How much data a health system integrates into a new EHR will depend on the records needed in each department to support ongoing patient care. For example, an outpatient clinic may choose to convert only the most recent imaging scans with normal results, as well as a more comprehensive set of abnormal scans. Similarly, while a hospital may need to save detailed patient histories, converting progress notes, operative reports, or completed lab orders may serve little benefit for patients.¹

The amount of data converted may also depend on an organization’s size and capabilities. Large datasets may require electronic transfer, whereas much smaller datasets can be converted manually.

The challenge

A public safety net healthcare system in New York City served more than one million people through 11 hospitals, five acute care facilities, a home care agency, more than 70 neighborhood centers, and a correctional health services unit. The disparate EHR systems that were serving this sprawling system had led to a lack of standardization and fragmented care. Providers had to share patient information across multiple EHR systems, leading to wasted time, a high administrative burden, and the potential for data errors and leaks.

The solutions

The healthcare system:

• Appointed an interim executive leadership team that was fully responsible for both setting project strategy and ensuring project success.
• Deployed a team of physicians, nurses, pharmacists, and technical experts to support specific needs — including ensuring acceptable ED throughput, physician education, and pharmacy needs — at each stage of the implementation process.
• Created a steering committee and governance structure to identify issues and mitigate risks.

Results

Manual conversion involves extracting data from the legacy EHR and entering it into the new EHR by hand, whereas electronic conversion uses mediating software (often called an “interface engine”) to transfer large amounts of data automatically. Because of the complexity of healthcare data, EHR conversion processes are almost always a hybrid of electronic and manual approaches. For example, lab test codes need manual clinical review during data conversion, because test codes are not standardized among hospitals. Appointment records are another common example. Because appointment data is entangled with other patient data — such as Medicare or Medicaid eligibility — it is too complex to be transferred automatically.

How heavily an organization relies on either approach will depend — among many other things — on the amount of data to be converted, organizational budget, and capacity.¹

• Manual conversion doesn’t need an interface engine to transfer data between EHRs. Health systems can also more easily choose which aspects of each record to convert. That said, manual transfer requires a lot of time and personnel. Although using manual transfer avoids extensive testing after the data is transferred, health systems must still account for human error. All records need to be verified by a second person during the transfer process. Because the cost of manual transfer is based on personnel capacity and calculated per provider, larger health systems face a much higher cost.¹
• Electronic conversion requires a different set of skills and capabilities. After transfer and before the system goes live, the new EHR needs extensive testing to ensure that the data is complete and accurate. Because electronic conversion is expensive, it may not be a practical option for smaller practices with fewer records to convert.¹ In some cases, AI support tools may help speed up the data conversion process and prevent inaccurate or missing data.

The challenge

When a large New Jersey medical group merged with a New York-based provider organization, they decided to sunset multiple legacy EHRs and integrate with a single Epic instance used nationwide. To do this, the New Jersey medical group had to redesign certain workflows to align with a single Epic service area. On top of that, the medical group was working against the clock and would have to pay additional expenses if they didn’t transition from their legacy systems by the end of 2023.

The solution

The medical group partnered with Optum Advisory to access a cost-effective resource model and develop dynamic staffing, instead of creating a new Epic team from scratch.³ As a result, the group was able to integrate with a modernized Epic instance on time and under budget.

Results

If not planned and implemented carefully, EHR data conversion can lead to lost or corrupted data, increased patient wait times, disrupted provider recommendations and workflows — and can even threaten patient safety.^1,4 Every health system will encounter different risks and obstacles, so identify the challenges and solutions that are specific to your organization for a smooth conversion process.⁵

AI tools are increasingly available to accelerate data conversion and improve quality by automating parts of the data conversion process. For example, AI tools can map fields in both the legacy and destination EHRs, analyze their similarities and differences, and remove conflicts between data models.

And, because AI allows health systems to review and analyze large amounts of data, generative AI can help refine the data conversion process itself. By training AI on historical data conversions, health systems can leverage past processes to create new conversion models — and continue to learn and refine those models for future data conversions.

Using automated support can ensure a smooth transition and uninterrupted care delivery by:

• Shortening the conversion timeline and reducing workflow disruptions
• Reducing inaccurate, missing, or corrupted data
• Alleviating the administrative burden of data conversion for staff
• Lowering costs

As innovative as generative AI can be for EHR migration, it’s important to ground the data conversion process with human support and oversight. That could mean including IT teams to manage data conversion, having a physician-led team available to support the process, or deploying EHR-trained medical coders to validate data.    

The process of EHR data conversion is still challenging, but careful planning and setting up a framework of technological and human support can create a positive experience for health systems.