Use of Digital Peer Support for Employee Well-Being: Retrospective Analysis Across Five Large Employers

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Large percentage improvements but clinical significance unclear due to unvalidated measurement

• Primary Outcome: Reduction in negative sentiment scores: loneliness (46%), sadness (45%), stress (46%), anxiety (39%), depression (40%)

• Nnt Or Sensitivity: 77% increase in optimism scores; resource sharing associated with 2.9-8.8% additional improvement across domains

• Confidence Interval: Not reported

Clinical Application

Platform appears scalable but requires validation studies before clinical recommendation Would need evaluation against Australian privacy laws, workplace mental health frameworks, and integration with existing EAP providers. TGA regulation unclear for AI-based mental health tools Working adults in large organisations seeking anonymous peer support for subclinical mental health concerns

Abstract

BACKGROUND: Anonymous, 24/7 digital peer support (DPS) offers a scalable solution to support employees' emotional well-being. Understanding sociobehavioral factors, such as timing of engagement and the impact of shared resources, can help employers and employee assistance programs (EAPs) integrate digital tools to better support workforce well-being. OBJECTIVE: This study examines (1) outcomes via overall sentiment changes among DPS users; (2) sociobehavioral differences between employees who accessed real-time, anonymous DPS during versus outside business hours; (3) differences in sentiment outcomes based on time of use; and (4) the impact of in-session resource sharing on sentiment improvement. METHODS: Using OpenAI's large language model (LLM) GPT-4o-mini with a few-shot learning approach, 24,818 anonymous chat conversations from 13,879 employees at 5 large employers were evaluated for subclinical sentiment variables, including loneliness, sadness, stress, anxiety, depression, despair, helplessness, and optimism. RESULTS: Distinct activity patterns were observed between employees during and outside business hours, with a median user age of 36 years. During business hours, employees reported higher baseline stress (Δ1.6%), whereas outside business hours, baseline depression (Δ1.7%) and loneliness (Δ1.2%) were higher. After DPS use, LLM-derived negative sentiment scores decreased (loneliness 46% reduced, sadness 45%, stress 46%, anxiety ~39%, depression ~40%, despair ~40%, and helplessness ~38%), and optimism increased (~77%). Outside business hours, users engaged more (~68% of all sessions), remained in sessions 19% longer, and discussed 13% more topics, whereas users during business hours reported greater improvements in depression (Δ2.3%), helplessness (Δ1.9%), and loneliness (Δ0.5%). Resource sharing was associated with greater improvements in loneliness (Δ2.9%), stress (Δ1.3%), anxiety (Δ1.9%), depression (Δ7.3%), despair (Δ3.0%), helplessness (Δ2.8%), and optimism (Δ8.8%), but not sadness. CONCLUSIONS: DPS complements employers' EAPs by addressing employee engagement gaps, reducing barriers to mental health care, and promoting emotional well-being among the workforce.