TY - T1的癫痫流行病学病例定义的开发和验证使用澳大利亚行政健康数据(S6.007) JF -神经学乔-神经学六世- 84 - 14补充SP - S6.007 AU -迈克尔·谭盟-伊恩·威尔逊AU -瓦内萨布拉加莎盟苏菲Ignatiadis AU -雷蒙德波士顿盟Vijaya Sundararaj首页an AU -马克·库克盟Wendyl D’索萨Y1 - 2015/04/06 UR - //www.ez-admanager.com/content/84/14_Supplement/S6.007.abstract N2 -目的:Data-linkage是一个新兴的强大的工具支持医疗疾病和健康状况可能使用定期收集集中的数据库连接。然而,编码诊断癫痫诊断的有效性可能会限制任何派生的效用估计基于这些数据。我们报告的诊断有效性选择算法使用ICD-10AM(澳大利亚修改)编码仅癫痫和的医药信息识别癫痫病例。设计/方法:回顾性验证icd - 10编码医院记录和制药研究数据采样300连续潜在的癫痫病例和300随机选择通过3/7/2012 10/7/2013。两个癫痫专家独立分类诊断。评定等级的协议确认诊断与分歧了三分之一的癫痫专家,前达成最终共识。多变量逻辑回归模型拟合来确定最优编码算法癫痫和内部验证。敏感性,特异性,阳性预测值(PPV),阴性预测值(NPV)和area-under-ROC-curve (AUC)计算。结果:157/300 (52.3 [percnt])病例和0/300 (0 [percnt])控制被证实患有癫痫症。kappa两分的协议为0.89 (95 [percnt]可信区间0.81 - -0.97)。 ICD-10 codes alone had reasonable accuracy with G40 and G41 conferring specificity of 67.7[percnt] (95[percnt]CI 63.1-72.1) and PPV of 50.9[percnt] (95[percnt]CI 45.0-56.7). The model utilising G40, G41, and 蠅1 anti-epileptic drug (AED) conferred the highest PPV of 81.4[percnt] (95[percnt]CI 73.1-87.9) and specificity of 95.0[percnt] (95[percnt]CI 92.6-96.9). The AUC was 0.90 (95[percnt]CI 0.88-0.93). CONCLUSIONS: Diagnostic coding precision of epilepsy in an Australian hospital setting is similar to other high-income countries. When combined with the number of AEDs the precision of case identification for epilepsy data-linkage design is considerably improved, but at the cost of sensitivity, suggesting case-control rather than incidence or prevalence designs maybe more suitable.Disclosure: Dr. Tan has nothing to disclose. Dr. Wilson has nothing to disclose. Dr. Braganza has nothing to disclose. Dr. Ignatiadis has nothing to disclose. Dr. Boston has nothing to disclose. Dr. Sundararajan has nothing to disclose. Dr. Cook has received research support from UCB Pharma, Pfizer Inc., and Sanofi-Aventis. Dr. D'Souza has nothing to disclose.Tuesday, April 21 2015, 1:00 pm-2:45 pm ER -