In 1999, the esteemed Institute of Medicine, in a landmark report titled To Err Is Human, noted that between 44,ooo and 98,000 people die yearly in U.S. hospitals from treatment errors.  An error was defined as "the failure of a planned action to be completed as intended or the use of a wrong plan to achieve an aim," so both errors of omission and commission counted. Emphasizing the high end of the estimate, hospitals endorsed the 100,000 Lives Campaign with lapel buttons, lists of "to do's," checklists and acronyms to help guide medical, and to an extent nursing, care.

In 2005, the New England Journal of Medicine published a critique of the oft-cited hospital error figures. The studies that led to the "100,0o0 lives" figure had problems: they didn't require the raters to agree that an error had occurred (in research parlance, this is inter-rater reliability.) If one reviewer was particularly harsh and another particularly lenient, the error rate was distorted. The 2005 paper stated, "if it was stipulated that three reviewers had to agree that an error had occurred, the error rate was less than 1/10 of the rate when the vote of only one of the three was required to make the determination." Another hospital improvement group, Leapfrog, did research on quality-improvement interventions that claimed 50,000 lives could be saved annually in intensive care, but that was about 10 times the number of preventable deaths that studies suggested occur in intensive care units.

So maybe there are only 4400 to 9800 people killed by errors every year.  We want zero people to die this way. But there  is something discomfiting about the need to inflate numbers in order to capture the attention of researchers, providers, and policymakers. Why isn't one unnecessary death too many?

Part of the reason is the difficulty in defining and measuring "unnecessary." If an 80-year-old with severe sepsis, one of the most deadly ICU syndromes, gets the wrong antibiotic and dies, was the error lethal, or would the person have died anyway?  It becomes difficult to impress upon practitioners the urgent need to reduce error rates when we so often see non-harmful or equivocal errors in practice. 

Non-harmful and near-miss errors are notoriously underreported. Although risk management teams urge practitioners to tell them about these events, we rarely take the time (or have the time) to do so.  In a hectic 12 1/2-hour day, when I'm lucky to sit down for half an hour for lunch, I cannot ethically choose to stop caring for my patients long enough to submit a near-miss report. Yet it is under these circumstances, on these days, that I am most likely to make an error. 

There is the problem with defining error. Is it a near miss every time I tell a physician "You need to write an order for _______"?  This happens countless times as such words are uttered by countless nurses to countless physicians, but this sort of mutual cross-checking is not considered error prevention.  (Which is probably to my benefit, because I would spend half my time writing up near-miss reports if reminding a physician colleague to write a certain order were worthy of reporting.) 

Yesterday, as I cared for a very sick 63 year old, it  dawned on me that when I tell a physician what to do, it is not called "an order." Most nurses do not bother to write in the medical record that they told the physician to order something, in fact. I do. I want a modicum of credit for being as knowledgeable (or more) about the right course of action.  I want the regulators, administrators, and policy makers who might one day read that medical record to know that I do not need to be "ordered" to do what my ICU patient needs. So although I used to write "Suggested that physician order additional IV fluid," now I write "Told physician to order more IV fluid."  Because it's what really happens with nurses and physicians.  

And that has prevented at least 100,000 errors.