An ABA clinical decision rule worksheet should show the question the team is trying to answer, the evidence that must be available before review, and who owns the final decision. It should not turn a number into an automatic treatment change. The copyable form below keeps the planned rule beside the actual data, visual review, implementation conditions, client feedback, exceptions, and signed phase-change record.

Clinicians & ABA Professionals / Data, Outcomes and Clinical Decision-Making.

That separation matters. A prospective rule can reduce hindsight bias and make a review easier to reconstruct. It cannot decide whether a goal still matters to the client, whether the measure represents the relevant behavior, whether implementation was safe and faithful, or whether another explanation fits the pattern. A qualified reviewer still has to work through those questions with the client and relevant stakeholders.

What this worksheet is for

Use this artifact when a qualified reviewer wants to state in advance what will open a clinical review and later document what was considered. It can support a treatment-planning discussion, a supervision review, or a quality check on a phase transition. It is not a stand-alone treatment plan, graphing system, consent record, safety plan, research protocol, authorization request, or payer rule.

The BCBA Test Content Outline, Sixth Edition identifies measurement validity and reliability, representative measurement, graphing, graph interpretation, and critique of single-case designs as relevant entry-level BCBA knowledge. The BACB test-content outline page identifies the current outline. Examination content describes professional knowledge; it does not endorse this form or set a required phase length.

The current Ethics Code for Behavior Analysts, reached through the BACB's ethics-code access page, addresses client responsibility, appropriate assessment, data collection, continual evaluation, documentation, and stakeholder involvement. Exact duties also depend on competence, licensure, law, contracts, organizational policy, and the facts of the case. The CASP ABA Practice Guidelines Version 3 access page describes a licensed practice-guideline resource for autism treatment. This page neither reproduces that licensed text nor claims that the worksheet is a CASP form.

A trigger should start a review, not end one

A useful rule has two parts. The first says when the record is ready for review. The second states that a named qualified person will interpret the evidence and decide what happens next. For example, “Open a clinical review after four consecutive eligible observations at or below the demonstration boundary, provided the measurement definition and phase label are current.” That is very different from “Change treatment after four low points.”

The rule should also say what blocks or redirects review. Missing opportunities, an altered definition, a device failure, an illness, a major context change, poor procedural integrity, new distress, a withdrawn willingness to participate, or an unresolved safety concern can change what the data mean. A blocked trigger is not a failed client response. It is a signal that the record needs another kind of attention.

Research and reporting guidance helps explain why these distinctions belong in the record. A review of single-case design, analysis, and quality assessment describes examination of level, trend, variability, immediacy, overlap, and consistency, alongside design quality. A paper on systematic protocols for visual analysis discusses efforts to make visual review more explicit while warning that weak design or missing integrity and reliability evidence can make a structured result misleading. These dimensions are prompts for judgment, not a universal scoring formula.

The SCRIBE 2016 statement asks single-case researchers to report phase sequences, whether changes were planned or data driven, phase-change criteria when applicable, procedural changes, analyses, raw outcomes, adverse events, and limitations. Clinical service documentation is not automatically research, but that transparency model is useful: preserve what was planned, what actually happened, and why the reviewer departed from the plan.

Copyable decision-rule contract

Complete this section before anyone starts looking for the preferred result. Protected health information belongs in the team's governed record, not in an uncontrolled copy of the worksheet.

FieldEntryClient-selected purpose and why it matters nowDirect client input and preferred communication methodCaregiver or other stakeholder input, kept separatelyClinical question this rule can help reviewResponse or outcome definition and versionUnit, denominator, opportunity, and observation windowPhase labels and intervention versionMinimum eligible information before reviewProspective review triggerConditions that block, pause, or redirect the triggerRequired visual-analysis viewsRequired measurement-reliability and integrity evidenceAssent, access, distress, safety, and contextual checksQualified decision owner and backupPeople whose input must be soughtRule version, author, approval date, and next review date

Write the trigger in observable terms. “Data look better” is not reconstructable. “Open review after four consecutive eligible daily observations at or below 4, with the same response definition and no unresolved integrity or safety flag” is testable. The number 4 and the sequence length are placeholders selected for a fictional example, not recommended clinical values.

Session and condition ledger

The ledger preserves the evidence that produced a trigger state. Keep raw or source data in the governed clinical record and use references here when duplication would increase privacy or transcription risk.

Date or sequencePhase and versionValue and unitEligible?Definition and measure versionIntegrity or reliability evidenceClient/access/context noteException codeSource-record referenceyes / no / unclearyes / no / unclearyes / no / unclearyes / no / unclearyes / no / unclearyes / no / unclear

An exception code needs a definition. “NO-OPP” might mean the planned opportunity never occurred; “INVALID-TIME” might mean the measurement device failed; “DEF-V2” might indicate a definition change; and “CTX” might flag a material setting change. Do not silently enter any of these as zero. Record whether the observation remains usable for the specific question and who made that determination.

Visual and contextual review panel

The graph should retain raw points, phase lines, measurement units, dates or sequences, and annotations for material changes. Reviewers can then describe the pattern without pretending that a checklist has established a functional relation.

Review viewWhat is visible?Interpretation and uncertaintyFollow-up neededWithin-phase levelWithin-phase trendWithin-phase variabilityChange in level across phasesChange in trend across phasesImmediacy or latency of changeOverlap across adjacent phasesConsistency across comparable contrastsData density and missingnessMeasurement reliability or observer agreementProcedural integrity, dosage, and exposureClient experience, access, burden, and unanticipated effects

“Immediate” should not be a casual label. Research on defining and assessing immediacy shows that the construct and its operationalization can vary. A review comparing visual and statistical analysis in single-case studies also illustrates that analytic methods can disagree. State what the reviewer examined, what remained ambiguous, and what the design can support.

Exception and competing-explanation log

Observation or periodWhat changed?Could it affect the measure, intervention, or opportunity?Immediate actionOwnerResolved before decision?yes / noyes / noyes / no

Common entries may include a revised definition, schedule disruption, medication or health change reported through the appropriate channel, staffing change, altered materials, new communication support, low implementation integrity, a shift in opportunities, or a client request. The presence of a possible competing explanation does not itself prove what caused the pattern. It prevents the record from implying that nothing else changed.

Phase-change review record

Decision fieldEntryTrigger state: met, not met, blocked, or not applicableEvidence window reviewedGraph version and source referencesClient input and communication accessCaregiver and other stakeholder input, separately attributedIntegrity, reliability, safety, and context findingsReasonable alternative explanationsDecision: continue, modify, pause, gather more information, consult, refer, or otherRationale and uncertaintyChanges made, with treatment-plan/version referenceMonitoring and reconsideration dateQualified reviewer, credentials, signature, and date

The form should point to the approved treatment-plan change rather than becoming that plan. If the decision falls outside the reviewer's competence or authority, the record should identify consultation, referral, escalation, or deferral instead of forcing a choice from the menu.

Fictional worked example

The following numbers exist only to demonstrate arithmetic and documentation. They are not a client record, recommended measure, phase length, decision threshold, or treatment claim.

A fictional team tracks a count during daily-living practice. The baseline values are 7, 6, 8, 5, 7, and 6. They total 39, so the mean is 39 / 6 = 6.5. Sorted values are 5, 6, 6, 7, 7, and 8, so the median is the average of the third and fourth values: (6 + 7) / 2 = 6.5.

The comparison-phase values are 5, 4, 4, 3, 3, and 2. They total 21. The mean is 21 / 6 = 3.5. Sorted values are 2, 3, 3, 4, 4, and 5, so the median is (3 + 4) / 2 = 3.5.

PhaseValuesSumMeanMedianRangeBaseline7, 6, 8, 5, 7, 6396.56.55 to 8Comparison5, 4, 4, 3, 3, 2213.53.52 to 5

The fictional prospective rule says to open review after four consecutive eligible values at or below 4, if the definition and phase version remain unchanged and no unresolved access, integrity, safety, or context flag exists. The last four comparison values are 4, 3, 3, and 2. All four are at or below 4, so the record says “trigger met; qualified review opened.”

It does not say “treatment succeeded” or “change phases automatically.” A simple A-B contrast does not by itself demonstrate experimental control. The six-point summaries also hide sequence details, and no reliability, integrity, exposure, client-experience, or competing-explanation information has been supplied. The reviewer must inspect those fields and the graph before documenting a decision.

Reconsider the rule when the question changes

A rule should be versioned when the response definition, unit, denominator, observation window, intervention, opportunity structure, client priority, setting, data system, or decision purpose changes. Preserve the old rule and explain the transition. Retrospectively rewriting the rule to fit the observed data makes the record look more certain than it was.

Reconsideration is also appropriate when the rule repeatedly opens reviews that do not help, fails to notice meaningful client feedback, creates burdensome measurement, or rewards teams for producing a preferred number. A tool that is easy to complete can still be poorly matched to the clinical question.

Use the worksheet as an audit trail

The most useful result from this artifact is a transparent chain from the planned question to source data, exceptions, review, decision, and follow-up. A later clinician should be able to see what was known, what remained uncertain, and why the qualified reviewer acted.

That record supports accountability without overstating evidence. It also leaves room for the client to say that the goal, procedure, burden, or outcome no longer fits. When that happens, the correct next step is a real clinical and stakeholder conversation, not a more elaborate formula.

Related resources

Sources