NCT04705363

A Sociolinguistic-enabled Web Application to Develop Precision Health Intervention for African Americans

• Status: Completed
• Phase: N/A
• Study type: Interventional
• Enrollment: 751 (actual)
• Sponsor: University of Florida · Academic / Other
• Sex: All
• Age: 45 Years – 75 Years
• Healthy volunteers: Not accepted

Summary

This pilot study will explore the preliminary efficacy of a colorectal cancer (CRC) screening intervention delivered by Virtual Human Agents (VHAs). Seven hundred fifty participants aged 45 to 75 will be recruited through Qualtrics panels. The study examines how different levels of dialectal linguistic features willingness to be screened for colorectal cancer. Participants will be randomly assigned to interact with one of four VHA conditions: a VHA using non-dialectal linguistic features, a VHA with a low level of dialectal linguistic features integrated, a VHA with a high level of dialectal linguistic features integrated, or a text-only control condition. Following the interaction, participants will complete survey measures to assess perceived willingness to be screened.

Detailed description

African Americans experience significant health inequities, including higher morbidity and mortality rates due to colorectal cancer (CRC) compared to White Americans. While the causes of these disparities are complex, regular screening can help reduce them. However, adherence to CRC screening guidelines remains low, particularly among African Americans. One strategy to reduce CRC screening disparities is using strategic communication interventions to promote the fecal immunochemical test (FIT). FIT is a low-cost, non-invasive screening method that alleviates common patient barriers to CRC screening and is as effective as colonoscopy in reducing CRC incidence and mortality. Tailored messaging interventions have been shown to improve CRC screening rates. However, two critical questions must be addressed before implementing tailored screening interventions within healthcare systems: (1) To what extent must message content be tailored to be effective? and (2) How can participants be effectively engaged? This study builds upon an existing project that utilizes mobile Virtual Human Technology (VHT) to deliver tailored CRC screening messages. Virtual Human Agents (VHAs) provide a unique opportunity to customize communication strategies, including linguistic adaptation, to align with patient preferences. Such interventions can help mitigate CRC screening barriers such as cultural mismatch and low self-efficacy. This study investigates explicitly the role of dialectal linguistic features in shaping willingness to be screened for CRC. The pilot study is exploratory in nature and seeks to examine the following aim: To assess how tailoring the dialectal variety of VHA speech affects willingness to be screened for CRC. We aim to recruit 750 participants, each of whom will interact with a VHA that varies in speech style across four conditions: (1) non-dialectal linguistic features, (2) a low-level integration of dialectal linguistic features, (3) a high-level integration of dialectal linguistic features, or (4) a voiceless, text-only control. Following the interaction, participants will assess the VHA's credibility using survey-based measures.

Conditions

• Colorectal Cancer

Interventions

Type	Name	Description
BEHAVIORAL	High Dialectal	A virtual health assistant that will consist of an interactive computer-generated doctor with voice that used high dialectal variation that will guide participants through the interaction.
BEHAVIORAL	Low Dialectal	A virtual health assistant that will consist of an interactive computer-generated doctor with voice that used low dialectal variation that will guide participants through the interaction.
BEHAVIORAL	Non-Dialectal	A virtual health assistant that will consist of an interactive computer-generated doctor with voice that used no dialectal variation that will guide participants through the interaction.
BEHAVIORAL	Text-only	A virtual health assistant that will consist of photos of the computer-generated doctor with text that will guide participants through the interaction. No voice will accompany the photos or text.

Timeline

• Start date: 2021-06-29
• Primary completion: 2023-12-18
• Completion: 2023-12-18
• First posted: 2021-01-12
• Last updated: 2025-04-18
• Results posted: 2025-04-18

Locations

1 site across 1 country: United States