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

The interventions used in the present study are based on social distancing. Surveys have shown that a higher proportion of low-income countries have social distancing interventions planned in the case of a pandemic, compared to high income European countries due to their feasibility and cost-effectiveness. Pharmaceutical-based interventions were not considered in this study due to cost and delivery issues found with low-income countries. [4, 8] Consideration may need to be given to some form of antiviral agent use, possibly for case treatment, given the poor performance of purely social distancing interventions in developing countries as suggested by this study. The advantage of antiviral use is that it may be targeted at reducing within-household transmission, a location of transmission otherwise not impacted by social distancing interventions. The present study shows that larger numbers of transmissions occur within households due to the larger household sizes in countries such as PNG. Furthermore, low-income countries generally have population health profiles not seen in developed countries, including the prevalence of diseases such as malaria, tuberculosis and HIV. [44] Antiviral treatment of influenza cases co-infected with such diseases may be an appropriate intervention strategy given the poorer health outcomes which co-morbidity may impose. [12] The provision and use of antiviral agents will have challenges not faced by industrialised countries, due to the cost of providing an antiviral stockpile and subsequent distribution of antivirals in countries which have health systems already stretched to capacity. [15] The importance of suitable pandemic influenza preparedness plans for low-income countries is recognized by the WHO [45, 46] . Disease transmission models will play a role in guiding policy makers in determining the effectiveness of possible pandemic countermeasures. History tells us that low-income countries are likely to be more affected, yet almost all models deal with mitigation strategies for developed nations with good health infrastructure and low prevalence of endemic diseases. Daily case incidence for the no intervention, school closure and rigorous social distancing (school closure and workplace and community contact reductions) scenarios is shown. The blue, red and green curves represent the Albany, Madang and Madang-nnh models respectively. Compliance with home isolation of school children following school closure intervention has been examined and compliance reduction from 100% to 50% considered. Effects of reduced compliances on the illness attack rates and their corresponding 95% CI (in parentheses, shaded rows) are presented in Table A4 . For the purely school closure intervention, the illness attack rate increases by ~3% in Albany as reduced isolation permits additional child-to-child contact. In the Madang models, reduced compliance with home isolation during the period of school closure has minimal effect on the attack rate. This is due to ~50% school age children not attending school. In all three models, reduced compliance has minimal effect if additional rigorous social distancing interventions are also applied. Alternative intervention activation timing assumptions for school closure, workforce reduction and community contact reduction are examined and two different settings are considered. Effects of these alternative assumptions on the illness attack rates and their corresponding 95% CI (in parenth