This paper demonstrates the use of data-driven analytics for improving the delivery of care to patients suffering from treatment-resistant depression (TRD), a severe form of the major depressive disorder, that necessitate a referral to a specialized mental health clinic. In particular, this paper focuses on two important decisions, i.e., (1) the pharmacotherapy treatment selection at the initial visit to this specialized clinic and (2) the selection of the frequency of follow-up appointments. While these two decisions have an important effect on the patients (e.g., side effects, risk of suicide) and clinic (e.g., waiting list), there is unfortunately limited evidence on how to take these decisions, even though depression is a growing global health concern. To study these decisions, this paper uses various methods (e.g., causal inference, Cox regression, Little’s law) on an observational data set of 463 adult patients obtained from an outpatient clinic treating patients suffering from TRD. No significant results are obtained for the treatment decision at the initial visit, but we obtain insightful results for the appointment frequency decision. In particular, we show that, while variations in appointment frequencies do not appear to have a major impact on clinical outcomes, they can be managed to achieve significant improvements in the accessibility of the clinic. The higher resolution representation of the appointment frequency decision, in comparison with the initial treatment decision, seems to be a determinant of the significance of the results for the former problem.