Can This Search Tool Make Doctors Love EHR?

Surgical Robots: Look Who's Coming To The OR

The sum of "Google, plus CliffsNotes," might be the formula for making electronic health records software more usable, particularly in large hospital networks that use multiple EHR systems.

That formula is QPID Health CEO Mike Doyle's shorthand for what his company does. It adds search and summarization technology as a layer on top of EHR software to provide more convenient access to patient data when needed most -- the time doctors are making clinical decisions. The EHR world today is like "the Internet 20 years ago when we had all this data but no Google," Doyle said in an interview. "EHRs have done a great job of capturing all this data, but not at making it particularly useful."

A few months ago, in a column called Why Doctors Hate EHR Software, I quoted a pediatrician named Dave Denton on his frustration with EHR software and particularly the "treasure hunt" he found himself going on to find which tab of which screen might contain clinically relevant information about any given patient. Denton sits on his hospital's IT advisory board and believes in the potential of health IT, even as he is dismayed by the reality of it. Although the theory of EHR software is about getting all the information about a patient in one place, finding that information again is harder than it ought to be, he complained.

QPID just might be the map that makes the hunt a lot easier.

[Want more on how some EHRs can be tweaked for easier use? Read Medication Cabinets 'Talk' To Cerner EHR. ] 

The EHR software designer's standard strategy for making information easier to retrieve is to add more structured database fields. But if there is anything doctors hate more than wasting time trying to find information in an EHR, it is wasting time checking boxes on a complex data-entry form.

Like Google search, QPID is designed to find information regardless of whether it is neatly tagged and classified or all stored in the same place by using contextual clues.

QPID, which stands for "queriable patient inference dossier" (but is pronounced "cupid," which you've got to love), was developed at Massachusetts General Hospital by Michael Zalis, an interventional radiologist, and Mitch Harris, the computer scientist who led development of the natural language search technology and medical ontology. When trying to find the clinical context for the images he was sent to read, Zalis found he spent too much time trying to dig relevant information out of the hospital's information systems. He approached Harris, thinking they ought to be able to find a better way.

Figure 1: "Formula" from a QPID marketing slide deck.

The software spread from Mass General to the other major hospitals in the Boston-based Partners HealthCare system. It has started to be adopted elsewhere. QPID Inc. was launched in 2013 as a Partners spin-off venture backed by Matrix Partners, Partners Innovation Fund, Massachusetts General Physicians Organization (MGPO), and Cardinal Partners. The company is now in the process of building a sales force, with a couple of customer implementations in progress at medical centers in Virginia and Connecticut, which Doyle said he is "not at liberty to name" just yet.

In one particularly advanced application of the technology at Massachusetts General, QPID sorts through both clinical and research data about a specific cardiac procedure and compares the patient's risk factors to best-practice recommendations. The software then gives the physician a simple green light, yellow light, or red light indicator on whether the procedure should be recommended for that patient, a recommendation the doctor has the power to override. If the decision is made to go ahead, the application generates the required consent forms, with the risk factor rating clearly spelled out to help the patient make his or her own best decision.

This workflow is sufficiently airtight, taking advantage of the principles of

Next Page

evidence-based medicine, that one Boston-area insurer has agreed to waive the requirement for preapproval of procedures where the process is followed.

Used as a search engine, QPID can pull all the records for a given patient that contain the word "cancer," for instance. If a clinician searches for "*cancer" using QPID, it will also find records that don't contain that exact word, but do relate to the concept of malignant tumors. In addition to EHR records, it can search lab results containing diagnostic indications of cancer.

Figure 2: The QPID dashboard.

That's the Google part. To be clear, QPID doesn't actually use Google technology. Nor does it use CliffsNotes technology, but the QPID dashboard is designed to help a doctor be a quick study of a patient's condition. In this mode, rather than searching, the clinician is presented with a screen that can be customized by specialty showing an overview of all the most relevant information about that patient.

Instead of CliffsNotes, you might think of it as a sort of Google News -- all the top headlines for the patient. Except, instead of being classified into business and sports news, they're broken down, for example, by cardiovascular and pulmonary conditions, and by common warning signs such as chest pain. To get a better idea of how this fits into a practice, you can look at a goofy holiday video the QPID staff produced about diagnosing Santa's ailments with the help of the tool.

Another way Mass General has put the technology to work is by proactively checking for risk factors before performing a procedure. For example, the hospital schedules about 100 colonoscopies per day. It was having to cancel 10 to 15 of those because of risk factors that were unknown before the patient walked in the door and the doctor asked some questions, Doyle said. There are a half dozen of these risk factors. If a patient is on blood thinner medication, the procedure won't be performed, because if a polyp is found, the surgeon wouldn't be able to snip it without the risk of the patient bleeding to death. Using QPID to do a comprehensive records search, a nurse can discover that the patient has one of these risk factors and either cancel the procedure or, in the case of blood thinners, tell the patient to discontinue taking the medication for a couple of weeks prior to the procedure.

Similarly, a patient who has a pacemaker or any other metallic medical implant should never be given an MRI -- a diagnostic test that uses powerful magnets -- and QPID helps make sure clinicians are aware of such complications prior to the day of a procedure. In this way, Mass General has dramatically reduced the number of appointments that have to be canceled, Doyle said.

"It's pretty obvious EHR software was not made for physicians to practice," said Sid Govindan, another former Mass General radiologist, now serving as QPID's director of clinical content.

Often, the information doctors need to find is in structured fields, and they can only find it if they know exactly where to look. Further, even where structured fields are designated, they might not be used consistently. A do not resuscitate order might be recorded by checking off DNR on a form in some cases. In others it might be mentioned in a free-form physician's note. QPID helps make sure the information will be found either way, he said.

"EHRs today are like the Web before Google," Govindan said.

Medical data breaches seem to show up on the 6 o'clock news almost every week. If you think it wouldn't happen to you -- or the financial impact will be minor -- think again. Download the Healthcare Data Breaches Cost More Than You Think report today. (Free registration required.)