Burden of disease and identification of intervention targets in extended-spectrum β-lactamase-associated infections

Abstract

Extended-spectrum β-lactamase-producing Enterobacteriaceae (EPE) have surged globally since the early 2000s, partly due to dissemination of CTX-M-15 producing Escherichia coli (E. coli) sequence type 131 and its sub-lineages H30-R and H30-Rx. The overarching aim of the thesis was to increase the knowledge of the disease burden in and risk factors for bloodstream infection (BSI) with EPE.

For papers I and II, 408 children were enrolled in a cross-sectional study upon seeking emergency medical care at a referral hospital in Bissau with signs of systemic infection (fever or tachycardia). In paper I, we reported that the proportion of these children that were intestinally colonized with E. coli or Klebsiella pneumoniae (K. pneumoniae) that produced extended-spectrum β-lactamases (ESBLs) was 33% and that co-resistance to other antibiotic drugs was common. There was considerable genetic heterogeneity among the isolates, suggesting wide community dissemination. This descriptive knowledge of EPE dispersion in Guinea-Bissau can help inform local antibiotic treatment guidelines for infections potentially caused by Gram-negative bacteria. In Paper II, we assessed the etiology of severe infections in the same population. Pathogenic bacteria were identified from the bloodstream of 12% of the children. Staphylococcus aureus was the commonest species (26 findings), followed by non-Typhoidal Salmonella (5) and Streptococcus pneumoniae (4). Bloodstream infection (BSI) was common among non-febrile children. Malaria was clinically diagnosed in 64% of the subjects, but could only be laboratory-verified in 5%. The findings can be used to inform treatment in patients with signs of systemic infection and support previous studies indicating fever as an insufficient predictor of BS.

In Paper III, the risk of EPE BSI was studied in a cohort of Swedes with a previous finding of EPE in urine (urine cohort) or feces (feces cohort) from 2007-12 and compared to the risk in the Swedish population. The incidence rate in the urine cohort peaked at 22.4 events per 1000 person-years 31-90 days after the initial finding and fell to 1.8 events after 2-6 years. The cumulative 6-year incidence was 3.8% and 1.6% for the urine and feces cohorts, respectively. The relative risk of incident EPE BSI in the respective cohorts compared to the Swedish population was 62- and 32-fold. Urological disorders and other underlying morbidities were associated with increased risk. Paper IV was a nation-wide case-control study of disease burden and risk factors for community-onset EPE BSI. The incidence rate of EPE BSI in Sweden was 1.7 events per 100 000 person-years during 2007-12. Male sex, high age and urological morbidity were strong risk factors. The 30-day mortality after EPE BSI was 11.7%. Fluoroquinolone consumption 8-91 days before the outcome was associated with 5.5-fold odds, which in a causal interpretation yields a population attributable fraction of 14%. The findings from papers III and IV can be used to identify high-risk populations that are suitable targets for interventions such as eradication therapy and antibiotic stewardship programs.

In conclusion, the thesis contributes to the knowledge of BSI and EPE dissemination in Guinea-Bissau and can be used to inform treatment guidelines and public health policy. It also improves our understanding of the natural history of EPE colonization in terms of subsequent risk of EPE BSI and the attribution of morbidity, antibiotic consumption and other risk factors to that outcome.