PRACTICAL GUIDE · 9 OCTOBER 2026
RBQM in clinical trials: a practical guide
Learn how Risk-Based Quality Management (RBQM) connects critical-to-quality factors, study risks, KRIs and QTLs. Follow a six-step workflow, compare fictional examples, and put the process into practice with the free RBQM tool.
RBQM means Risk-Based Quality Management. It focuses on what could affect participants or the reliability of results, from design to closeout. Risk-based monitoring (RBM) is part of that work.
ICH E6(R3) promotes a proportionate approach. In the EU, the Principles and Annex 1 have been effective since 23 July 2025; Annex 2 becomes effective on 15 January 2027. The guideline does not turn a matrix or software product into a guarantee of compliance. [1]
First, define what must not fail
In a fictional study, the primary endpoint depends on a week-12 assessment. The critical factor is obtaining a valid assessment within the specified window. Poor visit scheduling could cause that assessment to be missed. This connection is more useful than a generic inventory of every possible problem.
KRIs and QTLs: two levels of interpretation
| Element | Purpose | Fictional example |
|---|---|---|
| KRI | Indicator that helps detect a risk signal. It may be assessed by site or at another level. | Proportion of queries open for more than 30 days at each site. |
| QTL | Prespecified study-level parameter limit linked to a critical factor. A deviation requires evaluation. | Overall proportion of missing primary assessments. |
A metric name is not enough: document the population, numerator, denominator, time window, source, frequency, limits and response. In this Workbench, KRIs can be configured by site or as aggregates and QTLs always operate at trial level. This is an explicit product simplification, not the only way to organize RBQM. TransCelerate provides reference resources on risk assessment and indicators. [2]
A practical six-step workflow
- Identify critical factors. Involve clinical, statistics, data, operations and safety teams; include participant burdens.
- Describe specific risks. Record the cause or failure mode, consequences, owner and existing controls. Consider detectability too.
- Prioritize and prevent. Estimate probability and impact, choose mitigations and estimate residual risk. Severe consequences deserve attention even when unlikely.
- Define usable indicators. Justify limits using the protocol, experience or simulation. Set a minimum base and maximum acceptable data age.
- Review signals in context. Check completeness, timeliness, denominators and trends before attributing a problem to a site.
- Respond and verify. Record investigation, decision, owner and due date. Check effectiveness before closure and review the plan periodically.
Example: the same percentage, different evidence
One site has 1 aged query out of 5 and another has 20 out of 100: both show 20%. They do not have the same stability. With a fictional minimum base of 20, the first is marked insufficient. Stale data quality is marked separately; the limit signal remains visible. For a study aggregate, 1 + 20 out of 5 + 100 gives 21/105 = 20%; in general, do not average site percentages without weighting. Counts must use comparable definitions and periods.
How to respond to an exceeded limit
In this example, investigate whether there is a loading delay, inconsistent definitions or genuinely missed visits. Assess the impact on the endpoint and consult the risk owner. The response may be to clarify data, strengthen a process, train the team or open a corrective action. Document why you act—or why action is not warranted—and when you will assess the effect. A traffic light does not prescribe a clinical decision.
What this tool supports
The Workbench links risks, KRIs/QTLs, measurements and actions; retains measurement history; identifies missing, insufficient or stale data; and exports CSV, JSON and reports. The P×I matrix and demo thresholds are illustrative. There is no EDC connection, shared user management, immutable version history, electronic signature, centralized statistical monitoring or GxP validation. For controlled use, the sponsor must assess and document system and process suitability.
Open free RBQM WorkbenchFor deviation, root-cause and follow-up documentation, explore the Quality pack.
View Quality packRBM impact: eight metrics and context
The RBM impact panel includes major/critical findings per audit; unreported confirmed SAEs; significant deviations per treated participant; average monitoring cost per site; interval between on-site visits; and medians of eCRF entry, query closure and significant/major issue closure. Definitions adapt TransCelerate RBM Metric Narratives (2015).
Deviations are events per participant, with possible values above 1. Cost uses one currency. Intervals are divided by consecutive visit pairs, not visits. Aggregate medians need an explicit base and are not averaged across sites; query median can be calculated from raw snapshots and excludes unresolved queries.
These metrics are impact KPIs and have no automatic limits. Configure an earlier cutoff as baseline and confirm comparability of population, definition, coverage and currency to view absolute and relative changes. An initial increase in findings or deviations may reflect better detection. Fewer visits or lower cost do not prove better quality. A lower median must be reviewed alongside outstanding cases and safety signals.
TransCelerate RBM Metric Narratives (2015) · RBM Metrics Report (2019)
Sources and scope
- EMA — ICH E6(R3), Principles and Annex 1; current and future effective dates
- TransCelerate — Risk Based Monitoring Solutions
Educational content and original PharmaTemplate examples. No affiliation or endorsement by the cited organizations. Reviewed 9 October 2026.