Nursing: Inseparably Linked to Patient Safety
Over the last two decades, substantial changes have been made in the organization and delivery of health care. These fast-paced changes have resulted from multiple, concurrent events, including (1) major modifications in the ways in which government and private health insurance programs reimburse health care providers (including hospitals, nursing homes, home health care agencies, and individual practitioners); (2) cost-containment efforts of health care organizations (HCOs) in response to these changes in reimbursement; (3) growth in and increased demand for new health care technologies; and (4) changes in the health care workforce. HCOs have responded in a variety of ways that, in turn, have affected the work and work environment of nurses. Some of these changes have resulted, for example, in greater numbers of more acutely ill and technology-dependent patients being assigned to individual nurses; changes in how licensed and unlicensed nursing staff are deployed; and a growing number of competing demands on nurses’ time, such as increased paperwork and documentation requirements. Many individuals and organizations have expressed concern that these and other changes have adversely affected nurses’ ability to provide safe patient care (Aiken et al., 2001a; Service Employees International Union, 2001; Shindul-Rothschild et al., 1996).
In response to such concerns, the U.S. Department of Health and Human Services’ Agency for Healthcare Research and Quality (AHRQ) asked the Institute of Medicine (IOM) to conduct a study to identify key aspects of the work environment for nurses that likely have an impact on patient safety, and to identify potential improvements in health care working conditions that would likely increase patient safety. AHRQ further directed
that the study be conducted “in the context of current policy debates on regulation of nursing work hours and nursing workload … [and] cover such topics as: extended work hours and fatigue, including mandatory overtime; workload issues, including state regulation of nurse-to-bed ratios; workplace environmental issues, including poorly designed care processes; … workplace systems, including reliance on memory and lack of support systems for decision-making; and workplace communication, including social, physical, and other barriers to effective communication among care team members.” The Committee on the Work Environment for Nurses and Patient Safety was formed to carry out this study. This report presents the study results.
In responding to its charge, the committee reviewed and built upon recommendations for increasing patient safety contained in two earlier IOM reports—To Err Is Human: Building a Safer Health System (IOM, 2000) and Crossing the Quality Chasm: A New Health System for the 21st Century (IOM, 2001). In this introductory chapter, we first summarize and update the evidence presented in To Err Is Human about the magnitude and etiology of health care errors affecting patient safety. We then present evidence of the key role played by nurses in patient care and safety, and briefly describe some of the characteristics of the current health care delivery system that shape the work and work environment of nurses, particularly in in-patient facilities. Evidence is then presented showing that nurses are not immune to the problems that plague health care delivery in the United States—problems that foster the occurrence of errors in which all health care providers, not just nurses, are involved. The chapter ends with a call for a substantial transformation in the work environment of nurses to better safeguard patients.
THOUSANDS OF HEALTH CARE ERRORS
I was a “new” nurse. I’d been practicing only a few months when I was assigned an elderly patient who was scheduled for abdominal surgery that morning and needed a urinary catheter inserted. I knew about, but hadn’t performed, this procedure before, and neither had the other nurses on the floor—we all were new graduates and fairly inexperienced. I asked my head nurse if she would supervise me while I placed the catheter, but she was late for a meeting and assured me that it wasn’t difficult and I would be fine.
I went to get the supplies I needed, but there were no prepackaged catheterization trays on the floor. I ran the stairs to the floors above and below me, but they were out, too. As I passed the nursing station, the clerk called out to me that the OR [operating room] wanted to know where the patient was. I began to round up the materials needed on an item-by-item basis.
I got a sterile prep tray (the last one), sterile catheter and gloves, antiseptics for cleansing, and drainage bag. I opened the sterile prep tray, prepared the patient, put on the sterile gloves, and realized I hadn’t opened the bottles of antiseptic before putting on the sterile gloves and that the routine sterile prep tray didn’t contain what I had expected. There were no more gloves in the patient’s room. I went to get more, cautioning the patient to not move, and leaving my sterile field unattended.
As I passed the nurses’ station, the clerk again called out: “The OR called again and they are really angry and want to know what’s keeping your patient. You are backing up the entire OR schedule!” I got the gloves and with trembling hands, uncertainty about the sterility of my “sterile field,” and not the best of technique, inserted the catheter.
A day or two later, I was charting on my patients and seated next to the patient’s resident, who exclaimed, “Mrs. X has the worst UTI [urinary tract infection] I’ve ever seen!”
I didn’t say anything. I was ashamed and afraid, and besides, the resident was already writing an order for antibiotics. There was nothing more to be done. What would be gained if I told anyone?
What happened to Mrs. X in the above (true) incident was a mistake—an error. Her urinary tract infection was an adverse event likely resulting from (at least in the opinion of the nurse performing the procedure) that error. While this error involved an inexperienced nurse, errors are committed by individuals with all levels of experience.
To Err Is Human helped the United States (and other countries) come to a better understanding of the likely hundreds of thousands of health care errors and adverse events that occur in the United States every year in which nurses, physicians, pharmacists, dentists, nurse aides, and assistants—in fact, all health care providers—are involved. First, To Err Is Human presented the vocabulary necessary to begin to better understand the problem:
Errors are failures of planned actions to be completed as intended, or the use of wrong plans to achieve what is intended.
Adverse events are injuries caused by medical intervention, as opposed to the health condition of a patient. A large proportion of adverse events are the result of errors. When the adverse event is the result of an error, it is considered a preventable adverse event.
Sometimes an error, such as giving a patient the wrong medication, may lead to no detectable adverse event. Other errors can temporarily or permanently harm the health of the patient or cause the person’s death. In
the incident described above, the catheterization of the patient was not completed as intended. The process was replete with errors, including the nurse’s technique in catheterization, the nurse manager’s assumption that the new nurse could perform the procedure safely, and the supply department’s failure to stock prepackaged catheterization trays on the floor. The patient received an injury—a urinary tract infection—an adverse event that was likely preventable. The infection likely caused discomfort and possibly even pain. It required the administration of antibiotics, which carries the risk of side effects, adverse reactions, and medication errors. Moreover, the administration of antibiotics may have prolonged the patient’s stay in the hospital. Urinary tract infections can also lead to more serious kidney infections and, if undetected or occurring in a patient with a weakened immune system, can lead to sepsis (an infection in the blood), which can cause death.
To Err Is Human also calls attention to the magnitude of adverse events that occur every day to patients in the hospital. The report estimates that adverse events (involving all health care providers) occur in 2.9 to 3.7 percent of acute care hospitalizations, and that approximately half of these events are likely due to errors (i.e., preventable adverse events). The report further estimates that each year, between 44,000 and 98,000 hospitalized Americans die as a result of medical errors—more than die from motor vehicle accidents, breast cancer, or AIDS. Indeed, To Err Is Human presents evidence that these numbers are likely underestimates of the numbers of people injured by errors in health care. These numbers also do not include persons injured as a result of medical errors in nursing homes, home health care, and other health care settings. Earlier studies of medical errors have indicated similarly high rates of adverse events (Steel et al., 1981).
The IOM’s estimates of high rates of errors have been reaffirmed more recently by two different sources—practicing physicians and the public at large. In a 2002 national survey of practicing physicians and the American public, 35 percent of surveyed U.S. physicians and 42 percent of the public reported experiencing an error either in their own care or in that of a family member. Moreover, 18 percent of the physicians and 24 percent of the public reported an error that had caused serious health consequences, including death (reported by 7 percent of physicians and 10 percent of the public), long-term disability (6 percent and 11 percent, respectively), and severe pain (11 percent and 16 percent). These were not the biased perceptions of distraught family members. About one-third of the respondents who reported experience with an error stated that the health professionals involved had told them about the error or apologized to them (Blendon et al., 2002).
The United States is not alone in its high rate of health care errors; research in other countries also has found high error rates. It is estimated that 10 percent of hospital patients in Great Britain and 16.6 percent of
such patients in Australia experience an adverse event (WHO, 2002). No one receiving health care—young or old; severely or slightly ill; patients in hospitals, in nursing homes, or in their doctors’ offices; wealthy, middle class, poor, or near poor; those receiving health insurance through Medicare, Medicaid, or private health insurance—is immune to health care errors and adverse events.
Most important, To Err Is Human has helped concerned individuals and organizations better understand the reasons behind this profusion of health care errors and how it can best be addressed.
WHY HEALTH CARE ERRORS OCCUR
Two very different views are often held about why errors in health care, like errors in other industries, occur (Reason, 2000).
The first view holds individuals as primarily responsible for any error or unsafe action. Unsafe acts are viewed as arising principally from an individual’s faulty mental processes or weaknesses of character, such as forgetfulness, inattention, poor motivation, carelessness, negligence, and recklessness. Bad outcomes are viewed largely as the result of bad behavior by people, behavior that should be corrected through workplace policies and procedures, safety campaigns, disciplinary measures, the threat of litigation, retraining, and “naming, blaming, and shaming.” In this view, when workplace errors occur, the person most directly involved in the work at the time the error is thought to have taken place (often known as “the last person to touch the patient”) might well be blamed. In the above example, the nurse inserting the urinary catheter would be blamed for causing the urinary tract infection. After all, she inserted the catheter—a highly likely candidate for the introduction of bacteria causing the infection.
Such assignment of blame is the approach historically used in health care, as has been the case in other industries, and is deeply rooted in Western civilization (Reason, 2000). The 2002 survey of practicing physicians and the public cited earlier revealed that the public believes individuals, and not organizations, should be held responsible for errors with serious consequences through lawsuits, fines, and suspension of their professional licenses. Similarly, the majority of physicians surveyed believe that individual health professionals, as opposed to health care institutions, are more likely responsible for preventable medical errors (Blendon et al., 2002). This human tendency to blame bad outcomes on an individual’s personal inadequacies rather than on situational factors beyond the individual’s control (identified in social psychology as “fundamental attribution error”) is a serious obstacle to preventing or mitigating the inevitable errors that occur in complex organizations such as those delivering health care (Reason, 1990). It fails to acknowledge that, indeed, “to err is human.”
The contrasting systems view of errors and error prevention is based on research findings from a variety of fields, including studies of accidents and breaches of safety in a variety of industries, studies of “high-reliability organizations,” and research into effective organizational and managerial practices. In all of this work, the interdependent interaction of multiple human and nonhuman (equipment, technologies, policies, and procedures) elements of any effort to achieve a stated purpose is regarded as a “production process” or “system.” These interrelated human and nonhuman system elements are required to operate in synchrony if a given goal is to be achieved. As the elements of the production process or system are changed, the likelihood of error also changes. This research has revealed that errors typically result from problems within the system in which people work—not from poor individual worker performance—and typically originate in multiple areas within and external to an organization. Error results when these multiple problems converge and impair an organization’s performance (Perrow, 1984; Reason, 2000). Not surprisingly, errors increasingly are attributed to the hyper-complex organizations that emerged in the last half of the twentieth century in response to technological and social changes (Perrow, 1984).
A fundamental principle of the systems approach to error reduction is the recognition that all humans make mistakes and that “errors are to be expected, even in the best organizations” (Reason, 2000:768). To Err Is Human endorses the systems approach to understanding and reducing errors and notes that failures in large systems, such as hospitals or their various patient care units, nursing homes, or ambulatory practice sites, are most often due to unanticipated events or factors occurring within multiple parts of the system. In most cases, the accumulation of these factors, as opposed to the actions of a single individual, is what leads to an error or accident. In the above example, these multiple factors include the inexperience of the nurse; the lack of available supervision; the unavailability of the tools needed to perform the task; and the nurse’s possible perception of her lack of authority to call attention to and change the unsafe situation by, for example, sending the patient to the OR without a catheter and directing OR staff to catheterize the patient. Addressing any one of these factors might have prevented the urinary tract infection. Blaming the individual nurse would not change these factors and would not result in increased safety for the next patient in need of catheterization on the nursing unit. As Reason notes, when an error occurs, the question should not be “Who is at fault?” but rather “Why did our defenses fail?” (Reason, 2000).
At the same time, even though errors are understood to be the result of multiple factors within a system, the human component of systems in all industries has been identified as one of the largest contributors to the occurrence of accidents. Reason explains that since people design, manufacture, operate, maintain, and manage complex technological systems, it is hardly
surprising that human decisions and actions are implicated in all organizational accidents. Human beings contribute to the commission of errors in two ways: through the commission of active failures and the creation of latent conditions1 (Reason, 1997).
Active failures occur at the level of the front-line worker (e.g., airplane pilots; control room operators; health care workers, such as nurses, physicians, and pharmacists; and other operators of technology interfacing with people). Such failures are sometimes called the “sharp end” of an error. The types of errors committed by front-line workers involve such phenomena as lapses in memory, misreading or misinterpretation of written data, incorrect performance of a routine activity as a result of a distraction or interruption, or simply human variations in fine motor skills. The consequences of these actions are experienced almost immediately. In the above example, the nurse is the front-line worker at the sharp end of the work process. Her insertion of the catheter using poor processes and tools represents an active failure.
In contrast, latent conditions are factors in the production process or system that are not under the direct control of front-line workers. These factors include poor design of work or equipment, inadequate training, gaps in supervision, insufficient supply of equipment to perform work, undetected manufacturing defects or faulty maintenance, inadequate personnel deployment, and poorly structured operations. They arise from strategic and other top-level decisions made by entities at the “blunt end” of an organization or production system, such as government regulators, manufacturers, system designers, and high-level managers and decision makers.
The error described above resulted from multiple latent conditions. First, the new nurse had not had practical experience in either her nursing school or her workplace in the performance of this specific task. A mechanism for identifying the presence or absence of core nursing skill competencies would have detected this lack of experience, so that the nurse could have received instruction to fill this gap in her skill set. Further, the mechanism used to deploy staff created a situation in which all the nurses on duty in the unit at the time of the event were similarly new and inexperienced. Thus the nurse committing the error had no source of clinical expertise to whom she could turn for advice. Necessary supplies also were not available; the nurse was forced to improvise using equipment not specifically designed for the procedure, thereby creating opportunities for faulty technique. It is important to note, moreover, that the nurse did not give evidence of feeling
empowered to call a halt to an unsafe practice that was putting the patient at risk. Finally, the nurse’s statement that she felt ashamed and afraid indicates that the workplace environment did not possess a culture of safety that would encourage the reporting, analysis, and remediation of error-producing situations. Because the nurse did not come forward, none of these latent conditions were recognized as threats to patient safety, and the potential remained that future patients admitted to this unit would face a similar risk to their safety. Indeed, latent conditions such as these are present in all organizations and have been identified as posing the greatest risk to safety in complex or high-technology systems because of their capacity to result in multiple types of active failures. Their impact spreads throughout an organization, creating error-producing factors within individual workplaces (Reason, 1990).
Unfortunately, when errors are discovered, attention tends to focus on the more visible “sharp end” of the activity (the person associated with the error) because latent conditions are less visible, often hidden in routine practices or in the structure or management of an organization. As a result, responses to errors tend to focus on retraining, “discipline” (reprimanding, firing, or suing), or other responses aimed at specific individuals. Although a punitive response may be appropriate in cases of willful wrongdoing, evidence has shown that it is not an effective way to prevent subsequent errors. Focusing only on the sharp end allows latent conditions to remain undetected in the system, and their accumulation makes the system more prone to additional accidents and errors in the future.
Efforts to discover and fix latent system conditions are more likely to result in safer systems than attempts to minimize active errors at the point at which they occur (Institute of Medicine, 2000). Reason (2000:769) uses the analogy of mosquito control to illustrate this argument: “Active failures are like mosquitoes. They can be swatted one by one, but they will still keep coming.” The best remedies involve creating more effective defenses to target and prevent the conditions that allow them to breed and flourish in the first place.
However, viewing errors as resulting solely from either individual or systemic errors has its dangers. Attributing errors predominantly to the deficiencies of individuals fails to recognize the findings of safety studies estimating that the majority of unsafe acts—90 percent or more—arise from system failures in which individuals are not to blame (Reason, 1997). Focusing exclusively on individuals misses an essential part of the error story, and blocks the path to effective remediation.
On the other hand, an extreme systems perspective that recognizes no individual contributions to patient safety presents problems such as “learned helplessness” and failure to address instances of individual deficits in competencies or willful wrongdoing. With regard to the phenomenon of
“learned helplessness,” although most health professionals are highly motivated to provide safe patient care, there is a possibility that if the systems perspective becomes the sole explanation for unsafe practices, health care practitioners may be tempted to lessen their personal vigilance and striving for personal excellence and think, “It’s the system—there’s nothing I can do about it.” But safe and effective health care depends upon each professional continuing the struggle under less-than-ideal local circumstances (Reason, 1997). Further, health care practitioners vary in their expertise, competency, and exercise of necessary care. To attribute all adverse events to system failings ignores the fact that some erroneous actions, albeit a relatively small proportion of the total, are the product of reckless or incompetent individual behaviors. An exclusive focus on the systems approach will not remedy these few, but significant, threats to patient safety. It also ignores the unsung and undocumented heroes.
Thus a number of patient safety experts believe we need to strive for fair and just systems of safety that acknowledge both the individual and system contributions to successful as well as adverse events while emphasizing the systems approach to error reduction (Reason, 1997). This perspective is reflected in To Err Is Human, which concludes that efforts to prevent errors and improve patient safety will be most successful if they emphasize a systems over an individual approach, focused on modifying the conditions within the system that contribute to errors. Protecting patients from errors and adverse events therefore requires an examination of health care delivery systems to identify defects and create stronger system-level defenses. As nurses are the largest component of the health care workforce, and are also strongly involved in the commission, detection, and prevention of errors and adverse events, they and their work environment are critical elements of stronger patient safety defenses.
THE CENTRAL ROLE OF NURSES IN PATIENT SAFETY
Nurses: The Largest Component of the Health Care Workforce
Nursing personnel represent the largest component of the health care workforce. Licensed nurses2 and unlicensed nursing assistants (NAs) repre-
sent approximately 54 percent of all U.S. health care workers (e.g., physicians, nurses, dentists, allied health professionals, technicians and technologists, and other health care assistants) (Bureau of Labor Statistics, undated). Registered nurses (RNs) alone constitute approximately 23 percent of the entire health care workforce. These 2.2 million RNs, along with 683,800 licensed practical nurses (LPNs) or licensed vocational nurses (LVNs) and 2.3 million nursing aides, orderlies, attendants, and personal and home care aides, provide health care to individuals in virtually all locations in which health care is delivered—hospitals; long-term care facilities; ambulatory care settings, such as clinics or physicians’ offices; and other settings, including the private homes of individuals, schools, and employee workplaces. In most of these settings, the nurse or NA is the health care provider who has the greatest amount of direct contact with patients. In U.S. hospitals, approximately one of every four hospital employees is a licensed nurse (AHA, 2002). In nursing homes, the majority of patient care is provided by NAs, under the supervision of a licensed nurse. Efforts to detect and remedy error-producing defects in health care systems will be severely constrained without the assistance of the eyes, ears, cognitive powers, and interventions of over half the health care workforce.
Surveillance and “Rescue” of Patients
A primary activity performed by nursing staff in all hospitals, long-term care facilities, and ambulatory settings is ongoing patient surveillance (sometimes referred to as patient “assessment,” “evaluation,” or “monitoring”)—an important mechanism for the detection of errors and the prevention of adverse events. If a patient’s status begins to decline, the decline will be detectable though the nurse’s observation of changes in the patient’s physical or cognitive status. Performance of this patient monitoring requires great attention, knowledge, and responsiveness on the part of the nurse.
Patient assessment is the basis for all licensed nursing care (ANA, 1998). Indeed, ongoing patient assessment and evaluation are the two guideposts of licensed nursing care between which hands-on nursing treatments, patient education, and care planning are delivered. In acute care hospitals, this bedside monitoring or surveillance of the condition of patients prior to, during, and following medical procedures such as surgery, initiation of new medications, or a course of medical therapy typically includes, for example, monitoring patients’ vital signs (temperature, heart rate and rhythm, breathing rate and character, blood pressure), airway, risk/presence of infection, fluid intake and output, electrolytes, and pain (Bulechek et al., 1994). In intensive care units, the monitoring is more frequent, invasive, and technologically complex, as illustrated in Box 1-1.
Another nurse and I were assigned two patients: a 2-day-old infant born 3 1/2 months prematurely and a full-term, 3-day-old infant named Dan. A congenital bacterial infection had invaded Dan’s blood and lungs after his birth, and his condition had deteriorated so badly during the night that he had to be placed on a heart-lung bypass machine known as extra-corporeal membrane oxygenation, or ECMO. In his brief life, Baby Dan already had suffered multiple ruptures of his lung tissue, the result of the high pressures needed by the mechanical ventilator to push air into his diseased lungs. Two tubes, inserted between his ribs on both sides, removed the air leaking into his chest cavity. A third tube, exiting below his sternum, removed fluid collecting in the sac around his heart to prevent compression of the heart. The ECMO machine, used only as a last resort in dire cases, functionally replaced Dan’s failing heart and lungs. The machine drained his blood from a small tube inserted into a vein in his neck, passed it through plastic tubing to an artificial lung for gas exchange, and returned it under pressure to his body through a second tube in his aorta.
Blood flowing outside the body involves a great risk of clotting, which is controlled by continuous infusion of a blood-thinning medication, heparin, into the ECMO circuit. However, too much thinning of the blood can lead to uncontrolled bleeding, and the fluid oozing from Dan’s incision sites showed that his blood’s ability to clot was already severely impaired. I had to test his blood’s clotting ability every 10 minutes to adjust the heparin infusion. In addition, he was on two other medication infusions to address his failing blood pressure and required frequent transfusions of various blood products to supply clotting factors and improve his blood pressure. He further was receiving several antibiotics to combat the infection and required constant sedation to keep him from fighting us. Caring for an ECMO patient typically required two nurses—one trained as a specialist to monitor the ECMO circuit continuously, the other to provide constant assessment of the patient’s vital signs and other health status indicators and manage the other aspects of patient care.
Over the course of our 12-hour shift, we started to rein in his many problems, and Baby Dan slowly improved. Although he would remain on ECMO several more days to recuperate, he eventually overcame his infection and was discharged.
SOURCE: Bingham (2002).
A review of 81 research papers published predominantly since 1990 examining the relationship between organizational structures/processes and patient mortality/adverse events revealed that nursing surveillance was one of three organizational process variables consistently related to lower mortality (Mitchell and Shortell, 1997). Studies of quality of care before and after implementation of the Medicare prospective payment system for hospitals found better-quality nursing surveillance to be predictive of lower severity-adjusted Medicare mortality (Kahn et al., 1990; Rubenstein et al., 1992).
Although the type and frequency of patient assessment and monitoring activities carried out by licensed nurses vary by the setting of care, the clinical condition, and other characteristics of the patient, such activities are performed by nurses for each patient in every setting in which health care is delivered—ambulatory primary care sites, hospitals, schools, workplace health sites, home health agencies, and nursing homes. In nursing homes, each resident receives a comprehensive assessment performed or coordinated by an RN upon admission and at regularly scheduled intervals thereafter. This assessment employs a federally prescribed minimum data set (MDS)3 to document each resident’s diagnoses and health conditions, dental and nutritional status, skin condition, medications, discharge potential and other special treatments or conditions needed, customary routines, cognitive patterns, communication, vision, mood and behavior patterns, psychosocial well-being, physical functions, continence, and other physical and psychosocial characteristics. When this assessment detects areas of concern, a more detailed resident assessment protocol is initiated (Morris et al., 1995).
For chronically ill homebound patients, home health nurses assess the health status and responses to treatments of individuals too ill to leave their home using a wide array of assessment instruments and tools. Examples of these include stethoscopes, sphygmomanometers (blood pressure measurement devices), Doppler fetal monitors, depression screening tools, Denver Developmental Screening tests, pain scales, the Braden scale for pressure ulcer prevention, wound measurement instruments, diet recall checklists, glucose tests, urine tests, fall risk assessment tools, an Alcohol Consumption Questionnaire, functional independence measurements, safety checklists, the SF-12 and other health surveys, tools for measuring activities of daily living (ADLs) and instrumental ADLs, a Mini-Mental Status Examination, a Family Assessment for School Nurses, and vision and hearing assessment tools (Martin, 2002). In addition, since 1999 the Medicare pro-
gram has required that an RN perform a comprehensive, detailed assessment of each Medicare beneficiary receiving Medicare-covered home health care at the initiation of home health care services and at regular intervals thereafter.4 The nurse performing this assessment must assess the patient’s health status and health care and support needs, as well as items included in the Medicare Outcome and Assessment Information Set (OASIS) that address the patient’s history, and “sensory status, integumentary status, respiratory status, elimination status, neuro/emotional/behavioral status, activities of daily living, [and] medications,” among other information.5
While performing these assessments (and also when delivering therapeutic treatment and patient education), nurses are functioning at the “sharp end” of the health care system because of their immediate link to the patient. This ongoing vigilance function often thrusts nurses into a role that has been described as the “front line” of patient defense (JCAHO, 2001). Studies of organizations with a strong track record of high reliability and safety have shown that such vigilance by front-line workers is essential for detecting threats to safety before they actually become errors and adverse events (Roberts, 1990; Roberts and Bea, 2001). Because licensed nurses and NAs work at the “sharp end” of health care delivery, they are key instruments for carrying out such vigilance in health care.
The goal of this nursing surveillance or vigilance function is the early detection of a downturn in a patient’s health status or the advent of an adverse event, and the initiation of activities to “rescue” the patient and restore health. When this does not happen, “failure to rescue” is said to occur. The concept of failure to rescue has been tested and validated as an indicator of the quality of acute hospital care for surgical patients (Silber et al., 1992). When higher levels of nurse staffing are present, the incidence of failure to rescue is reduced (Aiken et al., 2002; Needleman et al., 2002). Further evidence of the effectiveness of nurse surveillance is found in studies of medication errors. A systems analysis of 334 medication errors associated with 264 preventable adverse events occurring in two hospitals over a 6-month period revealed that nurses were the health care personnel most likely to intercept errors in the ordering of a medication by a physician, the transcription of the drug order by a clerk, or the dispensing of the drug by a pharmacist before such errors resulted in an adverse event. Nearly half of all physician errors examined in this study had been intercepted before they resulted in an adverse event; 87 percent of those interceptions were by nurses. About one-third of transcription and dispensing errors had been
intercepted prior to administration, again largely by nurses. Overall, nurses were responsible for intercepting 86 percent of all medication errors made by those in all disciplines (Leape et al., 1995).
Coordination and Integration of Care and Services from Multiple Providers
In addition to providing surveillance of patients, therapeutic nursing interventions, and treatments to carry out medical orders, licensed nurses serve as the integrator or coordinator of patient care. These integrating activities include implementing physician treatment orders and explaining them to the patient; planning for patients’ discharge from hospitals or other health care facilities to enable continued care in the home, school, or nursing home; providing health care treatment in the home or other setting of care; and educating the patient and family about the patient’s disease, course of therapy, medications, self-care activities, and other areas of concern to the patient. In addition, while such practices are not desirable, nurses are also pressed into performing a variety of non-nursing patient care activities because of their ever-present availability in inpatient facilities. For example, when delivery of medications, medical equipment or supplies, blood products, or laboratory specimens is required for the patient, and transport staff are not available for the purpose, this activity often is carried out by the nurse. This practice, relying on the “inevitable availability” of nurses, occurs frequently (Prescott et al., 1991; Upenieks, 1998). Large proportions of nurses report spending time delivering and retrieving food trays; performing housekeeping duties; transporting patients; and ordering, coordinating, or performing ancillary services (Aiken et al., 2001a).
The amount of time nurses spend integrating or coordinating care is suggested by the amount of time they spend on “indirect” as opposed to “direct” patient care. Direct patient care encompasses activities carried out in the presence of the patient and family, such as performing a physical examination and other assessments of the patient, administering medications, and performing treatments and procedures. Indirect care involves those activities that are performed away from but on behalf of the patient, such as documenting care, communicating with other health care providers, seeking consultations, and preparing medications (Division of Nursing, 1978). Although numerous work sampling studies of nursing care have been conducted—with varying degrees of divergence from these definitions—and the location of some indirect care activities may be shifting to the bedside (as is the case with automated patient records), the vast majority of studies agree that nurses spend a greater percentage of their time in indirect versus direct care (Hendrickson et al., 1990; Linden and English, 1994; Upenieks, 1998). As a result of all these indirect activities, nurses have substantial
contact with all health care personnel providing care to the patient—across multiple units, divisions, services, institutions, and providers constituting the health care delivery system—and are able to detect and take action to fill gaps in patient care in order to protect the patient.
Distinguished physician and author Lewis Thomas, former Dean of the Yale and New York University medical schools and chief executive officer of the Sloan-Kettering Institute in New York City at the time of his death in 1993, well describes this integrating and coordinating function of nurses in The Youngest Science: Notes of a Medicine Watcher:
One thing the nurses do is to hold the place together. It is an astonishment, which every patient feels from time to time, observing the affairs of a large, complex hospital from the vantage point of his bed, that the whole institution doesn’t fly to pieces. A hospital operates by the constant interplay of powerful forces pulling away at each other in different directions, each force essential for getting necessary things done, but always at odds with each other…. My discovery, as a patient … is that the institution is held together, glued together, enabled to function as an organism, by the nurses and nobody else. (Thomas, 1983:66–67)
PATIENT SAFETY RISK FACTORS IN NURSES’ WORK AND WORK ENVIRONMENTS
Because nurses carry out the responsibilities described above, they potentially are well positioned to observe and influence how the health care system functions across all aspects of patient care, and thereby to detect and address threats to patient safety. However, nurses’ work and work environments have changed over the last two decades, and these changes have been cited as having implications for patient safety.
More Acutely Ill Patients
Nurses, health care industry associations, and numerous other entities have observed that hospital and nursing home patients are more severely ill than in the recent past. Although the truth of this observation is widely accepted, its extent and its implications for nursing are difficult to determine. First, there is no standard method used across hospitals to measure the severity of illness of all hospital patients. Although many hospitals use patient acuity systems to estimate the amount of nursing care their patients will require, those systems are not standardized, and there is no external reporting to produce national trend data. Second, where other severity-of-illness measurements are collected (i.e., for Medicare patients), the severity of a patient’s medical illness does not necessarily correlate with the level of nursing care that a patient requires. For example, a patient with pneumonia
might not have a high score on a medical severity-of-illness algorithm but still could require a large amount of nursing care.
Nonetheless, Medicare data and a limited amount of state-specific hospital data support the observation that, beginning in the mid-1980s following implementation of the Medicare prospective payment system (PPS) for hospitals and continuing into the late 1990s, patients admitted to hospitals were increasingly more acutely ill. Data on all Medicare hospital admissions for 1985–1997 show an annual increase in the complexity of cases treated in acute care hospitals as measured by the Medicare case mix index (CMI),6 while a review of patient data for all payors and all acute care general hospitals in Pennsylvania during 1994–1997 revealed that the severity of illness of patients admitted to those hospitals increased in the aggregate by 4.5 percent over the 4-year period (Unruh, 2002b). The annual increases were highest in the early years just after implementation of the PPS and slowed fairly steadily until 1998, when a decline in severity as measured by the CMI was observed. This decline continued into 1999, the last year for which these data are available. It was determined that the CMI decrease of 0.5 percent in 1998 likely reflected changes in coding practices; however, this was not the case for the 0.4 percent CMI decline in 1999 (Medicare Payment Advisory Commission, 2001).
This increase in the severity of illness of hospital patients has had a ripple effect throughout all health care settings. Evidence indicates that patients receiving care in long-term care facilities, in their homes, and in other community-based settings are more ill and debilitated and/or require more technologically complex medical care than in the past. In nursing homes, the proportion of patients who are more frail (i.e., need assistance with three or more ADLs, such as bathing, dressing, eating, and toileting) and therefore need more skilled and/or specialized care increased from 72 percent in 1987 to 83 percent in 1996. As a consequence, over the last few years, nursing homes have developed specialized units to care for patients who need more extensive care, such as those with dementia, rehabilitation needs, ventilator dependency, or brain injury. Approximately 12.6 percent of all nursing homes in 1996 had units devoted to the specialized care of
individuals with Alzheimer’s disease (the most common type of specialized unit); more than half had been in operation for 5 years or less (Rhoades and Krauss, 2001).
While there is no precise way to measure trends in the numbers of nursing home patients having more complex medical needs—necessitating intervention from a licensed nurse as opposed to ADL support from an NA—changes can be inferred from the proportion of residents whose care is covered by Medicare, because Medicare coverage of nursing home care is limited to payment for rehabilitation care and skilled nursing services. Between 1987 and 1996, the percentage of nursing home patients whose care was paid for by Medicare increased from 3 to 9 percent, and the proportion of nursing homes certified to receive Medicare reimbursement increased from 28 to 73 percent, indicating that the number of nursing homes planning to take in residents with more acute illness or more complex needs increased substantially (Rhoades and Krauss, 2001). This increase in resident dependency and medical complexity has important implications for the work of nurses and NAs. Staff time required to meet basic patient care needs (such as feeding, toileting, and ambulation) increases with the level of dependency of residents (CMS, 2002). Since Medicare residents often have complex health conditions or are recovering from serious health events, a more sophisticated knowledge base is required to care for these residents, and a higher level of vigilance and monitoring is required.
Shorter Hospital Stays
In addition to the likelihood that patients in hospitals are sicker, evidence is clear that when patients are admitted to the hospital, their hospital stays are for shorter periods of time than in the past. This combination of increased patient severity of illness and shorter inpatient stays has given rise to the expression that nurses are asked to care for patients “sicker and quicker.” From 1980 to 2000, the average length of a patient’s stay in the hospital (for nonfederal short-term general hospitals and other special hospitals) declined from 7.6 to 5.8 days (AHA, 2002). Although it is likely that these shorter stays in part reflect improvements in care, their implication in the context of nursing is that as patients’ lengths of stay decrease, the less demanding initial patient workup and post-treatment recovery periods are foregone. The remaining patient days in the hospital involve caring for patients in need of a greater intensity of care. Further, these reduced lengths of stay allow less time for nurses to become acquainted with their patients’ baseline health and to readily detect changes in health status, educate patients and families about health conditions, and fully prepare patients and families for discharge. Shorter lengths of inpatient stays also transfer the risk for adverse events from the hospital setting to the home, where such
events may be less readily detected and result in more serious consequences for the patient.
Labor costs are the largest component of hospital expenses, and nursing staff represent the largest category of hospital labor (AHA, 2002). As hospitals tried to respond to the cost pressures generated by new reimbursement methods in the 1980s, many of their approaches targeted more efficient use of nursing staff. These initiatives (referred to as restructuring, reengineering, or redesign initiatives)7 continued through the 1990s into the present and have been widely adopted (Gelinas and Manthey, 1997). Redesign initiatives typically have changed the ways in which licensed nurses and NAs are organized to provide patient care, through, for example, personnel reductions; cross-training of personnel to perform additional duties; changes in the mix of nursing staff (RN, LPN/LVN, or unlicensed staff); reassignment of support services (e.g., laboratory, radiology) to nursing units; redistribution of patients across nursing units; redesign of patient care processes; and other changes in organization structure, decision-making processes, and the responsibilities of management and patient care staff (Aiken et al., 2000; Norrish and Rundall, 2001; Ritter-Teitel, 2002; Walston et al., 2000; Walston and Kimberly, 1997). Use of multiskilled workers who are not RNs to perform such activities as making beds, giving patients baths, positioning patients too ill to position themselves, performing electrocardiograms, and drawing blood was identified as a core feature of redesign initiatives by 61 percent of 360 hospital nurse executives surveyed in 1995 (Gelinas and Manthey, 1997).
The outcomes of these redesign initiatives are not clear (Walston et al., 2000). Formal measurements of the results of these multifaceted restructuring, reengineering, and redesign initiatives have been few, and findings have been contradictory with respect to the consequences for nurses’ work and work environment, including nursing staff satisfaction, control over work environment, concern over changes in responsibilities, and work group relationships. However, role conflict and ambiguity are consistent issues in redesigned work settings (Ingersoll et al., 2001; Walston et al., 2000), and such changes have been well documented as contributing to error-producing situations because they involve departures from well-established routines and create new situations for which workers have no preplanned solutions (Reason, 1990).
Changes in Deployment of Nursing Personnel to Care for Patients
Declining numbers of nursing staff available to care for inpatients in health care facilities have been widely reported by the press, labor publications, and professional journals (Aiken et al., 2001b; Hurley, 2000; Shindul-Rothschild et al., 1996). Quantitative analyses to explore this perception have been hampered by the limitations of available data on nurse staffing8 and patient acuity. As a result, national studies have not yet produced a fully clear picture of changes in nurse staffing levels. An analysis of national hospital staffing data from 1981 through 1993 (while total hospital employment was growing steadily) revealed that total nursing personnel (RNs, LPNs, and NAs) per 1,000 adjusted patient days, also adjusted for case mix, declined nationally by 7.3 percent. This decrease in the number of nursing caregivers per patient was accomplished primarily through the loss of non-RN personnel (Aiken et al., 1996). A follow-up study of RN staffing between 1990 and 1996 found that the number of hospital RNs increased nationally by 15 percent, and that the percentage of RNs among all hospital employees increased from approximately 22 percent to 25 percent. During this period, however, LPN full-time equivalents (FTEs) decreased by 14 percent (data were not available on NAs) (Kovner et al., 2000).
Several explanations have been advanced for the mismatch between reports of declining RN staffing and the quantitative data generated by analyses such as those cited above. The first is that while levels of RN staffing may have held constant or even increased, they have not been adequate to compensate for the loss of LPN/LVN and NA staff whose duties likely have fallen to RNs. The further increase in patient acuity and shortened hospital stays compounds the workload of RNs. Another explanation is that inadequate staffing data cannot fully document the extent to which RNs are or are not available to provide direct care to patients. Data on RN hospital staffing often include RNs engaged in administrative duties who have no patient care responsibilities, as well as RNs providing care in outpatient hospital settings, and therefore cannot provide a clear picture of changes in the numbers of RNs providing direct care to inpatients. Finally, these studies have not always distinguished between full-time and part-time nurses; two part-time nurses may be counted as two nurses despite equaling only one full-time nurse.
Another important factor is the extent to which national statistics mask the variation that exists across individual hospitals. A recent and detailed analysis of nurse staffing levels at the aggregate level across facilities and at
the level of individual hospitals illustrates this point. This study of nurse staffing in all general, acute care Pennsylvania hospitals from 1991 to 1997 found that, although the statewide ratio of all nursing staff (RNs, LPNs, and NAs) to patient days of care increased from 3.86 to 4.04 between 1991 and 1997, examination of staffing at each hospital individually revealed that 32 percent of hospitals reduced the ratio of all nurses (RNs, LPNs, and NAs) to patients by more than 10 percent; and, with adjustments for the increased acuity of patients, more than 50 percent of hospitals decreased their ratio of nursing staff to patient days by more than 10 percent (Unruh, 2002a). Such declines are worrisome because health services research continues to produce strong evidence that nurse staffing in the aggregate is an important factor in the prevention of adverse events in both acute hospitals (Kovner et al., 2002; Needleman et al., 2002; Seago, 2001) and nursing homes (CMS, 2002).
Frequent Patient Turnover
High patient turnover rates contribute to increased workload for hospital nurses. Patient turnover refers to the phenomenon in which a given hospital bed may be occupied by more than one patient in a 24-hour period. For example, a patient may be discharged at 10:00 in the morning and a new patient admitted to the same bed during the same nursing shift. The number of patients in need of care is typically counted at a point in time during a 24-hour period (e.g., midnight). However, this patient census does not indicate the true number of patients in need of care because it does not reflect the actual number of patients cared for or the admissions and discharges taking place on a given day. Assessment and stabilization of patients upon admission and patient education and planning upon discharge are time- and personnel-intensive.
The patient turnover rate has increased as the numbers of available hospital beds and lengths of stay have declined. In one study of 20 medical–surgical units in five hospitals, the number of admissions, discharges, and transfers averaged between 25 and 70 percent of the midnight census (Lawrenz, 1992). Patient turnover rates as high as 40–50 percent also have been reported during an 8- to 12-hour period (Norrish and Rundall, 2001).
High Staff Turnover
High rates of turnover characterize the nursing staff of both hospitals and nursing homes. Such high turnover can have adverse consequences for patient safety. Evidence from non–health care industries shows that new or substitute staff are less familiar with work processes, and that the potential for errors thereby increases (Rousseau and Libuser, 1997). In nursing
homes, high turnover rates have been hypothesized to result in low staff morale, staff shortages, and poor quality of care (CMS, 2002).
A 2001 survey of directors of nursing of all U.S. nonfederal acute care hospitals found (for the 14.7 percent of hospitals responding) that, on average, 21.3 percent of all full-time registered hospital nurses had resigned or been terminated during the preceding year. While most hospitals reported turnover rates of 10–30 percent, some cited much higher rates. For example, 2 percent of responding hospitals reported turnover rates of 50 percent or higher (The HSM Group, 2002). Turnover rates among nursing staff in nursing homes are even greater. A national survey conducted by the American Health Care Association (AHCA) in 2001 found annual turnover rates of 78 percent for NAs, 56 percent for staff RNs, 54 percent for LPNs/ LVNs, and 43–47 percent for directors of nursing and RNs with administrative duties (AHCA, 2002).
Long Work Hours
Nursing staff working in in-patient facilities traditionally have worked in 8-hour shifts, but increasingly work longer hours. Reasons include a desire for increased compensation (“elective overtime”), requirements by facilities to work overtime (“mandatory overtime”) to compensate for insufficient staffing, and a desire for more flexible work hours (e.g., 10- or 12-hour shifts) to accommodate the needs of either facilities or nurses or both. Scheduled shifts may be 8, 10, or 12 hours, and may not follow the traditional pattern of day, evening, or night shifts. Moreover, nurses working on specialized units, such as the OR, dialysis units, and some intensive care units, may be required to be on call in addition to their regularly scheduled shifts (Rogers, 2002).
A 2002 study funded as part of AHRQ’s initiative to examine the effects of working conditions on patient safety documented the work patterns of a national sample of hospital staff nurses who are members of the American Nurses Association. The study measured each nurse’s work hours, length of shifts, and amount of overtime hours worked and the effects of these factors on nurses’ commission of errors. It was found that although the majority (84.3 percent) of scheduled shifts were 8 or 12 hours in duration, 3.5 percent were for periods greater than 12 hours, some lasting as long as 22.5 hours.9
Research on the work hours of nursing staff in nursing homes also has revealed extended work hours. In site visits to 17 nursing facilities in Ohio, Colorado, and Texas in 2001, researchers found that double shifts (i.e., two consecutive 8-hour shifts totaling 16 hours) and extra shifts were performed in many of these facilities on a regular basis. Double shifts in particular were pervasive. In 13 of the 17 facilities, at least one nursing staff member, but frequently more, had worked between one and three double shifts in the previous 7 days. In five facilities, at least one staff member had worked between four and seven double shifts in the last 7 days. In one of the facilities, more than a third of the interviewed nursing staff had worked between eight and eleven double shifts in the last 14 days (CMS, 2002).
The number of hours worked has been identified as a contributing factor to the commission of errors by nurses (Narumi et al., 1999). The AHRQ-funded study mentioned above found that shift durations of greater than 12 hours were significantly associated with increased errors among nurses.
Rapid Increases in New Knowledge and Technology
The IOM (2001) report Crossing the Quality Chasm cites the growing complexity of science and technology, resulting from the tremendous advances made in clinical knowledge, drugs, medical devices, and technologies for use in patient care, as one of the four main attributes of the U.S. health system affecting health care quality. Since the results of the first randomized controlled clinical trial were published more than 50 years ago, health care practitioners have been increasingly inundated with information about what does and does not work to achieve good clinical outcomes. Over the last 30 years, such trials have increased in number from 100 to nearly 10,000 annually. The first 5 years of this 30-year period accounts for only 1 percent of all the articles in the medical literature, while the last 5 years accounts for almost half. Although part of this growth in the literature can be attributed to factors other than new findings and knowledge, there is no doubt that as the knowledge base has expanded, so, too, has the number of drugs, medical devices, and other technological supports (IOM, 2001).
Such increases in technology are beneficial and likely to continue. In a study of hospital organizational and structural features associated with patient mortality, only the presence of high technology or its proxies has been consistently associated with lower mortality (Mitchell and Shortell, 1997). However, these developments also have implications for patient safety and health care providers, including nursing staff. First, as stated in the Quality Chasm report, “Today, no one clinician can retain all the information necessary for sound, evidence-based practice. No unaided human being can read, recall, and act effectively on the volume of clinically relevant scientific
literature” (IOM, 2001:25). If nurses are not aided with information and decision support at the point of care delivery, the likelihood of errors increases. Second, this growth of technology, much of it involving high-risk systems, creates changes in the work nurses are asked to perform. In particular, as systems (e.g., medication administration) become more automated, the technology makes work less transparent and creates opportunities for new types of errors (Reason, 1990).
Increased Interruptions and Demands on Nurses’ Time
Changes such as those described above have resulted in increases in the types and amount of work required of nurses. In addition to the heavier patient care loads borne by nursing staff, evidence cited above indicates that large proportions of nurses spend time performing activities that can disrupt their primary patient care responsibilities, such as delivering and retrieving food trays; performing housekeeping duties; transporting patients; and ordering, coordinating, or performing ancillary services, such as delivery of medical equipment or supplies, blood products, or laboratory specimens (Aiken et al., 2001a; Prescott et al., 1991; Upenieks, 1998). It is clear that interruptions and interference occur frequently in nursing care from these and other nursing unit activities (Bowers et al., 2001; O’Shea, 1999; Wakefield et al., 1998; Walters, 1992). To the extent that such interruptions and distractions take place, patient safety is threatened. When health professionals have been asked to report their perceptions of why medical errors occur, interruptions and distractions have frequently been cited (Ely et al., 1995; Gladstone, 1995).
Documentation and Paperwork
Documenting nursing work and other activities to meet facility, insurance, private accreditation, state, and federal requirements, as well as to furnish information needed by other providers, is uniformly cited across all care delivery settings as imposing a heavy demand on nurses’ time. The types of required documentation vary. Some may be characterized as administrative, that is, not treatment-specific; examples are providing insurance certifications, obtaining permission for the release of information, and informing patients of their rights. Other documentation pertains to nursing care; examples here are recording medications and treatment given, performing nursing assessments, and preparing discharge plans. Nurses in particular settings must also complete setting-specific documentation. For example, as discussed earlier, home health care nurses must complete a
federally required OASIS assessment instrument for each Medicare beneficiary receiving Medicare home health care services, while nursing home nurses must complete a similar federally prescribed MDS for nursing home residents. These data sets are not always maximally compatible with internal documentation systems used by HCOs (e.g., the OMAHA system for home health care) and can create redundancies. Finally, nurses sometimes practice lengthy narrative charting as a defense against increasing litigation.
To the extent that paperwork and other documentation requirements lessen the time nurses have for direct contact with patients, they contribute to the reduced availability of nurses that has been shown to affect patient safety. Estimates from work sampling studies and surveys of nurses within individual hospitals of the amount of time spent in patient care documentation range from 13 to 28 percent (Korst et al., 2003; Pabst et al., 1996; Smeltzer et al., 1996; Upenieks, 1998; Urden and Roode, 1997). Home care nurses are estimated to spend a much greater proportion of their time in documenting care. According to some estimates, home health nurses spend approximately twice as much time in documenting patient care as do hospital nurses, in part because of more prescriptive federal regulatory requirements (Trossman, 2001). Completion of required paperwork is also cited as one reason nurses work overtime; because it cannot be accomplished in an 8-hour shift, it becomes a form of unpaid mandatory overtime (Trossman, 2001).
THREATS TO PATIENT SAFETY POSED BY WORK ENVIRONMENT FACTORS
All of the changes affecting the work environment of nurses described above can constitute latent factors conducive to health care errors. This fact is dramatically expressed in the text, but not the title, of a widely cited Chicago Tribune article, “Nursing Mistakes Kill, Injure Thousands Annually” (Berens, 2000). This article reports the results of an analysis of records from the U.S. Food and Drug Administration and other Department of Health and Humans Services agencies, federal and state files of annual hospital surveys and complaint investigations, court and private health care files, and nurse disciplinary records for every state. The analysis detected 1,720 deaths and 9,584 injuries among hospital patients resulting from the actions or inactions of RNs over a 5-year period, and 119 deaths and 564 patient injuries due to errors on the part of unlicensed NAs. Because of incomplete reporting, the article notes that these numbers should be interpreted as underestimates. Despite its title, the article does not point to willful wrongdoing or carelessness on the part of the RNs and NAs associated with these errors. Instead, it calls attention to their working conditions as the underlying causes (latent conditions) of the errors, prominently citing
inadequate nurse training and insufficient monitoring of patients because of too few nurses being assigned to patient care.
These findings are underscored by an analysis of data on serious health care errors that are reported to the Joint Commission on the Accreditation of Healthcare Organizations (JCAHO) database on sentinel events. JCAHO defines a sentinel event as an “unexpected occurrence involving death or serious physical or psychological injury, or the risk thereof” (JCAHO, 2003:53). The JCAHO database is relatively small and subject to underreporting. Nevertheless, for 19 percent of the total errors reported to the database from 1995 to 2002, nurse staffing levels are cited as one of the four major causal factors for reported serious errors/adverse events, such as patient falls, medication and transfusion errors, delays in treatment, and operative and postoperative complications. Inadequate staff orientation and training and competency assessment, as well as breakdowns in communication, were also revealed as frequent contributors to errors; communication-related factors were the most frequently identified root cause of all types of sentinel events (Croteau, 2003).
Preventing errors associated with such conditions requires that strong defenses be built into the work environment of nurses. As noted by Reason (2000:769), “We cannot change the human condition, but we can change the conditions under which humans work.”
TRANSFORMING NURSES’ WORK ENVIRONMENTS: ESSENTIAL TO PATIENT SAFETY
The evidence cited above and in succeeding chapters makes clear that (1) patient safety continues to be threatened; (2) latent conditions in work environments are the primary sources of those threats; and (3) nurses are the largest contingent of health care workers and perform critical patient safety functions while operating at the “sharp end” of health care. Given these facts, it is clear that the latent conditions present in the work environment of nurses must be addressed if patient safety is to be improved. This conclusion validates AHRQ’s charge to the IOM to identify key aspects of the work environment for nurses likely to have an impact on patient safety, and potential improvements in health care working conditions that would likely increase patient safety.
In carrying out this charge, the committee reviewed published research and other evidence from a variety of disciplines: health services and nursing research; behavioral and organizational research on work and workforce effectiveness; human factors analysis and engineering; studies of organizational disasters and their evolution; and studies of high-risk industries (e.g., nuclear power production, chemical processing, transportation) with low accident rates (often called “high-reliability organizations”). The commit-
tee also commissioned papers and received expert testimony. (Appendix A contains a description of the committee’s membership and the process used to conduct this study.)
This process revealed that identifying and remediating latent factors in the work environment of nurses and increasing patient safety are not likely to be achieved by any single action. Instead, it will be necessary to implement bundles of mutually reinforcing practices—changes that support each other in altering the context of worker behavior within a work environment. Such bundles of changes are needed within each of the four fundamental components of all organizations: (1) management and leadership, (2) workforce deployment, (3) work processes, and (4) organizational culture. The changes needed in each of these components are essential to building stronger patient safety defenses in HCOs. Evidence also indicates that they are basic to efficient organization practices in the twenty-first century and to recruitment and retention of nurses in a time of nursing shortages, and indeed are fundamental to the effective deployment of all health care workers, not just nurses. However, evidence further indicates that many of these fundamental changes have not yet occurred in the work environments of nurses; thus there is a need not merely for small changes in those environments, but for a broad transformation.
Many individual aspects of the necessary transformation in these four bundles of practices are identified in To Err Is Human (IOM, 2000) and Crossing the Quality Chasm (IOM, 2001). This report is intended to serve as a companion to those earlier reports. It delves more deeply into some of their recommendations, and addresses some issues not discussed in those reports, such as worker fatigue and staffing levels. It also emphasizes the role health care organizations can play in increasing patient safety—a role addressed less fully in To Err Is Human and Crossing the Quality Chasm (Berwick, 2002; IOM, 2001).
In Chapter 2, we focus on the underlying framework linking the needed bundles of changes in management and leadership, workforce deployment, work processes, and organizational culture. We also describe further how this report relates to To Err Is Human and Crossing the Quality Chasm. Chapter 3 describes the characteristics of the nursing workforce and its work that are important factors in reshaping nursing work environments. Chapters 4 through 7 address the above four organizational components and the evidence base supporting the committee’s recommendations for change: Chapter 4 examines the need for evidence-based management and leadership; Chapter 5 calls for strengthening workforce capability; Chapter 6 speaks to the need to design nurses’ work and workspace to prevent errors; and Chapter 7 describes the need to create and sustain cultures of safety within organizations. Finally, Chapter 8 reviews the study findings in light of the turbulence that is characteristic of the U.S. health care system. It
presents a case for making these changes despite the many difficulties facing HCOs, policy makers, and other components of the health care system. It asserts the committee’s position that it is not just necessary, but also possible, to transform the work environment of today’s nurses. It further provides evidence that in addition to benefiting patients, such changes will benefit nurses, other health care workers, and the organizations in which they practice.
AHA (American Hospital Association). 2002. Hospital Statistics 2002. Chicago, IL: Health Forum LLC, an affiliate of AHA.
AHCA (American Health Care Association). 2002. Results of the 2001 AHCA Nursing Position Vacancy and Turnover Survey. Washington, DC: AHCA.
Aiken L, Sochalski J, Anderson G. 1996. Downsizing the hospital nursing workforce. Health Affairs 15(4):88–92.
Aiken L, Clarke S, Sloane D. 2000. Hospital restructuring: Does it adversely affect care and outcomes? Journal of Nursing Administration 20(10):457–465.
Aiken L, Clarke S, Sloane D, Sochalski J, Busse R, Clarke H, Giovannetti P, Hunt J, Rafferty A, Shamian J. 2001a. Nurses’ reports on hospital care in five countries. Health Affairs 20(3):43–53.
Aiken L, Clarke S, Sloane D, Sochalski J. 2001b. An international perspective on hospital nurses’ work environments: The case for reform. Policy, Politics, & Nursing Practice 2(4):255–263.
Aiken L, Clarke S, Sloane D, Sochalski J, Silber J. 2002. Hospital nurse staffing and patient mortality, nurse burnout, and job dissatisfaction. Journal of the American Medical Association 288:1987–1993.
ANA (American Nurses Association). 1998. Standards of Clinical Nursing Practice. Washington, DC: ANA.
Berens M. 2000 (September 10). Nursing Mistakes Kill, Injure Thousands. Chicago Tribune. News Section. P. 20.
Berwick D. 2002. A user’s manual for the IOM’s “Quality Chasm” report. Health Affairs 21(3):80–90.
Bingham R. 2002. Leaving nursing. Health Affairs 21(1):211–217.
Blendon R, DesRoches C, Brodie M, Benson J, Rosen A, Schneider E, Altman D, Zapert K, Herrmann M, Steffenson A. 2002. Views of practicing physicians and the public on medical errors. The New England Journal of Medicine 347(24):1933–1940.
Bowers B, Lauring C, Jacobson N. 2001. How nurses manage time and work in long term care facilities. Journal of Advanced Nursing 33:484–491.
Bulechek G, McCloskey J, Titler M, Denehey J. 1994. Report on the NIC project: Nursing interventions used in practice. American Journal of Nursing 94(10):59–62, 64, 66.
Bureau of Labor Statistics, Department of Labor. Undated. 2001 National Occupational Employment Statistics. [Online]. Available: http://stats.bls.gov/oesdate [accessed December 13, 2002].
CMS (Centers for Medicare and Medicaid Services). 2002. Report to Congress: Appropriateness of Minimum Nurse Staffing Ratios in Nursing Homes—Phase II Final Report: U.S. Department of Health and Human Services. [Online]. Available: www.cms.gov/medicaid/reports/rp1201home.asp [accessed June 25, 2002].
CMS. 2003. Acute Inpatient Prospective Payment System: U.S. Department of Health and Human Services. [Online]. Available: “Last modified on Tuesday, September 9, 2003.” Available: www.cms.gov/providers/hipps/ippspufs.asp [accessed October 4, 2003].
Croteau R. 2003 (March 26). Lessons Learned From Review of Sentinel Events 1995–2002: Issues Relevant to Nursing. Joint Commission on the Accreditation of Healthcare Organization Nursing Advisory Council Meeting.
Division of Nursing. 1978. Methods for Studying Nurse Staffing in a Patient Unit. DHEW Publication No. HRA 78-3. Washington, DC: U.S. Government Printing Office.
Ely J, Levinson W, Elder N, Mainous A, Vinson D. 1995. Perceived causes of family physicians’ errors. Journal of Family Practice 40(4):337–344.
Gelinas L, Manthey M. 1997. The impact of organizational redesign on nurse executive leadership. Journal of Nursing Administration 27(10):35–42.
Gladstone J. 1995. Drug administration errors: A study into the factors underlying the occurrence and reporting of drug errors in a district general hospital. Journal of Advanced Nursing 22(4):628–637.
Hendrickson G, Doddato T, Kovner C. 1990. How do nurses use their time? Journal of Nursing Administration 20(3):31–37.
Hurley M. 2000. Workload, UAPs, and you. RN 63(12):47–49.
Ingersoll G, Fisher M, Ross B, Soja M, Kidd N. 2001. Employee response to major organizational redesign. Applied Nursing Research 14(1):18–28.
IOM (Institute of Medicine). 2000. To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press.
IOM. 2001. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press.
JCAHO (Joint Commission on Accreditation of Healthcare Organizations). 2001. Front Line of Defense: The Role of Nurses in Preventing Sentinel Events. Oakbrook Terrace, IL: Joint Commission Resources.
JCAHO. 2003. 2003 Hospital Accreditation Standards. Oakbrook Terrace, IL: Joint Commission Resources.
Kahn K, Rogers W, Rubenstein L, Sherwood M, Reinisch E, Keeler E, Draper D, Kosecoff J, Brook R. 1990. Measuring quality of care with explicit process criteria before and after implementation of the DRG-based prospective payment system. Journal of the American Medical Association 264(15):1969–1973.
Korst L, Eusebio-Angeja A, Chamorro T, Aydin C, Gregory K. 2003. Nursing documentation time during implementation of an electronic medical record. Journal of Nursing Administration 33(1):24–30.
Kovner C, Jones C, Gergen P. 2000. Nurse staffing in acute care hospitals, 1990–1996. Policy, Politics, & Nursing Practice 1(3):194–204.
Kovner C, Jones C, Zhan C, Gergen P, Basu J. 2002. Nurse staffing and post surgical adverse events: An analysis of administrative data from a sample of U.S. hospitals, 1990–1996. Health Services Research 37(3):611–629.
Lawrenz E. 1992. Are patient classification systems still the best way to measure workload? Perspectives on Staffing and Scheduling XI(5):1–4.
Leape L, Bates D, Cullen D, Cooper J, Demonaco H, Gallivan T, Hallisey R, Ives J, Laird N, Laffel G, Nemeskal R, Petersen L, Porter K, Servi D, Shea B, Small S, Sweitzer B, Thompson B, Vander Vleit M. 1995. Systems analysis of adverse drug events. Journal of the American Medical Association 274(1):35–43.
Linden L, English K. 1994. Adjusting the cost-quality equation: Utilizing work sampling and time study data to redesign clinical practice. Journal of Nursing Care Quality 8(3):34–42.
Martin K. 2002 (September 24). The Work of Nurses and Nursing Assistants in Home Care: Public Health, and Other Community Settings. Paper commissioned by the Institute of
Medicine Committee on the Work Environment for Nurses and Patient Safety and presented to the Committee.
Medicare Payment Advisory Commission. 2001. Report to the Congress: Medicare Payment Policy. [Online]. Available: www.medpac.gov/publications/congressional_reports/Mar01%Entire%20report.pdf [accessed February 24, 2003].
Mitchell P, Shortell S. 1997. Adverse outcomes and variations in organization of care delivery. Medical Care 35(Supplement 11):NS19–32.
Morris J, Murphy K, Nonemaker S. 1995. Long Term Care Resident Assessment Instrument User’s Manual, Version 2.0. Baltimore, MD: Health Care Financing Administration, U.S. DHHS.
Narumi J, Miyazawa S, Miyata H, Suzuki A, Kohsaka S, Kosugi H. 1999. Analysis of human error in nursing care. Accident Analysis & Prevention 31(6):625–629.
Needleman J, Buerhaus P, Mattke S, Stewart M, Zelevinsky K. 2002. Nurse-staffing levels and the quality of care in hospitals. The New England Journal of Medicine 346(22):1715–1722.
Norrish B, Rundall T. 2001. Hospital restructuring and the work of registered nurses. Milbank Quarterly 79(1):55.
O’Shea E. 1999. Factors contributing to medication errors: A literature review. Journal of Clinical Nursing 8(5):496–505.
Pabst M, Scherubel J, Minnick A. 1996. The impact of computerized documentation on nurses’ use of time. Computers in Nursing 14(1):25–30.
Perrow C. 1984. Normal Accidents. New York, NY: Basic Books.
Prescott P, Phillips C, Ryan J, Thompson K. 1991. Changing how nurses spend their time. Image 23(1):23–28.
Reason J. 1990. Human Error. Cambridge, UK: Cambridge University Press.
Reason J. 1997. Managing the Risks of Organizational Accidents. Burlington, VT: Ashgate Publishing Company.
Reason J. 2000. Human error: Models and management. British Medical Journal 320(7237): 768–770.
Rhoades J, Krauss N. 2001. Chartbook #3: Nursing Home Trends, 1987 and 1996. [Online]. Available: http://www.meps.ahrq.gov/papers/cb3_99-0032/cb3.htm [accessed October 4, 2003]. Rockville, MD: Agency for Healthcare Research and Quality.
Ritter-Teitel J. 2002. The impact of restructuring on professional nursing practice. The Journal of Nursing Administration 32(1):31–41.
Roberts K. 1990. Managing high reliability organizations. California Management Review Summer:101–113.
Roberts K, Bea R. 2001. When systems fail. Organizational Dynamics 29(3):179–191.
Rogers A. 2002. Work Hour Regulation in Safety-Sensitive Industries. Paper commissioned by the Institute of Medicine Committee on the Work Environment for Nurses and Patient Safety. Washington, DC: IOM.
Rousseau D, Libuser C. 1997. Contingent workers in high risk environments. California Management Review 39(2):103–123.
Rubenstein L, Chang B, Keeler E, Kahn K. 1992. Measuring the quality of nursing surveillance activities for five diseases before and after implementation of the drug-based prospective payment system. In: Patient Outcomes Research: Examining the Effectiveness of Nursing Practice. Proceedings of the State of the Science Conference. Bethesda, MD: NIH, National Center for Nursing Research. Washington, DC: U.S. Government Printing Office.
Seago J. 2001. Nurse staffing, models of care delivery, and interventions. In: Shojania K, Duncan B, McDonald K, Wachter R, eds. Making Health Care Safer: A Critical Analysis of Patient Safety Practices. Evidence Report/Technology Assessment No. 43. Rockville, MD: AHRQ.
Service Employees International Union. 2001. The Shortage of Care: A Study by the SEIU Nurse Alliance. Washington, DC: Service Employees International Union.
Shindul-Rothschild J, Berry D, Long-Middleton E. 1996. Where have all the nurses gone? Final results of our patient care survey. American Journal of Nursing 96 11:25–39.
Silber J, Williams S, Krakauer H, Schwartz S. 1992. Hospital and patient characteristics associated with death after surgery: A study of adverse occurrence and failure to rescue. Medical Care 30(7):615–627.
Smeltzer C, Hines P, Beebe H, Keller B. 1996. Streamlining documentation: An opportunity to reduce costs and increase nurse clinicians’ time with patients. Journal of Nursing Care Quality 10(4):66–77.
Steel K, Gertman P, Crescenzi C, Anderson J. 1981. Iatrogenic illness on a general medical service at a university hospital. The New England Journal of Medicine 304(638-42):100–110.
The HSM Group. 2002. Acute Care Hospital Survey of RN Vacancy and Turnover Rates. Chicago, IL: American Organization of Nurse Executives.
Thomas L. 1983. The Youngest Science: Notes of a Medicine-Watcher. New York, NY: The Viking Press.
Trossman S. 2001. The documentation dilemma: Nurses poised to address paperwork burden. The American Nurse 33(5):1, 9, 18.
Unruh L. 2002a. Nursing staff reductions in Pennsylvania hospitals: Exploring the discrepancy between perceptions and data. Medical Care Research and Review 59(2):197–214.
Unruh L. 2002b. Trends in adverse events in hospitalized patients. Journal for Healthcare Quality 24(5):4–10.
Upenieks V. 1998. Work sampling: Assessing nursing efficiency. Nursing Management 29(4):27–29.
Urden L, Roode J. 1997. Work sampling: A decision-making tool for determining resources and work redesign. Journal of Nursing Administration 27(9):34–41.
Wakefield B, Wakefield D, Uden-Holman T, Blegen M. 1998. Nurses’ perceptions of why medication administration errors occur. MEDSURG Nursing 7(1):39–44.
Walston S, Kimberly J. 1997. Reengineering hospitals: Evidence from the field. Hospital and Health Services Administration 42(2):143–163.
Walston S, Burns J, Kimberley J. 2000. Does reengineering really work? An examination of the context and outcomes of hospital reengineering initiatives. Health Services Research 34(6):1363–1388.
Walters J. 1992. Nurses’ perceptions of reportable medication errors and factors that contribute to their occurrence. Applied Nursing Research 5(2):86–88.
WHO (World Health Organization). 2002. Quality of Care: Patient Safety. [Online]. Available: www.who.int/gb/EB_WHA/PDF/WHA55/ea5513.pdf [accessed February 6, 2003].