Time to Tackle the Tough Issues in Patient Safety

Evidence that medication errors occur commonly in all patient care settings continues to mount. Kozer et al¹ demonstrate in this issue of Pediatrics that the pediatric emergency department (ED) is no exception, with errors noted in 10.1% of records during a 12-day study period. As the authors point out, this error rate almost certainly is an underestimate of the magnitude of the problem, because the study was a retrospective chart review. For example, errors intercepted before an order actually was written and errors occurring while dispensing or administering drugs would not have been detected by the methodology used by the investigators. Although only 2 errors were “severe” (had the potential to cause death or decrease chances of successful treatment of a life-threatening condition), this translates into >100 “severe” mistakes annually in just 1 ED. Moreover, nearly half of the errors were “significant” and could have had important clinical consequences. And these are just medication errors. They do not include other important types of errors and near misses, such as diagnostic and procedure-related errors that undoubtedly occur in every ED every day.

Although troubling, these findings are not surprising. The ED is a particularly error-prone environment. There has been a striking increase in acuity and volume in many EDs in recent years, accompanied by increased emphasis on efficient, rapid throughput. Patients present with acute, unanticipated, sometimes baffling illness. Generally, the treating physicians have never seen the patient waiting for them in an ED examination room, and previous medical records may not be available. Cultural, socioeconomic, and linguistic barriers may impede accurate transfer of information, such as medication allergy or previous conditions that might affect drug disposition or efficacy. The patient may be taking prescription drugs, over-the-counter remedies, or complementary therapies that have important interactions with the drugs the ED physician will prescribe.

Pediatric patients pose special challenges to ED caregivers.^2,3 Pediatric medication doses must be adjusted for the child’s weight and stage of development. Many medications are supplied to the pharmacy in standard dosage forms that must be diluted or subdivided for administration to children. Young children are unable to warn care providers that they are about to receive the wrong medication or are experiencing the onset of an adverse reaction to drug administration. Infants have limited physiologic reserves to buffer mistakes in dosing, dispensing, or administering medications.

The experience and training of clinical personnel in EDs caring for children also may influence the risk of committing or failing to detect errors. Residents or fellows with very limited pediatric expertise are responsible for the bulk of direct patient care in some EDs. Even the most prestigious pediatric hospitals, such as the institution studied by Kozer et al, are not immune to this problem. Apparently, the majority of residents who see patients in this hospital’s ED have had no previous pediatric training. Moreover, many are in just their first year of training. In all likelihood, pediatric dosing is unfamiliar to such residents, who would have had little experience with the weight- and age-based dosing required for infants and children. Rigorous systems for verifying the competence of trainees in prescribing medications accurately are not available in most hospitals, increasing the onus on staff physicians and nurses to detect mistakes.

Individual lack of training and experience is only part of the problem. It also is important to emphasize that trainees in the ED—indeed throughout the hospital—usually have very short rotations on any one service. Therefore, it is virtually impossible for trainees to develop a sense of team identity, understand the systems of care on the ward or unit, become comfortable with interdisciplinary communications, or practice coordinated response to emergencies. Kozer et al note that >80 physicians were involved in ED care on just the 12 study days—a staggering number that virtually precludes formation of smoothly functioning, experienced teams.

Environmental and organizational factors also may increase the risk of errors. Fatigue, stress, anxiety, fear of blame, ergonomically substandard working environments, distractions, noise, poor lighting, lack of standardization of equipment or location of critical supplies—these and many other factors confront the trainee with substantial barriers to “patient safety.” The availability and quality of supervision is particularly important for preventing and catching mistakes by trainees. Prompt, direct faculty supervision is available in most EDs, and close supervision of trainees is fast becoming the expected standard of care. However, there is great variability in the style and quality of supervision. Real or perceived authority gradients (eg, seniority-based, gender-specific, cultural) are particularly problematic because they inhibit trainees from seeking help and dampen enthusiasm for open dialogue and feedback.

Undoubtedly, none of the physicians who made the errors enumerated in the Kozer report intended to make a mistake, yet many errors occurred. What can be done to help physicians and nurses avoid medication errors and other medical mistakes? Kozer et al propose a few widely advocated potential systems improvements. For example, computerized physician order entry (CPOE) has been shown to be an effective technology for reducing prescribing errors.^4–7 Although most studies have been performed in academic centers with well-developed, home-grown systems, CPOE clearly has enormous potential for improving patient safety. At the most basic level, CPOE ensures that orders are complete, legible, and in a standardized format. When decision support is incorporated, CPOE can guide drug dosage, frequency, and choice of route or administration, as well as perform checks for drug allergy and drug-drug interactions. Handheld devices have considerable promise in hectic environments such as EDs and clinics, where it may be difficult to access a computer terminal while providing care. Robots have promise in reducing errors in drug dispensing, smart intravenous pumps may reduce administration errors, and bar coding can improve the reliability of the entire medication system.

Low-tech approaches also can help physicians and nurses make correct medication choices. Practice guidelines and protocols with evidence-based order sets can provide physicians with real-time guidance for treating some high-volume conditions, such as asthma in children or myocardial infarction in adults. Participation of clinical pharmacists on patient care rounds also has shown encouraging results,⁸ although it has not been tested in the ED setting, where the absence of formal rounds makes it less efficient for pharmacists to offer decision support. Pharmacists could still assist individual physicians faced with complex decisions in critically ill patients; optimally such expertise should be available on all shifts. Pharmacists also could facilitate drug dispensing and monitor drug administration. Unit dosing of medications prepared by the pharmacy takes dilution and preparation of doses out of the hands of nurses, who have less training than pharmacy personnel in this important aspect of medication dispensing. Formatted templates or check lists, either paper or electronic, can remind clinicians to ask patients for critical information such as drug allergies and medication regimens. Some investigators have suggested that the patients or parents themselves can record such information on handheld devices or laptop computers so that the physician has important information at the fingertips when the patient is examined.

These systems improvements are likely to have a major impact on patient safety, albeit at considerable cost. Regulation, legislation, and pressure from organizations such as the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) will be powerful incentives for hospitals to finance these improvements. The Leapfrog Group, a consortium of Fortune 500 companies, has been particularly aggressive in pushing hospitals to adopt CPOE, an expensive intervention that can cost millions of dollars if properly integrated into an institutional information system.⁹ It will be difficult for hospitals to make such large capital investments in this era of fiscal constraint in health care. Short of assistance from the government, a long-range time horizon will be necessary for institutions to visualize the cost-benefit of these expenditures. Moreover, technology is evolving rapidly as the computer and medical equipment industries try to respond to public pressure for safer health care delivery systems. The computer system or automated intravenous fluid pump that a hospital purchases today may be outdated in a few years.

Chief executives and boards of trustees that have taken the financial plunge and hired clinical pharmacists or invested in costly technology may be surprised to discover that they have only begun the complex and difficult journey toward improved patient safety. This journey requires not only ongoing refinement of technologic interventions, but also a deep, comprehensive commitment to understanding the human factors that contribute to error. Kozer et al briefly mention some of the human factors that probably contributed to the medication errors that they observed. Although a comprehensive analysis of the human factors that lay at the root of these errors was beyond the scope of their study, they nevertheless touch on a critical issue that until recently received insufficient attention in the public dialogue on patient safety.

What, then, is human factors research? A useful definition of human factors research is “the study of the interrelationships between humans, the tools they use, and the environment in which they live and work.”¹⁰ Many errors occur when people subconsciously misapply the preprogrammed, virtually automatic actions that they rely on to negotiate common daily tasks successfully. Other errors have their origin in faulty application of ingrained mental models that individuals use to evaluate complex situations. Still others derive from failed efforts to apply incomplete knowledge or to call on remote memory when a tired, stressed, distracted individual attempts to deal with a crisis or complex decision. Faulty human communication is at the root of many medical errors. Medicine, with the notable exception of anesthesiology, has only begun to tap the riches of human factors research, a field populated by scientists such as sociologists, psychologists, anthropologists, and industrial engineers. Other industries, particularly aviation, have been far more attentive to human factors principles and have achieved remarkable improvements in safety (although attention to human factors principles has been notably absent in aviation security).

Reason¹¹ and Vincent et al¹² have been particularly vocal and effective advocates of the systematic application of human factors principles to medicine. Vincent’s paradigm of errors as “organizational accidents” is particularly compelling. He stresses the importance of not focusing entirely on active failures by health care professionals at the “sharp end” of the system, but rather looking carefully for “latent errors” that are upstream from the event itself and are not under the direct control of the person making the error. Often, multiple defenses, buffers, and checks are inserted into the patient care system in an effort to prevent “accidents,” but because the true character of latent problems may not have been identified, failures still occur. For example, multiple problems with working conditions may lurk behind a fatal drug administration error during a cardiac arrest. The proximate cause of the active failure may have been a nurse’s slip, memory lapse, or mistake based on lack of knowledge. But the error may have been fostered by heavy workload, fatigue, stress, inadequate supervision, faulty communication, unfamiliar or poorly maintained equipment, poorly designed space, unsatisfactory teamwork, lack of training in simulated emergencies, or failure to verify the nurse’s knowledge and competency, to name just a few potential problems.

Digging still deeper, these issues may not have been addressed because of basic problems with management decisions or organizational processes. It certainly would be fair to ask hospital leadership whether the institutional culture is conducive to risk identification and reduction, or if there are barriers to communication of potential risk factors and events.

Does leadership make it clear that prevention of adverse outcomes is a high priority? Most importantly, when critical institutional decisions are being made, perhaps regarding staffing, capital expenditures, facility modernization, or the budget for environmental upkeep and equipment maintenance, is the potential impact on patient safety always part of the equation? Such a detailed diagnostic inquiry into mistakes and preventable adverse events—a so-called “root cause analysis” of “critical incidents” is now required by the JCAHO. All too often, however, root cause analysis as currently practiced falls short of asking and answering the toughest questions.

It is important to recognize that root cause analyses are by definition reactive, because they explore the reasons for errors that already have happened. They are extremely time consuming and invariably are selective. The errors that are investigated may or may not reflect fundamental problems in key hospital systems. A more proactive approach to preventing errors is needed. Systems engineering principles offer important lessons for health care. Failure mode effects analysis (FMEA) is a powerful approach in which error-prone points in the system are identified, failures are anticipated, and the potential for error is designed out. In brief, a process flow diagram of the system is constructed and the system is scrutinized for points at which an error is likely to occur, is likely to have serious sequelae, and is unlikely to be detected. Interventions to lower the risk of error are implemented, and the vulnerable points in the system are monitored closely. The Food and Drug Administration uses a similar systems-oriented approach to improving food safety—Hazard Analysis Critical Control Point (HACCP). HACCP was pioneered by Pillsbury (Minneapolis, MN) in cooperation with NASA, the US Army, and the US Air Force Space Laboratory in the 1960s to ensure delivery of safe food to astronauts. It forms the basis for good manufacturing practice standards for pharmaceuticals and is an integral component of blood banking quality standards in the United States. JCAHO now requires hospitals to perform FMEA, but many institutions lack the resources or expertise to mount a meaningful effort at the present time. The Veterans Administration National Center has been particularly innovative in adapting FMEA and HACCP to health care and has started to pilot test its program in the Veterans Administration hospital system.¹³

If the task ahead seems daunting, it is. It will be expensive, not only for hospitals, but for society as a whole. But health care’s most important customers—our patients and parents—are demanding safer care. As we prepare to address the public’s legitimate demands, we should not forget our internal customers, the hospital’s own staff. Hospitals go to great lengths to hire the most talented, best trained, and most highly motivated personnel they can find. All too often, the staff’s best efforts are foiled by imperfect, complex, error-prone hospital systems. It is time to treat hardworking hospital personnel, from the most exalted cardiac surgeon to the person who empties the sharps container in the ED, with the respect they deserve. From a safety perspective, true respect involves more than just creating a “safety culture” in which individuals who report errors and accidents are not blamed or punished. Hospital organizations need to work much harder to make the comprehensive systems improvements that will be required to protect staff members from their own human fallibility.

• Copyright © 2002 by the American Academy of Pediatrics