EMR data integration: Still a long way to go
The advent of Electronic Medical Records (EMR) has been a boon to the operational and analytical abilities of healthcare analytics firms and medical providers, allowing them to more accurately document, track and create the analytics required to better care for patients. Likewise, patients have benefited from the ability to track and monitor their care needs across time, allowing proactive actions to take place when identified and recommended by their healthcare providers. One of the core goals of an EMR-based medical practice is to improve the overall quality of care over time across their particular patient population. This is welcome news all around, but it ultimately entirely depends on the effective and efficient creation, organization, normalization and distribution of medical practice and patient-related data. And that’s where the trouble begins.
For all its benefits—realized and potential—the process of aggregating and consuming EMR (and its offspring the Electronic Health Records or EHR) information is still problematic at best. To date, there is no single technical standard for creating, formatting and distributing EMR data between healthcare analytics firms and EMR vendors. The net result of that is the sub-optimization of the accuracy and usefulness of EMR data. That’s not only an operational issue for healthcare analytics firms but also a serious quality of care issue for patients.
One of the benefits of EMR data is that it can be converted into insightful and actionable analytic data, both on an individual patient level and on a patient population level. However, the current lack of standards and therefore consistency amongst EMR vendors and healthcare analytics firms continues to be a challenge to producing high-quality analytic information. That means that for most healthcare analytics firms the conversion of EMR data to analytical insights is still a cumbersome one.
Most healthcare analytics firms consume EMR data from multiple vendors in this space. The EMR vendors typically extract and aggregate data from various available sources that they then send to healthcare analytics firms that have contracted with to receive the data on a regular or ad hoc basis. However, the lack of standards by which the EMR vendors create, format and transmit that data makes it very difficult for healthcare analytics firms to consume it efficiently.
For example, things like Continuity of Care Documents (CCD)—a potential valuable data element—are often sent to healthcare analytics firms with missing or incomplete data, or with “standard” data fields in one place from one EMR vendor, and in another place from a different EMR vendor. Also, the data is not usually well edited or normalized, so initial input errors are seldom corrected before the data is redistributed for consumption. This process is further complicated by the fact that EMR vendors each use different transmission methods. Some use a web service approach, others use a file transfer protocol hybrid, while others create proprietary transmission channels. This all adds up to burdensome technical complexity and overhead for any healthcare analytics firm trying to consume the data and turn it into useful analytical insights.
The Clinical Document Architecture (CDA) is one of the ways the industry has tried to achieve standardized data, but to date these efforts have fallen short. For healthcare analytics firms needing to consume this data, that leaves few options. Most use consultants or other technical expertise to create front-end data consumption processes that translate and configure EMR data into something more useful to their data purposes. And there are ways to do that, even as it remains incumbent on healthcare analytics firms and other consumers of the data to do so. From a variety of lessons learned in this area, a few steps stand out as critical.
- First, with EMR vendors providing data in a number of formats, the practice of data profiling is essential. More often than not there are too many assumptions made on the healthcare analytics firm side about what type of data is being sent and how that data will be formatted. Is it test or live data? Will data be sent via the same channel and in the same format as the last time? Has the data been put into a format that can be used for integration on the healthcare analytics firm end? A good practice is to create a pre-integration process that requires the EMR vendor to send live data samples before any regular transmissions commence. This will go a long way toward ensuring that all of the necessary data elements are included and that the data will be ready for integration when it is received.
- Second, once the above is in place, an automated routine should be created that addresses any potential integration issues before they reach the data store by categorizing incoming EMR data along a data quality rubric. The quality check should result in the creation of a data feedback report that can be used by healthcare analytics firms and vendors alike to improve their data input and formatting accuracy.
- Third, any integration process should be built in such a way that it’s flexible enough to handle similar—but not the same—data types from different EMR vendors. For example, CCDs from different EMR vendors can be formatted completely different, even though there is a standard for CCDs. That means it’s critical for healthcare analytics firms to build their integration platforms with as much flexibility as possible to handle the same but different “standard format” of CCDs.
Taken as a whole, these and other similar steps—including the standardization of EMR data distribution channels by EMR vendors—will improve the quality and usability of EMR data for healthcare analytics firms. However, these steps are also an acknowledgment that the data quality problems associated with EMR data are likely to persist into the foreseeable future.
At present, there are not any overriding market, regulatory or operational forces strong enough to motivate EMR vendors, healthcare analytics firms, insurers or any other stakeholders to invest the time and resources necessary to improve things dramatically. Although, a recent effort to improve the accuracy and the quality of EMR data as part of a process to score medical providers’ eligibility for incentives and benefits for the delivery of proactive patient care holds some promise, but it’ll likely be a long time in coming.
In the interim, it remains incumbent on EMR vendors and healthcare analytics firms to put the necessary technologies and associated processes in place to improve the overall quality, integrity and integration of EMR data into their operational analytics.
Tags: EMR data integration