Advanced Operating Models
Mar 20, 2017

Master Data Management – Mystery, Myth or Magic?

Master Data Management (MDM) can be defined as something that “comprises the processes, governance, policies, standards and tools that consistently define and manage the critical data of an organization to provide a single point of reference"—and that single point of reference is also known as “the golden record." MDM is therefore a very far-reaching concept, one that has the potential to be misunderstood and misconstrued, whether perceived as something too large and complex to ever implement effectively, or as something that can solve almost any data management problem it is presented with.

Examination of each component of MDM can help considerably in understanding what MDM can and should do, as well as what it should not do.

What is being managed?

The thing that MDM exists for is to manage the “critical data of an organization." Different types of organizations will by their very nature have different types of critical business data. For pharmaceutical companies, some of the main areas of critical data include:

  • Products - Medicinal products, delivery devices
  • Components for products - Ingredient substances, packaging
  • Studies - All safety activity and study data
  • Processes - Submissions, authorizations, and study management
  • People – In various roles, such as employees, investigators, and study subjects
  • Other organizations, such as regulatory agencies and suppliers
  • Resources and facilities, including buildings and equipment

Companies can start with one or more key areas, and expand into other areas as experience and business need requires, leveraging opportunities from business drivers (e.g., compliance requirements, or provision from other areas such as the EMA's Organization registry) to manage the scope.

Why is this being managed?

The value of the golden record to an enterprise is that it increases both data quality (by reducing or eliminating inconsistent data, duplicate data, and unreliable data) and efficiency (by reducing or eliminating effort spent in correcting or cleaning data, and therefore reducing or eliminating misunderstandings around data). The increase in data quality and efficiency will manifest itself in different ways for the different process that occur within the various parts of an enterprise. For example, the single definition of a medicinal product globally will mean that a regulatory department can more easily track what is authorized for that product in any one jurisdiction and across all jurisdictions, thereby managing international birth dates and updating the required regular submission documentation more efficiently. Similarly, that single definition supports the supplies/logistics departments, enabling them to better track how their products are being sold; to anticipate and therefore address product shortages before they become critical; and to have a clearer view of issues related to counterfeiting that may arise. Having a single description of the processes and procedures that are undertaken to conduct the business is important especially for tracking and benchmarking across different areas in a global organization.

Note that traditionally, MDM was first implemented to support large-scale analytics (“big data" and its forerunners), often sitting on top or alongside a large data warehouse, providing data mapping between the same data items and aiming to resolve inconsistencies (e.g,, when a source had a slightly different nuance for a data item) as data flowed into the warehouse from different sources. Current thinking, especially for medicinal product data, is that rather than trying to resolve those nuances, it is both possible and more desirable to eliminate them through the use of a standard information model (ISO 11615).

How is it being managed?

MDM is achieved through a combination of processes, governance, policies, standards and tools. The standards are most clearly seen used in the MDM tools. All the things that can be classified as “master data" have various features (“attributes") that describe them: for example, a medicinal product has a dose form, and if that dose form is a tablet, the tablet will have a size, shape, and color. They also have relationships between them: for example, a medicinal product will have an authorization to allow its supply within one or more jurisdictional markets.

Unfortunately, each element classfied as master data, and its attributes and relationships may be described in different ways in different systems, which often means that it is hard to find the single source of truth. For example, is the tablet 6mm in diameter, or is it 6.2 mm? Was the 6mm a rounded number because system one can only take an integer value? Sometimes, the description is only available in a document, which means it is even harder to find a piece of critical information quickly. In large documents, using a keyword search and scrolling through is very time-consuming.

MDM uses a standard pattern (a model) to describe the critical business data of an enterprise – the elements that can be classified as master data, their attributes, and the relationships between them, to remove the ambiguity and to support the creation and maintenance of a living golden record for everyone in the enterprise to use. This is where the conceptual information model of IDMP in ISO 11615 and the supporting standards of 11238 and its implementation in the GINAS/SRS System for Substances and 11239 for Dose Forms, Units of Presentation, etc., play such a key role; as international standards, these draw on a wide range of expertise to define the core entities, attributes, and their relationships, in order to fully describe medicinal products in life sciences. The FDA's Structured Product Label (SPL) specification complies with the information model of IDMP and is an implementation of it. The IDMP model is also compatible with other standards in the clinical research domain, particularly the BRIDG (Biomedical Research Integrated Domain Group) Model which is an ISO standard that can be widely applied (e.g., for clinical study management and for submission management). It also critically supports pharmacovigilance and the Individual Case Safety Report (ICSR). Standards for regulatory information continue to be developed and refined: for instance, the Regulated Product Submission (RPS) standard (which should eventually also incorporate version four of the electronic Common Technical Document (ICH's eCTD)) and the Clinical Trial Registry standards (CTR) used for and EudraCT.

Note that standards that define and describe both the critical objects and their relationships are the most valuable, which is why some of the CDISC standards, where relationships between data items are less clearly defined, can have limited applicability. There are also standards for the management and use of the controlled terminology that is used in MDM. Standards themselves usually come with principles (policies and process) for their implementation and governance, either explicitly (e.g., ISO TR 20443 for the implementation of IDMP) or implicitly (for example, any controlled terminology standard will draw upon the principles of terminology maintenance standards).

By using standards—and in particular, by using the IDMP model in MDM—individual companies do not need to develop a product MDM for themselves, which would be expensive and inefficient. The standards also means that an MDM scope can be reliably expanded over time, as experience grows and benefits are realized, without costly reworking of foundations.

Any enterprise implementing MDM must take these standards and develop them into a set of processes to ensure relevant governance principles for their master data are adhered to. These processes, usually described in Standard Operating Procedures (SOPs) and responsibility matrices, describe how change can be managed (new items created, existing items updated, expired items retired) and how change itself can be requested.

How can MDM be supported?

An MDM system has two main objectives:

  • To build and maintain the master data (the golden record) for each element classfied as master data using the pattern required as well as the appropriate controlled terminology
  • To make the master data available for use across the enterprise

MDM systems provide a set of functions (capabilities) to achieve this. Some systems provide all the capabilities themselves, while others integrate capabilities together from subsystems to achieve the goal. For example, to build and maintain the master data, an MDM system may

  • Undertake to manage its information model itself or it may provide facilities to import an information model from a repository
  • Manage and version its controlled terminology itself, or it may integrate with a specialist terminology services system
  • Manage and version its registries itself, or it may provide facilities to import registry data from an external source—or it may do both, providing the necessary matching and ranking services for that

But all MDM systems must provide life cycle management for master data and support the business process for that life cycle (including roles and responsibilities and enforcing governance) and support impact analysis when changes to core information occurs. In addition, when data is imported into an MDM, the system should maintain traceability to the source(s) and provide reconciliation and remediation facilities for data cleansing. To support the use of the master data, all MDM systems must provide

  • Search and view facilities (including compare) for various views into the data
  • Data-sharing facilities:
    • Integration with other systems in the enterprise as a core data source 
    • Data extraction


When implemented effectively, based on well-accepted information standards, with a clear scope and practical governance processes, MDM is a powerful tool within an enterprise, increasing data quality and efficiency in data management and business analytics.

About the author

Julie M. James

Julie M. James

IDMP Consultant, Global Regulatory Affairs, Genpact Pharmalink