ISPE s Process Capability Team

Transcription

1 4 September 7 INDUSTRY MATURITY IN THE ASSESSMENT AND USE OF PROCESS CAPABILITY Arne Zilian Head MS&T Processes & Standards Novartis Pharma AG Process Validation Statistics Conference 5 September 7 ISPE s Process Capability Team - Performance and Compliance Applications PQLI ISPE Process Capability Team - Statistical Considerations 3 - Maturity Model Development Team Goals (per Team Charter): To develop a position on process capability as a performance and/or compliance metric Reportable to FDA to support a risk-based inspection program as outlined in sections 75 to 76 of US Food and Drug Administration Safety and Innovation Act (FDASIA). Acceptable to industry, addressing challenges for implementation. To communicate this position & gain support for implementation across pharmaceutical companies via: A series of articles and/or white papers. Potential baseline guide. Sessions at ISPE and other meetings.

2 4 September 7 Goal: Improve Quality and Ensure Reliable Supply.. Performance and Compliance Applications. Self-audit of process performance helps to prioritize improvements Embed a product robustness review process in the organization (at site level/ supply chain end-to-end/ for overall product portfolio). Statistical Considerations. Process Capability is an easy and meaningful metric to capture and report but there are challenges: Real life is drastically different from textbook examples Difficult to predict failures using Cpk/Ppk Many failures are not related to Cpk/Ppk 3. Maturity Model Development. Such a model can be used to guide through the implementation journey Survey was conducted to assess where companies are today, where they want to be Common opportunities for improvement have been identified 3 Process Capability Indices: The Good, The Bad, and The Ugly ASTM defines process capability as the natural or inherent behavior of a stable process that is in a state of statistical control, which is achieved when the process exhibits no detectable patterns or trends. (ASTM E8). Statistical tools such as control charts are utilized in the determination of process capability, however capability indices are often commonly used. The Good: Dimensionless index which can be used to compare across products, attributes, etc. The Bad: Highly dependent on meaningful specifications and well-behaved data sets. The Ugly: Process Capability is not the whole story. Complexity associated with practical application of these tools limits their effectiveness as a compliance requirement. 4

3 4 September 7 Three Levels of Product Robustness Monitoring Rely on Associates with Product Understanding 3. Quarterly Portfolio Review Sr. Management Scorecard looking at all sites and products. End to End Product Reviews --Cross functional team across Drug Substance, Drug Product, Analytical, Stability and Quality --Deeper Statistical Analysis applied to Hot Spots Attribute X Drug Product Manufacturing Site -Approximately Monthly inter batch data reviews -Voice of the Process at the Shop Floor -Site Focused but also feeds data upward 5 Product Robustness Involves Many Elements Bika, D., Walsh, J., Nestor, T., and Allenspach, C. Product Robustness: A comprehensive Product Quality Strategy, AAPS Newsmagazine, p 6-9, August

4 4 September 7 Process Robustness Continuous Improvement Over Time Release Attributes Summary Low <. Medium.-.33 High >.33 Significant improvement in robustness performance observed over the past 5 years Example from Peter G. Millili, BMS. Y-axis: proportion of CQAs with high/medium/low Cpk 7 Statistical Considerations Gap between Reality and Textbook A simple interpretation of process capability metrics is often not possible Must understand details of the data, and how results will be used. Otherwise, actions may be improper! ISPE Discussion Paper focuses on: ) use of process capability for risk based decsions for action and ) identification and practical treatment of non-textbook situations 8 4

5 4 September 7 Using Ppk for Risk Based Decision for Action Select relevant CQA and IPC DRAFT Here risk is negligible hence no further action required. Here risk is acceptable hence no further action required. Further analysis required. Maybe action required to remediate risk. Maybe accept risk based on cross-functional review. 9 Statistical Considerations: Textbook Example Individual Value Process Capability Sixpack Report for Assay 3 I Chart UCL=3.73 _ X=99.5 LCL=94.36 LSL 96 Capability Histogram 98 4 USL LSL 95 USL 5 5. Moving Range Chart UCL=5.76 Normal Prob Plot AD:.565, P:.3 Moving Range.5 MR=.763. LCL= Last 5 Observations Capability Plot Values 4 StDev.563 Cp.7 Cpk.86 PPM StDev.57 Pp.9 Ppk.88 Cpm * PPM Observation 5 3 Specs 5

6 ISPE Process Validation Conference 4 September 7 Statistical Considerations - Reality Small data sets Small number of unique values, including case with many values less than LOQ Non-independence resulting from non-random use of sources of variability Non-normality based on process performance, not customer driven (clinically relevant) Process Shifts Due to Non Random Use of Sources of Variability Individual Value Moving Range Process Capability Sixpack Report for CQA I Chart Moving Range Chart UCL=.343 _ X=.338 LCL=.9853 UCL=.355 MR=.73 LCL= Capability Histogram USL Normal Prob Plot AD:., P: <.5.3 USL.3.3 Individual Value I Chart of CQA4 by Batch UCL=.33 _ X=.9755 Values Last 5 Observations 55 6 Observation 65 7 StDev.95 Cp * Cpk 3. PPM. Capability Plot Specs StDev.398 Pp * Ppk.7 Cpm * PPM Observation 6 36 LCL=

7 4 September 7 Process Shifts Due to Non Random Use of Sources of Variability Process Capability Sixpack Report for CQA3 I Chart Capability Histogram Time Series Plot of CQA3 Individual Value Moving Range Moving Range Chart UCL=.64 _ X=.583 LCL=8.55 UCL=.495 MR=.764 LCL= LSL USL 4 Normal Prob Plot AD:.64, P: <.5 LSL 8 USL 4 CQA Last 5 Observations Capability Plot 9 Values Observation 65 7 StDev.6769 Cp.48 Cpk.7 PPM 68.5 Specs StDev.975 Pp.3 Ppk.89 Cpm * PPM Index Data with underlying non-normal distribution Moving Range Individual Value Process Capability Sixpack Report for Tailings CQA 5 (kg) I Chart Moving Range Chart UCL=47.7 _ X=6. LCL=-5.3 UCL=38.67 MR=.84 LCL= Capability Histogram USL Normal Prob Plot AD: 9.7, P: <.5 USL 5 8 Moving Range Individual Value Process Capability Sixpack Report for Tailings CQA 5 (kg) Using Box-Cox Transformation With λ = I Chart 96 5 Moving Range Chart UCL=4.55 _ X=.539 LCL=.57 UCL=.47 MR=.757 LCL=.. Capability Histogram USL* 4. Normal Prob Plot AD:.4, P: USL* Values 5 5 Last 5 Observations StDev.49 Cp * Cpk.7 PPM Capability Plot StDev.48 Pp * Ppk.9 Cpm * PPM Values Last 5 Observations StDev.679 Cp * Cpk.68 PPM Transformed Capa Plot StDev.6969 Pp * Ppk.66 Cpm * PPM Observation 85 9 Specs Observation 85 9 Specs 4 7

8 4 September 7 Statistical Considerations - Other Topics Influence of Measurement Variability Manipulated Data Survey of Other Methods 5 Why a Maturity Model? As shown in the example, continuous improvement success comes after a multi- year effort. During the course of the program, Sponsor and workstream leaders may change Culture changes slowly Systems contribute to inertia New ways of working need to be implemented A compass is needed to find true north 6 8

9 4 September 7 What is a Maturity Model? A maturity model points out areas for improvement starting from a current ( ) level, going to a target level ( ) Level - Initial. Exclusively a business process, not in quality systems. Lacks some rigor. Ad hoc, reactive, driven by users or events. Level Repeatable. Proactive within some organizations/groups. Proactive but selectively applied. Level 3 Defined. Proceduralized. There are sets of defined and documented standard procedures established and subject to some degree of improvement over time. Level 4 Managed. Rolled out across sites, networks, organizational units and functions. Level 5 Optimizing. Proactive, integrated into the culture and operations management review. Comprehensive view of leading indicators and recurring root causes. 7 Maturity Matrix Consists of Nine Key Process Areas In order to use process capability and concept tools optimally, it is important to achieve a certain level of proficiency and/or clarity in 9 specific areas: Policy SOPs are established. Data Management a system for collecting, managing and accessing data. Frequency how often process capability indices are calculated. Basis for Specification manner in which specifications are developed and linked to clinical studies. Calculation Consistency use of process capability approach, calculations and metrics. Response thresholds with action required, attention is shifted to products with low capability. Organization Skill Set and Execution level of process capability knowledge across the organization. Risk-based Context part of an overall risk management framework. Commercialization sources of variability are well understood. Process capability at initial launch is high and site resources focus on continuous improvement. 8 9

10 4 September 7 ISPE Process Capability Maturity Model: Key Process Areas Optimization Commercialization Basis for Response Calculation Consistency Organization and Skill sets Policy Frequency Risk-based Approach Data Management Foundational For each key process area five levels of maturity are defined with detailed descriptions 9 Maturity Model Survey - biggest opportunities according to participating companies are Commercialization and Data Optimization Management Commercialization Big change planned: Lowest starting point Medium ambition Basis for Foundational Response Policy Calculation Consistency Risk-based Approach Big change planned: Low starting point High ambition Organization and Skill sets Frequency Publication submitted to Pharm Eng

11 4 September 7 Main Areas for Improvement Commercialization challenges - Small data set before/ after commercial launch - Little experience with applying process capability at this stage in the lifecycle survey responses: Majority of data is CQAs, some CPP data available., program is in early stages of maturity, starting to integrate risk based methods, using scale up models for site transfers and commercialization. (5 yrs)* Data Management challenges - Complex IT system landscape - Manual data management - Cost of IT system upgrade survey responses: Some databases are structured; some are manually entered/updated. Network roll-up not available (% manual) Product specific, may or may not be automated. (5 yrs)* On demand pull of capability data (5 yrs)* (*Number of years using Process Capability Measurements) KEY TAKE AWAYS Process Capability is only a small part of the story A sustainable business process is needed to be successful Textbooks are good to get started It takes years to reach maturity and some already got there!

12 4 September 7 ISPE PROCESS CAPABILITY COMMUNITY OF PRACTICE PERFORMANCE AND COMPLIANCE APPLICATIONS WORKSTREAM Peter Millili Team Lead, BMS Jason Orloff PharmStat Kevin Roberson ABC Labs Abizer I. Harianawala Ariad Julia O Neill Tunnell Consulting Abe Germansderfer SHL Philippe Cini Tunnell Consulting George Millili Roche Genentech STATISTICS WORKSTREAM Tara Scherder Team Lead, Synolo Stats James Crichton Cook Pharmica Mahesh Shivhare Novartis Andreas Dander Novartis Aaron Goerke Roche Genentech MATURITY MODEL WORKSTREAM Philippe Cini Team Lead, Tunnell Consulting Gretchen Allison Pfizer Gerald Leister Eli Lilly and Company Eda Ross Montgomery Shire Pharmaceuticals Julia O Neill Tunnell Consulting Mike Thomas Tunnell Consulting Eric Thostesen Janssen Pharmaceuticals Arne Zilian Novartis