Depression and anxiety during and after episodes of COVID-19 in the community

Abstract

Understanding the connection between physical and mental health with evidence-based research is important to inform and support targeted screening and early treatment. The objective of this study was to document the co-occurrence of physical and mental health conditions during and after the experience of symptomatic SARS-CoV-2 illness episodes. Drawing from a national symptoms' surveillance survey conducted in the UK in 2020, this study shows that individuals with symptomatic forms of SARS-CoV-2 (identified by anosmia with either fever, breathlessness or cough) presented significantly higher odds of experiencing moderate and severe anxiety (2.41, CI 2.01-2.90) and depression (3.64, CI 3.06-4.32). Respondents who recovered from physical SARS-CoV-2 symptoms also experienced higher odds of anxiety and depression in comparison to respondents who never experienced symptoms. The findings are robust to alternative estimation models that compare individuals with the same socioeconomic and demographic characteristics and who experienced the same local and contextual factors such as mobility and social restrictions. The findings have important implications for the screening and detection of mental health disorders in primary care settings. They also suggest the need to design and test interventions to address mental health during and after physical illness episodes.

Figure 1

COVID-19 symptomatic episodes and mental health (adjusted odds ratios, logistic regression). Notes. Authors’ estimations from the UK COVID-19 symptoms tracker survey (April–December 2020). Adjusted ORs and 95% CI. X-axis on log scale. Sample size: 16,771. “Ongoing illness” identifies respondents with COVID-19 symptoms; “Past illness” identifies respondents that recovered from symptomatic COVID-19. Other covariates: region fixed effects and a constant. Outcomes (binary) are moderate and severe anxiety or depression, based on GAD-7 and PHQ-9 metrics. Regressions include age-gender-specific probability weights. Standard errors are heteroskedasticity-robust.