1. Prevalent cases are all persons who were existing cases in the population each year.
2. Prevalent cases from 2023 to 2032 were forecasted using ARIMA (Auto Regressive Integrated Moving Average) models, trained on historical data from 2014 to 2022.
3. Prevalence rate was calculated by dividing the number of existing cases for that year by the midyear population estimate.
4. Age standardization was performed using the direct method, with the 2022 Population in Hong Kong as the reference.
5. Prevalence rates from 2023 to 2032 were forecasted using ARIMA models, trained on historical data from 2014 to 2022.

Abbreviations: PI: Prediction interval.

1. The cost refers to the all-cause cost for that calendar year, including all healthcare resource utilization expenses across all settings.
2. Unit costs are based on fees and charges for public hospital services set by the Hospital Authority, effective from 2017.
3. Cost projections (2023–2032) were forecasted using ARIMA (Auto Regressive Integrated Moving Average) models trained on 2014–2022 data, adjusted for the 2019–2022 pandemic and social fluctuation.

1. Patients with unmet needs are defined as those who had taken at least two antidepressant regimens for a minimum of 28 days each (allowing gaps of up to 14 days) and required a third regimen to confirm failure of the previous two.

Note: Each curve tracks a cohort of patients (in the particular year) newly diagnosed with depression with each data point representing the cumulative number of individuals from that cohort who have developed treatment-resistant depression by that calendar year.