Objectives: (a) to estimate the accuracy of International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) code for amyotrophic lateral sclerosis (ALS) in the Hospital Discharge Database (HDD) of the Italian region Friuli-Venezia Giulia; (b) to identify the predictors of a true positive ALS code; (c) to compare incident and prevalent cases obtained from HDD with those identified in a retrospective population-based study. Methods: Records of all patients discharged 2010–2014 with an ICD-9-CM code for ALS and other motor neuron diseases were extracted from the HDD. For each record, all the available clinical documentation was evaluated to confirm or reject the diagnosis of ALS. ALS incident and prevalent cases were identified. Validity measures were calculated both overall and stratified by patient and hospitalization characteristics. Adjusted odds ratio (aOR), with 95% confidence interval (95%CI), of a true positive code was estimated using unconditional logistic regression. Results: ALS code had sensitivity 92.9%, specificity 75.3%, positive predictive value (PPV) 92.3%, and negative predictive value (NPV) 76.8%. A true positive ALS code was predicted by concurrent codes for respiratory interventions (aOR: 3.82; 95%CI: 2.09–6.99), primary position code (2.78; 1.68–4.62), non-programed hospitalization (2.06; 1.18–3.61), male patient (1.56; 1.06–2.29), and hospitalization length <14 days (1.42; 1.07–2.84). Two hundred and thirty-six prevalent and 187 incident cases were identified, 84% of those detected in the population-based study. Conclusion: ALS code shows very good accuracy and identifies a high percentage of true positive, incident and prevalent cases, but additional sources and an algorithm based on selected variables may further improve case identification.

目标：（a）在意大利弗留利-威尼斯地区医院出院数据库（HDD）中评估国际疾病分类第9次修订，临床修改（ICD-9-CM）编码肌萎缩性侧索硬化（ALS）的准确性朱利亚; （b）识别ALS编码为正的预测因子；（c）将从HDD获得的突发事件和流行病例与基于回顾性人群研究的结果进行比较。方法：从HDD中提取了2010-2014年出院的所有具有ALS和其他运动神经元疾病的ICD-9-CM码的患者的记录。对于每个记录，评估所有可用的临床文档以确认或拒绝ALS的诊断。确定了ALS事件和流行病例。总体上计算有效性指标，并根据患者和住院特征进行分层。使用无条件逻辑回归来估计真实阳性代码的校正比值比（aOR），置信区间为95％（95％CI）。结果：ALS代码的敏感性为92.9％，特异性为75.3％，阳性预测值（PPV）为92.3％，阴性预测值（NPV）为76.8％。并发的呼吸道干预代码（aOR：3.82; 95％CI：2.09–6.99），主要位置代码（2.78; 1.68–4.62），未计划住院（2.06; 1.18–3.61）预测出一个真实的ALS代码阳性，男性患者（1.56; 1.06-2.29），住院时间<14天（1.42; 1.07-2.84）。确定了236例流行病例和187例事件病例，其中84％在基于人群的研究中被发现。结论：ALS代码显示出非常好的准确性，并且可以识别出高百分比的真实阳性，偶发事件和流行病例，但是基于选定变量的其他来源和算法可以进一步改善病例识别。