Fig. 3

Model outputs showing potential impact of having changed the IRS campaign start date. The model estimates what the average impact of the IRS campaigns from 2012 to 2014 might have been if the campaign start dates had shifted a 2 months earlier, to May; b 1 month earlier, to June; c no shift, illustrating the actual impact of the campaigns that started in July; d 1 month later, to August; and e 2 months later, to September. The table shown in f shows the estimated number of cases per 10,000 person-months that were averted in each scenario, *based on the cumulative incidence from the 6 months after each modelled campaign