BACKGROUND: Detrusor overactivity (DO) is characterised by the symptoms of overactive bladder (OAB) including incontinence, urgency, and voiding episodes; though, identification of a cardinal symptom still remains unclear, where treatment efficacy is often evaluated across multiple outcomes. As a result, ascertainment of the most patient impacting symptom is crucial. Joint modelling allows us to evaluate the association between surrogate outcome measures such as patients’ symptom profiles and time to perceived treatment failure. OBJECTIVES: To determine the most associated OAB symptom with patients’ perceived time of treatment failure using a novel application of a joint modelling framework. METHODS: To account for the potential correlation between symptoms, we applied a joint model of multivariate longitudinal and time-to-event data to a randomised clinical trial of patients receiving botulinum toxin for DO. We investigate the association of incontinence, urgency and voiding episodes collected at baseline, 6 weeks, 3 months, and 6 months, on patients’ perceived time of treatment failure. RESULTS: The multivariate joint model was applied to 122 patients (270 data points) with 69 (56.6%) patients reporting treatment failure. The results identified urgency as the most impactful surrogate measure on patients’ perceived time of treatment failure, with an estimated hazard ratio of 1.26 (95% CI: 0.97 to 1.64) for every additional urgency episode. Incontinence, voiding episodes and the interaction between symptom measures had a non-significant association. However, in this example, sensitivity to the model choice produced inconsistent estimates due to the limited nature of the data set and therefore results should be interpreted cautiously. CONCLUSIONS: This approach illustrates the advantage of applying a joint modelling framework to identify symptoms most associated with time to treatment failure. With an increasing need to identify and assess core symptoms for varying medical conditions, the novel use of a joint modelling approach would appear to be extremely promising.