How would you design risk-based monitoring for a multisite trial with limited resources?

Short answer

Start from the factors critical to quality, assess what could go wrong with each, and spend monitoring effort where harm is likely and hard to detect. Combine centralized review of accumulated data with targeted remote and on site visits, and set rules that increase monitoring where signals appear.

Why

E6(R3) asks the sponsor to set the extent and nature of monitoring from identified risks, considering design, blinding, endpoints and the safety profile. Monitoring can be on site, remote or centralized, and centralized monitoring may complement site visits or be used on its own. FDA's 2023 question and answer guidance on risk based monitoring says the risk assessment should weigh likelihood, detectability and severity, should be documented, and may be requested at inspection.

With limited resources the logic is to stop spending visits on low value checks. Full source data verification of every field finds many trivial errors and few important ones. Instead, verify consent, eligibility, primary endpoint data and SAE reporting closely, and let data analytics watch for outliers, late entry and odd variability across sites.

The monitoring plan should state triggers, such as a site with high screen failure or missing endpoint data, that lead to a targeted visit. Risks and the plan are reviewed as the trial runs.

How to work it through

List the critical to quality factors from the protocol, such as consent, eligibility, dosing, the primary endpoint and SAE reporting.
Score each risk for likelihood, detectability and impact, and document the method.
Assign controls, choosing centralized analytics for patterns, remote review for documents and on site visits for what only a visit can check.
Define key risk indicators and thresholds that trigger escalation or extra visits.
Write these into the monitoring plan with clear owners.
Review indicators monthly and update the risk assessment when the trial changes.

An example

A 25 site trial of 400 participants has budget for about 120 site visit days. The team picks 5 critical factors and sets indicators, such as SAE reporting over 24 hours and missing primary endpoint data over 5 percent. Every site gets a remote consent and eligibility review for its first 3 participants.

Central statisticians compare site distributions each month. In month 4, site 17 shows identical blood pressure readings across visits and late entry, so a targeted visit is triggered and finds values copied forward. The site is retrained and its monitoring frequency raised.

Where the answer stops holding

Centralized monitoring depends on timely data entry and good data standards, so it is weak early in a trial or with paper based sites. Some risks, such as drug storage conditions or source that exists only on paper, still need someone on site. Thresholds set without historical data may trigger too often or too rarely and should be tuned.

A strong answer covers

Critical to quality factors as the starting point
Likelihood, detectability and impact in a documented risk assessment
Centralized, remote and on site methods matched to risk
Key risk indicators with escalation triggers
Continuous review of risks and plan

Common mistakes

Treating risk-based monitoring as simply visiting sites less often.
Keeping 100 percent source data verification and adding analytics on top.
Writing a risk assessment once and never updating it.

The interviewer might ask next

Which key risk indicators would you choose for an oncology trial with a time to event endpoint?