The impact of COVID-19 lockdown announcements on mental health: quasi-natural experiment in Lombardy, Italy

Abstract

Background: Evidence showed that mental health problems have risen markedly during COVID-19. It is unclear if part of the mental sufferings relates to the climate of uncertainty and confusion originated from rough communication by health officials and politicians. Here, we test the impact of unanticipated policy announcements of lockdown policies on mental health of the older population.

Methods: We used a representative telephone-based survey of 4400 people aged 65 years or older in Italy's Lombardy region to compare information on self-reported symptoms of anxiety, depression and poor-quality sleep of subjects interviewed on the days of the policy announcement with that of subjects interviewed on other days. We used regression models adjusting for potential socio-demographic confounders as well study design with inverse probability weighting.

Results: On days when policymakers announced to extend the lockdown, mental health deteriorated on average by 5.5 percentage points [95% confidence interval (CI): 1.1-9.8] for self-reported anxiety symptoms and 5.1 percentage points (95% CI: 2.7-7.4) for self-reported depressive symptoms. The effect of the announcement to shorten the lockdown is more moderate but statistically significant. These associations were short term in duration; after just 1 day, self-reported mental health and sleep quality return to levels better than pre-announcement until a new policy change.

Conclusions: Our research shows that lockdown policy announcements are associated with short-term worsening in mental distress, highlighting the importance of appropriate communication strategies and political determinations in crisis times.

Figure 1

Trends in levels of anxiety symptoms, depressive symptoms and poor-quality sleep, 17–30 November 2020. Notes: This figure reports the percentage prevalence (%) of subjects having anxiety symptoms, depressive symptoms and poor-quality sleep and their 95% CIs conditioning on the interview date. The vertical-dotted reference lines indicate the announcement days

Figure 2

Coefficient plots for interview days from probit model after risk adjustment. Notes: This figure reports the coefficient plots with 95% confidence interview from the probit model from Supplementary appendix table SA9. The horizontal line represents the reference level of the lowest mean of the outcome variable across all interview dates