Chapter 16. Workforce Issues

In many discussions of patient safety, it is assumed that the workforce is up to the task—in training, competency, and numbers. In these formulations, a combination of the right processes, information technology, and culture is enough to ensure safe care.

However, as any frontline worker can tell you, neglecting issues of workforce sufficiency and competency omits an important part of the equation. For example, a nursing or physician staff that is overworked and demoralized will breed unsafe conditions, even if good communication, sound policies, and efficient computers are in place.

In this chapter, I will discuss some key issues in workforce composition and organizational structure, including the nursing workforce, Rapid Response Teams, and trainee-related matters such as duty-hour restrictions and the so-called “July effect.” I'll close with a discussion of the “second victim” phenomenon: the toll that errors take on caregivers themselves. In the next chapter, I'll discuss issues of training and competency.

Much of our understanding of the interaction between workforce and patient outcomes and safety comes from studies of nursing. The combination of pioneering research exploring these associations,1–5 a U.S. nursing shortage that emerged in the late 1990s, and effective advocacy by nursing organizations (because most hospital nurses in the United States are salaried and employed by the hospitals, they have a strong incentive to advocate for sensible workloads; contrast this to physicians, most of whom are self-employed and therefore calibrate their own workload) has created this focus.

Substantial data suggest that medical errors increase with higher ratios of patients to nurses. One study found that surgical patients had a 31% greater chance of dying in hospitals when, on average, a nurse cared for more than seven patients. For every additional patient added to a nurse's average workload, patient mortality rose 7%, and nursing burnout and dissatisfaction increased 23% and 15%, respectively. The authors estimated that 20,000 annual deaths in the United States could be attributed to inadequate nurse- to-patient ratios.1 A recent study, the most methodologically rigorous to date, confirmed the association between low nurse staffing and increased mortality, further demonstrating that high rates of patient turnover were also associated with mortality even when average staffing was adequate.5

Unfortunately, these studies point to a demand for nurses that cannot be met by the existing supply. Despite some easing of the U.S. nursing shortage, in 2012 the demand for nurses continued to exceed supply by over 100,000; this mismatch was projected to grow (to a shortfall of more than 250,000 registered nurses) with the aging of the American population.6 While more young people have entered the nursing profession in recent years, nearly one million of the nation's nurses (nearly one in four) are over age 50, adding to the challenge.

Efforts to address the nursing shortage have centered on improved pay, benefits, and working conditions. In the future, technology could play an important role as well, particularly if it relieves nurses of some of their paperwork burden and allows them to spend more time with patients. In Chapter 13, I described the lunacy that results when a nurse takes a digital blood pressure reading and then begins an error-prone struggle to transcribe it in multiple places. Situations like this, which go on dozens of times during a typical shift, are one reason many nurses believe so much of their time and training are wasted.

Importantly, even as the nursing shortage has catalyzed efforts to improve pay, hours, staffing, and technology, it has also brought long-overdue attention to issues of nursing culture and satisfaction. Studies in this area often point to problematic relationships with physician colleagues. In one survey of more than 700 nurses, 96% said they had witnessed or experienced disruptive behavior by physicians. Nearly half pointed to “fear of retribution” as the primary reason that such acts were not reported to superiors. Thirty percent of the nurses also said they knew at least one nurse who had resigned as a result of boorish—or worse—physician behavior, while many others knew nurses who had changed shifts or clinical units to avoid interacting with particular doctors.7–9 These concerns have been an important driver for the interdisciplinary training discussed in Chapter 15, and for increasing efforts to enforce acceptable behavioral standards among all healthcare professionals (Chapter 19).10,11

Driven by the strong association between nurse staffing and outcomes, 15 states (including California, Texas, and Illinois) have legislated minimum nurse-to-patient ratios (usually 1:5 on medical-surgical wards and 1:2 in intensive care units), and 16 states restrict mandatory overtime. The jury is still out on whether these legislative solutions are enhancing safety. Some nurses complain that ancillary personnel (clerks, lifting teams) have been released in order to hire enough nurses to meet the ratios, resulting in little additional time for nurses to perform nursing-related tasks. Commentators have pointed out that the determination of adequate nurse staffing is complex and fluid, since it needs to take into account patient acuity, availability of adequate support staff, and a host of other factors that may vary shift-by-shift.12,13

In addition to these staffing standards, there has been an effort to develop quality and safety measures linked to the adequacy of nursing care. These “nursing-sensitive” measures include rates of falls, urinary tract infections, and decubitus ulcers.14,15 While the Joint Commission assesses nurse staffing and working conditions in its accreditation process, many hospitals aspire to the even higher standards of the American Nurses Credentialing Center's “magnet status.” Limited research suggests an association between magnet status and patient outcomes.16

Analyses of deaths and unexpected cardiopulmonary arrests in hospitals often find signs of patient deterioration that went unnoticed for hours preceding the tragic turn of events. Measures of these so-called “failure to rescue” cases are included among the AHRQ Patient Safety Indicators (Appendix V).17–19 Identifying warning signs may require better monitoring systems or improved nurse staffing.

But such cases may also reflect a cultural problem: analyses of failure to rescue cases sometimes reveal that a nurse noticed that something was awry but was either unable to find an appropriate responder or reluctant to bypass the traditional hierarchy (i.e., the standard procedure to call the private physician at home before the in-house ICU physician, or the intern before the attending). Table 16-1 lists a variety of causes of failure to rescue.19

The concept of Rapid Response Teams (RRTs) was developed to address these problems. Now sometimes called “Rapid Response Systems” (RRSs) to emphasize the importance of both the monitoring and response phases, such teams have been widely promoted over the past decade, most notably when they were included among the six “planks” of the Institute for Healthcare Improvement's “100,000 Lives Campaign” in 2005 (Table 20-2).20,21

Despite the face validity of RSSs, the evidence supporting them remains limited. While several single-center, before-and-after comparisons found a reduction in hospital cardiac arrests,19,22–24 a large randomized multisite study demonstrated no impact.25 Two meta-analyses have also questioned the benefit of RRSs.26,27

At the present time, the optimal criteria to trigger an RRT response are unknown.19 Some institutions use strict objective criteria, while others emphasize that anyone with concerns should feel free to activate the RRT. Some hospitals even allow patients and families to call the team. One Australian hospital's poster illustrating its vital sign criteria for activating its Medical Emergency Team (another name for RRT) is shown in Figure 16-1.19

Similarly, the optimal composition of the responding team remains uncertain. Some hospitals have used physician-led teams, while other RRTs are led by nurses, often with respiratory therapists as partners.19 RRTs have a positive impact on nursing satisfaction and perhaps retention, although the magnitude of this benefit has not been fully quanified.28

While the Joint Commission does not specifically require the presence of an RRT, it does mandate that accredited organizations select “a suitable method that enables health care staff members to directly request additional assistance from a specially trained individual(s) when the patient's condition appears to be worsening.”29 Ongoing research will hopefully answer the many questions surrounding RRTs, including the cost-effectiveness of the intervention. An RRT with round-the-clock dedicated staffing can easily cost a hospital hundreds of thousands of dollars per year.

For now, RRTs remain attractive in theory, and their underlying rationale—that hospitals should have robust systems to monitor patients for deterioration as well as a structure and culture that ensure that appropriately qualified staff and necessary technology are available to care for deteriorating patients—seems unassailable. That said, the absence of hard evidence of benefit is a concern, and RRTs have become Exhibit A in the debate among patient safety experts regarding the level of scientific evidence required before a so-called “safe practice” is promoted or required (Chapter 22).30–32

Amazingly, there is virtually no research linking safety or patient outcomes to the workload of practicing (i.e., nontrainee) physicians. Dr. Michael DeBakey, the legendary Texas heart surgeon, was once said to have performed 18 cardiac bypass operations in a single day! The outcomes of the patients (particularly patient number 18) have not been reported. But the implicit message of such celebrated feats of endurance is that “real doctors” do not complain about their workload; they simply soldier on. With little research to question this attitude and with the dominant payment system in the United States continuing to reward productivity over safety, there have been few counterarguments to this machismo logic.

The one area in which this has changed is in the workload of physician trainees, particularly residents. At the core of residency training is the ritual of being “on call”—staying in the hospital overnight to care for sick patients. Bad as things are now, they have markedly improved since the mid-1900s, when many residents actually lived at the hospital (giving us the term “house officer”) for months at a time. Even as recently as 20 years ago, some residents were on call every other night, accumulating as many as 120 weekly work hours—which is particularly remarkable when one considers that there are only 168 hours in a week.

Although resident call schedules became more reasonable over the past generation, until recently many residents continued to be on call every third night, with workweeks of more than 100 hours. But change has not come easily. As efforts to limit house staff shifts gained momentum, some physician leaders argued that traditional schedules were necessary to mint competent, committed physicians. For example, in 2002 the editor of the New England Journal of Medicine wrote that long hours:

… have come with a cost, but they have allowed trainees to learn how the disease process modifies patients' lives and how they cope with illness. Long hours have also taught a central professional lesson about personal responsibility to one's patients, above and beyond work schedules and personal plans. Whether this method arose by design or was the fortuitous byproduct of an arduous training program designed primarily for economic reasons is not the point. Limits on hours on call will disrupt one of the ways we've taught young physicians these critical values … We risk exchanging our sleep-deprived healers for a cadre of wide-awake technicians.33

Therein lies the tension: legitimate concerns that medical professionalism might be degraded by “shift work” and that excellence requires lots of practice and the ability to follow many patients from clinical presentation through work-up to denouement, balanced against worries regarding the effects of fatigue on performance and morale.34–37 The latter concerns are well grounded. One study showed that 24 hours of sustained wakefulness results in performance equivalent to that of a person with a blood alcohol level of 0.1%—above the legal limit for driving in every state in the United States.38

Although early researchers felt that this kind of impairment occurred only after very long shifts, we now know that chronic sleep deprivation is just as harmful. Healthy volunteers performing math calculations were just as impaired after sleeping 5 hours per night for a week as they were after a single 24-hour wakefulness marathon.39 The investigators in this study didn't check to see what happened when both these disruptions occurred in the same week, the norm for many residents operating under traditional training models.

Although it defies common sense to believe that sleep-deprived brains and bad medical decisions are not related, hard proof of this intuitive link has been surprisingly hard to come by. Most studies using surgical simulators or videotapes of surgeons during procedures show that sleep-deprived residents have problems with both precision and efficiency, yet studies of nonsurgical trainees are less conclusive. One showed that interns who averaged less than 2 hours of sleep in the previous 32 hours made nearly twice as many errors reading ECGs and had to give each tracing more attention, especially when reading several in a row.40 Another found that medical interns performed better on a simulated clinical scenario after a 16-hour shift than after a traditional extended duration shift.41 Yet other studies showed that tired radiology residents made no more mistakes reading x-rays than well-rested ones, and sleepy ER residents performed physical examinations and recorded patient histories with equal reliability in both tired and rested conditions.42

In 2003, the Accreditation Council for Graduate Medical Education (ACGME, the group that blesses the 7800 residency programs in the United States, involving over 100,000 trainees) stepped into the breach, limiting residents to a maximum of 80 hours per week, with no shift lasting longer than 30 consecutive hours and at least one day off per week (Table 16-2).43 Studies that followed these changes yielded surprisingly mixed results, with the weight of the evidence showing no significant change in patient safety or overall outcomes, but improved quality of life for residents.43–47

Notwithstanding the ambiguous evidence, pressure on the ACGME to do more continued to mount, driven by public and professional advocacy and by the results of a prominent study that found that residents on ICU rotations who worked in shifts averaging 16 hours committed fewer errors than those working traditional overnight shifts.43,48 In 2008, the Institute of Medicine released a study recommending that overnight shifts for all residents be eliminated and that round-the-clock supervision be required.49 In 2010, the ACGME responded by codifying the IOM's recommended supervision requirements, but prohibiting overnight shifts only for interns (first-year residents), not all residents (Table 16-2).43,50

As with the 2003 regulations, time will tell whether these changes, which went into effect in July 2011, will have the desired effects. We already know that they are unpopular with training program directors, who fear that they will soon begin to graduate undertrained physicians.37 Residents, interestingly, are conflicted—confident that the new rules will improve their own quality of life but skeptical that they will improve their training and patient safety.47 This debate is likely to continue, particularly since most other developed countries have duty-hour limits far more stringent than even the new U.S. standards (Table 16-3).49

In discussions regarding duty hours and fatigue, analogies are often made to other industries that have far more severe limits on consecutive work hours. For example, U.S. commercial pilots cannot fly for more than eight hours straight without relief, and truck drivers must pull off the road after 10 hours for a mandatory break. But between these shifts, the machines either sit idle or are operated by another crew. Fumbles may occur in the handoffs of the machines but they are unusual, and no one hands over the reins in the middle of a critical function (such as an airport landing approach). In medicine, on the other hand, handoff errors are routine (Chapter 8). This means that the real patient safety question is not whether fatigued doctors make more mistakes (they probably do), but whether the errors caused by fumbled handoffs and more subtle information seepage will exceed those caused by fatigue.

Moreover, a full accounting of the safety impact of duty-hour limits must also weigh the quality of care that physicians deliver after they enter full-time practice, and the substantial costs (well into the billions of dollars) to pay for new nonteaching services staffed by hospitalists, nurse practitioners, and physician assistants.51 Finally, there is another, more subtle risk of the new standards: as the New England Journal editor feared, the new rules will contribute to a change in culture among physicians, from a “do whatever it takes” mindset to a “shift work” mentality. Residents themselves remain torn by their understandable desire for more rest and their deeply felt ethical commitment to their patients.52,53

Even as we ponder the implications of the duty-hour changes, there can be little debate on one issue: 100-hour workweeks must be bad for young doctors (and older ones too, although no one is regulating their hours presently), their loved ones, and ultimately their patients. Because of this, some form of duty-hour limits are welcome, as long as they are accompanied by active efforts to make handoffs safer, to relieve residents of superfluous paperwork, to provide decent rest facilities in hospitals, to hire alternate providers to pick up the slack, and to promote professionalism among trainees. Moreover, these changes must be accompanied by rigorous research to assess both their benefits and harms.

Though not as vocal as the debate over duty hours, another potential patient safety risk associated with trainees has received more attention of late: the safety of patients cared for by new residents. Dubbed the “July effect” in the United States and the more macabre “August killing season” in the United Kingdom,54 the evidence supporting the association between new residents and harm has long been debated. However, a 2011 systematic review of 39 studies by Young et al. found that the more methodologically rigorous studies did, in fact, find increased mortality (averaging about 8%) and decreased efficiency (longer lengths of stay).55 The July effect could likely be mitigated by a number of interventions, including improved supervision, a limitation on trainees' workload in the first few weeks, and staggered starting times for new residents.

Young and his colleagues (including me) also recently raised concerns about another trainee-related transition: the year-end handover of outpatients that occurs in teaching settings.56 Residents typically accrue large ambulatory panels, and then abruptly hand these patients off to new interns at the completion of training. With this handoff, patients experience a vast drop in the experience level of their physicians—both clinical experience and practical knowledge regarding the inner workings of the healthcare organization. We estimate that nearly 13,000 trainees hand their clinic populations off every July, so this handoff affects about 2 million unsuspecting patients.

The solutions to this problem are likely to be similar to those described above, including improved supervision and limited initial schedules. But here, the safety issues surrounding handoffs also assume great importance (Chapter 8), which means that fixes will also involve Web-based ambulatory sign-out systems and mandatory discussions between outgoing residents and incoming interns, at least regarding the most active and challenging patients. This issue may well be addressed by healthcare system or training program accreditors in the not-so-distant future.57

It is self-evident that hospitals are 24-7-365 enterprises, yet few hospitals staff the nights and weekends the way they do daytimes and weekdays. This is not completely without rationale: while the patients don't change, staffing needs partly relate to patients' requirements for transport or the intensity of testing and consultations. If more is being done to and for patients during the day or on weekdays because of the greater availability of personnel and equipment, richer staffing on these shifts may be appropriate.

That said, there are likely to be other explanations for lighter staffing on nights and weekends, relating to caregiver preferences or constrained hospital budgets. Such understaffing could easily put patients at risk on these “off hours.” One hospital CEO put it well, after making a series of late-night administrative rounds.58 “I've come to appreciate the fact that I work in two distinct places,” the CEO, Dr. David Shulkin, observed, a “day hospital” and a “night and weekend hospital.” He continued:

The weekday hospital has a full administrative team, department chairs and service chiefs, experienced nurse managers, and a full complement of professional staff. The off-hours hospital, on the other hand, rarely, if ever, has senior managers present. Nurse-to-patient ratios are significantly lower. Even the number of residents is considerably lower—certainly lower than during my days of training—because of mandated work-hour restrictions.

Empirical data generally support Shulkin's concerns. The majority of studies that have scrutinized the outcomes of care on weekends and nights have documented higher death rates, higher readmission rates, and more surgical complications.59–63 Moreover, survival rates after in-hospital cardiac arrest are lower, and medical mistakes are more common.64,65 The evidence regarding the hazards of nights is slightly less robust than that of the weekends, but still of concern.66,67

The challenge will be to find the resources to provide adequate staffing on nights and weekends. In areas such as radiology or critical care, telemedicine may provide a partial answer by making physician expertise available without requiring on-site presence. In others, physicians will need to provide in-house coverage. Many hospitals have augmented their weekend and night staffing with a combination of hospitalists and intensivists. Selected groups of patients might benefit from being managed in specialized centers that offer adequate staffing at all times. For example, a recent study found that while stroke patients fared worse when admitted on weekends to general hospitals, this difference was eliminated when patients were admitted to specialized stroke centers.68

Some experts have argued for a more regulatory approach, and several states do in fact regulate nurse-to-patient ratios (see earlier discussion); these standards typically do not vary by time of day or day of the week. Ultimately, an integrated approach—including adequate staffing, appropriate oversight of trainees, and thoughtful use of technology, guided by evidence and supported by appropriate transparency and, when needed, regulatory or accreditation standards—will be needed to address the “two hospitals” problem.

While the healthcare system, correctly and understandably, focuses on the patient and family after a serious error, caregivers also require empathy and attention. In a 2000 article, Albert Wu of Johns Hopkins coined the term “second victim” to describe how caregivers themselves are affected after they commit an error.69 In a 2011 interview with me, he expanded on this concept:

I think that there are two ways that healthcare workers are traumatized. The first are things we do to ourselves; these are self-inflicted wounds that result from internalized judgments .... we feel anxious, depressed, and demoralized. On the other hand, some healthcare workers are actively victimized by elements of the system or by other organizations. Those are people who work in places that still believe in shame-and-blame or crime-and-punishment, and they're sometimes official bodies like nursing boards and so forth. They actually punish people and, I think, add insult to those self-inflicted injuries .... Some people have persistent problems—the incidents go on to cause lasting harm. Some people are so depressed that they need to be hospitalized; some people are suicidal. People quit medicine or nursing because of what they did …. In our hospital ... a boy with head trauma came in and was badly injured. He needed to be intubated, could not be intubated and could not be trached, and died in the emergency room. It was just a horrific incident and everyone felt unimaginably bad. But even though the emergency physician got some support, others were a little critical of his actions, and he quit clinical medicine within the year.70

Addressing the second victim issue is not easy. Colin West of the Mayo Clinic suggests that an opportunity be given to the caregiver to process the error, such as in an informal small group or a traditional M&M conference.71–73 Whichever forum is chosen, it is critical that participants model the importance of error acknowledgment and open discussion. Emotional support provided by a trusted colleague or a professional counselor is also helpful; West recommends that this person should not be part of the clinician's performance evaluation team.

While such counseling can be obtained informally after an error, too often it is not, leaving caregivers to suffer in silence.74,75 Because of this, some institutions have developed programs designed to sensitize all staff to the issues of second victims, including understanding that errors happen to every clinician and perfection is not an attainable goal. Such educational programs should arm clinicians with coping strategies, and caution against maladaptive strategies that are often employed.

A few programs go even further. One of the most advanced is the second victim support team (the forYOU Team) developed by Susan Scott and colleagues at the University of Missouri.76,77 Based on interviews with clinicians after adverse events, the Missouri investigators identified six stages in the second victim recovery process: (i) chaos and accident response, (ii) intrusive reflections, (iii) restoring personal integrity, (iv) enduring the inquisition, (v) obtaining emotional first aid, and (vi) moving on.78 To support caregivers through these stages, after a serious adverse event a trained supervisor or peer is immediately dispatched to offer support (such trained individuals are embedded on every high-risk service). The goal is to prevent clinicians from going home before they receive initial counseling. From there, referrals are made for additional support. Any investigation of the incident, including root cause analysis if appropriate, is conducted by different individuals than those on the support teams.

A few other institutions have instituted similar programs.79–81 Early responses from clinicians to these programs are positive and the programs clearly fill an important gap. However, additional research is needed to determine the optimal structure, financing, and timing of second victim programs, and how to ensure that caregivers receive needed support while also allowing for rapid and effective incident investigations and responses (Chapter 14).

• Increasing attention is being paid to the importance of a well-trained, well-rested workforce to patient safety.
• In nursing, good studies have linked longer work hours, lower nurse-to-patient ratios, and higher patient turnover to poor outcomes.
• Many hospitals have organized specialized teams of caregivers available to respond to unexpected patient deteriorations (RRTs). While the concept has face validity, studies of this strategy have yielded mixed results.
• Although research has linked long hours and fatigue in physicians to errors, it has not yet led to widespread efforts to regulate physician staffing. The exception is in residency training programs, where regulations now limit weekly hours and shift lengths. These limits have improved house staff well-being, but the impact on safety or education is not yet clear.
• Attention is turning to the safety hazards of hospital care on nights and weekends, and to the risk of receiving care early in the training year in teaching hospitals.
• Clinicians involved in serious errors are deeply affected by the experience. Several programs designed to support such “second victims” have been launched.