Pandemic influenza in Papua New Guinea: a modelling study comparison with pandemic spread in a developed country

Milne, George J; Baskaran, Pravin; Halder, Nilimesh; Karl, Stephan; Kelso, Joel

General comparative trends are presented in Table 1 , were the Albany and Madang models are compared. For a pandemic in a developed country setting having a basic reproduction number R0 of 1.5 and an illness attack rate of 31.7%, the same pandemic characteristics applied in a PNG setting with no interventions activated resulted in higher rates of influenza transmission, with an attack rate of 46.6% and reproduction number R0 = 1.88 (Table 1) . Comparing the attack rates when interventions were activated, both intervention strategies (solely school closure and rigorous social distancing) were much less effective in the PNG setting compared to the Australian one. The mitigated illness attack rates in the PNG model were approximately 3 times higher than those in the Australian model with the interventions being 60% -70% less effective. We also conducted experiments with the Madang-nnh model, where neighbourhood contact hubs were omitted, to contrast the transmission characteristics with and without these contact hubs. The results of these simulation experiments are included in Tables 1 and 2 . It should be noted that the Madang model with neighbourhood hubs included is believed to be more representative of population mixing characteristics in PNG, following local knowledge. Neighbourhood hubs were introduced to reflect the known mixing patterns occurring with adults not working in designated workplaces, such as found in the regular contact which may occur among individuals in marketplaces. The addition of such contact hubs increased individual-to-individual contacts and hence infection transmission opportunities (see Table 2 ) with approximately 2,000 additional symptomatic infections resulting, for all three (mitigated and non-mitigated) scenarios. In terms of cumulative illness attack rate, the Madang model gave a basic reproduction number R0 of 1.88 and an illness attack rate of 46.6%, compared to that of the Madang-nnh (no neighbourhood hub) model with R0 = 1.74 and an illness attack rate of 40.8%. The daily case incidence resulting from simulating the Albany, Madang and Madang-nnh models is presented in Figure 2 for the three scenarios. It is apparent from the results presented in Table 1 that there is a greater proportion of influenza cases occurring within households in both Madang models note that Madang has a total population of ~35,000 compared to ~30,000 in Albany.