4.5 Fidelity, Adaptation, and Quality Monitoring

Preserving what matters while responding to context

Fidelity is the degree to which a program is delivered as designed. It matters because the objectives, theory, activities, dose, and evaluation measures were engineered to work together; if implementation changes the core components, the program may no longer test the original idea, and outcome data become hard to interpret. A CHES professional must know what must be preserved and what can be flexed.

Core components are the elements believed to drive the program's effect, often called the active ingredients or theory of change. In a skill-building program they might include demonstration, supervised participant practice, feedback, and a minimum number of sessions. In a peer-education program they might include trained peer facilitators, structured discussion, and referral information. In a reminder intervention, message timing and content may be core. These elements should be identified during planning, before delivery begins, so the implementation team knows the non-negotiables.

When adaptation helps and when it harms

Adaptation is not automatically bad. It is often necessary for cultural relevance, language access, disability access, schedule fit, or setting constraints. Reasonable adaptations include swapping in locally relevant food examples, offering sessions at different times, translating and then reviewing materials, adding captions, or moving from an agency office to a trusted community site. These improve reach and effectiveness when they keep the mechanism of change intact, and researchers distinguish such surface adaptations from deeper changes.

Adaptation becomes risky when it removes or weakens the active ingredient (sometimes called program drift). If a curriculum relies on role play to build refusal or negotiation skill, replacing role play with a lecture undermines the core method. If an intervention requires three coaching contacts and only one is delivered, the dose has changed. If credibility depends on peer leaders, swapping in an unknown outside speaker may reduce fit. Such changes should be reviewed and approved before proceeding, then documented.

Fidelity monitoring should be planned and supportive, not punitive. Tools include session checklists, facilitator self-reports, observation forms, audio or video review when appropriate and consented, participant feedback, attendance logs, and supervision meetings. Monitoring focuses on the agreed elements: content accuracy, sequence, time on task, participation, practice opportunities, referral steps, and safety procedures.

Quality monitoring also requires responsiveness: if participants are confused, attendance declines, or a partner reports logistics problems, the team investigates and adjusts, documenting the reason, date, person responsible, and expected effect.

Fidelity data are most critical when outcomes are weak. If a program did not improve a skill, was it because the theory was wrong, the objective was unrealistic, or facilitators skipped the practice step? Without fidelity and process-evaluation data the team cannot answer, and may abandon a sound program or keep a flawed one. Fidelity also supports fairness across multi-site delivery: if sites receive very different versions, participants do not get equal access to the intended benefit, so standard materials, facilitator training, coaching, and shared monitoring tools become essential.

On exam items, when a facilitator changes activities on the spot, judge whether the change improves fit while preserving core elements; when in doubt, choose the answer that consults the plan, documents the change, and monitors its effect.

Types of fidelity and a decision rule for adaptations

Implementation-science frameworks describe several dimensions of fidelity worth recognizing on the exam. Adherence is whether the prescribed components were delivered. Dose or exposure is how much was delivered and received. Quality of delivery is how well facilitators delivered it. Participant responsiveness is how engaged participants were. Program differentiation is whether the unique active ingredients were present and not blurred with other content.

A scenario describing a facilitator who delivered every component but rushed and lost the group is an adherence-versus-quality contrast; one describing two sites delivering different core content is a differentiation problem.

Use a simple decision rule when a scenario proposes a change. Ask: does the change touch a core component (the active ingredient), or only the surface (examples, wording, scheduling, venue, format)? Surface changes that improve fit are encouraged, documented, and monitored. Core changes are paused for review and approval, because they can break the link between the program and its expected effect. A second question follows: is the change planned and documented, or unplanned and silent?

Even an acceptable adaptation loses value if no one records what was actually delivered, since the later evaluation will not be able to interpret the results.

Worked example: a peer-led HIV-prevention curriculum requires trained peer educators, structured small-group skill practice, and three sessions. A site, short on peers, proposes using a clinic nurse to lecture all content in one longer session. Apply the rule: this removes the peer-educator credibility component, replaces skill practice with lecture, and cuts the dose from three sessions to one, touching three core components at once.

The right response is to pause, train additional peers or recruit from a partner, and preserve the three-session practice structure, documenting any interim plan, rather than proceeding with a version that no longer reflects the program theory.

• Adherence: were the prescribed components actually delivered?
• Dose/exposure: how much was delivered and received versus the plan?
• Quality of delivery: how skillfully and respectfully was it delivered?
• Participant responsiveness: how engaged were participants?
• Program differentiation: were the unique active ingredients present and distinct?