Results of the generalized additive model.

Background
Social restrictions and vaccination seem to have shaped the pandemic development in Europe, but the influence of geographical position is still debated. This study aims to verify whether the pandemic spread through Europe following a particular direction, during the period between the start of the pandemic and November 2021. The existence of a spatial gradient for epidemic intensity is also hypothesized.
Methods
Daily COVID-19 epidemiological data were extracted from Our World in Data COVID-19 database, which also included vaccination and non-pharmacological interventions data. Latitude and longitude of each country’s centroid were used as geographic variables. Epidemic periods were delimited from epidemic surge data. Multivariable linear and Cox’s regression models were performed for each epidemic period to test if geographical variables influenced surge dates. Generalized additive models (GAM) were used to test the spatial gradient hypothesis with three epidemic intensity measures.
Results
Linear models suggest a possible west-east shift in the first epidemic period and features a significant association of NPIs with epidemic surge delay. Neither latitude nor longitude had significant associations with epidemic surge timing in both second and third periods. Latitude displays strong negative associations with all epidemic intensity measures in GAM models. Vaccination was also negatively associated with intensity.
Conclusions
A longitudinal spread of the pandemic in Europe seems plausible, particularly concerning the first wave. However, a recurrent trend was not observed. Southern Europe countries may have experienced increased transmissibility and incidence, despite climatic conditions apparently unfavourable to the virus.

研究背景
社交限制措施与疫苗接种似乎对欧洲新冠大流行的发展轨迹产生了显著塑造作用，但地理位置对疫情的影响仍存在广泛争议。本研究旨在验证在新冠大流行起始至2021年11月的时段内，欧洲境内的大流行传播是否遵循特定方向，并同时验证疫情强度存在空间梯度这一研究假说。

研究方法
本研究从世界数据项目（Our World in Data）新冠数据库中提取每日新冠流行病学数据，该数据库同时涵盖疫苗接种与非药物干预措施（non-pharmacological interventions，以下简称NPIs）相关数据。以每个国家几何中心的经纬度作为地理变量，基于疫情暴发数据划定各疫情周期。针对每个疫情周期构建多元线性回归与考克斯（Cox）回归模型，以检验地理变量是否对疫情暴发日期产生影响。采用广义可加模型（Generalized Additive Models, GAM）结合三项疫情强度评估指标，对空间梯度假说进行验证。

研究结果
线性模型结果显示，首个疫情周期可能存在自西向东的传播趋势，且NPIs与疫情暴发延迟存在显著关联。在第二与第三疫情周期中，经纬度均与疫情暴发时间无显著关联。在广义可加模型中，纬度与所有三项疫情强度评估指标均呈现显著负相关关系。疫苗接种也与疫情强度呈负相关。

研究结论
欧洲新冠大流行存在纵向传播的可能性，在第一波疫情中尤为明显，但并未观察到持续的重复传播趋势。尽管气候条件似乎不利于病毒存活，但南欧国家的病毒传播能力与感染发病率或已上升。