Search strategy
We performed a systematic review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol (4). The protocol for this systematic review was registered on PROSPERO (CRD42019136298) and can be accessed at http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42019136298 (5).
We identified quantified estimates of transmission routes from humans, animals and environmental reservoirs that resulted in an acquisition of ARB. An acquisition is conceptualized as one or more of: exposure, intake, infection, acquisition, carriage and colonisation. Antibiotic resistance is defined as the resistance of bacteria to one or more antibiotics for which the bacterium is not intrinsically resistant.
PubMed was searched using the search terms shown in Additional file 1 in section S1 Appendix, resulting in 4576 articles published until January 1st 2019. Secondly, Embase was searched using search terms shown in S1 Appendix, which resulted in 5195 articles. Articles were not excluded based on publication date, publication type, sample size, significance level or quality Removing duplicates in Endnote and Rayyan resulted in 6054 articles for title/abstract screening. Two researchers separately in- or excluded a random subset of 50 of the 6054 articles, there was an agreement of 100%. Thereafter, one researcher performed the title/abstract screening in Rayyan (6). In cases of doubt on in- or excluding an abstract/article, the second researcher was consulted. Five-hundred-and-twelve articles were included for full-text screening. After the full-text screening, 277 articles were included. Figure 1 shows a flow diagram of the articles in the review process. The PRISMA 2009 Checklist (4) can be found in Additional file 1 in section S2 Appendix.
Data synthesis and analyses
Reservoirs. Twelve reservoirs and cross-reservoirs in which transmission occurs were identified, 1) animal, 2) animal and food, 3) animal and human, 4) animal, human and water, 5) animal and environment, 6) animal and water, 7) human, 8) human and environment, 9) human and food, 10) human and water, 11) environment, and 12) environment and water. Travelling and studies researching an intervention to reduce transmission, e.g. handwashing, were not categorized as a reservoir. Although travelling frequently is considered a “route”, multiple transmission routes, such as food, people and environment, could cause acquisitions during travelling and therefore, we did not appoint travelling to a reservoir.
Transmission routes. In all 48 transmission routes were identified and frequencies are shown in the table in Additional file 1 in section S3 Table.
Methods of estimation. Four methods for quantifying transmission routes were identified; (1) based on genetic similarity between ARB or resistance genes in source and receiver; (2) quantifying the dosage of ARB ingested by humans; (3) mathematical modelling to estimate transmission rates; and (4) statistical inference. The latter included calculations of OR, risks, prevalence ratio’s (PR), risk ratio’s (RR) and transmission rates (TR). Methods using genetic similarity estimated percentages of transmissions that had occurred (T, transmission percentage) or estimated a risk being the percentage in which acquisition with a genetically identical strain to the source had occurred from the total sample at risk (R, risk percentage). Modelling methods included R0, attributable percentage of the total number of cases to this route, transmission rates (TRs), cases per day, incidence, incidence rate ratios (IRRs) and acquisition rates (ARs).
Pathogens. To estimate transmission routes per ARB, each estimate was categorized per bacterial species if two or more estimates on this species were present, otherwise the estimate was categorized as “other”. The following 19 groups were identified Staphylococcus aureus (S. aureus), Escherichia coli (E. coli), Enterobacteriaceae (estimates pooled for multiple species of Enterobacteriaceae), vancomycin resistant enterococci (VRE), staphylococci, Pseudomonas aeruginosa (P. aeruginosa), Acinetobacter baumannii (A. baumannii), Acinetobacter calcoaceticus (A. calcoaceticus), Campylobacter, Enterococcus faecalis (E. faecalis), Enterococcus faecium (E. faecium), Streptococcus pneumoniae (S. pneumoniae), group B streptococci, salmonella, Staphylococcus epidermidis (S. epidermidis), Staphylococcus pseudintermedius (S. pseudintermedius), Staphylococcus haemolyticus (S. haemolyticus) and other (Enterococci, Mycoplasma hominis, Ureaplasma urealyticum, Klebsiella pneumoniae, Staphylococcus hominis, Staphylococcus cohnii, Klebsiella oxytoca, Haemophilus influenzae, genes of a bacterial species (e.g. blaTEM)).
Travelling. For travelling countries of destination were grouped in one of the following 11 groups; South Asia (Afghanistan, Bangladesh, Bhutan, Maldives, Nepal, India, Pakistan and Sri Lanka), South East Asia (Indonesia, Cambodia, Laos, Myanmar, Malaysia, Thailand, Vietnam, Australia, Brunei, the Philippines and Singapore), West Asia (Bahrain, Iran, Israel, Jordan, Kuwait, Lebanon, Oman, Qatar, Palestinian territories, Syria, Turkey, Iraq, Saudi Arabia, United Arab Emirates, and Yemen), Central & East Asia (China, Hong Kong, Mongolia, Uzbekistan, Turkmenistan, Tajikistan, Kyrgyzstan, Kazakhstan, East Turkestan, North Korea, South Korea, Japan, and Taiwan), Asia unspecified, Latin America (central Amerika, Mexico and south Amerika), North America, America unspecified, Africa (entire African continent), Europe and multiple regions (multiple of the above regions).
Meta-analysis. The initial plan was to perform random effects meta-analyses using the maximum-likelihood estimator per methodology per transmission route per pathogen for studies with the same outcome measure or outcome measures that could be lumped. Yet, as there were not enough studies providing high-quality data on the frequency of exposure and the probability of colonisation, we were unable to quantify the link between probability of colonisation and frequency of exposure. In addition, high heterogeneity in methods, study designs, and analysis performed within categories of ARB transmission routes investigated precluded performance of pooled analyses in sensible manner. We, therefore, limit the reporting to a descriptive analysis of studies included in the systematic review. Our script for the initial meta-analysis in R version 3.6.1. and the datasets generated during the current study are available in the GitHub repository, https://github.com/NoorGo/Transmission_routes_of_antibiotic_resistant_bacteria_systematic_review and as R files and Excel files in Additional file 2, 3 and 4.