AMDIS 2014: The problem with problem lists

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - Clinical Documentation IMAGE REUSE DEMO

The need to maintain accurate problem lists in the EMR is not only a requirement of the Joint Commission, but is also a requirement for Meaningful Use, and on more than one occasion has been the topic of discussion on the Association of Medical Directors of Information Systems listserv.

During a session at the recent AMDIS Physician-Computer Connection Symposium in Ojai, Calif., William Galanter, MS, MD, PhD, associate chief health information officer at the University of Illinois Hospital and Health Sciences System, talked about his institution’s efforts to automate problem list placement as a means to improve documentation, while “making life a little easier” at the same time.

His institution automated the problem list by using CPOE and computerized decision support to link new medication orders to corresponding diagnoses in the list. The way it worked was if the patient’s medical record already contained a diagnosis that was an indication for the medication, the system wouldn’t respond with an alert. However, if there wasn’t a corresponding diagnosis an alert would appear asking the physician if he would like to add a problem, and a list of problems would appear associated with the drug from which the physician could pick.

According to Galanter, the system works really well with “easy” meds, i.e., those that are commonly prescribed and have few indications. A study of how well the system performed found that 96 percent of alerts were appropriate, and that 76 percent of the time doctors placed the problem on the problem list with 95 percent accuracy. “We thought this worked quite well,” Galanter said. “Not perfect, but not bad.”

The system, however, wasn’t quite as accurate with more problematic medications—those that may have had many off-label uses or multiple indications.

Still, not only did the automated system promote problem list placement, said Galanter, but it also provided “an added bonus in that it intercepted wrong medication and wrong patient errors.”

During the same discussion, John Lee, MD, medical director, clinical informatics, Edward Hospital and Health Services in Napierville, Ill., observed that the problem with problem lists in the past had been that they were “sparsely populated,” and that while the implementation of EHRs has solved that problem, it has had inadvertent consequences.

For example, Lee said, the ease with which doctors can put information into problem lists can leave them long and confusing. “You have all of this information and data, and you have try to collate all this stuff and distill it down so that the end user can use it,” he said, adding that it may be easier than reading through a bunch of notes, like doctors used to do in the pre-EHR days, “but not much easier, and not particularly useful.”

A major issue with problem lists, he said, is that while it’s easy to get the information into the list, it’s difficult to get it out. “Physicians I work with find it easy to put problems on the list, but don’t go in and pull out the inactive ones,” he said. 

“What I would like to see, and haven’t seen yet, is a problem list that is dynamic, intelligent, able to automate and make things active and inactive,” Lee observed. “What we really want to try to get to is some kind of intelligent context so the problem list becomes more useful to the end user. And we’re not there yet.”