Book review: The Checklist Manifesto by Atul Gawande

This is a great book that I would recommend anyone in the quality field read.  The key message of the book is that the complexity we face in our jobs is immense, and no amount of learning can help error-proof our jobs as much as well-designed checklists can.  Dr. Gawande follows two main professions and how checklists affect their jobs: airline pilots and surgeons.  He starts out by showing how the first implementation of checklists in each of these areas appeared: for airline pilots, the crash of the first B-17 test flight in the 1930s; for surgeons, the invention of a checklist to prevent IV line infections in the 2000s.  He shows how these checklists were for fairly “simple” problems (ensuring all airplane systems are in the right configurations at takeoff or taking all the proper steps to prevent contamination during IV line penetration), and then goes on to describe two other classes of problems that checklists can be applied to: complicated problems (a collection of simple problems) and complex problems (many independent variables and outside influences, a great number of which may beyond prediction).  While simple problem checklists tend to focus on ‘do this, do that’, complex problem checklists also focus on communication.  He illustrates this with the checklists required to build a skyscraper: first there is the Gantt chart that is a summation of checklists, but then there is the bin list of deviations (abnormalities) that the experts have to come in to evaluate and agree upon the correct next steps – but do so in an expeditious manner to keep the overall project on schedule.

Before he talks about the formation of more powerful checklists in his surgical field, Dr. Gawande highlights some of the properties of checklists.

• The checklists have to be unambiguous.  If it’s a “READ-DO” checklist (those that address simple problems), the DO items need to be simple to execute.  If it’s a “DO-CONFIRM” checklist (focusing on communication, letting the team be accountable for ensuring all necessary steps are completed), the CONFIRM needs to have a clear single yes/no answer.
• If a checklist needs to be followed, it should be followed – not analyzed.  If a plane is executing a recovery maneuver from a depressurization event, the pilots should be executing the checklist, not trying to figure out whether a depressurization alarm is really going off or is just malfunctioning.  (There will be always be time later to deal with it if it was malfunctioning; if not, precious time has been wasted troubleshooting rather than reacting.)
• Checklists should also be updated frequently with key learnings – not only are post-mortems very powerful events, but good checklists incorporate learnings from other people’s failures.  (The example he uses here is other well-known value investors learning lessons from Berkshire Hathaway’s failed investment in Cort Furniture; and yes, he cites a number of investing houses, both value and VC, that use checklists to help guide their investing.)
• Checklists for complex problems come with a caveat: if written properly, they actually promote teamwork (between pilot and co-pilot, between surgeon, nurse and anesthesiologist), which leads to lower error rates from improved communication and accountability.

When Dr. Gawande spearheaded the development of surgery checklists for World Health Organization, there was an initial disaster in the implementation – the first checklist he tried was too vague, and frustrated the doctors using it (including himself; he chose his team for the first pilot of the checklist).  But he learned a couple of key take-aways:

• The checklist should only have the “killer items”.  Even though operating room fires happen in surgery and are a major risk, their frequency is so low that having steps for prevention in the checklist slowed it down beyond his target execution time (no more than a minute per checklist execution).  But even though there were other steps for ensuring it was the right patient getting the right surgery, the bad publicity from such a mishap was worth doing a quick triple-check before starting the procedure.  So he used FMEA-style criticality (severity x occurrence) with an “is it a low-hanging fruit” approach to decide whether or not to include specific items.
• He made sure that he trained the management (hospital chief of staff, chief surgeon) when the checklist was implemented at each hospital, so that each facility could implement it within their own culture and structure.

Once his checklists were implemented, they saw improvements across the range of hospitals that were part of the pilot implementation (4 "high income" and 4 "low income" hospitals).  Even the most  highly accomplished hospitals (such as University of Washington Medical Center in Seattle) saw enormous improvements in avoiding “simple” errors that could lead to disaster.  And despite still having detractors (20% of the surgeons involved in an 8-hospital pilot in 2008 still didn't feel the checklists were necessary), the value was uniformly there (33% to nearly 50% reduction in post-surgical complications at each of the pilot hospitals); and even the detractors saw some of the benefit (93% of the surgeons in the pilots said that if they were going into surgery as a patient, they would want the checklists used for their surgery).  This demonstrates that even in a high-knowledge, high-expertise, fast-paced environment, checklists can be implemented in a way that minimizes bureaucracy, even improves efficiency (the time to complete the checks actually sped up surgeries because later double-checks weren’t needed) and delivers the ultimate benefit in the end. 

There are numerous parallels that can be gleaned from this book and applied to my business in semiconductor materials quality:

• the power of post-mortems relying on improving the checklist the next time around so others can benefit from the previous mistake;
• the use of checklists to promote expert review, not just execution of menial tasks;
• and in the larger picture, the fact that a well-designed checklist can provide a benefit in eliminating errors from any project no matter how complex (semiconductors are very complex, but last I checked, a patient under the knife in a surgical ward is infinitely moreso, in addition to far less predictable).

Dr. Gawande's work show great opportunities for application in a wide range of businesses (product and service) for driving error elimination toward achieving the goal of Quality Incident Free.