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


134 ISCB 2014 Vienna, Austria • Abstracts - Poster PresentationsWednesday, 27th August 2014 • 15:30-16:00 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust ence in NAR between two groups (McNemar´s test: χ2=0.235; P=.628; OR=1.429; 95% CI: 0.544, 3.753; P=.469). Acknowledgement: This study funded by the Korea Healthcare Technology R&D Project, Ministry of Health and Welfare, Republic of Korea (HI13C00040200), the Seoul National University Bundang Hospital Research Fund (04-2011-004), and Dasol Life Science Inc (06-2013-107), all of which have no role other than funding.   P4.4.99 The mediating role of mental health in the relations between dietary behaviors and general health MR Maracy1 , S Nematollahi1 1 Isfahan University of Medical Sciences, Isfahan, Islamic Republic of Iran   Aim: The aim of this study is to find a pathway to explaining the effect of dietary behaviors on mental health and psychological functioning. Material and Methods: This cross-sectional study contains 4763 par- ticipants from the employees of Isfahan University of Medical Sciences (IUMS). In this study 4 questionnaires were administered which were provided as followed: 21-item dietary behaviors questionnaire, Hospital Anxiety and Depression Scale questionnaire (HADS), and 12-item General Health Questionnaire (GHQ-12). We started the analysis with the SEM model with 4 unobserved latent and 16 observed variables. Results: The results show that the regression coefficient (SE) of diet be- havior on mental health and mental health on GHQ were -1 (0.37) and 0.02 (0.01) with their p-values as 0.007 and 0.01, respectively. For this model the estimated RMSEA is 0.062 with 90% confidence interval of (0.060-0.065). In addition, the comparative fit indices were as the following: 0.860 for NFI, 0.810 for RFI, 0.866 for IFI, 0.818 for TLI, and 0.866 for CFI. All these indices represent fairly good fit of the model to the data. Conclusion: We can conclude that“lower scores on diet behavior - higher scores on mental health problems”and“ higher mean scores in depression and/or anxiety - higher scores in GHQ domains”. The SEM results showed that dietary behaviors have significant impact on depression and/or anxi- ety and general health status.   P4.4.104 The distribution of apolipoprotein E genotype in relation to age and origin of birth K Mehlig1 , S Kern2 , J Kern2 , H Zetterberg2,3 , DS Thelle1,4 , I Skoog2 , L Lissner1 , K Blennow2 , A Börjesson-Hanson2 1 Public Health Epidemiology, University of Gothenburg, Gothenburg, Sweden, 2 Dept of Psychiatry and Neurochemistry, Gothenburg University, Gothenburg, Sweden, 3 UCL Inst of Neurology, London, United Kingdom, 4 Dept of Biostatistics, University of Oslo, Oslo, Norway   The e4 allele of the apolipoprotein E gene (APOE) has been associated with increased risk for age-dependent diseases, e.g. atherosclerosis and Alzheimer’s disease. In contrast, the same allele has been suggested to protect against age-related macular degeneration. When examining the age-dependent risk for a disease, it is important to consider the differen- tial risk of genotype-related mortality. Several studies examining the proportion of APOE alleles over lifespan ob- served a stable prevalence of e4 in midlife and a decrease from a certain age, however, the value of this turning point differed between studies. In addition, geographical variation of e4 prevalence has been found, e.g. along the European-African axis, with a minimum at Mediterranean lati- tudes and maxima near equator and North Pole. We examined APOE allele prevalences in 4570 subjects from four Swedish population studies in relation to age (25-99 years) and origin of birth. E4 prevalence varied between 16% in Scandinavia, 5% in Southern Europe, and 6% in the Middle East, with corresponding changes in e3 prevalence. When combined in two groups, e4 prevalence was age independent and constantly higher in subjects from Nordic compared to non-Nordic coun- tries up to 80 years. After age 92, e4 prevalence decreased from 15% to 9%, observed in subjects from Nordic countries only. The high turning point is explained by presence of participants with APOE-related diseases such as dementia in our gerontologic population studies. We show that age-dependent variation of ethnic composition and selec- tion bias could explain conflicting results observed previously. P4.4.126 Estimating the variance of a propensity score matching estimator: another look at right heart catheterization data R Pingel1 1 Uppsala University, Uppsala, Sweden   This study considers the implementation of a variance estimator when es- timating the asymptotic variance of a propensity score matching estima- tor for the average treatment effect. We investigate the role of smoothing parameters in the variance estimator and also propose using local linear estimation. Simulations demonstrate that large gains can be made in terms of mean squared error by properly selecting smoothing parameters and that local linear estimation may lead to a more efficient estimator of the asymptotic variance. The choice of smoothing parameters in the vari- ance estimator is shown to be crucial when evaluating the effect of right heart catheterization, i.e., we either show a negative effect on survival or no significant effect depending on the choice.   P4.4.171 Sensitivity analysis of dissociative principal strata effect: application to a bone fracture prevention trial Y Uemura1 1 Tokyo University, Tokyo, Japan   In medical studies, there are situations in which the interest is to esti- mate the treatment effect on outcome not due to change in the variable caused by the treatment. For example, in randomized trial for osteopo- rosis subjects, researchers may be interested in the preventive effect of the treatment on the fracture not mediated through bone mineral density. For such research question, we may use a framework of principal stratifi- cation, and estimate the casual effects within principal strata where the posttreatment variable (BMD) is unaffected by the treatment. This disso- ciative principal strata effects (DPSE) may be estimated using the methods proposed by Hyden et al (Biometrics, 2005), Chiba (Biom J,2011), and Diag et al (JASA, 2012). However, the DPSE cannot be identified without making any assumptions. Different assumptions are required for each estimation method proposed, although some of which are quite difficult to interpret and may not hold in actual studies. Using the above example of osteoporosis patients, I inter- pret each assumption. Also, I propose a sensitivity analysis for the estima- tion method proposed by Diang. This method uses a pretreatment covariate that assumes to be inde- pendent of outcome variable given principal strata and the treatment randomized. The above methods of estimating DPSE and the proposed sensitivity analysis will be applied to the data of randomized trial of os- teoporosis study.  

Pages Overview