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

ISCB2014_abstract_book

122 ISCB 2014 Vienna, Austria • Abstracts - Poster PresentationsWednesday, 27th August 2014 • 11:00-11:30 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust In the South Korea, most of the interventions with an ICER below KRW 30,000,000won/QALY are recommended routinely. Below on threshold 30,000,000won/QALY, pap smear test with 2 year interval (aged 20 to 79 years) was most cost-effectiveness strategy. P3.4.113 Dichotomising highly skewed outcome data using a distributional method: a simulation study M Ofuya1 , O Sauzet2 , JL Peacock1 1 King’s College London, London, United Kingdom, 2 Universität Bielefeld, Bielefeld, Germany   Researchers commonly dichotomise outcome data for ease of interpreta- tion. The distributional method provides a way to dichotomise a continu- ous outcome without losing power by considering the proportion below a given cut-off as a function of the parameters of the Normal distribution (Peacock et al., Stat Med, 2012). This method works when data are Normal or can be transformed to Normal. However, if the data is highly skewed, the commonly used log transforma- tion may not completely remove the asymmetry and in such situations, the Gamma distribution may fit better than the Log-normal. In this study, simulations are used to compare the results provided by the distributional method using log transformed data to the true gamma distributional val- ues, with standard error obtained through resampling methods. Simulations were performed using published data parameters based on a study investigating the effect of Hepatitis A vaccine on antibody titre levels and 10 mlU/ml was considered to be a clinically relevant cut-point. Random gamma variables for a two sample design were generated and the simulated data were log transformed for varied distribution param- eters. Distributional estimates of differences in proportions, risk and odds ratios and their standard errors were obtained for both distributions and compared. Provisional results indicate the distributional method on log-transformed data provides acceptable estimates of standard error for both distribu- tions but for differences in proportions, the effect sizes have bias 25%, risk ratios bias 8.5%.   P3.4.143 Comparison of classification models for sex determination of Polish skulls K Salapa1 , KA Tomaszewski1 , R Chrzan1 , M Pliczko1 , M Gomulska1 , P Fraczek1 , A Sliwinska1 , S Baczkowski1 1 Jagiellonian University Medical College, Cracow, Poland   Sexual dimorphism reveals in the whole skeleton. Many studies shown that sex can be determined by measurements of single bones. One of the most reliable bone structure for this is the skull. Various studies were per- formed to determine the sex by employing different measurements of the skull and usually discriminant analysis or logistic regression were applied. The purpose of this study is to determine whether there are significant differences in correct sex identification based on those two algorithms. The study consisted of 500 archived Polish adult head CT scans (237 [48.18%] males and 255 [51.82%] females), age >21, without any malfor- mation.The measurements of both right and left sides of the palatal bones and skull base, in millimeters, were considered in analysis: the depth of the greater palatine canal, distances between greater palatine foramen and incisive foramen, median palatine suture and posterior nasal spine. A model was created using SPSS Modeler v15.0. Its tasks involved calculat- ing both descriptive characteristics, correlations coefficients, mean com- parison in males and females and stepwise logistic and discriminant (with a leave-one-out cross-validation) functions. The input data was randomly divided into training and testing samples (using a ratio of 70%:30%). Final assessment of model quality was based on percentages of correct sex identification obtained using testing sets (N=152). Percentages of cor- rect sex identification for all model were very similar and slightly exceeded 69%. The best classification function was derived from discriminant analy- sis, which used as predictors the measurements only of the left side of skulls. P3.4.180 Quality of life in Portuguese cancer patients. A structural equation modeling application E Vilhena1,2,3 , JL Pais Ribeiro4,5 , I Silva6 , H Cardoso7 , RF Meneses6 , L Pedro8 , AM da Silva7 , D Mendonça2,3 1 Polytechnic Institute of Cavado and Ave, Barcelos, Portugal, 2 ICBAS, University of Porto, Porto, Portugal, 3 ISPUP, Institute of Public Health, University of Porto, Porto, Portugal, 4 Faculty of Psychology and Educational Sciences, Porto, Portugal, 5 UIPES, Lisboa, Portugal, 6 University of Fernando Pessoa, Porto, Portugal, 7 UMIB/ICBAS and Hospital Santo António/CHP, Porto, Portugal, 8 ESTeSL Polytechnic Institute of Lisbon (IPL), Lisboa, Portugal   Living with a chronic disease is a demanding experience that may affect multiple aspects of an individual’s life. In general, chronically ill patients are responsible for the management of a wide range of psychosocial fac- tors which contribute to their quality of life (QoL). QoL has become an important concept for health care. Cancer can produce many different symptoms. An increasingly important issue in oncology is to evaluate QoL in these patients. The aim of the present study was to test the hypothetical model to evalu- ate the simultaneous impact of optimism, treatment adherence and social support on QoL (general well-being, physical and mental health), control- ling for socio-demographic and clinical variables. This study included a sample of 210 in Portuguese cancer patients ap- proached by their physicians, in outpatient departments of the main hospitals in Portugal. All patients completed self-report questionnaires to assess socio-demographic and clinical, psychosocial and QoL variables. Structural Equation Modeling (SEM) was used to test the quality of the hy- pothesized model. Results (performed using EQS 6.1) showed that the hypothesized model fitted the data reasonably, CFI=0.85, RMSEA=0.06, x2 /df=1.77. All factors had a simultaneous independent statistically significant impact in QoL, demonstrating that an attitude more optimistic, a better treatment adher- ence and more social support contribute to a better general well-being, a better physical health and a better mental health. Structural Equation Modeling techniques are considered a major component of applied multi- variate statistical analysis for addressing complex scientific questions. P3.4.181 A structural equation modeling application to test mediation of optimism between stigma and quality of life in Portuguese obese patients E Vilhena1,2,3 , JL Pais Ribeiro4,5 , I Silva6 , H Cardoso7 , D Mendonça2 1 Polytechnic Institute of Cavado and Ave, Barcelos, Portugal, 2 ICBAS, University of Porto, Porto, Portugal, 3 ISPUP, Institute of Public Health, University of Porto, Porto, Portugal, 4 Faculty of Psychology and Educational Sciences, Porto, Portugal, 5 UIPES, Lisboa, Portugal, 6 University of Fernando Pessoa, Porto, Portugal, 7 UMIB/ICBAS and Hospital Santo António/CHP, Porto, Portugal   Quality of life (QoL) has become an important concept for health care. It is a construct composed of a number of factors that contribute to individu- al’s well-being and adjustment to chronic diseases. Obesity is considered one of the more relevant problems of public health in modern societies, as it is a factor predominant risk for the development of various diseases. They are patients are forced to live with the limitations imposed by their conditions.

Pages Overview