesis about the equality of the data’s average or data’s mean. Sum of squares – sum of the variance (an individual value minus the average squared). Null Hypothesis – we test to demonstrate that the difference between the means is null (not 0) or insignificant. Linear Regression and Correlation. Used to Predict unknown values within a bracket (minimum and maximum operating ranges qutside a bracket – future (be careful). Line of Best Fit (Regression Line), the slope and the Y intercept define the regression line, the idea behind finding a regression line (line of best fit) is based on the assumption that the data is scattered randomly about a single straight line. Correlation Coefficient – 1 is perfect – either positive or negative – 0 indicates the data have no correlation. Correlation is the term used to indicate how close to the single line the data fall