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


108 ISCB 2014 Vienna, Austria • Abstracts - Poster PresentationsTuesday, 26th August 2014 • 10:30-11:00 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust undertaken to address the research aim. This consists of a) Retrospective Audit based on existing data, b) Prospective Survey (to inform continuum of results, prevalence and severity of BAM) and c) Multi-Centre Prospective Observational Study (comparing diagnostic accuracy of SeHCAT, help to develop diagnostic threshold of SeHCAT and assess efficacy of BAS treat- ment).   P2.2.156 The value of using modern epidemiological approaches in studying past influenza pandemics: combining history, war and statistical methods JA Summers1 1 King’s College London, London, United Kingdom   Knowledge of pandemic influenza risk factors, the likely speed and pat- tern of spread, and the expected impact on a healthcare system, is based on the study of past pandemics. Therefore the ability to transform histori- cal records into a quantifiable form for epidemiological/statistical analysis is imperative for such research. As the 1918-19 H1N1 influenza pandemic occurred during World War One (WW1), many of the best-documented aspects from this period are held within military files. Several recent studies have used military records from diverse settings (such as New Zealand (NZ), Australian, British, Japanese, US, and Canadian sources). These records are relevant to understanding influenza epidemiology, en- abling exploration of mortality risk for sub-populations. In particular, a range of statistical methods (i.e. multivariate logistic regres- sion) has been used to assess variables extracted from enlistment data, using both cohort and case-control designs to assess the following: oc- cupational class, geographic variables, health status, military variables, ethnicity, body mass index, marital status and even complexion. The use of archival military sources for epidemiological research is grow- ing; both because of the increasing availability of records, but also be- cause of the increased scientific need to better understand the nature of influenza pandemics. Further research into past influenza pandemics will improve our under- standing of influenza, especially regarding the control of its impact and the optimal targeting of limited health care resources during a pandemic. Additionally, given the severity of the 1918-19 pandemic, this event could arguably be described as a worse-case scenario for guiding future popula- tion-based pandemic planning.   P2.2.163 Assessment of neighbourhood effect on neonatal mortality: translation of area level variance in odds ratio scale in multilevel logistic regression B Thakur1 , V Sreenivas1 , S Dwivedi1 , A Pandey2,3 1 All India Institute of Medical Science, New Delhi, India, 2 Indian Council of Medical Research, New Delhi, India, 3 National Institute of Medical Statistics, New Delhi, India   Introduction: Logistic regression is frequently used in epidemiologi- cal and public health research to measure the binary outcome. The vari- ability at different levels is not directly comparable in multilevel model. Quantifying area-level variance in a meaningful way is a challenge in mul- tilevel logistic regression. Method: We obtained individual and district level information on the bi- nary outcome neo-natal mortality from District Level Household Survey-3. The exploration of data structure confesses the consideration of only two-level structure in analysis, conceptualized as children nested within districts. Estimations of Variance Component Model (empty model) and Random Intercept Model in multilevel logistic analysis were carried out. The me- dian odds ratio translates the area level variance on the odds ratio scale. Result: The median odds ratio was equal to 1.60, in the empty model which shows if a person moves from one district to another district with a higher probability of neonatal mortality, their risk of mortality will increase by 1.6 times, when randomly picking out two persons in different districts. Adjusting the individual effect in random intercept model, this ratio re- duced to 1.54. Area level variance and Intra-class correlation were 0.246 and 0.067 in the empty model as well as 0.210 and 0.059 in the subsequent model respectively. Conclusion: The usual odds ratio are not proper interpretable for district- level covariates because it is impossible to make comparison within dis- trict. As MOR quantifies cluster variance in terms of odds ratios, it is com- parable to the fixed effects odds ratio and can be useful in epidemiological studies. P2.2.182 Regression models for rare events − stroke mortality rates over the last 30 years in Hungary K Virág1 , T Nyári1 , K Boda1 1 University of Szeged, Szeged, Hungary   Modeling the number of occurrences of a disease is a common task in epi- demiological investigations. When the dependent variable describes the counts of rare events then its distribution is skewed to the right, hence the use of the ordinary linear regression is inappropriate. The natural model for count data is a Poisson regression, which fits Poisson distribution to the number of occurrences (or rates) of the event. Poisson model assumes that the mean is the same as the variance. In epidemio- logical studies we often find that the variance is greater than the mean, therefore the data is not well modeled by Poisson regression. Negative Binomial regression can deal with overdispersion. The aim of this study is to compare different count regression models for rare events using SAS 9.2, SPSS 22.0, STATA 9 and R statistical softwares. We perform simulations to compare the coverage probabilities of the con- fidence intervals given by different methods under different conditions. As an empirical application, we analyze stroke mortality rates (ICD: 1981- 1995: 430-438; 1996-2010: I60-I69) in Hungary between 1981 and 2010. Acknowledgement: Katalin Virág was supported by the European Union and the State of Hungary, co-financed by the European Social Fund in the framework of TÁMOP 4.2.4.A/2-11/1-2012-0001 ‘National Excellence Program’.   P2.2.183 Dependence of the effect of altitude on infant as well as maternal related variables on birth weight T Waldhoer1 , K Klebermass-Schrehof2 1 Medical Univiversity of Vienna, Center for Public Health, Epidemiology, Vienna, Austria, 2 Medical University of Vienna, Dept. of Pediatrics, Vienna, Austria   Decrease of birth weight with increasing altitude has been described numerously and shown for different populations as well as for different types of studies. Only few studies have attempted to test whether the effect of altitude depends on other risk factors. Corresponding results showed rather clear independence on other variables. In this study we use about 2 million Austrian birth certificates in order to test for interaction terms with infant and maternal related variables as e.g. gestational age, sex, education, age of mother, year of birth, parity, time to previous birth. Results show that there obviously exist significant as well as relevant in- teractions which can not be detected in small sized studies because of lack of power.  

Pages Overview