by
Jeffery L. Cawley, Northwest Analytical, Inc.
Robert B. Gregory, Abbott Laboratories E-Ink, Inc.
John C. Bates, Abbott Laboratories

Abstract

The study examines the combination of an SPC web server with a LIMS to extend the reach of product quality reporting within a manufacturing plant. The LIMS manages the test data for high purity water production within the Chemical and Agricultural Products Division of Abbott Laboratories and the SPC web server delivers SPC charts and associated product quality reports to internal customers. The system provides control point quality data critical to processes up- and downstream from the affected sites, enabling more rapid quality decisions regarding process monitoring and management.

The scalability inherent in the SPC web server is particularly advantageous in expanding both the accessibility and the content of process quality reporting. Scalability factors include lower time and cost overhead to add additional users to the reporting system when compared to standard LIMS configurations, lower overhead to implement increased analysis capabilities for users, and reduced overhead for creation or expansion of content for reports using web page technology.

This improved access to process data enables better and more timely systematic investigation of process quality by more areas of the enterprise. Because the web pages integrate control charts and process capability reports, and include numeric data for more in-depth analyses, each user is provided greater information content and better communication of process performance.

Enterprise-wide SPC Reporting

High purity water production at the North Chicago, IL Chemical & Agricultural Products Division, Abbott Laboratories facility provides one of the critical raw materials for the finished products. It is critical that all sections of the plant have access to water quality data in order to best manage production.

The water quality data is collected in a Beckman Lab Manager LIMS. The problem was to both provide alarming for out of control conditions and deliver water quality information to other sections of the facility.

An existing database system provided provides data tracking in parallel to the LIMS system and delivered simple data trending reports. The desire was to eliminate redundant databases to reduce overhead and provide a greater certainty of data integrity.

The goal was also to reduce the costs and manpower requirements of quality reporting. The decision was made to integrate a web based SPC reporting system, NWA Quality Analyst Web Server, with the LIMS.

This system was designed to transfer quality data to both upstream and downstream processes and to enable rapid and correct quality decisions.

Web based reporting (Figure 1) provides greater scalability than conventional network based systems. Adding additional users requires only a personal computer equipped with a browser and access to the company intranet. The result is less overhead both to add users and to maintain the system.

Since the web server only limits access to the provided control chart images to on the web pages, the users were are separated from the production data. This simplifies providing a high level of data security.

The promise of the web server was improved access with enhanced reporting in the form of control charts and process capability reports instead of numerical reporting and simple trend charts which provide less useful information. In general, the system produces better, more timely decision making.

Figure 1
SPC Web Server simplifies collecting data from laboratory and process measurements and distributing SPC reports to all parts of the enterprise and its supply chain.

Microbial Environmental Control

In the pharmaceutical industry, the quality of the ambient environment, as well as materials used in manufacture which are produced in-house or supplied in bulk, must be controlled and monitored to provide the highest level of cleanliness for the production of drug substances and products. To guarantee the level of cleanliness, sampling is performed in a variety of modes to test as many critical variables as necessary. (Figure 2)

Figure 2
Samples are taken at all critical points to monitor for microbial contamination.

Gasses piped throughout the plant are sampled and tested for purity and other quality parameters. This ensures that each supply drop will deliver the required substances at the required quality at all times during manufacture.

The air and the surfaces in rooms that are critical to the manufacture of pharmaceutical materials is are sampled and tested for microbial levels and particulate counts. Supplied gases are tested against specifications. This testing ensures that the materials manufactured in the rooms will remain uncontaminated from extraneous ambient conditions. In some areas, worker clothing is sampled for microbial contamination.

Water systems and their corresponding drops and faucets are tested against specifications at the source and at the spigot. All surfaces are swabbed and tested for microbial contamination, and in some areas, worker clothing is sampled for microbial contamination This testing includes microbial, purity, and impurity testing designed to guarantee that the water used meets or exceeds the level of organic and inorganic purity required for the intended use, and that no microbial contamination exists. The water system represents the most interconnected system in the environmental monitoring program, and will be used to illustrate the construction of this new quality system.

Monitoring Distilled Water Production

The water used in manufacture at Abbott in Lake County is produced by a central facility that continuously monitors the quality of the outflow stream. This system distributes the water to individual water systems used in the manufacturing buildings. These systems use break tanks to buffer the central supply for the manufacturing line. These tanks are then connected to the usage sites through a closed-loop distribution system. The system supplies the water to the manufacturing line through a series of drops located at the point of use.

Figure 3
Pure water production system at Abbott CAPD plant.

The water sites and break tanks (water source from corporate system) are individually sampled for Aerobic Microbial Count (AMC) and LAL (endotoxin) on a twice weekly basis for hot-loop systems, daily for ambient temperature systems.

Sites are individually sampled monthly for "chemical" tests (pH, conductivity, total organic carbon, etc.), and sampled for coliform on a quarterly basis. In total there are over 3000 sampling sites, with over 500 for water alone.

Sampling is done randomly throughout the sampling period for each test. Each site is monitored for history against an action limit (based on compendial specifications) and an alert limit (performance based limit determined on an individual site basis). Each building is monitored in the same way, and all results are recorded in Beckman Lab Manager.

Each building has a quality assurance person (MQA) who has responsibility for the quality decisions regarding products manufactured in the building. Each building also has a building manager who has responsibility for the physical systems in the building. These two individuals also have the combined responsibility to investigate and maintain the quality related systems in the building.

Currently, exception data (over-action, over-alert events) are investigated after their occurrence, followed by corrective action, and monitoring of the site for trouble-free operation. The limitation of to this system is that it does not recognize trends in data prior to failure that might predict future performance. The use of a more advanced data trending approach is necessary to achieve a higher level of control.

The solution is to develop a package which assembles statistical control charts and utilizes statistical process control of the sites so that the MQA and building managers can proactively monitor and improve the performance of their systems.

In the first phase, the LIMS system assembles statistical quality control charts to facilitate the assessment of the history of a sampling site and its parent water system. This history helps the owner determine the severity of the loss of control of the process. The owner specifies which sites will be monitored using which chart types. At prescheduled intervals, a batch process is initiated which runs the specified charts, and posts them to the building managers' electronic mail accounts. At this point, the existing system of using alert- and action levels remains the method to document the outage and corrective action responses for individual sites.

In the second phase, the data are posted automatically to an internal web site along with the data files used to produce them. The data and plot files are also posted to a document vault for secure storage. The user has access to the original chart, while specific individuals have access to the plotted data to enable more in-depth analysis using alternate ways of presenting the data. No user has access to the stored data in the LIMS system, as the web site is independent of the underlying database, and no user has the ability to alter any data used to produce the web chart. This ensures data integrity and provides assurance that the electronic data remain accurate throughout their useful life.

In the third phase, the trends extracted from the data are used to initiate investigations to proactively improve performance. These trends are generated on a regular basis.

Additionally, when a site actually produces an exception event, the system will automatically generate a chart documenting performance to assist in the investigation of the event. At this point, the system will use regular control charts and, eventually, automated alarm functions based on statistical evaluations to analyze performance deviations earlier than the current alert- and action level-based system.

Conclusion

The addition of web-based SPC reporting to the LIMS system has enabled enterprise-wide quality reporting. This extends the effective use of the water quality data beyond the laboratory and immediate production management. This in turn supports effective process management and underwrites process improvement efforts.