Keith A. Willoughby, University of Saskatchewan, and Dale Schattenkirk, Five Hills Health Region, “Make No Mistake: Reducing Errors in Hospital Pathology Samples and Information Flow”

Health care in Saskatchewan:

  • quarter million square miles (bigger than Texas), just a million people
  • 12 health regions plus 1 health authority
  • Five Hills has 1700 employees, serves 55,000 people in and around Moose Jaw (Al Capone had tunnels there… or so goes the legend)

Pursuing Excellence: initiative launched 2007 after CEO saw Lean/Six Sigma at work in Washington state (Virginia Mason)

Unionized workforce was worried “lean” meant “job cuts”; admins were careful to word things to make clear that wasn’t what was coming.

Pathology: studying and diagnosing disease by assessment of tissues or organs. Errors cause incorrect diagnoses, more tests, unnecessary patient anxiety, extra work/inefficiency. (Meier et al. 2005) Most likely source of errors: defective errors.

Interesting: when Willoughby started this project, there was a seeming surge of stories of medical errors in Canada. Nice timing! But note: these guys are focusing on errors in information flow, not the typical medical error. They found anecdotal evidence but no rigorous analysis (there’s the gap! go find out!). And note that when those news stories came out, none were from Alberta, which is pushing electronic medical records.

Process maps, check sheets, Pareto analysis: this is basic textbook ops mgmt stuff, may seem trivial, but Willoughby says try it! Using these tools is remarkably challenging in a health care setting. A previous effort had used these tools and come up with a majority category for errors of “Other,” which is a sign of bad data collection. That shouldn’t happen in a Pareto analysis!

W&S assembled team of surgeon, pathologist, lab services director, OR nurse, lab specialist — that’s a trail of who’s involved. W&S wanted a knowledgeable team to get buy-in from participants.

Problem Statement: Lack of standards and poor communication channels between OR and pathology may lead to misdiagnosis and frustration.

Aim Statement: Perfection! 100% complete and accurate information

Process map is complicated: shows 30+ elements! Lots of folks involved had no grasp of the processes out in other units. Lots of silos! Process map helped researchers understand where mistakes could happen. (Post-It notes are key! They foster creativity and experimentation; other team members could mess with it, contribute to it, learn from it.)

Five error points in the process:

  1. first matching: Area where sample is first received in lab area
  2. LaboratoryInformation System matching
  3. proper preparation
  4. errors in specimen
  5. final check

Researchers developed check sheets for hospital staff to complete at each of the five process points to say whether there was an error and, if so, what type.

Pareto analysis: 80% of errors at first matching (31), few at final check (4) and proper prep (2), none at errors in specimen or LIS.

  • So I wonder: given this filtering process, did finding the first matching error first and stopping the show (did they stop the show?) somehow prevent us from catching the later errors?

97% of first matching errors were name of procedure (most serious), doctor’s name, and date of procedure.

  • Question: W. mentioned that name of procedure is serious: to what extent did we quantify the relation of those errors to negative outcomes?

Response: researchers tried redesigning the form for pathology reports, but three-week trial with the improved form produced an insignificant reduction in error rate (down from 31% to 29%). The improved form did result in near elimination of errors in name of procedure, but doctor name and date of procedure errors increased, as did body part errors!

  • I’m inclined to see the glass half full: their form produced mixed results, but it produced darn good results on an error that mattered! We need to measure the results as error rate x cost of error, a deeper analysis that Willoughby says hasn’t been done yet.

Note that 90% of errors came from staff not on the team! So maybe you need to keep people in the loop to make the information flow better.

This isn’t just “one and done”; we’re using this knowledge for additional projects in Saskatchewan. And remember: you can’t shut down health care to do a study — that makes it fun and challenging.