Annika Linde

Web queries as a source for syndromic surveillance.

In the field of syndromic surveillance, various sources are exploited for outbreak detection, monitoring and prediction. This paper describes a study on queries submitted to a medical web site, with influenza as a case study. The hypothesis of the work was that queries on influenza and influenza-like illness would provide a basis for the estimation of the timing of the peak and the intensity of the yearly influenza outbreaks that would be as good as the existing laboratory and sentinel surveillance. We calculated the occurrence of various queries related to influenza from search logs submitted to a Swedish medical web site for two influenza seasons. These figures were subsequently used to generate two models, one to estimate the number of laboratory verified influenza cases and one to estimate the proportion of patients with influenza-like illness reported by selected General Practitioners in Sweden. We applied an approach designed for highly correlated data, partial least squares regression. In our work, we found that certain web queries on influenza follow the same pattern as that obtained by the two other surveillance systems for influenza epidemics, and that they have equal power for the estimation of the influenza burden in society. Web queries give a unique access to ill individuals who are not (yet) seeking care. This paper shows the potential of web queries as an accurate, cheap and labour extensive source for syndromic surveillance.

SMS versus telephone interviews for epidemiological data collection: feasibility study estimating influenza vaccination coverage in the Swedish population

This study compared the use of Short Message Service (SMS) on mobile phones and the use of telephone interviews in collecting self-reported data about influenza vaccination. Through random selection from the Swedish population registry, 2,400 individuals were assigned to be contacted through SMS (SMS-group), and 2,150 were assigned to undergo personal telephone interviews (TI-group). Both groups were asked three questions about influenza and influenza vaccination. Mobile phone numbers were found for 1,055 persons in the SMS-group of whom 154 (6% of the original sample; 15% of all who had a listed mobile phone number) responded. Landline or mobile phone numbers were found for 1,636 persons in the TI-group and 1,009 (47% of the original TI sample; 62% of those where a telephone number was found) responded. The vaccination data collected via SMS was not statistically significantly different from data collected through telephone interviews, and adjustment for different background factors did not change this. Compared to the original sample, there was an under representation of elderly and less educated individuals among the participants in the SMS-group, and under representation of less educated in the TI-group. Though the participation rate was low, SMS is a feasible method for collection of information on vaccination status data among the Swedish population compared to telephone interviews.

Predictions by early indicators of the time and height of the peaks of yearly influenza outbreaks in Sweden.

Aims: Methods for prediction of the peak of the influenza from early observations are suggested. These predictions can be used for planning purposes. Methods: In this study, new robust methods are described and applied to weekly Swedish data on influenza-like illness (ILI) and weekly laboratory diagnoses of influenza (LDI). Both simple and advanced rules for how to predict the time and height of the peak of LDI are suggested. The predictions are made using covariates calculated from data in early LDI reports. The simple rules are based on the observed LDI values, while the advanced ones are based on smoothing by unimodal regression. The suggested predictors were evaluated by cross-validation and by application to the observed seasons. Results: The relationship between ILI and LDI was investigated, and it was found that the ILI variable is not a good proxy for the LDI variable. The advanced prediction rule regarding the time of the peak of LDI had a median error of 0.9 weeks, and the advanced prediction rule for the height of the peak had a median deviation of 28%. Conclusions: The statistical methods for predictions have practical usefulness.

Interactive Voice Response and web-based questionnaires for population-based infectious disease reporting

The authors aimed to evaluate the web and an Interactive Voice Response (IVR) phone service as vehicles in population-based infectious disease surveillance. Fourteen thousand subjects were randomly selected from the Swedish population register and asked to prospectively report all respiratory tract infections, including Influenza-like Illness (ILI—clinical symptoms indicative of influenza but no laboratory confirmation), immediately as they occurred during a 36-week period starting October 2007. Participants were classified as belonging to the web or IVR group based on their choice of technology for initial registration. In all, 1,297 individuals registered via IVR while 2,044 chose the web. The latter were more often young and well-educated than those registered via IVR. Overall, 52% of the participants reported at least one infection episode. The risk of an infectious disease report was 14% (95% CI: 6, 22%) higher in the web group than in the IVR group. For ILI the excess was 27% (95% CI: 11, 47%). After adjustments for socio-demographic factors, statistically non-significant excesses of 1 and 8% remained, indicating trivial differences potentially attributable to the two reporting techniques. With attention to confounding, it should be possible to combine the web and IVR for simple reporting of infectious disease symptoms.

Spatiotemporal characteristics of pandemic influenza

BackgroundPrediction of timing for the onset and peak of an influenza pandemic is of vital importance for preventive measures. In order to identify common spatiotemporal patterns and climate influences for pandemics in Sweden we have studied the propagation in space and time of A(H1N1)pdm09 (10,000 laboratory verified cases), the Asian Influenza 1957–1958 (275,000 cases of influenza-like illness (ILI), reported by local physicians) and the Russian Influenza 1889–1890 (32,600 ILI cases reported by physicians shortly after the end of the outbreak).MethodsAll cases were geocoded and analysed in space and time. Animated video sequences, showing weekly incidence per municipality and its geographically weighted mean (GWM), were created to depict and compare the spread of the pandemics. Daily data from 1957–1958 on temperature and precipitation from 39 weather stations were collected and analysed with the case data to examine possible climatological effects on the influenza dissemination.ResultsThe epidemic period lasted 11xa0weeks for the Russian Influenza, 10xa0weeks for the Asian Influenza and 9xa0weeks for the A(H1N1)pdm09. The Russian Influenza arrived in Sweden during the winter and was immediately disseminated, while both the Asian Influenza and the A(H1N1)pdm09 arrived during the spring. They were seeded over the country during the summer, but did not peak until October-November. The weekly GWM of the incidence moved along a line from southwest to northeast for the Russian and Asian Influenza but northeast to southwest for the A(H1N1)pdm09. The local epidemic periods of the Asian Influenza were preceded by falling temperature in all but one of the locations analysed.ConclusionsThe power of spatiotemporal analysis and modeling for pandemic spread was clearly demonstrated. The epidemic period lasted approximately 10xa0weeks for all pandemics. None of the pandemics had its epidemic period before late autumn. The epidemic period of the Asian Influenza was preceded by falling temperatures. Climate influences on pandemic spread seem important and should be further investigated.

The validity of self-initiated, event-driven infectious disease reporting in general population cohorts.

Background The 2009/2010 pandemic influenza highlighted the need for valid and timely incidence data. In 2007 we started the development of a passive surveillance scheme based on passive follow-up of representative general population cohorts. Cohort members are asked to spontaneously report all instances of colds and fevers as soon as they occur for up to 9 months. Suspecting that compliance might be poor, we aimed to assess the validity of self-initiated, event-driven outcome reporting over long periods. Methods During two 8 week periods in 2008 and 2009, 2376 and 2514 cohort members in Stockholm County were sent one-week recall questionnaires, which served as reference method. Results The questionnaires were completed by 88% and 86% of the cohort members. Whilst the false positive proportion (1–specificity) in the reporting was low (upper bound of the 95% confidence interval [CI] ≤2% in each season), the false negative proportion (failure to report, 1–sensitivity) was considerable (60% [95% CI 52%–67%] in each season). Still, the resulting epidemic curves for influenza-like illness compared well with those from existing General Practitioner-based sentinel surveillance in terms of shape, timing of peak, and year-to-year variation. This suggested that the error was fairly constant. Conclusions Passive long-term surveillance through self-initiated, event-driven outcome reporting underestimates incidence rates of common upper respiratory tract infections. However, because underreporting appears predictable, simple corrections could potentially restore validity.

Self-sampling for analysis of respiratory viruses in a large-scale epidemiological study in Sweden

Viral diagnosis of respiratory tract infections has so far required sampling by health professionals,hampering large-scale epidemiological studies of virus-specific disease outcomes. As part of a population-based, prospective study of work-related risk factors for transmission of viral infections (SWEDE-I), we developed a scheme for self-sampling with nasal swabs. Random selection from the gainfully employed population of a medium-sized town in central Sweden resulted in a study cohort of 2,237 men and women aged 25 to 63 years. From September 2011 through May 2012, the cohort reported all instances of respiratory tract infection or gastroenteritis and participants concomitantly sent self-sampled nasal swabs for analysis using regular mail. Diagnosis of 14 viruses was performed. A total of 1,843 samples were received. The week-wise average delay between disease on set and arrival of the specimens at the laboratory varied between four and six days, and the corresponding median delay was between 3.5 and six days. In line with previous community-based studies, picorna- and coronaviruses dominated in specimens obtained from the self-sampling scheme. The results of self-sampling were contrasted to those from contemporaneous routine clinical sampling, on the same age group, in the adjacent Stockholm county. Although higher proportions of positive samples for respiratory syncytial virus and influenza were observed in the clinical sampling scheme, estimations of seasonality for influenza A and picornaviruses derived from both schemes were similar. Our findings show that nasal self-sampling is feasible in large-scale surveillance of respiratory infections and opens new prospects for population based,virologically verified research on virus spread,burden of disease, and effects of environmental factors or interventions.

Evaluation of an Internet-based monitoring system for influenza-like illness in Sweden.

To complement traditional influenza surveillance with data on disease occurrence not only among care-seeking individuals, the Swedish Institute for Communicable Disease Control (SMI) has tested an Internet-based monitoring system (IMS) with self-recruited volunteers submitting weekly on-line reports about their health in the preceding week, upon weekly reminders. We evaluated IMS acceptability and to which extent participants represented the Swedish population. We also studied the agreement of data on influenza-like illness (ILI) occurrence from IMS with data from a previously evaluated population-based system (PBS) with an actively recruited random sample of the population who spontaneously report disease onsets in real-time via telephone/Internet, and with traditional general practitioner based sentinel and virological influenza surveillance, in the 2011–2012 and 2012–2013 influenza seasons. We assessed acceptability by calculating the participation proportion in an invited IMS-sample and the weekly reporting proportion of enrolled self-recruited IMS participants. We compared distributions of socio-demographic indicators of self-recruited IMS participants to the general Swedish population using chi-square tests. Finally, we assessed the agreement of weekly incidence proportions (%) of ILI in IMS and PBS with cross-correlation analyses. Among 2,511 invited persons, 166 (6.6%) agreed to participate in the IMS. In each season, 2,552 and 2,486 self-recruited persons participated in the IMS respectively. The weekly reporting proportion among self-recruited participants decreased from 87% to 23% (2011–2012) and 82% to 45% (2012–2013). Women, highly educated, and middle-aged persons were overrepresented among self-recruited IMS participants (p<0.01). IMS (invited and self-recruited) and PBS weekly incidence proportions correlated strongest when no lags were applied (ru200a=u200a0.71 and ru200a=u200a0.69, p<0.05). This evaluation revealed socio-demographic misrepresentation and limited compliance among the self-recruited IMS participants. Yet, IMS offered a reasonable representation of the temporal ILI pattern in the community overall during the 2011–2012 and 2012–2013 influenza seasons and could be a simple tool for collecting community-based ILI data.

Difference in immune response in vaccinated and unvaccinated Swedish individuals after the 2009 influenza pandemic

BackgroundPrevious exposures to flu and subsequent immune responses may impact on 2009/2010 pandemic flu vaccine responses and clinical symptoms upon infection with the 2009 pandemic H1N1 influenza strain. Qualitative and quantitative differences in humoral and cellular immune responses associated with the flu vaccination in 2009/2010 (pandemic H1N1 vaccine) and natural infection have not yet been described in detail. We designed a longitudinal study to examine influenza- (flu-) specific immune responses and the association between pre-existing flu responses, symptoms of influenza-like illness (ILI), impact of pandemic flu infection, and pandemic flu vaccination in a cohort of 2,040 individuals in Sweden in 2009–2010.MethodsCellular flu-specific immune responses were assessed by whole-blood antigen stimulation assay, and humoral responses by a single radial hemolysis test.ResultsPrevious seasonal flu vaccination was associated with significantly lower flu-specific IFN-γ responses (using a whole-blood assay) at study entry. Pandemic flu vaccination induced long-lived T-cell responses (measured by IFN-γ production) to influenza A strains, influenza B strains, and the matrix (M1) antigen. In contrast, individuals with pandemic flu infection (PCR positive) exhibited increased flu-specific T-cell responses shortly after onset of ILI symptoms but the immune response decreased after the flu season (spring 2010). We identified non-pandemic-flu vaccinated participants without ILI symptoms who showed an IFN-γ production profile similar to pandemic-flu infected participants, suggesting exposure without experiencing clinical symptoms.ConclusionsStrong and long-lived flu-M1 specific immune responses, defined by IFN-γ production, in individuals after vaccination suggest that M1-responses may contribute to protective cellular immune responses. Silent flu infections appeared to be frequent in 2009/2010. The pandemic flu vaccine induced qualitatively and quantitatively different humoral and cellular immune responses as compared to infection with the 2009 H1N1 pandemic H1N1 influenza strain.