Did the intervention actually happen? Measuring fidelity
A project plan describes the intervention you intended to deliver. Your results depend on what was actually delivered. Without some record of implementation, an unchanged outcome can be difficult to interpret.
Implementation fidelity concerns how closely delivery follows the intended intervention. Carroll and colleagues describe content, coverage, frequency, and duration as important aspects of delivery, while also considering quality and participant responsiveness. [1] For a DNP project, choose a manageable record that captures the components most important to your question.
Distinguish reach from delivery
Hypothetical example. You plan a communication workshop with a demonstration, practice scenario, and feedback. There are 24 eligible staff members. Eighteen attend, but only 12 receive all three components.
| Question | Result in the fictional project |
|---|---|
| Who was reached? | 18 of 24 eligible staff attended, or 75%. |
| What dose was delivered? | Record the sessions or practice opportunities each person received. |
| Were the core components delivered? | 12 of 18 attendees received all three components, about 67%. |
| What proportion of the eligible group received the full intervention? | 12 of 24, or 50%. |
All these numbers can be correct at the same time because they have different denominators. Reporting only “the workshop was completed” hides the variation in exposure.
Keep a small, useful implementation log
Before starting, identify the core components and how you will know they occurred. A log might record the date, planned component, delivery status, number reached, and reason for a deviation. Use project codes only when individual linkage is needed and approved.
A brief observation checklist may be useful when quality of delivery matters. If you use one, describe its criteria and train the people applying it. A locally created checklist should not be described as a validated instrument without supporting evidence.
Record adaptations, not just failures
Real settings change. Moving a session to another time may preserve its core content. Removing practice and feedback may change what the intervention actually is. Document the modification, reason, timing, and whether the required approval was obtained.
Low fidelity does not prove that an intervention would have worked if delivery had been perfect. High fidelity does not prove that any observed benefit was caused by the intervention. The log helps you interpret these possibilities rather than settle them automatically.
Your next decision
Write a short “delivered as intended” definition for your project and choose a feasible way to check it. Review delivery while the project is running so that correctable problems can be addressed. See Outcome, process, and balancing measures for process measures and Sustaining the change after your project ends for a workflow that can continue after you leave.