Evaluating the Preferences and Tradeoffs of AI-based Electronic Consultations for Older Adults in Primary Care

University of Pennsylvania220 enrolled

Overview

Electronic consultations, or e-consults, let a primary care doctor request medical advice from a specialist without requiring the patient to attend a separate visit. Artificial intelligence (AI) systems may be able to provide this type of advice, potentially making e-consults faster and less costly. However, whether AI-based e-consults are acceptable may depend on how patients and clinicians weigh factors such as who provides the advice, how quickly it is received, its cost, and its quality. The purpose of this study is to examine how older adult patients and primary care clinicians weigh these factors when comparing e-consults produced by a human specialist with those produced by an AI system. In the study, participants will complete a one-time survey in which they compare sets of two hypothetical e-consults. Each e-consult will include a different combination of features, such as whether the advice comes from a human or AI, the expected wait time, the cost, and the quality of the advice. This study will estimate how much patients and clinicians value each feature in an e-consult. The findings will inform the potential future AI e-consults, so they better reflect patient and clinician preferences.

Study Type

INTERVENTIONAL

Allocation

Purpose

HEALTH_SERVICES_RESEARCH

Masking

NONE

Enrollment

220

Conditions

Primary Care Patients Primary Care Provider

Eligibility

Sex: ALLMin age: 65 Years

Medical Language ↔ Plain English

PATIENTS: Inclusion Criteria: * Adult, 65 years of age or older * Had at least two office visits at Penn Medicine in the past 12 months * Has an upcoming appointment in a Penn Medicine primary care setting, including family medicine, internal medicine, and geriatric medicine Exclusion Criteria: * Documented Alzheimer's, dementia, or related condition CLINICIANS: Inclusion Criteria: * Adult, 18 years of age or older * Works at Penn Medicine as one of the following: physician, nurse practitioner, or physician's assistant * Works in a primary care Penn Medicine setting including: family medicine, internal medicine, geriatrics medicine Exclusion Criteria: * Actively in medical school, internship, or residency

Outcomes

Primary Outcomes

Preferences and Tradeoffs of Attributes of Electronic Consults

Participants will review 12 sets comparing 2 hypothetical e-consults. Each e-consult is defined by five attributes: the source (i.e., AI vs. human), wait time, cost, and quality of the e-consult, as well as, whether follow-up questions can be asked. For each set, the outcome is the participant's selection of one e-consult over the other. We will analyze the results of the discrete choice experiment using mixed-effects logistic regression adjusted for each attribute value with crossed random effects (intercepts) for participant and task.

Time frame: From enrollment to the end of the survey, approximately 15 minutes

Secondary Outcomes

Health Numeracy Measured by the Short Numeracy Understanding in Medicine Instrument

Health numeracy will be measured using the 8-item Short Numeracy Understanding in Medicine instrument (S-NUMi). Each correct response contributes one point to the total score. Total scores range from 0 to 8, with higher scores indicating greater health numeracy.