Developmental regression

Regression means loss of an established skill

A useful report names the response, prior use, observation window and opportunities, onset, current use, and comparable AAC and access conditions. “Stopped communicating” is too broad and does not identify a cause.

Several patterns can look similar:

A brief gap or marked reduction does not prove regression. Compare definitions, opportunities, prompting, access, and health while contacting the medical clinician; do not delay referral for more baseline. Patterns can coexist, and none identifies a cause.

A 2026 study proposed a working definition

A July 2026 two-round international Delphi study by Gawade and colleagues proposed a working definition: skill loss during childhood lasting at least four weeks and occurring in at least one of six domains, namely verbal communication, nonverbal communication, functional motor skills, cognition, social skills, or self-help. Seventy-six clinicians completed round one; 41 of 76 (53.9%) completed round two. The study set consensus at 70% or more. In round two, 35 of 41 (85.4%) agreed on loss in one or more domains, and 30 of 41 (73.2%) agreed that verbal skill loss lasting four or more weeks should count. The authors applied the four-week requirement to skill loss generally in the published working definition.

This broad proposed threshold is not a validated diagnostic test, causal classification, or triage rule. Round-one respondents included 48 of 76 (63.2%) from Australia; all represented countries were high income. Lived-experience perspectives were sought during study design; survey respondents were clinicians. The panel did not reach consensus on required prior duration or frequency of skill use or whether stasis, slowing, or fluctuation belongs in the definition. The authors call for testing and refinement. Do not wait four weeks before seeking medical advice.

Autism and regression are related but distinct

The current CDC autism overview says some autistic children gain skills and meet milestones until around 18 to 24 months, then stop gaining or lose skills. Many autistic children have no reported regression, and regression alone neither establishes nor rules out autism.

The 2020 AAP autism clinical report says medical evaluation of lost milestones is indicated and the underlying processes remain unclear. It says the usual 18- to 24-month language or social regression pattern in autistic children is unlikely to be attributable to seizures or neurodegenerative disorders. It treats loss after age two, motor loss, repeated regressions, or seizure and neurological concerns as atypical features that can change specialist evaluation. Population findings cannot determine one child's cause.

The CDC autism FAQ distinguishes regression, meaning skill loss, from regressive encephalopathy, meaning a brain disease or disorder causes the loss. Most autistic children have not had encephalopathy, though some have regression and some have both. An autism diagnosis or ABA observation cannot establish encephalopathy; that requires medical evaluation.

Monitoring supports evaluation without delaying it

Write down what you know while contacting the responsible clinician:

• the exact skill or response form that changed
• prior frequency, independence, settings, and duration of established use
• last known use and the onset, pace, and pattern of loss
• languages, partners, demands, and ordinary supports
• AAC, hearing, vision, mobility, and sensory access
• sleep, eating, drinking, pain, illness, fever, injury, medication, or exposure changes
• seizures, staring episodes, weakness, balance changes, unusual movements, or altered awareness
• what the person directly communicates through speech, AAC, gesture, or movement, with other observable behavior recorded separately without inferred meaning

Do not delay urgent action for a full baseline. Existing records or notes may help. New video requires consent, privacy protection, and safe priorities.

The CDC milestone page tells families to act early after skill loss. The CDC clinical screening page distinguishes screening from assessment; a screen neither explains regression nor replaces medical evaluation.

Preserve communication and ordinary support

Loss of speech does not mean loss of all communication. Keep speech, sign, gesture, writing, pictures, movement, behavior, and AAC available and record what remains reliable. The ASHA AAC practice portal says AAC users should always have access to their tools or devices.

Avoid removing supports or creating test conditions to determine whether a skill is “really gone.” Maintain food, water, bathroom access, pain care, prescribed health care, mobility, sensory regulation, rest, relationships, communication, and safety. Never make basic access contingent on demonstrating a skill or participating in an assessment. Pause nonessential demands when pain, illness, new neurological signs, or distress makes participation unsafe.

The BACB Ethics Code requires behavior analysts to work within a defined role and competence and, when medical or biological variables may influence behavior, help ensure medical needs are assessed, document medical referrals, and follow up. ABA staff may record time-stamped observations, opportunities, supports, access conditions, and reported health or safety signs; follow supervisor and emergency policy; and share records under consent and privacy rules. They cannot diagnose regression, encephalopathy, or its cause, order or interpret medical tests, or use a behavioral function as a medical explanation.

A fictional report of skill loss

Ishan is a fictional five-year-old who uses speech and AAC. In 10 planned morning routines last month, each routine included the same opportunity to communicate, and he independently selected at least one of six familiar AAC messages before a prompt in 8 routines (8/10). In 10 routines this week, he did so in 1 (1/10). The numerator is routines with an independent selection; the denominator is routines containing the planned opportunity. The documented device, vocabulary, positioning, and ordinary prompting conditions appear unchanged.

His family also observes three new episodes of unsteady walking over two days. That is an event count within a stated observation window, not a fraction of undefined walking opportunities. The family contacts his pediatric clinician that day and follows the urgent instructions provided. They report the counts, dates, access conditions, movement change, recent health information, and what Ishan communicates. They do not wait four weeks, assign a behavioral function, remove AAC to test performance, or presume autism explains the change.

The two small communication samples document a difference; reported comparability may still miss relevant variables. Neither the communication counts nor the gait count establishes developmental regression, cause, severity, or prognosis. Qualified medical and developmental professionals determine the evaluation while Ishan's communication, basic access, and safety supports remain available.