Impact of AI Care Summaries on Caregiver Comprehension

Impact of AI Care Summaries on Caregiver Comprehension of Hospital Course

The primary objective is to assess if an AI-generated plain language summary of an inpatient progress note can improve caregiver understanding of their child's hospitalization by measuring differences in Hospital Course Comprehension scores, based on their responses to a survey regarding their child's hospital course.

Secondary objectives include exploring the caregiver's perspective on the use of AI in clinical care, as well as 7-day and 30-day readmission rates, 7-day and 30-day emergency department (ED) re-utilization rates, length of stay, and environmental impact measures.

Study Overview

• Status: Recruiting
• Conditions: Patient Satisfaction; Health Literacy; Comprehension
• Intervention / Treatment: Other: AI Care Summary

Detailed Description

This randomized controlled trial evaluates whether AI-generated plain-language summaries of inpatient progress notes improve caregiver understanding of a child's hospitalization. The intervention uses a Children's National HIPAA-compliant large language model hosted within the institution's Azure environment to generate caregiver-facing summaries from attending-signed inpatient progress notes.

Eligible caregivers are randomized to receive either standard hospital communication alone or standard communication plus an attending-approved AI-generated care summary. Before distribution to caregivers, attending physicians review generated summaries for clinical appropriateness and accuracy.

Caregiver understanding of the hospitalization is assessed using a structured survey administered during the hospitalization. Responses are compared with information documented in the medical record to evaluate comprehension. Additional exploratory analyses evaluate implementation outcomes, caregiver perceptions of AI-assisted communication, physician acceptance of generated summaries, healthcare utilization outcomes following hospitalization, and operational measures related to AI summary generation.

Statistical analyses will compare outcomes between randomized groups while controlling for relevant covariates when appropriate.

• Study Type: Interventional
• Enrollment (Estimated): 100
• Phase: Not Applicable

Contacts and Locations

• Study Contact: Name: Kristen E Johnson, MD, MS; Phone Number: 571-235-6524; Email: kejohnson@childrensnational.org

Study Locations

• United States
  • District of Columbia
    • Washington D.C., District of Columbia, United States, 20010
      • Recruiting
      • Children's National Hospital
      • Contact: Kristen Johnson, MD, MS

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 18 years and older (Adult, Older Adult)
• Accepts Healthy Volunteers: No
• Genders Eligible for Study: All

Description

Inclusion Criteria: Caregiver of a patient 10 years and younger admitted to Children's National Hospital, whose preferred language is English

Exclusion Criteria:

• Caregivers of patients older than age 10 years
• Caregivers of patients not admitted to the Hospital Medicine service
• Caregivers of admitted patients whose preferred language is non-English
• Patients whose care summaries are deemed inaccurate by their attending physician and not eligible for sharing with their caregiver.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Health Services Research
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: Single

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
No Intervention: Standard Hospital Communication; Standard hospital communication
Experimental: Standard Hospital Communication Plus AI Care Summary; Attending-approved written AI Care Summary provided to caregiver	Other: AI Care Summary; Written AI-generated plain-language summary of the patient's hospital course, reviewed and approved by the attending physician and provided to the caregiver during the hospitalization.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Hospital Course Comprehension Score	Percentage of caregiver survey responses that align with information documented in the medical record (range 0%-100%; higher scores indicate greater caregiver understanding of the hospital course).	Within 24 hours after receipt of the AI care summary or completion of standard care communication, during the index hospitalization

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
LOS	Length of stay	Throughout study duration, on average one year
Readmission	7-day and 30-day readmission rates: readmissions within 7 or 30 days of the index hospitalization, when related to problems addressed during the initial hospitalization.	During study duration, on average one year
ED reutilization	7-day and 30-day ED re-utilization rates: return visits to the emergency department within 7 or 30 days for complaints related to the initial hospitalization.	7 days after index hospital discharge and 30 days after index hospital discharge, respectively
Tokens	Token burden: the number of tokens used per AI-generated care summary.	WIthin 1 hour following generation of each AI care summary (during index hospitalization)
Energy usage	Energy use: estimated average kilowatt-hour use per AI-generated summary.	WIthin 1 hour of generation of the AI care summary (during index hospitalization)
Acceptability: Physician	AI care summary rejection rate: the proportion of AI-generated summaries that attending physicians determine are not appropriate to share with caregivers.	Within 1 hour of attending physician review of each AI-generated summary, prior to caregiver distribution, during index hospitalization
Reason for Rejection	Reason for rejection: the primary reason an attending rejects a summary. One of five categorical options for rejection (contains sensitive information, contains factual errors, tone is not right, prefer not to provide written information, concerned may cause confusion); or category "other" with a free text response.	WIthin 1 hour of attending physician review of the AI care summary, during the index hospitalization
Caregiver-Rated Usefulness of AI Care Summary (5-Point Likert Scale)	Caregiver rating of how helpful the AI-generated care summary was for understanding the child's hospitalization, measured on a 5-point Likert scale ranging from 1 (Not Helpful at All) to 5 (Extremely Helpful). Higher scores indicate greater perceived usefulness.	Within 24 hours after receipt of the AI care summary, during the index hospitalization
Engagement: Caregiver	Caregiver interest in AI summaries as standard care: whether caregivers would want this type of summary incorporated into routine care.	Within 24 hours after receipt of the AI care summary, during the index hospitalization

Collaborators and Investigators

• Sponsor: Children's National Research Institute

Publications and helpful links

General Publications

• Greenberg JA, Basapur S, Quinn TV, Bulger JL, Schwartz NH, Oh SK, Ritz EM, Glover CM, Shah RC. Daily Written Care Summaries for Families of Critically Ill Patients: A Randomized Controlled Trial. Crit Care Med. 2022 Sep 1;50(9):1296-1305. doi: 10.1097/CCM.0000000000005583. Epub 2022 May 23.
• Bedi S, Liu Y, Orr-Ewing L, Dash D, Koyejo S, Callahan A, Fries JA, Wornow M, Swaminathan A, Lehmann LS, Hong HJ, Kashyap M, Chaurasia AR, Shah NR, Singh K, Tazbaz T, Milstein A, Pfeffer MA, Shah NH. Testing and Evaluation of Health Care Applications of Large Language Models: A Systematic Review. JAMA. 2025 Jan 28;333(4):319-328. doi: 10.1001/jama.2024.21700.
• Schoonbeek RC, Workum JD, Schuit SCE, Hoekman AH, Mehri T, Doornberg JN, van der Laan TP, Bootsma-Robroeks CMHHT; Applied Artificial Intelligence in Healthcare Consortium. Quality and efficiency of integrating customised large language model-generated summaries versus physician-written summaries: a validation study. BMJ Open. 2025 Sep 4;15(9):e099301. doi: 10.1136/bmjopen-2025-099301.
• Johnson AM, Brimhall AS, Johnson ET, Hodgson J, Didericksen K, Pye J, Harmon GJC, Sewell KB. A systematic review of the effectiveness of patient education through patient portals. JAMIA Open. 2023 Jan 18;6(1):ooac085. doi: 10.1093/jamiaopen/ooac085. eCollection 2023 Apr.
• Rea KE, Rao P, Hill E, Saylor KM, Cousino MK. Families' Experiences With Pediatric Family-Centered Rounds: A Systematic Review. Pediatrics. 2018 Mar;141(3):e20171883. doi: 10.1542/peds.2017-1883. Epub 2018 Feb 6.

Study record dates

Study Major Dates

• Study Start (Actual): May 5, 2026
• Primary Completion (Estimated): June 30, 2027
• Study Completion (Estimated): June 30, 2027

Study Registration Dates

• First Submitted: July 23, 2026
• First Submitted That Met QC Criteria: September 15, 2026
• First Posted (Actual): September 21, 2026

Study Record Updates

• Last Update Posted (Actual): September 21, 2026
• Last Update Submitted That Met QC Criteria: September 15, 2026
• Last Verified: September 1, 2026

More Information

Terms related to this study

• Keywords: communication; health literacy; plain language
• Additional Relevant MeSH Terms: Behavior; Treatment Adherence and Compliance; Health Behavior; Communication; Patient Satisfaction
• Other Study ID Numbers: STUDY00001841

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

IPD Plan Description

Our IRB protocol does not permit IPD sharing for subject privacy

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No