A recent study demonstrated a high rate of copy and paste in electronic medical records, which highlights the flawed design of our current electronic medical record (EMR) systems in healthcare. The introduction of EMRs into healthcare was supposed to lead to better patient care and outcomes – they have not, but why?
Major design flaws. Current EMRs are designed into the fragments of care based on our fragmented care delivery system. The EMRs are also designed as a one-size-fits-all system for all different types of patients and disease processes.
Too much data. Because of the overemphasis on documentation for billing, we collect too much data for each patient encounter. The vast majority of data has little or no impact on patient outcomes. Since there is so much data that is unimportant (noise), it is difficult to identify what data matters, the data that actually impacts patient outcomes.
Data not in context. Much of the EMR data is aggregated into large data repositories and analysis is performed from this centralized data without attention to the many different care processes contained in the data repositories.
Inaccurate data. This is highlighted in the article where the authors found that about half of the documentation is just copied and pasted. The bottom line is that the current method for healthcare data documentation results in very poor-quality data. Any company utilizing data from healthcare documentation, especially administrative/coding and billing data, will suffer from the well-known data problem of garbage in, garbage out.
CQInsights is not the first group to point out these significant flaws in healthcare data documentation and the design of EMRs. After working with engineers and data scientists for over a decade, we’ve learned that there are solutions to these flaws:
Complete cycle of care design. Data management tools should be designed around the most critical process in healthcare: the patient’s entire cycle of care. The data (information) should travel with the patient through their whole cycle of care and be available to any caregiver, including the patient, at any time.
Only pertinent data. Document what matters and leave out the garbage. Do we really need to document the vital signs or reflexes of a person who looks fine and is just interested in having an elective hernia repair?
In context of patient care process. The data that matters should be based on the context of each whole, definable patient care process. For example, measuring the value of care (which should be the most important outcome to measure, yet no one we know of in healthcare measures it) will differ for each type of care process. The value of an elective hernia repair will be measured differently than the value of a breast cancer process. There is no way to measure data the same for different care processes. This is why it’s impossible to have a data dictionary for a data lake or data warehouse containing data from different care processes.
Constant improvement of data. Data curation with feedback loops is critical to have accurate data. The key is to keep data decentralized in the context of each patient care process and in each local clinical environment. There is no other way to ensure high-quality, accurate data. There are no shortcuts.