• 
  

  Most studies addressing staffing of ICUs have had significant limitations, and this literature does not yet provide a consistent view of the best model to use. This subject is complicated by the fact that optimal ICU staffing may depend on ICU characteristics.

  
  
• 
  

  Despite calls for all ICUs to function as closed-model units with intensivists as the primary physician of record, evidence supporting this view is contradictory. Likewise, studies of around-the-clock intensivist presence have not consistently shown that it is associated with superior outcomes.

  
  
• 
  

  The data do not supply a consistent answer to the question of whether ICUs would obtain better outcomes if they added nurses to reduce their patient:nurse ratios.

  
  
• 
  

  Increasingly, nonphysician providers are playing innovative roles in the ICU, and care provided by teams including nurse practitioners or physician assistants appears to be safe and comparable to that provided by other staffing models.

  
  
• 
  

  The conditions of ICU staffing will continue to change under the stresses of shortages of a variety of health care workers relevant to ICU care, and increasing duty hour limitations for physician trainees. Nonphysician providers, innovative staffing models, telemedicine, and other technologies will be increasingly used to cope with these realities.

  
  
• 
  

  Since only quantitative evaluation can tell us whether one staffing model is better than another, we need more research from multiple sites to develop a consistent and integrated understanding of this complex topic.

Like all complex organizations, intensive care units (ICUs) have numerous variable elements of organization and structure, including how they are staffed. Outside of medicine, it is widely accepted that most of the opportunities to improve the performance of complex organizations derive from improving the structures and processes of which they consist. Within this systems-based concept, every aspect of what we do and how we do it is a candidate for study and change, including all aspects of ICU staffing.¹ Though a variety of types of health care workers (HCWs) collaborate in caring for ICU patients,² relatively little is known about the relationships between ICU staffing and outcomes. It is rare for staffing patterns to be the result of an evidence-based assessment of what works best; they usually reflect historical precedents, combined with practical necessities and growth by accretion.

Staffing options can be framed as a number of questions, such as: Who does it?, How many of them are there to do it?, What do they do?, and How do they do it? These address the type, training, experience, and other characteristics of the HCWs; details of work schedules, including workload, duty hours, shiftwork, and coverage for nights and weekends; details of assigned tasks; and interfaces between different types of HCW. Not only is all of this highly complex and interacting, but the optimal staffing structure for a given ICU may well differ based on ICU type, size, case mix, and other differences of baseline structure.

In this chapter, we review existing evidence addressing relationships between ICU staffing and outcomes. We will discuss intensivists, hospitalists, house officers, physician extenders, nurses, respiratory therapists, pharmacists, and telemedicine. As it cannot be assumed that what works best is the same in ICU and non-ICU settings, we will restrict ourselves to data derived from ICUs, and generally from adult ICUs. Although the major focus will be on clinically relevant outcomes for patients, where available we will also include considerations of other outcomes that are important to HCWs and society, such as job satisfaction and costs.³ As highlighted below, this literature has numerous limitations, and at the current time does not supply much guidance on how best to staff ICUs.

Many investigators have tried to address whether ICU patients have better outcomes if there are intensivists (attending physician specialists in critical care medicine) involved in their care, or if they receive a higher “dose” of intensivist care. This literature is difficult to interpret and apply because determining whether intensivist involvement produces better outcomes would require all physicians to round in a similar controlled fashion, and for there to be no other differences in ICU administration, team composition or function. No such experiments have been done.

OPEN VERSUS CLOSED ICUS, AND RELATED TOPICS

Most of these studies relate to “open” versus “closed” ICU structures. Usually, an open ICU has patients cared for by multiple generalists with or without assistance of intensivists, while in a closed ICU the attending physician of record for all the patients is a single intensivist. Open ICUs are more common in the United States and some other countries,^4-6 while closed ICUs are the rule in other areas.^7,8 However, this delineation does not adequately define the intensivist staffing model, as there are countless variations and intermediate models for both types.^6,9 Furthermore, there are often a host of other differences between open and closed ICUs, making it difficult to say that any benefits of closed-model ICUs are due to the involvement of intensivists; for example, closed ICUs are more common in larger, academic hospitals that typically have residents and ICU fellows working in them.^4,6

More than 30 studies have compared outcomes in open versus closed ICUs. Most of these have been from single ICUs, and most used before versus after study design that in all cases changed from open to closed units. These have yielded mixed results. In a systematic review and meta-analysis of 27 studies involving 27,000 patients, Pronovost et al tried to make sense of the organizational diversity of ICUs by dividing intensivist involvement into high versus low “intensity.”¹⁰ High intensity comprised both closed-model ICUs and open ICUs where consultation by an intensivist was mandatory. They found that high-intensity intensivist involvement was associated with lower ICU and hospital mortality (unadjusted, pooled risk ratios of 0.61 and 0.71, respectively) and also shorter ICU and hospital length of stay (LOS). An economic evaluation based on these findings indicated that costs would be lower with high-intensity intensivist involvement.¹¹ In contrast, Levy et al performed a cross-sectional study using the Project IMPACT database, including 101,000 patients in 123 ICUs.¹² Adjusting for severity of illness, they analyzed hospital mortality according to whether or not patients were under the care of an intensivist, without distinguishing whether the intensivist acted as the primary attending physician or a consultant. They reported that patients with intensivists involved in their care had higher hospital mortality (odds ratio [OR] 1.42, p <.001). The accompanying editorial speculated about possible reasons for the disparity between Pronovost’s and Levy’s studies.¹³

AROUND-THE-CLOCK INTENSIVIST PRESENCE

A growing movement around the world has intensivists physically present around-the-clock (24/7), usually effected via shiftwork where different intensivists are present during days and nights. Some have opined that 24/7 intensivist staffing is the ideal.^14,15 Such staffing exists in a minority of North American ICUs,^4,16-18 but is common in some European countries.⁸

The main rationale for 24/7 intensivist presence is reasonable, that at night critically ill patients need as much expert care as during the day. Some have also pointed to data showing that care is worse at night; however, that literature is contradictory, with some studies showing such an effect^19-25 and others not.^26-32 Others have called for 24/7 intensivist presence by suggesting that it will provide better end-of-life care, and benefit trainees and nurses.¹⁵

Despite abundant opinions, there are sparse data on this topic, most of which were observational, or used before versus after designs that suffer from the pitfalls of historical controls.³³ Blunt et al, reported that the standardized hospital mortality ratio after changing to the 24/7 staffing model among 824 patients in a single ICU in the United Kingdom declined from 1.11 to 0.81.³⁴ For 4388 patients in a medical ICU in the United States such a change was not associated with improved ICU survival (10.2 vs 10.4%, p = 0.83), hospital survival (17% vs 19%, p = 0.33), or family satisfaction, though it was associated with some improvements in ICU-acquired complications, processes of care, and reduced intensivist burnout.³⁵ Of note, in this latter situation, the standard ICU staffing model had ICU fellows present overnight. In the only interventional study, which did not use historical controls, Garland et al³⁶ utilized an alternating crossover study design in two closed-model, intensivist-run ICUs, one academic and one in a community hospital without house staff. Inclusion of the community ICU is valuable because it is more comparable to the majority of ICUs than are the large, academic units, which are the subject of most ICU research.^16,37 In this study, 24/7 intensivist presence did not produce better patient outcomes or family satisfaction in either ICU. The main effect of the shiftwork model was on the intensivists, for whom it was associated with lower job and life stresses. In the largest study to date, Wallace et al performed a retrospective, cross-sectional analysis of 49 ICUs participating in the APACHE database project.³⁸ Their sophisticated analysis indicated that nocturnal intensivist presence was associated with lower hospital mortality in ICUs with low-intensity involvement of intensivists in daytime care (OR = 0.62, p = 0.04), but not in those with high-intensity daytime intensivist involvement (OR = 1.08, p = 0.78).

These observations highlight the fact that the impact of 24/7 intensivist coverage may depend on ICU type, and preexisting staffing. Also, it is important to note that 24/7 staffing requires more intensivists, a serious challenge given the worsening intensivist shortage.^39,40 While no complete analysis of costs has been done, Banerjee et al reported that around-the-clock intensivist staffing led to lower direct costs, but only for the sickest patients.⁴¹

INTENSIVIST WORKLOAD

Workload is related to staffing, and there has been concern about intensivists’ workload. This concern derives from data showing that job burnout among intensivists is not rare,^42-44 and that trainees’ perception of high workload discourages them from going into this subspecialty.⁴⁵ Though there has been great attention to reducing the workload of physician trainees, little attention has been paid to the consequences for attending physicians.⁴⁶

Although the European Society of Intensive Care Medicine has stated that the optimal size of an ICU is 8 to 12 beds,⁴⁷ little is known about the workload intensivists should have in order to improve outcomes for patients, and for themselves. In a preliminary study, we found no clear relationship between job burnout and self-reported workload.⁴⁴ Dara et al sought to assess patient outcomes in relation to intensivist workload.⁴⁸ They studied 2492 patients in a medical ICU over 18 months, during substantial changes in ICU size and team composition, such that the ratio of beds per intensivist varied from 7.5 to 15. The results suggested that ICU LOS was longer when the ratio was 15, while hospital LOS and mortality rates were not different. An observational study with many methodologic limitations suggested better patient outcomes when ICU doctors worked 12 hours rather than 8-hour shifts.⁴⁹ In a cluster randomized study in five medical ICUs, Ali et al studied the effect of weekend cross-coverage for intensivists doing half-month rotations.⁵⁰ This form of weekend respite to reduce workload had no detrimental effects on mortality or LOS, but produced less burnout and job distress for the intensivists.

Telemedicine involves having ICU clinicians, who may be physicians and/or nurses, remotely provide real-time care. This may be nighttime only, or both days and nights. Remote clinicians have electronic access to a data stream typically including telemetry, diagnostic tests, information from devices such as ventilators, and if they exist, an electronic medical record and computer order entry.⁵¹ In some systems, software continuously analyzes the data for early identification of worrisome trends. The eClinicians in these eICUs can see patients via video cameras, and talk to HCWs in the ICU via telephone or intercom; they may even have a robotic presence in the ICU.⁵² The remote physicians may or may not have order writing authority. Several hundred hospitals in the United States have implemented ICU telemedicine.⁵³ The psychosocial aspects of working in an eICU environment are substantially different than bedside work.⁵⁴

A number of studies, most commonly assessing the VISICU system, have evaluated how introduction of eICU care influenced mortality, LOS, costs, and complications^55-58; these individual studies have had contradictory results. A recent meta-analysis of 13 studies including 35 ICUs and over 41,000 patients⁵³ identified the limitations of this literature: all had (a) simple before versus after study designs, (b) modest study quality, (c) large heterogeneity in baseline ICU structures and eICU implementation, and (d) potential for bias by virtue of vendor involvement or support for many of these studies.⁵⁹ With these problems in mind, the meta-analysis indicated that eICU implementation led to lower ICU mortality (OR = 0.80, p = 0.02) and ICU LOS (difference 1.3 days, p = 0.01), without concomitant changes in hospital mortality or LOS.

Hospitalists are attending physicians who specialize in the care of hospitalized patients.⁶⁰ Many US hospitals now have hospitalists involved with ICU care.¹⁷ Two studies have assessed outcomes related to these nonintensivists caring for ICU patients. An observational study compared outcomes in two adult medical ICUs, one staffed by intensivists and the other by hospitalists supported by an intensivist-led consultation service.⁶¹ After adjustment for large differences in case mix, there were no significant differences in hospital mortality (OR = 0.80, p = 0.22), ICU mortality (OR = 0.80, p = 0.41), or ICU LOS (mean difference −0.3 days, p = 0.32). In a before versus after study in an intensivist-led pediatric ICU, Tenner et al compared outcomes when night coverage was provided by residents versus hospitalists.⁶² After adjusting for major differences in case mix, nocturnal care by hospitalists was associated with lower ICU mortality (OR = 0.36, p = 0.01) and ICU LOS (mean difference −21 hours, p = 0.01).

Historically, house officers have been a vital part of ICU workforces, functioning under supervision as an extension of attending physicians. A small and disjointed literature has addressed how house officers, that is, residents and critical care subspecialty fellows influence outcomes in ICUs.

A study of the impact of ICU fellows evaluated outcomes in two academic, closed-model, medical-surgical ICUs.⁶³ These units had ICU fellows about half the time, though they always had a full complement of less senior house staff. Results indicate no differences on mortality rates or LOS related to the presence of ICU fellows.

Two studies evaluated outcomes in relation to the level of training of ICU residents. The first reported on 2274 patients in two open-model ICUs in Taiwan that were covered by a single surgical resident.⁶⁴ In an unadjusted analysis, hospital mortality of patients cared for by first-year residents was significantly higher than those cared for by more advanced residents (25 vs 18%, p = 0.002). In a study of 5415 children admitted to 16 pediatric ICUs, mortality was higher in patients cared for by first- and second-year than third-year residents, and was also higher earlier in the educational year for each resident level.⁶⁵

Several authors have addressed work-hour limits that have been increasingly placed on house officers in many countries.⁶⁶ This is of great consequence since teaching ICUs have historically relied on house staff for patient care services, especially overnight. A consequence of these limits is changes in house staff scheduling that reduces continuity of care. Also, teaching hospitals have attempted to fill the gaps by increased use of hospitalists and nonphysician providers.⁶⁶ While a detailed study in two ICUs found that reduced working hours resulted in a lower rate of serious errors by first-year residents,^67,68 a large study of 104 ICUs was unable to detect a change in severity-adjusted mortality attributable to the work rule limits implemented in the United States in 2003.⁶⁹

Nonphysician providers, mainly nurse practitioners (NPs) and physician assistants (PAs), are increasingly involved in the care of ICU patients.^17,70,71 While different by way of background and training (Table 3-1), these two classes of providers have been used, sometimes interchangeably, in the ICU setting in a variety of ways. In some academic ICUs, NPs and/or PAs have been integrated into house staff–based ICU teams^72-74; in others, they have been used to staff entirely separate ICUs.^75,76 Alternatively, NPs have been employed on specialty-based (eg, heart failure,⁷⁷ trauma,^78,79 transplantation⁸⁰) teams, which assist in the care of some ICU patients. Finally, NPs have been added in novel roles as overseers/outcomes managers and to provide unit-based care in previously open-model ICUs.^81-83

Studies assessing the impact of NPs and/or PAs are shown in Table 3-2. Patient morbidity, mortality, and quality of care have been seen to improve with the addition of NPs in novel roles in the ICU. With an NP acting as overseer/outcomes manager, hospital mortality, hospital and ICU LOS, duration of mechanical ventilation, complications (including skin breakdown and urinary tract infections), and costs were all reduced.^81,83 Adding an NP in an open-model surgical ICU resulted in increased adherence to clinical practice guidelines when compared with mandatory critical care consultation without a physician or physician extender tied to the unit.⁸²

At the University of Pittsburgh, the use of NPs as an alternative to ICU fellows has been examined in a step-down ICU. NPs were more engaged in coordination of care and less involved in off-unit projects.⁸⁴ There were no differences in mortality, LOS, or difference in duration of mechanical ventilation.⁸⁵ For patients with respiratory failure and a tracheostomy, there was no difference in LOS, success in ventilator weaning, or ICU readmissions.⁸⁶

Three studies have been published attesting to the comparability of nonphysician providers to house staff in the adult ICU setting. Using a historical controlled design, in 1991 Dubaybo et al reported similar mortality rates for medical ICU patients cared for by house staff compared with PAs.⁸⁷ Two more recent studies also support the notion that care by nonphysician provider-based ICU teams is similar to that provided primarily by house staff. Both of these two studies compared a medical ICU staffed by house staff to another staffed by NPs/PAs operating simultaneously in a single academic institution. Kawar et al reported that patients in the unit staffed by PAs had similar hospital LOS as well as ICU, hospital, and 28-day mortality to those cared for by house staff.⁷⁶ Gershengorn et al found that ICU patients whose care was provided by NPs/PAs versus house staff experienced comparable ICU and hospital LOS, hospital mortality, and discharge destination for hospital survivors.⁷⁵

Additional benefits are potentially realizable with the use of nonphysician providers in the care of ICU patients. First, unlike rotating residents, these providers become a consistent workforce in the ICU. This consistency may improve communication between ICU staff members and, thereby, ICU culture and safety. Improvements in care quality have been associated with better communication⁸⁸ and a more safe work environment.⁸⁹ Second, procedural proficiency is known to increase with practice^90-92; as a consistent presence in the ICU, nonphysician providers will have more opportunity to hone procedural skills than more transient care providers. Further, by being present in the ICU consistently, nonphysician providers will be repeatedly exposed to ICU-specific interventions (eg, continuous renal replacement therapy, mechanical ventilation, extracorporeal membrane oxygenation) for which comfort only comes with practice and experience. Finally, over long periods of time, nonphysician providers in the ICU should become expert in care of critically ill patients. While fewer than 5% of internal medicine house staff choose to pursue a career in intensive care medicine,⁴⁵ ICU-based nonphysician providers have already selected the discipline and are, therefore, likely to be more invested in learning the nuances of diagnosis and management of the critically ill patient.

As with all staffing models, there are several potential downsides of the use of nonphysician providers in the ICU worthy of consideration. First, often NPs and PAs receive less formal in-school education on the pathophysiology, evaluation, synthesis, and presentation of complex medical cases than do medical school graduates. As such, a significant upfront investment of time and resources for on-the-job training of these providers has been used^72,74 and may be needed. Second, as consistent ICU staff members, nonphysician providers may be at high risk for developing job burnout.⁹³ This syndrome is known to be highly prevalent in ICU care providers and, when present, to negatively impact care quality and to push providers to seek other lines of work.⁹⁴ Finally, employment of nonphysician providers can be financially burdensome. The salary of an NP or PA is nearly twice that of a resident.^95,96 Some of this increased cost, however, may be offset by cost savings associated with nonphysician provider-based care.⁷⁷

The nonphysician provider is an often untapped resource, which should be considered for use in the care of adult ICU patients. While data are limited, care provided by NP/PA-based teams appears to be safe and, in many instances, comparable to that provided using other staffing models.

Nurses are the lynchpin of ICU care. Unlike other ICU HCWs, they are with the patients most of the time.⁹⁷ They perform most of the numerous interventions conducted daily on ICU patients.⁹⁸

In this section, we review the existing knowledge about ICU staffing by nurses, and nurse extenders. We will limit our review to ICU-specific data; wards differ from ICUs in many ways (eg, patient types, structures, nursing tasks, prevalent patient:nurse ratios), making it unwise to assume that findings in one area will apply in the other.

Although the purpose of this section is to review the literature on issues related to ICU nurse staffing, the overarching nurse staffing issue is the nursing shortage. This shortage is substantial and worldwide.^99,100 ICUs are not immune to this shortage^101,102; indeed, they may be particularly affected.¹⁰³

PATIENT:NURSE RATIOS

Numerous studies have tried to measure the association between the workload of ICU nurses and clinical outcomes (Table 3-3). Most of these analyzed patient:nurse ratios (PNRs), though a few used the related measure of nurse hours per patient-day.