Search results
During the selection process, Scopus yielded 1712 records, Web of Science Core Collection 1557 records, PubMed 1512 records and SpringerLink 2242 results. All these added to 7023 articles, of which 2198 were duplicates, leaving a total of 4825 distinct reports. ProMED mail had an additional 198 records fitting our research strategy criteria.
Study selection
The study selection process is summarised in a flow diagram (Fig. 1). As mentioned, the search provided initially 7023 citations. After eliminating duplicates, 5023 remained. After the initial review of all these studies by the first and last author and after subsequent discussion and consensus of the remaining authors, 2113 articles/records were discarded after screening, both the title and the abstract while 2572 articles/records were discarded as they did not meet the eligibility criteria, after a full-text screening or/and during the data extraction process. Three additional studies were screened, which met the eligibility criteria but were discarded as the full texts were not available. A total of 173 publication articles from the electronic databases mentioned, and 198 records from ProMED mail were eventually deemed eligible for inclusion in this systematic review. Unpublished related literature was not included.
Assessment of risk of bias in included studies
The overall agreement percentage was calculated and considered as substantial for all sample, analyses and measurement criteria. Only 13 of the included studies received the maximum score for all criteria.
Characteristics of included studies
The greatest number of studies and records (144, 22.2%) occurred in year 2016, while 142 (21.9%) occurred in 2015, 105 (16.2%) in 2014 and 82 (12.6%), 56 (8.6%), 67 (10.3%) and 52 (8%) in 2013, 2012, 2011 and 2010, respectively. Taking into account the outbreaks outsourced by ProMED mail, which were instantly (real-time) announced, most of the records (143, 27.3%) were published in 2016, 134 (25.6%) in 2015, 74 (14.1%) in 2014, 52 (9.9%) in 2013, 33 (6.3%) in 2012, 47 (9%) in 2011 and 39 (7.4%) in 2010.
Regarding the “country of epidemics”, we estimated that the largest number of outbreaks (88, 13.6%) occurred in China, while 53 (8.2%), 32 (4.9%), 29 (4.5%), 27 (4.2%), 25 (3.9%) and 23 (3.5%) occurred in Viet Nam, Egypt and Germany, India and the USA, Taiwan, the Netherlands and Israel, respectively. AIV subtypes recorded per country are presented in Fig. 2. Subsequently, we conducted a world spot map, based on outbreaks recorded in each country during 2010–2016 [25] (Fig. 3). The largest outbreak reported in the scientific literature and ProMED mail took place in Jalisco, Mexico. This outbreak occurred in 2012. During this outbreak, it was estimated that 3,987,160 birds from commercial poultry species were infected by HPAI H7N3.
AIV cases
The studies and records (ProMED mail) included a total of 58,709,463 individual birds. Most of the outbreaks (364), which are included in this systematic review, originated from commercial poultry farms (56.1%), followed by wild bird species (103 outbreaks, 15.9%), backyard domestic poultry (87 outbreaks, 13.4%), mixed (commercial and wild) species (26 outbreaks, 4%), live poultry market species (17 outbreaks, 2.6%), live bird market species (16 outbreaks, 2.5%), village species (11 outbreaks, 1.7%), natural park species (7 outbreaks, 1.1%), wet market species (5 outbreaks, 0.8%), zoo species (4 outbreaks, 0.6%), wild habitat species (3 outbreaks, 0.5%), slaughterhouse species (1 outbreak, 0.2%) and finally, game birds (1 outbreak, 0.2%).
The flock size ranged between 14 and 7,498,221 but was not available in almost half (292 out of 649) of the outbreaks included. In 44.99% of the outbreaks, the flock size was not reported either because there was no need as the data was referring to wild species or because this information was indeed missing.
In only 138 outbreaks (21.3%), a vaccination programme was mentioned; however, no access to more precise information was feasible concerning the subtypes covered, the doses received, etc. Regarding the administration of the surveillance programme at the time of the outbreak included, only in 252 outbreaks (38.8%), which were assessed, it was reported that a surveillance programme was implemented, without precisely mentioning the syllabus and the format of the programme (age, species, subtypes covered, etc.).
The bird age was between 20 and 420 days; however, the factor “age” was mentioned only in 5.39% of the studies and records (35) included; 94.6% of the outbreaks failed to outline the age, and for this reason, the value age could not be reliably evaluated. Among those mentioning the specific case age, 61.8% were aged between 0 and 100 days old, 29.4% between 101 and 200 days old and 2.9% between 301 and 400, 201 and 300 and > 401 days old.
The symptoms of AIV were not mentioned in 58.4% of the outbreaks, while 16.9% outlined mixed symptoms (a drop in feed and water intake, a drop in egg production, lack of vocalisation, depression, mortality, coughing, disability in breathing, respiratory signs, fever), 15.7% of the cases mentioned mortality of the birds, 2.3% of the cases referred symptoms of the respiratory system, 2.2% of the cases cited other symptoms and 0.2% of the cases pointed out symptoms of the reproductive system and general symptoms.
The sample type tested was missing in almost 63% of the outbreaks. Whenever stated, 36.7% were characterised as mixed (blood, oropharyngeal, conjunctiva, pharynx and faecal swabs, tissue, faeces, carcass), 35.4% of the samples were stated as “other” (like feathers, due to feather tropism of specific AIV subtypes noted in experimentally infected avian species) [26], 20.4% were swabs (cloaca, pharynx and conjunctiva), 3.8% blood samples, 2.5% faeces and 1.3% specific tissue.
Information regarding “vicinity to water” in relation to the outbreak area was missing in 75.5% of the outbreaks. However, whenever vicinity to water was mentioned, water was present in 97.5% of the cases; vicinity to water is considered as a significant risk factor because it is regarded as a wild species habitat for most of the cases [27]. The subtypes reported in outbreaks, where vicinity of water was mentioned, were H5N1 (43 outbreaks, 28.5%), H5N8 (34 outbreaks, 22.5%), mixed (24 outbreaks,15.9%), H5 (18 outbreaks, 11.9%), H5N6 (9 outbreaks, 6%), non-specified subtype (5 outbreaks, 3.3%), H7N9 (5 outbreaks, 3.3%), H3N8 (2 outbreaks, 1.3%), H10N7 (1 outbreak, 0.7%), H5N9 (1 outbreak, 0.7%), H11N9 (1 outbreak, 0.7%), H1N2 (1 outbreak, 0.7%), H7N2 (1 outbreak, 0.7%), H7N7 (1 outbreak, 0.7%), H4N6 (1 outbreak, 0.7%), H9N2 (1 outbreak, 0.7%) and H5N2 (1 outbreak, 0.7%).
Almost all (97.68%) scientific articles and records found in ProMED mail mentioned the type of testing employed to document AIV infection; 62.6% of the samples were identified by advanced laboratory testing [namely real-time polymerase chain reaction (real-time PCR), real-time reverse transcriptase/polymerase chain reaction (RRT-PCR), virus isolation, virus sequencing], 31.1% were identified by molecular methods, 3.5% of samples were tested with both serological and molecular methods (in most cases, screening was prepared with serological methods and verification for positive and suspect samples was confirmed with molecular methods), 2.5% were tested with serological methods only (enzyme-linked immuno-sorbent assay, Immunoblot) and 0.3% with specific pathogen-free embryonate eggs.
Only 3.5% of the included articles did not specify the AIV subtype isolated. AIV subtypes investigation during 2010 to 2016 is clearly demonstrated in Fig. 4, where the presence of H5N1, H5N2 and H5 and other subtypes remains constant and strong. The most often (229 outbreaks, 38.2%) isolated AIV subtype was H5N1, followed by H5N8 (78 outbreaks, 13%), H5 (61 outbreaks, 10.2%), H5N2 (49 outbreaks, 8.2%), mixed subtypes (33 outbreaks, 5.5%), H5N6 (26 outbreaks, 4.3%), H7N9 (25 outbreaks, 3.9%), H7N7 (18 outbreaks, 3%), H7N3 (13 outbreaks, 2.2%), H9N2 (8 outbreaks, 1.3%), H7 (8 outbreaks, 1.3%), H7N1 (7 outbreaks, 1.2%), H7N2 (4 outbreaks, 0.7%), H5N9 (4 outbreaks, 0.7%), H5N3 (3 outbreaks, 0.5%), H3N8 (2 outbreaks, 0.3%), H5N5 (2 outbreaks, 0.3%), H3N2 (1 outbreak, 0.2%), H10 (1 outbreak, 0.2%), H1N2 (1 outbreak, 0.2%), H7N6 (1 outbreak, 0.2%), H4N6 (1 outbreak, 0.2%), H10N7 (1 outbreak, 0.2%),, H9N1 (1 outbreak, 0.2%), H1N1 (1 outbreak, 0.2%) and H11N9 (1 outbreak, 0.2%). The distribution between patterns of HPAI and LPAI infection in bird populations is summarised in Fig. 5. In almost 82.5% of wild bird outbreaks HPAI, was recorded, followed by 9.7% of LPAI, 6.8% not mentioned and 1% mixed outbreaks. Concerning commercial poultry farms, 73.1% were HPAI, 21.7% LPAI and 5.2% were not mentioned. A more detailed case and quantitative approach, namely the precise number of cases of each AIV subtype upon the country, epidemiological unit and year is summarised in (see Additional file 1: Table S1). AIV subtype distribution upon avian species is summarised in Fig. 6, where commercial poultry seems to be “hit” more than any other species and mostly by H5N1 subtype, while wild birds were, also, mostly “hit” by H5N1, but also by various other AIV subtypes. Notably, H5N8 subtype entered Europe with its first appearance in Germany.