Search

BackgroundDuring the COVID-19 pandemic, health care systems were faced with the urgent need to implement strategies to address the behavioral health needs of health care workers. A primary concern of any large health care system is developing an easy-to-access, streamlined system of triage and support despite limited behavioral health resources.ObjectiveThis study provides a detailed description of the design and implementation of a chatbot program designed to triage and facilitate access to behavioral health assessment and treatment for the workforce of a large academic medical center. The University of California, San Francisco (UCSF) Faculty, Staff, and Trainee Coping and Resiliency Program (UCSF Cope) aimed to provide timely access to a live telehealth navigator for triage and live telehealth assessment and treatment, curated web-based self-management tools, and nontreatment support groups for those experiencing stress related to their unique roles.MethodsIn a public-private partnership, the UCSF Cope team built a chatbot to triage employees based on behavioral health needs. The chatbot is an algorithm-based, automated, and interactive artificial intelligence conversational tool that uses natural language understanding to engage users by presenting a series of questions with simple multiple-choice answers. The goal of each chatbot session was to guide users to services that were appropriate for their needs. Designers developed a chatbot data dashboard to identify and follow trends directly through the chatbot. Regarding other program elements, website user data were collected monthly and participant satisfaction was gathered for each nontreatment support group.ResultsThe UCSF Cope chatbot was rapidly developed and launched on April 20, 2020. As of May 31, 2022, a total of 10.88% (3785/34,790) of employees accessed the technology. Among those reporting any form of psychological distress, 39.7% (708/1783) of employees requested in-person services, including those who had an existing provider. UCSF employees responded positively to all program elements. As of May 31, 2022, the UCSF Cope website had 615,334 unique users, with 66,585 unique views of webinars and 601,471 unique views of video shorts. All units across UCSF were reached by UCSF Cope staff for special interventions, with >40 units requesting these services. Town halls were particularly well received, with >80% of attendees reporting the experience as helpful.ConclusionsUCSF Cope used chatbot technology to incorporate individualized behavioral health triage, assessment, treatment, and general emotional support for an entire employee base (N=34,790). This level of triage for a population of this size would not have been possible without the use of chatbot technology. The UCSF Cope model has the potential to be scaled, adapted, and implemented across both academically and nonacademically affiliated medical settings.

To present the structure and outcomes to date for the Psychiatric Assessment and Brief Intervention (PABI), a pilot program developed at University of California, San Francisco, to improve access of primary care patients to mental health services. PABI offers diagnostic evaluations and brief (up to 3 months) evidence-based treatment, including pharmacologic management and psychotherapy, to medical patients 18 years of age and older. Core PABI features are ensuring prompt access, actively partnering with patients and referring providers, and coordinating seamless transitions of care. Demographic and clinical variables and outcome indicators were collected for all patients seen in PABI from October 2015 to June 2017. Descriptive statistics and mixed-effects linear models were used to analyze the data. During the study period, 139 patients (54% women, mean [SD] age of 48.2 [17.5] years) with a mean of 2 DSM-5 psychiatric diagnoses each (range, 1-5) were seen. Mean time to access was 8 days, with a mean length of stay in the program of 11 weeks. Compared to baseline, final behavioral health measure scores showed significant improvement: the mean Patient Health Questionnaire-9 score decreased by 5.9 points (95% CI, 4.6-7.2), and the mean 7-item Generalized Anxiety Disorder scale score was 4.4 points lower (95% CI, 3.2-5.6; both P values < .0001). This brief psychiatric treatment program provides prompt access to quality mental health care for patients with medical comorbidities. Results to date suggest that this program leads to significantly improved clinical outcomes. Further research is needed to determine its long-term sustainability and generalizability.