Stepwise Success : Diagnostics

Rules and Algorithms

Scott Warner — Fri, 06 Nov 2009 11:26:00 GMT

Just the other day a question of how to alert physicians of laboratory reflex rules came up. After all, everyone's rules are slightly different. Examples: if the dipstick is positive for blood or leukocyte esterase, perform a microscopic examination of...(read more)

Teachable Moments

Scott Warner — Wed, 14 Oct 2009 10:25:00 GMT

Much was made of the President's "beer summit" last July, a meeting between a black professor and the white policeman who arrested him, something the Wall Street Journal labeled a "teachable moment." Aside from wondering what kind of beer goes best with...(read more)

Calculations

Scott Warner — Tue, 08 Sep 2009 10:17:00 GMT

Most of us aren't mathematicians. Well, I'm not. I've known people who are just better at "seeing" math than myself. I've always struggled to add a column of figures in my head, forget solving Fermat's Last Theorem . Which has been solved already. Just...(read more)

Package Inserts

Scott Warner — Wed, 02 Sep 2009 10:14:00 GMT

In writing a procedure, I strive to make it the reference on the bench. Yet no matter how good a written procedure is, the package insert is always invaluable. There are two reasons for this. One, the package insert is the main reference for any procedure...(read more)

A New Marker

Scott Warner — Wed, 08 Apr 2009 08:26:00 GMT

When I spoke with the diaDexus rep the other day on the telephone, he told me the PLAC test , a new test that measures lipoprotein-associated phospholipase A 2 (Lp-PLA 2 ), is being done in forty labs across the US, with another added every two weeks....(read more)

I Got My Answer From the Nurse

Scott Warner — Wed, 28 Jan 2009 20:02:00 GMT

A hospital receptionist concerned about a lab report on her son came into the lab asking what a "bicarb" level meant. I printed a page from Lab Tests Online, a resource I often use for patients. When I walked it down the hall to her desk, she was putting down a telephone. "That's OK," she said. "I just got my answer from the ER nurse."

She more or less dismissed me with a nod to place the literature on her desk. Clearly, in her mind I knew less about a lab test than the nurse who happened to answer the telephone in the emergency room.

Why was a nurse able to answer her question about a lab test better than the lab?

Nurses interact with patients in more ways and more often than anyone else in the hospital. They also work with clinicians, absorbing a provider's unique tone, temperament and knowledge. And a large part of this is learning to read patients.

Perhaps, the receptionist wasn't looking for an explanation of the test result. What she wanted was more practical: a simple, direct statement of what her son's doctor was looking for. Why is the doctor worried about this test? is a completely different question than What does a bicarb level mean?, for instance. Answering the former may soothe a mother's angst; the latter generates jargon. Apparently, the nurse read the need of this receptionist and answered her question appropriately.

How often do nurses explain lab results to patients? This may be an opportunity for the lab to be proactive about nursing education, but we may also take a lesson in what a patient needs. That human connection can make all the difference in patient care.

Smear Art

Scott Warner — Mon, 11 Aug 2008 15:48:00 GMT

Little these days is not automated. The art of laboratory medicine is vanishing, from spectrophotometers to Folin-Wu filtrates to Lee-White clotting times. I'd wax poetic about the meditative lure of bench work, but I don't miss mouth pipetting, spit tubes (pg. 20), or Miller discs.

The wedge smear technique for making blood smears survives. Like handwriting, it is learned early and changed slowly. And while most colleagues make "acceptable" slides, others seem to use the heel of their shoe as the spreader.

Does it matter? This study concludes automated preparation methods show less variability with automated counts, particularly with monocyte populations. This makes sense, as cell size affects how cells are rolled along and eventually flattened by the edge, angle and force of a spreader slide. Sample hematocrit, room temperature and humidity, and slide cleanliness can all make a difference.

Two thoughts.

One, modern analyzers classify thousands of cells, not 100 cells from a few dozen fields. We don't need to match an automated count, just to scan a feathered edge long and wide enough to locate and identify abnormalities. How long and wide are disputable.

Two, if we reduce variability in smear preparation, does it significantly increase precision or accuracy? While difficult to get techs to agree what a smear should look like, it's even harder to get them to change. Experience teaches that day-to-day, tech-to-tech variation in identifying and counting cells is greater than, and not necessarily caused by, millimeters on feathered edges.

Maybe, that's just the artist in me talking. A well-made blood smear is a pretty thing, after all. How do you make your smears? And if you've switched, are automated methods really any better?

Understanding Tests

Scott Warner — Mon, 28 Apr 2008 17:20:00 GMT

It is the laboratory scientist's responsibility to make a test choice clear for the physician. We can be expected to reasonably interpret an order that doesn't match a menu name (AST for SGOT, for example). Other times, a questionable order is a methodology choice. A simple telephone call to the provider, a pathologist or a reference laboratory can clarify ambiguity.

The FDA and the Institute for Safe Medication Practices (ISMP) recognize the error potential of ambiguous abbreviations. Their education focuses on medications, but it isn't a stretch to apply this to the laboratory. Reference laboratory test menus, in particular, can create a bewilderment of choice.

Less often, a test report isn't clear or is difficult to interpret. There are even those times when the provider hasn't ordered what he thought he ordered.

As this small study published in a British biochemistry journal suggests, not all physicians are equally up to speed on laboratory tests. Junior doctors--similar to American interns--order a majority of laboratory tests, but nearly one in six admitted not being able to fully interpret the results.

How can laboratories help? A proactive approach can help make a laboratory, regardless of size, an approachable and responsive resource.

A laboratory newsletter is an excellent way to let providers know what the laboratory offers. Reference materials such as specimen collection charts, media guides and directory of service booklets are invaluable for providers and office staff.

Much of this material can be on a hospital Web site, emailed or distributed on a CDROM. Test synonyms in computerized physician order entry (CPOE) systems are helpful. Finally, it's useful to periodically survey providers to make sure a particular service responds to their needs.

As engaged professionals we can help improve patient care. We need to build partnerships that help providers understand laboratory test menus and reports.

Quick and Dirty, Part One

Scott Warner — Thu, 13 Mar 2008 17:21:00 GMT

As a programmer on the coast of Maine 20 years ago, I liked a quick and dirty solution--the inelegant, simplistic, but often workable idea. Scientists call this lex parsimoniae or Occam's Razor (after a 14^th century English logician), which states the simplest solution is best, all things being equal. It surfaces in Toyota's Lean philosophy, pushing the automobile giant ahead of General Motors.

In the next few entries I'll describe such a solution.

Here's the problem: requisitions without diagnoses generate telephone calls or faxes to doctors' offices, causing delays. If the office isn't open, the patient may be turned away. Data has to be collected that answers at least the following:

What time of day does the disruption occur?
Who is the ordering physician?

Requisitions can be tagged and tallied at day's end, but this invites transcription error. An information system "marker" can be used--a dummy test, location or other tag--nice if a user does it and patient billing doesn't mind. Or a paper log can be kept--ugh.

Above all, we need good data--complete and accurate as possible--and process disruptions corrupt data proportionally. Minimal disruption is also crucial to maintain patient throughput at a busy phlebotomy station; we are especially interested in interruptions at busy times. Before we start fitting solutions to the problem, we need to imagine what the ideal solution is.

Data is captured at point of order entry
Minimal extra steps
Collection is novel and easy
Data is easily mined

A lowly DOS batch file--quick and dirty--does the job. Next entry, I'll describe what that is.