Audit Treatment-Integrity Data Before Reporting

To audit treatment integrity data before reporting, validate client or cohort identity, procedure version, source, component, eligible opportunity, denominator, timestamps, observer, agreement, missingness, adaptations, corrections, access, exposure, and linkage to outcomes. Reconcile duplicates and preserve the original evidence. Hold unsupported rows. A polished dashboard cannot repair an ambiguous definition or mismatched denominator.

Validate identity and version

Match each row to the correct client or cohort, procedure, effective dates, setting, implementer, and source record.

Check stable identifiers rather than names alone and reconcile duplicates, merged records, late entries, and sessions that cross a version boundary. Confirm the procedure actually available to the implementer, including cached or printed copies. Hold rows with uncertain identity or version. A technically valid percentage attached to the wrong person or procedure is unusable and potentially harmful.

Compare the observation log with schedules, source records, and version distribution history using a documented matching rule. Flag reused identifiers, impossible overlaps, and records outside the effective period. Limit access to identifiable records according to the report's purpose. When a correction changes cohort membership, preserve both the original and revised counts with the reason.

Recalculate the denominator

Confirm which components and opportunities were due and which states were missing, invalid, adapted, or inapplicable. Recompute from raw counts when possible.

Apply the rules in effect at the observation date and preserve version-specific denominators. Distinguish not observed from not due, and retain client-declined or inaccessible opportunities according to the defined measure. Compare component totals with session summaries and investigate impossible values. Never reconstruct missing raw counts from a displayed percentage when more than one denominator could produce it.

Recalculate a reproducible sample and every outlier before trusting automated summaries. Check conditional components, sessions with zero opportunities, partial observations, and rounding. Document how adapted components are treated. If raw records are insufficient, hold the calculation and state the limitation rather than replacing it with a plausible denominator selected to fit the published value.

Check observation and correction history

Verify observer, time, mode, agreement sample, original value, amendment, reason, author, and date. Preserve the source and audit trail.

Review whether observation consent, privacy, visibility, and observer competence matched the intended use. Link agreement data to the exact observations sampled. Corrections should append a reason and responsible author without erasing the original. Flag unexplained late edits, copied timestamps, or bulk changes for review. Separate data-quality findings from conclusions about clinical implementation.

Release with a hold ledger

Report valid rows separately from held records. Give each hold a reason, owner, age, repair, and decision about whether it can enter a later report.

Show the number and relevant denominator of held observations by cause, version, setting, and period so omission is visible. State how holds affect interpretation and prohibit downstream ranking or decision uses the evidence cannot support. Require independent verification before release from hold. Keep unresolved critical client concerns on their responsible safety or clinical path while reporting remains paused.

Put the pre-report audit into practice

Wen's audit finds two duplicated observations, three rows tied to an expired procedure version, and four percentages without opportunity counts. Those rows remain held. The final dashboard reports valid records, holds, missingness, settings, sample coverage, client outcomes, and limitations instead of silently dropping the problems.

Keep supervision, quality, and employment roles separate for Wen

Wen's pre-report integrity data audit identifies the clinical decision owner, observer, supervisor, quality reviewer, operations owner, privacy contact, and employment owner. State which process is active, what notice and evidence it uses, who may decide, and where the record belongs. One integrity observation should not silently change purpose after collection.

Protect client access during Wen's review

Wen's process keeps AAC, interpreters, mobility, food, water, bathroom use, prescribed care, health support, rest, relationships, and emergency help available. Obtain required consent and assent when applicable and follow the governing response to withdrawal or distress. Client protection can require immediate action while broader measurement and personnel conclusions remain pending.

Use raw evidence and matched denominators for Wen

Report Wen's correctly implemented components divided by eligible components due, observations completed divided by observations due, agreement pairs divided by pairs due, and repairs verified divided by repairs due. Keep procedure versions, critical steps, exposure, client outcomes, direct experience, access, missingness, adaptations, drift, burden, and adverse effects in separate series.

Ask six supervision and review questions for Wen

Use these questions in the pre-report integrity data audit:

Which client decision, procedure version, component, and eligible opportunity apply?
Which people, settings, risks, supports, observation modes, and sample limits matter?
Which integrity, agreement, exposure, outcome, experience, access, and adverse-effect series stay separate?
Which skill, clarity, resource, workflow, safety, burden, supervision, or plan-fit barrier is supported?
Which clinical, quality, operations, privacy, employment, certification, or payer owner may act?
Which repair, stop rule, follow-up evidence, and review date close the loop?

Keep unresolved evidence visible with a state, owner, age, and next action.

A fictional supervision example for Wen

Wen is fictional and involved in an organizational dashboard combining home, center, and community observations. Reviewers freeze 41 identity, version, source, component, opportunity, time, observer, agreement, missing, adaptation, correction, access, exposure, and linkage fields and complete 29 of 41 by the checkpoint. Missing, unobservable, inapplicable, adapted, drifted, disputed, failed, and pending fields keep their defined states.

The pre-report integrity data audit measures evidence completeness. It does not establish effectiveness, safety, acceptability, competence, employment performance, certification compliance, medical necessity, authorization, payment, generalization, maintenance, or causation.

Use current sources within their scope for Wen

For Wen's pre-report integrity data audit, the CASP public summary supplies high-level individualized assessment, implementation, and evaluation scope for ABA treatment of autistic people. The BACB Ethics Code addresses competence, client involvement, consent and assent when applicable, supervision, risk, documentation, and evaluation for covered behavior analysts. The BACB supervision page routes readers to role-specific supervision and training resources; those requirements do not replace licensure, payer, employment, or case authority.

For Wen, the BCBA Test Content Outline addresses procedural fidelity as examination content; case protocols require qualified case-specific authority. The Ferguson practitioner guide supports observable components and eligible opportunities. The Essig, Rotta, and Poling review supports caution in fidelity-report interpretation. ASHA says AAC users should always have access to their tools or devices. Set coaching triggers, comparison rules, and performance thresholds from the current case and setting.

Close Wen's review

Ask Wen, the implementer, and the responsible clinician to review the pre-report integrity data audit through accessible communication. Record the evidence, perspectives, limitations, selected repair, responsible system, clinical decision, follow-up observation, client outcome and experience, and review date. Reopen the analysis when the procedure, client priority, access, health, setting, implementer, supervisor, risk, or outcome changes.

How to Audit Treatment-Integrity Data Before Reporting