5 reasons data inaccuracies occur in EMRs
Studies have shown in recent years that the quality of data in many electronic medical records is often not very good. According to Peter Witonsky, president and chief sales officer at iSirona, this is largely due to simple inaccuracies that occur more often than we think.
"A lot of these fall into the same category, in my mind, but it's different ways of getting to that category," said Witonsky. "That latency of data is terrible. We have customers, prior to us, with eight to 10 hours in latency of data, and that's not uncommon. It's not the end of the world, but there are tons and tons of examples of what latency of data will do to decision making on the other side."
Witonsky highlights five reasons why data inaccuracies occur in EMRs.
1. Simple miskeying. Although it may be easy and "quite common," said Witonsky, the main way data inaccuracies tend to occur is because of simple miskeying. "If you look at any nurse of any floor, there's about 1,000 or over 1,000 data elements a shift that person is responsible for," he said. "So if you're an ICU nurse, and you're taking vitals and other critical information every 15 minutes, or if you're a low acuity nurse and you have four patients to be responsible for, it seems to average out just north of 1,000 data elements." And to expect a nurse to key in those elements with 100 percent accuracy isn't a realistic goal, Witonsky said. "The idea any person [can do that] is ludicrous," he said.
[See also: EMR implementation requires right planning.]
2. Miscommunication from the patient. Bad information or miscommunication from the patient is another all-too-common way these inaccuracies can occur, said Witonsky. And this can include the patient not telling which drugs they're on, not knowing the name of the drug or the dosage or even the patient lying about his or her weight. "So it's sort of a garbage in, garbage out theory," said Witonsky. "If you don't tell me that you're allergic [to a drug] and I give you Penicillin and it's a bad result, again, that's bad data in the EMR." It's for that reason, he pointed out, that most of today's EMRs have allergies highlighted at the top of every patient screen.
3. Wrong entry or lack of entering device data. Looking back to simple miskeying, said Witonsky, 1,000 data elements, over time, is "an awful lot of work," he said. "So you have something called smoothing, [which is] a long practice for smoothing data where a nurse of physician is expecting to see normal [results], and they put in normal regardless of what the device is telling you." These generic readings tend to bring out inconsistencies in data, he continued, which wouldn't occur if the person inputting data took the actual information from the device. "That's not intended to be a knock," he said. "That's intended to say, in performing the hardest job on the planet, if they knew [a patient] was healthy, they leave all the vitals on the machine and may choose to put [the patient] in as a normal patient, as opposed to the exact answers."