Patient care as a performance system

A recent interview with Dr. Peter J. Pronovost dealt with safer ways to care for patients in hospitals.  Pronovost is the medical director for the Quality and Safety Research Group at Johns Hopkins Hospital in Baltimore.

The interview’s worth reading on its own merits.  I saw in it good examples of performance analysis and efforts to improve performance–with relative few attempts to train people out of non-training problems.

For example, for cardiac catheterization, Hopkins had an infection rate of 11 per 1,000 procedures.  According to Pronovost, at the time that “put us in the worst 10% of the country.”

Here’s a diagram I created to illustrate some influences on performance:

And here are points that Pronovost makes:

  • Hopkins developed a checklist to standardize what to do before catheterization (wash hands, clean skin with chlorhexidine, drape the patient, etc.).  To me, this is support for item 3 above.
  • Supplies, which had been stored in as many as eight places, were prepped in a cath cart–with someone assigned to make sure it was stocked and handy.  Item 2, equipment and materials.
  • The hospital asked nurses to remind doctors to wash their hands–and empowered nurses to stop procedures if this didn’t happen.  Item 8 (standards) and item 9 (feedback) — and, you could argue, item 7 (consequences).

Note also that the Hopkins project defined a specific problem (a high rate of infection), analyzed likely causes, chose action based on those causes, and measured the results.

Pronovost forcefully describes another barrier to performance: workplace culture:

As at many hospitals, we had dysfunctional teamwork because of an exceedingly hierarchal culture…

…in every hospital in America, patients die because of hierarchy. The way doctors are trained, the experiential domain is seen as threatening and unimportant. Yet, a nurse or a family member may be with a patient for 12 hours in a day, while a doctor might only pop in for five minutes.

I mention this not to single out doctors but to emphasize that performance problems usually have multiple causes.  Some you can address in a straightforward fashion (rethinking where to keep the supplies).  Others, you have to keep working at.  In commercial aviation, use of preflight checklists is maintained not only by regulations but by the active support of those who use them: it’s not smarter or more efficient to try memorizing the checklist.  In fact, it’s seen as counterproductive.

(Note what the Skout Group says about workplace culture–and checklists–in terms of USAir 1549, the plane that Sullenberger and Skiles managed to set down in the Hudson River last year, with no loss of life.)

Back to the hospital: isn’t there some need for training?

I couldn’t say; Pronovost’s interview doesn’t have enough detail.  It could be that some hospital staff need training in preparing for catheterization.  If that’s the case, I suspect that inside the generalization of “preparing for catheterization,” there are distinct subtasks: identify and obtain the supplies, prep yourself, prep the patient, assist (or be assisted by) a specialist, and so on.

And perhaps there’s a meta-skill: make sure the individual assigned to this task can first demonstrate an acceptable level of skill.  In other words, something like “we expect you learned this in nursing school (or wherever); here are our standards; we’ll observe you and tell you how you did.”

I don’t know that I’d put the necessary culture change under “training.” I’m pretty sure the label is less important than the goal: having doctors (most not hospital employees) and hospital staff work together to reduce the rate of preventable infection.

Word of the day: nosocomial, meaning “occurring in a hospital.” I came across it in this 2001 CDC report, The Impact of Hospital-Acquired Bloodstream Infections.  Its low estimate for life-threatening bloodstream infections acquired in the hospital is 87,500 per year.  The low estimate of deaths from these bloodstream infections: 8,750.

(And bloodstream infections are estimated at 10% of all nosocomial infection.)