COVID-19 is the first pandemic in human history in which it has been possible to have real-time access to data associated with its spread at different time and spatial scales in distinct regions of the world such as countries, states, counties, and provinces, among others. By analyzing data on confirmed COVID-19 cases and deaths in distinct regions of the globe, in this work, we show that some complex system properties are universally present in the spatial and temporal spread of the COVID-19 pandemic. Specifically, by using the cumulative daily time series of confirmed cases and deaths of the different sub-regions which can be divided into 14 distinct regions of the world, we show as quasi-equilibrium states, ensemble fluctuation scaling and the emergence of log-normal distribution patterns are properties that are universally present in the spatial spread of COVID-19. Also, we analyze the daily time series of confirmed cases and deaths of the 14 regions of the world which are considered in this work, and then we show as cross or long-range correlations, temporal fluctuation scaling, temporal Theil scaling, and multifractality are universal properties present in the temporal spread of COVID-19. Additionally, a spatial and temporal stochastic model is proposed that takes into account these stylized facts of the spread of COVID-19.