How Should Repeated ABA Safety Events Be Reviewed?

Review patterns with fixed measures

Define an eligible event and exposure unit before comparing periods. Report count, severity, injuries, near misses, client communication, setting, staff configuration, health or medication changes, AAC availability, service hours, and missing data. Segment categories before combining them.

Keep emergency and communication routes clear

The ASHA AAC portal says AAC users should always have access to their communication tools or devices. Preserve a reliable way to report pain, danger, stop, help, and what happened.

SAMHSA directs anyone in danger or having a medical emergency in the United States to call 911 or go to the nearest emergency room. Immediate care comes before routine review work.

Separate records, review, and decision authority

The CASP public summary frames assessment, planning, implementation, and evaluation within its autism-treatment scope. The BACB Ethics Code addresses competence, risk, client involvement, documentation, and data-based evaluation for covered behavior analysts.

For a HIPAA covered entity, HHS access guidance and 45 CFR 164.526 create distinct access and amendment routes for records in scope. Other records, laws, and internal correction processes can follow different rules.

A practical example

A practice records four elopement events across 20 community arrivals, then three across 40 later arrivals. The count falls by one and the rate changes from 20% to 7.5%. Route, staffing, and AAC also changed, so the review avoids a causal claim.

Define the event and exposure before counting

Write an observable event definition that different reviewers can apply. Define the exposure unit that creates a real opportunity for the event, such as a community arrival, transport leg, session hour, meal, or medication administration. State the reporting period and which records were eligible.

Keep near misses, injuries, restrictive actions, emergency calls, and other event types in separate fields before combining them. A change in label can make the trend look better or worse without changing actual safety. Preserve records that were missing or excluded and explain why.

Compare rates, severity, and context together

Counts alone can mislead when service volume changes. Four events across 20 arrivals is 20%. Three across 40 is 7.5%. Report both the count and rate, plus the numerator and denominator. A lower rate can coexist with a serious injury, so include severity, health outcome, emergency response, and service disruption.

Segment by setting, route, time, staff configuration, communication access, health change, medication, activity, weather, transportation, and other relevant conditions. Avoid slicing the data so narrowly that individual people can be identified unnecessarily or the sample becomes meaningless.

Look for system and partner conditions

Review whether AAC was available, the planned route was open, staff assignments were clear, supervision occurred, health and mobility instructions were current, and the person's stop or help message was honored. A repeated event may reflect a combination of environmental, clinical, operational, and access conditions.

Do not assume recurrence proves the person's behavior worsened. Schedule changes, unfamiliar staff, an inaccessible setting, pain, missing equipment, or a changed counting rule may contribute. The appropriately qualified clinician makes clinical interpretations, while other qualified roles review their domains.

Include the person's account across events

Offer an accessible way for the client to describe what happened, what felt unsafe, and what helped. Compare statements over time without treating consistency as a test of truth. The person may communicate more after recovery, with a different partner, or through another mode.

Record AAC, interpreter, support, question wording, and timing. A caregiver or staff member can provide context without replacing the client's account. If the person declines to participate, respect that choice and continue reviewing other evidence.

Use an action register that can be validated

For each corrective action, name the risk or condition, owner, due date, affected setting, evidence of completion, and validation measure. “Retrain staff” is incomplete without the specific skill, participants, competency evidence, and follow-up observation. An environmental change should be inspected where it will be used.

Keep open, overdue, failed, and inapplicable actions visible. If an event recurs, ask whether the action was absent, implemented differently, insufficient, or unrelated. Reopen the review when new evidence changes the earlier conclusion.

Work through the arrival data

In the first period, four defined roadway-entry events occur across 20 eligible arrivals. In the second, three occur across 40. The practice reports 4/20 and 3/40 rather than saying only that events decreased. It also records that the route changed, two staff roles were reassigned, and AAC availability improved from 14/20 arrivals to 39/40.

The review cannot isolate which change affected the event rate. It does show that one roadway event and one missing-AAC arrival still occurred. The practice keeps the route and communication checks active, investigates the remaining failures, and schedules another review after a predeclared number of arrivals rather than waiting for a serious injury.

Decide when outside review is needed

Repeated injuries, restrictive actions, missed reporting, unexplained clusters, medical concerns, possible abuse or neglect, or conflicts of interest may require escalation beyond the immediate team. Follow emergency and mandated-reporting duties without waiting for an internal review. Use qualified clinical, medical, privacy, legal, licensing, payer, insurer, or safeguarding expertise as the facts require.

Tell the family which service decisions are immediate, which analysis remains open, and how the person will be protected while the review continues. A repeated-event trend should drive timely action, not become a reason to normalize unsafe conditions.

Do not let a favorable average hide a serious event

Aggregate rates can improve while one subgroup, setting, or severity category worsens. Review injuries, emergency calls, restrictive actions, missed communication, and near misses separately. A single high-severity event may require immediate action even when the overall rate is lower.

Keep the person-level review connected to the system trend. The family should know what protects this client now, which service is held or released, and who made that decision. Population data can identify patterns, while it cannot replace a case-specific clinical, medical, safety, or legal review.

Predeclare the next review period and stop conditions. If a serious injury, missing safeguard, or reporting failure occurs, reopen the review immediately rather than waiting for the planned denominator. Report both the trend and the exceptions in plain language.

Document it.

Questions families can use

Ask whether definitions stayed fixed, what denominator fits, how severity changed, which contexts cluster, what the client reports, which health or access issues were checked, what action started when, and what evidence supports continuation or revision.