The Grant, the Fidelity, and the Funder
You are the clinical lead on a state-funded grant to implement a manualized trauma-focused intervention for adolescents in a rural school district. The funder requires 'fidelity to the model,' and your contract specifies that clinicians must complete all 12 modules in sequence. Midway through the year, your data show that many adolescents are dropping out after module 6, and clinicians report that the rigid sequence conflicts with the realities of rural life - transportation gaps, seasonal farm work, and a high rate of acute crises (e.g., a peer suicide in the district) that interrupt the protocol. You believe a more flexible, phased approach would serve these youth better, but deviating from the manual risks violating the grant and losing funding that supports 60 families. Your outcome data so far are inconclusive because of the dropouts.
1. How do you weigh fidelity to the evidence-based model against adaptation to context, and what would you do?
Read the worked answer
This is the classic fidelity–adaptation tension in implementation science. Neither rigid fidelity nor free adaptation is automatically correct. Key principles:
- Distinguish core components from peripheral features. Evidence-based models have active ingredients (e.g., trauma psychoeducation, exposure/processing, skills) and surface features (sequence, number of sessions, delivery format). Adapting surface features while preserving core components is generally defensible; stripping core components is not.
- The evidence base is context-bound. A model tested in urban clinics may not transfer unchanged to rural, crisis-prone settings. The question is whether the mechanism of change is preserved.
- Your data are currently uninterpretable - dropouts confound outcomes. Before changing the model, improve your measurement: track why youth drop out (transportation, crisis, scheduling), and collect outcome data on those who complete.
- Engage stakeholders. Talk to the funder early and transparently - many grants allow documented adaptations with justification, and funders often prefer a learning approach to silent noncompliance. Consult the model developers, who frequently have guidance on adaptations.
- Pilot, don't overhaul. Test a phased or modular adaptation with a small cohort, with pre-specified fidelity and outcome metrics, before scaling.
- Document everything - the rationale, the adaptations, the fidelity monitoring, and the outcomes. This protects the grant and generates knowledge.
Decision: Do not simply abandon the manual. Propose a documented, core-component-preserving adaptation (e.g., modular delivery, crisis-responsive pause/resume, transportation supports) to the funder, pilot it, and monitor fidelity and outcomes. If the funder refuses, weigh the harm of rigid delivery against the harm of lost funding - and advocate with data.
2. What ethical issues arise in using incomplete outcome data to justify a program change, and how would you address them?
Read the worked answer
Using inconclusive data to justify change is a research-integrity and stewardship problem. Specific issues:
- Selection bias / missing data: Dropouts are not missing at random - they may be the youth with the worst outcomes or the greatest barriers. Concluding 'the model doesn't work' from dropout-confounded data is unsupported. Address by analyzing dropout reasons, comparing completers vs. non-completers, and using intention-to-treat logic where feasible.
- Confirmation bias: You already believe flexibility is better; that belief can shape how you read ambiguous data. Address by pre-specifying what evidence would count for and against adaptation, and by involving a skeptical colleague or methodologist.
- Stakeholder harm: 60 families depend on the funding. A unilateral change that jeopardizes the grant harms them. Address through transparent funder communication and a phased, low-risk pilot.
- Informed consent and transparency: Families should know if the program is being modified and why; adaptations should be disclosed.
- Equity: Rural youth face structural barriers (transportation, seasonal work). Rigid fidelity may systematically disadvantage them - an equity argument for adaptation, but one that must be evidenced, not assumed.
- Dissemination honesty: Report adaptations and outcomes accurately; do not overstate the evidence for the adapted model.
Practical response: Build a small evaluation plan - fidelity checklist, dropout tracking, pre/post outcomes, and a comparison of adaptation cohorts - and share interim findings with the funder. Treat the program as a learning system, not a compliance exercise.