The first year of the project has been dedicated to the development of the Population Based Approach. Characteristics of population-based approach surveillance include:

• Random (and thus representative) sample of the population from complete and continuously
updated population register
• Selected individuals approached via regular mail and asked to commit to self-reporting
infectious disease symptoms during 6-12 months
• Initial questionnaire about background data
• Self-reporting of infectious disease symptoms on an event-driven basis through automated
technologies (secure website or Interactive Voice Response telephone service – IVRS – with a
toll-free phone number)
• Self-selected communication technology
• Symptoms specified through a tree-structured set of questions with initial gate questions
• Algorithms for clinical diagnosis
• Repeated reminders

Due to the intervening new influenza pandemic that emerged during the spring of 2009, the set-up and implementation of the PBA was expedited.

The secure website for participants (www.sjukrapport.se) is in Swedish and contains a home page, a registration module, a reporting module, and a quitting module. In this document , an English user guide to the web based reporting system can be downloaded. There are links to Sjukrapport’s weekly newsletter hosted by the website for the Swedish Institute for Infectious Disease Control (SMI) (http://www.smittskyddsinstitutet.se/publikationer/smis-nyhetsbrev/sjukrapport/) and the embedded link to the results of the reporting (http://www.smittskyddsinstitutet.se/publikationer/smis-nyhetsbrev/sjukrapport/sasongen-
20092010/). Both of the latter pages are open to the public. To get in to the registration and reporting modules of the participant website, participants have to enter their individually unique National Registration Number (NRN) as authentication. This number is compared to the NRNs of all invited persons in the study database, and if it doesn’t match, further access is denied. Before disease reporting is allowed, participants first have to register by entering their NRN in the registration module.

Upon reporting of a disease episode, regardless of communication technology, the participants are required to answer a brief tree-structured symptom questionnaire. The questionnaire probes into symptoms that are part of the case definition for Influenza-like Illness (ILI), as proposed by the European Centre for Disease Prevention and Control (http://ecdc.europa.eu/en/activities/surveillance/EISN/Pages/AbouttheNetwork_Influienzacasedefin
itions.aspx). The questions are shown in translated form below. In addition there is a question about the number of days since onset of the symptoms.

Important components of the scheme are the monthly newsletters that keep the participants alerted to their task. In addition to serving as reminders, they provide the participants with feedback on their reporting, along with various kinds of “infotainment” related to infectious diseases. For participants who have given us their e-mail address, the newsletters are distributed electronically, but for the remainder, they are sent out as printed pamphlets via regular mail.

Thank to the participation of the population, the study has been able to compare the epidemic curves for ILI generated by the population-based approach surveillance (Sjukrapport) with those obtained through the regular sentinel physician
reporting system. It should be emphasized that the raw data represent two totally different things – the sentinel figures denote the percentage among all patients at sentinel units who have symptoms indicative of influenza, while the Sjukrapport data represent the incidence per 100 person-weeks. This notwithstanding, the pattern and timing of peaks, and the relative difference between 2008 and 2009 are strikingly similar between sentinel reports and population-based ILI reporting (see graphs below).

Interestingly, the incidence recorded in Sjukrapport is approximately 100 times higher than the estimated incidence that is reported to the European Influenza Surveillance Scheme (EISS) based on the Swedish sentinel data (the latter incidence is merely an educated guess). It is quite obvious that the Swedish sentinel-based incidence figures are gross underestimations, while the data generated by Sjukrapport appear to be much more reasonable and in line with international data. The tendency for sentinel data to be underestimations has been noted in other countries (e.g. the Netherlands, UK), where alternative and more population-centered methods for influenza surveillance have been
tested.

Due to the emerging A(H1N1) influenza pandemic we have, over and above the prescribed tasks for this first year, drawn a random population sample and included approximately 5500 active participants (participation rate 46%) who started real live reporting in early September 2009. The PBA surveillance system, thus, has been in full operation for more than 6 months. In the graphs below (taken from the most recent account of the Sjukrapport data – http://www.smittskyddsinstitutet.se/publikationer/smis-nyhetsbrev/sjukrapport/sasongen-
20092010), the solid red line represents the epidemic curve for the 2009-2010 pandemic influenza (which, unfortunately, was already widespread when our registration started).

The left panel shows the incidence of ILI for all ages combined, and the right panel shows the same curve broken down into age groups. The dotted line in the left graph represents the 2007-2008 seasonal influenza and the dashed line the 2008-2009 influenza. It can be seen that the pandemic  A(H1N1) influenza activity was rather unimpressive, compared to the seasonal influenza epidemics, notably the one in 2008- 2009. In week 47 (third week of November), a sharp decline in the incidence began. A similar decline
was noted also in other European countries, albeit it began 1 or 2 weeks earlier. The age-specific incidence curves show that the peak in mid-fall was driven to a large extent by children and their parent generation (age 15-39 years). In the youngest age group, the peak incidence was well above 2000 per 100,000 person-weeks. In middle-aged and old people, on the other hand, the incidence was considerably lower and declining. This pattern is in good agreement with qualitative reports from the Stockholm schools and from the sentinel reporting system. Thus, the data from the PBA was quite reasonable, and in terms of numbers much more reasonable than the existing sentinel reporting. These data have been extremely helpful in SMI’s monitoring of the pandemic.