A regression discontinuity analysis of the social distancing recommendations for older adults in Sweden during COVID-19

Abstract

Background: This article investigates the impact of a non-mandatory and age-specific social distancing recommendation on isolation behaviours and disease outcomes in Sweden during the first wave of the coronavirus disease 2019 (COVID-19) pandemic (March to July 2020). The policy stated that people aged 70 years or older should avoid crowded places and contact with people outside the household.

Methods: We used a regression discontinuity design-in combination with self-reported isolation data from COVID Symptom Study Sweden (n = 96 053; age range: 39-79 years) and national register data (age range: 39-100+ years) on severe COVID-19 disease (hospitalization or death, n = 21 804) and confirmed cases (n = 48 984)-to estimate the effects of the policy.

Results: Our primary analyses showed a sharp drop in the weekly number of visits to crowded places (-13%) and severe COVID-19 cases (-16%) at the 70-year threshold. These results imply that the age-specific recommendations prevented approximately 1800-2700 severe COVID-19 cases, depending on model specification.

Conclusions: It seems that the non-mandatory, age-specific recommendations helped control COVID-19 disease during the first wave of the pandemic in Sweden, as opposed to not implementing a social distancing policy aimed at older adults. Our study provides empirical data on how populations may react to non-mandatory, age-specific social distancing policies in the face of a novel virus.

Figure 1

Regression discontinuity plots for the impact of Sweden’s age-specific isolation recommendations on social distancing behaviours at the 70-year threshold with binned means (grey dots) and fitted values (black lines) from local linear (A) and quadratic (B) regressions estimated within mean squared error optimal bandwidths around the threshold, for three social distancing measures: (i) mean weekly visits to crowded places, (ii) mean weekly outdoor episodes with no or limited interaction and (iii) mean weekly visits to healthcare providers

Figure 2

Regression discontinuity plots for the impact of Sweden’s age-specific isolation recommendations on COVID-19 disease incidence per 1000 population at the 70-year threshold with binned means (grey dots) and fitted values (black lines) from local linear (A) and quadratic (B) regressions estimated within the mean squared error optimal bandwidths around the threshold, for two disease outcome measure: (i) severe cases (hospitalized or dead) and (ii) all confirmed cases. The y-axis is plotted on a logarithmic scale to enable better visualization of the data close to the 70-year threshold