In Lessons from a Sunken Ship, I recounted the story of a 1628 shipwreck that occurred in the aftermath of a failed stability test, a test result known to at least 30 shipbuilders (who nearly put the Vasa underwater during a preliminary test of seaworthiness) and the ranking military leader who observed the aborted test. Signing off on the launch, the ranking officer lamented the absence of the King, apparently the only person with authority to cancel the launch.
This story lends itself to talking about the dangers of rigid hierarchies, and I'll probably return to it at some point, rigid hierarchies having sunk more than a few ships in healthcare. But the Vasa also illustrates principles about the hierarchy of error and harm prevention:
1. Eliminate or prevent mistakes. A better design would have prevented the Vasa from going down.
2. Make mistakes that have been set in motion visible. The ship did not perform as expected when subjected to simulated sea-like conditions. Not launching a ship with dubious stability would have prevented the Vasa from going down.
3. Mitigating the hazard should a mistake occur. Lifeboats prevented some people on the sinking ship from going down.
4. Education/re-education about how to manage known hazards. Swimming lessons might have helped some people save themselves.
Healthcare has been criticized for the tendency to bottom feed when it comes to risk reduction, meaning that we tend to rely on risk-reduction strategies low on the hierarchy. This doesn't mean that, as individuals, healthcare professionals don't care about risk or don't want things to turn out well. It simply means that we're more likely to select and implement interventions like "review policy" with individuals who make errors than to examine the underlying factors that allowed frontline workers to err. We spend a lot of energy attempting to teach front line clinicians how to save themselves.
So if we got out of the lifeboats and headed north on the risk-reduction hierarchy, how far could we go and what would the consequences be?
Health and healing are complex, and it's fair to say that we're sailing more than a few badly designed ships. 1 in 7 Americas lack healthcare insurance. Healthcare disparities are rampant. Patients are older, sicker, and rounder than they used to be. Our system does not incentivize prevention. A better design would avert many crises. But redesign of healthcare--something that appears to be emerging as a national priority--is outside the locus of control of individual clinicians, irrespective of how often or how nobly we face the consequences of the current bad design.
So how can front line clinicians prevent a poorly designed vessel from sinking? One answer is: embrace processes and procedures that make mistakes set in motion visible. Be able to identify emerging practice changes as the higher-level risk-reduction strategies they are. Get ready to cancel the launch!
A bar-code scan reveals a mismatch between ordered medications and a similarly packaged one in the patient’s drawer: you’ve cancelled a launch. A pre-procedure time-out reveals a site-of-surgery discrepancy: you’ve cancelled a launch. Reading back and verifying a telephone order (insulin 50, five-zero, units sub-cutaneously now) reveals the prescriber on the crackly line said one-five (15), not 50 units of insulin: you’ve cancelled a launch.
Cancelling a launch is not as good as preventing mistakes from occurring. But this approach trumps lifeboats and swimming lessons. Right now, healthcare is adopting, occasionally adapting, risk-reduction strategies from other industries, industries more reliable than ours.
The best risk-reduction strategies, in my opinion, are yet to come. As healthcare workers--bright, caring, and competent individuals—come to understand the principles that drive reliable performance, participate in developing highly-reliable processes, demand these be vigorously applied, and eventually come-of-age in an environment where reliability is the norm, it will no longer be necessary to report preventable adverse health events as aggregate data!
See you there!