How should health plans evaluate autism care outcomes?

Health plans evaluating autism care programs need to understand three things: whether members can access the service, whether the care is helping them make meaningful progress, and whether the available evidence supports the conclusions being drawn.

A reported improvement percentage rarely answers all three questions. Its meaning depends on who was assessed, what was measured, how the measurement was conducted, and how much time passed between assessments. A result from a small group of participants who completed every assessment, for example, may not represent the experience of everyone who enrolled.

Autism care outcomes also cover more than changes in specific behaviours or the number of treatment hours delivered. Depending on the individual and the program, they may include communication, daily living skills, participation, well-being, quality of life, and family experience.

For health plans, the task is to establish a consistent way to evaluate these outcomes without overlooking differences in members’ needs, treatment goals, and circumstances.

Alt text - Interactive neurodevelopmental therapy for young people

What counts as an autism care outcome?

A useful starting point is to separate the measures used to evaluate a program into three categories: structure, process, and outcomes. This framework, developed by health services researcher Avedis Donabedian, is widely used in healthcare quality measurement. The Agency for Healthcare Research and Quality (AHRQ) explains how each category answers a different question.

All three matter, but they should not be treated as interchangeable.

A program may have qualified staff, well-documented processes and high participation rates without yet demonstrating meaningful improvements in the outcomes it intends to address. Conversely, an individual may make progress while waiting long periods or experiencing inconsistent access to services.

Process measures can help explain whether care is being delivered as intended. They do not, on their own, establish that the care improved outcomes. AHRQ also notes that factors beyond the provider’s control can affect patient outcomes, which makes context important when comparing performance.

For health plans, this means evaluating the full picture rather than relying on one metric to represent program quality.

Alt text: Structure process and outcome measures in autism care

Which autism outcome measures should health plans consider?

The right autism outcome measures depend on the purpose of the program, the population it serves, and the changes it is designed to support. There is no single score that captures every aspect of an autistic person’s development or wellbeing.

The International Consortium for Health Outcomes Measurement (ICHOM) developed an autism-specific Standard Set of Patient-Centered Outcome Measures to support more consistent assessment across care settings. The published standard recommends nine outcomes across four domains and includes 15 risk-adjustment factors to help account for differences between people receiving care.

That is a useful starting point for health plans designing an evaluation framework. The next step is to connect each measure to a clear purpose.

Clinical and functional progress

Clinical measures can help track changes in the areas addressed by an intervention. Depending on the care plan, these may include adaptive functioning, communication, social responsiveness, or specific skills identified during assessment.

The important question is whether the selected measure can capture the type of change the program is intended to support.

For example, a diagnostic assessment may help establish whether a person meets diagnostic criteria, but that does not automatically make it the right instrument for tracking treatment-related change over a short period. A review of outcome measures for young children with autism discusses limitations in using some commonly used instruments to measure treatment response, including concerns about sensitivity to change and the purpose for which a measure was originally developed.

Health plans should therefore ask providers to explain why a particular measure was selected, what it is designed to assess, and whether it is appropriate for the population and evaluation period.

Daily functioning and participation

A change in a test score is easier to interpret when it is considered alongside its relevance to everyday life.

Depending on the individual’s goals, this might mean looking at changes in daily living skills, the ability to communicate needs, participation in activities, or performance of a specific functional task.

The measure should reflect the goal being evaluated. A program focused on a particular functional skill should not rely exclusively on a broad measure of autism-related characteristics to demonstrate progress in that skill.

This distinction also helps prevent health plans from comparing programs that measure different things under the same label of “improvement.”

Wellbeing and quality of life

Behavioral and clinical measures provide important information, but they may not capture every outcome that matters to an individual.

Catalight’s discussion of autism outcome measurement highlights this issue, including why reducing a behaviour does not necessarily indicate improved wellbeing. Some repetitive movements, for instance, may serve a self-regulatory purpose. Their reduction should not automatically be interpreted as a positive outcome without understanding the individual and the context.

Health plans should understand how the program defines meaningful progress and whether its chosen measures reflect outcomes relevant to the people receiving care.

This does not mean replacing clinical measures with a broad wellbeing score. It means avoiding the assumption that any one measure can represent the entire care experience.

Family and caregiver experience

Where relevant to the program, caregiver experience can provide additional context about the practical effects of care. Measures may examine caregiver stress, family functioning, or the experience of navigating and participating in services.

These measures should be reported separately from clinical outcomes. A change in caregiver stress, for example, should not be presented as proof of a change in a member’s clinical status.

The value comes from understanding how different dimensions of care relate to one another, not combining them into a single score without a clear rationale.

Five questions to ask before accepting reported autism treatment outcomes

Once the outcome domains are clear, health plans can assess whether a program’s evidence supports the claims being made.

1. Who was included in the evaluation?

The population matters because autism care programs may serve people with different ages, support needs, functional abilities, co-occurring conditions, and prior treatment experiences.

A result from a narrowly defined group should not automatically be applied to a broader membership population.

Ask for the number of participants enrolled, their relevant characteristics, the eligibility criteria, and the setting in which care was delivered. If the program is intended for a particular age group or level of support need, the evidence should make clear whether the evaluated population resembles that group.

2. What exactly was measured?

A percentage improvement is only meaningful when the underlying measure is understood.

Health plans should request the name of the assessment instrument, the outcome it measures, the person completing it, and the scoring method used. They should also distinguish between a standardized assessment, a clinician’s observation, a caregiver report, a program-specific task score, and an operational metric.

These sources can all be useful, but they do not carry the same meaning.

For example, a program’s internal task score may show that a participant performed better on a particular activity. It does not, by itself, establish a broader change in adaptive functioning or quality of life. That conclusion requires evidence appropriate to the broader claim.

3. What was the starting point?

Baseline data establish the starting position against which later results are assessed.

Without a clear baseline, a health plan cannot reliably determine how much change occurred during the evaluation period. It may also be difficult to distinguish improvement from differences in how participants were assessed or selected.

The baseline should be collected before the intervention period being evaluated, using a defined method. Reports should identify the assessment date, the intervention start date, and any relevant prior treatment.

Where possible, the same measure and collection method should be used at follow-up. If the method changes, the report should explain how that affects interpretation.

4. What happened to participants who did not complete follow-up?

This is an important question because reported results can look different depending on who remains in the analysis.

Suppose a program enrols 100 participants, but only 60 complete the follow-up assessment. If the program reports that 45 of those 60 improved, it can accurately say that 75% of the participants with completed follow-up assessments improved, assuming that is how improvement was defined.

It cannot automatically say that 75% of everyone enrolled improved. The result for the full enrolled group is not known for the 40 participants without follow-up data.

This example is illustrative, not a real program result.

Missing follow-up data do not necessarily invalidate an evaluation. However, the report should state how many people completed each assessment, how missing data were handled, and whether those who completed follow-up differed from those who did not.

A systematic review of autism assessment research has also noted that studies reporting only participants with both baseline and follow-up data can make it difficult to understand the effects of attrition.

For health plans, transparency about completion and attrition is essential to judging how representative the results may be.

5. What does the evidence allow us to conclude?

The strength of a conclusion depends on the study design, the measure, the sample, the duration of follow-up, and the availability of a comparison group.

A pre- and post-evaluation can show that a measured outcome changed over time. By itself, it may not establish that the program caused the change. Other factors, including maturation, concurrent services, changes in assessment, or differences in who completed follow-up, may also have contributed.

A comparison group, when appropriately designed, can help address some of these questions. Randomised controlled studies can provide stronger evidence of causal effects, although their design, execution, population and relevance to the intended setting still need to be assessed.

Health plans should also distinguish between a statistically significant result and one that is meaningful in practice. A change can be statistically detectable without necessarily representing a substantial or relevant improvement for the person receiving care.

The appropriate question is not simply whether a program reported improvement. It is whether the study design and data support the specific claim being made.

Alt text: Five checks before comparing autism care programs

Why access and participation belong in an autism care outcomes review

A program's outcomes cannot be interpreted fully without understanding who was able to access and participate in it.

Health plans should track the journey from referral through assessment, eligibility determination, onboarding, participation and follow-up. These stages help identify where members may be waiting, disengaging, or encountering barriers.

Useful operational measures include:

• Time from referral to initial assessment.

• Time from eligibility confirmation to program start.

• The proportion of referred members who begin care.

• Participation and completion rates over the defined evaluation period.

• The proportion of participants with complete baseline and follow-up assessments.

• Reasons for non-start, discontinuation or missing follow-up, where available.

These measures should be reported separately from clinical outcomes. Faster access is important, but it does not establish that a program is clinically effective. Similarly, high participation is useful information, but it does not prove that participants made meaningful progress.

The relationship between these measures is what makes them valuable. If a program reports strong outcomes among participants who complete care, the health plan should also understand how many members reached that stage and whether the reported results are representative of the broader eligible population.

Aetna’s 2026 discussion of autism care takes a similar payer-oriented approach, describing a quality framework that combines claims-based measures of structure and process with analysis of data beyond claims to develop more meaningful outcome measures.

For health plans, the practical implication is to evaluate access, delivery, and outcomes together while keeping each measure’s purpose clear.

What should an autism care outcomes report include?

A useful report should allow a reviewer to understand what was measured, how the result was calculated, and how confidently it can be interpreted.

At a minimum, health plans should request the following information.

Reports should present the underlying values alongside percentages wherever possible. If a program reports that 70% of participants improved, a reviewer should be able to identify the number of participants represented by that percentage, how improvement was defined, and how many enrolled participants were included in the analysis.

Results should also be separated by outcome domain rather than rolled into one overall improvement figure without a transparent method. Clinical progress, daily functioning, caregiver experience, and access are related, but they measure different aspects of care.

For aggregate reporting, health plans should also consider whether outcomes differ across relevant participant groups. Differences may reflect variation in needs, access, intervention exposure, or measurement completion. Any comparisons should be interpreted carefully, especially when subgroup sizes are small.

How should health plans evaluate autism care outcomes during a pilot?

A pilot allows a health plan to assess whether a program can be implemented as intended and whether its evaluation approach produces useful evidence. It should not begin with an assumed clinical or financial result.

Before enrolment, the health plan and program partner should agree on five elements.

1. Define the evaluation question. Identify the decision the pilot is meant to inform. Is the health plan assessing feasibility, access, participation, change in a specific outcome, or the suitability of the program for a particular member population? These are different questions and require different evidence.

2. Agree on the population and eligibility criteria. Document who may participate, how members will be referred, and how the enrolled group will be described in the final report.

3. Select the measures in advance. Choose measures that match the intended outcomes and are appropriate for the population and timeframe. Establish who will administer them, who will complete them, and how the data will be recorded.

4. Set the measurement schedule. Define baseline, interim, and follow-up assessments before the pilot begins. The schedule should reflect the measure and program being evaluated rather than using an arbitrary interval for every outcome.

5. Agree on reporting and decision criteria. Establish how participation, missing data, outcome changes, limitations and any comparison data will be reported. Decide in advance what evidence would support continuing, adapting, expanding or ending the pilot.

The review should then consider implementation and outcomes separately. A pilot may demonstrate that a service can reach eligible members and collect reliable data, while leaving questions about longer-term clinical effects unanswered. That is still useful evidence, provided the conclusions remain within the scope of what was measured.

Financial evaluation requires the same discipline. If a pilot includes a cost analysis, the report should specify which costs are included, the period being assessed, the comparator and the assumptions used. A modelled estimate should be labelled as a projection, not presented as observed savings. Likewise, changes in service use should not automatically be attributed to the program without an appropriate evaluation design.

The goal is to make a decision based on evidence that is relevant to the health plan’s population and the question it set out to answer.

Alt text: BrainyAct team discussing autism care program’s

Making autism care outcomes useful for health plan decisions

A strong evaluation framework makes it easier to distinguish between a program that delivers its planned activities and one that has demonstrated meaningful change in the outcomes it measures.

Health plans should look for clearly defined measures, appropriate baseline and follow-up assessments, transparent reporting, and enough information to understand who the results represent. They should also examine access and participation alongside clinical and functional outcomes, without treating any one of these as a substitute for the others.

Standardised frameworks such as ICHOM’s Autism Spectrum Disorder Standard Set can help establish a consistent starting point. They do not remove the need to check whether individual measures fit the program, population and intended decision.

For health plans considering newer models of neurodevelopmental support, the same principles apply. The evaluation should begin with a defined population, an agreed measurement plan and clear reporting expectations. Claims about clinical progress, program effectiveness or cost should follow the evidence collected, not precede it.

BrainyAct works in intelligent, interactive neurodevelopmental therapy for children and young adults ages 6 to 30. Health plans evaluating this or another care model can use these questions to establish what should be measured, how results should be reported, and what evidence is needed before making broader decisions. Looking at a new approach to autism care? Book a call with us to explore BrainyAct’s approach to interactive neurodevelopmental therapy.

FAQs

What outcomes should a health plan monitor in autism care?

Health plans should consider relevant clinical and functional outcomes, daily living and participation, wellbeing or quality of life, and caregiver or family experience where appropriate. Access, engagement and follow-up completion should also be tracked, but reported separately from clinical outcomes.

How can health plans compare autism care programs?

Start by checking whether the programs serve comparable populations and measure the same outcomes using appropriate methods. Review baseline data, follow-up periods, completion rates, study design and limitations before comparing reported improvement percentages.

What makes an autism outcome measure useful?

A useful measure assesses an outcome relevant to the care program, is appropriate for the population, and can be administered and interpreted consistently. Health plans should understand what the measure captures, who completes it, and whether it is suitable for tracking change over the period being evaluated.

Why does baseline data matter when evaluating autism care?

Baseline data establish the starting point for measuring change. Without them, it is difficult to determine how much progress occurred during the evaluation period or compare results across participants and programs.

How should health plans interpret reported improvement in autism treatment outcomes?

They should examine the outcome definition, measure, starting values, participant population, follow-up period, missing data, and study design. A reported change may demonstrate progress on a particular measure, but it does not automatically establish broader clinical benefit, causation, or cost savings.