Optimising predictive modelling of Ross River virus using meteorological variables

BackgroundStatistical models are regularly used in the forecasting and surveillance of infectious diseases to guide public health. Variable selection assists in determining factors associated with disease transmission, however, often overlooked in this process is the evaluation and suitability of the statistical model used in forecasting disease transmission and outbreaks. Here we aim to evaluate several modelling methods to optimise predictive modelling of Ross River virus (RRV) disease notifications and outbreaks in epidemiological important regions of Victoria and Western Australia.Methodology/Principal findingsWe developed several statistical methods using meteorological and RRV surveillance data from July 2000 until June 2018 in Victoria and from July 1991 until June 2018 in Western Australia. Models were developed for 11 Local Government Areas (LGAs) in Victoria and seven LGAs in Western Australia. We found generalised additive models and generalised boosted regression models, and generalised additive models and negative binomial models to be the best fit models when predicting RRV outbreaks and notifications, respectively. No association was found with a model’s ability to predict RRV notifications in LGAs with greater RRV activity, or for outbreak predictions to have a higher accuracy in LGAs with greater RRV notifications. Moreover, we assessed the use of factor analysis to generate independent variables used in predictive modelling. In the majority of LGAs, this method did not result in better model predictive performance.Conclusions/SignificanceWe demonstrate that models which are developed and used for predicting disease notifications may not be suitable for predicting disease outbreaks, or vice versa. Furthermore, poor predictive performance in modelling disease transmissions may be the result of inappropriate model selection methods. Our findings provide approaches and methods to facilitate the selection of the best fit statistical model for predicting mosquito-borne disease notifications and outbreaks used for disease surveillance.     

Application of space technologies to the surveillance and modelling of waterborne diseases

Earth observing satellites, global positioning and geographic information systems are new tools that currently enable the scientific community to integrate ecological, environmental and medical data to develop predictive models for disease surveillance and modelling. A number of investigators have explored remotely sensed environmental factors that might be associated with waterborne disease ecology and human transmission risk. However, health specialists have not been fully familiarized with the capabilities of space technology, and in some cases it has not proved to be the wonder tool that scientists expected. New satellite capabilities and new sensors now allow exploration of risk factors previously beyond the capabilities of remote sensing and put researchers in a position to analyze the effects of environment on disease outbreaks.     

Suitability and limitations of portion-specific abattoir data as part of an early warning system for emerging diseases of swine in Ontario

Background  

Abattoir data have the potential to provide information for geospatial disease surveillance applications, but the quality of the data and utility for detecting disease outbreaks is not well understood. The objectives of this study were to 1) identify non-disease factors that may bias these data for disease surveillance and 2) determine if major disease events that took place during the study period would be captured using multi-level modelling and scan statistics. We analyzed data collected at all provincially-inspected abattoirs in Ontario, Canada during 2001-2007. During these years there were outbreaks of porcine circovirus-associated disease (PCVAD), porcine reproductive and respiratory syndrome (PRRS) and swine influenza that produced widespread disease within the province. Negative binomial models with random intercepts for abattoir, to account for repeated measurements within abattoirs, were created. The relationships between partial carcass condemnation rates for pneumonia and nephritis with year, season, agricultural region, stock price, and abattoir processing capacity were explored. The utility of the spatial scan statistic for detecting clusters of high partial carcass condemnation rates in space, time, and space-time was investigated.     

Syndromic Surveillance for Local Outbreaks of Lower-Respiratory Infections: Would It Work?

Background

Although syndromic surveillance is increasingly used to detect unusual illness, there is a debate whether it is useful for detecting local outbreaks. We evaluated whether syndromic surveillance detects local outbreaks of lower-respiratory infections (LRIs) without swamping true signals by false alarms.

Methods and Findings

Using retrospective hospitalization data, we simulated prospective surveillance for LRI-elevations. Between 1999–2006, a total of 290762 LRIs were included by date of hospitalization and patients place of residence (>80% coverage, 16 million population). Two large outbreaks of Legionnaires disease in the Netherlands were used as positive controls to test whether these outbreaks could have been detected as local LRI elevations. We used a space-time permutation scan statistic to detect LRI clusters. We evaluated how many LRI-clusters were detected in 1999–2006 and assessed likely causes for the cluster-signals by looking for significantly higher proportions of specific hospital discharge diagnoses (e.g. Legionnaires disease) and overlap with regional influenza elevations. We also evaluated whether the number of space-time signals can be reduced by restricting the scan statistic in space or time. In 1999–2006 the scan-statistic detected 35 local LRI clusters, representing on average 5 clusters per year. The known Legionnaires'' disease outbreaks in 1999 and 2006 were detected as LRI-clusters, since cluster-signals were generated with an increased proportion of Legionnaires disease patients (p:<0.0001). 21 other clusters coincided with local influenza and/or respiratory syncytial virus activity, and 1 cluster appeared to be a data artifact. For 11 clusters no likely cause was defined, some possibly representing as yet undetected LRI-outbreaks. With restrictions on time and spatial windows the scan statistic still detected the Legionnaires'' disease outbreaks, without loss of timeliness and with less signals generated in time (up to 42% decline).

Conclusions

To our knowledge this is the first study that systematically evaluates the performance of space-time syndromic surveillance with nationwide high coverage data over a longer period. The results show that syndromic surveillance can detect local LRI-outbreaks in a timely manner, independent of laboratory-based outbreak detection. Furthermore, since comparatively few new clusters per year were observed that would prompt investigation, syndromic hospital-surveillance could be a valuable tool for detection of local LRI-outbreaks.     

Challenges to global surveillance and response to infectious disease outbreaks of international importance

This article presents a notional scheme of global surveillance and response to infectious disease outbreaks and reviews 14 international surveillance and response programs. In combination, the scheme and the programs illustrate how, in an ideal world and in the real world, infectious disease outbreaks of public health significance could be detected and contained. Notable practices and achievements of the programs are cited; these may be useful when instituting new programs or redesigning existing ones. Insufficiencies are identified in four critical areas: health infrastructure; scientific methods and concepts of operation; essential human, technical, and financial resources; and international policies. These insufficiencies challenge global surveillance of and response to infectious disease outbreaks of international importance. This article is intended to help policymakers appreciate the complexity of the problem and assess the impact and cost-effectiveness of proposed solutions. An assessment of the potential contribution of appropriate diagnostic tests to surveillance and response is included.     

Use of the internet to enhance infectious disease surveillance and outbreak investigation

Modernization of electronic communication systems to facilitate infectious disease surveillance and outbreak investigation became a priority after the 2001 anthrax attacks. However, the extent to which communicable disease investigators are using web-based information resources, e-mail notifications, or secure information exchange systems to facilitate surveillance is unknown. To address this question, we conducted a survey in 2004 of state and local communicable disease investigators responsible for infectious disease surveillance and outbreak investigation in three states. The majority (70.7%) of the 297 respondents accessed the Internet for information regarding infectious disease surveillance and outbreaks at least weekly. Most (74%) respondents who searched for information from the Centers for Disease Control and Prevention (CDC) website reported that they found what they were looking for 75-100% of the time, compared with 54% who found the information from their state health department websites 75-100% of the time. One-third of respondents read e-mail notifications regarding outbreaks under investigation in their state less frequently than monthly; 34% of those enrolled in CDC's Epidemic Information Exchange (Epi-X) read e-mail notifications of new reports less frequently than monthly. Forty-seven (18%) respondents read ProMED-mail at least monthly, while 46% indicated they had never consulted MEDLINE/PubMed. Some progress has been made in use of the Internet to facilitate communication in infectious disease surveillance and outbreak investigation. Addressing barriers to access and usability of new information systems in conjunction with training and technical support could enhance infectious disease surveillance and timely investigation of outbreaks and bioterrorism events.     

Bringing together emerging and endemic zoonoses surveillance: shared challenges and a common solution

Early detection of disease outbreaks in human and animal populations is crucial to the effective surveillance of emerging infectious diseases. However, there are marked geographical disparities in capacity for early detection of outbreaks, which limit the effectiveness of global surveillance strategies. Linking surveillance approaches for emerging and neglected endemic zoonoses, with a renewed focus on existing disease problems in developing countries, has the potential to overcome several limitations and to achieve additional health benefits. Poor reporting is a major constraint to the surveillance of both emerging and endemic zoonoses, and several important barriers to reporting can be identified: (i) a lack of tangible benefits when reports are made; (ii) a lack of capacity to enforce regulations; (iii) poor communication among communities, institutions and sectors; and (iv) complexities of the international regulatory environment. Redirecting surveillance efforts to focus on endemic zoonoses in developing countries offers a pragmatic approach that overcomes some of these barriers and provides support in regions where surveillance capacity is currently weakest. In addition, this approach addresses immediate health and development problems, and provides an equitable and sustainable mechanism for building the culture of surveillance and the core capacities that are needed for all zoonotic pathogens, including emerging disease threats.     

Prediction of Dengue Outbreaks Based on Disease Surveillance and Meteorological Data

Research is needed to create early warnings of dengue outbreaks to inform stakeholders and control the disease. This analysis composes of a comparative set of prediction models including only meteorological variables; only lag variables of disease surveillance; as well as combinations of meteorological and lag disease surveillance variables. Generalized linear regression models were used to fit relationships between the predictor variables and the dengue surveillance data as outcome variable on the basis of data from 2001 to 2010. Data from 2011 to 2013 were used for external validation purposed of prediction accuracy of the model. Model fit were evaluated based on prediction performance in terms of detecting epidemics, and for number of predicted cases according to RMSE and SRMSE, as well as AIC. An optimal combination of meteorology and autoregressive lag terms of dengue counts in the past were identified best in predicting dengue incidence and the occurrence of dengue epidemics. Past data on disease surveillance, as predictor alone, visually gave reasonably accurate results for outbreak periods, but not for non-outbreaks periods. A combination of surveillance and meteorological data including lag patterns up to a few years in the past showed most predictive of dengue incidence and occurrence in Yogyakarta, Indonesia. The external validation showed poorer results than the internal validation, but still showed skill in detecting outbreaks up to two months ahead. Prior studies support the fact that past meteorology and surveillance data can be predictive of dengue. However, to a less extent has prior research shown how the longer-term past disease incidence data, up to years, can play a role in predicting outbreaks in the coming years, possibly indicating cross-immunity status of the population.     

Medical Surveillance for Biological Terrorism Agents

The first recognition of a bioterrorist attack may be sick patients. Recognition and subsequent notification of an attack at the earliest possible moment will assist in rapidly instituting protective measures. New methods of medical surveillance can assist with rapid detection. Syndromic surveillance is the use of newly created or pre-existing databases to track the health of communities. The data can include medical and non-medical information, but typically contains non-specific indicators of health status. These systems are currently being tested and fielded in multiple locations, and nationwide systems are being developed in the United States. Retrospective and prospective analyses have demonstrated the ability of the data to detect disease outbreaks, but the systems have yet to be proven to lead to more effective interventions.     

Avoidable errors in the modelling of outbreaks of emerging pathogens, with special reference to Ebola

As an emergent infectious disease outbreak unfolds, public health response is reliant on information on key epidemiological quantities, such as transmission potential and serial interval. Increasingly, transmission models fit to incidence data are used to estimate these parameters and guide policy. Some widely used modelling practices lead to potentially large errors in parameter estimates and, consequently, errors in model-based forecasts. Even more worryingly, in such situations, confidence in parameter estimates and forecasts can itself be far overestimated, leading to the potential for large errors that mask their own presence. Fortunately, straightforward and computationally inexpensive alternatives exist that avoid these problems. Here, we first use a simulation study to demonstrate potential pitfalls of the standard practice of fitting deterministic models to cumulative incidence data. Next, we demonstrate an alternative based on stochastic models fit to raw data from an early phase of 2014 West Africa Ebola virus disease outbreak. We show not only that bias is thereby reduced, but that uncertainty in estimates and forecasts is better quantified and that, critically, lack of model fit is more readily diagnosed. We conclude with a short list of principles to guide the modelling response to future infectious disease outbreaks.     

Using Hawkes Processes to model imported and local malaria cases in near-elimination settings

Developing new methods for modelling infectious diseases outbreaks is important for monitoring transmission and developing policy. In this paper we propose using semi-mechanistic Hawkes Processes for modelling malaria transmission in near-elimination settings. Hawkes Processes are well founded mathematical methods that enable us to combine the benefits of both statistical and mechanistic models to recreate and forecast disease transmission beyond just malaria outbreak scenarios. These methods have been successfully used in numerous applications such as social media and earthquake modelling, but are not yet widespread in epidemiology. By using domain-specific knowledge, we can both recreate transmission curves for malaria in China and Eswatini and disentangle the proportion of cases which are imported from those that are community based.     

Can hyperparasitoids cause large‐scale outbreaks of insect herbivores?

Synchronous population fluctuations occur in many species and have large economic impacts, but remain poorly understood. Dispersal, climate and natural enemies have been hypothesized to cause synchronous population fluctuations across large areas. For example, insect herbivores cause extensive forest defoliation and have many natural enemies, such as parasitoids, that may cause landscape‐scale changes in density. Between outbreaks, parasitoid‐caused mortality of hosts/herbivores is high, but it drops substantially during outbreak episodes. Because of their essential role in regulating herbivore populations, we need to include parasitoids in spatial modelling approaches to more effectively manage insect defoliation. However, classic host‐parasitoid population models predict parasitoid density, and parasitoid density is difficult to relate to host‐level observations of parasitoid‐caused mortality. We constructed a novel model to study how parasitoids affect insect outbreaks at the landscape scale. The model represents metacommunity dynamics, in which herbivore regulation, colonisation and extinction are driven by interactions with the forest, primary parasitoids and hyperparasitoids. The model suggests that parasitoid spatial dynamics can produce landscape‐scale outbreaks. Our results propose the testable prediction that hyperparasitoid prevalence should increase just before the onset of an outbreak because of hyperparasitoid overexploitation. If verified empirically, hyperparasitoid distribution could provide a biotic indicator that an outbreak will occur.     

Genomic and bioinformatics analysis of HAdV-4, a human adenovirus causing acute respiratory disease: implications for gene therapy and vaccine vector development

Human adenovirus serotype 4 (HAdV-4) is a reemerging viral pathogenic agent implicated in epidemic outbreaks of acute respiratory disease (ARD). This report presents a genomic and bioinformatics analysis of the prototype 35,990-nucleotide genome (GenBank accession no. AY594253). Intriguingly, the genome analysis suggests a closer phylogenetic relationship with the chimpanzee adenoviruses (simian adenoviruses) rather than with other human adenoviruses, suggesting a recent origin of HAdV-4, and therefore species E, through a zoonotic event from chimpanzees to humans. Bioinformatics analysis also suggests a pre-zoonotic recombination event, as well, between species B-like and species C-like simian adenoviruses. These observations may have implications for the current interest in using chimpanzee adenoviruses in the development of vectors for human gene therapy and for DNA-based vaccines. Also, the reemergence, surveillance, and treatment of HAdV-4 as an ARD pathogen is an opportunity to demonstrate the use of genome determination as a tool for viral infectious disease characterization and epidemic outbreak surveillance: for example, rapid and accurate low-pass sequencing and analysis of the genome. In particular, this approach allows the rapid identification and development of unique probes for the differentiation of family, species, serotype, and strain (e.g., pathogen genome signatures) for monitoring epidemic outbreaks of ARD.     

Combinatorial decomposition of an outbreak signature

We use mathematically rigorous definitions of epidemiological concepts in order to derive a sequential combinatorial model of disease outbreak decomposition. We define the idea of a population specific 'disease signature' and use this in order to decompose and further understand outbreaks as incidents of spatial and temporal spread of disease exposure both in, and across, populations. This allows us to differentiate between different disease spread scenarios with a level of sensitivity that previous models were unable to provide. This perspective leads us to propose a new practical definition for 'outbreak'. In addition, we are able to use this model to understand, estimate, and, in some cases, correct for, the likely instances of reporting error inherent in disease surveillance. We demonstrate our model first with a hypothetical outbreak scenario and then in an analysis of suspected outbreaks of waterborne diseases in Massachusetts (MA) in 1995.     

Electronic outbreak surveillance in Germany: a first evaluation for nosocomial norovirus outbreaks

Background

In Germany, surveillance for infectious disease outbreaks is integrated into an electronic surveillance system. For 2007, the national surveillance database contains case-based information on 201,224 norovirus cases, three-quarters of which are linked to outbreaks. We evaluated the data quality of the national database in reflecting nosocomial norovirus outbreak (NNO) data available in 19 Hessian local public health authorities (LPHAs) and the influence of differences between LPHA''s follow-up procedures for laboratory notifications of Norovirus positive stool samples on outbreak underascertainment.

Methods

Data on NNO beginning in 2007 and notified to the 19 LPHAs were extracted from the national database, investigated regarding internal validity and compared to data collected from LPHAs for a study on NNO control. LPHAs were questioned whether they routinely contacted all persons for whom a laboratory diagnosis of norovirus infection was notified. The number of outbreaks per 1,000 hospital beds and the number of cases within NNOs for acute care and rehabilitation hospitals were compared between counties with and without complete follow-up.

Results

The national database contained information on 155 NNOs, including 3,115 cases. Cases were missed in the national database in 58 (37%) of the outbreaks. Information on hospitalisation was incorrect for an estimated 47% of NNO cases. Information on county of infection was incorrect for 24% (199/820) of cases being forwarded between LPHAs for data entry. Reported NNO incidence and number of NNO cases in acute care hospitals was higher in counties with complete follow-up (incidence-rate ratio (IRR) 2.7, 95% CI 1.4–5.7, p-value 0.002 and IRR 2.1, 95% CI 1.9–2.4, p-value 0.001, respectively).

Conclusions

Many NNOs are not notified by hospitals and differences in LPHA procedures have an impact on the number of outbreaks captured in the surveillance system. Forwarding of case-by-case data on Norovirus outbreak cases from the local to the state and national level should not be required.     

Comparison of Human and Animal Surveillance Data for H5N1 Influenza A in Egypt 2006–2011

Background

The majority of emerging infectious diseases are zoonotic (transmissible between animals and humans) in origin, and therefore integrated surveillance of disease events in humans and animals has been recommended to support effective global response to disease emergence. While in the past decade there has been extensive global surveillance for highly pathogenic avian influenza (HPAI) infection in both animals and humans, there have been few attempts to compare these data streams and evaluate the utility of such integration.

Methodology

We compared reports of bird outbreaks of HPAI H5N1 in Egypt for 2006–2011 compiled by the World Organisation for Animal Health (OIE) and the UN Food and Agriculture Organization (FAO) EMPRESi reporting system with confirmed human H5N1 cases reported to the World Health Organization (WHO) for Egypt during the same time period.

Principal Findings

Both human cases and bird outbreaks showed a cyclic pattern for the country as a whole, and there was a statistically significant temporal correlation between the data streams. At the governorate level, the first outbreak in birds in a season usually but not always preceded the first human case, and the time lag between events varied widely, suggesting regional differences in zoonotic risk and/or surveillance effectiveness. In a multivariate risk model, lower temperature, lower urbanization, higher poultry density, and the recent occurrence of a bird outbreak were associated with increased risk of a human case of HPAI in the same governorate, although the positive predictive value of a bird outbreak was low.

Conclusions

Integrating data streams of surveillance for human and animal cases of zoonotic disease holds promise for better prediction of disease risk and identification of environmental and regional factors that can affect risk. Such efforts can also point out gaps in human and animal surveillance systems and generate hypotheses regarding disease transmission.     

A space-time permutation scan statistic for disease outbreak detection

BackgroundThe ability to detect disease outbreaks early is important in order to minimize morbidity and mortality through timely implementation of disease prevention and control measures. Many national, state, and local health departments are launching disease surveillance systems with daily analyses of hospital emergency department visits, ambulance dispatch calls, or pharmacy sales for which population-at-risk information is unavailable or irrelevant.ConclusionIf such results hold up over longer study times and in other locations, the space–time permutation scan statistic will be an important tool for local and national health departments that are setting up early disease detection surveillance systems.     

Improving the Modeling of Disease Data from the Government Surveillance System: A Case Study on Malaria in the Brazilian Amazon

The study of the effect of large-scale drivers (e.g., climate) of human diseases typically relies on aggregate disease data collected by the government surveillance network. The usual approach to analyze these data, however, often ignores a) changes in the total number of individuals examined, b) the bias towards symptomatic individuals in routine government surveillance, and; c) the influence that observations can have on disease dynamics. Here, we highlight the consequences of ignoring the problems listed above and develop a novel modeling framework to circumvent them, which is illustrated using simulations and real malaria data. Our simulations reveal that trends in the number of disease cases do not necessarily imply similar trends in infection prevalence or incidence, due to the strong influence of concurrent changes in sampling effort. We also show that ignoring decreases in the pool of infected individuals due to the treatment of part of these individuals can hamper reliable inference on infection incidence. We propose a model that avoids these problems, being a compromise between phenomenological statistical models and mechanistic disease dynamics models; in particular, a cross-validation exercise reveals that it has better out-of-sample predictive performance than both of these alternative models. Our case study in the Brazilian Amazon reveals that infection prevalence was high in 2004–2008 (prevalence of 4% with 95% CI of 3–5%), with outbreaks (prevalence up to 18%) occurring during the dry season of the year. After this period, infection prevalence decreased substantially (0.9% with 95% CI of 0.8–1.1%), which is due to a large reduction in infection incidence (i.e., incidence in 2008–2010 was approximately one fifth of the incidence in 2004–2008).We believe that our approach to modeling government surveillance disease data will be useful to advance current understanding of large-scale drivers of several diseases.     

Highly Pathogenic Avian Influenza Virus among Wild Birds in Mongolia

Mongolia combines a near absence of domestic poultry, with an abundance of migratory waterbirds, to create an ideal location to study the epidemiology of highly pathogenic avian influenza virus (HPAIV) in a purely wild bird system. Here we present the findings of active and passive surveillance for HPAIV subtype H5N1 in Mongolia from 2005–2011, together with the results of five outbreak investigations. In total eight HPAIV outbreaks were confirmed in Mongolia during this period. Of these, one was detected during active surveillance employed by this project, three by active surveillance performed by Mongolian government agencies, and four through passive surveillance. A further three outbreaks were recorded in the neighbouring Tyva Republic of Russia on a lake that bisects the international border. No HPAIV was isolated (cultured) from 7,855 environmental fecal samples (primarily from ducks), or from 2,765 live, clinically healthy birds captured during active surveillance (primarily shelducks, geese and swans), while four HPAIVs were isolated from 141 clinically ill or dead birds located through active surveillance. Two low pathogenic avian influenza viruses (LPAIV) were cultured from ill or dead birds during active surveillance, while environmental feces and live healthy birds yielded 56 and 1 LPAIV respectively. All Mongolian outbreaks occurred in 2005 and 2006 (clade 2.2), or 2009 and 2010 (clade 2.3.2.1); all years in which spring HPAIV outbreaks were reported in Tibet and/or Qinghai provinces in China. The occurrence of outbreaks in areas deficient in domestic poultry is strong evidence that wild birds can carry HPAIV over at least moderate distances. However, failure to detect further outbreaks of clade 2.2 after June 2006, and clade 2.3.2.1 after June 2010 suggests that wild birds migrating to and from Mongolia may not be competent as indefinite reservoirs of HPAIV, or that HPAIV did not reach susceptible populations during our study.