Depression and anxiety during COVID-19 - Authors' reply.

We thank Michael Daly and Eric Robinson for their comments on our Article. We share Daly and Robinson's caution against generalising mental disorder prevalence estimates solely from the very early phases of the COVID-19 pandemic. Our methods took this timing into account in several ways. Every estimate that informed our model had corresponding values of the impact of the pandemic (estimates of human mobility and SARS-CoV-2 infection rate) for the period in which the survey was done, whether it was early or later during the course of the pandemic. We extrapolated the change in prevalence (by age, sex, and location) for each day of the year 2020, on the basis of daily estimates of the COVID-19 impact indicators. We then calculated the average daily prevalence for the year 2020 to represent the annual point prevalence for that year. This approach meant that our extrapolated annual point prevalences incorporated the prevalences during the months leading up to the pandemic, the very early phases of the pandemic, the time between waves of infection, and subsequent waves and prevention measures.

Although Daly and Robinson are correct that the most common month of data collection was April, 2020, we also had studies providing data for every month between March, 2020 and January, 2021 (appendix). We had 12 studies reporting on prevalence from June, 2020 onwards. Many of these studies about the later months in 2020 still reported elevated COVID-19 prevalence. For example, data from Knudsen and colleagues suggested a 70% increase in the prevalence of major depressive disorders (via diagnostic interview) in Trondheim, Norway, during August and September, 2020, compared with prepandemic estimates. Also, the Household Impacts of COVID-19 Survey in Australia reported a 71% increase in the prevalence of psychological distress in November, 2020 compared with prepandemic estimates.

The broader literature on the effect of past population shocks also shows substantial increases in mental disorder prevalence. For example, prevalences of major depressive episodes doubled after the 2009 financial crisis in Greece, and increased by more than 50% following the 2008 financial crisis in Hong Kong. Elevated prevalence of depressive and anxiety disorders have also been observed in conflict-affected populations.

However, our methods contain several important caveats, and we acknowledge the large bounds of uncertainty around the prevalence estimates produced. We need more high-quality mental health survey data across many parts of the world throughout 2020 and 2021 to better understand the effect of COVID-19 on the prevalence of mental disorders. Our method and results reflect the best approach and best estimates available, given the limitations and sparsity of available data. We appreciate the work by researchers like Daly and Robinson in doing these surveys during challenging circumstances brought about during the pandemic. We hope to see more work of this kind in the future.