Search

Introduction: Major depressive disorder (MDD) has a significant impact on global burden of disease. Complications in clinical management can occur when response to pharmacological modalities is considered inadequate and symptoms persist (treatment-resistant depression (TRD)). We aim to investigate inflammation, proxied by C-reactive protein (CRP) levels, and body mass index (BMI) as putative causal risk factors for depression and subsequent treatment resistance, leveraging genetic information to avoid confounding via Mendelian Randomization (MR). Methods: We used the European UK Biobank subcohort (n = 451, 025), the mental health questionnaire (MHQ) and clinical records. For treatment resistance, a previously curated phenotype based on GP records and prescription data was employed. We applied univariable and multivariable MR models to genetically predict the exposures and assess their causal contribution to a range of depression outcomes . We used weak and pleiotropy, robust estimation techniques within univariable, multivariable and mediation MR models in order to address our research question with maximum rigour. In addition, we developed a novel statistical procedure to apply pleiotropy robust multivariable MR to one sample data and employed an unfamiliar bootstrap procedure to accurately quantify estimate uncertainty in mediation analysis which outperforms standard approaches. Results: In univariable MR models, genetically predicted BMI was positively associated with depression outcomes, including MDD and TRD, with a larger magnitude in women and with age acting as a moderator of the effect of BMI on PHQ9 (p = 0.011). Multivariable MR analyses suggested an independent causal effect of BMI on TRD not through CRP. Our mediation analyses suggested that the effect of CRP on PHQ9 was partly mediated by BMI. Individuals with TRD (n = 2, 199) observationally had higher CRP and BMI compared with individuals with MDD alone and healthy controls. A positive causal association was noted for both the exposures, but in multivariable analyses only BMI retained statistical significance at the 5% level. Discussion: Our work supportst the assertion BMI exerts a causal effect on a range of clinical and questionnaire-based depression phenotypes, with the effect being stronger in women and in younger individuals. We show that this effect is independent of inflammation proxied by CRP levels as the effects of CRP do not persist when jointly estimated with BMI. This is consistent with previous evidence suggesting that overweight contributed to depression even in the absence of any metabolic consequences. It appears that BMI exerts an effect on TRD that persists when we account for BMI influencing MDD.

Positivity for islet autoantibodies is used for diagnosis of type 1 diabetes. However, the importance of the autoantibody level at diagnosis of type 1 diabetes is not clear. Here, we assessed the association of glutamate decarboxylase (GADA), islet antigen-2 (IA-2A) and zinc transporter 8 (ZnT8A) autoantibody levels, measured using radiobinding assays, on genetic and clinical characteristics at diagnosis of 1536 participants with diabetes who were positive for these autoantibodies. We show that GADA and IA-2A levels had bimodal distributions, but ZnT8A level did not. The comparison of genetic and clinical characteristics between high and low level categories showed high GADA level was associated with older age at diagnosis, female sex and HLA-DR3-DQ2, whereas high IA-2A level was associated with younger age of diagnosis, ZnT8A positivity and HLA-DR4-DQ8. We replicated our findings in an independent cohort of 427 people with type 1 diabetes where autoantibodies were measured using enzyme-linked immunosorbent assays. In conclusion, Islet autoantibody levels provide additional information over positivity in type 1 diabetes at diagnosis. The bimodality of islet autoantibody levels highlights the novel aspect of heterogeneity of type 1 diabetes which may have implications on prediction, treatment and prognosis.Islet autoantibodies are commonly used in the diagnosis and prediction of type 1 diabetes. They are well established as the biomarkers of the underlying autoimmune pathogenesis (1). Autoantibodies to islet cell antigen (ICA), glutamate decarboxylase (GADA), islet antigen-2 (IA-2A), insulin (IAA) and zinc transporter 8 (ZnT8A) are the most commonly used islet autoantibodies at diagnosis (2). As detectable islet autoantibodies overlap between health and disease, a test is usually considered positive for a given islet autoantibody when the antibody level is higher than a 97.5–99th centile of a control population (3; 4). In routine clinical practice, quantitative islet autoantibody results are usually interpreted as positive or negative, and the level of the islet autoantibody, is not thought to be clinically meaningful.Islet autoantibody levels may provide additional information over positivity in type 1 diabetes at diagnosis. Similar to type 1 diabetes, autoantibodies to a specific antigen are commonly used for diagnosis in many other autoimmune diseases (such as TSH receptor antibodies in Graves’ disease and Rheumatoid Factor and Citrullinated Protein in rheumatoid arthritis). For Graves’ disease and rheumatoid arthritis, along with autoantibody positivity for these antigens, autoantibody level at diagnosis is associated with disease severity, prognosis and treatment success (5; 6). Multiple studies have shown a role for islet autoantibody level in the prediction of onset of type 1 diabetes, those with a higher levels of IA-2A, IAA and ICA have an increased risk of developing type 1 diabetes in at-risk populations (1; 7-9). However, it is not clear if the islet autoantibody level at diagnosis of type 1 diabetes, in addition to its interpretation as ‘positive’, is associated with the clinical phenotype similar to other autoimmune diseases.In this study, we undertook an analysis of GADA, IA-2A and ZnT8A levels at diagnosis in a large cohort of participants with type 1 diabetes, assessing the association of islet autoantibody levels on genetic and clinical characteristics at diagnosis in people with type 1 diabetes.

BackgroundMissing data is a common problem in regression modelling. Much of the literature focuses on handling missing outcome variables, but there are also challenges when dealing with missing predictor information, particularly when trying to build prediction models for use in practice.MethodsWe develop a flexible Bayesian approach for handling missing predictor information in regression models. For prediction this provides practitioners with full posterior predictive distributions for both the missing predictor information and the outcome variable, conditional on the observed predictors. We apply our approach to a previously proposed treatment selection model for type 2 diabetes second-line therapies. Our approach combines a regression model and a Dirichlet process mixture model (DPMM), where the former defines the treatment selection model and the latter provides a flexible way to model the joint distribution of the predictors.ResultsWe show that under missing-completely-at-random (MCAR) and missing-at-random (MAR) assumptions (with respect to the missing predictors), the DPMM can model complex relationships between predictor variables, and predict missing values conditionally on existing information. We also demonstrate that in the presence of multiple missing predictors, the DPMM model can be used to explore which variable(s), if collected, could provide the most additional information about the likely outcome.ConclusionsOur approach can provide practitioners with supplementary information to aid treatment selection decisions in the presence of missing data, and can be readily extended to other types of response model.Key MessagesMissing predictor variables present a significant challenge when building and implementing prediction models in clinical practice.Removing individuals with missing information and performing a complete case analysis can lead to imprecision and bias. Multiple imputation approaches typically translate uncertainty through prediction model parameter standard errors, as opposed to a consistent joint probability model.Alternatively, a Bayesian approach using Dirichlet process mixture models (DPMMs) offers a flexible way to model complex joint distributions of predictor variables, which can be used to estimate posterior (predictive) distributions for the missing predictors, conditional on the observed predictors.Using a DPMM, in this way allows uncertainties around missing predictor data to be propagated through to a prediction model of interest using a Bayesian hierarchical framework. This allows prediction models to be developed using datasets with incomplete predictor information (assuming missing-completely-at-random/missing-at-random). Furthermore, predictions can be made on new individuals even if they have incomplete predictor information (under the same assumptions).This approach provides full posterior predictive probability distributions for both missing predictor variables and the outcome variable, allowing a wide range of probabilistic models outputs to be derived to support clinical decision making.

Gut microbiome disequilibrium is increasingly implicated in host fitness reductions, including for the economically important and disease‐challenged western honey bee Apis mellifera. In laboratory experiments, the antibiotic tetracycline, which is used to prevent American Foulbrood Disease in countries including the US, elevates honey bee mortality by disturbing the microbiome. It is unclear, however, how elevated individual mortality affects colony‐level fitness.We used an agent‐based model (BEEHAVE) and empirical data to assess colony‐level effects of antibiotic‐induced worker bee mortality, by measuring colony size. We investigated the relationship between the duration that the antibiotic‐induced mortality probability is imposed for and colony size.We found that when simulating antibiotic‐induced mortality of worker bees from just 60 days per year, up to a permanent effect, the colony is reduced such that tetracycline treatment would not meet the European Food Safety Authority's (EFSA) honey bee protection goals. When antibiotic mortality was imposed for the hypothetical minimal exposure time, which assumes that antibiotics only impact the bee's fitness during the recommended treatment period of 15 days in both spring and autumn, the colony fitness reduction was only marginally under the EFSA's threshold. Synthesis and Applications. Modelling colony‐level impacts of antibiotic treatment shows that individual honey bee worker mortality can lead to colony mortality. To assess the full impact, the persistence of antibiotic‐induced mortality in honey bees must be determined experimentally, in vivo. We caution that as the domestication of new insect species increases, maintaining healthy gut microbiomes is of paramount importance to insect health and commercial productivity. The recommendation from this work is to limit prophylactic use of antibiotics and to not exceed recommended treatment strategies for domesticated insects. This is especially important for highly social insects as excess antibiotic use will likely decrease colony growth and an increase in colony mortality., Modelling colony‐level impacts of antibiotic treatment shows that individual honey bee worker mortality can lead to colony mortality. To assess the full impact, the persistence of antibiotic‐induced mortality in honey bees must be determined experimentally, in vivo. We caution that as the domestication of new insect species increases, maintaining healthy gut microbiomes is of paramount importance to insect health and commercial productivity. The recommendation from this work is to limit prophylactic use of antibiotics and to not exceed recommended treatment strategies for domesticated insects. This is especially important for highly social insects as excess antibiotic use will likely decrease colony growth and an increase in colony mortality.

Background Variation in host attributes that influence their contact rates and infectiousness can lead some individuals to make disproportionate contributions to the spread of infections. Understanding the roles of such ‘superspreaders’ can be crucial in deciding where to direct disease surveillance and controls to greatest effect. In the epidemiology of bovine tuberculosis (bTB) in Great Britain, it has been suggested that a minority of cattle farms or herds might make disproportionate contributions to the spread of Mycobacterium bovis, and hence might be considered ‘superspreader farms’. Objectives and Methods We review the literature to identify the characteristics of farms that have the potential to contribute to exceptional values in the three main components of the farm reproductive number ‐ Rf: contact rate, infectiousness and duration of infectiousness, and therefore might characterize potential superspreader farms for bovine tuberculosis in Great Britain. Results Farms exhibit marked heterogeneity in contact rates arising from between‐farm trading of cattle. A minority of farms act as trading hubs that greatly augment connections within cattle trading networks. Herd infectiousness might be increased by high within‐herd transmission or the presence of supershedding individuals, or infectiousness might be prolonged due to undetected infections or by repeated local transmission, via wildlife or fomites. Conclusions Targeting control methods on putative superspreader farms might yield disproportionate benefits in controlling endemic bovine tuberculosis in Great Britain. However, real‐time identification of any such farms, and integration of controls with industry practices, present analytical, operational and policy challenges., In the epidemiology of bovine tuberculosis (bTB) in Great Britain, a minority of ‘superspreader farms’ might make disproportionate contributions to the maintenance and spread of infection. A minority of farms act as trading hubs that greatly augment connections in trading networks, herd infectiousness might be increased by the presence of supershedding individuals, and herd infectiousness might be prolonged due to undetected infections or repeated local transmission, via wildlife or fomites. Targeting control methods on putative superspreader farms might yield disproportionate benefits in controlling endemic bovine tuberculosis in Great Britain.

The cetacean conservationist is often faced with evaluating population trends from abundance data that are either sparse or recorded at different times in different years. The presence of diel or seasonal patterns in the data together with unplanned gaps is often problematic. Such data are typical of those obtained from static acoustic monitoring. We present a simple and transparent non-parametric trend evaluation method, ‘Paired Year Ratio Assessment (PYRA)’ that uses only whole days of data wherever they are present in each of successive pairs of periods of 365 days. We provide a quantitative comparison of the performance of PYRA with traditional generalised additive models (GAMS) and nonparametric randomisation tests that require a greater level of skill and experience for both application and interpretation. We conclude that PYRA is a powerful tool, particularly in the context of identifying population trends which is often the main aim of conservation-targeted acoustic monitoring.

Whether an infectious disease threat to wildlife arises from pathogen introduction or the increased incidence of an already-present agent informs mitigation policy and actions. The prior absence of a pathogen can be difficult to establish, particularly in free-living wildlife. Subsequent to the epidemic emergence of the fungus, Batrachochytrium salamandrivorans (Bsal), in mainland Europe in 2010 and prior to its detection in captive amphibians in the United Kingdom (UK), we tested archived skin swabs using a Bsal-specific qPCR. These samples had been collected in 2011 from 2409 wild newts from ponds across the UK. All swabs were negative for Bsal. Bayesian hierarchical modelling suggests that Bsal was absent from, or present at very low levels in, these ponds at the time of sampling. Additionally, surveillance of newt mortality incidents, 2013–2017, failed to detect Bsal. As this pathogen has been shown to be widespread in British captive amphibian collections, there is an urgent need to raise awareness of the importance of effective biosecurity measures, especially amongst people with captive amphibians, to help minimise the risk of Bsal spreading to the wild. Continued and heightened wild amphibian disease surveillance is a priority to provide an early warning system for potential incursion events.

Highlights • We analysed high resolution proximity networks and GPS locations of dairy cattle. • Contact rates varied markedly in time and space, and in relation to husbandry. • Contacts were more frequent and longer in buildings. • Few cows exhibited evidence of consistent relationships. • When given free access across the farm, cows showed greater variation in contacts., The nature of contacts between hosts can be important in facilitating or impeding the spread of pathogens within a population. Networks constructed from contacts between hosts allow examination of how individual variation might influence the spread of infections. Studying the contact networks of livestock species managed under different conditions can additionally provide insight into their influence on these contact structures. We collected high-resolution proximity and GPS location data from nine groups of domestic cattle (mean group size = 85) in seven dairy herds employing a range of grazing and housing regimes. Networks were constructed from cattle contacts (defined by proximity) aggregated by different temporal windows (2 h, 24 h, and approximately 1 week) and by location within the farm. Networks of contacts aggregated over the whole study were highly saturated but dividing contacts by space and time revealed substantial variation in cattle interactions. Cows showed statistically significant variation in the frequency of their contacts and in the number of cows with which they were in contact. When cows were in buildings, compared to being on pasture, contact durations were longer and cows contacted more other cows. A small number of cows showed evidence of consistent relationships but the majority of cattle did not. In one group where management allowed free access to all farm areas, cows showed asynchronous space use and, while at pasture, contacted fewer other cows and showed substantially greater between-individual variation in contacts than other groups. We highlight the degree to which variations in management (e.g. grazing access, milking routine) substantially alter cattle contact patterns, with potentially major implications for infection transmission and social interactions. In particular, where individual cows have free choice of their environment, the resulting contact networks may have a less-risky structure that could reduce the likelihood of direct transmission of infections.

Infectious diseases such as avian influenza pose a global threat to human health. Mathematical and statistical models can provide key insights into the mechanisms that underlie the spread and persistence of infectious diseases, though their utility is linked to the ability to adequately calibrate these models to observed data. Performing robust inference for these systems is challenging. The fact that the underlying models exhibit complex non-linear dynamics, coupled with practical constraints to observing key epidemiological events such as transmission, requires the use of inference techniques that are able to numerically integrate over multiple hidden states and/or infer missing information. Simulation-based inference techniques such as Approximate Bayesian Computation (ABC) have shown great promise in this area, since they rely on the development of suitable simulation models, which are often easier to code and generalise than routines that require evaluations of an intractable likelihood function. In this manuscript we make some contributions towards improving the efficiency of ABC-based particle Markov chain Monte Carlo methods, and show the utility of these approaches for performing both model inference and model comparison in a Bayesian framework. We illustrate these approaches on both simulated data, as well as real data from an experimental transmission study of highly pathogenic avian influenza in genetically modi fied chickens., This work was supported by the Biotechnology and Biological Sciences Research Council (grants BB/G00479X/1, BBS/B/00239, and BBS/B/00301) and by the Cambridge Infectious Diseases Consortium Department for Environment, Food and Rural Affairs-Higher Education Funding Council for England (grant VT0105). AJKC was supported by BBSRC grant BB/I024550/1.

Trading animals between farms and via markets can provide a conduit for spread of infections. By studying trading networks, we might better understand the dynamics of livestock diseases. We constructed ingoing contact chains of cattle farms in Great Britain that were linked by trading, to elucidate potential pathways for the transmission of infection and to evaluate their effect on the risk of a farm experiencing a bovine tuberculosis (bTB) incident. Our findings are consistent with variation in bTB risk associated with region, herd size, disease risk area and history of previous bTB incidents on the root farm and nearby farms. However, we also identified effects of both direct and indirect trading patterns, such that connections to more farms in the England High-Risk Area up to three movements away from the root farm increased the odds of a bTB incident, while connections with more farms in the England Low-Risk Area up to eight movements away decreased the odds. Relative to other risk factors for bTB, trading behaviours are arguably more amenable to change, and consideration of risks associated with indirect trading, as well direct trading, might therefore offer an additional approach to bTB control in Great Britain.

Structure-from-Motion Multi View Stereo (SfM-MVS) photogrammetry is a technique by which volumetric data can be derived from overlapping image sets, using changes of an objects position between images to determine its height and spatial structure. Whilst SfM-MVS has fast become a powerful tool for scientific research, its potential lies beyond the scientific setting, since it can aid in delivering information about habitat structure, biomass, landscape topography, spatial distribution of species in both two and three dimensions, and aid in mapping change over time – both actual and predicted. All of which are of strong relevance for the conservation community, whether from a practical management perspective or understanding and presenting data in new and novel ways from a policy perspective.For practitioners outside of academia wanting to use SfM-MVS there are technical barriers to its application. For example, there are many SfM-MVS software options, but knowing which to choose, or how to get the best results from the software can be difficult for the uninitiated. There are also free and open source software options (FOSS) for processing data through a SfM-MVS pipeline that could benefit those in conservation management and policy, especially in instances where there is limited funding (i.e. commonly within grassroots or community-based projects). This paper signposts the way for the conservation community to understand the choices and options for SfM-MVS implementation, its limitations, current best practice guidelines and introduces applicable FOSS options such as OpenDroneMap, MicMac, CloudCompare, QGIS and speciesgeocodeR. It will also highlight why and where this technology has the potential to become an asset for spatial, temporal and volumetric studies of landscape and conservation ecology.

Selecting between competing statistical models is a challenging problem especially when the competing models are non-nested. An effective algorithm is developed in a Bayesian framework for selecting between a parameter-driven autoregressive Poisson regression model and an observation-driven integer valued autoregressive model when modelling time series count data. In order to achieve this a particle MCMC algorithm for the autoregressive Poisson regression model is introduced. The particle filter underpinning the particle MCMC algorithm plays a key role in estimating the marginal likelihood of the autoregressive Poisson regression model via importance sampling and is also utilised to estimate the DIC. The performance of the model selection algorithms are assessed via a simulation study. Two real-life data sets, monthly US polio cases (1970–1983) and monthly benefit claims from the logging industry to the British Columbia Workers Compensation Board (1985–1994) are successfully analysed.\ud \ud

The emergence and wide-spread circulation of mosquito-transmitted viral diseases, such as dengue, Zika and Chikungunya, is a global public health concern. In the absence of effective vaccines, current control measures are mostly targeted against the mosquito vector and have so far only shown limited success. The reliance on mosquitoes for transmission also imposes strong ecological constraints that can introduce significant spatial and temporal variations in disease incidence. However, the way that epidemiological and ecological factors interact and determine population-level disease dynamics is only partially understood. Here we fit a spatially-explicit individual based model defined within a Bayesian framework to Zika incidence data from Feira de Santana, allowing us to more precisely quantify the relationships between socio-ecological factors and arboviral outbreaks. Our results further demonstrated that the virus was likely introduced into multiple spatially segregated locations at the start of the outbreak, highlighting the benefits that spatio-temporal incidence data would bring in making modelling approaches more realistic for public health planning.

Selecting between competing statistical models is a challenging problem especially when the competing models are non-nested. In this paper we offer a simple solution by devising an algorithm which combines MCMC and importance sampling to obtain computationally efficient estimates of the marginal likelihood which can then be used to compare the models. The algorithm is successfully applied to a longitudinal epidemic data set, where calculating the marginal likelihood is made more challenging by the presence of large amounts of missing data. In this context, our importance sampling approach is shown to outperform existing methods for computing the marginal likelihood.

Background: Uganda changed its antiretroviral therapy guidelines in 2014, increasing\ud the CD4 threshold for antiretroviral therapy initiation from 350 cells/µl to 500 cells/µl.\ud We investigate what effect this change in policy is likely to have on HIV incidence,\ud morbidity, and programme costs, and estimate the cost-effectiveness of the change\ud over different time horizons.\ud Methods: We used a complex individual-based model of HIV transmission and\ud antiretroviral therapy scale-up in Uganda. 100 model fits were generated by fitting the\ud model to 51 demographic, sexual behaviour, and epidemiological calibration targets,\ud varying 96 input parameters, using history matching with model emulation. An\ud additional 19 cost and disability weight parameters were varied during the analysis of\ud the model results. For each model fit, the model was run to 2030, with and without the\ud change in threshold to 500 cells/µl.\ud Results: The change in threshold led to a 9.7% (90% plausible range: 4.3%-15.0%)\ud reduction in incidence in 2030, and averted 278,944 (118,452-502,790) DALYs, at a\ud total cost of $28M (-$142M to +$195M). The cost per disability adjusted life year\ud (DALY) averted fell over time, from $3238 (-$125 to +$29,969) in 2014 to $100 (-$499\ud to +$785) in 2030. The change in threshold was cost-effective (cost

Approximate Bayesian Computation (ABC) and other simulation-based\ud inference methods are becoming increasingly used for inference in\ud complex systems, due to their relative ease-of-implementation. We briefly\ud review some of the more popular variants of ABC and their application\ud in epidemiology, before using a real-world model of HIV transmission to\ud illustrate some of challenges when applying ABC methods to high dimensional,\ud computationally intensive models. We then discuss an alternative\ud approach—history matching—that aims to address some of these issues,\ud and conclude with a comparison between these different methodologies.

Slaughterhouse surveillance through post-mortem meat inspection provides an important mechanism for detecting bovine tuberculosis (bTB) infections in cattle herds in Great Britain (GB), complementary to the live animal skin test based programme. We explore patterns in the numbers of herd breakdowns detected through slaughterhouse surveillance and develop a Bayesian hierarchical regression model to assess the associations of animal-level factors with the odds of an infected animal being detected in the slaughterhouse, allowing us to highlight slaughterhouses that show atypical patterns of detection. The analyses demonstrate that the numbers and proportions of breakdowns detected in slaughterhouses increased in GB over the period of study (1998–2013). The odds of an animal being a slaughterhouse case was strongly associated with the region of the country that the animal spent most of its life, with animals living in high-frequency testing areas of England having on average 21 times the odds of detection compared to animals living in Scotland. There was also a strong effect of age, with animals slaughtered at > 60 months of age having 5.3 times the odds of detection compared to animals slaughtered between 0–18 months of age. Smaller effects were observed for cattle having spent time on farms with a history of bTB, quarter of the year that the animal was slaughtered, movement and test history. Over-and-above these risks, the odds of detection increased by a factor of 1.1 for each year of the study. After adjustment for these variables, there were additional variations in risk between slaughterhouses and breed. Our framework has been adopted into the routine annual surveillance reporting carried out by the Animal Plant Health Agency and may be used to target more detailed investigation of meat inspection practices.

Introduction: UNAIDS calls for fewer than 500,000 new HIV infections/year by 2020, with treatment-as-prevention being a key part of their strategy for achieving the target. A better understanding of the contribution to transmission of people at different stages of the care pathway can help focus intervention services at populations where they may have the greatest effect. We investigate this using Uganda as a case study.\ud Methods: An individual-based HIV/ART model was fitted using history matching. 100 model fits were generated to account for uncertainties in sexual behaviour, HIV epidemiology, and ART coverage up to 2015 in Uganda. A number of different ART scale-up intervention scenarios were simulated between 2016-2030. The incidence and proportion of transmission over time from people with primary infection, post-primary ART-naïve infection, and people currently or previously on ART was calculated.\ud Results: In all scenarios, the proportion of transmission by ART-naïve people decreases, from 70% (61%-79%) in 2015 to between 23% (15%-40%) and 47% (35%-61%) in 2030. The proportion of transmission by people on ART increases from 7.8% (3.5%-13%) to between 14% (7.0%-24%) and 38% (21%-55%). The proportion of transmission by ART dropouts increases from 22% (15%-33%) to between 31% (23%-43%) and 56% (43%-70%).\ud Conclusions: People who are currently or previously on ART are likely to play an increasingly large role in transmission as ART coverage increases in Uganda. Improving retention on ART, and ensuring that people on ART remain virally suppressed, will be key in reducing HIV incidence in Uganda.\ud

Summary Complex stochastic models are commonplace in epidemiology, but their utility depends on their calibration to empirical data. History matching is a (pre)calibration method that has been applied successfully to complex deterministic models. In this work, we adapt history matching to stochastic models, by emulating the variance in the model outputs, and therefore accounting for its dependence on the model's input values. The method proposed is applied to a real complex epidemiological model of human immunodeficiency virus in Uganda with 22 inputs and 18 outputs, and is found to increase the efficiency of history matching, requiring 70% of the time and 43% fewer simulator evaluations compared with a previous variant of the method. The insight gained into the structure of the human immunodeficiency virus model, and the constraints placed on it, are then discussed.

History matching is a model (pre-)calibration method that has been applied to computer models from a wide range of scientific disciplines. In this work we apply history matching to an individual-based epidemiological model of HIV that has 96 input and 50 output parameters, a model of much larger scale than others that have been calibrated before using this or similar methods. Apart from demonstrating that history matching can analyze models of this complexity, a central contribution of this work is that the history match is carried out using linear regression, a statistical tool that is elementary and easier to implement than the Gaussian process--based emulators that have previously been used. Furthermore, we address a practical difficulty with history matching, namely, the sampling of tiny, nonimplausible spaces, by introducing a sampling algorithm adjusted to the specific needs of this method. The effectiveness and simplicity of the history matching method presented here shows that it is a useful tool for the calibration of computationally expensive, high dimensional, individual-based models.

Background With ambitious new UNAIDS targets to end AIDS by 2030, and new WHO treatment guidelines, there is increased interest in the best way to scale-up ART coverage. We investigate the cost-effectiveness of various ART scale-up options in Uganda. Methods Individual-based HIV/ART model of Uganda, calibrated using history matching. 22 ART scale-up strategies were simulated from 2016 to 2030, comprising different combinations of six single interventions (1. increased HIV testing rates, 2. no CD4 threshold for ART initiation, 3. improved ART retention, 4. increased ART restart rates, 5. improved linkage to care, 6. improved pre-ART care). The incremental net monetary benefit (NMB) of each intervention was calculated, for a wide range of different willingness/ability to pay (WTP) per DALY averted (health-service perspective, 3% discount rate). Results For all WTP thresholds above $210, interventions including removing the CD4 threshold were likely to be most cost-effective. At a WTP of $715 (1 × per-capita-GDP) interventions to improve linkage to and retention/re-enrolment in HIV care were highly likely to be more cost-effective than interventions to increase rates of HIV testing. At higher WTP (> ~ $1690), the most cost-effective option was ‘Universal Test, Treat, and Keep’ (UTTK), which combines interventions 1–5 detailed above. Conclusions Our results support new WHO guidelines to remove the CD4 threshold for ART initiation in Uganda. With additional resources, this could be supplemented with interventions aimed at improving linkage to and/or retention in HIV care. To achieve the greatest reductions in HIV incidence, a UTTK policy should be implemented. Electronic supplementary material The online version of this article (doi:10.1186/s12879-017-2420-y) contains supplementary material, which is available to authorized users.

UNLABELLED: A case-control investigation was undertaken to determine management and health related factors associated with pleurisy in slaughter pigs in England and Wales. METHODS: The British Pig Executive Pig Health Scheme database of abattoir pathology was used to identify 121 case (>10% prevalence of pleurisy on 3 or more assessment dates in the preceding 24 months) and 121 control units (≤5% prevalence of pleurisy on 3 or more assessment dates in the preceding 24 months). Farm data were collected by postal questionnaire. Data from respondents (70 cases and 51 controls) were analysed using simple logistic regression models with Bonferroni corrections. Limited multivariate analyses were also performed to check the robustness of the overall conclusions. RESULTS AND CONCLUSIONS: Management factors associated with increased odds of pleurisy included no all-in all-out pig flow (OR 9.3, 95% confidence interval [CI]: 3.3-29), rearing of pigs with an age difference of >1 month in the same airspace (OR 6.5 [2.8-17]) and repeated mixing (OR 2.2 [1.4-3.8]) or moving (OR 2.2 [1.5-3.4]) of pigs during the rearing phase. Those associated with decreased odds of pleurisy included filling wean-to-finish or grower-to-finish systems with piglets from ≤3 sources (OR 0.18 [0.07-0.41]) compared to farrow-to-finish systems, cleaning and disinfecting of grower (ORs 0.28 [0.13-0.61] and 0.29 [0.13-0.61]) and finisher (ORs 0.24 [0.11-0.51] and 0.2 [0.09-0.44]) accommodation between groups, and extended down time of grower and finisher accommodation (OR 0.84 [0.75-0.93] and 0.86 [0.77-0.94] respectively for each additional day of downtime). This study demonstrated the value of national-level abattoir pathology data collection systems for case control analyses and generated guidance for on-farm interventions to help reduce the prevalence of pleurisy in slaughter pigs.

1. Understanding the demography of domestic dog populations is essential for effective disease control, particularly of canine-mediated rabies. Demographic data are also needed to plan effective population management. However, no study has comprehensively evaluated the contribution of demographic processes (i.e. births, deaths and movement) to variations in dog population size or density, or determined the factors that regulate these processes, including human factors. 2. We report the results of a 3-year cohort study of domestic dogs, which is the first to generate detailed data on the temporal variation of these demographic characteristics. The study was undertaken in two communities in each of Bali, Indonesia and Johannesburg, South Africa, in rabies-endemic areas and where the majority of dogs were free-roaming. None of the four communities had been engaged in any dog population management interventions by local authorities or animal welfare organizations. All identified dogs in the four communities were monitored individually throughout the study. 3. We observed either no population growth or a progressive decline in population size during the study period. There was no clear evidence that population size was regulated through environmental resource constraints. Rather, almost all of the identified dogs were owned and fed regularly by their owners, consistent with population size regulated by human demand. Finally, a substantial fraction of the dogs originated from outside the population, entirely through the translocation of dogs by people, rather than from local births. These findings demonstrate that previously reported growth of dog populations is not a general phenomenon and challenge the widely held view that free-roaming dogs are unowned and form closed populations. 4. Synthesis and applications. These observations have broad implications for disease and population control. The accessibility of dogs for vaccination and evaluation through owners and the movement of dogs (some of them infected) by people will determine the viable options for disease control strategies. The impact of human factors on population dynamics will also influence the feasibility of annual vaccination campaigns to control rabies and population control through culling or sterilization. The complex relationship between dogs and people is critically important in the transmission and control of canine-mediated rabies. For effective management, human factors must be considered in the development of disease and population control programmes.

A powerful and flexible method for fitting dynamic models to missing and censored data is to use the Bayesian paradigm via data-augmented Markov chain Monte Carlo (DA-MCMC). This samples from the joint posterior for the parameters and missing data, but requires high memory overheads for large-scale systems. In addition, designing efficient proposal distributions for the missing data is typically challenging. Pseudo-marginal methods instead integrate across the missing data using a Monte Carlo estimate for the likelihood, generated from multiple independent simulations from the model. These techniques can avoid the high memory requirements of DA-MCMC, and under certain conditions produce the exact marginal posterior distribution for parameters. A novel method is presented for implementing importance sampling for dynamic epidemic models, by conditioning the simulations on sets of validity criteria (based on the model structure) as well as the observed data. The flexibility of these techniques is illustrated using both removal time and final size data from an outbreak of smallpox. It is shown that these approaches can circumvent the need for reversible-jump MCMC, and can allow inference in situations where DA-MCMC is impossible due to computationally infeasible likelihoods.

Wildlife hosts are important reservoirs of a wide range of human and livestock infections worldwide, and in some instances, wildlife populations are threatened by disease. Yet wildlife diseases are difficult to monitor, and we often lack an understanding of basic epidemiological parameters that might inform disease management and the design of targeted interventions. The impacts of disease on host survival are generally associated with age, yet traditional epidemiological models tend to use simplistic categories of host age. Mortality trajectory analysis provides the opportunity to understand age-specific impacts of disease and uncover epidemiological patterns across complete life histories. Here, we use Bayesian survival trajectory analysis (BaSTA) software to analyse capture-mark-recapture data from a population of wild badgers Meles meles naturally infected with Mycobacterium bovis, the causative agent of tuberculosis in badgers and cattle. We reveal non-constant mortality trajectories, and show that infection exaggerates an age-dependent increase in late-life mortality. This study provides evidence for actuarial senescence in badgers, a species previously believed to display constant mortality throughout life. Our case study demonstrates the application of mortality trajectory analysis in wildlife disease research, but also highlights important limitations. We recommend BaSTA for mortality trajectory analysis in epidemiological research, but also suggest combining approaches that can include diagnostic uncertainty and the movement of hosts between disease states as they age. We recommend future combinations of multi-state and multi-event modelling frameworks for complex systems incorporating age-varying disease states.

Campylobacter jejuni, the most common cause of bacterial diarrhoeal disease, is normally helical. However, it can also adopt straight rod, elongated helical and coccoid forms. Studying how helical morphology is generated, and how it switches between its different forms, is an important objective for understanding this pathogen. Here, we aimed to determine the genetic factors involved in generating the helical shape of Campylobacter. A C. jejuni transposon (Tn) mutant library was screened for non-helical mutants with inconsistent results. Whole genome sequence variation and morphological trends within this Tn library, and in various C. jejuni wild type strains, were compared and correlated to detect genomic elements associated with helical and rod morphologies. All rod-shaped C. jejuni Tn mutants and all rod-shaped laboratory, clinical and environmental C. jejuni and Campylobacter coli contained genetic changes within the pgp1 or pgp2 genes, which encode peptidoglycan modifying enzymes. We therefore confirm the importance of Pgp1 and Pgp2 in the maintenance of helical shape and extended this to a wide range of C. jejuni and C. coli isolates. Genome sequence analysis revealed variation in the sequence and length of homopolymeric tracts found within these genes, providing a potential mechanism of phase variation of cell shape.

Salmonella enterica causes a range of diseases. Salmonellae are intracellular parasites of macrophages, and the control of bacteria within these cells is critical to surviving an infection. The dynamics of the bacteria invading, surviving, proliferating in and killing macrophages are central to disease pathogenesis. Fundamentally important parameters, however, such as the cellular infection rate, have not previously been calculated. We used two independent approaches to calculate the macrophage infection rate: mathematical modelling of Salmonella infection experiments, and analysis of real-time video microscopy of infection events. Cells repeatedly encounter salmonellae, with the bacteria often remain associated with the macrophage for more than ten seconds. Once Salmonella encounters a macrophage, the probability of that bacterium infecting the cell is remarkably low: less than 5%. The macrophage population is heterogeneous in terms of its susceptibility to the first infection event. Once infected, a macrophage can undergo further infection events, but these reinfection events occur at a lower rate than that of the primary infection.

Since the demonstration that almost 80% of human immunodeficiency virus type 1 (HIV-1) infections result from the transmission of a single variant from the donor, biological features similar to those of HIV mucosal transmission have been reported for macaques inoculated with simian immunodeficiency virus (SIV). Here we describe the early diversification events and the impact of challenge doses on viral kinetics and on the number of variants transmitted in macaques infected with the chimeric simian/human immunodeficiency virus SHIV sf162p4 . We show that there is a correlation between the dose administered and the number of variants transmitted and that certain inoculum variants are preferentially transmitted. This could provide insight into the viral determinants of transmission and could aid in vaccine development. Challenge through the mucosal route with high doses results in the transmission of multiple variants in all the animals. Such an unrealistic scenario could underestimate potential intervention measures. We thus propose the use of molecular evolution analysis to aid in the determination of challenge doses that better mimic the transmission dynamics seen in natural HIV-1 infection.

Salmonella enterica serovar Typhi, the agent of typhoid fever in humans, expresses the surface Vi polysaccharide antigen that contributes to virulence. However, Vi expression can also be detrimental to some key steps of S. Typhi infectivity, for example, invasion, and Vi is the target of protective immune responses. We used a strain of S. Typhimurium carrying the whole Salmonella pathogenicity island 7 (SPI-7) to monitor in vivo Vi expression within phagocytic cells of mice at different times after systemic infection. We also tested whether it is possible to modulate Vi expression via the use of in vivo -inducible promoters and whether this would trigger anti-Vi antibodies through the use of Vi-expressing live bacteria. Our results show that Vi expression in the liver and spleen is downregulated with the progression of infection and that the Vi-negative population of bacteria becomes prevalent by day 4 postinfection. Furthermore, we showed that replacing the natural tviA promoter with the promoter of the SPI-2 gene ssaG resulted in sustained Vi expression in the tissues. Intravenous or oral infection of mice with a strain of S. Typhimurium expressing Vi under the control of the ssaG promoter triggered detectable levels of all IgG subclasses specific for Vi. Our work highlights that Vi is downregulated in vivo and provides proof of principle that it is possible to generate a live attenuated vaccine that induces Vi-specific antibodies after single oral administration.

Several classes and multiple subclasses of immunoglobulins are produced towards protein and polysaccharide antigens in response to Salmonella infection and play a key role in protection against systemic disease. The targeting of Salmonella to Fc receptors (FcR) on phagocytes is a key step in the antibody-mediated antibacterial functions of host cells. We wished to compare the relative efficiency of different human IgG subclasses, which targeted the Salmonella enterica OmpA surface protein in modulating the interaction of bacteria with human phagocytes. To this end, we developed a novel system by tagging OmpA with a foreign CD52 mimotope (TSSPSAD) and opsonizing the bacteria with a panel of humanized CD52 antibodies that share the same antigen-binding V-region, but have constant regions of different subclasses. Our data revealed that opsonization with all the IgG subclasses increases Salmonella uptake by human phagocytes. IgG3 resulted in the highest level of bacterial uptake and the highest average bacterial load per infected cell, which was closely followed by IgG1, then IgG4 and lastly IgG2. Phagocytosis mediated by IgG1, IgG3 and IgG4 had a higher dependency on FcγRI than FcγRIIA, whereas IgG2-mediated phagocytosis required FcγRIIA more than FcγRI. The results show that IgG binding to OmpA increases the uptake of Salmonella by human phagocytic cells and that the efficiency of this process depends both on the subclass of the IgG and the type of FcR that is available for antibody binding.

The interaction between Salmonella enterica and the host immune system is complex. The outcome of an infection is the result of a balance between the in vivo environment where the bacteria survive and grow and the regulation of fitness genes at a level sufficient for the bacteria to retain their characteristic rate of growth in a given host. Using bacteriological counts from tissue homogenates and fluorescence microscopy to determine the spread, localization, and distribution of S. enterica in the tissues, we show that, during a systemic infection, S. enterica adapts to the in vivo environment. The adaptation becomes a measurable phenotype when bacteria that have resided in a donor animal are introduced into a recipient naïve animal. This adaptation does not confer increased resistance to early host killing mechanisms but can be detected as an enhancement in the bacterial net growth rate later in the infection. The enhanced growth rate is lost upon a single passage in vitro , and it is therefore transient and not due to selection of mutants. The adapted bacteria on average reach higher intracellular numbers in individual infected cells and therefore have patterns of organ spread different from those of nonadapted bacteria. These experiments help in developing an understanding of the influence of passage in a host on the fitness and virulence of S. enterica .

Determining the evolutionary basis of cross-species transmission and immune evasion is key to understanding the mechanisms that control the emergence of either new viruses or novel antigenic variants with pandemic potential. The hemagglutinin glycoprotein of influenza A viruses is a critical host range determinant and a major target of neutralizing antibodies. Equine influenza virus (EIV) is a significant pathogen of the horse that causes periodical outbreaks of disease even in populations with high vaccination coverage. EIV has also jumped the species barrier and emerged as a novel respiratory pathogen in dogs, canine influenza virus. We studied the dynamics of equine influenza virus evolution in horses at the intrahost level and how this evolutionary process is affected by interhost transmission in a natural setting. To this end, we performed clonal sequencing of the hemagglutinin 1 gene derived from individual animals at different times postinfection. Our results show that despite the population consensus sequence remaining invariant, genetically distinct subpopulations persist during the course of infection and are also transmitted, with some variants likely to change antigenicity. We also detected a natural case of mixed infection in an animal infected during an outbreak of equine influenza, raising the possibility of reassortment between different strains of virus. In sum, our data suggest that transmission bottlenecks may not be as narrow as originally perceived and that the genetic diversity required to adapt to new host species may be partially present in the donor host and potentially transmitted to the recipient host.

Objective: To (1) discover kinematic variables that differ between dogs with cranial cruciate ligament rupture (CCLR) and control dogs and (2) to identify suitable outcome variables for longitudinal studies on the effects of surgical interventions for CCLR in dogs. Study Design: In vivo case–control comparison. Animals: Dogs with unilateral CCLR (n=16), and no other detectable neurologic or orthopedic disease and 10 dogs without CCLR, neurologic, or orthopedic disease. Methods: Kinematic data was collected from dogs as they walked at constant speed on a treadmill, using an infrared camera system and adhesive reflective markers applied to the pelvic limbs. Data on 5 selected variables was extracted and comparisons made between paired pelvic limbs in both CCLR and control dogs and between control and CCLR dogs. Results: Comparisons in CCLR dogs revealed significant differences between affected and unaffected limbs for many examined variables; the largest differences were in paw velocity and stifle angular velocity. There were highly significant differences between CCLR and control dogs when examining paired limb ratios for 4 of the variables, the largest differences were in stride length and paw velocity. Conclusions: Kinematic analysis provides straightforward and objective methods for defining the lameness associated with CCLR in dogs. Paw velocity and stride length were most notably reliable and not susceptible to systematic alterations in stifle joint dynamics that might be associated with specific surgical procedures. Clinical Relevance: This study identifies several variables that can provide the objective measurements essential to evaluate the efficacy of surgical interventions for CCLR in dogs.

During systemic disease in mice, Salmonella enterica grows intracellularly within discrete foci of infection in the spleen and liver. In concomitant infections, foci containing different S. enterica strains are spatially separated. We have investigated whether functional interactions between bacterial populations within the same host can occur despite the known spatial separation of the foci and independence of growth of salmonellae residing in different foci. In this study we have demonstrated that bacterial numbers of virulent S. enterica serovar Typhimurium C5 strain in mouse tissues can be increased by the presence of the attenuated aroA S. Typhimurium SL3261 vaccine strain in the same tissue. Disease exacerbation does not require simultaneous coinjection of the attenuated bacteria. SL3261 can be administered up to 48 hr after or 24 hr before the administration of C5 and still determine higher tissue numbers of the virulent bacteria. This indicates that intravenous administration of a S. enterica vaccine strain could potentially exacerbate an established infection with wild-type bacteria. These data also suggest that the severity of an infection with a virulent S. enterica strain can be increased by the prior administration of a live attenuated vaccine strain if infection occurs within 48 hr of vaccination. Exacerbation of the growth of C5 requires Toll-like receptor 4-dependent interleukin-10 production with the involvement of both Toll/interleukin-1 receptor-domain-containing adaptor inducing interferon-beta and myeloid differentiation factor 88.