RESUMEN
Green spaces may be protective against COVID-19 incidence. They may provide outdoor, ventilated, settings for physical and social activities and therefore decrease transmission risk. We examined the association between neighborhood greenness and COVID-19-like illness incidence using individual-level data. Methods: The study population includes participants enrolled in the COVID Symptom Study smartphone application in the United Kingdom and the United States (March-November 2020). All participants were encouraged to report their current health condition and suspected risk factors for COVID-19. We used a validated symptom-based classifier that predicts COVID-19-like illness. We estimated the Normalized Difference Vegetation Index (NDVI), for each participant's reported neighborhood of residence for each month, using images from Landsat 8 (30 m2). We used time-varying Cox proportional hazards models stratified by age, country, and calendar month at study entry and adjusted for the individual- and neighborhood-level risk factors. Results: We observed 143,340 cases of predicted COVID-19-like illness among 2,794,029 participants. Neighborhood NDVI was associated with a decreased risk of predicted COVID-19-like illness incidence in the fully adjusted model (hazard ratio = 0.965, 95% confidence interval = 0.960, 0.970, per 0.1 NDVI increase). Stratified analyses showed protective associations among U.K. participants but not among U.S. participants. Associations were slightly stronger for White individuals, for individuals living in rural neighborhoods, and for individuals living in high-income neighborhoods compared to individuals living in low-income neighborhoods. Conclusions: Higher levels of greenness may reduce the risk of predicted COVID-19-like illness incidence, but these associations were not observed in all populations.
RESUMEN
BACKGROUND: There is uncertainty about the mode of transmission of the severe acute respiratory syndrome (SARS) virus. We analyzed the temporal and spatial distributions of cases in a large community outbreak of SARS in Hong Kong and examined the correlation of these data with the three-dimensional spread of a virus-laden aerosol plume that was modeled using studies of airflow dynamics. METHODS: We determined the distribution of the initial 187 cases of SARS in the Amoy Gardens housing complex in 2003 according to the date of onset and location of residence. We then studied the association between the location (building, floor, and direction the apartment unit faced) and the probability of infection using logistic regression. The spread of the airborne, virus-laden aerosols generated by the index patient was modeled with the use of airflow-dynamics studies, including studies performed with the use of computational fluid-dynamics and multizone modeling. RESULTS: The curves of the epidemic suggested a common source of the outbreak. All but 5 patients lived in seven buildings (A to G), and the index patient and more than half the other patients with SARS (99 patients) lived in building E. Residents of the floors at the middle and upper levels in building E were at a significantly higher risk than residents on lower floors; this finding is consistent with a rising plume of contaminated warm air in the air shaft generated from a middle-level apartment unit. The risks for the different units matched the virus concentrations predicted with the use of multizone modeling. The distribution of risk in buildings B, C, and D corresponded well with the three-dimensional spread of virus-laden aerosols predicted with the use of computational fluid-dynamics modeling. CONCLUSIONS: Airborne spread of the virus appears to explain this large community outbreak of SARS, and future efforts at prevention and control must take into consideration the potential for airborne spread of this virus.