Giuseppe
Emeritus
ITALY, PANDEMIC PREPAREDNESS: Diffusion and Control of an hypothetic pandemic [ECDC Health Content]
SCIENTIFIC ADVANCES ? PANDEMIC INFLUENZA
Hypothetic Influenza Pandemic in Italy: diffusion and control
Scenarios of diffusion and control of an influenza pandemic in Italy
C. Rizzo, A. Lunelli, A. Pugliese, A. Bella, P. Manfredi, G. Scalia-Tomba, M. Iannelli, M.L. Ciofi Degli Atti
Epidemiol. Infect. (2008), 136, 1650-1657
Description:
The study describes the predicted outcome of pandemic infections in Italy, in a baseline scenario with no interventions. It then considers scenarios with some interventions.
The model takes the population of Italy (n = 57 million) as its population.
People are assumed to be travelling according to data on air passengers domestically and the population is divided into region of residence in Italy, and within each region, there is a division into six age-groups: 0-2. 3-14. 15-18- 19-39. 40-64. >65 years.
These groups are assigned different values on how they interact with each other in terms of contacts with infectious people. Also contacts were also distributed to consider, households, school / workplaces and random contacts.
The model is initiated by assuming 5 infected persons arrive simultaneously at an international airport and no more imported cases follow. The model has both random and deterministic elements. At the beginning of the outbreak a random simulation is performed, leading sometimes to the result that no larger outbreak is initiated by the five index cases. However if the index cases create a larger outbreak, a deterministic model (a model with no random components) is applied to predict the impact on the society. Incubation period of one day and infectious period of 3.9 days were assumed.
Assumed interventions included antiviral prophylaxis (AVP), vaccination and social distancing. Vaccine effectiveness of 50% and 70% was assumed. The use of AVP was assumed to reduce susceptibility by 30% and infectiousness by 70%. To measure effects of social distancing, closing schools for three weeks, public offices for four weeks and public meeting places for eight weeks were considered. Schools closure was assumed to mean reduction of the contact rates among children by 75%. Workplace closure by 16% and closure of public places would reduce the transmission rate by 50% of random contacts on the society. The simulations were undertaken by assuming reasonable values for infectivity, the basic reproduction number (Ro) and then calculating the attack rates and the time to the peak of the epidemic. ?Low? value on Ro around 1.6, were associated with greater effect from the countermeasures than when there were high value around 2.0. For higher values of Ro it was early timing of the interventions that demonstrated an effect on delaying the peak and reducing the number of infected at the peak.
ECDC comment (03/12/2008):
Developing a model for predicting how a pandemic will happen and investigating how some assumptions for the national pandemic plan is a welcome idea. However the study confirms much of what has been known already suggesting predictable benefits of interventions, whose major aim (especially social distancing) is to reduce the burden on the society by spreading the peak of the epidemic over a longer time period. It would have been beneficial to also estimate the burden for the health care during these weeks, and possible effects of spatial distribution of antivirals.
The model relies on a number of assumptions; in the study sensitivity analysis is performed to estimate the effects on varying the basic reproductive number R0, at the values of 1.6, 1.8 and 2.0; the study adds some interventions whose effects are widely discussed. Like school closure, here only a three week period and a reduction of infectious contacts by 75% among the children are considered. This is optimistic and assumes that children out of school do not mix which does not always seem to be the case. One empirical investigation of school closures found lower effects of this social distancing than might have been expected from modelling alone[1] Added value would have been testing more the sensitivity of these assumptions, especially when some other studies assume that school closure reduces the infectious contacts among school children by only 10%. [2]. Using antivirals and the assumptions about their effect on susceptibility and transmission would also have benefited the study.
References.
1. Cauchemez S, Valleron A-J, Boelle P-Y, Flahault A and Ferguson N Estimating the impact of school closure on influenza transmission from Sentinel data Nature 452, 750-754 (10 April 2008)
2. Vynnycky, E. and W.J. Edmunds, Analyses of the 1957 (Asian) influenza pandemic in the United Kingdom and the impact of school closures. Epidemiol Infect, 2008. 136(2): p. 166-79.
Comment to: influenza@ecdc.europa.eu
-
<cite cite="http://ecdc.europa.eu/en/health_content/sciadv/081211_sciadv.aspx">ECDC Health Content</cite>Hypothetic Influenza Pandemic in Italy: diffusion and control
Scenarios of diffusion and control of an influenza pandemic in Italy
C. Rizzo, A. Lunelli, A. Pugliese, A. Bella, P. Manfredi, G. Scalia-Tomba, M. Iannelli, M.L. Ciofi Degli Atti
Epidemiol. Infect. (2008), 136, 1650-1657
Description:
The study describes the predicted outcome of pandemic infections in Italy, in a baseline scenario with no interventions. It then considers scenarios with some interventions.
The model takes the population of Italy (n = 57 million) as its population.
People are assumed to be travelling according to data on air passengers domestically and the population is divided into region of residence in Italy, and within each region, there is a division into six age-groups: 0-2. 3-14. 15-18- 19-39. 40-64. >65 years.
These groups are assigned different values on how they interact with each other in terms of contacts with infectious people. Also contacts were also distributed to consider, households, school / workplaces and random contacts.
The model is initiated by assuming 5 infected persons arrive simultaneously at an international airport and no more imported cases follow. The model has both random and deterministic elements. At the beginning of the outbreak a random simulation is performed, leading sometimes to the result that no larger outbreak is initiated by the five index cases. However if the index cases create a larger outbreak, a deterministic model (a model with no random components) is applied to predict the impact on the society. Incubation period of one day and infectious period of 3.9 days were assumed.
Assumed interventions included antiviral prophylaxis (AVP), vaccination and social distancing. Vaccine effectiveness of 50% and 70% was assumed. The use of AVP was assumed to reduce susceptibility by 30% and infectiousness by 70%. To measure effects of social distancing, closing schools for three weeks, public offices for four weeks and public meeting places for eight weeks were considered. Schools closure was assumed to mean reduction of the contact rates among children by 75%. Workplace closure by 16% and closure of public places would reduce the transmission rate by 50% of random contacts on the society. The simulations were undertaken by assuming reasonable values for infectivity, the basic reproduction number (Ro) and then calculating the attack rates and the time to the peak of the epidemic. ?Low? value on Ro around 1.6, were associated with greater effect from the countermeasures than when there were high value around 2.0. For higher values of Ro it was early timing of the interventions that demonstrated an effect on delaying the peak and reducing the number of infected at the peak.
ECDC comment (03/12/2008):
Developing a model for predicting how a pandemic will happen and investigating how some assumptions for the national pandemic plan is a welcome idea. However the study confirms much of what has been known already suggesting predictable benefits of interventions, whose major aim (especially social distancing) is to reduce the burden on the society by spreading the peak of the epidemic over a longer time period. It would have been beneficial to also estimate the burden for the health care during these weeks, and possible effects of spatial distribution of antivirals.
The model relies on a number of assumptions; in the study sensitivity analysis is performed to estimate the effects on varying the basic reproductive number R0, at the values of 1.6, 1.8 and 2.0; the study adds some interventions whose effects are widely discussed. Like school closure, here only a three week period and a reduction of infectious contacts by 75% among the children are considered. This is optimistic and assumes that children out of school do not mix which does not always seem to be the case. One empirical investigation of school closures found lower effects of this social distancing than might have been expected from modelling alone[1] Added value would have been testing more the sensitivity of these assumptions, especially when some other studies assume that school closure reduces the infectious contacts among school children by only 10%. [2]. Using antivirals and the assumptions about their effect on susceptibility and transmission would also have benefited the study.
References.
1. Cauchemez S, Valleron A-J, Boelle P-Y, Flahault A and Ferguson N Estimating the impact of school closure on influenza transmission from Sentinel data Nature 452, 750-754 (10 April 2008)
2. Vynnycky, E. and W.J. Edmunds, Analyses of the 1957 (Asian) influenza pandemic in the United Kingdom and the impact of school closures. Epidemiol Infect, 2008. 136(2): p. 166-79.
Comment to: influenza@ecdc.europa.eu
-