The Fidelity Problem in a Stepped-Care Clinic

The current protocol's outcomes are below benchmark, but the audit also shows a plausible implementation explanation: 62% adherence is low, and low fidelity is a well-established moderator of outcome. Abandoning an evidence-based protocol because it was delivered poorly would be a category error - the problem may be the delivery, not the intervention. The 45% comorbidity rate is a legitimate signal that a standard protocol may need adaptation, but adaptation should be principled, not improvised. The proposed replacement has weak evidence: a single pre-post study of 22 participants with no control group cannot separate treatment effect from regression to the mean, natural remission, or expectancy; pre-post panic data are especially vulnerable because panic symptoms fluctuate. There is no evidence of durability, no replication, and a commercial interest. The defensible position: do not adopt the replacement on this evidence; do not abandon the incumbent either. Instead, run a fidelity improvement cycle - training, supervision, recorded session review - and re-audit, while developing a documented adaptation for the comorbid pain subgroup and evaluating it. If the replacement is to be piloted, do so as an evaluated pilot with a comparison and pre-specified outcomes, not as a de facto switch.

Address it directly and without blame, but firmly. First, clarify the governance position: a change in treatment pathway is a service-level decision requiring approval, and unilateral adoption creates inconsistent care, confounds the audit, and exposes the service to risk if the programme's claims are not supported. Second, hear the staff rationale - they may be responding to genuine unmet need in the chronic pain subgroup, which is useful clinical intelligence. Third, agree a transition: either they revert to the current protocol with fidelity support, or their cases are formally enrolled in a supervised pilot with consent, outcome measurement, and a defined review date. Fourth, communicate the decision to the whole team so the standard is consistent and nobody is singled out unfairly. Document the decision, the rationale, and the plan. Model the evidence-based reasoning you expect from staff: the point is not that innovation is forbidden, but that it must be evaluated rather than assumed.

A workable minimum: (1) Define the primary outcome (PDSS change) and a secondary functional measure, with pre-specified collection at baseline, mid, end, and three-month follow-up. (2) Fix fidelity measurement - independent rating of a random sample of recordings each quarter, with a target and a supervision response when adherence falls below it. (3) Stratify by comorbidity so the chronic pain subgroup is analysed separately rather than averaged away. (4) For any adapted or new protocol, use a concurrent comparison group - stepped-wedge or a matched usual-care cohort - rather than a single-arm pre-post design, because the pre-post design cannot answer the causal question. (5) Pre-register the analysis plan and stopping rules so the decision is not made post hoc on a favourable run of cases. (6) Report to the team quarterly. This design is feasible in a small service, distinguishes fidelity failure from intervention failure, and gives the manager a defensible basis for a decision rather than a marketing claim.