Source attribution of human echinococcosis: A systematic review and meta-analysis

Abstract

Background: A substantial proportion of echinococcosis transmission to humans via contamination of food has been assumed. However, the relative importance of food as a transmission vehicle has previously been estimated through expert opinion rather than empirical data.

Objective: To find and evaluate empirical data that could be used to estimate the source attribution of echinococcosis, in particular the proportion that is transmitted through contaminated food.

Methods: A systematic review was undertaken to identify reports on the risk factors for human cystic (CE) and alveolar (AE) echinococcosis. Data bases searched included PubMed, Scopus, Web of Knowledge, Cab Direct, Science Direct, Google Scholar, eLIBRARY.RU, CyberLeninka, CNKI and VIP. Search terms included Echinococc*, hydatid, epidemiology, logistic regression, risk factors, odds ratio, relative risk, risk factors. Reports, including grey literature where available, that had suitable data were selected and data were extracted. The main pathways of transmission were hypothesised to be contact with the definitive host, contaminated water, contaminated food and contaminated environment (other than food). For each study the attributable fraction for these potential sources of infection was calculated from the data presented. A meta-analysis was then undertaken to obtain pooled estimates for the relative contribution of these transmission pathways.

Results: Data from 28 cross-sectional studies and 14 case-control studies were extracted. There was strong evidence for transmission by direct contact with dogs for both CE and AE. The estimated attributable fractions were 26.1% (CI 13.8%-39.6%) and 34.4% (CI 20.7% -48.2%) respectively. Transmission through contaminated water was estimated to be responsible for approximately 29.4% (CI 12.1%-51.7%) for CE and 24.8% (CI 10.6% to 42.6%) for AE. Contaminated food may be responsible for approximately 23.4% of CE cases (CI 2.1%-47.3%). Globally, there was insufficient evidence to conclude AE can be transmitted by food, although case control studies from low human incidence areas suggested that possibly 32.5% (CI 10.0%-53.2%) could be transmitted by food. There was also insufficient evidence that direct contact with foxes was a significant source of human disease. There were no suitable studies with a risk of environmental contact reported, but the residual attributable fraction thatwould likely include this pathway was approximately 21.1% for CE and 11.1% for AE.

Conclusions: The results support the hypothesis that dog contact and drinking contaminated water are major pathways of transmission of both CE and AE. For contaminated food, the results are less consistent, but suggest that it is an important transmission pathway and provide better evidence than expert elicitations as previously used.

Fig 1. Prisma flow diagram with the search strategy steps.

Fig 2. Geographical distribution of studies.

For CE (red, size proportional to number of selected studies) this ranged from 17 studies from China, 2 from Argentina, Uganda, and Peru, and 1 for other countries. For AE (blue), there were 10 studies from China, and one each from Kyrgyzstan, Germany, Austria, France and USA (Alaska). The map was generated using the open source software R [13] using a shape file downloaded from Natural Earth (www.naturalearthdata.com).

Fig 3. Forest plot for the attributable fraction for 24 cross-sectional studies and 9 case-control studies that analysed potential contact with infected dogs as a potential risk factor for CE.

* and # indicate different studies from the same published report. $ Data not reported in study, but supplied by the authors.

Fig 4. Forest plot for the attributable fraction for 11 cross-sectional studies and 4 case-control that analysed contaminated water as a potential risk factor for CE.

$ Indicates different studies in the same published report the same source. *Data not reported in study, but supplied by the authors.

Fig 5. Forest plot for the attributable fraction for 8 cross-sectional studies and 5 case-control studies that analysed contaminated food as a potential risk factor for CE.

Fig 6. Funnel plot illustrating the relative risk (or OR as an estimate of RR) of studies of CE associated with food.

Black circles are the selected studies, and open circles represent missing data.

Fig 7. Forest plot for the attributable fraction for 10 cross-sectional studies and 5 case-control studies that analysed contact with infected dogs as a potential risk factor for human AE.

Fig 8. Forest plot for the attributable fraction for 6 cross-sectional studies and 1 case-control study that analysed contaminated water as a potential risk factor for human AE.

*Data not reported in study, but supplied by the authors.

Fig 9. Forest plot for the attributable fraction for 6 cross-sectional studies and 4 case-control studies that analysed contaminated food as a potential risk factor for human AE.

*Data not reported in study, but supplied by the authors.

Fig 10. Forest plot for the attributable fraction for 5 cross-sectional studies and 3 case-control studies that analysed contact with foxes as a potential risk factor for human AE.

*Data not reported in study, but supplied by the authors.

Fig 11. Transmission pathways for E. granulosus and E. multilocularis between the definitive host and humans.

The 95% range (CI) for the attributable fraction of each pathway is given. Red lines indicate statistical evidence for the reported AF. Orange lines indicate insufficient statistical evidence (ie CI include 0). Dotted lines indicate the assumed pathway that is the residual AF after others are deducted. Green solid line indicates indirect pathway to human infection.