Scoring Risk and Recommending a Site

Part of Clinical Research Associate (CRA) Training: Evaluating and Qualifying a Site.

About 5 minutes. Free, in your language, no account needed.

Why this matters

After the pre-study visit you have impressions, notes and documents. The sponsor needs a decision-grade recommendation.

Turning one into the other is a skill, and it is where CRAs most often produce something unusable: either a bland endorsement or a list of observations with no conclusion.

Risk levels

Most sponsors use three levels. Learn the shape even if the labels differ.

The middle row is where most sites sit and where your work matters most. "Moderate risk" is only useful when it is accompanied by named, dated, checkable mitigations.

What to score

Score consistently across sites, using the same categories:

• Investigator and staff capability, including protected time and backup.
• Patient population and realistic recruitment potential.
• Facilities and equipment adequacy for this protocol.
• Operational capacity: workload, processes, scheduling, cover.
• Compliance and regulatory history.
• Technology capability.
• Financial and contractual readiness.

The point of a scorecard is comparability. Two CRAs assessing two sites should produce reports a sponsor can put side by side.

The risks that recur

Across studies and countries, the same risks appear:

• Overworked coordinators. The most common and the most predictive.
• Investigator disengagement. The most damaging, and the hardest to fix later.
• Poor documentation habits, visible in how the site describes its own record-keeping.
• Incomplete regulatory files, which usually indicate no owner rather than no capability.
• Uncalibrated or unmonitored equipment.
• Weak safety processes, particularly no defined route for out-of-hours events.
• Unrealistic recruitment projections that have not been revised after validation.

Writing the recommendation

Three options, and say which one plainly:

• Select.
• Conditionally select, with a mitigation plan.
• Do not select.

Then support it. A usable recommendation has five parts.

1. The conclusion, first. One sentence.
2. The evidence, specific and attributable. Not "storage is inadequate" but "study product requires 2 to 8 degrees Celsius; the fridge alarm is not monitored outside working hours and there is no written excursion procedure."
3. The risks, each with a likely consequence for this protocol.
4. The mitigations, each with an owner and a date. Mitigations without owners are wishes.
5. The realistic enrollment estimate, with the arithmetic behind it.

Avoid three habits that make a report unusable: adjectives without evidence; listing every observation without ranking it; and softening a serious risk because you liked the people. You will meet these people again, and the way to keep the relationship is to be accurate now.

Comparing sites

Once all sites are assessed, the sponsor needs a comparison. Provide:

• A ranking by capability and by risk.
• Recruitment potential per site, on the same basis.
• Whether resources match the protocol's complexity.
• A view on top choices and backups.
• Any dependencies between sites, such as a shared laboratory or a shared patient catchment.

Be explicit about competition for patients. Two sites drawing from the same catchment do not add together.

When you are pressured

You will sometimes be pressured to upgrade a site: a timeline is slipping, or the investigator is well-known, or the country needs representation.

Do not change your assessment. Change the mitigation. "This site is moderate risk; it can be selected if a second coordinator is funded and the first monitoring visit occurs within six weeks of first enrollment" is a constructive answer that keeps your evidence intact.

Selection is the sponsor's decision. Your obligation is that the decision is informed.

In practice

Leah assesses three candidate sites for VT-301.

Site 1 is straightforward: experienced, well documented, uncontested patients. Low risk, select.

Site 2 has strong patients and staff but a fridge alarm nobody hears after 18:00 and no backup coordinator. Moderate risk, conditionally select, with two mitigations and dates.

Site 3 claimed 60 participants and could not produce any arithmetic. The investigator was unavailable for the pre-study visit and sent apologies twice. Two staff members separately described a different consent process. High risk, do not select.

The sponsor's project manager pushes back on Site 3, because the country has no other candidate.

Leah does not upgrade the assessment. She writes a supplementary note setting out what would have to be true for the site to become acceptable: a documented recruitment analysis, direct engagement from the investigator, a written consent procedure with training, and a first monitoring visit within four weeks of activation.

The sponsor selects the site on those terms. Two are met, two are not, and the sponsor makes that decision knowingly.

Your job is not to be the gatekeeper. It is to make sure nobody walks through the gate by accident.

Key takeaways

• Three risk levels — low, moderate and high — and moderate risk is only useful when accompanied by named, dated, checkable mitigations.
• Score every site on the same categories so a sponsor can compare reports side by side.
• A usable recommendation states the conclusion first, then specific evidence, ranked risks, mitigations with owners and dates, and a realistic enrollment estimate with its arithmetic.
• The recurring risks are overworked coordinators, investigator disengagement, poor documentation habits, incomplete files, uncalibrated equipment and weak out-of-hours safety routes.
• When pressured, change the mitigation rather than the assessment, because selection is the sponsor's decision and your obligation is that it is informed.