Please activate JavaScript!
Please install Adobe Flash Player, click here for download


ISCB 2014 Vienna, Austria • Abstracts - Poster Presentations 105Tuesday, 26th August 2014 • 10:30-11:00 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust Results: Most of the patients (56%) came to the hospital in early terms and in a timely manner has received antiviral therapy. Flu-positive patients experienced more chills, myalgia, nasal congestion, dry cough, conjuncti- val injection, dyspnea than patients withother ARI (χ2, p<0,01). The most frequent complications of admission were bronchitis (13.7%), pneumonia (5.6%), sinusitis (1.9%). For therapy of the most commonly prescribed Umifenovir (Arbidol): at the pre-hospital stage 10.2%, in the hospital from 65.2% in monotherapy, and at 14.3% in combination with Oseltamyvir (2.6%) and recombinant inter- feron (11.7%). The early beginning therapy of influenza (Umifenovir, Oseltamyvir) to minimize the risk development of complications and showed the effective ness of the basic criteria. Conclusion: Results complement the criteria for a rational and effective choice of the drug. An early beginning of antiviral therapy (Umifenovir or/ and Oseltamyvir) is one of the major factors ensuring the effectiveness of influenza therapy and reduction of the risk for complications.   P2.2.41 Sensitivity analysis for possible bias due to event-dependent observation periods in self-control case series analysis MAJ de Ridder1 , GMC Masclee1 , CN Dodd1 , S Romio1,2 , MCJM Sturkenboom1 1 Erasmus University Medical Center, Rotterdam, The Netherlands, 2 University of Milano-Bicocca, Milan, Italy   The self-controlled case series (SCCS) method analyses the association be- tween a time-varying exposure and an event, using data on successive time episodes of cases only, including information on exposure status and occurrence of an event. A key assumption is independence between total length of observation and time of event. To allow event-dependent censoring, e.g. event-related death, a modified method was presented (Farrington 2011). After modelling the censoring process, weights are al- located to all episodes, depending on sequence. However, when using public health databases, this sequence may be unknown due to removal of dates for privacy regulations. We did sensitivity analyses to investigate possible bias caused by event-dependent observation times in a study on the effect of concomitant drug use on the risk of upper gastrointestinal bleeding. Survival after the event was modelled with a mixture of an exponential (short-term) and a Weibull (long-term survival) distribution, with param- eters depending on age at event. Random and extreme sequences of epi- sodes were generated and estimates of unmodified and modified SCCS analyses using these different sequences studied. Compared to unmodified, estimated RRs using both random and extreme sequences were moderately decreased and most were within the origi- nal 95% CI. As example, extreme RRs for concomitant use of non-selective NSAIDS and steroids were 2.81 and 4.42, random RRs were between 3.69 and 3.82, versus unmodified RR 4.60 (95% CI 2.43 - 8.68). Using the modified SCCS analysis we were able to show in our data small bias due to ignoring event-dependency of the observation time.   P2.2.52 Smoking statistics in the mid-1920s birth cohorts I Funatogawa1 1 The Institute of Statistical Mathematics, Tokyo, Japan   Early smoking initiation may greatly increase mortality, but the subse- quent deaths from smoking typically become evident in elderly indi- viduals. This obscures the impact of early smoking on mortality in most nations. Although mortality in older ages has not yet been observed for younger birth cohorts, the mid-1920s birth cohorts are currently in their late 80s. Therefore, we showed changes with age in smoking statistics (smoking ini- tiation, smoking prevalence, and lung cancer mortality) of the 1925 birth cohorts of men and women in Japan and the United States (Funatogawa et al. Bull WHO 2013). In this study, we further compare changes in smoking statistics of the mid-1920s birth cohorts across several countries, includ- ing Japan, the United States, Great Britain, Spain, Italy and Norway based on WHO mortality database and literatures. Spanish women showed ex- tremely low smoking prevalence and low lung cancer mortality which levels are similar to those in non-smokers. Japanese and Italian women showed modest smoking prevalence and modest lung cancer mortality. Although Norwegian women showed higher smoking prevalence, lung cancer mortality was similar to those in Japanese and Italian women. American, British and Italian men showed highest lung cancer mortal- ity. American and British men showed high proportion of early smoking initiation and high smoking prevalence in young adulthood. Italian men showed lower smoking prevalence compared to the other two countries. Generally smoking initiation and prevalence are corresponded to lung cancer mortality. P2.2.53 Non-monotonic trends in smoking statistics I Funatogawa1 1 The Institute of Statistical Mathematics, Tokyo, Japan   Many studies concerning smoking and health implicitly assume smoking habit spreads monotonically and then declines. However, Japanese wom- en showed a long-term (i.e., six or seven decades) trend with a decrease and increase in smoking prevalence with birth year, instead of the increase and decrease (Funatogawa et al. Bull WHO 2013). Short-term decreases and increases with birth year (i.e., within two decades) were also seen in Japanese men and British men and women (Funatogawa et al. Bull WHO 2013, Funatogawa et al. BMJ Open 2012). In this study, we further examine changes in age-specific smoking indi- cators (smoking initiation, smoking prevalence, lung cancer mortality) with birth year in other nations based on WHO mortality database and literatures. Chinese women showed a trend with an increase, decrease and increase again in lung cancer mortality with birth year. Although a long-term decreasing trend in smoking prevalence with birth year was seen, the mean age of smoking initiation became younger with birth year. The effect of early smoking initiation on mortality has not yet been fully observed due to the long time lag. Smoking habit does not nec- essary spread monotonically. Therefore, age-specific smoking measures should be paid more attention than summary smoking statistics which ignore non-monotonic trends with birth year.   P2.2.65 Exemplifying the usefulness of combining difference and equivalence tests in spatial maps T Waldhör1 , H Heinzl2 1 Medical Univiversity of Vienna, Center for Public Health, Epidemiology, Kledering, Austria, 2 Medical University of Vienna, CeMSIIS, Vienna, Austria   A choropleth map provides a widely accepted method for graphically pre- senting spatially distributed health data. Quite often local health authori- ties and the general public are concerned whether the health situation in certain areas is relevantly worse than a reference standard. Highlighting statistically significant areas allows the conclusion that the health situation in these areas is better or worse than the reference value. However, this approach does not allow concluding that the situation in some areas is sufficiently close to the reference value. Since statements with regard to the relevance of observed extreme results are of interest for health policy making as well, a combined integration of statistical dif-

Pages Overview