An Evaluation of the Use of Automated Chatbots Within One Psychological Therapy and Counselling Service

Abstract

Objectives: This study evaluated the effectiveness of the use of automated chatbots within one NHS Talking Therapies provider. Two chatbots were designed and piloted: (1) to help prepare patients for therapy and (2) to provide solution-based coaching to patients regarding a problem or concern. This study aimed to evaluate the effectiveness of the use of these chatbots in the NHS Talking Therapies service. Methods: The chatbots were made available to patients (N = 4329) who had been referred for therapy and were waiting for their first assessment. Data from a control group of patients (N = 4333) who did not use a chatbot were also analysed. The evaluation adopted a mixed methods approach to capture data relating to the conversations had by patients with the chatbot, feedback regarding satisfaction with its use, data relating to patient attendance and standardised outcome measures relating to patients' improvements in anxiety and depression. Results: Qualitative data showed that the chatbots were received positively by patients who gave feedback. Quantitative data showed non-attendance was significantly lower in the control group (p ≤ 0.001), cancellations were significantly lower in the control group (p ≤ 0.001), but days spent in therapy were significantly higher in the chatbot groups (p ≤ 0.001). All groups saw significant improvements in their GAD-7 (p = 0.000) and PHQ-9 (p = 0.000) scores. Conclusions: In NHS Talking Therapy services, waiting times have been problematic; thus, chatbots can be a useful tool to assist patient engagement while they wait for their appointment. Therefore, they have the potential to reduce anxiety, frustration or disappointment in patients who are unable to wait for too long and may drop out before they start therapy.