Improving Supply Chains Through Analytics

Increasingly complex supply chains and source globalization in Life Sciences manufacturing has created an environment of uncertainty regarding the quality of incoming materials and the need to monitor these materials more closely. Because vendor-supplied quality checks submitted in the form of C of A always show that the material is within the required specification limits, the goal is to detect when these test results may be faulty, identify unusual and potentially dangerous trends, and reveal when a vendor may have changed their process, source material or manufacturing location. Such due diligence requires constant monitoring and the application of analytics to all appropriate data, yet many Life Sciences companies do not have comprehensive programs for monitoring raw material supplier quality using current-generation analytics and reporting tools. Data Access DeniedWhile it is universal to require that vendors perform quality checks on all materials and to provide Certificates of Analysis, this data is made available in a wide range of non-standard formats (paper, proprietary-format text files, Excel, etc.) and is not often collected in an accessible data system for immediate or long-term analysis. Most life science companies also perform some incoming materials testing, with the test results entered into a LIMS or equivalent system. Not all incoming materials are subjected to comprehensive testing, and the data from tests that are performed are not always subject to routine or real-time analytics. The combination of inaccessible supplier data and incomplete internal test data makes it difficult to routinely check supplier results with internal testing and nearly impossible to perform long-term analysis to detect trends or process changes. Closing these gaps becomes a strategic imperative and several of NWA’s customers sought an assist in improving their Vendor/Raw Materials monitoring programs through the application of analytics. These solutions have ranged from applying analytics and notifications services accessing existing databases to providing methods for collecting and managing vendor-supplied information.For NWA’s customers, initiating a comprehensive Raw Materials data management, analysis and reporting program generally began by taking one or more of the following steps:Step #1 – Collect vendor-supplied test results into an accessible, validated database. This has required a combination of approaches, depending on how vendors provide the data, which is usually in the form of a Certificate of Analysis. Methods include:

Manual data entry from printed or electronic images of Certificates of Analysis.

Automated import from electronic submissions via Excel (XLS) and other user-generated files.

Automated import from function-specific files and file formats such as CSV and XML.

The system must also include data validation and review to meet GMP and other regulatory standards, and Steps 2 through 4 frame an effective process: Step #2 – Integrate analytics applications with the systems managing Vendor Raw Materials test data via standard connection technology such as ODBC, OLE-DB, or database-specific API. Step #3 – Define specifications and the appropriate analytics for each data element acquired in Step #2. Step #4–Create Dashboards, Reports and Notification Services for each user or group of users that need to monitor Vendor Raw Materials quality.Completing these steps successfully provides a comprehensive, real-time monitoring capability for a wide range of vendor-supplied raw materials results integrated with in-house testing programs. Users will be notified of out-of-specification results, unusual trends and patterns, and significant differences between vendor-supplied and in-house test results before they manifest themselves in process and/or product quality issues.Compliance and Good SenseRegulatory requirements, GMP and good sense require methods for tracking issues, the steps taken to resolve them, and the steps taken to prevent their re-occurrence. A good AC, CA, CAPA (Assignable Cause, Corrective Action, Preventative Action) program also accumulates knowledge over time and presents this knowledge to assist in quickly identifying solutions to future problems. It is essential to dynamically collect AC/CA/PA information at the source — the users responsible for identifying, characterizing, correcting and preventing problems, and not acquire this information after the fact. Advanced Univariate AnalyticsConventional (and relatively easy to implement) univariate analytics can detect issues that are readily identified by analyzing individual measurement parameters. This will provide the quickest, most accessible results and will detect the most common and easily addressed issues. However, since vendor-supplied data is controlled artificially (only shipping material that tests “in specification”), more subtle issues may not be detected, such as a change in the manufacturing process or the sourcing of critical components. Also, a material may be manufactured in multiple locations with different processes that deliver material with different characteristics. This can affect production and product quality even when a raw material appears to be in specification according to typical testing. One approach to addressing this issue is to include multivariate techniques in the range of analytics applied to the data. Models based on multivariate statistical techniques such as PCA and PLS can deliver analytics similar in form and interpretation to the univariate analytics already deployed. These techniques can detect more subtle changes — identifying when materials move outside their “design space” and indicating that a significant, but difficult to detect, change has been made.NWA is currently implementing systems that perform the four basic steps of a comprehensive Raw Materials/Supply Chain Analytics system with a major pharmaceutical manufacturer, as well as a large food services company and a prominent, leading specialty chemical firm. Notable is the fact that all three presented the same set of problems at nearly the same time, reflecting a growing concern in multiple market segments.For NWA’s recent customers, the Raw Materials/Supply Chain application provided:

Configurable Data Entry forms tailored for keying in test results from Certificates of Analysis sent by vendors. The software immediately writes the results to a centralized Quality Information System database hosted on MSSQL Server. The input forms are designed to reduce the chances of keystroke errors and provide immediate feedback when values entered are outside of the expected range. A second set of forms is used to review and approve the data to meet GMP, also flagging data that appears to be outside expected ranges.

Automatic import of Excel files sent by vendors into product-specific forms that allow manual review prior to moving the data to the same QIS database. This data is also reviewed/approved by another staff member on a second set of forms to meet GMP.

Connecting “off-line” and “real-time” analytics and reporting applications to the Incoming Raw Materials database to access the results of vendor-supplied raw materials data. Only data approved during the “GMP” step can be accessed.

Connecting the Analytics and Reporting applications to the LIMS to access the results of in-house testing of the incoming raw materials.

Creating role-specific analytics reports, real-time dashboards and email/SMS notification to put the right information into the right hands at the right time.

The analytics results from the Raw Materials database and LIMS are evaluated and any results meeting pre-defined criteria are flagged. A pre-selected group of staff is notified via real-time Dashboards, email, and text (SMS) messages. Authorized staff can access historical data to further evaluate potential problems or specific suppliers. This will be the first time that the full range of vendor-supplied QC data and in-house laboratory test results have been instantly available and the data accessible for other analysis and reporting tasks.BenefitsAs a result of readily available data, applied analytics and immediate notification of potential problems, the company realizes benefits by:

Avoiding quality issues reported as adverse patient outcomes that could have been prevented by better use of vendor-supplied information and internal testing.