Chau Minh Bui

A Systematic Review of the Comparative Epidemiology of Avian and Human Influenza A H5N1 and H7N9 – Lessons and Unanswered Questions

The aim of this work was to explore the comparative epidemiology of influenza viruses, H5N1 and H7N9, in both bird and human populations. Specifically, the article examines similarities and differences between the two viruses in their genetic characteristics, distribution patterns in human and bird populations and postulated mechanisms of global spread. In summary, H5N1 is pathogenic in birds, while H7N9 is not. Yet both have caused sporadic human cases, without evidence of sustained, human-to-human spread. The number of H7N9 human cases in the first year following its emergence far exceeded that of H5N1 over the same time frame. Despite the higher incidence of H7N9, the spatial distribution of H5N1 within a comparable time frame is considerably greater than that of H7N9, both within China and globally. The pattern of spread of H5N1 in humans and birds around the world is consistent with spread through wild bird migration and poultry trade activities. In contrast, human cases of H7N9 and isolations of H7N9 in birds and the environment have largely occurred in a number of contiguous provinces in south-eastern China. Although rates of contact with birds appear to be similar in H5N1 and H7N9 cases, there is a predominance of incidental contact reported for H7N9 as opposed to close, high-risk contact for H5N1. Despite the high number of human cases of H7N9 and the assumed transmission being from birds, the corresponding level of H7N9 virus in birds in surveillance studies has been low, particularly in poultry farms. H7N9 viruses are also diversifying at a much greater rate than H5N1 viruses. Analyses of certain H7N9 strains demonstrate similarities with engineered transmissible H5N1 viruses which make it more adaptable to the human respiratory tract. These differences in the human and bird epidemiology of H5N1 and H7N9 raise unanswered questions as to how H7N9 has spread, which should be investigated further.

An overview of the epidemiology and emergence of influenza A infection in humans over time

In recent years multiple novel influenza A strains have emerged in humans. We reviewed publically available data to summarise epidemiological characteristics of distinct avian influenza viruses known to cause human infection and describe changes over time. Most recently identified zoonotic strains have emerged in China (H7N9, H5N6, H10N8) – these strains have occurred mostly in association with visiting a live bird market. Most zoonotic AIVs and swine influenza variants typically cause mild infections in humans however severe illness and fatalities are associated with zoonotic H5N6, H10N8, H7N9 and H5N1 serotypes, and the H1N1 1918 Spanish Influenza. The changing landscape of avian influenza globally indicates a need to reassess the risk of a pandemic influenza outbreak of zoonotic origin.

Influenza A H5N1 and H7N9 in China: A spatial risk analysis

Background Zoonotic avian influenza poses a major risk to China, and other parts of the world. H5N1 has remained endemic in China and globally for nearly two decades, and in 2013, a novel zoonotic influenza A subtype H7N9 emerged in China. This study aimed to improve upon our current understanding of the spreading mechanisms of H7N9 and H5N1 by generating spatial risk profiles for each of the two virus subtypes across mainland China. Methods and findings In this study, we (i) developed a refined data set of H5N1 and H7N9 locations with consideration of animal/animal environment case data, as well as spatial accuracy and precision; (ii) used this data set along with environmental variables to build species distribution models (SDMs) for each virus subtype in high resolution spatial units of 1km2 cells using Maxent; (iii) developed a risk modelling framework which integrated the results from the SDMs with human and chicken population variables, which was done to quantify the risk of zoonotic transmission; and (iv) identified areas at high risk of H5N1 and H7N9 transmission. We produced high performing SDMs (6 of 8 models with AUC > 0.9) for both H5N1 and H7N9. In all our SDMs, H7N9 consistently showed higher AUC results compared to H5N1, suggesting H7N9 suitability could be better explained by environmental variables. For both subtypes, high risk areas were primarily located in south-eastern China, with H5N1 distributions found to be more diffuse and extending more inland compared to H7N9. Conclusions We provide projections of our risk models to public health policy makers so that specific high risk areas can be targeted for control measures. We recommend comparing H5N1 and H7N9 prevalence rates and survivability in the natural environment to better understand the role of animal and environmental transmission in human infections.

Rolling epidemic of Legionnaires' disease outbreaks in small geographic areas

Legionnaires’ disease (LD) is reported from many parts of the world, mostly linked to drinking water sources or cooling towers. We reviewed two unusual rolling outbreaks in Sydney and New York, each clustered in time and space. Data on these outbreaks were collected from public sources and compared to previous outbreaks in Australia and the US. While recurrent outbreaks of LD over time linked to an identified single source have been described, multiple unrelated outbreaks clustered in time and geography have not been previously described. We describe unusual geographic and temporal clustering of Legionella outbreaks in two cities, each of which experienced multiple different outbreaks within a small geographic area and within a short timeframe. The explanation for this temporal and spatial clustering of LD outbreaks in two cities is not clear, but climate variation and deteriorating water sanitation are two possible explanations. There is a need to critically analyse LD outbreaks and better understand changing trends to effectively prevent disease.

Quantified degree of poultry exposure differs for human cases of avian influenza H5N1 and H7N9

Preliminary evidence suggests that direct poultry contact may play a lesser role in transmission of avian influenza A(H7N9) than A(H5N1) to humans. To better understand differences in risk factors, we quantified the degree of poultry contact reported by H5N1 and H7N9 World Health Organization-confirmed cases. We used publicly available data to classify cases by their degree of poultry contact, including direct and indirect. To account for potential data limitations, we used two methods: (1) case population method in which all cases were classified using a range of sources; and (2) case subset method in which only cases with detailed contact information from published research literature were classified. In the case population, detailed exposure information was unavailable for a large proportion of cases (H5N1, 54%; H7N9, 86%). In the case subset, direct contact proportions were higher in H5N1 cases (70·3%) than H7N9 cases (40·0%) (χ 2 = 18·5, P < 0·001), and indirect contact proportions were higher in H7N9 cases (44·6%) than H5N1 cases (19·4%) (χ 2 = 15·5, P < 0·001). Together with emerging evidence, our descriptive analysis suggests direct poultry contact is a clearer risk factor for H5N1 than for H7N9, and that other risk factors should also be considered for H7N9.

Highly Pathogenic Avian Influenza Virus, Midwestern United States.

To the Editor: Novel highly pathogenic avian influenza (HPAI) viruses of subtypes H5N2, H5N8, and H5N1 have recently caused numerous outbreaks in commercial poultry farms in the United States and Canada (1). Risk for zoonotic transmission is low; humans are affected primarily from the extensive economic repercussions of suspending poultry-farming activities (1). Large-scale research is under way, including case-control studies of infections on poultry farms and modeling studies to investigate the spread of virus in waterfowl (1,2). The US Department of Agriculture has published a report that summarizes various biosecurity measures of affected farms, results of airborne pathogen testing, and geospatial analyses correlating wind speed and direction to outbreaks (1). These studies found insufficient evidence to support any particular modes of virus spread and suggest that farms are contaminated from infected migrating waterfowl and/or unauthorized movements (e.g., of vehicles, equipment, persons, or animals) between farms and that unusually high wind speeds are the likely mechanism of spread (1). The spread from farm to farm, but not from barn to barn within a single farm (3), further adds to the puzzle of how infection has been transmitted. To better understand the outbreak behavior, we used publicly available sources (4–6) to create maps of outbreaks of HPAI virus, subtype H5, infections in relation to poultry distribution and wild bird migratory patterns (Figure; Technical Appendix Figures 1, 2; Video). From November 30, 2014, through June 17, 2015, a total of 280 outbreaks caused by HPAI virus subtype H5 in Canada and the United States were reported to the World Organisation for Animal Health (4). Most outbreaks occurred during April (n = 116) in commercial turkey farms (n = 154) and were caused by HPAI virus subtype H5N2 (n = 256) (Technical Appendix Figure 3). Related reassortant HPAI subtypes H5N8 and H5N1 were also found among infected poultry; however, these appeared infrequently. Subtype H5N1 appeared in 4 of 21 outbreaks in backyard and commercial farms and was found in 1 of 3 infections in a backyard farm. Backyard farms generally contain flocks for local consumption and implement fewer biosecurity measures (4). Figure Distribution of outbreaks caused by highly pathogenic avian influenza (HPAI) virus, subtype H5, in domestic poultry compared with domestic poultry flock density and direction of wild waterfowl migration. Triangles represent outbreaks caused by HPAI virus, ... Video Time-series map showing the cumulative daily geographic distribution of highly pathogenic avian influenza (HPAI) virus, subtype H5, events in North America, from November 30, 2014, to May 18, 2015. Events are symbolized by triangles. The different colors ... Initial outbreaks on poultry farms that began in November 2014, near the British Columbia–Washington State border, have been associated with timing of waterfowl migration and reported infection in migratory waterfowl (7,8). Subsequent surveillance of avian influenza virus in wild birds in the Pacific flyway has also shown sporadic infections caused by HPAI virus subtype H5, primarily in waterfowl species of the family Anseriformes (4) (Technical Appendix Table 1). In late February 2015, however, HPAI virus subtype H5, emerged in US midwestern states, leading to a substantial number of outbreaks in commercial poultry farms in the region. The spread from west to east does not correlate with the direction of typical waterfowl migration, in which movement occurs from south to north. Unlike the earlier outbreaks in poultry in Canada, in the outbreaks in midwestern states, corresponding high numbers of virus were not detected in samples of wild birds in surrounding regions (despite increased surveillance). Of 3,300 samples tested, 1 sample tested positive for HPAI virus subtype H5 (4,9). In addition, most poultry farms were affected in April, and migratory waterfowl typically appear in Minnesota in March and April (Technical Appendix Figure 1). This February introduction of virus to Minnesota may be explained by an earlier-than-usual spring (10). Minnesota and Iowa lie within regions where migrating waterfowl spend their breeding season, and waterfowl densities on commercial poultry farms are particularly high (Technical Appendix Figure 2). In southern parts of the United States, where poultry density is also high, isolated outbreaks of HPAI have occurred in poultry, although the introduction of virus into these regions did not result in a surge of outbreaks. The timing of waterfowl migration enables the mixing of highly dense populations of wild waterfowl and poultry, which likely plays a key role in spreading virus onto farms. Of particular note, outbreaks in poultry were densely concentrated within Minnesota and Iowa in a spatial pattern inconsistent with the much more geographically dispersed spread of infection in wild birds. The magnitude and clustered distribution of poultry outbreaks are suggestive of local spread, rather than multiple introductions from passing migratory waterfowl. Genetic analyses have similarly shown evidence for concurrent multiple introductions as well as common source exposures, and surveys of affected farms have shown that local spread could be facilitated by the sharing of equipment by multiple farms or through animals entering barns (1). The combination of high poultry densities and timing of waterfowl migration have likely predisposed Minnesota and Iowa to outbreaks of avian influenza among poultry flocks. However, consistent with US Department of Agriculture findings, local factors have likely also contributed to the large number of outbreaks in these states. We suggest that network modeling analyses would be valuable in exploring how virus may spread from farm to farm.

A systematic review of early modelling studies of Ebola virus disease in West Africa.

Phenomenological and mechanistic models are widely used to assist resource planning for pandemics and emerging infections. We conducted a systematic review, to compare methods and outputs of published phenomenological and mechanistic modelling studies pertaining to the 2013-2016 Ebola virus disease (EVD) epidemics in four West African countries - Sierra Leone, Liberia, Guinea and Nigeria. We searched Pubmed, Embase and Scopus databases for relevant English language publications up to December 2015. Of the 874 articles identified, 41 met our inclusion criteria. We evaluated these selected studies based on: the sources of the case data used, and modelling approaches, compartments used, population mixing assumptions, model fitting and calibration approaches, sensitivity analysis used and data bias considerations. We synthesised results of the estimated epidemiological parameters: basic reproductive number (R 0), serial interval, latent period, infectious period and case fatality rate, and examined their relationships. The median of the estimated mean R 0 values were between 1·30 and 1·84 in Sierra Leone, Liberia and Guinea. Much higher R 0 value of 9·01 was described for Nigeria. We investigated several issues with uncertainty around EVD modes of transmission, and unknown observation biases from early reported case data. We found that epidemic models offered R 0 mean estimates which are country-specific, but these estimates are not associating with the use of several key disease parameters within the plausible ranges. We find simple models generally yielded similar estimates of R 0 compared with more complex models. Models that accounted for data uncertainty issues have offered a higher case forecast compared with actual case observation. Simple model which offers transparency to public health policy makers could play a critical role for advising rapid policy decisions under an epidemic emergency.

Publicly available software tools for decision-makers during an emergent epidemic—Systematic evaluation of utility and usability

Epidemics and emerging infectious diseases are becoming an increasing threat to global populations-challenging public health practitioners, decision makers and researchers to plan, prepare, identify and respond to outbreaks in near real-timeframes. The aim of this research is to evaluate the range of public domain and freely available software epidemic modelling tools. Twenty freely utilisable software tools underwent assessment of software usability, utility and key functionalities. Stochastic and agent based tools were found to be highly flexible, adaptable, had high utility and many features, but low usability. Deterministic tools were highly usable with average to good levels of utility.

Pandemics, public health emergencies and antimicrobial resistance - putting the threat in an epidemiologic and risk analysis context

Public health messaging about antimicrobial resistance (AMR) sometimes conveys the problem as an epidemic. We outline why AMR is a serious endemic problem manifested in hospital and community-acquired infections.AMR is not an epidemic condition, but may complicate epidemics, which are characterised by sudden societal impact due to rapid rise in cases over a short timescale. Influenza, which causes direct viral effects, or secondary bacterial complications is the most likely cause of an epidemic or pandemic where AMR may be a problem. We discuss other possible causes of a pandemic with AMR, and present a risk assessment formula to estimate the impact of AMR during a pandemic. Finally, we flag the potential impact of genetic engineering of pathogens on global risk and how this could radically change the epidemiology of AMR as we know it.Understanding the epidemiology of AMR is key to successfully addressing the problem. AMR is an endemic condition but can play a role in epidemics or pandemics, and we present a risk analysis method for assessing the impact of AMR in a pandemic.

Adherence to anti-vectorial prevention measures among travellers with chikungunya and malaria returning to Australia: comparative epidemiology

ObjectiveCompare the adoption and adherence to health protection behaviours prior to and during travel among international Australian travellers who return to Australia with notified chikungunya or malaria infection. This information could inform targeted health promotion and intervention strategies to limit the establishment of these diseases within Australia.ResultsSeeking travel advice prior to departure was moderate (46%, N = 21/46) yet compliance with a range of recommended anti-vectorial prevention measures was low among both chikungunya and malaria infected groups (16%, N = 7/45). Reasons for not seeking advice between groups was similar and included ‘previous overseas travel with no problems’ (45%, N = 9/20) and ‘no perceived risk of disease’ (20%, N = 4/20). Most chikungunya cases (65%, N = 13/20) travelled to Indonesia and a further 25% (N = 5/20) visited India, however most malaria cases (62%, N = 16/26) travelled to continental Africa with only 12% (N = 3/26) travelling to India. The majority (50%, N = 10/20) of chikungunya cases reported ‘holiday’ as their primary purpose of travel, compared to malaria cases who most frequently reported travel to visit friends and family (VFR; 42%, N = 11/26). These results provide import data that may be used to support distinct public health promotion and intervention strategies of two important vector-borne infectious diseases of concern for Australia.