Integrated summaries are among the most demanding deliverables in clinical development. They require sponsors to bring together data from multiple studies, often conducted across different phases, geographies, designs, vendors, standards versions, endpoints, and operational contexts. An Integrated Summary of Safety and Efficacy is not simply a large pooled dataset — it is a structured evidence package that supports regulatory understanding of a product’s safety and efficacy profile. For sponsors, ISS and ISE work should begin long before the final submission timeline. If integration planning starts too late, teams may face inconsistent data standards, coding differences, endpoint misalignment, missing documentation, and programming complexity that could have been reduced earlier.

ISS and ISE serve different purposes

The ISS focuses on the safety profile across studies. It may include adverse events, serious adverse events, deaths, laboratory abnormalities, vital signs, ECG, exposure, concomitant medications, special safety topics, and subgroup analyses.
The ISE focuses on efficacy evidence across the development program. It may include primary and secondary endpoints, supportive analyses, subgroup analyses, sensitivity analyses, exposure-response relationships, and consistency of treatment effect across studies.
Both require careful statistical and programming planning. However, the safety and efficacy integration challenges are different. Safety data often involve broad pooling and consistent coding across studies. Efficacy data may require deeper attention to study design, endpoint definitions, estimands, populations, comparators, and statistical methods.

Integration starts with a program-level data strategy

The strongest ISS and ISE packages are supported by a program-level data integration strategy. This should define how studies will be standardized, what data will be pooled, which standards versions will be used, how differences will be documented, and how cross-study analyses will be performed.
Important questions include:

• Which studies are included in each integrated summary?
• Are all studies using compatible SDTM and ADaM standards?
• How will legacy studies be converted or reconciled?
• Are adverse events coded using consistent MedDRA versions?
• Are medications coded consistently?
• How will treatment arms be grouped across studies?
• How will exposure be summarized?
• Are endpoint definitions consistent enough to pool?
• Are populations defined consistently across studies?
• Which subgroup variables are available across the program?
• How will missing or non-comparable data be handled?

These questions should be answered before programming begins.

Cross-study consistency is the central challenge

ISS and ISE programming is complex because studies are rarely identical. Even within a single development program, studies may differ by design, population, dosing regimen, comparator, endpoint, visit schedule, data collection method, and standards version.
Integration requires sponsors to distinguish between differences that can be harmonized and differences that must be preserved and explained.

For example, adverse event pooling may require treatment-emergent definitions that are consistent across studies. Laboratory data may require unit conversions and reference range handling. Efficacy endpoints may require harmonized derivations or separate analysis strata. Exposure data may need consistent dose grouping across formulations or regimens.

Poor integration can lead to misleading results. Over-harmonization can obscure meaningful differences. Under-harmonization can make pooled analyses difficult to interpret.

Coding consistency matters

Medical coding is a major ISS consideration. If adverse events or medications were coded using different dictionary versions, sponsors need a strategy for alignment. This may involve up-versioning, recoding, reconciliation, or careful documentation.
Coding decisions can materially affect safety summaries. Differences in preferred terms, grouping, or coding conventions may change event counts and safety signal interpretation.
The integration plan should address:

• MedDRA version consistency.
• WHODrug version consistency.
• Handling of verbatim term differences.
• Recoding strategy for legacy studies.
• Quality control of coded terms.
• Documentation of coding changes.

Coding strategy should be aligned with medical review, safety, data management, and programming.

Integrated ADaM datasets require strong specifications

ISS and ISE deliverables often require integrated ADaM datasets. These datasets must support pooled analyses while preserving study-level traceability.
Integrated ADaM specifications should define:

• Study identifiers and pooling variables.
• Treatment group mapping.
• Population flags.
• Analysis parameters.
• Baseline rules.
• Visit or timepoint harmonization.
• Endpoint derivations.
• Exposure definitions.
• Subgroup variables.
• Handling of study-specific differences.
• Traceability to study-level SDTM and ADaM.

Without strong specifications, integrated programming becomes vulnerable to inconsistent assumptions and repeated rework.

ISS and ISE require close biostatistics and programming collaboration

Integrated summaries are not only programming exercises. They require statistical judgment.
Biostatisticians should define pooling strategy, analysis populations, subgroup strategy, multiplicity considerations, sensitivity analyses, and interpretation of heterogeneity across studies. Statistical programmers operationalize these decisions in integrated datasets, outputs, metadata, and documentation.
The two functions need to work closely, especially when studies differ in design or when endpoint harmonization is not straightforward.

Reviewer documentation is critical

Integrated summaries require clear documentation. Reviewers need to understand which studies were included, how data were pooled, what differences existed between studies, how those differences were handled, and how results should be interpreted.
Documentation may include reviewer guides, define metadata, integrated SAPs, programming specifications, pooling rules, and explanatory materials for major assumptions.
This documentation should not be generic. It should explain the actual integration decisions made for the program.

Common ISS and ISE pitfalls

Common issues include:

• Starting integration too late.
• Underestimating legacy study conversion work.
• Inconsistent coding across studies.
• Weak treatment group mapping.
• Endpoint definitions that are not harmonized.
• Population flags that differ across studies.
• Missing traceability from integrated datasets to study-level data.
• Insufficient documentation of pooling decisions.
• Quality control focused only on outputs instead of integration logic.
• Lack of early regulatory and statistical alignment.

These issues can create pressure close to submission and complicate reviewer interpretation.

Bioforum’s Approach to ISS and ISE Programming

Bioforum supports ISS and ISE programming, integrated ADaM datasets, TFLs, legacy data conversion, CDISC standards implementation, reviewer documentation, and independent quality review. Our teams combine statistical programming, biostatistics, and data management expertise to help sponsors build integrated packages that are traceable, consistent, and submission-ready.
This work is especially important for sponsors moving from individual study reporting to program-level evidence generation.

Building an Integration Strategy That Holds Up at Submission

ISS and ISE delivery requires more than pooling datasets. It requires a program-level integration strategy that connects clinical meaning, statistical analysis, programming implementation, data standards, coding, documentation, and quality control.
For sponsors, early planning is the best way to reduce late-stage risk. The stronger the integration strategy, the more confidently the program can move toward regulatory submission.

Ready to Strengthen Your Integrated Data Package?

Preparing an ISS, ISE, pooled analysis, or multi-study submission package? Bioforum’s statistical programming and biostatistics experts support integrated datasets, analysis outputs, reviewer documentation, and independent quality review.