Can economic indicators predict infectious disease spread? A cross-country panel analysis of 13 European countries

Abstract

Aims: It is unclear how economic factors impact on the epidemiology of infectious disease. We evaluated the relationship between incidence of selected infectious diseases and economic factors, including economic downturn, in 13 European countries between 1970 and 2010. Methods: Data were obtained from national communicable disease surveillance centres. Negative binomial forms of the generalised additive model (GAM) and the generalised linear model were tested to see which best reflected transmission dynamics of: diphtheria, pertussis, measles, meningococcal disease, hepatitis B, gonorrhoea, syphilis, hepatitis A and salmonella. Economic indicators were gross domestic product per capita (GDPpc), unemployment rates and (economic) downturn. Results: GAM models produced the best goodness-of-fit results. The relationship between GDPpc and disease incidence was often non-linear. Strength and directions of association between population age, tertiary education levels, GDPpc and unemployment were disease dependent. Overdispersion for almost all diseases validated the assumption of a negative binomial relationship. Downturns were not independently linked to disease incidence. Conclusions: Social and economic factors can be correlated with many infections. However, the trend is not always in the same direction, and these associations are often non-linear. Economic downturn or recessions as indicators of increased disease risk may be better replaced by GDPpc or unemployment measures.

Association between ambient temperature and COVID-19 infection in 122 cities from China

“…We also controlled the relative humidity (rhu il ), air pressure (prs il ), and wind speed (win il ) during the same period for the possible confounding effect. log(y i,t−1 ) is the log-transformed COVID-19 counts lagged one day in city i to account for potential serial correlation in the data (Hunter et al, 2019;Liu et al, 2020). city i is the city fixed effect variable and day t captures day fixed effect (Amuakwa-Mensah et al, 2017).…”

A multimethod approach for county-scale geospatial analysis of emerging infectious diseases: a cross-sectional case study of COVID-19 incidence in Germany

“…Due to nonlinear relationships expressed by model covariates, General Additive Models (GAMs) provide a useful semiparametric technique for modelling nonlinear associations [ 60 ]. GAMs operate as an extension of GLMs, but allow for the inclusion of smoothing terms, which can be explained by the following general form [ 61 ]: where is the row of the parametric model matrix of the model with parameters , and the smooth terms constitute the nonparametric part of the model.…”

Big Data Analytics in the Fight against Major Public Health Incidents (Including COVID-19): A Conceptual Framework

“…For a smaller domain-specific input corpus, the Twitter corpus is better than general pre-training methods such as Word2Vec (from Google News) or GloVe (from the Stanford NLP group) in terms of extracting meaningful semantic relationships, and skip-gram’s accuracy is better than that of CBOW [ 54 ]. Using this method, we can unearth the relevant factors that will affect the epidemic, even including abnormal weather changes [ 55 ], economic development [ 56 ], and other information, for a more comprehensive early warning. The BlueDot automatic epidemic surveillance system established by Kamran Khan monitors outbreaks of infectious diseases by collecting more than 100,000 articles every day [ 57 ].…”