Masks and distancing during COVID-19: a causal framework for imputing value to public-health interventions

Area of Science:

• Epidemiology
• Public Health
• Data Science

Background:

• Predictive models for governmental interventions during COVID-19 are common.
• Assessing the actual effects of these interventions retrospectively is less advanced.

Purpose of the Study:

• To propose a novel data-driven framework for retrospectively assessing the causal effects of public health interventions.
• To quantify the impact of specific interventions, such as mask mandates, on disease transmission.

Main Methods:

• Utilized regularized regression to develop a parsimonious model fitting observed data.
• Employed the do-operator to simulate counterfactual scenarios, isolating intervention effects.
• Quantified intervention impact by comparing actual outcomes with simulated counterfactuals.

Main Results:

• The framework successfully identifies and quantifies the impact of interventions.
• Demonstrated that recommending universal mask usage could avert approximately 200,000 COVID-19 cases in three US states.
• The model is applicable to scenarios with sparse temporal cause-and-effect relationships.

Conclusions:

• The proposed data-driven framework offers a robust method for retrospective analysis of intervention effectiveness.
• This approach can inform future public health policy by providing quantitative evidence of intervention impacts.
• The methodology is generalizable to various situations requiring the assessment of time-sparse causal relationships.