Pandemic Influenza, Reopening Schools, and Returning to Work

Davey and Robert Glass present a paper (1) in which they consider the question of when to " switch off " community based interventions designed to reduce the spread of pandemic infl uenza. These authors attempt to answers questions such as when it would be optimal to reopen schools that have been closed as part of a nonpharmaceuti-cal, communitywide infl uenza mitigation strategy. The authors use a mathematical model, previously described in this journal (2), to simulate the spread of pandemic infl uenza throughout a community that represents the US population. This model is similar to another model that was used to examine the effectiveness of closing schools to slow the spread of infl uenza pandemic (3). Both models simulate the spread of infl uenza by dividing a representative population into households. The models then track each household member with each member having a defi ned number of random contacts (per day) that are allocated within a network of possible contacts. Once a contact is calculated to have occurred, the probability of infl uenza transmission is calculated. Also included in the calculations are variables such as infl uenza incubation and infectiousness periods. What does the model " say " ? Davey and Glass considered what would happen if schools were reopened and community-wide sequestering were halted when infl uenza cases in a community fall below preset thresholds (e.g., 1, 2, or 3 cases in 7 days). Sequestering strategies would be restarted if the epidemic pandemic resurged and >10 cases occurred in a 7-day period. This " pulsing technique " would reduce the number of days needed to sequester schoolchil-dren and the community by 6% to 32%. The authors maintain that for a given pandemic scenario, the reduction in days sequestered would not notably affect the number of persons infected. The implication is that reduction in days sequestered will reduce the economic impact and social disruption caused by community-wide, nonpharmaceutical interventions. Are the results " believable " ? As with all mathematical models, some potential technical problems exist. First, almost all models that simulate individual person-to-person infl uenza transmission use 1 or 2 databases that record the probability of infl uenza transmission. One database was recorded in the early 1970s in Tecumseh, Michigan (4), and the other among ≈400 households across France (5). Is it reasonable to use these estimates to simulate in-fl uenza transmission in every community, town, city, and …