1.0 What is the frequency of reporting o f zoonotic pathogens that cause fever?
1.2.1 Filtering and reference classification
The abstracts of the 840 full- text articles were manipulated in an Excel spreadsheet. A list of 10 common zoonotic pathogens (Table 1) was selected based on a prior procedure of
quantifying the frequency of reporting in a trial dataset of references that came up in a search for fever and zoonoses (Halliday et al., 2014). An Excel filter tool was then used to create a subset of abstracts that refer to these 10 common zoonotic pathogens or their synonyms (Table 1). The majority of the synonyms were identified in the articles used for the review. In order to identify synonyms that may not have been indexed in Medline, an additional search for each pathogen was carried out using the internet browser ‘Google’. This enabled the identification of words or statements relating to the zoonotic pathogen, which have not been indexed, but may have been mentioned as a local name or term used by the scientific
community when referring to the pathogen. The references from the ‘Google’ search were manually screened to identify other words or phrases that have been used in the articles reviewed to describe the pathogen. After identifying key words and possible synonyms for each pathogen, a search query was developed by a Boolean combinatory ‘OR’ of the key words and synonyms. This was done for all the 10 pathogens. For example, the query developed for Rickettsia spp. was ‘Rickettsia’ (key word), and ‘Rickettsiae, Rickettsiosis, Typhus, Spotted fever, Orientia tsutsugamushi (Scrub typhus)’, as synonyms. Using an Excel filter tool, the search queries were used to identify all abstracts in the dataset tha t included either the keyword and/or any of the listed synonyms. The procedure was repeated for each pathogen and all identified abstracts were recorded.
Table 1: List of the 10 Pathogens and their synonyms Key search te rm relating
to the pathogen
Additional search team Leptospira Leptospirosis, Leptospires, Rat fever
Coxiella Q fever, C. burnetti
Brucella Brucellosis, Undulant fever, Malta fever, Rock fever Rickettsia Rickettsiae, Rickettsiosis, Typhus, Spotted fever, Orientia
tsutsugamushi (Scrub typhus) Hantavirus HPS, Pulmonary syndrome
West Nile Virus WNV
Borrelia Borreliosis, Lyme, (Lyme disease), Tick-borne relapsing fever (TBRF), Louse-borne relapsing fever (LBRF)
Francisella F. tularensis, Tularemia, Tularaemia, Rabbit fever
Salmonella Salmonellosis, non-typhoidal Salmonella (NTS) Rabies virus Rabies, Rabid
Manual screening
The abstracts that mentioned a zoonotic pathogen were manually screened to confirm investigation of a zoonotic pathogen. This process also enabled classification of abstracts, identification of abstracts that described risk factors and those that specifically referred to an animal group. The outcome of the manual screening was recorded with a further column created for each animal species. To validate the Excel filter process, 20% of randomly selected abstracts from the pool that did not mention a zoonotic pathogen were manually screened for confirmation. The manual screening procedure was also used to remove duplicate filtered abstracts.
Abstract classification
The second part of the first step involved classification of abstracts with confirmed pathogen inclusion. Abstracts were grouped based on whether the pathogen was the target of the study or mentioned in a more general context. The abstracts relating to studies that specifically investigated these zoonotic pathogens were further classified into three major categories: (a) country and geographical regions where the study was conducted; (b) journal of publication; (c) type of study design.
(a) Country and geographical regions where study was conducted
The country where the study was conducted was identified for each abstract. The main review focused on malaria endemic countries, which are located largely in Africa, Asia and Latin America and the Caribbean continents (WHO, 2015b). Countries were then grouped into regions and continents based on the current United Nations (UN) groupings (United Nations, 2016). At this stage, a further manual screening was carried out because, although the original inclusion criteria were for studies conducted in malaria endemic countries, several studies were included among the abstracts that had involved travellers who developed the zoonotic disease on returning to their home (non- malaria endemic) country. For these studies, the geographic region was classified according to the country where people had acquired the infection.
(b) Journal of publication
The journal of publication was classified into: Medical, Veterinary or Life Sciences based on the main theme of the journal. However, there are limitations in this classification because some journals may have more than one theme: specialist and general sections. For example, the journal of clinical microbiology (with life sciences as the main theme), which also has a dedicated section for veterinary related publicatio ns, would be classified under the main theme.
Medical Journal: This includes articles that were published in the journal of general
medicine and journals of specialist medicine.
Veterinary Journal: This includes articles published in the veterinary and other allied animal
health specialties.
Life Sciences: This includes journals of general life sciences, and others that could not be
classified as medical or veterinary.
The full list of journals where these studies were published, and the list of countries where zoonotic pathogens were reported are attached in the appendix.
(c) Type of study design
To allow examination of risk factors, abstracts were classified into the following study designs: case–series, cohort study, case-control, and cross-sectional study. The remaining studies, which included mainly diagnostic test evaluation, randomized controlled trials,
clinical trials and experimental studies were grouped together as a separate category (‘Other’). These classifications were assigned on the basis of how the study was described by the
authors in the abstract (Table 2). By definition and concept, case-control studies, cohort studies and cross-sectional studies, are expected to consider risk factors for zoonotic disease transmission and were further screened for animal related risk factors.
Table 2: Classification of abstracts by study design
Case-series A case series represents an observational study that reports data from a subject group without a comparison population (Dekkers et al., 2014; Vandenbroucke et al., 2007). In this review, case-series information was derived mostly from cases in health care settings, with all included individuals having the same case definition. Clinical cohorts where the study population was followed to allow sufficient time for clinical and laboratory diagnosis were classified in this category.
Cohort study The reviewed abstracts with cohort study design describe a study
population with similar exposure risks (e.g. sampled based on living in a rural area, agriculture workers, cleaners etc) and were followed over time starting from a baseline point. Those that developed febrile illness and other related symptoms of disease were further investigated for aetiology and relevant risk factors for zoonotic diseases and absolute risks calculated (Dekkers et al., 2014; Vandenbroucke et al., 2007).
Cross-sectional study
This group contains abstracts with descriptive and analytical study designs where demographical information of the sampled febrile population were collected at a defined time as detail of the cause of the febrile illness for each recruited subject. This category includes abstracts that describe cause of fever, source of exposures and risk factors, prevalence and/or burden of zoonotic pathogens in the febrile population.
Case-control study
This classification include abstracts where febrile population with specific case definition for the investigated disease was compared with a suitable control group without the zoonotic disease, but would have been included as cases if they had. The associations or risk factors for the disease
between the two groups were assessed.
Other For the RCT abstracts, each subject in the sampled febrile population was randomly allocated to a particular group (treatment being evaluated group or control/group with alternative treatment). Studies that compared the sensitivity and specificity and effectiveness of different diagnostic methods to identify zoonotic pathogens in a febrile population were included in this group and other types of study not classified above.