Figure 2

Cumulative number of weeks with alarms over year-long periods (week 14 in one year to week 13 in the next), for the periods (a) 2008/9, (b) 2009/10 – split between the first (dotted lines) and second (solid lines) parts of the year to show the pH1N1 introduction in the summer of 2009, (c) 2010/11, (d) 2011/12 and (e) 2012/13 for hospital staff-ILA (blue) and London-ILI (red). Alarms are generated using a Bayesian approach, taking the previous 6 weeks as reference. With the exception of pH1N1 introduction in the summer of 2009, hospital staff-ILA generates alarms at the same or earlier times than London-ILI; however, this is not generally associated with a higher number of alarms (i.e. false positive alarms) over the course of the year.