Health care providers have access to more data than ever, but the challenge is how to draw meaningful conclusions from this information to deliver better care.
According to Julian M. Goldman, MD, an anesthesiologist at Massachusetts General Hospital and Harvard Medical School, and medical director of Partners HealthCare Systems Biomedical Engineering, all in Boston, this means refining the electronic medical record (EMR) to limit spurious data collection and errors of omission while connecting individual medical devices into more integrated systems.
At the American Society of Anesthesiologists’ INSIGHTS + INNOVATIONS 2017 Conference, Dr. Goldman described a familiar scenario: A blood pressure cuff and pulse oximeter are placed on the same arm, and as the blood pressure cuff inflates, blood flow to the finger is momentarily impeded, generating an erroneously low oxygen saturation signal. When the cuff deflates, blood returns to the finger along with a pulse.
“As clinicians we see this all the time, and we just ignore it,” Dr. Goldman said. “But if we’re [using] automated data acquisition systems and we happen to pluck that data point at that moment in time, that’s what gets put into the EMR.”
Although this kind of spurious data entry happens frequently, Dr. Goldman said, problems like this are difficult to solve. A physician present at the time of the event can simply mark the data as erroneous, but it’s impractical to rely on manual intervention to ensure data quality. Moreover, while it would be fairly straightforward to program an algorithm to recognize this mistake, the design of current medical devices hinders this type of connectivity.
“The way the EMR is implemented and medical devices operate prevents providers from writing algorithms like that, which makes it difficult to acquire data for future use,” Dr. Goldman explained. “It’s a fundamental problem in the way that we’re currently implementing technology for patient care.”
In addition to recording false information, electronic record sampling may not capture transient events. An experiment conducted with a pulse oximetry simulator by Dr. Goldman and his colleagues demonstrated the effect of averaging time setting on measured value. When the simulator created a transient desaturation from 99% to 70% and back to 99%, only devices set to a two-second averaging time accurately captured the lowest oxygen saturation.
“On most pulse oximeters you can change the averaging time, but metadata is not stored as part of the data set on the EMR, so we cannot recreate this scene,” Dr. Goldman said. “This is the tree that fell in the forest that no one knows about. We don’t know whether desaturation is missed because of an averaging time issue or something else.”
Challenges With Consumer Technology
Newness of the latest technology does not ensure superior performance, either. “Just because the latest device is equipped with Bluetooth technology does not mean it’s going to deliver better data,” Dr. Goldman said. “In fact, we’re probably going to get worse data.”
A comparison of two blood pressure monitors available on Amazon illustrates this point. While the conventional cuff goes on the arm, the newer, “fancier” model is worn on the wrist and boasts several conveniences: It’s lighter and more comfortable, has longer lasting batteries and communicates with a smartphone via Bluetooth. It’s also twice the price. Data quality and accuracy, however, aren’t as impressive.
“The cuff that’s on the arm is naturally at the level of the heart, so no matter how you move your arm, measurements will typically be accurate,” Dr. Goldman explained. “If it’s on the wrist, however, you have to worry about the angle of the arm and the forearm to determine whether the cuff is at heart level. One of the tradeoffs with newer, consumer-friendly technology is a higher likelihood of use error.”
Moreover, while it may be possible to monitor everyone in the future, ubiquitous health sensing poses its own challenges. Data interpretation without “eyes” on the patient, for example, may be incomplete.
“If you have a sensor on a person and see that person is lying down, that person could be unconscious on the floor or comfortable in bed, but the sensor will indicate the exact same thing,” Dr. Goldman noted. “Clearly, we need more information about context and can’t look at data blindly.”
Advancing Interoperability Of Medical Devices
Finally, lack of access to complete, accurate data is an issue. The EMR does not include most data generated by devices, Dr. Goldman said, and currently there is no national clearinghouse for system safety problems. Nevertheless, several organizations, such as the American Society of Anesthesiologists, have created grassroots efforts to change the technology ecosystem by advancing the intercommunication and interoperability of electronic medical devices. According to a statement issued by the American Medical Association, “seamless intercommunication” among devices could lead to important advances in patient safety and patient care, and as Dr. Goldman reported, participation of the FDA has been a powerful incentive for medical device manufacturers to explore innovative medical technology solutions, especially those benefiting from data sharing between manufacturers.
As director and founder of the Medical Device “Plug-and-Play” Interoperability Program, Dr. Goldman also works collaboratively on procurement processes to make it easier to access data. A project called MD FIRE (Medical Device Free Interoperability Requirements for the Enterprise) is attempting to simplify purchasing specifications so that hospitals can order equipment with open interfaces that are better equipped to share data.