Retention of Records when Transitioning to an EMR

Over the past week I participated in a conference on practice efficiency and technology hosted by the Canadian Medical Association. As one of the presenters, I had an opportunity to talk with a number of physicians throughout the week. One of the most interesting discussions focused on the retention of paper and electronic medical records as required by law in Canada.

There is some variance in different provinces in terms of how long medical records need to be retained. It is my understanding that the general rule of thumb is 7 years from the date of last entry in a medical record in a private physician office or 7 years from the age of majority (which is currently 19 years of age). The British Columbia College of Physicians and Surgeons Policy on Retention of Records can be read by clicking on this link. In practical terms, this means that a physician with a newborn patient in his/her practice has to retain that record for 26 years.

There are a number of situations that are worthy of further discussion. I encourage physician readers of this entry to add your thoughts by clicking on the 'Comments' link at the bottom of this posting. These situations include the following:

• When a physician converts from a paper practice to an EMR, a core data set is usually entered into the EMR. This includes demographic information, long term medications, allergies and alerts, relevant clinical information and relevant diagnostic imaging, pathology and specialist reports. While some physicians may enter more information than others, the paper chart is generally not entered into the EMR in its entirety. During early visits, the physician may use both the EMR and paper record during encounters. Within 3 to 4 visits, the paper record is usually no longer required and is ultimately is placed into storage. For healthy young patients (children), very little information may be entered from the paper record into the EMR. However, this group is the one that requires the longest retention of the patient record by the physician. Even though the physician may be using an EMR, there is still a legal requirement to store the paper patient record for anywhere up to 26 years. I am not sure how one could avoid this situation without some mechanism to archive the entire patient record. Chart storage adds up to $$$ in the longer term. Thoughts?? How have physicians using EMR dealt with this issue?
• The second situation is one in which a physician who is using an EMR makes the decision to move to a new system. If the information in the first EMR is stored in a proprietary format (a format that is not standards based), it may not be possible to extract all the information from the EMR. In this situation, the physician may be faced with the need to maintain a copy of the original EMR at the same time as running the second EMR. This could be further complicated by the fact that the original EMR may run on an older operating system (e.g. an old version of Windows) and the fact that hardware needs to be replaced after a certain time cycle. The purpose of this second parallel system is simply to be available in the event that data is required for medico-legal or other clinical purposes based on the legal requirements in that province at that time. An alternative may be to print all of the patient records and store a paper copy, however this could create significant difficulty in finding the information month or years later. Any thoughts on how this situation could be managed?
• The last situation that we discussed was that of scanning consultant and diagnostic reports into the EMR. It is the practice in some EMR systems to scan consultant reports using Optical Character Recognition (OCR) software and then copy and paste the scanned text directly into the EMR. The benefit of this approach is the ability to search the text at the later time as it is in digital format. However there are also disadvantages in that the accuracy of the scanning needs to be checked (e.g. a Hyper being read as a hypo if certain fonts are used to print the document). In addition, if the source of the original document is then referenced in the EMR and the original document is then shredded, it is no longer possible to produce an exact copy of the original report, only a scanned version (which may or may not represent the entire specialist or diagnostic report). In addition, this is further complicated by the fact that the majority of specialists tend to maintain episodic relationships with patients. A patient is referred for a specific reason and may or may not follow up at a later stage with that specialist. If the specialist policy is to retain medical records for 7 years (in those over 19 years of age) and then the record is destroyed, it may not be possible to reproduce an original version of that specialist report beyond the 7 year time frame. This could have potential medico-legal ramifications as there would be no way to reproduce the exact report at a later time, as the original may no longer exist. Perhaps the only way to deal with this is to have a standard policy requiring that physicians retain an image of that document (e.g. as a .pdf file) and if there is a desire to have some of the information searchable in the EMR using OCR software, then an additional OCR copy can be placed directly into the EMR for reference or search purposes. This is a more robust answer, but the cost implications are significant as greater sophistication would be required in terms of scanning technology and it would take twice as long (or longer) to scan the reports directly into the EMR. The ideal end-state would be one in which only digital reports are inserted directly into the EMR, thus doing away with the need for scanning... however this is still some time in the future.

These are complex issues, but ones that have significant implication as we move to the EMR in physician practices.

The readers of this Blog would be most interested to learn the thoughts and experiences of others. Please add your comments by clicking on the 'Comments' link below.

Comments

[Allan Horii]

This is an issue which will become much more apparent here in BC in the near future. With the recent BCMA/provincial government agreement to fund EMRs, a decision will be forthcoming regarding which specific vendors will be approved for funding. Ultimately, many physicians may be faced with migrating their data to a new EMR system in order to qualify for compensation. Likewise, offices making the step towards electronic medical records for the first time will need to decide how to deal with their paper records.

Personally, my paper charts still occupy their usual place on the office wall since my switch to EMR 2 years ago. They are still periodically pulled to look at an old report or document. I have wondered, on occasion, about the cost and utility of paying a professional service to scan all those files onto CDs or DVDs.

[Michelle Greiver]

I have had all my old charts scanned into pdf, and we shredded the paper record. The scanned charts are kept on an external hard drive, accessible within my office network. We keep a copy on a DVD, and I took a second DVD home, locked in a filing cabinet. I hired a student for the summer. Cost was about $160 for the hard drive, $150 for the external DVD reader, about 120 hours of labour for the student (he did about 10 charts per hour, and taught my staff how to do it, so they did some too). These are not outrageous costs. The cost of retention, even for children's charts, is now nil. We will be scanning all transferred charts, inactive charts, deceased charts over time, but the most important part was scanning the active charts, which is now done. There are no chart pulls for me at my office, and I have access to the old charts at home or office.

For new incoming reports, we scan all consultant/DI/other reports to the EMR in .pdf format. This ensures that we have an exact copy of the original, which is then shredded. However, if I need to have quick access to the report within the EMR, then I will OCR the document (in my case, print to MS document imaging, then OCR). As most of us know, there is very little within most reports that needs to be OCR'd. The majority of reports are not sent to OCR; instead, I'll put in a quick note in Comments. For example: CXR normal. All Comments are visible in the EMR without loading the scanned file, which gives me a nice summary. I'll only OCR if there are comments of significant length that need to be copied and pasted. I wish consultant and DI reports were written using consistent database fields; for example: vitals, diagnosis, recommendation. This would then make it more possible to import into EMR (if we had common standards). To reiterate: the majority of reports contain only two to three things that need to be visible (Dx, recommendation). It is better and much faster to scan to pdf, and summarize the 2 bits of info in the comments section. Only use OCR if there is significant/complex information that needs to be visible.

Michelle

[Adam Chen]

We also scan only .pdf images without OCR, as OCR searchable .pdf files take twice the memory as non-searchable .pdf files.

I had pulled boxes of charts 10 years old (Ontario Statute of Limitation) ready to be destroyed, only to find out now that CPSO recommends storing them for 15 years. Ughhhh!

I am contemplating scanning and shredding. However, I will probably scan only reports and keep my original handwritten notes and batch file them for compact storage (without thick folders). In case of legal proceedings, original handwritten notes can better demonstrate that there has not been record tampering.

[Stephen McLaren]

Is it time to rethink the importance of the medical record?
The discussion below re-addresses the dialogue about archiving paper / quality of data entry into EMR and transferring to a new EMR.
Let's be evidence based in our thoughts.

What is the true value of "aged" data?
Having been chartless 7 years with thousands of paper records offsite, there exists 1.5 linear feet of paper charts that were pulled to meet a request for information. Can I tell you that any of that data made a difference to current care? I cannot. How often do medical records departments need to pull charts dormant for 5, 10 15 years? What percentage of their day to day chart pulls are archived volumes? What impact on current and future care is derived from having data 10,15, 20 years old that is not summarized in the current record?
The record exists to serve the patient, the provider(s), the legal community and historians. Is there an example in the legal community where data is not kept forever? It is not in my knowledge base that all police records are archived and kept indefinitley.

The value of EMR (over paper based charts) to patient care is readily apparent. The quality of the record far surpasses any paper based record I have seen. Not only is it infintely more legible but our experience has been one of the record being far more inclusive of the thoughts going into patient care than existed in the handwritten chart. Aggregation and compliation of data allows trends to indicate care progress and intervention/outcome analysis.
The searchable record allows groups to cross-cover and provide after hours care from a more knowledgable place.
We scan and OCR every incoming paper consultants note other than their hand written notes. All OCR "errors" actually got traced back to "errors" in the original documents. The searchable EMR is a far more functional tool than the non-searchable. Having made strides to embrace a better quality of care the EMR docs should not be saddled with advice or direction necessitating them to run dual systems. (paper, pdf or jpeg copy?)

On moving to a new EMR I would hope that basic demographic data gets converted to the new system. Beyond that I would propose a simple soultion for migration of the "meat" of the chart. The EMR should be able to provide one document in chronological or categorical order that shows every entry into the EMR in descending sequence. This document should be exported in total as an uneditable pdf which can then be appended within a new EMR as a searchable attachment.(Vol.II)(VolI=paper record)
Does this ability exist? Yes it does for us.

As I read about some of the challenges of EMR I am often found thinking that we should be as stringent in our setting of standards and criteria for all our colleagues in the paper world.
The paper record is, after all, a medical record and is deserved of the same privacy, legibility, comprehensiveness and accuracy standards.
Having been less than diligent in enforcing high standards for paper based medical records why are we so determined to save them?

[Ray Simkus]

This a new comment on an old message thread related to an ongoing problem. As more and more physicians start using EMRs there will be occasions when physicians change from one EMR vendor to another. I have heard that there are issues related to keeping the old system available because the new system may not be a complete and accurate copy of the old data. The current status of data exchange between systems is rudimentary and idiosyncratic. The official MSVA specs in BC covers only the very basic demographics, the Alberta specs are more detailed but there are errors in some of the specs. Data transfer based on these specs will not produce a true and complete copy of the original EMR record. In the UK they have the GP2GP project that allows for the exchange of the entire patient file between systems.

Considering the ongoing changes in hardware, operating systems and office staff what are the chances of being able to read data 7 or 25 years down the road. Think of what computer systems you had in 1983. These retention requirements are not based on usefulness but are imposed on us by the medicolegal system. What happens to my 4 year old computer running Windows 98 stops working 6 years from now. What are the chances that I could find parts or get a new one running with the old data. How faithful is the PDF copy to what was actually seen on the screen when decisions were made? I suspect that what is 'printed' out to a PDF is really not the 'original' either.

What we need are EMRs that have a sophisticated enough data structure that can accommodate all our needs and that this structure is designed well enough to remain unchanged for a very long time. It is the partial measures that are the problem.

Ray Simkus