TL;DR
This study causally links COVID-19 lockdowns in the US to a significant increase in mental health facility usage, particularly for women and specific diagnoses like panic disorder. The findings suggest that policy interventions like lockdowns had a more profound impact on mental health than the pandemic itself, with effects persisting and even growing throughout 2020.
What Problem Does This Paper Address?
The COVID-19 pandemic and the subsequent lockdowns were widely expected to negatively impact mental health. While anecdotal evidence and smaller studies suggested this, a large-scale, causal investigation into the impact of these mitigation measures on mental health resource utilization across the United States was lacking. Specifically, the authors aimed to disentangle the effect of the lockdowns themselves from the effect of the pandemic's presence on mental health outcomes, using robust causal inference methods. Understanding this distinction is crucial for informing future public health policy and resource allocation during crises.
Methods
The core of the study employs a difference-in-differences (DID) analysis. This econometric technique allows researchers to estimate the causal effect of an intervention (in this case, lockdown policies) by comparing the changes in outcomes over time between a "treated" group (regions with lockdowns) and a "control" group (regions without lockdowns).
To ensure the validity of the DID approach, the authors:
- Validated parallel trends: They used an "event-study" model to confirm that mental health trends in treated and control regions were similar before the lockdown policies were implemented.
- Controlled for confounding factors: They incorporated a time-varying adjusted (TVA) model to account for the influence of COVID-19 case counts, acknowledging that the pandemic's progression could also affect mental health.
- Used fixed effects: Regional and date fixed effects were included in the models to control for unobserved, time-invariant regional characteristics (e.g., local healthcare infrastructure) and time-varying factors (e.g., general societal changes).
Datasets
The study utilized three primary datasets:
- Mental Health Claims Data: A large, de-identified medical claims corpus from Change Healthcare, covering approximately two-thirds of the US population and over 50% of private insurance claims from 2019-2020. This dataset provided daily patient visits to mental health facilities and emergency department (ED) visits for mental health issues.
- COVID-19 Cases Data: State- and county-level data on confirmed COVID-19 cases, sourced from the New York Times database.
- Lockdown Data: Information on the dates of stay-at-home and school closure orders at state and county levels, obtained from the COVIDVis project.
Key Insights and Ideas
- Lockdowns, not just the virus, drove mental health strain: The study provides strong evidence that the policy interventions (lockdowns) themselves, rather than solely the presence of the virus, significantly increased mental health facility usage. This is a critical distinction for policy design.
- Mental health is sensitive to policy: The findings suggest that mental health is more reactive to policy changes like lockdowns than to the direct threat of the pandemic itself, highlighting the profound societal impact of these measures.
- Long-term and escalating effects: The impact of lockdowns wasn't a short-term blip; it significantly increased mental health resource usage, and these effects continued to grow throughout 2020, even months after initial implementation.
- Disproportionate impact on women: Female populations experienced a more pronounced negative effect on their mental health due to lockdowns compared to males.
- Specific diagnoses surged: Conditions like panic disorder and reactions to severe stress saw significant increases, indicating a direct link between lockdown stress and specific mental health challenges.
Key Results
- Increased Facility Usage: Regions with lockdowns saw an 18% increase in mental health resource usage compared to a 1% decline in regions without lockdowns.
- Gender Disparities: In counties, stay-at-home orders led to a 24% increase in mental health patient visits for females, compared to a 3% increase in non-locked counties. For males, there was a 5% decline in non-locked counties.
- Diagnosis-Specific Increases: Patients diagnosed with panic disorder saw the largest increase, with an estimated effect of 3.3 (per 10^5 population) in counties with lockdowns.
- Emergency Department Visits: ED visits for mental health issues also showed a significant positive trend in response to lockdown measures, indicating an increase in urgent treatment-seeking.
- State-Level Consistency: Similar positive and significant effects of lockdowns on mental health were observed at the state level, reinforcing the robustness of the findings.
Limitations and Future Work
The authors acknowledge several limitations:
- Non-random Lockdown Adoption: Lockdowns were not implemented randomly, and their timing was influenced by factors like COVID-19 case counts and regional capacities, which could introduce bias. However, the DID methodology and regional fixed effects were employed to mitigate this.
- Economic Confounding: Job losses and economic disadvantage due to lockdowns could confound the results, as economic hardship is linked to mental illness.
- Data Coverage: The medical claims data did not include Medicare and Medicaid insurance programs, potentially limiting the representativeness of the findings for certain demographic groups (e.g., low-income beneficiaries). Additionally, the data lacked race and ethnicity demographics.
- Future Work: The study suggests the need for policy interventions to be accompanied by strategies that ensure mental health treatment reachability, especially during restrictive periods, to mitigate the exacerbated effects of delayed care. Further research could explore the long-term mental health consequences and the effectiveness of different policy durations.
Summary updated