How to Ask for the Evidence Behind an ABA Recommendation

Start with the exact recommendation

Ask for the service, goal, setting, procedure, schedule or intensity, responsible role, start point, review date, and rationale. A broad statement such as “more therapy is better” cannot show how a proposal fits this person. Clarify which parts are clinical recommendations and which reflect coverage, capacity, or scheduling.

The CASP public summary covers ABA treatment for people diagnosed with autism and points to licensed detailed guidance. Its population-level scope does not replace individualized review.

Ask what evidence was used

Request the assessment question, operational definitions, direct observations, interview sources, record review, health or access factors, baseline period, opportunity denominators, and relevant research. Ask what evidence was unavailable or conflicting. A score without its measure, date, informant, and interpretation has limited use.

For covered professionals, the BACB Ethics Code addresses assessment, client-informed goals, intervention selection, risk, data evaluation, referrals, and documentation.

Include the person's priorities and access

Ask how the person participated, which goals matter to them, and how assent, dissent, fatigue, pain, communication, culture, school, other care, travel, and family time shaped the proposal. Keep AAC available throughout the discussion. ASHA guidance says users should always have access to their tools or devices.

Compare alternatives and uncertainty

Ask what other options were considered, including ordinary supports, environmental changes, medical or interdisciplinary referral, another setting, a different schedule, or no immediate change. Request the expected benefit, likely burden, known risks, uncertainty, and evidence that would lead the clinician to revise the recommendation.

Review one recommendation at a time

Noah's family asks eight evidence questions about a schedule change. Six receive source-linked answers, one needs updated direct observation, and one concerns a payer rule rather than clinical evidence. Evidence-answer completeness is 6 of 8 questions. The other two move to the correct owners.

Convert the recommendation into testable parts

Write the recommendation as a concrete proposal: service, goal, method, setting, frequency, duration, responsible role, and review date. Ask which part is based on direct assessment, record review, client report, family report, professional judgment, payer criteria, organizational capacity, or another source. This prevents one confident sentence from hiding several different claims.

Request the relevant date range and comparison. A percentage needs a numerator, denominator, opportunity definition, and context. A score needs the instrument version, qualified interpreter, limitations, and reason it answers the referral question. A narrative observation should say when, where, with whom, and under what supports it occurred.

Ask for evidence in layers

Use questions such as:

• What exact problem or desired outcome does this recommendation address?
• Which client priorities, preferences, assent, dissent, and access needs informed it?
• What direct and indirect evidence supports it, and what evidence conflicts?
• Which alternatives were considered, with what benefits, burdens, and risks?
• What assumptions remain uncertain, and what would change the recommendation?
• How and when will the team measure benefit, unwanted effects, feasibility, and fit?

The answer may involve records the family can request through an applicable access process, but a record copy alone may not explain the clinician's reasoning. Ask for an understandable explanation linked to the evidence. Keep AAC and other communication supports available so the person receiving services can question the proposal directly.

Check fit and authority separately

A qualified clinician should own the clinical interpretation within scope. The payer may make a separate coverage determination. The family or legally authorized person handles required consent, while the client's assent and withdrawal remain relevant when applicable. Operations can confirm staffing and schedule feasibility. Record these decisions separately so coverage, capacity, and clinical appropriateness do not masquerade as one another.

Look for burdens beyond session hours. Travel, preparation, missed school, caregiver participation, sensory cost, communication effort, other care, rest, and recovery can change whether a recommendation is workable. Feasibility information can justify redesign even when a procedure has supporting evidence in another context.

Work through a recommendation example

Noah's team recommends increasing weekly hours after a progress review. The initial explanation cites “slow progress,” but the family sees that the last eight weeks include two staff transitions, three cancelled sessions, and inconsistent AAC availability. The family asks for the eligible sessions, defined measures, actual delivered hours, and the evidence connecting added time to the proposed goals.

The clinician separates lack of opportunity from performance within completed sessions, reviews Noah's priorities, and compares three options. One adds hours, one stabilizes staffing first, and one narrows goals while communication access is repaired. The recommendation is revised to a four-week stabilization period with a predeclared review. The review measures twelve scheduled sessions, AAC available in 11 of 12, and assigned staff present in 10 of 12. Those process measures inform the next decision without proving a treatment effect.

Record the disposition and revisit it

Ask for a written response that identifies accepted, modified, declined, and still-open parts of the recommendation. Preserve the original recommendation and subsequent revision with dates and authors. State what remains in effect and which urgent safety or medical routes operate independently.

At the review date, compare the promised conditions with what occurred. Examine outcomes, burden, access, preference, unwanted effects, and missing data. A strong evidence discussion can end with uncertainty or a different plan. Its value lies in making the reasoning inspectable and responsive to the person, rather than manufacturing certainty from incomplete evidence.

Distinguish evidence strength from relevance

Evidence can be carefully collected and still answer the wrong question. A clinic observation may show performance in a quiet room without explaining community access. A caregiver report may describe a meaningful daily pattern without establishing behavioral function. A standardized score may compare performance with a norm group while offering limited guidance about one preferred goal. Ask what inference each source supports and where that inference stops.

Check whether the evidence reflects the conditions the recommendation is supposed to change. If the proposal concerns independent communication, ask whether AAC was available, partners recognized the person's messages, and eligible opportunities were defined. If it concerns safety, ask whether health conditions, pain, environmental hazards, and emergency routes were assessed by the appropriate roles. Missing conditions can make a neat percentage misleading.

Ask the clinician to explain causal uncertainty. A before-and-after difference may follow several simultaneous changes. A small sample can be useful for planning while remaining too limited for a confident conclusion. A lack of measured improvement may reflect an ineffective intervention, poor implementation, insufficient opportunity, inaccessible measurement, or a goal that no longer fits. The recommendation should state how the team will distinguish these possibilities.

Use a one-page evidence matrix with columns for source, date, setting, measure, finding, limitation, and decision impact. Add the person's and family's priorities as evidence, while labeling them accurately rather than converting preference into a clinical score. Mark stale or conflicting sources for review. The matrix should help discussion, not automate a clinical answer.

Before agreeing to a recommendation, ask for a stop or review rule. Examples include unexpected distress, loss of communication access, repeated implementation failure, new medical information, family burden beyond the planned range, or no meaningful progress within the stated period. A recommendation is more trustworthy when it explains what could prove it wrong and how the team will respond.