December 2018, Volume XXXII, No 9

Electronic Health Records

Leveraging your information resources

Put your data to work

ver the past decade, the transition to electronic health records (EHRs) has increased the amount of health data at our fingertips. At the same time, the pressure to improve access to care and patient outcomes is steadily increasing. Fortunately, advances in technology now allow providers to more easily leverage the data in the EHR to make improvements in care, recognize trends in their patient population, improve clinical processes, and publish and disseminate research.

Leveraging existing EHR data

EHRs contain a wide range of information about patients and their visits: demographic characteristics, diagnoses, vital signs, lab values, medications, patient-reported outcomes, dates and departments of appointments, billing codes, and more. The more complete and accurate the data being entered into the EHR, the more this wealth of information can be used to enhance the following:

Reimbursement. As Medicare and private insurers move to value-based reimbursement, providers will need to submit not just the appropriate billing codes, but documentation of improvement in patient outcomes.

Individual patient outcomes. During each visit, the patient’s EHR is pulled up for documentation of the visit. Having the chart up provides access to the patient’s complete medical history, making it easier to see trends in vital signs and symptoms over time and helping to diagnose diseases. EHRs can also alert providers when a new medication might be contraindicated—preventing a possible adverse event—and when the patient is due for preventive screenings.

Clinic-wide patient outcomes. With consistent and complete information being captured in the EHR for individual patients, providers can look at outcomes and trends among all patients who meet certain criteria. For example, providers can find out which chronic conditions are most prevalent in their clinic; what percentage of their patients with diabetes are meeting blood sugar goals; and what percentage of patients with hypertension are prescribed anti-hypertensive medications. With this information, providers can update clinic processes, refine standards of care, and target their continuing education to better treat their patients.

Research studies and quality improvement projects. Data from the EHR can also be used to support scholarships for research studies, quality improvement projects, and other initiatives. For example, if you are conducting a study on a rare condition, it may be helpful to use EHR data to estimate how many patients with that condition attend your clinic. This could inform your project’s recruitment plan and feasibility. As described below, the EHR can also be the source of data for research projects.

Seeking outside help

If you’re interested in using EHR data, seek out assistance to obtain and analyze the data. In-house IT teams are a good place to start, but they may not have the statistical background needed to fully analyze and interpret the data. Contact your in-house research department, if you have one, to see what statistical resources are available. If in-house resources are not available, consider retaining an outside data analyst. When interviewing external analysts, make sure they are familiar with the unique strengths and challenges of health care and EHR data, as well as HIPAA requirements.

Data analysts and statisticians can advise you regarding the best study design to answer your research question, calculate sample sizes, decide which pieces of EHR data to collect and how to collect them, analyze statistics, interpret results, and prepare manuscripts for journal submission. In the cases of clinical trials, statisticians can help with randomizing patients to study arms and submissions for FDA approval. Statisticians and analysts can also help initiate quality improvement efforts, streamline clinic processes, consult on how to use EHR data for clinic improvements, identify best practices, analyze and interpret quality improvement data, and prepare presentations and posters for medical conferences and key stakeholders.

Beyond the EHR

collect more information to support all of the goals we have identified. There are many data collection methods, including EHR chart abstraction (“chart reviews”), surveys (via phone, paper, or web), key informant interviews, and focus groups. Select the best method based on your research questions and the data required to answer them. For example, chart reviews are well suited for understanding patient demographics, co-morbid conditions, health care utilization, health outcomes, and tracking trends in these metrics over time. However, chart reviews may not be able to explain why trends are occurring, provide patient satisfaction or patient-reported outcomes, or describe the patient or staff experience. To obtain these types of qualitative data, a survey or interviews would be more appropriate.

General data collection tips

Regardless of the data collection method you choose, it is important to gather accurate and appropriate data. Haphazard data collection methods could lead to inaccurate or indefensible findings, wasted resources, and potential harm to patients. Some general tips:

Patient data … can lead to meaningful improvements in patients’ experiences and health.

Haphazard data collection methods could lead to inaccurate or indefensible findings.

Have a plan for your data. Collect the data you need to answer your questions and describe the characteristics of the patients. Brainstorm the data points you expect to use and then plan how you will use them in your analysis or final report/presentation. If a piece of data does not serve a clear purpose, consider not collecting it. You may wish to involve a statistician or data analyst.

Devote time to project planning. Dedicate time to plan the logistics of the project with the study team before you begin data collection. For chart reviews, decide who and how many people will collect the data, how they will record the data (paper forms that are entered into a database or data entered directly into a database), and where the data will be stored.

Standardize data collection procedures. Thoroughly document the process, procedures, and decisions made by staff during data collection. If the data collection is complex, consider creating a training guide for staff. This will help ensure that all staff follow the same procedures, provide a resource for new team members, and serve as a starting point for the write-up of your methods in a presentation or publication.

Protect patient data. Train your study team on your organization’s policies for storing and transferring sensitive data. Be conscious of potential HIPAA violations when collecting data. You may need to get approval (or a waiver) from your organization’s Institutional Review Board (IRB) before you start collecting data. This will help determine if your patients need to provide informed consent before you can use their data.

Chart review considerations

Before conducting an EHR chart review, develop clear data collection guidelines and project goals. When deciding which pieces of data to collect, include those that relate directly to your project aims as well as participant characteristics that will allow you to describe your sample—these may vary by field of study or project topic. Even if you are not going to look for subgroup differences in your project, these characteristics can help you understand if you could apply your findings to other clinics and patient populations. In addition, identify the time period you’ll be collecting from and which patients are eligible to be included.

To ensure consistency in data collection, create working definitions for each piece of data you want to collect, how to record it, and where in the record you will find that information. That way, if two people are doing the reviews, they are searching the EHR in the same way for the same information (e.g., searching only flowsheets, searching flowsheets first and then looking at specialty visit notes, etc.). They are also recording it consistently. Be sure to involve at least one provider with content-area expertise to help plan and to answer questions that arise during the chart review process. This expert will know the feasibility (and credibility) of the data in the EHR.

While abstracting the data from the EHR, you’ll likely encounter situations you didn’t think of beforehand. Keep detailed notes regarding the decisions you make and update the working definitions with these decisions.

If multiple people will collect data from the charts, have them work together for the first three patients following the working definitions; then independently review three additional cases and compare the data they found to verify that they used the same process to collect the data. If they did not find the same data for the second set of cases, they should discuss where they found the information, agree on where to look for that data going forward, and revise the definitions.

Project planning should also include setting up a database that will be used for data entry and analysis. Regardless of which system you use, the database should correspond to the data collection forms and working definitions and should be set up to protect patient data (e.g., password protected or de-identified). To the extent possible, data validation rules should be set up in the database to help prevent data entry errors (e.g., if all included patients were over the age of 65, the database should not allow an age of 40 to be entered). A statistician or analyst can help you develop the database, create the electronic data entry forms, and set up the validation logic.

If the data are being collected on paper forms and manually entered into the database, you may want to consider an additional step to verify the accuracy of the data entry. One way to do this is through a double data entry process, where each form is entered twice and then compared for discrepancies. Another, less resource-intensive way is to randomly check 10–20 percent of the forms for data entry errors and keep track of the number of errors found. If the number of errors is concerning, consider retraining of staff and double data entry.

Survey administration considerations

With survey administration, simple is often better. Try to choose survey collection methods that seem feasible and fit within the clinic workflow. Patients could fill out a survey in the waiting room or while they are waiting in the exam room, or they could be mailed/emailed/called. Depending on your institution and project, you may not be able to connect with patients via email and web. Make sure you review your organization’s specific policies. If you use an online survey, check with your IT department to make sure the survey site is HIPAA compliant.

To decrease survey respondent burden, ask only the “need to know” questions. It is difficult to write good questions, so first search for questions from reputable sources or validated questionnaires rather than creating your own. There are many free, validated questionnaires available, saving you the time and hassle of creating your own questions. If you create your own questions, become familiar with survey development best practices. Many organizations and universities have such guidelines, including the Harvard University Program on Survey Research and the American Association for Public Opinion Research’s “Best Practices for Survey Research.”


Planning, consistency, documentation, and data privacy are critical in all data collection. Whether you’re using the wealth of data in your EHR or asking patients to fill out a short survey before they leave the clinic, patient data collected purposefully doesn’t have to be difficult and can lead to meaningful improvements in patients’ experiences and health.

Sara Richter, MS, is a senior statistician at Professional Data Analysts, Inc. She has over 10 years of experience in health care research, including health services research, clinical trials, and quality improvement.

Samantha Carlson, MPH, is an associate analyst at Professional Data Analysts, Inc. She received her master’s degree in epidemiology with a minor in biostatistics, and has experience working in clinical trials and public health research, as well as quality improvement in primary care. 


PO Box 6674, Minneapolis, MN 55406

(612) 728-8600

follow us

© Minnesota Physician Publishing · All Rights Reserved. 2019


Samantha Carlson, MPH, is an associate analyst at Professional Data Analysts, Inc. She received her master’s degree in epidemiology with a minor in biostatistics, and has experience working in clinical trials and public health research, as well as quality improvement in primary care.