Vanderbilt Genome-Electronic Records (VGER) Project: QRS Duration

An important potential enabling resource for Personalized Medicine is the combination of a DNA repository with Electronic Medical Record (EMR) systems sufficiently robust to provide excellence in clinical care and to serve as resources for analysis of disease susceptibility and therapeutic outcomes across patient populations. The Vanderbilt EMR is a state of the art clinical and research tool (that includes >1.7 million records), and is associated with a DNA repository which has been in development for over 3 years; these are the key components of VGER, the Vanderbilt Genome-Electronic Records project, a part of NHGRI's eMERGE network. The VGER model acquires DNA from discarded blood samples collected from routine patient care, and can link these to de-identified data extracted and readily updated from the EMR. The phenotype we analyze here is the QRS duration on the electrocardiogram, since slow conduction (indicated by longer QRS duration) is a marker of arrhythmia susceptibility.]]>
Patient Population: Patients with DNA whose first electrocardiogram (ECG) is designated as "normal" and lacking an exclusion criteria. For this study, case and control are drawn from the same population. The only difference will be the QRSd. From this population, we will select 3000 patients with "normal" ECGs, and analyze QRSd as a continuous variable. Hypothetical timeline for a single patient: Notes: ECG1 is the study ECG since it is the first normal ECG and preceded by no abnormal ECGs The "Mildly abnormal" ECG cannot be abnormal by presence of heart disease. It can have abnormal rate, be recorded in the presence of Na-channel blocking meds, etc. For instance, a HR >100 is OK but not a bundle branch block. Y duration = from first entry in the electronic medical record (EMR) until one month following normal ECG Z duration = most recent clinic visit or problem list (if present) to one week following the normal ECG. Labs values, though, must be +/- 48h from the ECG time Exclusion criteria: Criterion Source/Method Impression diagnoses contains: bundle branch blocks intraventricular conduction delay myocardial infarction and ischemia cardiac pacing extrasystoles (PVCs, PACs, etc) tachycardia (ventricular, atrial, supraventricular, etc) arrhythmias AV blocks (any degree) aberrant conduction hypertrophy (LVH,RVH) electrocardiographic wave form abnormalities suggestive of disease (e.g., q waves and ST changes). Natural Language Processing (NLP) on ECG impression. Excludes all but negated terms (e.g., exclude those with possible, probable, or asserted bundle branch blocks). Also excludes normalization negations like "LBBB no longer present." No Known heart disease at or before ECG time NLP for notes, Problem Lists at or near ECG time, ignoring Family Medical History and Allergy sections ICD9 and CPT codes at or near ECG time describing heart disease EMR Lab calues: Positive cardiac enzymes (CPK-MB > 8, Troponin > 0.05) BNP > 100 ECGs recorded during treatment with sodium channel blocking drugs are excluded: Presence of drug names (detected via NLP) from problem list or notes at or near ECG time. We used coded inpatient medication data for ECGs done while an inpatient. Absence of both a problem list and note containing medications and past medical history before or immediately after the time of the ECG. Note section tagging to detect non-empty past medical history and medication sections. ECGs recorded during these lab abnormalities are ignored EMR Lab values: K > 6, K < 3.5 Ca < 8 or Ca > 11 Mg < 1.7 Definitions: Term Definition Normal ECG QRSd between 65-120 ms ECG designated as "NORMAL" Heart Rate between 50-100 Sodium channel blocking drugs flecainide, propafenone, mexiletine, lidocaine, quinidine, procainamide, disopyramide, amiodarone, imipramine, amitryptiline (>25mg total daily dose), lithium, encainide, moricizine, quinine, desipramine, propoxyphene, tocainide Heart disease Presence of concepts (must be "probable" or "asserted"): Any presence of coronary disease concepts Any type of heart failure Any type of valvular disease Any type of cardiomyopathy Ventricular hypertrophy Any type of arrhythmia Any type of cardiac conduction problem Heart transplants Presence of ICD9 near ECG: (codes representing concepts above) Presence of CPT codes representing cardiac surgery. ]]>
QRS Duration Time Line February 2007 - BioVU became operational January 2008 - Began QRS phenotyping process January/March 2009 - Selected samples to be genotyped April 2009 - Samples shipped and received at The Broad Institute ]]>

个性化医疗（Personalized Medicine）的一项重要潜在支撑性资源，是DNA数据库与电子病历系统（Electronic Medical Record, EMR）的有机结合——后者需具备足够的稳定性与先进性，以提供优质临床服务，并可作为跨患者群体开展疾病易感性与治疗结局分析的研究资源。范德堡大学的电子病历系统是当前顶尖的临床与研究工具，收录超过170万条病历记录，其配套的DNA数据库已开发超过3年；二者均为范德堡基因组-电子病历项目（Vanderbilt Genome-Electronic Records Project, VGER）的核心组成部分，该项目隶属于美国国家人类基因组研究所（National Human Genome Research Institute, NHGRI）的eMERGE联盟网络。

VGER项目的样本获取模式依托常规临床诊疗中废弃的血液样本提取DNA，并可将这些样本与从电子病历系统中提取、可随时更新的去标识化数据进行关联。本研究分析的表型为心电图（electrocardiogram, ECG）的QRS时限（QRS duration, QRSd），因为传导减慢（表现为QRS时限延长）是心律失常易感性的标志物。

### 患者人群
携带DNA且首次心电图被判定为"normal"、无排除标准所列情况的患者。本研究的病例组与对照组均来自同一人群，唯一差异仅为QRS时限。我们将从该人群中筛选3000名心电图为"normal"的患者，并将QRS时限作为连续变量进行分析。

### 单例患者假设时间线
#### 备注
本研究的研究心电图为ECG1，因其为患者首次正常心电图且此前无异常心电图记录。"Mildly abnormal"心电图的异常情况不得包含心脏疾病相关异常，仅可表现为心率异常、使用钠通道阻滞剂类药物时的记录异常等。例如，心率>100次/分为可接受的异常，但束支传导阻滞则不符合该定义。
Y时长：从电子病历系统首次记录患者信息起，至正常心电图检查完成后1个月止。
Z时长：从最近一次门诊就诊或问题清单（若有）起，至正常心电图检查完成后1周止。
需注意：实验室检查结果的采集时间需距心电图检查时间±48小时内。

### 排除标准
1. **心电图印象诊断包含以下情况**：束支传导阻滞、心室内传导延迟、心肌梗死与心肌缺血、心脏起搏、期前收缩（室性早搏PVC、房性早搏PAC等）、心动过速（室性、房性、室上性等）、心律失常、房室传导阻滞（任何程度）、异常传导、心室肥厚（左心室肥厚LVH、右心室肥厚RVH）、提示疾病的心电图波形异常（如Q波与ST段改变）。
排除方法：对心电图印象报告进行自然语言处理（Natural Language Processing, NLP），仅排除明确存在或经证实的相关术语（例如排除存在可能、疑似或确诊束支传导阻滞的患者），同时排除"LBBB no longer present."这类表示异常已消除的否定表述。
2. **心电图检查时或此前无已知心脏疾病**。
排除方法：对心电图检查前后的病历笔记、问题清单进行自然语言处理，忽略家族病史与过敏史章节；结合心电图检查前后的ICD-9与CPT编码中描述心脏疾病的条目；同时参考电子病历系统中的实验室检查结果：心脏酶学指标阳性（肌酸激酶同工酶CPK-MB>8，肌钙蛋白>0.05）、B型钠尿肽BNP>100。
3. **服用钠通道阻滞剂类药物期间记录的心电图被排除**。
排除方法：通过自然语言处理识别心电图检查前后的问题清单或病历笔记中的药物名称；对于住院期间完成的心电图，则使用编码后的住院患者用药数据进行筛选。
4. **心电图检查前后既无问题清单，也未留存包含用药史与既往病史的病历笔记**。
排除方法：通过病历章节标记识别非空的既往病史与用药章节，若二者均缺失则符合排除条件。
5. **存在以下实验室异常期间记录的心电图被忽略**：电子病历系统中的实验室检查结果显示血钾>6mmol/L或<3.5mmol/L、血钙<8mg/dL或>11mg/dL、血镁<1.7mg/dL。

### 术语定义
| 术语 | 定义 |
| --- | --- |
| 正常心电图 | QRS时限为65~120ms；心电图被判定为"NORMAL"；心率为50~100次/分 |
| 钠通道阻滞剂类药物 | 包括氟卡尼、普罗帕酮、美西律、利多卡因、奎尼丁、普鲁卡因胺、丙吡胺、胺碘酮、丙米嗪、阿米替林（每日总剂量>25mg）、锂盐、恩卡尼、莫雷西嗪、奎宁、地昔帕明、右丙氧芬、妥卡尼 |
| 心脏疾病 | 需满足以下"probable"或"asserted"的概念存在：任何冠心病相关概念、任何类型心力衰竭、任何类型瓣膜疾病、任何类型心肌病、心室肥厚、任何类型心律失常、任何类型心脏传导问题、心脏移植；或心电图检查前后存在ICD-9编码对应上述概念；或存在代表心脏手术的CPT编码。

### QRS时限研究时间线
2007年2月：BioVU数据库正式上线
2008年1月：启动QRS表型分析流程
2009年1月/3月：筛选用于基因分型的样本
2009年4月：样本运送至博德研究所（Broad Institute）并完成接收