Susan Helbig, MA, RHIA
Introduction
The Department of Veteran Affairs (VA) has been implementing an electronic health record for several years. Beginning in 1997, VA Puget Sound Health Care System (VAPSHCS) began the systematic implementation of the first version of its computerized patient record system (CPRS). CPRS includes orders, diagnostic test results, progress notes, consultation reports, discharge summaries operative and procedure reports, and imaged documents such as digital radiology films, EKGs, and scanned documents. Although traditional dictation for initial assessments (history and physicals), operative reports, and discharge summaries continued, direct text entry by clinicians for progress notes and discharge summaries was strongly encouraged. The goal was to transform pen and paper documentation into an electronic format so that the many benefits of an electronic health record could be realized. Document scanning was implemented later in the process as it was preferred to have information in a true electronic format.
One of the functionalities in a computerized environment is the ability to re-use information previously documented elsewhere in the medical record. A primary method of re-using information is to copy and paste the original text into a new document. Busy clinicians use this functionality to speed up their documentation process.
Health information management (HIM) professionals in many VA medical centers began to notice that the copy/paste functionality was creating problems in medical record integrity. They noticed that discharge summaries and progress notes were becoming exceedingly long as large amounts of diagnostic data and earlier progress notes were being "copied" into newer documents of the same patient. HIM staff also noticed that sometimes information was being copied from one clinician's note to another clinician's note without any attribution to the original author.
Imagine going to court and having to explain pages of copied material that may appear repeatedly in the record!
Coding hospitalizations or long notes that included copied portions of earlier notes presented difficulties in following the patient's course of treatment during hospitalization and/or what happened during a specific encounter. Sometimes, it was difficult to code because it was not always clear what had been done at each encounter. There was much discussion among VA HIM professionals about this problem.
Overview of Quality Improvement Project
It was evident from HIM discussions that an opportunity for study existed. There was a clear need to investigate and quantify the extent, exact nature, and possible effects of the problem. A quality improvement project was designed, proposed, and implemented at VA Puget Sound Health Care System in Washington State. In recognition of its quality improvement programs, Veterans Integrated Service Network 20 (VISN 20), located in the Pacific Northwest, received a $550,000 Kenneth Kizer Quality Improvement Award to support local quality improvement projects.
The purpose of the quality improvement project was to identify plagiarism events: that is, those instances where text was copied within a document without attribution to the original author. Plagiarism software, Copyfind, which was originally developed by Louis Bloomfield from the University of Virginia for use by university faculty to detect plagiarism in student papers was modified to detect copy/paste events in medical record progress notes, discharge summaries, and consultations.
As of June 15, 2002, there were 4.5 million electronic documents in VAPSHCS' electronic health record. More than 3,000 of these text documents were being added each day. As part of the project, documents were downloaded into a Microsoft Access read-only database and grouped by patient. A sample of 1,479 patients was selected from June 2002. The modified version of Copyfind was run against this sample to identify all copy events for each patient. For this sample of patients, there were 167,076 notes written between 1990 and 2002. Of these, there were 90,762 copy events. A physician and an HIM professional reviewed 243 patients and classified and rated the "risk" of each copy event. The risks were classified as follows:
Table 1. Risk Classifications | Code | Risk Description |
1 | Artifact, not misleading, no risk |
2 | Artifact, minimally misleading, minimal risk |
3 | Human, not misleading, no risk |
4 | Human, minimally misleading, minimal risk |
5 | Human, misleading, some risk |
6 | Human, clinically misleading, major risk |
Risks 5 and 6 are the most potentially dangerous for patients. Risk 4 may have a significant impact on accurate coding and could result in either under coding or fraudulent over coding. Coding impact was not investigated in this study.
Findings
One early observation was that the copy/event ratio to total documents exceeded 50 percent. Decisions were made to filter out trivial and nontrivial but frequent duplications. It is important to realize that the data presented below is an underestimation of the true number of copy events. Discharge summaries were eliminated because there is justification for copying into these documents. In all, 63 note types of 874 note titles were eliminated, thus reducing the number of copy events for manual review by 68 percent.
The remaining sample consisted of 1,479 patients with 29,386 notes. Of these notes, 2,645 were found to contain a total of 29,009 copy events. A manual review of 6,322 copy events for 243 patients was conducted. Sixty-three percent of copied notes, 5.2 percent of the total notes, were found to have human copy events. The process yielded the following results:
Table 2. Data Description Sample Data | Notes | included
167,076 | reviewed
29,386 | Patients | included
1,479 | reviewed
243 | Copy Events | included
29,009 | reviewed
6,322 | | Table 3. Copy Event Statistics Copy Event Patient Record Data | Patient records manually reviewed | 243 | Copy risk 5 notes | 75 | Copy risk 6 notes | 25 | Total copy risk 5 & 6 records | 89 | Percentage of reviewed records w/risk 5 & 6 | 37% | Percentage of patients w/risk 5 & 6 | 6% | Copy Event Data | Copy events manually reviewed | 6,322 | Copy risk 5 events | 294 | Copy risk 6 events | 44 | Percentage of copy events w/risk 5 & 6 | 5% | Copy Event Note Data | Notes reviewed | 29,386 | Notes containing copied data | 2,645 | Notes containing risk 5 & 6 copy events | 338 | Percentage of notes w/risk 5 & 6 | 1.2% | |
Obviously, copying has some risk associated with it. This was a first attempt to quantify the amount and type of risk associated with copying and pasting text. Although we found a relatively low rate of high risk copying, it equated to at least one high risk event for every 10 cases reviewed. We had hoped that Copyfind would be a tool that we could use to monitor documentation on a regular basis and provide feedback to clinicians. We found, however, that the sheer volume of copy events overwhelmed this technology.
Current Status of Project
Results were described in an article and were presented at the November 2002 American Medical Informatics Association (AMIA) meeting. AMIA is a US organization that is dedicated to the development and application of medical informatics in support of patient care, teaching, research, and healthcare administration. AMIA was founded in 1990 through the merger of three existing health informatics associations. Since that time, AMIA has grown to more than 3,000 members from 42 countries worldwide. Together, these members represent all basic, applied, and clinical interests in healthcare information technology .
Previous to describing results to AMIA, the results were shared with clinicians at a medical staff meeting at VAPSHCS. Work continues on measuring the number of physical exams copied. This focus is related to the belief that a copied physical exam presents greater risk than other copy events. HIM and other stakeholders would like a flexible, easy-to-use tool to run on electronic documentation as part of on-going monitoring. When such a tool is developed and tested, more focused review on the impacts on coding and reimbursement can be accomplished. As this monitoring technology improves, we hope documentation practices improve.
Recommendations for Better Documentation
To improve the readability and integrity of documentation, clinicians need to be taught the appropriate information to include in documentation, what information may be copied and may not be copied, and how to attribute copied information to the original author. Clear reference to another part of the clinical record would obviate the need for much of the copying as would embedded "hot links" that allow one to jump to the source document. Clear identification of copied material will be done in different ways depending on the electronic medical record system being used. It may be possible to mark copied text by highlighting, italicizing , or placing copied text in quotation marks. Vividly acknowledging copied text may help to decrease the practice.
Summary
Copy/paste functionality in electronic health record systems can have both positive and negative effects. VAPSHCS recognized this as an opportunity for a quality improvement effort and developed a way to quantify and identify risk in copy/paste events. Findings and general recommendations were presented to clinical staff; however, the initial tool used to quantify copy/paste events is not practical for HIM departments to use as an on-going monitoring tool. Development continues on creating a practical tool that can be used to provide feedback to clinical staff.
Acknowledgements
Veterans Integrated Service Network (VISN) 20 VA Kenneth Kizer Quality Improvement Award 2002
References
Hammond KW, Helbig ST, Benson CC, Brathwaite-Sketo BM. "Are electronic medical records trustworthy? Observations on copying, pasting and duplication," Proceedings, 2003 AMIA Symposium : pp. 269-273.
Weir CR, Hurdle JF, Felgar MA et al. "Direct text entry in electronic progress notes: an evaluation of input errors," Methods Inf Med 2003;42:61-67.
Bloomfield, L. Plagiarism Resource Center, http://plagiarism.phys.virginia.edu/
| Source: 2004 IFHRO Congress & AHIMA Convention Proceedings, October 2004 |