Improving Contraceptive Counseling for College-Aged People with Uteruses: Implementing an Artificial Intelligence-Based Contraceptive Decision Aid at a University Health Clinic

Unintended pregnancy disproportionately affects college-aged people with a uterus, with rates of 76% among those aged 18-19 and 56% among those aged 20–24. Contributing barriers include digital misinformation, limited appointment time, and inconsistent provider training in contraceptive counseling. This quality improvement project evaluated the impact of an AI-based contraceptive decision aid (SARHAchat™) on patient and provider satisfaction with contraceptive counseling at a university health clinic. SARHAchat™ was implemented across four iterative, two-week Plan-Do-Study-Act cycles (September–November 2025) in the Gynecology department at UNC Campus Health. A pre-appointment link was distributed to patients for voluntary completion. Patient satisfaction and decisional conflict were assessed using the Person-Centered Contraceptive Counseling (PCCC) Measure and SURE tool. Provider satisfaction was evaluated via a Likert-scale survey. Of 91 eligible patients, 27.5% completed surveys and 19.8% of provider-reported encounters involved SARHAchat™ use. Decisional conflict was low (96% reported none), and contraception initiation was high (92%), though neither patient nor provider satisfaction met the 80% benchmark (68% and 69%, respectively). SARHAchat™ users were less likely to rate their provider as "Excellent" than non-users (33% vs. 79%), a clinically meaningful though statistically non-significant difference (p = 0.07), likely reflecting the small user sample. Low SARHAchat™ uptake (19.8%) limited the ability to draw conclusions, though high decisional clarity and contraception initiation suggest promise. Satisfaction benchmarks were not met, and lower provider ratings among SARHAchat™ users warrant further investigation. Future efforts should focus on improving pre-appointment engagement to enable more definitive evaluation of the tool's effectiveness.