Icon

This entry is part 26 of 44 in the series Words

In Icons, Pedagogic Vectors, Forms Design and Posture we briefly discussed icon design. (Icons, in this context, meaning the sketch-pictures on buttons that you can click.) The bottom line was that it’s hard to learn and remember what icons stand for.Icon

In Performance, Data Pixels, Location, and Preattentive Attributes we discussed how icons should be recognizable by preattentive attributes, so there is no Cognitive Friction to overcome when selecting the right icon on which to click.

In Color, we discussed the role of color in icons, coming down to the idea that icons shouldn’t use color, and should be grayscale. Read the rest of this entry
»

Share

Signal-to-Noise Ratio

This entry is part 27 of 44 in the series Words
finding the numbers can be hard

finding the numbers can be hard

I work at the University of Pittsburgh Medical Center. UPMC has prioritized IT, and compared with many other academic medical centers, the IT department is fairly well-funded and well-staffed. The central IT umbrella spreads wide, including 16 major hospitals and numerous other facilities. UPMC uses Cerner for an inpatient electronic medical record (EMR) (and for outpatient settings). For clinical charting in the ED, we use Cerner’s PowerChart 2G, dictating into it using Dragon speech recognition. PowerChart is pretty klunky, as are its templates, and in our ED we use our own very simple PowerChart templates, basically a blank page into which to dictate.We in the ED built some standard templates and macros in Dragon, and docs, including residents, can customize or add new templates or speech macros as they wish, which speeds up dictation quite a bit.

However, we actively discourage the use of the PowerChart templates. Why? Because PowerChart templates have a seductive feature that the vendor and our IT people used to tout. But as it turns out, that feature trashes the signal-to-noise ratio of the chart.

Read the rest of this entry
»

Share

Anti-Data Pixels

This entry is part 29 of 44 in the series Words

Less is More
Mies van der Rohe

In high school English class, many of my generation were forced to study a book about writing known as “Strunk and White.” Compared to many other books we were forced to read, it had many advantages. It was short. It was to-the-point. It was full of pithy sayings, the most pithy: omit needless words.

In Cognitive Friction, we extended the idea to graphical computer user interfaces as “omit needless pixels.” In Performance, Data Pixels, Location, and Preattentive Attributes we looked at Nielsen and Tahir’s analysis of the percentage of a home page’s area devoted to different purposes; in this way, we could determine which were valid data pixels, which were not, and the ratio of data to non-data pixels. Read the rest of this entry
»

Share

iPhones

This entry is part 30 of 44 in the series Words

On May 3, Steve Stack, Chair of the American Medical Association (and an emergency physician from Lexington, KY) gave a presentation on electronic health records (EHRs) to the Centers for Medicare and Medicaid Services. The paper is worth a close read. He observes that physicians are technology early-adopters, but that there had to be Federal financial incentives for physicians and hospitals to adopt an EHR. Why? EHRs suck. (I rephrase only slightly.) He points out that EHRs are immature products. If we judge by human development, and want to use a derogatory term, we might call them retarded, in this case invoking the original meaning of the word retarded, as in slowed development, compared to their peers.Signal

Though an 18 month-old child can operate an iPhone, physicians with 7 to 10 years of post-collegiate education are brought to their knees by their EHRs.

In 2010, a quarter of physicians who would not recommend their EHR to others. Two years later,  over a third were “very dissatisfied” with their EHR and would not recommend it.

When an EHR is deployed in a doctor’s office or hospital, physician productivity predictably, consistently and markedly declines. Even after months of use, many physicians are unable to return to their pre-EHR level of productivity – there is a sustained negative impact resulting in the physician spending more time on clerical tasks related to the EHR and less time directly caring for patients. In a way, it ensures the physician practices at the bottom of his degree.

He gives examples of how a physician’s medical note used to be:

  • 24 y/o healthy male. Slipped on ice and landed on right hand. Closed, angulated distal radius fracture. No other injuries. Splint now and to O.R. in a.m. for ORIF.
  • 18 y/o healthy female. Fever and exudative pharyngitis for 2 days. Exam otherwise unremarkable. Strep test +. Rx. Amoxil

He goes on to talk about how malpractice litigation, billing and coding, and CMS and other insurance requirements for payment have bloated the medical record. And further, how EHR features such as templates, macros, and cut-and-paste have homogenized medical records while (with the difficulty of typing or dictation) decreased the visit-appropriate essential information.

Seems to me that over the past 30-40 years (yes I’m that old) the medical-chart signal-to-noise ratio has gone from 0.99 (99% of the chart is signal, that is, clinically useful information) to, at least for EHR inpatient progress notes and ED notes, to 0.1 (10% signal, and 90% noise).Noise

One of his three conclusions was

ONC [Office of the National Coordinator for Health IT] should immediately address EHR usability concerns raised by physicians and take prompt action to add usability criteria to the EHR certification process.

Bravo!

Share

Suicide

This entry is part 31 of 44 in the series Words

Text MiningData mining has been a topic of interest to businesses and researchers for many decades. For physicians and other clinicians, and those designing systems for clinicians, data mining has been of less interest. Yes, you can use data mining to predict the volume of patients in your ED by day and hour. Yes, you can use data mining to order supplies more intelligently. But to improve patient care? Not so much.

Yes, research using data mining can provide us with some clinical information,  but such retrospective studies, especially subgroup analysis, can lead to egregious error, as summarized in a recent article in JAMA. It’s not a replacement for prospective, blinded and randomized studies. Read the rest of this entry
»

Share