One of the advantages of using an Electronic Health Record (EHR) is its ability to simplify research activities to the point that even small medical practices can do their own research, either to support quality improvement activities within their practice or purely for research purposes. Determining what type of research you are interested in and how you want to use the results has implications in terms of needed functionality for your EHR, and research policies that your practice needs to maintain. If you are interested in performing research using your EHR system, you should make sure that it can record patient consent as a structured data field. In addition, the EHR should have flexible and easy-to-use reporting features. It should allow you to set up prompts that can alert you to patients that may be eligible to participate in the studies you are undertaking.
The first thing you should decide is how you want to use the results of the research. If you are just looking to use the results for quality improvement purposes that are limited to your practice, you are generally not required by regulation/law to undertake an IRB approval process and the requirements that related to patient consent are more flexible. However, if you are looking to publish the results and/or are conducting prospective research you will need to go through more rigorous processes. Many professional societies encourage their members to follow the more rigorous standards whenever they are undertaking research activities.
The second thing you should determine is whether you intend to do a retrospective study (in which you will be using data that has already been collected) or a prospective study (collecting information going forward). Most randomized clinical studies are prospective in nature, although they may use retrospective data as part of the inclusion and exclusion criteria.
When dealing with research data, one of the best ways to ensure protection of patient confidentiality is through the analysis and use of de-identified or blinded data.* This also has the benefit of simplifying the IRB process.
Regardless of the type of research you are interested in, it is critical to ensure that you are entering the data as structured elements in the EHR system. I have worked with a number of practices that had been diligent in typing in all of the details of the encounter and relevant test results, but, as they were entering it in as free text, were unable to use the data to run reports without a manual extraction process. While some systems are starting to use natural language processing techniques to search through free text entries, this data can be less accurate and is difficult to combine with standardized data fields limiting analysis. For more guidance on using standardized data fields, read “Why Templates Are Essential When Using an EHR” and “Data Entry Strategies — Using Templates.”
Another critical component within a practice is to ensure that the data is recorded using the same data fields so that diagnoses (e.g. Type 2 Diabetes) are recorded consistently. This tends to be more of an issue when practices or providers have customized their EHR in order to create new or alternate fields that are incorrectly mapped to an equivalent field within the system. This can make it difficult to correctly identify data for extraction when running reports, especially when relying on the reporting capabilities found in most EHR systems in which relevant fields have been pre-mapped for a particular condition.
When evaluating an EHR system from a research perspective, you should determine how difficult it is to combine or separate data elements that you would want to include in aggregated data reports. You should also determine the system’s ability to produce reports that combine different data types — for instance a clinical diagnosis, plus a prescribed medication, plus a particular lab result. If you are interested in the process measures, you will also want to ensure that the system can incorporate demographic and payer information in the reports.
In conclusion, if you are interested in using your EHR for research you should plan ahead, both when selecting an EHR system (as they differ substantially in terms of how they handle data reporting and research requirements) and when implementing and using your EHR. Entering data into your EHR in a standardized and consistent manner will be critical to successfully using your EHR for research purposes.
* See the following for more information on using de-indentified data:
Standards for privacy of individually identifiable health information final rule. 67. Federal Register. 2002:53181–53273; Malin B, Benitez K, Masys D.
Never too old for anonymity: a statistical standard for demographic data sharing via the HIPAA Privacy Rule. J AM Med Inform Assoc 2011;18:3–10.