IHA Health Literacy Conference • May 12–14, 2026

Stress-Testing a Voice-Ready Pediatric Obesity Intake Workflow to Reduce Health Literacy Burden in Safety-Net Care

Vikki S. Small, MPH, LSSGB, CSM

DrPH Student • SUNY Downstate School of Public Health • Brooklyn, NY

Many clinical intake forms were never designed for the families completing them

Black and Latinx families in safety-net pediatric clinics often struggle to complete written intake forms. This can be because of reading level, time pressure, or not having easy access to digital tools.

Nearly 1 in 4 non-Hispanic Black adults has below basic health literacy, compared to 9% of non-Hispanic White adults — and federal data show that Black and Hispanic adults use health information technology less often than their White peers.

These gaps are not about families' abilities. They reflect structural inequities in how health tools are designed and who they are designed for. In pediatric obesity care, these barriers can lead to incomplete intake forms and more time spent during visits correcting paperwork rather than focusing on counseling and care planning (James et al., 2023).

Designing an intake workflow for how caregivers actually communicate

Working with Live Light Live Right (LLLR), a community-based pediatric obesity program in Brooklyn, we redesigned 19 intake questions. We aimed for a 6th grade reading level and adapted every question so it could be used in a conversational, voice-based tool.

We checked plain language quality using the CDC Clear Communication Index (CDC, 2019) and got a score of 94 out of 100. The scenarios were based on published qualitative research that describes how Brooklyn families in weight management care talk about their lives, challenges, and communication needs (Browne et al., 2022).

The workflow was designed to match how caregivers actually speak. It accepts rough estimates, short stories, and everyday number phrases instead of demanding exact or perfectly formatted answers. The goal was a tool designed for the families who will use it, rather than an idealized user that health systems often assume they are serving.

A four-step testing cycle: Design → Simulate → Evaluate → Refine

Step 1 Design 19 questions rewritten in natural language at 6th grade level
Step 2 Simulate 31 scripted scenarios based on published qualitative research
Step 3 Evaluate Completion rate, rule fidelity, edge case identification
Step 4 Refine Targeted rule updates verified through re-testing

Testing was conducted using Vertex AI, a cloud-based platform that supports the development and evaluation of conversational workflows. The platform was used to run each scripted scenario, capture how the workflow responded to varied caregiver inputs, and identify where decision logic broke down.

No families participated, no patient data were used, and the workflow was not connected to any clinical system.

The testing framework itself is platform-agnostic. While Vertex AI was used for this proof-of-concept, the same scripted scenario testing approach can be implemented using a range of conversational development tools. This allows teams to identify design and communication issues early, helping ensure that literacy-sensitive intake tools function as intended before families encounter them in real clinical settings.

All 31 scenarios completed. Zero dead ends.

31 Scripted scenarios completed
19 Intake questions redesigned
100% Workflow completion, zero dead ends
95% Decision rule accuracy across simulated caregiver responses

All planned feasibility targets were met. We found two unusual situations, both related to how the workflow handled numbers, and neither reflected caregiver misunderstanding or a system failure.

In the first, the workflow accepted fractional numbers when a whole-number estimate was needed for data consistency. We fixed this by adding a short, plain-language follow-up question instead of rejecting the caregiver's answer. In the second, the workflow accepted very unlikely values without checking them. We fixed this by adding a gentle range check that asks the caregiver to confirm their answer, instead of showing an error.

After these changes and re-testing, the workflow stayed conversational and focused on the caregiver's experience throughout.

This framework is designed to travel

Literacy-aligned intake redesign is possible even in small, under-resourced safety-net clinics. It does not require a large IT team or expensive technology. Testing the workflow in advance with scripted scenarios is a practical, low-risk way to find and fix design problems before families ever see the tool.

Building questions that accept how caregivers naturally talk — estimates, short stories, and everyday language — can lower burden without losing the clinical details providers need.

This approach is built to travel: clinics can adapt it to other safety-net settings, other chronic conditions, and other groups whose usual ways of communicating have often been treated as a problem to fix instead of a design requirement to respect.

Interested in testing this framework in your clinic?

Interested in adapting this framework for your safety-net setting? Want to share tools, ask questions, or follow the pilot? Reach out directly.