A One Health systematic review of diagnostic tools for Echinococcus multilocularis surveillance: Towards equity in global detection

Echinococcus multilocularis is a zoonotic cestode of canid definitive hosts that is emerging as a parasite of medical and veterinary concern in regions of North America, Europe and Asia. Infection with the metacestode stage (alveolar echinococcosis – AE) is life-threatening, especially for patients who reside in low resource countries and lack access to modern diagnostic tests and treatments. The overall objectives of this One Health review were to systematically describe the diagnostic tests currently employed in endemic countries to detect E. multilocularis in people, canids and the environment, and to report the test characteristics of new diagnostic techniques for population surveillance. In this systematic review of English and Chinese language databases, we identified 92 primary records of E. multilocularis surveillance in canids (N = 75), humans (N = 20) and/or the environment (food, soil; N = 3) and 12 grey literature records that reported E. multilocularis surveillance or health systems protocols between 2008 and 2018. Surveillance for E. multilocularis was conducted using a broad range of combined morphological, molecular, immunological and imaging techniques. Nine studies reporting diagnostic evaluations for cestode or metacestode detection were identified, including studies on copro-antigen ELISA, copro-PCR, intestinal examination, Western Blot, magnetic capture RT-PCR and immunochromatography. Our dataset includes prevalence estimates for E. multilocularis in canids, people, or environment in 27 of the 43 endemic countries and reports data gaps in surveillance, laboratory methods, and diagnostic sensitivity. International consensus on gold standard diagnostic techniques and harmonization of human, canid and environmental surveillance data across political boundaries are needed to comprehensively assess the global burden and distribution of this parasite.


Introduction
Alveolar echinococcosis (AE) is a debilitating medical condition that affects people and animals infected with the metacestode stage of Echinococcus multilocularis (Eckert et al., 2001). Such hosts are infected when they accidentally ingest cestode eggs that Food and Waterborne Parasitology 12 (2019) e00048 Searches of Google Scholar excluded patents but included citations. One author (JS) reviewed the first 10 pages of results, and each page thereafter containing at least one relevant citation.
One author (JS) searched multiple databases (GreyLit.org, OpenGrey, Science.gov, European Centres for Disease Control (ECDC), European Food and Safety Authority (EFSA), Government of Canada) for grey literature published from January 1st, 2008 up to October 14th, 2018 using the terms "Echinococcus multilocularis" OR "E. multilocularis" OR "Alveolar echinococcosis". Using the same terms, a second author (LM) searched The University of Toronto's custom Canadian government document search engine (available at: https://cse.google.com/cse?q=+&cx=007843865286850066037:3ajwn2jlweq) for relevant Canadian federal, provincial and municipal documents published from January 1st, 2008 up to October 14th, 2018. We were unable to access the search databases of relevant Chinese grey sources (National Institute of Parasitic Diseases, China Ministry of Health, National Health Commission).
Peer-reviewed and grey literature searches were not restricted by language. All collected studies were collated in Zotero reference manager (https://www.zotero.org/) and duplicates were removed according to title and author.

Study selection
Primary studies were included if they described diagnostic methods used for E. multilocularis surveillance and/or diagnostic method evaluation. Institutional laboratory or surveillance protocols and institutional surveillance reports outlining prevalence of E. multilocularis were also included. Reference lists of relevant grey literature were searched for additional sources. In addition, as part of our One Health approach, inclusion criteria selected studies that reported outcomes from humans, canids (i.e. wolves, foxes, coyotes, raccoon dogs and domestic dogs), and/or the environment (e.g. soil, plants, food, water). We excluded data related to non-canid definitive and aberrant hosts as population level surveillance is uncommon for these animals, and we excluded data related to intermediate hosts as post-mortem examination is a widely used and accepted method of diagnosis. We also excluded studies where diagnostic method efficacy was conducted using an early stage index test in Phase I or Phase II of evaluation (Boelaert et al., 2018). This review focused on articles published between 2008 and 2018 because we wanted to capture and report the diagnostic methods currently being used for surveillance, and because the accuracy of older diagnostic tests has been reported elsewhere. Further, studies must have reported on detection of E. multilocularis within an endemic country. Two global reviews of AE prevalence and distribution were completed in the last 10 years, and we included countries listed as endemic by these publications to create a list of countries to be included in this review Deplazes et al., 2017). When discrepancies occurred between the two reports, we considered whether a country shared a border with an endemic country, whether autochthonous animal or human cases had been recently reported, and whether the authors had found reliable secondary sources on which to base their decision. The 43 countries that we considered endemic for AE are listed in Table 1.
Case reports, case series, short reports, short communications, letters, conference abstracts, experimental infections, reviews or meta-analyses were excluded, as were studies that did not report the diagnostic test used. Finally, article selection was carried out using the following steps: (1) title and abstracts of all collected studies were screened for relevance according to the stated inclusion/exclusion criteria; (2) articles that were deemed to be relevant, or for which more information was required, were read in full; (3) any studies that met relevance criteria following review of the full article proceeded to data extraction. The eligibility of each study was assessed independently by two reviewers (JS, AN; LM, ET; DH, JPM; YH, HL). Disagreements regarding eligibility were resolved by consensus.

Data extraction
Data from each relevant article was extracted independently by two reviewers (JS, AN; LM, ET; DH, JPM; YH, HL), using pretested forms. Extracted information from primary studies included: first author, publication year, title, language, study objective, host/source, location of data collection, location of lab analysis, study design, sampling method, inclusion/exclusion criteria, sample size, reported prevalence, detection method(s), tests in series/parallel, criteria for assigning case status, test characteristics, method(s) for determining test characteristics, and if E. multilocularis samples were sequenced and/or submitted to GenBank® (Clark et al., 2016), an open access nucleotide sequence database. When a study reported the criteria to assign case status, this was extracted and recorded. If no explicit criteria were stated, we assumed that the results of a single test were used to assign case status but classified the case definition as Not Reported when multiple tests were used. Data elements extracted from grey literature included title, first author, publication year, report type, host/source, case definition(s), reportable/notifiable directive, diagnostic method(s) used and reported prevalence.

Quality assessment
Only studies that evaluated diagnostic test accuracy for detection of the parasite were assessed for quality. Each study that passed through quality assessment was independently evaluated by two separate reviewers (LM, JS) using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS 2) tool (Whiting et al., 2011). QUADAS 2 measures risk of bias and applicability to the review question within four domains: patient selection, index test, reference standard, and flow/timing. Disagreement was settled by discussion and consensus.   1996-1997: 39 2003: 195 1996-1997: 7.7 2003: 11.7 (i)

Data synthesis
We reported the geographic location of population level E. multilocularis surveillance, the detection tools used by such studies and the diagnostic test characteristics of detection tools evaluated as part of Phase III or Phase III field studies. Maps display one study location per citation and were created using ArcGIS v10.6. Elements of the QUADAS 2 assessment were rated as high risk of bias, low risk of bias or unclear according to the method's recommendations for analysis (Whiting et al., 2011).

Primary literature search
Our searches of English and Chinese language primary literature identified 1393 articles that matched our search terms for the years 2008-2018 (Fig. 1). Of these, 1295 records were excluded because they were duplicates, unavailable, did not meet inclusion criteria, or were written in a language other than English or Chinese. The primary literature search retrieved eight reports deemed to be grey literature, which were then included in grey literature relevance screening. In total, the search team extracted data from 92 peer-reviewed articles reporting E. multilocularis prevalence or diagnostic test evaluation in canids (N = 69; Table 2), humans (N = 14; Table 3), multiple hosts (N = 6; Table 4), or the environment (N = 3; Table 5). Prevalence estimates were most frequently reported for China (N = 21), France (N = 11), Poland (N = 10) and Canada (N = 7), with an overall dataset that spanned 27 of the 43 AE endemic countries in North America, Europe and Asia (Fig. 2a,b). The countries that reported some form of surveillance for this parasite were most frequently categorized as high (74%, 20 of 27) or upper middle (22%, 6 of 27) income by the World Bank (The World Bank, 2018). Countries not reporting data were evenly categorized as high, upper-middle, and low-middle (31%, 5 of 16, each) with one country categorized as low income. Only 34 of the 92 studies reported the location of sample analysis; of these, most research teams analyzed their samples in the same country where they were collected (85%, 29 of 34). Only two studies estimated E. multilocularis prevalence in a low income country (Kyrgyzstan) and these occurred through collaboration between Kyrgyz and Swiss researchers, with molecular analyses performed in Switzerland (Ziadinov et al., 2008;Ziadinov et al., 2010).

Surveillance for E. multilocularis in canids
Our search collected 75 primary research studies that carried out case detection and population surveillance of E. multilocularis cestodes in wild and domestic canids (Table 2). Of these, six studies reported simultaneous surveillance in canids and humans (Table 4). Canid surveillance was most frequently reported from France and China (N = 11 each), followed by Canada (N = 7), Poland (N = 6), and Iran (N = 5) (Fig. 2a). Many studies reported prevalence estimates for multiple canid species, and out of all canid or canid/human studies the majority focused on red foxes (49%), followed by dogs (32.5%) and foxes (unspecified, 12.5%). Just under 10% reported on each of coyotes, raccoon dogs and wolves, and our search also found studies that assessed prevalence in silver foxes, arctic foxes, grey foxes, Tibetan foxes, and a single study that reported on captive canids. Most studies used a variety of diagnostic techniques to identify and confirm E. multilocularis infection, including necropsy or arecoline hydrobromide purgation followed by morphological identification of cestodes, copro-PCR, fecal sieving/flotation followed by PCR, and coproantigen ELISA (Enzyme-Linked Immunosorbent Assay). Post-mortem examination of canid intestinal tracts was employed using various techniques (Sedimentation and Counting Technique -SCT, Intestinal Scraping Technique -IST, Segmental Sedimentation and Counting Technique -SSCT, Sedimentation, Filtration and Counting Technique -SFCT) to collect E. multilocularis adult cestodes for species level identification (N = 44). Molecular identification of taeniid eggs or Echinococcus cestodes, either as  a stand-alone assay or to confirm morphological results, was conducted by 55 research teams. These methods included a variety of PCR techniques -conventional simplex, nested, multiplex, copro-, fluorescent, qPCR, RFLP-PCR, magnetic capture RT-PCR, as well as microsatellite analysis. Various loci were targeted, including cytochrome c oxidase subunit 1 (co1), NADH dehydrogenase subunits 1 and 5 (nd1, nd5), ATPase subunit 6 (atp6), cytochrome b (cob), and the small and large subunit rRNA genes (rrnS, rrnL).
Multiplex PCR was the single most popular molecular technique (N = 20), with the majority carried out using Cest1/Cest2 primers (nd1) to differentiate E. multilocularis from E. granulosus and Taenia spp. (Trachsel et al., 2007). Only two studies used alternative multiplex primers (Jiang et al., 2012;Liu et al., 2018). Of the studies using PCR to detect E. multilocularis in canids, approximately half (57%, 25 of 44) sequenced PCR products and one-third (34%, 15 of 44) submitted these nucleotide sequences into GenBank®. The highest number of records that sequenced and/or submitted E. multilocularis sequences originated from Canada ( Although canid studies often did not explicitly state their sample collection strategy, most appeared to use convenience sampling and a cross-sectional approach. Furthermore, 36% (29/80) of studies reporting surveillance of E. multilocularis in canids did not describe the case definition (Tables 2, 3).

Surveillance for E. multilocularis in humans
In total, 20 studies from four countries (China, Kazakhstan, Poland, Austria) conducted population surveillance for AE in people (Tables 3-4). Prevalence based on abdominal ultrasound and serology was estimated to be 0-13.45% in China (Yang et al., 2008;Gao et al., 2018), and 0% in Kazakhstan (Torgerson et al., 2009). Seroprevalence was estimated to be 0% in Austria (Poeppl et al., 2013) and 0-1.7% in Poland (Karamon et al., 2016;Cisak et al., 2015). Most case definitions (70%, 14/20) were based on abdominal ultrasound and confirmatory serology (ELISA and/or Western Blot), while the remainder were based on serology only. Case definitions for AE were not reported for 30% of studies where humans were surveyed for infection (Ma et al., 2015;Karamon et al., 2016;Sen-Hai et al., 2008;Torgerson et al., 2009;Han et al., 2015;许阳阳 et al., 2015). Case definitions were reported or implied for all studies of human patients (Table 3) but were only provided for one study that addressed human/canid infection (Table 4). A variety of sampling methods were reported including convenience, purposive, randomized, and cluster. Only one study sequenced AE cyst tissue removed from human patients and submitted the DNA sequences to GenBank® (Ma et al., 2015).

Surveillance for E. multilocularis in the environment
Two studies conducted surveillance for E. multilocularis eggs in food (Lass et al., 2015;Lass et al., 2017), and one conducted surveillance in soil between 2008 and 2018 (Szostakowska et al., 2014; Table 5). All three studies collected taeniid eggs using ZnCl 2 flotation, identified the eggs using nested PCR and sequencing, and submitted amplified PCR product sequences into Genbank®. The reported contamination levels ranged from 6.73% to 23.3% in Poland where the studies took place; however, these levels have been called into question.

E. multilocularis diagnostic test evaluations
Our review identified nine studies that conducted Phase III or Phase III field surveillance to evaluate the diagnostic accuracy of techniques to detect E. multilocularis in people, canids or the environment ( Table 6). Seven of these protocols screened canids for cestode or metacestode infection, and were categorized as assays for (i) intestinal examination (N = 1), (ii) fecal analysis (N = 5), or (iii) serology (N = 1): (i) The SSCT is a modification of the gold standard SCT protocol that reduces processing time by examining cestode predilection sites in the intestinal tract (Umhang et al., 2011). The SSCT has a high sensitivity (b98.3%) and specificity compared to SCT (both depend on the intestinal segment), but lower accuracy in estimating infection intensity. (ii) Fecal analysis is a non-invasive method of sampling canids; however, the eggs shed by E. multilocularis cestodes are morphologically indistinguishable for those of other taeniid species and molecular analyses can lack sensitivity due to inconsistent shedding of eggs and the difficulty of extracting DNA from taeniid eggs. We identified one study that aimed to optimize molecular detection by testing different combinations of commercially available kits for DNA extraction with PCR and qPCR protocols (Maksimov et al., 2017). Using IST as the reference standard, the authors identified QIAamp® Fast DNA Stool Mini Kit with qPCR (using QuantiTect® Multiplex-Master Mix) as having the highest sensitivity (97%) of the combinations tested. An alternative technique, DNA fishing using magnetic probes to extract DNA in combination with RT-PCR, is less sensitive (88.2%) relative to the SCT but is appealing for mass surveillance as it can be partially automated for high throughput (Isaksson et al., 2014). This systematic review identified other techniques, such as arecoline purgation, copro-antigen ELISAs and PCR, which had even lower sensitivity but that generally had acceptable specificity (close to 100%; Maksimov et al., 2017;Ziadinov et al., 2008;Otero-Abad et al., 2017). (iii) Lastly, an evaluation of native and recombinant antigens highlighted EmVF-Western Blot and rEm95-ELISA as two highly sensitive (100%) and specific (100% and 98%, respectively) options for serological diagnosis of AE in canids (Frey et al., 2017).
Of the two human serological rapid diagnostic tests evaluated against abdominal ultrasound as a reference standard, the antibody gold immunochromatographic assay demonstrated higher sensitivity (97.8%) and specificity (95.8%) than the Em2-DIGFA assay. Both tests are commercially available, field stable, and can be used to rapidly screen large numbers of people ( Table 6).
Six of the nine studies reported the analytical approach used to calculate sensitivity and specificity, and these were as follows: standard 2 × 2 calculation and modelling methods (Receiver Operating Curve/Area Under the Curve, Hui-Walter maximum likelihood estimation, and Bayesian Latent Class model; Iskenderali Ziadinov et al., 2008;Jiang et al., 2012;Otero-Abad et al., 2017;Gao et al., 2018;Isaksson et al., 2014;Frey et al., 2017). Ziadinov et al., 2008 used maximum likelihood estimation to determine test characteristics in dog populations from different villages in Kyrgyzstan, while Otero-Abad et al., 2017 employed Bayesian latent class analysis to jointly estimate test characteristics of four tests, prevalence and covariates related to prevalence in foxes. Both methods allow for determination of sensitivity and specificity in the absence of a gold standard. No studies were identified that evaluated new methods of detecting E. multilocularis in environmental samples at the Phase III field level.

QUADAS 2 quality appraisal
Only one of the nine studies that evaluated diagnostic assays for E. multilocularis surveillance in people, canids or the environment was considered low risk of bias and concern across all QUADAS2 criteria (Table 7; Gao et al., 2018). High risk of bias related to index and reference tests was generally a result of poor transparency on interpreting test thresholds and/or lack of clarity on whether diagnosticians were blinded between index and reference test results. The high risk of bias associated with participant selection was due to the frequent use of case-control studies and unclear reporting on exclusion criteria. Similarly, lack of information regarding time intervals between reference and index tests contributed to the assessment of high bias risk for flow and timing for approximately half of studies. In general, few applicability concerns were identified, indicating that participant selection and the use and interpretation of index and reference tests matched the questions posed by the systematic review.

Grey literature search
Searches performed on grey literature search engines and government databases collected 276 reports, of which 12 were health system protocols or surveillance reports from endemic countries and deemed relevant ( Table 8). Nine of these were European Union (EU) reports on surveillance methodology and outcomes in canids and humans, of which two provided information on diagnostic methods (Madslien et al., 2013;Antolova et al., 2014). Annual surveillance for E. multilocularis in red foxes in Norway is conducted to maintain official "free-from" status, for which the 2011-2012 assessment utilized magnetic capture realtime PCR to detect positive cases (Madslien et al., 2013). Antolova et al., 2014 carried out multi-year surveillance in Slovakia, using nested PCR for dogs, SCT for red foxes as well as ELISA, western blot and imaging techniques for humans (Table 8). While diagnostic techniques were not reported in EFSA surveillance reports, the 2015 citation states that the human case definition of echinococcosis does not differentiate between the two forms of the disease (EFSA, 2015). The EU defines human echinococcosis as at least one of the following: (1) histopathology or parasitology compatible with E. multilocularis OR E. granulosus (direct visualization of the protoscolex in cyst fluid); (2) detection of E. granulosus pathognomonic macroscopic morphology of cyst(s) in surgical specimens; (3) typical organ lesions detected by imaging techniques (CT, sonography or MRI) AND confirmed by a serological test; (4) Echinococcus spp. specific serum antibodies by high-sensitivity serological test AND confirmed by a high specificity serological test; (5) detection of E. multilocularis or E. granulosus nucleic acid in a clinical specimen (ECDC, 2018b). Three Canadian health system protocols were found, including two 2018 Ontario Public Health Standards outlining guidance on the natural history, transmission and public health management of E. multilocularis in humans (MOHLTC, 2018a;MOHLTC, 2018c). Appendix B provided provincial case definitions and appropriate diagnostic methods for case detection and surveillance, Table 7 Quality appraisal of studies that reported diagnostic test characteristics of novel methods for Echinococcus multilocularis surveillance in people, canids, or the environment at the population level (2008-2018).

Title
Author, year  citing that serological testing for specimens collected in Ontario is analyzed using a combination of Em2-ELISA, II/3-10-ELISA or Em2Plus-ELISA at the Institute of Parasitology, University of Bern, Switzerland (MOHLTC, 2018a). The third report detailed public health guidance for sampling and detection of E. multilocularis in animals but did not provide details of testing methods (MOHLTC, 2018b).

Discussion
Despite recent improvements to diagnostic technologies, AE remains a life-threatening infection for humans and animals; in part because patients across endemic countries do not have equal access to modern diagnostics and treatments. This systematic review identified examples of case finding in 27 of 43 countries where E. multilocularis is known to be present, confirming that there continue to be knowledge gaps in the global distribution and burden of this parasite Torgerson et al., 2010;EFSA and ECDC, 2016). Prevalence estimates were most notably missing from Russia, which has the second highest reported annual incidence after China , and were also missing from other Central Asian countries (Azerbaijan, Georgia, Mongolia, Tadjikistan, Uzbekistan, Turkmenistan). Time period and language limitations of our search strategy might partially explain these gaps; however, it is also likely that limitations in reporting infrastructure, access and availability of diagnostic tests, poor physician and veterinarian awareness, long asymptomatic period, and presence of other closely related cestodes contribute to the general issues of under-diagnosis and under-reporting universal to characterizing this parasite (Eckert et al., 2001;EFSA and ECDC, 2016). These barriers to understanding the true burden of AE are cause for concern because E. multilocularis appears to be emerging in areas of North America, Europe and Asia. This review identified nine protocols that were evaluated through Phase III and Phase III field trials for accuracy in diagnosing AE. There is currently no international consensus on specific gold standard protocols to detect this parasite in humans, animals, or the environment. Recommendations by the World Health Organization-Informal Working Group on Echinococcosis state that diagnostic criteria for AE in humans require parasite detection using at least one of the following: (i) imaging, (ii) serology, (iii) histopathology, (iv) nucleic acid detection (Brunetti et al., 2010). Many laboratories consider SCT to be the gold standard for case detection in wild canids (Eckert et al., 2001;Conraths and Deplazes, 2015), and as a result, new diagnostic techniques are often evaluated against this standard. However, two recent studies, one comparing SCT and three other diagnostic techniques in latent class models (Otero-Abad et al., 2017) and the second comparing SCT directly to magnetic capture RT-PCT (Isaksson et al., 2014), provide evidence that the sensitivity of SCT is lower than originally thought (Eckert et al., 2001). Therefore, it would be sensible to reconsider the value of SCT in surveillance and as a reference standard for diagnostic test evaluations (Conraths and Deplazes, 2015). As importantly, intestinal examination requires death of the animal, which has ecological effects when conducted as part of mass surveillance, and which is also not suitable for diagnosing infection in domestic or captive canids. We noted some studies conducting intestinal examination as a stand-alone determinant of E. multilocularis infection did not report morphologically identifying cestodes, which would delay early detection of other emerging Echinococcus species. Our study suggests that the latest ante-mortem developments in commercial and in-house technology lack diagnostic sensitivity and/or specificity (Table 5), although older techniques excluded by the timeline of this study do exist (EFSA and ECDC, 2016). Magnetic capture RT-PCR performed on fecal matter had the highest reported diagnostic accuracy of these assays and can be semiautomated for mass surveillance but requires costly equipment and reagents (Isaksson et al., 2014). In contrast, several promising serological tests were identified for detecting metacestode infection in human and canid hosts, each with sensitivity and sensitivity exceeding 95%. These included a commercially available but modified EUROLINE®-Western Blot (based on IgG, rEm18, rEm95, rEgAgB), an in-house Western Blot (EmVF) and an in-house ELISA (rEm95) for use in dogs, and a commercially available antibody gold immunochromatographic rapid diagnostic test for use in people (Gao et al., 2018;Frey et al., 2017). Diagnostic tests have differing capacity for detecting pre-patent, early metacestode and low intensity infections (Conraths and Deplazes, 2015). Furthermore Phase III field evaluations of test accuracy are often performed on one host in one locale, ignoring potential differences in accuracy across host species and prevalence (Conraths and Deplazes, 2015). Therefore, mass screening campaigns should consider the epidemiological situation of a region and the detection limits of diagnostic tests when creating a surveillance strategy.
Laboratory capacity is an integral component of health system infrastructure, and such services play a key role in detection, assessment, response, notification, and monitoring of public health events. According to the World Bank, seven AE endemic countries were classified as low income (Kyrgyzstan, Tadjikistan) or low-middle income (Georgia, Mongolia, Uzbekistan, Moldova, Ukraine) at the mid-point of this review; of these, our search only captured prevalence estimates for Kyrgyzstan (The World Bank, 2018;Ziadinov et al., 2008;Ziadinov et al., 2010). Some E. multilocularis detection techniques, such as SCT, arecoline hydrobromide purgation and Em2-DIGFA, require minimal equipment and are feasible for a range of settings. However, most molecular and immunological detection techniques require significant investments to laboratory infrastructure and technician training as well as access to expensive reagents, making them inaccessible to diagnosticians in low resource regions. It is also important to consider the safety aspects of various diagnostic tests. For example, dogs treated with arecoline hydrobromide have died of bone splinters puncturing the intestinal tract. Technicians collecting and analyzing freshly purged fecal matter must be properly equipped with personal protective equipment, and environments housing purged animals must be thoroughly decontaminated from infectious E. multilocularis eggs (Eckert et al., 2001). High income countries currently engaged in E. multilocularis surveillance can play a greater role in building capacity for surveillance, diagnostics, research and treatment among lower resource countries, which would ultimately deliver mutual benefits given the ability of infected wild canid to move freely between endemic and non-endemic regions. Moreover, it is not only low income countries experiencing capacity shortages. The outsourcing of Ontario medical diagnostics to Swiss laboratories for confirmatory testing suggests the need for improved laboratory capacity in Canada (MOHLTC, 2018c). As well, E. multilocularis specimens collected from humans do not appear to be sequenced routinely (our study found only one example (Ma et al., 2015)). This represents a lost opportunity to explore parasite origin, to confirm the emergence of specific haplotypes into new areas, or to investigate the biological and clinical significance of parasite diversity.
While our findings show Europe as a leading region in AE surveillance and reporting, diagnostic methods were not explicitly stated within ECDC reports, likely due to the lack of standardization across Member State (MS) laboratories (EFSA Panel on Animal Health and Welfare, 2015). Moreover, there exists a high degree of discrepancy of diagnostic test characteristics reported between MS, and as reported by pooled meta-analysis evaluations (Conraths and Deplazes, 2015; EFSA Panel on Animal Health and Welfare, 2015; Casulli et al., 2015). One third of diagnostic accuracy evaluations in our study did not report how sensitivity and specificity were calculated, while others utilized advanced modelling methods to compare measures across prevalence parameters. These findings indicate the need for standardized internal AE protocols for laboratory diagnostics within endemic and newly emerging regions. Our team did not carry out meta-analysis for test characteristics given the small number of studies collected and the variation of test types, host populations and population prevalence. We chose not to report pooled estimates of population prevalence, as combining studies by higher-level geographical designations would misrepresent prevalence variations by regions.
These diagnostic considerations bear strongly on mechanisms of disease reporting, and particularly on case definitions for surveillance. Definitions utilized by the EU and Ontario (Canada) denote a case positive by the outcome of any one of a number of tests. Moreover, some components of the case definition algorithms require a positive result from multiple diagnostic tests run in series to yield a final positive classification. In our study, case definitions in primary surveillance studies were often not described, and it was difficult to interpret the relationships between multiple tests in a diagnostic procedure. The reliance of case definitions on multiple test outcomes, combined with substantial variation in test characteristics, predictive values and performance of gold standards, is a barrier to accurately assessing the confidence of reported prevalence estimates. This is further compounded by a lack of consistency in application of case definitions within health jurisdictions. For example, while the EU has an established annual reporting system for E. multilocularis, only 23 countries in 2016 reported echinococcosis cases using the 2008 or 2012 case definitions, neither of which are species specific (ECDC, 2018b). Species specific reporting is especially important for coendemic regions, and is also a problem in Canada, where patient records obtained from the Canadian Institute for Health Information show that physicians often do not differentiate between Echinococcus species (Schurer et al., 2015). Development of case definitions should consider the test characteristics of diagnostic procedures wherever possible, be standardized for classifying/counting cases consistently across reporting jurisdictions, and should, at minimum, differentiate between Echinococcus species.
Our literature search did not identify a source that summarized mandatory versus voluntary E. multilocularis reporting all AE endemic countries. In the EU, notification of human echinococosis is mandatory for 22 MS, although other countries do voluntarily report cases (EFSA and ECDC, 2016). Echinococcosis in animals is notifiable in 17 MS, and contamination of food is notifiable in 10 MS (EFSA and ECDC, 2016). Surveillance for E. multilocularis in Europe is usually carried out on red foxes, and is predominantly diagnosed using morphological (SCT) or molecular (PCR) methods. Four EU countries (Finland, Ireland, Malta and the United Kingdom) are considered free from E. multilocularis and must conduct annual surveillance to retain this status (as per Regulation (EU) No 1152/201117). Human AE is not nationally notifiable in Canada, but it became provincially reportable in 2018 in response to heightened prevalence in wild canids and the novel detection of AE in domestic dogs within Ontario (Government of Ontario, 2018). Detection of E. multilocularis in animals and food is not currently notifiable and we did not find any documents to suggest that the government was involved in active surveillance in people, animals, or food. Human AE is reportable in China (Ding and Li, 2016), Turkey (Altintas, 2008) and Kyrgyzstan (Usubalieva et al., 2013). Considering the increased concern for this parasite in regions of Canada, Europe and Asia, there exists a need for mandatory reporting frameworks based on consistent case definitions. Moreover, developers of surveillance and reporting frameworks should consider applying a One Health approach to create enhanced systems that work synergistically in monitoring human, animal and environmental sources, especially as this parasite continues to build in importance.

Conclusions
Individuals infected with AE require early, affordable, and accurate diagnosis as well as access to modern treatment to ensure a favourable prognosis. Our study identified barriers to this goal that included scarce surveillance in low and low-middle income AE endemic countries, lack of consensus on diagnostic gold standards, reliance on convenience sampling for human and canid studies, poor reporting of case definitions, genus versus species level diagnosis, and infrequent submission of nucleotide sequences public databases. Improving reporting infrastructure systems is an important next step to comprehensively defining the global health burden and geographic distribution of E. multilocularis, as well as to monitoring emergence of this parasite into new areas and host types. Mandatory reporting of human cases in endemic countries and animal cases in non-endemic countries, data sharing between government agencies engaged in human, animal and environmental surveillance, and strengthened partnerships between low and high resource countries are all strategies to optimize health equity for AE patients. Furthermore, national E. multilocularis control programs should consider diagnostic test limitations with respect to host species, estimated local prevalence, and presence of other Echinococcus species when designing surveillance strategies. The formation of a joint WHO/OIE world reference laboratory with a mandate to develop diagnostic protocols, identify Echinococcus species, and design integrated human-animal surveillance systems could go a long way in achieving these goals.