Managing Process Risk Line by Line: An FMEA Case Study
In applying Failure Mode and Effects Analysis to batch record line items, Lonza Biologics rooted out targeted process risks and instilled an ethic of continuous improvement.
By Fred Greulich, Maxiom Group, and Jonathan Hardy, Lonza Biologics, Inc.
This article provides a case study of the use of Failure Mode and Effects Analysis (FMEA) to identify and manage process risk in a biotech manufacturing environment. While FMEA is an established tool (see end of article) and is not new to biotech manufacturing, this case is unusual in tfhe level of process detail that was analyzed and the amount of rigor with which the tool was applied.
FMEA typically examines risk at the overall process level; in this situation, it was used to examine and manage the risk inherent in each line item in a set of batch records associated with a few specific manufacturing processes.
This FMEA project took place at Lonza’s Portsmouth, N.H. site, a large-scale mammalian cell culture facility, with the objective of mitigating process failures and reducing overall manufacturing process risk. The scope of the effort was to apply FMEA and root cause analysis (RCA) tools to the eight batch records for the purification process of a customer’s product. While the site has a comprehensive operational excellence program and has used FMEA and RCA tools in the past, this project represented the first time the tools were applied to batch record line items. As a result, the project team was able to develop and implement targeted solutions and mitigate specific areas of process risk.
A Three-Phase Approach
The project was undertaken in three phases (Table 1), starting with education and ending with implementation. The first two phases were conducted with the active involvement of cross-functional Lonza teams to increase quality and ownership of results. The teams typically met once or twice per week in working sessions of two to four hours. Because the solutions targeted for implementation were mainly focused on manufacturing, Phase III work was championed by the Manufacturing group with significant support from the Engineering, Controls, Validation, and Quality Assurance teams. The entire project was conducted over approximately one year.
Phase I: FMEA Assessment and RPN Scoring
Results from Phase I were captured in a typical FMEA spreadsheet format (Table 2). The spreadsheet was populated in real time during the working sessions.
The second row of Table 2 is a sample of the completed spreadsheet. Note that the Potential Failure Effects, Potential Causes, and Current Controls entries are all coded. For consistency in analyzing and summarizing the assessment results, the team identified a set of possible entries for each column and assigned codes for each. Using these codes sped up the process by reducing the amount of typing required to complete the spreadsheet.
Important to success in this phase was that all members of the team used the same assessment and scoring standards. Templates developed by Maxiom Group were customized during the initial Phase I team session. These templates—one each for the Severity rating scale, the Occurrence rating scale, and the Detection rating scale (Table 3)—used a 1 to 5 system with a clear description of each rating. Each team member kept copies of these templates and referenced them throughout the assessment sessions.
After the entire team of six to eight Lonza employees had completed the assessment and RPN scoring for a few batch records, patterns began to emerge. For example, each of the four chromatography-related batch records followed a pattern—set up equipment, set up skid, run multiple cycles, and complete post-chromatography activities. Based on these patterns, Maxiom Group proceeded to “pre-populate” the spreadsheets for the remaining batch records. The successive sessions with the Lonza team then focused on reviewing and editing, where necessary, the pre-populated spreadsheets. This saved significant time and energy for Lonza personnel.
During Phase I, the team began with the analysis of eight batch records, which eventually mapped to 25 critical failure modes. The critical failure modes were defined as those failure modes associated with the batch record steps that were determined to have high RPN’s.
Determining the high RPN’s was done by plotting each unique RPN (developed for all the line items for a specific batch record) on an X-axis and the RPN “values” on the Y-axis. For each of the eight batch records, we were able to overlay two lines representing slopes that approximately connected the plotted points. The first, more gradual line was associated with the majority of RPN values, while the steeper second line highlighted a smaller set of higher RPN values. It was these values that were considered the “high RPN’s.”
In all, 2,600 batch record line items were analyzed, and 266 high-risk RPNs assessed. Of the 25 critical failure modes identified, just six were selected to move into Phase II. Many of the other failure modes were already being addressed by Lonza through other initiatives, and Lonza leadership wanted to focus efforts only the vital few failure modes.
Phase II: Root Cause Analysis (RCA) and Solution Identification