By Amal A. Alzu’bi, PhD; Valerie J. M. Watzlaf, MPH, PhD, RHIA, FAHIMA; and Patty Sheridan, MBA, RHIA, FAHIMA
电子健康记录(EHR) abst的目的raction includes collection of data related to administrative coding functions, quality improvement, clinical registry functions and clinical research. This article examines the different abstraction methods, such as manual abstraction, simple query, and natural language processing (NLP). It also discusses the advantages and disadvantages of each of those methods. The process used for successful EHR abstraction is also discussed and includes the scope and resources needed (time, budget, type of healthcare professionals RHIA, RHIT, etc.). The relationship between EHRs and the clinical registry is also examined with a focus on validity of the data extracted. Future research in this area to examine abstraction methods across hospitals who do data abstraction are being finalized for a future publication.
关键字：电子健康记录、电子健康档案、抽象、自然l language processing, NLP, query, quality improvement, patient safety.
The widespread adoption of electronic health record (EHR) systems makes it possible to retrieve patient records digitally and to extract useful clinical data. Therefore, several secondary applications have become accessible such as quality management, health management and translational research.1All of these secondary applications aim to improve patient care.2,3The overall quality of healthcare and patient treatment depends heavily on the quality of data. Therefore, having inaccurate, incomplete, and inconsistent data and documentation can result in errors and adverse events4that may affect patient safety5, limit health information exchange (HIE), and hinder clinical research.6
In the remainder of this article, we discuss the methods of data abstraction. Advantage and disadvantages of data abstraction, the key factors for successful abstraction within EHRs and registries will also be discussed. We also discuss the importance of having a health information managementprofessional at the forefront of clinical data abstraction methods.
The Centers for Medicare & Medicaid Services (CMS) developed the Meaningful Use program, now called Promoting Interoperability, to help healthcare professionals and hospitals improve quality, efficiency, and safety of patient health through the use of certified EHRs.7,8Thus, it is increasingly important to get meaningful and efficient methods for data collection, sharing and reporting. Furthermore, it is important to have efficient methods for abstracting the needed clinical data from EHR systems and other clinical documents. Three methods that can be used for data abstraction include, manual abstraction, search engines, and abstraction using natural language processing (NLP). Several measures can be used to evaluate the quality of each one of these methods, including, completeness, correctness, concordance, currency, and plausibility.9
Manual Record Abstraction. Manual abstraction is the process of collecting important information from a medical record and transcribing it into discrete fields within the new EHR. Structured data parts (coded data) in the EHR such as, medications, diagnoses, and an active medication allergy list, helps to abstract the needed information from EHR systems.10Manual abstraction can be performed by health information management professionals, nurses, physicians, or other individuals who have training in data abstraction. The review of the entire medical record allows one to collect more specific clinical details, especially for the information that is not readily coded using the existing coding systems.11
Manual abstraction helps to integrate discrete patient data into the EHR and make them readily available for healthcare providers. It also allows for triggering some decision support alerts that are related to information integrated into the EHR.12
Although the manual abstraction method is convenient and easy to understand, it has several limitations:
1. Some outcome measures, such as those related to cancer recurrence, are usually documented in unstructured notes and reports. Thus, it will be hard to extract all the required information since there are limited structured parts.13.14
2. Manual abstraction can be time consuming and expensive.
3. Manual abstraction methods threaten the privacy of patient information.15
4.Manual abstraction may increase errors. For compliance with CMS regulations, clinical data abstractors need to review patient records to identify the ones that meet the guidelines.16回顾这增加了大量的数据risk of making errors. The situation will be worse when dealing with narrative information that lacks standardization. Therefore, data abstractors may get unreliable results.
搜索引擎（简单查询）。Several studies have shown that physicians prefer to enter their comments in some unstructured free-text entries even if there are options for structured coding elements in the system.17,18Additionally, unstructured free text entries are always required for some complex tasks such as, clinical trial recruitment.19Extracting some useful clinical data from such unstructured free-text entries is a complex task that face several challenges20including; the physician tendency to use some acronyms, abbreviations21, negation22,23, and hedge phrases.24Lack of standard grammar and punctuation may lead to ambiguity and misunderstanding.21The difficulty in automatically processing some context-sensitive meanings25and temporal relationships26is another challenge. There is a significant need for searching full text medical records.27有一些简单的SQL查询28and search engine tools can help in conducting this full-text search.
In order to solve the problem with abstraction and obtain useful information from the unstructured clinical notes, researchers at the University of Michigan have developed the Electronic Medical Record Search Engine (EMERSE).1This is a full-text search engine that is mainly designed to extract useful information from the narrative clinical notes in the EHR systems. CISearch is another tool for searching free-text reports within EHR systems.29
抽象的ion Using NLP
NLP can be defined as computation algorithms for analyzing machine readable unstructured text.15NLP can conclude the meaning behind the words. Automated data abstraction using NLP has the potential to convert the unstructured text-free notes into structured and codified format.14Thus, NLP is an efficient alternative for manual abstraction.30NLP-based systems can reduce the time and efforts of manual abstraction in large-scale population-based studies.
NLP有可能提取所有必要的信息并执行一些复杂的多变量查询。31它标记每个单词，并将其放入可用于报告的离散格式。此外，NLP可以识别相关的单词和短语。例如，高血压和高血压可以被认为是对高血压一词的整体描述。专家们认为，有意义的使用/促进互操作性可能是NLP采用的最大驱动因素。31This is due to the ability of the NLP system to search through a large volume of documents and extract information that are related to meaningful use data elements, such as, a problem list, procedures, medications, allergies, vital signs, social history, and quality measure information.
NLP has been successfully used to abstract useful information in several applications including; emergency medicine physician visit notes32, pathology reports33,34, identifying individuals based on cancer screening35, abstracting findings from imaging36，进行药物基因组学研究37, extracting cancer stage information from narrative EHR data38, and identification of breast and prostate malignancies described in pathology reports.39
Advantages of the abstraction process include:40
1. Ensure correct placement of data into their intended field in the EHR.
2. Speed up the go-live process for physicians since the abstraction can help to provide easy and rapid access to patient data.
3. Save electronic storage space since abstracting only the needed information requires less storage space than whole clinical documents.
4.抽象的ion is a source of supplemental information that supports claims information, which in turn provides more specific evidence for clinical care.41For example, for some measures, claims information is incomplete. So, information from the abstraction process can be used to supplement evidence of the service provided, to verify the population that is being measured.
5. Abstraction supports key processes such as coding and reimbursement, quality improvement, billing audits, and clinical research.42
Also, abstraction can have some disadvantages since it may take extra time and resources in order to enter all the patient information into the EHR.Table 1provides a summary of the advantages and disadvantages for each type of clinical data abstraction method.
Successful Abstraction Process:Care Communications，Inc。提供了成功的数据抽象的一个例子，该公司是提供数据抽象服务的领导者，现在已成为CIOX Health的一部分。根据他们在数据抽象方面的经验，重要的是要满足一些关键因素43: Increasing the medical records procurement rate, enhancing data integrity using an inter-rater approach and working with specialists in the field of health information management who are familiar with HIPAA, ongoing status reporting, and personalized project management.
1. Determine the scope of the abstraction, which means deciding what data should be abstracted and when and whether there are some special abstraction needs for sub-specialists.
3. Determine the budget for the abstraction process based on the scope of the abstraction.
4.Determine who will do the real abstraction of data and how the abstractors will be trained.
Scope of Abstraction:Examples of abstracted data include:12,44demographics, scheduled appointments, active orders, allergies, medications, immunizations, chronic conditions, problem lists, hospital discharge summaries, special studies (echocardiograms, pulmonary function tests, etc.), and patient history (medical, surgical, social and family). The chart abstraction process may also include the identification of key paper clinical documents that need to be included in the new EHR by scanning those records into the electronic chart prior to bringing the new EHR live.
Six important categories should be recorded when doing abstraction including:45
1. Impact on clinical outcomes (length of stay, morbidity, mortality, validated measure of health-related quality of life (HRQOL) or functional status, adverse events).
2. Impact on health care process outcomes (preventative care ordered/completed, clinical study ordered/completed, treatment ordered/prescribed, impact on user knowledge).
3. Impact on workload, efficiency, and organization of health care delivery (number of patients seen/unit time, clinician workload, efficiency).
4.影响relationship-centered出来comes (patient satisfaction).
5. Impact on economic outcomes (cost).
6. Impact on health care providers (HCP) use and implementation (HCP acceptance and satisfaction, implementation of clinical decision support system (CDSS).
In general, coders abstract Present on Admission (POA), Hospital-Acquired conditions (HAC), some patient safety indicators (PSI) and the Core Measures.46Additionally, many facilities require their coders to check the charges for services or enter charges altogether based on the type of record they code. Based on a survey done by Himagine solutions (www.himaginesolutions.com) on their field coders47, the coders reported that there are many more elements that are currently being abstracted in an effort to capture data, streamline the process, and assure the accuracy of input.
Time of Abstraction. The time required to complete the abstraction depends on the clinical practice48(NextGen Healthcare49). Generally, patients see their physician three to five times per year. Thus, the abstraction volume will decrease in the first two months. However, the abstraction volume will keep increasing when having new patients, and thus the time will also increase.
Budget. Generally, the data abstraction process is labor intensive and requires solid data validation and quality control mechanisms.50The budget for abstraction and the needed information varies depending on the scope, size, and the needs of the clinical practice.12Thus, the budget can range from very little to very high.
Who Will Do the Abstraction?Based on the NextGen Healthcare experience, the abstraction process needs to be delegated to either; nurses or medical abstractors which can include health information management professionals. Some of the NextGen Healthcare clients have used their current health information staff to do the abstraction. One benefit of using the current HIM staff as data abstractors is to reduce the time required to do abstraction since they will be familiar with the practice and the EHR. NextGen Healthcare clients have suggested that there might be a need to hire some temporary abstractors for the first two to three months. Through time, the amount of information that needs to be abstracted will decrease. Additionally, HIM staff, physicians, nurses will become more proficient with the abstraction process, and thus, they will be able to keep up with the abstraction.
Using credentialed HIM professionals (RHIAs and RHITs) and Registered Nurses (RNs) to do the data abstraction will be better than assigning the abstraction task to clinical coders.51该声称的原因是，健康信息管理专业人员（Rhias和RHITS）和RN始终专注于日常任务中的临床数据完整性。因此，他们将能够提供有关患者护理连续性的最有价值的细节。此外，Rhias，Rhits和RNS可以更广泛地了解患者数据，并且可以提取关键细节，因为他们了解所有不同的临床成分，这些临床成分塑造了个人的整个健康状况。51
Organizations can use their health information management professionals, internal nurses, and physicians to do the abstraction, or they can outsource the abstraction to other organizations that have some clinical experts who can do the abstraction.51尽管在内部进行抽象似乎是可行的和具有成本效益的方法，但它可能会降低内部人员的生产力。
The ideal clinical abstraction team can include:52
1. Project manager who can monitor all the abstraction project components such as budget and timeframe.
2. Research manager who can monitor the quality of the abstracted data. He/she needs to have a high clinical and technical expertise.
4.抽象的ors who will conduct the actual abstraction and they should have experience in clinical data abstraction and familiarity with the EHR.
Ideally, data abstractors need to have the following qualifications:53
2. Clinical and research experience relevant to the study being conducted
3. Advanced educational preparation in a health information and health care profession.
重要的是要确定所需的资源，预算和时间限制，并在真正的抽象过程之前和之前。一些研究是资源密集型，需要高级计划，以便为抽象过程的所有步骤提供。例如，护理通信进行的肺癌筛查研究中图表的抽象非常复杂且具有挑战性54and requires high level project management and clinical experience. This study requires screening thousands of medical records within more than twenty hospitals in the nation.
EHRs and Registries
A registry can be defined as an organized system that uses observational study methods to collect uniform clinical data to evaluate specified outcomes for a population defined by a particular disease, condition, or exposure, and that serves one or more predetermined scientific, clinical, or policy purposes. Registries are focused on populations and are designed to fulfill specific purposes defined before the data are collected and analyzed.55On the other hand, EHRs are focused on the collection and use of individual patient health-related information. Although, it seems that both registries and EHRs overlap in functionality, their roles are different. According to the Institute of Medicine (IOM), (which is now called the National Academy of Medicine), an EHR has four core functionalities: health information and data, results management, order entry and support, and decision support.56There are several obstacles to achieve the meaningful communication between systems such as, EHRs and registries. These obstacles are related to confidentiality, security, privacy and data access.55
Currently, there is an increasing demand for physicians to participate in the registries in order to manage safety, evaluate effectiveness, and measure and improve the quality of patient care. Therefore, it is becoming increasingly important that EHRs should serve as an interface for several registries with different purposes at the same time. EHRs can enable health care information to be available and accessible to registries. Additionally, EHRs can provide some relevant information fromthe registry to the physicians such as, information about natural history of disease, safety57, effectiveness, and quality.58Figure 1展示了EHR与注册表之间的关系。
Navaneethan et al.59have described the development of a registry for patients with chronic kidney disease (CKD) that is derived from EHR data. The benefits of this kind of patient registry can range from allowing better aggregation of patient data for practice assessment or quality improvement, to facilitating clinical research. The study shows that the quality of data in this registry is comparable to that of the data from a much more labor-intensive and expensive process of human abstraction. This registry can be used for quality improvement, clinical research, and other important tasks.
抽象有效性. Medical record abstraction is a primary mode of data collection in secondary data use. Abstraction is associated with high error rates.60It is important to validate the abstracted data and ensure that the data are abstracted correctly and consistently. There are several benefits of the validation process41包含：
1. Enhance the clarity of specification through the identification of specification ambiguities that are related to the abstraction process.
2. Help to ensure abstractor consistency through the ongoing monitoring.
4.Provide some information for future internal quality improvement. Strategies for improving the validity of data abstracted from medical records include61training abstractors, masking abstractors to study hypotheses, assessing interrater reliability and agreement, and the re-abstraction of records.62
There are three components of validation41包含：
1. Validating the currently used tools from different vendors and updating the existing tools.
2. Validating the abstraction process and this can be done during the data collection by taking a convenience sample of records and ensure that all measures are abstracted consistently by different vendors to uncover any specific ambiguities.
3. Validating at the end of data collection in order to ensure the integrity and accuracy of the abstracted data.
We have conducted a study to examine abstraction methods across hospitals usinginterviews of managers of abstraction within their healthcare organizations as well as a survey of clients of a large consulting company (Ciox Health) who do data abstraction. Those results are being finalized and will在未来出版物的第二部分中提供。
This research study was supported by a unique research entity between CIOX Health and the University of Pittsburgh, Department of Health Information Management, School of Health and Rehabilitation Sciences.
Amal A. Alzu’bi, PhD, (firstname.lastname@example.org) is Assistant Professor- Department of Computer Information Systems, Jordan University of Science and Technology in Irbid, Jordan.
Valerie J. M. Watzlaf, MPH, PhD, RHIA, FAHIMA, (email@example.com) is Associate Professor and Vice Chair of Education, Department of Health Information Management, University of Pittsburgh, School of Health and Rehabilitation Sciences, .
Patty Sheridan, MBA, RHIA, FAHIMA, (firstname.lastname@example.org) ) isPresident, Sheridan Leadership Consulting (formerlySenior Vice President, HIM Services, at Ciox Health).
1. University of Michigan’s nine-year experience in developing and using the Electronic Medical Record Search Engine (EMERSE). J Biomed Inform, 2015. 55: p. 290-300.
2. Friedman，C。和M. Rigby，《概念化和创建全球学习卫生系统》。Int J Med Inform，2013年。82（4）：p。E63-71。
3. Friedman, C.P., A.K. Wong, and D. Blumenthal, Achieving a nationwide learning health system. Sci Transl Med, 2010. 2(57): p. 57cm29.
4. Maurette，P。和S. Comite Analyse et Maitrise du Risque de la，[to Err是人类：建立更安全的卫生系统]。Ann Fr Anesth Reanim，2002年。21（6）：p。453-4。
5. Abramson，E.L。等人，电子健康记录之间的过渡：对门诊开处方安全的影响。J Gen Intern Med，2011年。26（8）：p。868-74。
6. Clark, J.S., et al., Assessing and improving EHR data quality (updated). J AHIMA, 2013. 84(3): p. 48-53.
7. Office of the National Coordinator for Health Information Technology, D.o.H. and S. Human, Health information technology: initial set of standards, implementation specifications, and certification criteria for electronic health record technology. Final rule. Fed Regist, 2010. 75(144): p. 44589-654.
8. Blumenthal, D. and M. Tavenner, The “meaningful use” regulation for electronic health records. N Engl J Med, 2010. 363(6): p. 501-4.
9. Weiskopf，N.G。和C. Weng，电子健康记录数据质量评估的方法和尺寸：促进临床研究的重用。J am Med Inform Assoc，2013年。20（1）：p。144-51。
10. Floyd, J.S., et al., Use of administrative data to estimate the incidence of statin-related rhabdomyolysis. JAMA, 2012. 307(15): p. 1580-2.
11. Leon-Chisen, N., Coding and quality reporting: resolving the discrepancies, finding opportunities. J AHIMA, 2007. 78(7): p. 26-30; quiz 33-4.
12. Hundt, A., Considerations for Successful Clinical Data Abstraction. HCI, 2015.
13. Meystre, S.M., et al., Extracting information from textual documents in the electronic health record: a review of recent research. Yearb Med Inform, 2008: p. 128-44.
14. Jha, A.K., The promise of electronic records: around the corner or down the road? JAMA, 2011. 306(8): p. 880-1.
15. Carrell, D.S., et al., Using natural language processing to improve efficiency of manual chart abstraction in research: the case of breast cancer recurrence. Am J Epidemiol, 2014. 179(6): p. 749-58.
16. Angi, P., The Challenges of Capturing Meaningful Use Data For The Record 2013. 25(15): p. 30.
17. Zheng, K., et al., Handling anticipated exceptions in clinical care: investigating clinician use of ‘exit strategies’ in an electronic health records system. J Am Med Inform Assoc, 2011. 18(6): p. 883-9.
18. Zhou, L., et al., How many medication orders are entered through free-text in EHRs?–a study on hypoglycemic agents. AMIA Annu Symp Proc, 2012. 2012: p. 1079-88.
19. Raghavan, P., et al., How essential are unstructured clinical narratives and information fusion to clinical trial recruitment? AMIA Jt Summits Transl Sci Proc, 2014. 2014: p. 218-23.
20. Edinger, T., et al., Barriers to retrieving patient information from electronic health record data: failure analysis from the TREC Medical Records Track. AMIA Annu Symp Proc, 2012. 2012: p. 180-8.
21. Barrows Jr, R.C., M. Busuioc, and C. Friedman, Limited parsing of notational text visit notes: ad-hoc vs. NLP approaches. Proc AMIA Symp, 2000: p. 51-5.
22. Agarwal, S. and H. Yu, Biomedical negation scope detection with conditional random fields. J Am Med Inform Assoc, 2010. 17(6): p. 696-701.
23. Mitchell, K.J., et al., Implementation and evaluation of a negation tagger in a pipeline-based system for information extract from pathology reports. Stud Health Technol Inform, 2004. 107(Pt 1): p. 663-7.
24. Hanauer，D.A。等人对冲他们的Mets：在临床文件中使用不确定性术语及其在与患者共享文件时的潜在影响。艾米亚·安努（Amia Annu）。SYMP。Proc，2012年。2012（2012）：p。321-330。
25.Jagannathan, V, et al ., commerc评估ial NLP engines for medication information extraction from dictated clinical notes. Int J Med Inform, 2009. 78(4): p. 284-91.
26. Tang, B., et al., A hybrid system for temporal information extraction from clinical text. J Am Med Inform Assoc, 2013. 20(5): p. 828-35.
27. Yang，L。等人，电子健康记录搜索引擎的查询日志分析。Amia Annu Symp Proc，2011年。2011年：p。915-24。
28. Botsis, T., et al., Secondary Use of EHR: Data Quality Issues and Informatics Opportunities. AMIA Jt Summits Transl Sci Proc, 2010. 2010: p. 1-5.
29. Natarajan，K。等人，对电子健康记录搜索实用程序中临床查询的分析。Int J Med Inform，2010年。79（7）：p。515-22。
30. Nadkarni, P.M., L. Ohno-Machado, and W.W. Chapman, Natural language processing: an introduction. J Am Med Inform Assoc, 2011. 18(5): p. 544-51.
31. Eramo, l., Natural Language Processing. For The Record, 2011. 23(8).
32. Kimia, A.A., et al., An Introduction to Natural Language Processing: How You Can Get More From Those Electronic Notes You Are Generating. Pediatr Emerg Care, 2015. 31(7): p. 536-41.
33. Crowley, R.S., et al., caTIES: a grid based system for coding and retrieval of surgical pathology reports and tissue specimens in support of translational research. J Am Med Inform Assoc, 2010. 17(3): p. 253-64.
34. Coden, A., et al., Automatically extracting cancer disease characteristics from pathology reports into a Disease Knowledge Representation Model. J Biomed Inform, 2009. 42(5): p. 937-49.
35. Denny, J.C., et al., Natural language processing improves identification of colorectal cancer testing in the electronic medical record. Med Decis Making, 2012. 32(1): p. 188-97.
36. Hripcsak, G., et al., Use of natural language processing to translate clinical information from a database of 889,921 chest radiographic reports. Radiology, 2002. 224(1): p. 157-63.
37. Wilke, R.A., et al., The emerging role of electronic medical records in pharmacogenomics. Clin Pharmacol Ther, 2011. 89(3): p. 379-86.
38. Warner, J.L., M.A. Levy, and M.N. Neuss, Feasibility and Accuracy of Extracting Cancer Stage Information From Narrative Electronic Health Record Data. J Oncol Pract, 2015.
39. Strauss, J.A., et al., Identifying primary and recurrent cancers using a SAS-based natural language processing algorithm. J Am Med Inform Assoc, 2013. 20(2): p. 349-55.
40. Hundt, A., Legacy Data Management: Strategic Considerations for Successful Clinical Data Abstraction.
41. Stieg, J., Medical Record Review, D.o. Health, Editor. 2015, New York County Health Services Review Organization (NYCHSRO) New York.
42. Medical records abstracting- Harness the power in your clinical data. 2015[cited 2015 12/15/2015]; Available from:http://solutions.3m.com/wps/portal/3M/en_US/Health-Information-Systems/HIS/Products-and-Services/Abstracting/.
43.临床研究服务。护理沟通s, Inc., 2015.
44. Gettinger, A. and A. Csatari, Transitioning from a legacy EHR to a commercial, vendor-supplied, EHR: one academic health system’s experience. Appl Clin Inform, 2012. 3(4): p. 367-76.
45. Lobach, D., G. Sanders, and T. Bright, Enabling Health Care Decisionmaking Through Clinical Decision Support and Knowledge Management. Evidence Report/Technology Assessments. Vol. 203. 2012, Rockville (MD): Agency for Healthcare Research and Quality (US).
47. himagine solutions. 2019[引用2020年4月21日]；可从：https://www.glassdoor.com/Interview/himagine-solutions-Interview-Questions-E920736.htm.
48. Noss，R。，C。Hoyles和S. Pozzi，《专业知识中的抽象：护士集中概念的研究》。数学教育研究杂志，2002年。33（3）。
49. NextGen.[引用2020年4月21日]；可从：NextGen Healthcare获得
50. Dinh，A。等人，在医师实践中从纸上迁移到EHR。J Ahima，2010年。81（11）：p。60-4。
51. Abatjoglou, G., Uncovering the Unexplored Role and Benefits of Clinical Data Abstraction HIS talk, 2013.
52. Medical Record Abstraction for Real World Data…Simplified Keeping your sanity when integrating medical record data into large scale research projects. 2014[cited 2015 11/29/2015]; Available from:http://www.slideshare.net/CareCommunications/medical-recordabstractionforrealworlddata.
53. Gregory, C. and L. Radovinsky, Research strategies that result in optimal data collection from the patient medical record. Appl Nurs Res, 2012. 25(2).
54. Gatsonis, C., Clinical Data Abstraction. Care Communications inc., 2015.
55. in Registries for Evaluating Patient Outcomes: A User’s Guide, R.E. Gliklich, N.A. Dreyer, and M.B. Leavy, Editors. 2014: Rockville (MD).
56. Medicine, I.o., in Key Capabilities of an Electronic Health Record System: Letter Report. 2003: Washington (DC).
57. Hersh, W., Electronic health records facilitate development of disease registries and more. Clin J Am Soc Nephrol, 2011. 6(1): p. 5-6.
58. Yang, W., et al., [How to read and understand Registries for Evaluating Patient Outcomes: A User’s Guide (2nd Edition)]. Zhongguo Zhong Yao Za Zhi, 2013. 38(18): p. 2958-62.
59. Navaneethan, S.D., et al., Development and validation of an electronic health record-based chronic kidney disease registry. Clin J Am Soc Nephrol, 2011. 6(1): p. 40-9.
60. Nahm，M。，医疗记录抽象的数据准确性。UT SBMI论文，2010年。
61. Reisch, L.M., et al., Training, quality assurance, and assessment of medical record abstraction in a multisite study. Am J Epidemiol, 2003. 157(6): p. 546-51.
62. Cook, E.A., et al., Field methods in medical record abstraction: assessing the properties of comparative effectiveness estimates. BMC Health Serv Res, 2014. 14: p. 391.