Geographical variability in the likelihood of bloodstream infections due to gram-negative bacteria: Correlation with proximity to the equator and health care expenditure

David N Fisman, Eleni Patrozou, Yehuda Carmeli, Eli Perencevich, Ashleigh R. Tuite, Leonard A. Mermel, the Geographical Variability of Bacteremia Study Group

Research output: Contribution to journalArticleResearchpeer-review

35 Citations (Scopus)


Objective: Infections due to Gram-negative bacteria exhibit seasonal trends, with peak infection rates during warmer months. We hypothesized that the likelihood of a bloodstream infection due to Gram-negative bacteria increases with proximity to the equator. We tested this hypothesis and identified geographical, climatic and social factors associated with this variability. Design: We established a network of 23 international centers in 22 cities. Setting: De-identified results of positive blood cultures from 2007-2011 and data sources for geographic, climatic and socioeconomic factors were assembled for each center. Participants: Patients at the 23 centers with positive blood cultures. Main outcome: Due to variability in the availability of total culture volumes across sites, our primary outcome measure was the fraction of positive blood cultures that yielded Gram-negative bacteria; sources of variability in this outcome measure were explored using meta-regression techniques. Results: The mean fraction of bacteremia associated with Gram-negative bacteria was 48.4% (range 26.4% to 61.8%). Although not all sites displayed significantseasonality, the overall P-value for seasonal oscillation was significant (P<0.001). In univariate meta-regression models, temperature, latitude, latitude squared, longitude, per capita gross domestic product and percent of gross domestic product spent on healthcare were all associated with the fraction of bacteremia due to Gram-negative bacteria. In multivariable models, only percent of gross domestic product spent on healthcare and distance from the equator (ie. latitude squared) were significantly associated with the fraction of bacteremia due to Gram-negative bacteria. Conclusions: The likelihood of bacteremia due to Gram-negative bacteria varies markedly between cities, in a manner that appears to have both geographic (latitude) and socioeconomic (proportion gross domestic product devoted to health spending) determinants. Thus, the optimal approach to initial management of suspected bacteremia may be geographically specific. The rapid emergence of highly antibiotic-resistant Gram-negative pathogens may have geographically specific impacts.

Original languageEnglish
Article numbere114548
JournalPLoS ONE
Issue number12
Publication statusPublished - 1 Dec 2014

Cite this