By 2018, all global vendors of medicinal products (pharmaceutical companies) that operate globally or in the European market will be obliged to ensure that all submissions of product information follow a standardized data model with standard vocabularies and code sets. This standardized data model is defined by the ISO IDMP set of standards. Companies can face hefty fines for non-compliance.
In the first of a series of articles, we present some basic IDMP concepts and why data governance is important for the industry to tackle the complexity and the data among their different systems.
IDMP – Identification of Medicinal Products – is a set of ISO global standards designed to support the unequivocal identification and description of medicinal products throughout their entire lifecycle. This lifecycle – from research to production, marketing and pharmacovigilance – involves a complex and variable set of attributes and a range of systems across departments. Even within the same company, this complexity already requires strong governance.
To support global identification of products, the IDMP standards harmonize key concepts, as well as their relations, meaning and attributes; this enables consistent terminologies, naming standards, etc.
Source: official graphic from ISO TC 215, WG6
The picture above shows the key concepts in IDMP, and how they relate to each other: Pharmaceutical Product (i.e. a formulation, which includes the substances and other attributes); Medicinal Product (the marketable form, or commercially available product); product characteristics like units of measure, dose form, etc.; and the many details and rules around these concepts.
Regulatory authorities (e.g. EMA in Europe) are relying on IDMP, and pharmaceutical companies are expected to submit their product information in an IDMP-compatible way. InThis means that pharmaceutical companies need to adopt IDMP and govern their data to ensure:
- They can produce the required attributes and identifiers
- They can use the right value sets
- Their internal systems respect the governance and rules dictated by the standards, as well as the internal rules of the company
- Their submissions are controlled, to conform to the rules
With such great complexity, and such business impact and penalties for non-compliance, IDMP adoption is not only an IT problem, but must be well governed by the business areas – Research, Regulatory, IP, etc.
This is where data governance shows its advantages. It brings the complexity to an understandable, traceable process, avoiding many spreadsheets and scattered technical communication.
Data governance does not mean ignoring details or creating shortcuts. It is about giving executives and non-technical people the insight into what data is need, where, and how much of that data respects the internal governance aspects.
Concrete IDMP example:
ISO 11616 defines some concepts and attributes and rules – e.g. the Pharmaceutical Product is identified bythe Pharmaceutical Product ID, and is defined by attributes like Substance, Strength, Dose Form, Device, etc. The details on substances are further harmonized by ISO 11238.
But the standard does not define the value sets for substances – they are defined elsewhere, e.g. by the GINAS system and value set, which is made available from the Global Substance Registration System (G-SRS)system, hosted by the WHO in Uppsala, Sweden. The same SRS value set is used in the UNII which can be accessed from the NLM/NIH. Similarly, the units of measure are harmonized by ISO 11240, and the code list is UCUM, which is available from unitsofmeasure.org. Dose Forms (among others) are normalized by 11239 and a reference value set is the Standard Terms dictionary, maintained by the EDQM. If multiple value sets can exist, this usually implies a need for governing each of these value sets.
ISO 11615 states that a Pharmaceutical product is present in the Medicinal Product, which is identified by a Medicinal Product ID, and has attributes like Marketing Authorisation Number, Classification, Name (where Name can have components like invented name, strength, intended use, etc). The Medicinal Product is packaged as Packaged Medicinal Product, which has other attributes… These many attributes are associated with a set of rules about cardinality, optionality, uniqueness, etc.
So this is fairly comprehensive, and companies must master this model and their IT systems must show to respect the rules. Instead of using many spreadsheets, data governance solutions enable these concepts to be visualized and traced in an accessible manner:
The figure above shows several types of concepts and their articulation:
- Business assets – concepts, in blue – and their relations
- Data assets – code sets, in yellow – and their relations
- System assets – systems, in red – and their relations
- Governance assets – standards, in green – and their relations
Data governance enables organizations to capture the same concepts and structure described above in a simple, traceable manner, which enables organizations to understand and monitor all aspects without delving into the IT aspects. Without data governance, these relations would be much harder to manage, even just for the small sub-set above.
Focusing on business aspects makes it possible to include other governance aspects like the rules – e.g. attributes may be optional or mandatory, identifiers have to be unique in some conditions, and more. Data governance enables not only to explore these rules and their traceability to the other assets, but also to actually check whether those rules are properly enforced in the systems that hold the data. In other words, not just in the model, but monitoring the actual data in the IT systems. But that is for another article.
Together with its customers, Collibra is establishing a reference implementation for IDMP (among other areas where healthcare and life science industries can benefit from data governance), to enable a fast adoption of data governance throughout the institution and where it matters for compliance.
We accelerate trusted business outcomes by connecting the right data, insights, algorithms and people for all Data Citizens.