The study of spatial autocorrelation for infectious disease epidemiology decision-making: a systematized literature review

Abstract In recent years, the global spread of communicable diseases such as Ebola and COVID-19 has stressed the need for clear, geographically targeted, and actionable public health recommendations at appropriate spatial scales. Country-level stakeholders are increasingly utilizing spatial data and spatial decision support systems to optimize resource allocation, and researchers have access to a growing library of spatial data, tools, and software. Application of spatial methods, however, varies widely between researchers, resulting in often unstandardized results, which may be difficult to compare across geographical settings. This literature review aims to compare epidemiological studies, which applies methods including spatial autocorrelation to describe, explain, or predict spatial patterns underlying infectious disease health outcomes, and to describe whether those studies provide clear recommendations. The results of our analysis show an increasing trend in the number of publications applying spatial analysis in epidemiological research per year, with COVID-19, tuberculosis and dengue predominantly studied (43% of n = 98 total articles), and a majority of publication coming from Asia (62%). Spatial autocorrelation was quantified in the majority of studies (72%), and 57 (58%) of articles include some form of statistical modeling of which 11 (19%) accounted for spatial autocorrelation in the model. Most studies (68%) provided some level of recommendation regarding how results should be interpreted for future research or policy development, however often using vague, cautious terms. We recommend the development of standards for spatial epidemiological methods and reporting, and for spatial epidemiological studies to more clearly propose how their findings support or challenge current public health practice.