After spending a lot of my time recently visiting my mom in a hospital, I couldn’t help but think every time the nurse came by, with pills and a small plastic cup of water, about what physicians, nurses and patients must confront: what is the right medication, when is it too much, what isn’t enough? What if they make a mistake about those medications?

Yes, despite electronic medical records, which have enabled sweeping changes in healthcare over the last few years to improve coordination efficiency, and quality of care, there are still wide gaps in the information that exists and not used properly.

That’s one of the reasons some numbers, like those of medication errors, are not where we want them to be.

According to the FDA, medication errors increased from 16,689 in 2010, to over 93,930 in 2016 – showing a whopping 463 percent jump. Those errors contributed to at least 230,000 deaths annually, some studies show, making them responsible for being the third largest cause of fatalities.

As in any error, there are many variables and reasons for each of them. Ironically, the increasing errors come at a time when hospital officials say there have been many data significant improvements in medical data systems. They include “alerts” in electronic medical records when the computers identify something that procedurally may be wrong, such as the kind of – or amount of – medications being given to a patient.

The latest statistics on medication errors show that more safeguards and tools are needed to help physicians, nurses and other staff overcome human flaws in assessing these alerts, such as “alert fatigue” when too many red lights go off, and the staff believe they know better than what machines are telling them, as if the computers are crying wolf once too often, and the physicians or nurses just walk on by to the next patient.

Shobha Phansalkar, RPh, PhD, director of informatics and clinical innovation at Wolters Kluwer, has spent years evaluating these so-called clinical information systems, especially the ins-and-outs of how healthcare handles medications and how they are given to patients. Also an assistant professor in the division of general medicine and primary care at Brigham and Women’s Hospital and Harvard Medical School, Phansalkar has worked diligently on analytical tools and yes, has examined the human factor in medication errors. In extensive studies she has carried out with colleagues over the years, and in an interview with HealthDataBuzz earlier this year, Phansalkar says the push toward adopting electronic medical records have been key initiatives for hospital quality and safety, but they are deeply flawed and their potential benefits have not been reached, not by any stretch.

In fact, Phansalkar’s studies on medication errors, particularly related to alerts related to drug-interaction, have exposed significant weaknesses in healthcare electronic medical systems and the so-called alerts. One of the major problems, she says, is that there are unclear messages in these alerts, and physicians decide to ignore or “override” them because they are not presented in a clear manner, or give context to what the doctors are doing. Studies by Phansalkar and her colleagues show that as many as 49 to 96 percent of all alerts are ignored, depending on the setting. Or as she and her colleagues have written: Alerts have “lagged behind expectations.”

As such, Phansalkar and her colleagues have been working to develop alert systems that have tools to overcome these limitations, with the “human element” very much a key focus. Wolters Kluwer Clinical Drug Information technology has developed programs to revise the alerting “logic” in electronic medical records to improve patient safety, she says.

“Errors are continuing to take place in every aspect of healthcare,” she says. “And they are taking place especially at the point of care such as when the right medications are chosen or supposed to be chosen. One of the problems is that providers are inundated with these alerts and are overriding a large number of them.”

Electronic medical records are limited in their ability to evaluate drug interactions, which could have sharp negative impacts on patients. This is becoming of greater importance as patients take more prescription medications than ever and often physicians still don’t have access to the entire patient history – despite the electronic medical records, Phansalkar says.

“It’s an age where we can exchange information easily, but somehow healthcare has lagged behind,” she says. “We’ve become more aware of the problem and we have better mechanisms in place, but with a fragmented system, and patients seeking care from multiple numbers of providers, whether it’s by choice or driven by health insurance, providers don’t have access to the entire patient profile. The physician doesn’t have access to all the medications and relies on a piece of the pie and that results in errors.

When a physician gets these “alerts” there are some built-in problems that Phansalkar and others are trying to fix. The lack of context about the medications being prescribed, such as the dose-toxicity or potential dependency of the patient, Phansalkar says.  Also, alerts should be identified as to whether they should be a priority or not, or the impact to the patient when a doctor decides to override them, she adds.  In addition, physicians have to evaluate similarities in different medications to ensure there is a “right bar code,” for instance, that is coordinated with pharmacies, according to Phansalkar.

Another problem in how these alerts are put together, she and colleagues say in a study. That relates to how “designers and vendors sharply limit the ability to modify alert systems because they fear being exposed to liability,” they wrote. That can be overcome through “finely tailored or parsimonious warnings that could ease alert fatigue without imparting a high risk of litigation for vendors, purchasers, and users,” the authors added in the study.  In addition, they said, more government regulation and development of international practice guidelines should highlight the warnings.

As Phansalkar wrote in Health System Management last year, “Alert fatigue remains one of the greatest hindrances to optimal use of clinical decision support at the point of care. Finding the right balance of the volume of alerts to display has been an elusive pursuit for EMR and CDS vendors.”

Phansalkar and her colleagues are working to change the alert system through various new models to improve patient care.

“Warnings could be tailored to a particular clinical environment, taking into account the individual care setting, such as adjusting the type or state of a medication interaction alert if a particular community of physicians is found to respond inappropriately to it,” they said in one study. “A tailored system might advise adjusting a patients medication dosage only when other patient data, such as age, or specific comorbidities, such as a rental function, raise concern, as opposed to alerting indiscriminately for all patients.”

Some healthcare organizations have taken steps to do just that – tinkering with their alert systems, such as the Group Health Cooperative of South Central Wisconsin, which used “filtering mechanisms” to address potential alert fatigue. Using guidelines developed by Wolters Kluwer, the healthcare system reduced the number of alerts from 87 percent to 27 percent, Phansalkar wrote.

MetroHealth in Ohio reduced and streamlined drug dose alerts by some 80 percent to avoid physician fatigue, she added.

Phansalkar is examining the impact of medication errors through and beyond the hospital stay or at a physician’s office.

“We’re are studying the downstream impact, what happens when a patient leaves the hospital, or the provider. Are they being counseled appropriately about medications or drug interactions?” she says. “There is a big opportunity to counsel these patients at that time.”

Indeed, health officials must ensure proper medication adherence from the beginning of a patient’s journey through the healthcare system, from the moment at “the gateway of patient interaction,” Phansalkar says . – Joe Cantlupe.