Digital Certificates Of Analysis: A Vision For The Transfer Of Quality Data
By BioPhorum
Biopharmaceutical supply chains have become increasingly complex, often involving multi-enterprise manufacturing collaborations and partnerships to supply lifesaving medicines to patients. This evolution beyond the single-source manufacturing model is driving a need for standardized approaches to data integration between organizations to support agile and predictive supply chains.
In this ever-evolving landscape, the integration of digital technologies has become paramount. One of the most significant advancements is the adoption of digital certificates of analysis (CoAs), which are transforming the way quality data is transferred between organizations in the pharmaceutical supply chain.
This article proposes a set of best practices for digital CoAs that promise to enhance efficiency, accuracy, and compliance across the sector.
Best Practice: Align Solutions Across All The Use Cases In The Supply Chain
There are many use cases of CoAs in the pharmaceutical value chain. In the future, digital CoAs might pass between raw material and packaging suppliers, shipping partners, contract manufacturers, contract laboratories, and a sponsor organization with affiliates and distributors, with the resulting data about the product examined by clinicians, patients, customs officers, and health authorities.
It is common to see projects established to digitalize certificates for any one of these use cases. However, best practice is to design the solution so it can be extended to serve all the other use cases as well in the future.
Crucial to the success of digital CoAs is to align the solutions across all the different use cases in the supply chain. This is what will allow a true end-to-end picture to be built up.
Best Practice: Design For Secondary Uses
The primary reason for implementing digital CoAs is to help the release manager review the batch quickly and efficiently. However, there is an important secondary use of the same data, exemplified in a user story for the data scientist.
A release manager wants to review the digital CoAs for the batch of raw materials or finished products received to easily verify and validate that the product received is correct and meets specifications. Consistent and reliable availability of digital CoAs for incoming raw materials, various manufacturing consumables, and finished products would significantly improve the overall efficiency of material receipt. The digital format eliminates issues with paper versions and can improve process efficiency by providing access to this data before shipment receipt. It also allows the initiation and preparation of the material review process, accelerating the real-time release as products arrive at the receiving dock.
A data scientist wants to identify potential trends, process or product variabilities, etc., that could impact final product specifications or effectiveness so they can predict deviations, prevent out-of-specification situations, and improve quality and process yield. To support and enhance these activities and objectives, digital CoAs should be provided in a format that can be easily ingested into existing process analytics tools and models and contain additional information to help predict process variability. This would also help minimize the errors due to manual transcriptions, resulting in more accurate process models and predictive analytical results.
Best Practice: Consider The Process Implications
Technology does not produce benefits; it is only when it is used to enhance business processes that the positive impact is felt. What changes will be needed to make the most of an increased digital maturity and standardization in the way digital CoAs and other test result data are passed between organizations? These changes include:
- The ability for digital CoA data to be sent ahead of the physical product allows the receivers of incoming material to start the review and approval process earlier.
- Optimizing the review-to-approvals process can then open the door to optimizing further related processes, such as procurement and warehouse receipt processes.
- Following data standards means aligning the digital CoA process and method with the rest of industry. A single approach is critical for industrywide adoption.
- With digital data transfer in place, it is important to change processes and information access (e.g., in meetings) to avoid relying on paper CoAs. Digital CoAs and their constituent data would need to be more readily accessible at any time required.
- Having ready access to data gives the opportunity to improve certain processes, but in organizations required to follow documented procedures, it needs to be made explicit that digital data can be used.
- With suitable technology enablers in place, the business needs to encourage and support partners to step up the maturity ladder.
- Once digital CoA data has been auto-aggregated with other information to support release, the release process needs to change to rely on any automatic checking and reduce human intervention.
- Once there is a mechanism to create digital CoAs in all the formats required to meet the regional requirements of regulators, the CoA production process needs to change to use the new mechanism rather than the old manual approach.
- Once there are data products based on digital CoA and other test results data, there needs to be people and processes in place to use the data assets to leverage benefits, identifying and implementing optimizations of the process and supply chain. Plans to transition to structured data approaches need to include updating standard operating procedures and other quality management system documentation.
Best Practice: Don’t Compromise On The Data Model
The general principles and related criteria guiding the modeling of data entities and the relationships between them include:
- Data entities should be representative of real-world use cases and contain comprehensive and common attributes.
- It should be possible to model the one-to-many relationships found in the real world.
- Message and data elements must be traceable to their sources to ensure the quality of the data.
- Data elements need to be accompanied by metadata to provide relationships and context for using the data.
- Data structures should be designed to make it possible to realize a message using PDF, CSV, XML, or JSON.
- Messages ought to be automatically extracted from or ingested into common systems used to capture, store, report and analyze data, so the entity model needs to be mappable to common data systems to facilitate the implementation.
- Entity names should be both general and understandable, matching what things are called in the real world.
- Entity models should be simple, scalable, and standardizable.
Best Practice: Adopt Controlled Vocabularies
Building on the converged logical data model (which identifies all the data items expected in each entity), we should move toward controlled vocabularies.
It is not only the names of the fields that need to be aligned but also the values that can be contained in those fields. For example, the status descriptor must be the same and mean the same for data analyzed from different sources to make sense. The same applies to many of the fields used in digital CoAs. Free text should be the occasional exception.
Of course, controlling vocabularies according to an external standard means that in a partnership both parties will need to map to and from their operational systems (rather than the contract organization mapping to each sponsor’s requirements). However, once done, that mapping should be good for future partnerships and be quicker to implement, with improved accuracy for everyone. It also makes it much easier to combine data from many partners.
In time, we expect the data managed by companies in data lakes and data products to increasingly conform to controlled vocabularies and, eventually, for these to be adopted into the operational systems.
Best Practice: Optionality Should Be By Use Case
Data fields in a standard are traditionally mandatory or optional. However, if a standard is to cover the range of situations found in pharmaceutical value chains, it needs to consider optionality by partner, use case, market, and modality.
As experience with a variety of use cases, markets, and modalities grows, we expect that the optionality statements will be refined into a matrix that shows what is recommended to be mandatory and optional for particular situations. Each partnership will then use that as the basis for deciding which of the optional items to implement.
Use cases may not all need the same data fields. For example, sponsors may require extremely detailed information to perform advanced analytics, but affiliates may only require the summary to release the batch. More details may be required to support an investigation.
Best Practice: Choose A Structured Data Serialization Format
The downside of being agnostic to the information container is the need for some senders and receivers to support more than one type for the benefit of partners at lower digital maturity. The primary driver for choosing a data serialization format is digital maturity at both ends of the data transfer. But if there is a choice, should it be PDF, CSV, XML, or JSON?
Structured data serialization formats (such as XML and JSON) lead to higher digital maturity, offering the potential for end-to-end validation, increasing efficiency by eliminating human transcription and checking and improving the cycle time.
Using a format standardized according to a common logical data model offers the potential for easier joined-up analytics combining data from multiple partners. Using common field names and semantics (albeit in different formats) makes it more straightforward to transform incoming data from several sources into a common destination data structure.
Best Practice: Support Different Levels Of Digital Maturity
There are varying levels of digital maturity in the many industry partnerships across which this data must flow. To support these, the vision is to have the flexibility to realize the data transfer as PDFs, CSVs, XML, or Javascript object notation (JSON), and for it to either be sent or pulled on demand. If all options are built on a common data model, it makes it easier to handle the variety of formats and paves the way for a future step up in digital maturity.
The partnering relationship between organizations needs to include a commitment in the contract to work toward the digital exchange of data. There is a useful outline of digital collaboration agreements in the BioPhorum DISCO playbook.1
Best Practice: Focus Implementation Effort To Deliver Key Benefits
The digitalization challenge is huge, and it is easy to attempt too much. The first implementation of digital CoAs may require new technologies and changes to existing business processes. It is important to focus implementation efforts to deliver key benefits on a specific need, before moving on to expand the scope to achieve full benefits for both the supplier of the data and the recipient. To secure the necessary funding and resources for implementation, organizations need to be clear on how this kind of project aligns with key business strategies, objectives, and priorities.
The benefits of implementing the technology enablers for digitally driven process improvements and making business changes include eliminating the potential for human error, releasing batches faster, reducing complexity, and enabling advanced analytics.
Call To Action
The digital exchange of CoAs and related data is generally at a low maturity level in the pharmaceutical industry. The above highlights some best practices for the use of digital CoAs, and business and digital leaders are encouraged to:
- review current strategies and plans for implementing digital CoAs and include data scientist perspectives
- champion the benefits of using a digital CoA, especially moving away from paper-based and PDF format and promoting a future where all levels of digital maturity can be satisfied
- identify and overcome roadblocks to progress, including a technical platform solution for sending and receiving digital CoAs as structured data.
This article summarizes some of the main points from a recent BioPhorum publication on this topic. To read more, check out the full paper, Digital certificates of analysis.
Reference