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

ISCB2014_abstract_book

106 ISCB 2014 Vienna, Austria • Abstracts - Poster PresentationsTuesday, 26th August 2014 • 10:30-11:00 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust ference and equivalence tests into choropleth maps has been suggested. The approach will be graphically exemplified comparing mortality rates for the circulatory system, respiratory system and Alzheimer disease. Data provided by Statistics Austria covering all 121 administrative Austrian dis- tricts for the years 1998 to 2004 will be used and compared with tradi- tional choropleth maps provided by Statistics Austria. Choropleth maps, which only show the variable of interest and their corre- sponding difference test results, can easily misguide local health authori- ties and the general public. The adding of equivalence test results enables a better understanding of regional health care results as the issue of rel- evance is explicitly addressed through a pre-defined equivalence range.   P2.2.79 Medical use of allergic rhinitis under two healthcare system in South Korea B-H Jang1 , I-H Choi1 , J-S Park1 , J-H Park2 , S-H Sun3 , Y Shin1 , S-G Ko1 1 College of Korean Medicine, Kyung Hee University, Seoul, Republic of Korea, 2 College of Korean Medicine, Gachon University, Seoul, Republic of Korea, 3 College of Korean Medicine, Sangji University, Seoul, Republic of Korea   In South Korea, there are two healthcare system - Western Medicine (WM) andTraditional Korean (TKM).This study aims to investigate medical use of allergic rhinitis (AR), especially under two healthcare system. We analyzed characteristics of AR patients from 2011 National Patients Sample (NPS) data of Health Insurance Review & Assessment Service. AR patients were defined as those who were diagnosed with AR as the primary disease (J30.x from ICD-10). We analyzed by dividing into three groups - visiting only the WM institute (A), and both WM and TKM (B), and only the TKM (C). Among 1,375,751 patients in 2011 NPS data, AR patient was 12.5%. A fe- male/male ratio was 1.20, and Average age was 31.8. More than half AR patients were diagnosed as J30.4 (Allergic rhinitis, unspecified), the pro- portion of primary clinics was 88.2%, and the most frequent in September and October. Outpatient clinics were visited by 99.9% of AR patients. Group A was 97.7%, and B was 0.9%, and C was 1.4% of AR patients. In Group A, 36.2% was below 20 age, but 66.5% in B, and 59.1% in C. The annual average cost and the mean visit time per person were the highest in group B. 5.6% of group A had diagnostic test of AR at least one time, and 10.8% of group B had it. According to this study, TKM was not widely used by AR patients in South Korea. In order to reflect the actual medical field, further study including uninsured items is needed.   P2.2.85 Multidimensional outcome. Does the interpretation change with the analysis method? C Klersy1 , L Scudeller1 , P Maras2 , S Doimo2 , A De Silvestri1 , C Tinelli1 1 IRCCS Fondazione Policlinico San Matteo, Pavia, Italy, 2 Ospedali Riuniti Trieste and University of Trieste, Trieste, Italy   Binary outcome measures that indicate the presence or absence of cer- tain medical conditions are widely used in epidemiologic investigations. Moreover, these studies frequently measure an array of such indicators for different medical conditions to make an overall assessment, making it multidimensional. The situation where multiple binary outcomes are simultaneously assessed on the same individual presents some basic methodological problems in that proper statistical modeling here should account for the following features: A) Each individual contributes to mul- tiple outcomes. Thus, the different outcomes are likely to be correlated. B) These multiple outcomes possibly measure the same underlying condi- tion or construct. C) Outcome-specific exposure (i.e., which specific out- comes are associated with the exposure of interest) may still be of scien- tific interest in many situations. To assess the association of any risk factor/exposure with these outcomes, several analysis methods are found in the literature: A) Each outcome is considered separately and independently (binary); B) Each outcome is considered in the framework of repeated measures (binary); C) All out- come components or a certain (acceptable) number of outcomes are satis- fied (binary); D) The number of satisfied components is computed (count); E) The number of satisfied components is computed (few, ordinal); F) All outcome components are expression of an underlying (latent) concept (continuous). Starting from a motivating example in outcome research for the second- ary prevention of ischemic heart disease, with the primary analysis, based on structural equation modeling, following scenario F, we will discuss the implications of using alternative methods for the interpretation of results.   P2.2.105 Sensitivity analysis for the misclassification of competing outcomes in a cohort study in Japan MN Mieno1 , N Tanaka2 , M Sawabe3 , T Arai4 , S Ishikawa1 1 Jichi Medical University, Shimotsuke, Japan, 2 National Center for Global Health and Medicine, Tokyo, Japan, 3 Tokyo Medical and Dental University, Tokyo, Japan, 4 Tokyo Metropolitan Geriatric Hospital, Tokyo, Japan   Background: In epidemiological and clinical studies, causes of death are often measured as outcome. However, the risk estimates are susceptible to bias due to misclassification of causes of death, and the degree of mis- classification actually remains unknown. In this study, we conducted a sensitivity analysis for quantifying the magnitude of bias to the risk esti- mates for three competing causes of death, such as cancer, cardiovascular disease and other causes of death. Methods: The motivated data was from a multi-center population-based cohort study in Japan (The Jichi Medical School cohort study, n=10,692) and we analyzed whether the low lipoprotein(a) [Lp(a)] concentration was related to mortality using competing risks approach. Based on the data- set, we conducted simulations for the sensitivity analysis with several set- tings assuming differential or non-differential misclassification proportion of cause of death as 10%, 20%, 30% and 40%, including covariates which could relate to the misclassification mechanism. Results: The observed risk ratio (low / not-low Lp(a)) for cancer as cause of death was 1.484 [95%CI: 1.147-1.919]. If the misclassification was non- differential and not with over-recording of a specific cause of death, the cumulative incidence rates and risk ratios from Fine & Gray proportional hazards model were not changed so much, whereas in the other differen- tial cases, the estimates were changed depending on the degree of mis- classification. Conclusions: When the outcome misclassification are expected, careful consideration must be given to the interpretation of the results and sen- sitivity analysis like demonstrated here is useful for evaluating the robust- ness of the results.  

Pages Overview