Although Hippocrates said “first, do no harm” over 2000 years ago and many hospitals have long hosted conferences to discuss errors (Morbidity and Mortality, or “M&M,” conferences), until recently medical errors were considered an inevitable by-product of modern medicine or the unfortunate detritus of bad providers. This began to change in late 1999, with the Institute of Medicine’s (IOM) publication of To Err is Human: Building a Safer Health System.1 This report, which estimated that 44,000 to 98,000 Americans die each year from medical mistakes, generated tremendous public and media attention, and set the stage for unprecedented efforts to improve patient safety. Of course, these seminal works built on a rich tapestry of inquiry and leadership in the field of patient safety (Appendix III), familiar to a small group of devotees but generally unknown to mainstream providers, administrators, policymakers, and patients.

The IOM death estimate, which was drawn from thousands of chart reviews in New York,2,3 Colorado, and Utah4 in the late 1980s and early 1990s, was followed by studies that showed huge numbers of medication errors, communication problems in intensive care units (ICUs), gaps in the discharge process, retained sponges in the operating room—in short, everywhere one looked there was evidence of major problems in patient safety. Moreover, accompanying this information in the professional literature were scores of dramatic reports in the lay media: errors involving the wrong patient going to a procedure or the wrong limb being operated on, chemotherapy overdoses, botched transplants, patients released from the emergency department (ED) only to die later from myocardial infarction or aortic dissection, and more (Table 1-1).

The patient safety literature contains many overlapping terms to describe safety-related issues. Although the terms sometimes confuse more than clarify, two key distinctions underlie most of the terminology and allow one to keep it relatively straight. First, because patients commonly experience adverse outcomes, it is important to distinguish adverse outcomes as a result of medical care from morbidity and mortality that patients suffer as a consequence of their underlying medical conditions. The former are known as adverse events or harm (the two terms are generally used interchangeably) and have been defined by the Institute for Healthcare Improvement as:

Unintended physical injury resulting from or contributed to by medical care (including the absence of indicated medical treatment) that requires additional monitoring, treatment, or hospitalization, or that results in death.5

Second, because patients may experience harm from their medical care in the absence of any errors (i.e., from accepted complications of surgery or medication side effects), the patient safety literature separates preventable adverse events from nonpreventable ones. Figure 1-1 shows a Venn diagram depicting these various terms.

Now, where do errors or mistakes fit in? The safety literature commonly defines an error as “an act of commission (doing something wrong) or omission (failing to do the right thing) leading to an undesirable outcome or significant potential for such an outcome.”6 Note that many errors do not result in adverse events (Figure 1-1)—we generally characterize the more serious ones as “near misses” or “close calls.” Note too that some errors involve care that falls below a professional standard of care—these are called negligence and may create legal liability or a duty to compensate the patient in some systems (Chapter 18). Finally, although most preventable adverse events involve errors, not all of them do (see “The Challenges of Measuring Errors and Safety,” below).

In the early days of the patient safety movement, most of the focus was on measuring and decreasing the incidence of errors. Increasingly, safety experts prefer to highlight preventable adverse events or preventable harm—rather than errors—as the main target of the field. Framing this issue in the language of the Donabedian process–outcome–structure triad (Chapter 3),7 one might think about harm as the “outcome” and errors as the “process.” Advocates of this new approach see several advantages to it. First, focusing on adverse events removes the implication that a specific provider was responsible for the harm, which may generate defensiveness by caregivers or an inordinate focus by the organization on dealing with the individual rather than the system (Chapter 2).8–10 Moreover, patients, quite naturally, care more about what happened to them than whether their doctor or nurse made a mistake.

Although these advocates make a strong argument (and I find myself in their camp with increasing frequency), it is worth appreciating the fear on the part of other experts that the terms “preventable adverse events” or “preventable harm” seem a bit too careful, perhaps even politically correct, lacking the visceral punch of “error” or “mistake” that helped launch the safety field. One can be sure that this vigorous debate—which has been with us since the early days of the field11,12—will continue. For now, it is clear that patient safety’s center of gravity has shifted toward adverse events/harm as the focus of measurements and interventions.

These distinctions are important to understand as one tries to wade through the safety literature or interpret individual cases. To help illustrate the point, let’s consider three scenarios involving a patient placed appropriately on warfarin for chronic atrial fibrillation who develops a gastrointestinal (GI) bleed. If the bleeding began while the patient’s international normalized ratio (INR) was therapeutic, he would be said to be the victim of an adverse event (often called a side effect when caused by a medication), but not a preventable adverse event or a medical error. Such cases should prompt us to reevaluate the benefit-to-risk ratio of the intervention (in this patient and others like him) and to try to improve the science (i.e., identify a strategy with a comparable benefit but lower risk), but not to infer that there was a mistake in the patient’s care or that the harm was preventable.

If, on the other hand, the patient had bled in the setting of a supratherapeutic INR, but there was no overt error on the part of the physician (she was following standard guidelines for warfarin dosing and monitoring), the patient would be said to have suffered a preventable adverse event, but the case would still not be characterized as having been caused by an error. Here too, the focus should be on reassessing the benefit-to-risk ratio in this and similar patients and on improving the science, although it is worth considering systems changes (such as more frequent monitoring) that might have made a difference in this case and ones like it.

Finally, it would be a preventable adverse event and an error if the INR was supratherapeutic because the physician prescribed a new medication without checking for possible drug interactions. As we’ll see, before focusing too narrowly on the failures of the individual caregiver (unless such carelessness was habitual or willful), the more productive stance will likely be to consider systems changes (a computerized order entry system with robust decision support,13 or a dedicated anticoagulation clinic14) that might prevent such errors and harm in the future. Ideally, of course, warfarin would be replaced by a medication that delivered the same therapeutic benefit but lacked its drug interactions and monitoring requirements.15

Before leaving the area of patient safety terminology, I should add that some safety experts bristle at the distinction between preventable and nonpreventable adverse events, reminding us that certain types of harm previously thought unpreventable are now known to be preventable with better systems. Some even contend that labeling such events “nonpreventable” is defeatist and self-fulfilling. Probably the strongest support for this argument comes from the story of central line–associated bloodstream infections, which were once felt to be inevitable consequences of modern medicine but are now known to be largely preventable with the consistent application of a “bundle” of safety practices16 (Chapter 10). Although this point of view has considerable merit, the distinction between preventability and nonpreventability permeates the literature, and blurring it risks a public perception that all adverse events result from errors, which they do not. As one point of reference, studies of cases of harm in hospitalized patients generally find about half to have been preventable.9,10

This debate is likely to continue because measurement is so fundamental to our efforts to improve safety. Robust and reliable measurement is the cornerstone for conducting sound research about patient safety practices, prioritizing areas of focus for study and implementation, holding individuals and institutions accountable for performance, and assessing how we are faring in our efforts to improve safety.8 So, although matters of terminology and measurement might seem arcane and of only marginal relevance to patients, caregivers, administrators, and policymakers, they are really quite central to the field.

At Douglastown Hospital, the patient safety officer has become concerned about the frequency of medication errors. One patient received a 10-fold overdose of insulin when the order “please give 10U regular insulin” was interpreted as “100 regular insulin.” Another patient received a cephalosporin antibiotic for pneumonia, despite being allergic to this class of drugs. A third suffered a GI bleed when an overdose of warfarin led to over-anticoagulation.

In response to these incidents, the hospital is considering whether to invest in a $50 million computerized provider order entry system. The Chief Financial Officer, knowing that this expense will mean that the hospital will have to forego its planned purchase of a new 256-slice computed tomography scanner and the construction of two new operating rooms (investments with near-guaranteed positive returns on investment), asks the safety officer, “How will we know that we’ve made a difference?”

The CFO’s question seems relatively straightforward, but is much harder to answer than you might think. Let’s consider various ways of measuring errors.

The most common method is through self-reports of errors by providers, usually referred to as incident reports. These reports have traditionally been completed by pencil and paper; they are increasingly being inputted through a computerized system. Incident reports (see also Chapter 14) might seem to be a reliable way of tracking errors, but there are several problems with using them to measure the frequency of errors.17 First, although nurses tend to report errors through incident reporting systems, few doctors do,18 either not reporting at all or reporting through informal channels (such as, in teaching programs, telling the chief residents). Secondly, as most reporting systems are voluntary, the frequency of reports will be influenced by many factors other than the number of errors. Let’s say the institution has improved its safety culture (Chapter 15), such that reporting of errors is now strongly encouraged by local leaders and incident reports result in tangible action. Under these circumstances, an increase in incident reports might well reflect the same number of, or even fewer, errors being reported more assiduously. This conundrum distinguishes measuring patient safety from measuring the quality of care, which is less dependent on voluntary reporting and thus can be done more reliably (Chapter 3).