V. “Bala” Balasubramanian, Orion innovation

Imagine this hypothetical scenario, which is not farfetched in the life sciences industry. Let us assume there is a safety event in a market due to a contaminated drug product and the health authority in that market issues a recall. If the drug sponsor would like to proactively withdraw it from all markets and if the company does not have good data about which markets receive the product  from a specific manufacturing site, then the sponsor is likely to be impacted by a number of challenges including loss of revenues, compliance and/or reputation risk, and potential penalties.

The information about market approvals and product manufacturing sites related to approvals is usually managed by regulatory in what is called a Regulatory Information Management (RIM) system.

stated that “While industries such as financial services have had a long track record of managing data, and applying analytics to optimizing customer relationships and developing new services, life sciences companies have only in recent years begun to fully embrace and seize upon the opportunities to organize and apply their data in a systematic way to a range of drug development and patient care challenges.” While the life sciences industry has begun to see the advantages of treating data as a strategic asset, pockets of the industry have always played a key role around data as a strategic asset. For example, the clinical function, due to its very data-driven nature, has always focused significantly on data as a strategic asset given that the safety and efficacy profile of a drug largely depend on good data management practices.

Data can be a strategic asset only if there are enough processes and support to govern and manage data quality. There are several areas of impact due to bad data quality.

For example, bad data may result in increased costs such as frequent remediation, increased resource needs, health authority penalties, etc.

Bad data quality leads to impaired decision-making or forecasting. It can result in potential loss of revenues due to lost opportunities or lack of patient access to latest therapies.

There can also be increased compliance or reputation risk  like the hypothetical example cited earlier.

In a recent World Class RIM survey of 65 companies, 62% of them reported that their data standards and data governance processes are inefficient (see figure).

Good data quality could be achieved through people, process and technology considerations (see figure below).

Define the right organizational and governance structures to encourage and sustain good data.

Embed data quality in performance objectives supported by incentives and/or penalties.

Provide change management/training on the importance of data ownership and inculcating a data quality culture.

Establish data governance processes along with data quality metrics.

Perform periodic process and information audits.

Define processes for data stewardship, data standards management, data quality reviews, data remediation and verification.

Develop appropriate standard operating procedures for ensuring and managing data quality.

Improve the overall user experience by enforcing data entry rules, constraints, automatic naming conventions, alerts?notifications for upcoming or overdue data entry tasks, etc.

Eliminate duplicate data entry where possible by integrating with other sources of truth, such as product master.

Leverage standard reference data dictionaries such as MedDRA, WHO, ISO, etc.

Leverage artificial intelligence(AI)/machine learning (ML) techniques to unify and cleanse data.

Strong data quality begins with strong data governance (see figure below). Based on our experience with large data quality initiatives in multiple industry sectors, we recommend the following:

Gain executive sponsorship to establish data governance with appropriate authority.

Create a cross-functional data governance team including business users, business analysts, data stewards, data architects, data analysts and application developers.

Establish strong data governance, managed and adequately funded and staffed by the right individuals, with sponsorship at an executive level.

Set up strong governance structures including data stewardship, proactive monitoring and periodic reviews of data.

Define business glossary data standards, nomenclature and controlled vocabularies. Increase adoption of controlled vocabularies established by WHO, MedDRA, ISO and others.

Establish and/or leverage strong Master Data Management for product-related data. There must be only one inclusive and common way of defining a product.

Define data validation and business rules which are embedded into the process and systems in place.

Define data stewards for various business domains and establish processes for the review and approval of data elements and/or data.

Eliminate data duplication by integrating data wherever possible through interfaces to other systems.

Here, we present a short case study of how a life sciences organization established a strong data quality program. Key objectives of the program were:

As part of a new global transformation initiative around RIM, there was a desire to get data quality right the first time.

Avoid issues faced with multiple data remediations over prior RIM implementation over two decades.

Increase adoption and usage of new processes and systems.

We supported this client with the following:

Defined data quality metrics and processes.

Recommended domain-based data stewardship.

Defined processes for periodic compliance checks on a weekly, monthly and quarterly basis.

Instituted frequent follow-ups with countries.

Provided frequent change management sessions and training.

The outcomes that very visible to the organization included:

Better data and metrics for management reporting and decision-making.

Significant reduction in number of compliance issues across various process areas.

Greater global adoption of new processes and systems.

As a company, we support various aspects of the customer data journey, including current state assessment, defining and implementing data strategy and data governance approaches, designing and implementing master data management, applying visualization and analytics to generate insights and supporting data quality operations.

In summary, a number of the challenges outlined in the introduction about a hypothetical scenario of a contaminated drug being withdrawn can be managed effectively through good data quality and integrated end-to-end visibility into data.

Good data quality starts with the realization that data is a strategic asset with the following key principles to be observed:

Data culture needs to become part of the psyche of the entire organization, right from senior leadership to operational personnel. This has to be supported by performance objectives and incentives.

Establish strong data strategy and governance mechanisms.

Advanced analytics and AI/ML make sense only if there is good quality of data and true digital transformation is possible only with sustained data quality.

V. “Bala” Balasubramanian, PhD, is Senior Vice President at Orion Innovation and heads the Life Sciences IT division  

Articles

Building in the right processes and support for data quality, standards, and governance