How Merck standardized its statistical computing environment

Q: How should a pharmaceutical organization evaluate a new SCE when no single platform meets all requirements?

Lock in non-negotiables before reaching out to vendors, then design the assessment to understand how those companies handle technical stress. Merck accepted early on that every product matrix would surface some functional gaps. What mattered was finding a provider willing to build missing architecture directly into their standard codebase, rather than treating the gap as an isolated customer patch. This active cooperation turns a software vendor into a long-term infrastructure partner.

Q: What signals indicate a legacy statistical computing environment has reached its functional end of life?

The clearest signal is when audit trail capture falls onto the programmer rather than being managed automatically by the environment. In a modern SCE, compliance is an intrinsic property of system architecture, not an administrative task for an individual scientist. When developers have to design manual workarounds to stitch fragmented tools together, the platform is no longer viable. Merck’s ARP built on Domino shifts this burden, letting the platform track behavior automatically while practitioners focus on data exploration.

Q: What is the SCE Coalition and what practical value does it deliver for pharma companies?

The SCE Coalition is a cross-company consortium of pharmaceutical organizations designed to combine industry requirements and coordinate long-term influence over platform vendor roadmaps. When an individual sponsor challenges a technology vendor alone, their ability to steer product development is limited. Uniting as a group enables member companies to carry significantly more weight when pushing vendors to fix capability gaps that affect the entire industry.

Merck transitioned from a 25-year-old legacy platform to a modern analysis and reporting platform built on Domino, shifting compliance from a manual tracking burden to a native property of their platform infrastructure.

Uday K Reddy Kandula

Principal Scientist, Statistical Programming

Merck

What you'll take away from this session

When statistical programmers must manage compliance by hand, the underlying architecture has failed

Forcing teams to bridge tool gaps manually with Excel introduces significant human error. Compliance must be a native property of the system design rather than a behavioral requirement.

Set non-negotiables before vendor sales cycles start

Merck established fixed data, tool, and compliance requirements before looking at external software, preventing the team from rationalizing platform gaps under sales pressure.

A vendor’s response to shortcomings reveals their true value as a partner

No platform met every Merck requirement. The critical signal is whether a vendor will actively engineer missing capabilities into their core product when challenged.

Running dual legacy and modern systems in parallel is expensive but protects regulatory submissions

Maintaining system parity and validating live data across both environments simultaneously is resource-heavy, but it is a fraction of the cost of a questioned FDA filing.

Live platform demonstrations shift user resistance far better than written training documentation

Skeptical statistical programmers rarely believe automated audit trails are real until they watch the system log metadata and track code transformations entirely on its own.

Merck runs one of the pharmaceutical industry's largest statistical programming operations, managing thousands of analytical artifacts under strict regulatory boundaries. Their legacy statistical computing environment, CPI, served the company reliably for 25 years. However, its aging architecture required researchers to manually maintain audit trails, which grew increasingly risky as modern clinical study data formats expanded. The breaking point arrived when Merck realized its scientists were spending too much time proving compliance rather than focusing on core scientific analysis.

To address this, Merck launched a multi-phase transition to build ARP, their new Analysis and Reporting Platform built on Domino. Uday Kiran Kandula, Principal Scientist of Statistical Programming, maps out this journey as a lesson in vendor management. Merck locked in rigid, non-negotiable criteria upfront: data ingestion flexibility across any database, open-source tool integration, and automated audit capture. When initial market reviews surfaced zero out-of-the-box matches, Merck refused to compromise. Instead, they challenged Domino on its process management gaps. By pushing back, Merck transformed a standard vendor relationship into a true development partnership, with Domino adding missing governance and audit features directly into its core product codebase.

Today, ARP connects live clinical data repositories with flexible R and Python environments. Every code transformation automatically creates an audit trail, creating an environment that is always inspection-ready. Uday details the deployment timeline, starting with right-sized proof-of-concept tests, advancing to total study reconstruction using historical datasets, and entering their current phase of running live data in parallel to guarantee system parity before fully decommissioning the old platform.

FAQ

How should a pharmaceutical organization evaluate a new SCE when no single platform meets all requirements?

What signals indicate a legacy statistical computing environment has reached its functional end of life?

What is the SCE Coalition and what practical value does it deliver for pharma companies?

Transform the work that matters most

See how Domino helps the world’s most regulated enterprises build, scale, and govern AI-powered applications.

Watch demo Talk to an expert