Patient Safety

Introduction

Although Hippocrates said “first, do no harm” over two millennia ago, and many hospitals have long hosted conferences to discuss errors (Morbidity and Mortality, or M&M, conferences), until recently there was little teaching about the nature of medical mistakes, investment in safety research, regulation of safety standards, or emphasis on safety improvements in healthcare. In late 1999, the Institute of Medicine published To Err is Human: Building a Safer Health System (1). This seminal report, which estimated that 44,000 to 98,000 Americans die each year from medical mistakes (the equivalent of a jumbo jet crashing each day), shocked patients, politicians, and policymakers (and more than a few doctors and nurses), and paved the way for a revolution in our approach to patient safety in the past decade.

Because patients can be injured while receiving perfect care, it is important to separate errors from adverse events. An error is usually defined as “an act or omission that leads to an unanticipated, undesirable outcome or to substantial potential for such an outcome.” Adverse events, on the other hand, are injuries due to medical management rather than the patient’s underlying illness. Although patients experiencing errors and adverse events may be equally harmed, the distinction is crucial because the fixes may be very different. For example, a patient appropriately treated with a blood thinner like warfarin (coumadin) for chronic atrial fibrillation (an irregular heart rhythm that can lead to blood clots) who develops gastrointestinal bleeding despite being on the correct dose of the blood thinner is the victim of an adverse event, but not a medical error. Preventing cases like this will require better science: developing safer blood thinners or new ways of protecting the gastrointestinal system from bleeding complications. On the other hand, it would be an error if the patient became over-anticoagulated (the blood became too thin) because the physician prescribed a new medication without checking for possible drug interactions. The fix here might be a computerized system that would remind the doctor of potential interactions or to have new medications double-checked by a pharmacist.

I discuss the broader issue of healthcare quality (how to measure and improve it) in a dedicated Knol.

The Modern Approach to Patient Safety

The traditional approach to medical errors has been to blame the provider who was at the “sharp end” (delivering) of care: the surgeon performing an operation, or the nurse administering an intravenous medication. This approach is quite natural: we seek individual accountability for our actions, particularly in countries like the United States whose core values emphasize individual responsibility. In days past, when medicine was much simpler and slower paced, the notion of individual responsibility for medical errors made some sense. As one example, it has been estimated that the average internal medicine physician in the 1960s had to master the use of about 10-15 medications (such as morphine, heparin, Lasix, and a handful of antibiotics).

Today, that number is estimated to be about 400. Dr. Marcus Welby (the 1960's analog of Dr. House today) would be bewildered by a modern intensive care unit. Take a look at the picture above – notice the portable dialysis machine in the left foreground, the mechanical ventilator (breathing machine) to the patient’s right, and a dozen different intravenous bags, many infusing medications that can kill in seconds if given to the wrong patient or at the wrong dose. In this environment, the idea that one can keep patients safe simply by being really, really careful is tragically naïve.

Over the last decade, we have recognized that the traditional approach of attributing errors to individual fault ignores the fact that most mistakes are committed by hard working, well-trained individuals, and thus are unlikely to be prevented by admonishing people to be more careful, or shaming and suing them. Instead, the modern approach, known as systems thinking, holds that humans will inevitably err and that safety depends on creating systems that anticipate errors and either prevent or trap them before they cause harm (2). Such an approach has been the cornerstone of safety improvements in other high-risk industries such as aviation and nuclear power, but has been ignored in medicine until recently.

British psychologist James Reason’s “Swiss cheese model” (below) of accidents has been widely embraced as a mental model for system safety (3). This model, drawn from dozens of accident investigations in non-healthcare fields (such as space shuttle crashes and train derailments), emphasizes that in complex organizations, “sharp end” errors are rarely sufficient by themselves to cause terrible harm. Instead, everyday human glitches must penetrate multiple incomplete layers of protection (“holes in the Swiss cheese”) to cause a tragedy. The Swiss cheese model highlights the need to focus less on trying to perfect human behavior (a futile undertaking) and more on aiming to shrink the holes in the cheese and create multiple overlapping layers of protection to decrease the probability that the holes will ever align and let an error slip through.

Even armed with this understanding, improving complex systems is really tricky. Sure, we could create systems in our emergency departments with so many double and triple checks that no patient with a life-threatening infection ever received the wrong antibiotic. But if the system was too time-consuming, one of two things would happen: either some patients would die of their infection while we were traversing step 12, or the nurses and doctors would begin ignoring the checks in order to attend to their patient (“workarounds”) (4). This doesn’t mean that creating new layers of Swiss Cheese is wrong (it is often precisely right), but rather that this needs to be done thoughtfully, paying tremendous heed to how real people do their work. Some fixes that look really elegant in a conference room prove completely unworkable when exposed to the harsh light of actual clinical practice.

How to Improve Patient Safety

Redundancies, Standardization, and Forcing Functions

Drawing on the Swiss cheese model and the emphasis on systems thinking, the modern approach to patient safety emphasizes the need to shore up systems to prevent or catch errors. For example, errors in routine behaviors (“slips”) can best be prevented by building in redundancies and cross checks, in the form of checklists (“Did you remember to wash your hands and drape the patient correctly before inserting the large catheter?”), read-backs (“Let me read your order back to you”), and other standardized safety procedures (e.g., counting the sponges and needles in the operating room to be sure that none have been left inside the patient, signing a surgical site prior to an operation to be sure to operate on the correct limb, asking patients their name before giving them a medication or drawing their blood) (2).

Recently, we have expanded our notion of systems thinking by focusing on new strategies to decrease errors when healthcare workers interact with complex technologies and machines. Here, the use of forcing functions – engineering solutions that decrease the probability of human error – is particularly promising (5). The classic example of forcing functions comes from the automobile industry. In the 1960s and 1970s, auto manufacturers learned of dozens of tragic instances of parents backing their cars up over their children. How could that happen? Easy. Picture a mother, distracted and rushed. She gets behind the wheel, while her child goes behind the car to get into the back seat. Mom absently shifts into reverse with her foot off the brake. The result is tragedy.

The traditional medical approach to an error like this would be to focus on the driver’s mistake – suing her, admonishing her to be more careful, etc. In the automobile industry, however, smart engineers saw this as a person-machine problem, and set out to create a system that anticipated this error and blocked it from causing harm. And that’s precisely what they did, by redesigning automobile braking systems to make it literally impossible to place a car in reverse when the driver’s foot is off the brake. In healthcare, an (all-too-rare) example of a forcing function was changing the gas nozzles and connectors so that anesthesiologists cannot mistakenly hook up the wrong gas (such as nitrogen instead of oxygen) to a patient. Given the ever-increasing complexity of modern medicine, building in these forcing functions (in intravenous pumps, defibrillators, mechanical ventilators, and computerized order entry systems) will be crucial to safety.

The Role of Computerization

Healthcare is extraordinarily information intensive. A large integrated healthcare system that possesses even a rudimentary information technology backbone will process many times more computerized transactions each day than the NASDAQ stock exchange.

But volume only begins the challenge. Think about trying to keep track of one patient’s current illnesses, past medical history, medications, allergies, and test results. Now stir in the fact that the patient will be seen by many different doctors, often working for different healthcare systems.

OK, now assume that the patient is in a car collision and taken to an emergency room far away from home. Ideally, the caregivers would have access to the patient’s key information, all of which would be presented in the form best suited for each individual provider (the orthopedic surgeon needs different information than the internist, who needs different data than the physical therapist and the nurse). And, of course, all of this information would be perpetually updated, effortlessly and in real time.

Contrast this vision of information heaven to the sorry state of most of today’s healthcare system, in which information is stored on paper, often illegibly and in free text, therefore rendering it immune from aggregation and analysis. Information rarely follows patients as they move from system to system – in fact, studies have shown the following remarkable, and terrifying, problems, many of which could be solved through a healthcare information network:

•    Half of all medical patients experience at least one error in the post-hospital-discharge period (6).
•    Two-thirds of post-discharge hospital visits take place without a hospital discharge summary available to the primary care doctor (7).
•    In more than two-thirds of outpatient subspecialty referrals, the specialist received no information from the primary care physician to guide the consultation (8).
•    Then again, in one-quarter of the specialty consultations, the primary care physician received no information back from the consultant within a month (8).

You’ll be pleased to know that healthcare, which lagged behind virtually every other facet of society in computerization, is finally entering the modern age, in part driven by the hope that computers can prevent many types of medical mistakes. Many hospitals and doctors offices are implementing computer systems – including computerized provider order entry (CPOE), electronic health records (some of which will be accessible by patients themselves, making them personal health records), and a variety of other information technology solutions (bar-coding for medication administration, “smart” intravenous pumps). It’s about time.

Because the prescribing process is one of the most visible safety hazards, efforts to computerize this process have long been a focus of efforts to decrease medical mistakes. Studies have shown that well constructed CPOE systems can decrease medication error rates by more than 50%, by getting rid of illegible prescriptions, alerting doctors or pharmacists to potential drug interactions, and ensuring that doses and frequencies are in acceptable ranges (9). These systems will become even more useful as they build in layers of decision support – prompting a doctor to use a preferred antibiotic for a given infection, for example (maybe even helping to choose that antibiotic based on that individual hospital’s patterns of antibiotic sensitivity), or to check a certain blood test after starting a given medication (such as a test of kidney function when a medication that can harm the kidneys is begun). Moreover, just the simple act of installing a CPOE system usually prompts organizations to standardize chaotic processes – the equivalent of cleaning out your closet before moving.

Ultimately, CPOE systems, when combined with electronic health records that are interoperable (meaning that, even if hospitals or doctors’ offices are using different computer vendors, data are stored in standardized fashion, allowing systems to “talk to each other” – the same kind of functionality that allows you to withdraw money from an ATM machine that isn’t your bank) will provide new levels of safety. As more and more computer systems are bought and used, their functions and user-friendliness are sure to improve. Increasingly, patients will be able to access their own records, pointing out mistakes, scheduling visits, and becoming educated about their care in new ways. (In the past year or so, computer giants like Google and Microsoft have created tools to build personal health records, one measure of the level of interest in this area and the size of the potential market). Whatever the motivation, the potential for improved safety is immense.

That said, it is important to point out that one of the lessons of the first decade of the patient safety movement has been that computerization by itself – if unaccompanied by other types of systems solutions, an appropriately staffed and trained workforce, and a “safe culture” (see below) – cannot ensure safety itself. Moreover, computerization has created new classes of errors that will have to be studied and attacked (10). At least in the paper universe, errors tend to harm one patient at a time. In the computerized world, some errors can harm dozens, even hundreds of patients, if they are not discovered and rectified.

Communication and Culture

In addition to systems enhancements such as simplification, checklists, forcing functions, and computerization, the last decade has seen a growing recognition of the importance of communication and teamwork in healthcare. In the Joint Commission’s (the main accreditor of U.S. hospitals) database, communication failures underlie the majority of errors. We have learned that many of these failures owe to dysfunctional relationships between doctors and nurses, between trainees and their supervisors, or between patients and their providers. Unlike medicine, “safe industries” (such as aviation and nuclear power) have learned to “flatten hierarchies” – to create environments and cultures in which it is not only acceptable for someone lower on the organizational totem pole to raise a concern, it is seen as essential.

With the recognition of the importance of creating a safety culture, healthcare has tried to adapt a model from commercial aviation, a field that, until a generation ago, had a similarly rigid hierarchy as that of medicine. Tragically, several airline crashes in the 1970s and early 1980s (including the famous collision of two 747s at Tenerife airport in the Canary Islands in 1977) convinced the public and the profession that poor communication could kill. In the case of the Tenerife accident, analysis of the black box audiotapes made clear that, while other members of the cockpit crew were not 100 percent sure that the runway was clear on that foggy morning, the captain was emphatically sure that it was. In 1970s aviation culture, one didn’t question the captain’s judgment unless they were certain the captain was wrong, and so he was allowed to take off, plowing his KLM 747 into the fuselage of a Pam Am 747 still on the runway, killing 583 people (11).

Now, all commercial flight crews take crew resource management courses, in which they train for emergencies with other crewmembers, learning to flatten hierarchies that might stifle open communication, communicate clearly using standard language, and utilize checklists, pre-flight briefings, and other systemic approaches (2). Although the evidence that such interventions will improve patient safety is still mixed (and changing the culture of a busy medical or surgical floor is orders of magnitude more difficult than doing so in a hermetically sealed cockpit), there is considerable enthusiasm about them in safety circles (12). I personally believe that this kind of training will be critical to safety improvement efforts. 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) is at work (13).

Learning From Mistakes

Another key principle in patient safety 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 M&M conferences. I am privileged to edit a federal web site in which this case-based approach is used to analyze errors (AHRQ WebM&M) and wrote two books that do more of the same (a more lay-oriented book [11] and a recent professionally-oriented patient safety primer [2]). Whatever form the discussions take, there is a new emphasis on making sure that they involve the appropriate disciplines (doctors, nurses, hospital administrators), point out errors when they occurred, emphasize systems thinking and solutions, and are not punitive (14).

In addition to open discussions at conferences, organizations that are serious about improving patient safety build in ways to hear about unsafe conditions and errors from front-line staff, often via “incident reporting systems” (15) or through Patient Safety Walk Rounds by institutional leaders (the CEO might wander up to a hospital floor and asked the nurses, “If we were to harm or kill someone tonight from an error, what would be the cause?”) (16). Moreover, many organizations have embraced a technique drawn from engineering called Root Cause Analysis – a blame-free forum to analyze major errors in detail to discover all of the layers of Swiss cheese that need to be shored up (17).

Creating a Safe Workforce

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 physician board certification to poor patient outcomes (18-20). A variety of interventions have been implemented in response to this evidence. For example, the State of California now mandates ratios of nurses to patients no higher than 1:2 in intensive care units, and 1:5 on general medical floors. And the organization that accredits U.S. residency training programs now mandates that residents get at least one day off per week, and work no more than 80 hours per week.

Although the impact of these regulations on patient safety is still debated, the new focus on the workforce is overdue. In fact, it is evidence of a more holistic view of patient safety, one that acknowledges that the implementation of safer systems will not prevent mistakes if healthcare providers have unmanageable workloads, or are poorly trained or supervised.

Preventing Diagnostic Errors

The modern patient safety movement has emphasized medication errors, handoff errors, infections, and surgical errors – all areas amenable to technological (e.g., computerized order entry), procedural (e.g., double checks), and policy (e.g., “sign-your-site”) solutions. Diagnostic errors have been less well emphasized, in part because they are more difficult to measure and to fix.

Yet a number of studies have demonstrated that diagnostic errors are common, and that they can be deadly (21). For example, nearly 1 in 25 patients with heart attacks who come to emergency rooms are mistakenly sent home with a different (incorrect) diagnosis (22). And autopsy studies have shown high rates of missed diagnoses, rates that haven’t really budged over the past 30 years, notwithstanding fancy new technologies such as CAT and MRI scanners (23).

At first glance, diagnostic errors would seem to represent human failings – simple bad thinking. And it is true that, perhaps more than in any other area in the field of patient safety, the training and skills of the physician are perhaps the most important factor. However, in keeping with our modern understanding of patient safety, there are systems fixes, likely combining better information technology with more physician training on common cognitive mistakes, that can decrease the frequency and consequences of diagnostic errors.

Since the days of the early computers, many experts looked forward to the day when computers would augment, and then even replace, physicians as diagnosticians. But early clinical artificial intelligence programs – in which doctors entered key elements from the history, physical examination, and laboratory studies and the computer offered a list of possible diagnoses – were disappointing, for a couple of reasons. First, in part because of the nuances of clinical diagnoses (for example, there is about a 90% overlap between the signs and symptoms of the common flu and those of bubonic plague), the computer-generated diagnostic lists mixed plausible diagnoses with absurd ones. Secondly, data entry was prohibitively inefficient. The doctor who wanted to use a computerized clinical decision support system literally had to re-enter all the information into the computer after having already written or dictated a chart note.

Recent advances have generated new interest in computerized diagnostic decision support (24). For example, some programs now pull data directly from the electronic medical record, bypassing the need for redundant data entry. Most modern decision support programs not only suggest possible diagnoses but link to helpful resources and references. Some programs cull their searches from thousands of journals and textbooks, obviating the labor-intensive requirement of early artificial intelligence systems to program in the possible matches. Building on these advances, it is likely that we’ll soon see computerized decision support that draws its information directly from the electronic medical record and suggests possible diagnoses, based on information that is automatically updated when new research becomes available. By way of analogy, consider a driver in need of directions. In the distant past, drivers relied on maps that were sometimes wrong or out of date. Then the maps became better, but, before their trip, drivers needed to log onto their computer, type in the “from” and “to” addresses, and wait for the map to print before they brought it with them into the car. Ultimately, GPS systems were built into the car itself, making it relatively easy to input the necessary information and see the results on a screen. Finally, systems improved to the point of updating the maps in real time, showing traffic jams and construction zones and giving drivers the appropriate new instructions. Computerized diagnostic decision support is not there yet, but the direction is clear.

But better computerized support is only part of the solution. Improving diagnostic accuracy also requires that we truly understand how doctors think, and thus how they err. Studies over the past 40 years have shown that even well-trained and highly competent doctors are capable of diagnostic errors because of their cognitive shortcuts, often taken in the name of efficiency (25). For example, many errors occur when doctors are too quick to come to a decision (“this is definitely a case of pneumonia”), and then defend that judgment too vigorously even in the face of contradictory evidence. This is human nature, of course; we tend to see what we expect to see rather than than what’s actually in front of our eyes. By the way, did you notice the word “than” used twice in a row in the previous sentence? Even when we don’t intend to do it, our brains take shortcuts to get us to our goal – whether it’s finishing a sentence or diagnosing a patient’s chest pain.

This act of getting “stuck” on a diagnosis, known as the anchoring bias, is only one of the many pitfalls that underlie many diagnostic errors. Other common biases include the availability heuristic (relying too strongly on memorable past cases), blind obedience (deferring too strongly to an authority figure’s opinion or the results of a high-tech test), and the framing bias (getting too strongly wed to a certain way of thinking about a case without considering alternative ways of “framing” the situation).

How can these errors caused by faulty thinking be prevented? As always, the answer will come through a systems approach, but here this means the creation of better systems to train physicians to avoid common diagnostic mistakes. As Canadian physician and safety expert Pat Croskerry put it:

One uniquely distinguishing characteristic of those who make high-quality decisions is that they can largely free themselves from the common pitfalls to which novices are vulnerable. A rite of passage in all disciplines of medicine is learning about clinical pitfalls that have been identified by the discipline’s experts. This [says] in effect, “Here is a typical error that will be made, and here is how to avoid it” (26).

What Can Patients Do to Keep Themselves Safe?

Given the toll of medical errors in modern healthcare, many people are understandably interested in what they, or their loved ones, can do to protect themselves. I wish there was a simple answer.

Assuming that the patient is alert and engaged (or has a loved one who is willing and able to play an advocacy role), there are some errors that can be prevented (27). Doing so requires an exaggerated state of vigilance, and a willingness to speak up when the patient or family member notices something that seems amiss – whether it is a medication that seems wrong or a doctor who has failed to wash his hands before seeing a hospitalized patient. Of course, the same “authority gradients” that might prevent a nurse from speaking up to a doctor can easily prevent a family member from speaking up to a nurse or physician, and so efforts to dampen hierarchies and welcome patient and family input is critical here. When this kind of culture is present and patients and families do speak up, I have no doubt that this can prevent some mistakes.

But I believe that the impact on overall patient safety will be limited. Why? First of all, many patients receiving healthcare are sufficiently confused or anxious that they simply cannot be active participants in their own care. Significant numbers do not speak English (or whatever the dominant language of healthcare is) or have low health literacy (the ability to understand basic health information) (28,29). Even when patients are capable of understanding health information and advocating for themselves, there is so much activity that takes place behind the scenes that most patients will be far-from-protected from errors. If you doubt this, ask any physician whether he or she felt protected from errors if they had ever been hospitalized or had major surgery.

Finally, there is evidence that, in part because of the idea that patients can and should protect themselves from medical mistakes, many feel guilty when they or a loved one are a victim of a medical error (30). This, of course, is just plain wrong. Although it is reasonable for patients to do what they can to try to prevent medical mistakes, why should it be up to a patient to be sure that we give them the right medicine, make the correct diagnosis, or operate on the right limb? This problem is ours to fix, and fix it we must.

Challenges to Improving Safety

Reviewing the above list of potential approaches to improving safety makes clear one of the great challenges: in the absence of comparative evidence, and in light of the high cost of some of the interventions (improved staffing, computerized order entry, teamwork training), even institutions committed to safety will be understandably confused about which approach to emphasize (31,32). As in quality improvement, where institutions naturally focus on the practices that are measured, publicly reported, and compensated, in safety, institutions and physician focus first on areas subject to regulation, and those with multiple potential benefits (such as computerization, which might improve both safety and efficiency). Since improving culture is hard to measure and to regulate, I worry that it will be perpetually shuffled to the bottom of the deck, notwithstanding its importance to safety.

Another set of challenges relate to workarounds, which I described earlier (4). For example, some organizations that have adopted the use of bar-coded wrist bands to improve medication safety (the nurse scans the patient’s wristband to ensure that the barcode matches that of the intended medication) were shocked to learn that some nurses took a handful of wristbands and scanned them all before going on their medication rounds, thus thwarting the safety system. When workarounds are discovered, it is important for safety leaders to ask whether the providers are forced into them to get their work done. If so, the right solution is to improve the process rather than blame the nurse or doctor.

A related issue is that of squeezing balloons – fixing one problem and seeing another emerge. Improving safety in organizations as complex as hospitals and clinics involves disturbances of existing work patterns that often create unintended, and negative, consequences. For example, commonsense regulations to decrease the number of consecutive hours that residents can work have probably prevented errors due to fatigue, but have undoubtedly increased errors due to poor handoffs. Similarly, computerization can prevent errors from illegible handwriting, but might create new errors as doctors pick the wrong medication (e.g., the chemotherapy penicillamine instead of the antibiotic penicillin) from a computer-generated list. As with workarounds, the key here is to scrutinize the implementation of even “easy” fixes for unanticipated consequences. For example, in the case of shortening residency duty hours, the solution was not to scrap the shift regulations, but rather to implement systems to improve physician-to-physician handoffs.

Another challenge is the general absence of a business case for safety. In most of American healthcare, an unsafe hospital is paid the same as a safe one, and patients have little ability to tell the difference. Recent efforts to report rates of certain kinds of errors (hospital-acquired infections and wrong-site surgery, for example) (2), and even to not reimburse hospitals for the cost of these adverse events, represent early initiatives to put “skin in the game” when it comes to safety (33). Since many safety fixes are extraordinarily expensive (fully computerizing a 500-bed hospital may run upwards of $100 million, and computerizing a doctor’s office may cost more than $50,000 per doctor), it will be important to create a business case for safety – appeals to professionalism and ethics, while important, are clearly insufficient.

Finally, improving patient safety is just plain hard. Although analogies from other industries can help (and we are learning from the lessons of commercial airlines, Toyota, and FedEx), caring for a desperately ill patient is far more complex to flying a plane safely from JFK to O’Hare, or building a car that won’t break down after 60,000 miles. We have made significant progress in the decade since the publication of the IOM Report on medical errors catalyzed the modern patient safety movement. But there is much more to do, and much more to come.

References

1.    Kohn L, Corrigan J, Donaldson M, eds. To Err is Human: Building a Safer Health System. Washington D.C.: Committee on Quality of Health Care in America, Institute of Medicine. National Academy Press, 2000.
2.    Wachter RM. Understanding Patient Safety. New York: McGraw-Hill, 2008.
3.    Reason J. Human Error. Cambridge, UK: Cambridge University Press, 1990.
4.    Spear SJ, Schmidhofer M. Ambiguity and workarounds as contributors to medical error. Ann Intern Med 2005; 142:627-30.
5.    Gosbee J. Human factors engineering and patient safety. Qual Saf Health Care 2002; 11:352-4.
6.    Moore C, Wisnivesky J, Williams S, McGinn T. Medical errors related to discontinuity of care from an inpatient to an outpatient setting. J Gen Intern Med 2003; 18:646-51.
7.    Kripalani S, LeFevre F, Phillips CO, Williams MV, Basaviah P, Baker DW. Deficits in communication and information transfer between hospital-based and primary care physicians: implications for patient safety and continuity of care. JAMA 2007; 297:831-841.
8.    Gandhi TK, Sittig DF, Franklin M, Sussman AJ, Fairchild DG, Bates DW. Communication breakdown in the outpatient referral process. J Gen Intern Med 2000; 15:626-31.
9.    Bates DW, Gawande AA. Improving safety with information technology. N Engl J Med 2003; 348:2526-34.
10.    Ash JS, Sittig DF, Poon EG, Guappone K, Campbell E, Dykstra RH. The extent and importance of unintended consequences related to computerized provider order entry. J Am Med Inform Assoc 2007; 14:415-23.
11.    Wachter RM, Shojania KG. Internal Bleeding: The Truth Behind America’s Terrifying Epidemic of Medical Mistakes. New York: Rugged Land, 2004.
12.    Salas E, Wilson KA, Burke CS, Wightman DC. Does crew resource management training work? An update, an extension, and some critical needs. Hum Factors 2006; 48:392-412.
13.    Pronovost P, Sexton JB. Assessing safety culture: guidelines and recommendations. Qual Saf Health Care 2005; 14:231-3.
14.    Pierluissi E, Fischer MA, Campbell AR, Landefeld CS. Discussion of medical errors in morbidity and mortality conferences. JAMA 2003; 290:2838-42.
15.    Vincent C. Understanding and responding to adverse events. N Engl J Med 2003; 348:1051-6.
16.    Frankel A, Graydon-Baker E, Neppl C, Gustafson M, Gandhi TK. Patient safety leadership walkrounds. Jt Comm Qual Improv 2003; 29:16-26.
17.    Wu AW, Lipshutz AKM, Pronovost PJ. Effectiveness and efficiency of root cause analysis in medicine. JAMA 2008; 299:685-7.
18.    Aiken LH, Clarke SP, Sloane DM, Sochalski J, Silber JH. Hospital nurse staffing and patient mortality, nurse burnout, and job dissatisfaction. JAMA 2002; 2898:1987-93.
19.    Landrigan CP, Rothschild JM, Cronin JW, et al. Effect of reducing interns’ work hours on serious medical errors in intensive care units. N Engl J Med 2004; 351:1838-48.
20.    Brennan TA, Horwitz RI, Duffy FD, Cassel CK, Goode LD, Lipner RS. The role of physician specialty board certification status in the quality movement. JAMA 2004; 292:1038-43.
21.    Graber ML, Franklin N, Gordon R. Diagnostic error in internal medicine. Arch Intern Med 2005; 165:1493-9.
22.    Goldman L, Kirtane AJ. Triage of patients with acute chest pain and possible cardiac ischemia: the elusive search for diagnostic perfection. Ann Intern Med 2003; 139:987-95.
23.    Shojania KG, Burton EC, McDonald KM, Goldman L. Changes in rates of autopsy-detected diagnostic errors over time: a systematic review. JAMA 2003; 289:2849-56.
24.    Graber ML, Mathew A. Performance of a web-based clinical diagnosis support system for internists. J Gen Intern Med 2008; 23 (suppl1): 37-40.
25.    Redelmeier DA. Improving patient care. The cognitive psychology of missed diagnoses. Ann Intern Med 2005; 142:115-20.
26.    Croskerry P. Achieving quality in clinical decision making: cognitive strategies and detection of bias. Acad Emerg Med 2002; 9:1184-1204.
27.    Vincent CA, Coulter A. Patient safety: what about the patient? Qual Saf Health Care 2002; 11:76-80.
28.    Flores G. Language barriers to health care in the United States. N Engl J Med 2006; 355:229-31.
29.    Institute of Medicine. Health Literacy: A Prescription to End Confusion. Washington, DC: National Academy Press, 2004.
30.    Delbanco T, Bell SK. Guilty, afraid, and alone – struggling with medical error. N Engl J Med 2007; 357:1682-3.
31.    Wachter RM. The end of the beginning: patient safety five years after 'To Err is Human'. Health Aff (Millwood) 2004 Jul-Dec;Suppl Web Exclusives:W4-534-45.
32.    Ranji SR, Shojania KG. Implementing patient safety interventions in your hospital: what to try and what to avoid. Med Clin North Am 2008; 92:275-93.
33.    Wachter RM, Foster NE, Dudley RA. Medicare’s decision to withhold payment for hospital errors: the devil is in the details. Jt Comm J Qual Patient Saf 2008; 34:116-23.

Suggested Readings

Leape LL. Error in medicine. JAMA 1994; 272:1851-7.
The article that first introduced concepts of systems thinking to health care practice.

Leape LL, Berwick DM. Five years after ‘To Err is Human’: What have we learned? JAMA 2005; 293-2384-90.
A thoughtful review of progress in safety and quality five years after the blockbuster IOM report.

Pronovost PJ, Nolan T, Zeger S, Miller M, Rubin H. How can clinicians measure safety and quality in acute care? Lancet 2004; 363:1061-7.
A good review of some of the key measurement issues.

Reason J. Human Error. Cambridge, UK: Cambridge University Press, 1990.
A classic work on how errors occur in complex enterprises.

Wachter RM, Shojania KG. Internal Bleeding: The Truth Behind America’s Terrifying Epidemic of Medical Mistakes. New York: Rugged Land, 2004.
Uses dramatic cases of medical errors to illustrate key concepts in patient safety.

Wachter RM. Understanding Patient Safety. New York: McGraw-Hill, 2008.
Bestselling primer on patient safety and medical errors.

Websites

AHRQ WebM&M (editor, Robert M. Wachter)
A monthly case-based journal included user-reported cases of medical errors, expert commentaries, and interviews with safety newsmakers. For the AHRQ WebM&M patient safety glossary, click here.

AHRQ Patient Safety Network (editor, Robert M. Wachter)
The main patient safety portal, combining weekly updates of patient safety literature, resources, conferences, tools, and more with the world’s most robust and searchable patient safety library. For the Agency for Healthcare Research and Quality’s (AHRQ) main website, click here; for their Patient Safety Indicators (PSIs), click here.

Institute for Healthcare Improvement
An important organization working to improve safety and quality.

The Joint Commission
The primary accreditor of U.S. hospitals. For the Joint Commission’s National Patient Safety Goals, click here.

National Quality Forum
Public-private partnership designed to endorse quality and safety standards. For the NQF’s list of “Never Events,” click here.