Chapter 2. Basic Principles of Patient Safety

The traditional approach to medical errors has been to blame the provider who delivers care directly to the patient, acting at what is sometimes called the “sharp end” of care: the doctor performing the transplant operation or diagnosing the patient's chest pain, the nurse hanging the intravenous medication bag, or the pharmacist preparing the chemotherapy. Over the last decade, we have recognized that this approach overlooks the fact that most errors are committed by hardworking, well-trained individuals, and such errors are unlikely to be prevented by admonishing people to be more careful, or by shaming, firing, or suing them.

The modern patient safety movement replaces “the blame and shame game” with an approach known as systems thinking. This paradigm acknowledges the human condition—namely, that humans err—and concludes that safety depends on creating systems that anticipate errors and either prevent or catch them before they cause harm. Such an approach has been the cornerstone of safety improvements in other high-risk industries but has been ignored in medicine until the past decade.

British psychologist James Reason's “Swiss cheese model” of organizational accidents has been widely embraced as a mental model for system safety1,2 (Figure 2-1). This model, drawn from innumerable accident investigations in fields such as commercial aviation and nuclear power, emphasizes that in complex organizations, a single “sharp-end” (the person in the control booth in the nuclear plant, the surgeon making the incision) error is rarely enough to cause harm. Instead, such errors must penetrate multiple incomplete layers of protection (“layers of Swiss cheese”) to cause a devastating result. Reason's model highlights the need to focus less on the (futile) goal of trying to perfect human behavior and more on aiming to shrink the holes in the Swiss cheese (sometimes referred to as latent errors) and create multiple overlapping layers of protection to decrease the probability that the holes will ever align and let an error slip through.

The Swiss cheese model emphasizes that analyses of medical errors need to focus on their “root causes”—not just the smoking gun, sharp-end error, but all the underlying conditions that made an error possible (or, in some situations, inevitable) (Chapter 14). A number of investigators have developed schema for categorizing the root causes of errors; the most widely used, by Charles Vincent, is shown in Table 2-1.3,4 The schema explicitly forces the error reviewer to ask whether there should have been a checklist or read back, whether the resident was too fatigued to think clearly, or whether the young nurse was too intimidated to speak up when she suspected an error.

Even though we now understand that the root cause of hundreds of thousands of errors each year lies at the “blunt end,” the proximate cause is often an act committed (or neglected, or performed incorrectly) by a caregiver. Even as we embrace the systems approach as the most useful paradigm, it would be wrong not to tackle these human errors as well. After all, even a room filled with flammable gas will not explode unless someone strikes a match.

In thinking about human errors, it is useful to differentiate between “slips” and “mistakes”, and to do this, one must appreciate the difference between conscious behavior and automatic behavior. Conscious behavior is what we do when we “pay attention” to a task, and it is especially important when we are doing something new, like learning to play the piano or program our DVD player. On the other hand, automatic behaviors are the things we do almost unconsciously—they may have required a lot of thought initially, but after a while we do them virtually “in our sleep.” Humans prefer automatic behaviors because they take less energy, have predictable outcomes, and allow us to “multitask”—do other things at the same time. Some of these other tasks are also automatic behaviors, like driving a car while drinking coffee or talking on the phone, but some require conscious thought. These latter moments—when a doctor tries to write a “routine” prescription while also pondering his approach to a challenging patient—are particularly risky, both for making errors in the routine, automatic process (“slips”) or in the conscious process (“mistakes”).

Now that we've distinguished the two types of tasks, let's turn to slips versus mistakes. Slips are inadvertent, unconscious lapses in the performance of some automatic task: you absently drive to work on Sunday morning because your automatic behavior kicks in and dictates your actions. Slips occur most often when we put an activity on “autopilot” so we can manage new sensory inputs, think through a problem, or deal with emotional upset, fatigue, or stress (a pretty good summing up of most healthcare environments). Mistakes, on the other hand, result from incorrect choices. Rather than blundering into them while we are distracted, we usually make mistakes because of insufficient knowledge, lack of experience or training, inadequate information (or inability to interpret available information properly), or applying the wrong set of rules or algorithms to a decision (we'll delve into this area more deeply when we discuss diagnostic errors in Chapter 6).

When measured on an “errors per action” yardstick, conscious behaviors are more prone to mistakes than automatic behaviors are prone to slips. However, slips probably represent the greater overall threat to patient safety because so much of what healthcare providers do is automatic. Doctors and nurses are most likely to slip while doing something they have done correctly a thousand times: asking patients if they are allergic to any medications before writing a prescription, remembering to verify a patient's identity before sending her off to a procedure, or loading a syringe with heparin (and not insulin) before flushing an IV line (the latter two cases are described in Chapters 15 and 4, respectively).

The complexity of healthcare work adds to the risks. Like pilots, soldiers, and others trained to work in high-risk occupations, doctors and nurses are programmed to do many specific tasks, under pressure, with a high degree of accuracy. But unlike most other professions, medical jobs typically combine three very different types of tasks: lots of conscious behaviors (complex decisions, judgment calls), many “customer” interactions, and innumerable automatic behaviors. Physician training, in particular, has traditionally emphasized the highly cognitive aspects with a small focus on the human interactions. But, until very recently, it has completely ignored the importance, and risky nature, of automatic behaviors.

With all of this in mind, how then should we respond to the inevitability of slips? Our typical response would be to reprimand (if not fire) a nurse for giving the wrong medication, admonishing her to “be more careful next time!” Even if the nurse did, she is just as likely to commit a different error while automatically carrying out a different task in a different setting. As James Reason reminds us, “Errors are largely unintentional. It is very difficult for management to control what people did not intend to do in the first place.”2 And it is not just managers whose instinct is to blame the provider at the sharp end—we providers blame ourselves! When we make a slip—a stupid error in something that we usually do perfectly “in our sleep”—we feel embarrassed. We chastise ourselves harder than any supervisor could, and swear we'll never make a careless mistake like that again.5 Realistically, though, such promises are almost impossible to keep.

Whatever the strategy employed to prevent slips (and they will be discussed throughout the book), a clear lesson is that boring, repetitive tasks can be dangerous and are often performed better by machines. In medicine, these tasks include monitoring a patient's oxygen level during a long surgery, suturing large wounds, holding surgical retractors steady for a long time, and scanning mountains of data for significant patterns. As anesthesiologist Alan Merry and legal scholar and novelist Alexander McCall Smith have observed:

people have no need to apologize for their failure to achieve machine-like standards in those activities for which machines are better suited. They are good at other things—original thought, for one, empathy and compassion for another. . It is true that people are distractible—but in fact this provides a major survival advantage for them. A machine (unless expressly designed to detect such an event) will continue with its repetitive task while the house burns down around it, whereas most humans will notice that something unexpected is going on and will change their activity. .6

The decisions made and actions taken by individual doctors and nurses occur in the context of the structure, policies, and culture of their organizations and healthcare systems. We will explore these issues later in this book because appreciating this context is often the key to improving safety. For now, as we think about mental models relevant to safety, it is worth introducing the topic of complexity theory.7

You'll recall that I referred to complexity earlier when discussing many of the decisions that frontline caregivers need to make every day. But another type of complexity—that of the entire ecosystem of care—is also at play when we consider patient safety. This area goes under the broad heading of complexity theory, a branch of management thinking that holds that large organizations don't operate like predictable and static machines, in which Inputs A and B predictably lead to Result C. Rather, they operate as complex adaptive systems, with elements of unpredictability, codependancy, and nonlinearity that must be appreciated as we try to make them work better.8

Complexity theory divides decisions and problems into three general categories: simple, complicated, and complex.9 Simple problems are ones in which the inputs and outputs are known; they can be managed by following a recipe or a set of rules. Baking a cake is a simple problem; so is choosing the right antibiotics to treat pyelonephritis. Complicated decisions involve substantial uncertainties: the solutions may not be known, but they are potentially knowable. An example is designing a rocket ship to fly to the moon—if you were working for NASA in 1962 and heard President Kennedy declare this as a national goal, you probably believed it was not going to be easy but, with enough brainpower and resources, it could be done. Finally, complex decisions are likened to raising a child. Although we may have a general sense of what works, the actual formula for success is, alas, unknowable (or certainly seems so most days!).

Understanding these differences is vital because we need to match our approaches to the types of problems at hand. A checklist may be a wonderful solution for a simple problem, but a distraction for a complex one. Enacting a series of rules and policies may seem like progress, but may actually set us back if it stifles innovation and collegial exchange among frontline workers. Sometimes the best approach to a complex problem is to try something that seems sensible, measure the results (making sure workers feel able to speak truthfully and staying attuned to unanticipated consequences), and repeat this cycle many times.

But, as organizational expert Paul Plesk points out, our recognition of the complexity of healthcare systems is not a call for chaos. “A somewhat surprising finding from research on complex adaptive systems,” he writes, “is that relatively simple rules can lead to complex, emergent, innovative system behavior.”8 Atul Gawande expands on this point in The Checklist Manifesto.9 A good checklist, by specifying a few behaviors and encouraging cross-talk among frontline caregivers, can lead to remarkable improvements that go well beyond adherence to the items on the list (Chapter 15).

Table 2-2 presents a distillation of some of the key lessons from studies of complex adaptive systems. We'll return to these issues several times over the course of the book, such as when we discuss culture and High Reliability Organizations (Chapter 15) and consider how to build strong institutional patient safety programs (Chapter 22).

Drawing on these mental models, the modern patient safety field emphasizes the need to shore up systems to prevent or catch errors rather than to create “goof-proof” individual providers. For example, errors in routine behaviors (“slips”) can best be prevented by building in redundancies and cross checks, in the form of checklists, read backs (“let me read your order back to you”), and other standardized safety procedures (e.g., signing the surgical site prior to an operation, asking patients their name before medication administration). It also emphasizes the need for standardization and simplification. For example, standardizing the process of taking a patient to the magnetic resonance imaging scanner makes it far easier to “bake in” the correct safety procedures.

Recently, there has been increased emphasis on decreasing errors at the person–machine interface through the use of forcing functions—engineering solutions that decrease the probability of human error. The classic example outside medicine was the 1980s-era modifications to automobile braking systems that rendered it impossible to place a car in reverse when the driver's foot is off the brake. In healthcare, forcing functions include changing the gas nozzles and connectors so that anesthesiologists cannot mistakenly hook up the wrong gas to a patient. Given the ever-increasing complexity of modern medicine, building in such forcing functions (in intravenous pumps, cardiac defibrillators, mechanical ventilators, and computerized order entry systems) will be crucial to safety (Chapter 7).

In addition to systems enhancements, there has been a growing recognition of the importance of improving communication and teamwork. Commercial pilots all participate in “crew resource management” courses in which they train for emergencies with other crew, learning to encourage open dialogue, communicate clearly using standard language, and utilize checklists, debriefings, and other systemic approaches. The evidence that such training enhances patient safety is increasingly persuasive.10 The term “culture of safety” is used as shorthand for an environment in which teamwork, clear communication, and openness about errors (both to other healthcare professionals and to patients) are operative (Chapter 15).

Another key patient safety principle is to learn from one's mistakes. This may take multiple forms. Safe systems have a culture in which errors are openly discussed, often in morbidity and mortality (M&M) conferences. There is a new push to make sure that such discussions include members of the appropriate disciplines (such as nursing and hospital administration, not just physicians), point out errors rather than gloss over them in the name of avoiding a punitive atmosphere, and emphasize systems thinking and solutions.11,12 In addition to open discussions at conferences, safe organizations build in mechanisms to hear about errors from frontline staff, often via incident reporting systems or unit-based safety teams,12,13 and to perform detailed (“root cause”) analyses14 of major errors (“sentinel events”) in an effort to define all the “layers of Swiss cheese” that need attention (Chapter 14). They also recognize the realities of complex adaptive systems: overmanaging workers through boatloads of top-down, prescriptive rules and directives may be more unsafe than tolerating some degree of flexibility and experimentation on the frontlines.

Finally, there is increasing appreciation of the importance of a well-trained, well-staffed, and well-rested workforce in delivering safe care. There is now evidence linking low nurse-to-patient ratios, long resident work hours, and lack of board certification to poor patient outcomes (Chapter 16).15–19 Such research is catalyzing a more holistic view of patient safety, recognizing that the implementation of “safer systems” will not create safe patient care if the providers are overextended, poorly trained, or undersupervised. Nor is “systems thinking” an excuse to allow caregivers or their leaders to remain unaccountable for being unwilling to follow appropriate and well-vetted safety rules (Chapter 19).20

This long list of potential approaches to improving safety (each of which will be discussed in greater detail later) highlights one of the great challenges in the field: in the absence of comparative evidence, and in light of the high cost of some of the interventions (e.g., improved staffing, computerized order entry, simulation, teamwork training), even organizations committed to safety can become bewildered as they consider which approach to emphasize.21 Institutions quite naturally focus on the practices that are measured, publicly reported, and compensated. As the next chapter will show, such a prioritization scheme will tend to elevate quality improvement strategies over those focused on patient safety because the results of the former are easier to measure. Thankfully, many of the approaches to improving quality, such as computerization and standardization, will also yield safety benefits. On the other hand, because improving culture is both difficult and hard to measure, it risks being shuffled to the bottom of the deck, notwithstanding its importance to patient safety. Pronovost, Miller, and I have suggested a measurement scheme for patient safety that takes these considerations into account (Table 2-3).22

• The modern approach to patient safety hinges on “systems thinking”—a recognition that most errors are made by competent, careful, and caring providers, and preventing these errors often involves embedding providers in a system that anticipates glitches and catches them before they do harm.
• James Reason's “Swiss cheese model” is the dominant paradigm for understanding the relationship between active (“sharp end”) errors and latent (“blunt end”) errors; it is important to resist the temptation to focus solely on the former and neglect the latter.
• A variety of strategies should be employed to create safer systems, including simplification, standardization, building in redundancies, improving teamwork and communication, and learning from past mistakes.
• There is a growing appreciation of the complexity of healthcare delivery organizations, which are vivid examples of complex adaptive systems. Such systems rarely respond in linear, predictable ways to prescriptive policies. This means that efforts to improve safety must combine rules and standards with messier activities that account for the importance of culture, innovation, and iterative learning.