mv

The project has started in June 2008, so that it has been going on for a year and a half. The activities planned for the first part of the project hinged on the acquisition of relevant data, and on the development of mathematical and statistical techniques suitable for the available data.

A consistent body of work has been achieved towards the first five objectives .

Concerning objective 1 (To improve characterization of population contact and travel patterns in models), existing population data from previous projects (such as MODELREL and POLYMOD) and new information from EuroStat and other international sources have been collected and brought into a unified database and repository (warehouse) for the use of the consortium. New sources of information on population movement data, in the form of origin-destination flows, have been identified and, as long as licensing issues allowed, added to the database or anyway their use made possible to the consortium. These commuting travel data have been integrated into models of epidemic spread to test the feasibility of using simple ‘gravity’ models in epidemic models. Contact matrices arising from the POLYMOD studies have been extensively used to analyse data from the pandemic outbreak, but other approaches have also been explored to complement them. In particular, data on age-specific contact patterns have been collected in the Netherlands, and are now being analysed to provide independent estimates. In another approach, an agent-based model parameterised using demographic, socio-economic as well as Time Use data, has been used to provide estimates of contact matrices; this approach will allow to assess predictable changes in the mixing pattern.

As for objective 2 (To evaluate behavioural responces to epidemics and social acceptance of restriction measures) a survey has been permormed in the period June-August 2009 in four European states (Finland, Italy, Romania and United Kingdom) in order to assess behavioural responses to the emergence of the novel influenza strain and the declaration of a pandemic, and to evaluate the acceptance of possible restriction measures directed at mitigating the influenza pandemic.
Such a survey of the general public had been prepared with questions concerning seasonal influenza, and hypothetical scenarios of a pandemic. The current situation has motivated us to revise the questionnaire to be updated to the knowledge about the novel A/H1N1 influenza epidemic and gives the added opportunity to address the behaviour of general public in real time during a pandemic. National results have already been presented to the respective health authorities a survey on behavioural responses to a pandemic and acceptance of restriction measures . After the spread of A(H1N1) influenza it was decided to repeat the survey during spring 2010, to be able to assess changes in perception after the major pandemic wave, and to compare the results with available data on actual behaviour.  The two surveys will be used for building realistic models of behavioural changes in response to a pandemic, to be incorporated in epidemiological models; an epidemic model with a hypotetical mechanism for behavioural changes has already been proposed and analysed. A survey was also carried out, before the emergence of the A/H1N1 influenza virus, among advisors to policy makers in the different member states; they were asked about their national pandemic plan, and also about their own perception about the likelihood that specific public health measures would be applied in the event of a pandemic in their country. With emergence of the novel flu, the results of the survey provide important insights into the likely management of the pandemic in over a third of EU countries, and also into the differences between plans and actual policy.

Concerning objective 3 (To develop a suite of models for the spatio-temporal spread of a new influenza pandemic) several improvements have been performed during the first part of the project to the existing models for spatiotemporal epidemic spread: simple ‘gravity’ models of human movement have been integrated into a new patch model of pandemic spread, showing that gravity models provide an adequate description of disease spread, though imperfect at a fine spatial scale. Age-structured patch models have been further analysed, studying in particular the sensitivity of the results to the exact structure of the age-contact matrix. Developments have been made to EU-scale individual-based models, improving on the parameterisation from available data, on the modelling of travel behaviour, and on the efficiency of the implementation.
These more theoretical advances have been delayed by the outbreak of the pandemic, that caused the focus to shift towards simpler models based on available data, or towards micro-simulation focussing on a single policy issue (e.g., effect of school closure).

As for objective 4 (To estimate model parameters and test model adequacy using data on seasonal flu and endemic diseases) outbreak data, initially from Mexico, then from other areas with a sustained transmission (US, UK, Southern Hemisphere) and from several other European countries, have been analysed by partners of the project. This body of work  has led to the estimation of epidemiological parameters including reproduction number, household secondary attack rates, age dependence of susceptibility to infection, generation time, and severity. More work is ongoing, using techniques borrowed from the field of image reconstruction, to extract and disentangle the information present in FF100 datasets, thus obtaining more accurate estimates of the distributions for infection date and incubation period.
Data from seasonal influenza epidemic and from previous pandemic provide important information both for guiding the data analysis and forecast about the current pandemic, and for planning for future ones. Data on seasonal influenza from several countries have been analysed to estimate the transmissibility and magnitude of seasonal variation in transmission; data from the surveillance system and from serological studies in Italy have been used to estimate the under-notification of the surveillance system.
We collected the available datasets from past pandemics, including a review of their possible use for parameter estimation: 30 reports presenting original data have been found: 20 included data for the 1918 pandemic, 9 for  the 1957 pandemic and 7 for the 1968 pandemic. All data have been collected in a spreadsheet electronic document, and PDF sources of the corresponding articles are provided, completing a project deliverable.
In particular, we have found reports of serological survey conducted in the Netherlands before and after the influenza pandemic in 1957 and 1958. Using these reports, we could reconstruct the infection dynamics of that pandemic in the Netherlands.
In order to have a better understanding of influenza natural history, nonlinear models have been developed to describe influenza viral kinetics (VK) as a surrogate of infectiousness, and symptoms dynamics (SD) correlated with cytokines level. The model was fitted to individual data of VK and symptoms from 31 volunteers experimentally challenged with A(H1N1) influenza virus, yielding estimates of generation time, latent period (defined by the time between inoculation and a threshold level of VK) and incubation time (defined by the time between inoculation to a threshold symptom score). This work, presented at the Epidemics2 meeting in Athens, is a step tying epidemiological work to experimental work, and helps in the design of new operational protocols to quantify aspects of the natural history of influenza relevant to potential intervention measures.

Finally, concerning the evaluation of the impact of intervention options (objective 5), general work on pandemic mitigation control options in an EU context has been performed and presented at a number of meetings. Work has been devoted to specific policy options:
- the efficacy of interventions based on age-prioritized use of antivirals has been evaluated in terms of cumulative attack rate and excess mortality reduction under different scenarios in a stochastic, spatially structured individual-based model;
- a novel algorithm, relying on an approximation of the contact network structure, has been developed to approximate optimal allocation patterns for vaccines during an influenza pandemic, using only information that can be observed during the pandemic.
- reactive short-duration school closure has been extensively studied by partners of the project; one study concentrated on the dynamics resulting from different rules for school closure, and on how to achieve maximum effects with a limited school closure; the other study, focussed on the English setting but with possible wider international public-health implications, concentrated on whether school closure could reduce the actual burden on ICU when these are close to capacity, finding that considerable local coordination of school closures would be required for this to happen.
The results obtained so far in the project have been presented at several scientific conferences, at meetings more policy-oriented organized by ECDC or by the Health Security Committee, at the meetings of the WHO and ECDC informal networks on pandemic flu; some results have also been published in academic journals.

MAIN RESULTS

The survey on behavioural responses to a pandemic and acceptance of restriction measures has provided important information, that are being currently analysed, on the beliefs and attitudes present in the population at the start of the pandemic in 4 distant European countries. Repeating the survey next spring will allow a detailed assessment of the changes occurred in public perception, comparing them, for instance, to actual vaccination rates.
Both patch and micro-simulation models of pandemic spread have been developed and improved in the first part of the project. Several aims have been achieved: an analysis of the dependence of spatiotemporal disease spread on the details of modelling travel behaviour; the analysis, through the development of an individual based model at European scale running in a sensible time on a single computer, of how socio-demographic heterogeneities within Europe influence the pattern of spread of pandemic flu; the coupling, through detailed spatial data on the destinations of travellers, of a stochastic meta-population model of global transmission to a model of internal transmission to assess the ongoing impact of importation on the early stages of the epidemic. These technical advances (some of which published in academic papers, others still in draft form) have been partially used as a guidance in the present pandemic, and will contribute to the development of the modelling tools that will be one of the project outputs.
Several partners have collaborated with WHO and national health authorities in the analysis of early data on the outbreak of pandemic A(H1N1) providing estimates of epidemiological parameters including reproduction number, household secondary attack rates, generation time, and severity (estimates published in several academic papers). The resulting estimates have been used by health planners to obtain plausible ranges of intensity and timing of the pandemic wave, and to advise intervention strategies.
Progress has been accomplished also in the statistical analysis of previous flu epidemics. From the analysis of seasonal epidemic data, estimates have been made of temporal variability in virus transmissibility, and of the reporting rates of the sentinel surveillance system. Such estimates are important for assessing the attack ratio of the H1N1 pandemic wave, and the potential for a second wave.
An organized dataset on available data on previous pandemics has also been organized. In particular, different dataset on the Asian Influenza pandemic in 1957-58 in the Netherlands have been integrated to reconstruct the infection dynamics exhibiting a second smaller wave after the first one.
Much effort has been devoted to the potential effect of specific intervention measures. Reactive school closure has been modelled in detail, and its potential effects on the transmission of H1N1 analysed. The potential effects of different strategies of using antivirals and vaccine in the context of the current pandemic has been studied extensively; in particular, we analysed age-prioritized use of antivirals, and developed a novel algorithm that approximates optimal allocation patterns for vaccines during an influenza pandemic with limited information.