The Intensive Care Specialist

Physician staffing in intensive care has not been rigorously studied. The literature is largely anecdotal or observational, usually detailing changes in costs and outcomes after planned changes in critical care staffing or configuration. Changes in physician staffing were usually accompanied by other alterations, for example, the introduction of a critical care team or an ICU director. Simultaneous changes in case mix or severity of illness require adjustment in statistical results. The definition of physician staffing varies from an intensivist doing daily rounds (often in collaboration with the primary care team) to a closed 24-hour critical care service ( Table 41-1 ). Both the American College of Critical Care and the Society of Critical Care Medicine recommend intensivist coverage 24 hours per day 7 days per week. However, with increasing numbers of ICU beds and decreasing numbers of trainees selecting the field of critical care, achieving 24-hour coverage has proven to be challenging. Although the idea of a 24-hour intensivist is appealing, the necessity for this approach may be questionable. Recently, Wallace and colleagues found that night-time intensivist staffing in low-intensity daytime-staffed ICUs was associated with a reduced mortality rate. However, this benefit was not seen in those units with a high-intensity daytime staffing model. A low-intensity model was one in which consultation with an intensivist was optional.

Different styles of critical care service that involve the intensivist may or may not use external physician consultants, may envelop consultation services such as nutrition or pharmacy, and may operate quite differently but carry the same “intensivist” label. Attention should also be paid to specialist nurse training, nurse-to-patient ratios, and the presence or absence of certified nurse practitioners.

Li and colleagues looked at outcomes and interventions in a community hospital ICU before ( n = 463) and after ( n = 491) the introduction of an ICU physician. There was a significant reduction in adjusted hospital mortality rate (adjusted for reason for admission, age, and mental status) after the change, with a concomitant increase in the use of invasive monitors.

Pollack and colleagues studied ICU mortality rates, the use of monitoring and therapeutic modalities, and efficiency of ICU bed utilization in the 3 months before ( n = 149) and after ( n = 113) the appointment of a pediatric intensivist and daytime ICU team. There was a clear improvement in the efficiency of bed utilization after the arrival of the intensivist. There was a reduction in the number of admissions for monitoring and for patients with low severity of illness and a parallel increase in therapeutic and monitoring interventions in the postintensivist period. Mortality rate, adjusted for case mix, was reduced in the intensivist period by 5.3% (number needed to treat to prevent one death [NNT], 19; odds ratio [OR], 0.51; 95% confidence interval [CI], 0.16 to 1.67).

Reynolds and colleagues studied outcomes in patients with septic shock in the year before ( n = 100) and after ( n = 112) the introduction of a critical care service, staffed by intensivists. A significant reduction was seen in the hospital mortality rate from 74% to 64% (absolute risk reduction [ARR], 10%; NNT, 10; OR, 0.46; 95% CI, 0.26 to 0.83), after introduction of the critical care service. The use of invasive monitors also significantly increased, but the number of external consultations did not change.

Brown and Sullivan performed a cohort analysis of patients admitted to the ICU before ( n = 223) and after ( n = 216) the introduction of an intensivist operating in an open model. The intensive care mortality rate decreased from 28% to 13% (ARR, 15%; NNT, 6.6; OR, 0.40; 95% CI, 0.25 to 0.66). The hospital mortality rate decreased from 36% to 25% (ARR, 11%; NNT, 9; OR, 0.59; 95% CI, 0.39 to 0.90). This effect was consistent irrespective of the severity of illness.

Hanson and colleagues undertook a cohort study comparing two parallel models of critical care. One group of patients was looked after by an on-site critical care team, supervised by an intensivist. The other cohort was managed by a surgical team, supervised by a general surgeon, that had commitments outside the ICU. Despite having higher Acute Physiology and Chronic Health Evaluation (APACHE) II scores, patients cared for by the critical care team spent less time in the SICU, had fewer complications, used fewer resources, and had lower total hospital charges. No significant difference was found in hospital or ICU mortality rates. Selection bias may have been an issue with this study.

Samuels and colleagues examined the impact of the implementation of a neurointensivist-led neurocritical care team on the discharge disposition of those patients ( n = 703) retrospectively found to have subarachnoid hemorrhage. Patients cared for after the change ( n = 386) were significantly more likely to be discharged home (25.2% versus 36.5%; p < 0.001) and less likely to be discharged to a rehabilitation facility (42.5% versus 32.4%, p < 0.01) than those admitted before ( n = 317) the service was installed. Shortcomings included the retrospective nature of the study and the prolonged (7-year) period of data collection, making it likely that many things other than the institution of a critical care team changed.

Blunt and Burchett compared outcomes in ICUs covered by intensivist versus nonspecialist consultants (anesthesiologists) covering multiple sites using standardized mortality ratios. The case mix–adjusted hospital mortality rate of intensive care patients improved significantly in the intensivist group compared with the nonspecialist group (standardized mortality ratios, 0.81 versus 1.11; OR, 0.73; 95% CI, 0.55 to 0.97).

Dimick and colleagues and Pronovost and colleagues, using similar methodology, studied outcomes after high-risk surgery in the state of Maryland via a large database. After esophageal resection, lack of daily rounds by an ICU physician was associated with longer lengths of stay (7 days; 95% CI, 1 to 15; p = 0.012), higher hospital costs (61% increase or $8839; 95% CI, $1674 to $19,192; p = 0.013), and increased frequency of postoperative complications. After aortic repair surgery, not having daily rounds by an ICU physician was associated with a threefold increase in the in-hospital mortality rate (OR, 3.0; 95% CI, 1.9 to 4.9) and in major postoperative complications, such as cardiac arrest (OR, 2.9; 95% CI, 1.2 to 7.0), acute renal failure (OR, 2.2; 95% CI, 1.3 to 3.9), and sepsis (OR, 1.8; 95% CI, 1.2 to 2.6). Thus daily rounds by an intensive care physician are efficient, effective, and economical.

Reriani and colleagues examined the impact of mandatory versus on-demand intensivist care on long-term patient mortality rates and quality of life. Baseline quality of life surveys were reviewed on discharge and again at 6 months. The baseline characteristics between the two groups did not vary greatly according to their respective APACHE III scores. After the institution of a 24-hour intensivist, no difference was seen in long-term survival rates of medical ICU patients. However, this same group had previously demonstrated that the change in staffing was associated with improved processes of care and staff satisfaction, as well as decreased ICU complication rates, hospital LOS, and hospital cost. In these two previous studies, there was no change in ICU or hospital mortality rates.

Numerous other studies have haphazardly appeared in the literature in abstract form. Pronovost and colleagues have completed a systematic review to include these data. ICU physician staffing was divided into low intensity (no intensivist or elective intensivist consultation) or high intensity (mandatory intensivist consultation). High-intensity staffing reduced the risk of ICU mortality (pooled relative risk [RR], 0.61; 95% CI, 0.50 to 0.75), hospital mortality (RR, 0.71; 95% CI, 0.62 to 0.82), and ICU and hospital LOS, regardless of whether it was adjusted for case mix.

Levy and colleagues studied the impact of intensive care specialists on hospital mortality rate using a large database (Project IMPACT) that had been designed to address resource use in 123 ICUs across the United States. The study was performed by intensivists using a database constructed by intensivists. Patients who were managed by intensive care specialists had greater severity of illness than those managed by the primary physician and they underwent more procedures. When outcomes were adjusted for illness severity and a propensity score was used, patients cared for by intensive care specialists had greater in-hospital mortality rates than those who were not. Critical care predicted the hospital mortality rate with a crude OR of 2.13 ( p < 0.001). The addition of SAPS II (a severity of illness scoring system) to this model reduced this OR to 1.42 ( p < 0.001). Further inclusion of the propensity score decreased the OR to 1.40 ( p < 0.001). Several potential limitations to this study should be noted. The study tests two different hypotheses. The first looked at outcomes, depending on whether an intensivist was chosen by the primary physician. This likely resulted in selection bias because chosen patients were likely to be less severely ill and intensivists were presumably consulted because of clinical concerns. The second study involved more robust groups: critical care for the entire stay (18,618 patients, critical care medicine [CCM] group) versus no critical care (22,870 patients, no CCM group), presumably because of lack of availability. The CCM group was more likely to be at academic medical centers in urban locations, indicating that selection bias, which included racial background, chronic health problems, and socioeconomic status, may have had an impact. Another form of selection bias may have been evident—that of the units themselves. It is likely that there is a cohort of nursing-led ICUs that may function at a very high level of care. This may result from strict adherence to protocols and guidelines, with meticulous attention to infection control and involvement in, and submission to, national benchmarking databases (such as Project IMPACT). Thus this study may illuminate the effectiveness of an elite group of ICUs, absent an intensive care specialist, that through tight organizational controls may have better outcomes.

In conclusion, the majority of studies have demonstrated that availability of an intensive care specialist may reduce mortality rate, LOS, and costs in intensive care. Interestingly, impressive epidemiologic data show that intensive care outcomes for many diagnoses are improving. This may reflect the overall increase in awareness of critical illness; improved vertical integration between emergency medicine, medicine, surgery, and anesthesia; and a problem-oriented, systems-based approach to medical education and practice.

Young and Birkmeyer have estimated that full implementation of intensivist-model ICUs would save approximately 53,850 lives each year in the United States. Conversely, Levy and colleagues have suggested that management of patients in “choice” ICUs by intensivists and in units with full critical care management of patients, compared with a no-intensivists model, may be associated with worse outcomes. No clear explanation for the adverse outcomes in this patient subgroup has emerged. However, it is worth noting that the presence of an intensive care specialist alone is not a “critical care service” and that improved outcomes may result from an integrated model of specialist and multidisciplinary team care, strategic management, and tight organizational structure.

Intensive Care Organization

As previously noted, the introduction of intensive care specialists is one part of a system, usually referred to as a critical care service. A critical care team, led by an intensivist and including residents, fellows, nurse practitioners, respiratory therapists, and a pharmacist, provide 24-hour care to the patient. This may be in full collaboration with the primary care team (the open model) or may replace that team as primary caregivers (the closed model).

Baldock and colleagues prospectively studied 1140 patients admitted into a mixed medical–surgical ICU over a 3-year period, during which time resident medical staff and a closed configuration were introduced. The ICU mortality rate was reduced from 28% to 19% (ARR, 9%; NNT, 11; OR, 0.61; 95% CI, 0.42 to 0.89). The hospital mortality rate was reduced from 36% to 24% (ARR, 12%; NNT, 8; OR, 0.54; 95% CI, 0.38 to 0.77).

Carson and colleagues studied change from an open ( n = 121) to a closed ( n = 124) format in a medical ICU. APACHE II scores indicated that patients admitted after closure of the unit were significantly sicker. Mortality rates increased after unit closure. However, the ratio of the actual mortality rate to the predicted mortality rate was lower in this system. Resource utilization remained similar, which is surprising in view of the increase in the severity of illness. Consequently, this article suggests the cost-effectiveness and probable clinical effectiveness of the closed unit format.

Ghorra and colleagues retrospectively studied the conversion of an SICU from an open ( n = 125) to a closed ( n = 149) format. Again, primary care was provided by an intensive care team. There was a significant reduction in mortality rate, from 14% to 6% (ARR, 8; NNT, 12; OR, 0.38; 95% CI, 0.17 to 0.88,), and in complications from 56% to 44% (ARR, 12; NNT, 8). This was accompanied by a reduction in the number of consultations (from 0.6 to 0.4 per patient). The incidence of renal failure and the use of low-dose dopamine were higher in the open format, reflecting outdated approaches to critical illness.

Multz and colleagues retrospectively looked at outcomes in a community hospital before and after conversion to a closed ICU model and prospectively compared outcomes with a nearby hospital’s open ICU. Although no significant differences in mortality rate were found in either arm of this underpowered study, there was a significant reduction in ICU LOS (retrospective, 6.1 versus 9.3 days; p < 0.05; prospective, 6.1 versus 12.6 days, p < 0.0001), hospital LOS (retrospective, 22.2 versus 31.2 days; p < 0.02; prospective, 19.2 versus 33.2 days; p < 0.008) and days of mechanical ventilation (retrospective, 3.3 versus 6.4 days; p < 0.05; prospective, 2.3 versus 8.5 days; p < 0.0005).

Treggiari and colleagues studied outcomes for patients with acute lung injury in open versus closed ICUs. A total of 24 ICUs were evaluated, and complete data were available for 23; 13 units were closed and 11 were open. The hospital mortality rate was improved significantly in the closed versus open units (adjusted OR, 0.68; 95% CI, 0.53 to 0.89; p = 0.004). The presence of a consulting pulmonologist, presumably with critical care training and thus an “intensivist,” did not appear to confer benefit in open ICUs.

Cooke and colleagues conducted a secondary analysis of the data presented by Treggiari and colleagues that examined the effect of a closed staffing model on tidal volume in patients with acute lung injury. The authors reviewed day 3 tidal volumes in open and closed units and found that those patients in closed ICUs received tidal volumes that were 1.40 mL/kg predicted body weight (PBW) lower than patients in open model ICUs (95% CI, 0.57 to 2.24 mL/kg PBW). Patients in closed ICUs were more likely (OR, 2.23; 95% CI, 1.09 to 4.56) to receive lower tidal volume (6.5 mL/kg PBW or less) and were less likely (OR, 0.30; 95% CI, 0.17 to 0.55) to receive a potentially injurious tidal volume (12 mL/kg PBW or greater) compared with patients cared for in open ICUs, independent of other variables.

Using data from a prospective cohort study, Nathens and colleagues looked at mortality rates in trauma patients across 68 ICUs. After adjustment for differences in baseline characteristics, the relative risk of death in intensivist-model ICUs was 0.78 (95% CI, 0.58 to 1.04) compared with an open ICU model. The effect was greatest in the elderly (RR, 0.55; 95% CI, 0.39 to 0.77), in units led by surgical intensivists (RR, 0.67; 95% CI, 0.50 to 0.90), and in designated trauma centers 0.64 (95% CI, 0.46 to 0.88). It is worth noting that in this study, as in other studies of SICUs, high-volume surgical centers are more likely to have intensivists, and these factors may reinforce one another.

Petitti and colleagues assessed the association between the change to a closed-unit, intensivist-led system and mortality in injured patients at an urban Level I trauma center. A total of 18,918 patients were admitted to the ICU during periods of preintensivist, partial intensivist, and full-intensivist care. Mortality for patients older than age 65 years in the partial intensivist period was decreased relative to the preintensivist period (OR, 0.51; 95% CI, 0.31 to 0.84, p < 0.05); however, no added benefit was seen with the addition of a full-time intensivist. Changing to a closed unit configuration brought about improved survival rates in patients with less severe injuries and patients older than 65 years, but no improvement was seen in the survival of the group as a whole.

Tai and colleagues retrospectively studied quality of patient care and procedure use in a MICU over two 3-month periods before ( n = 112) and after ( n = 127) change in unit organization. In the first period, an open model prevailed. In the second, an intensivist provided daytime care, acting as primary physician and gatekeeper, with rotational medical cover at night. There was a reduction in median LOS. Interestingly, the use of invasive monitors increased from 0% to 24% for arterial lines and from 0% to 5.5% for pulmonary artery catheters, without evidence of improvements in outcomes.

The introduction of a physician–manager for intensive care services (ICU director) has become universal. However, significant variability exists in the director’s day-to-day involvement in medical care, protocols, bed management, and audit.

Manthous and colleagues studied outcomes and educational standards in a medium-sized community hospital in the year before ( n = 459) and after ( n = 471) the appointment of a director of critical care. The ICU mortality rate was reduced from 21% to 15% (ARR, 6%; NNT, 16; OR, 0.66; 95% CI, 0.47 to 0.93). This reduction in mortality rate was consistent for most disease processes and severity of illness. In addition, a significant reduction was seen in the hospital mortality rate from 34% to 25% (ARR, 9%; NNT, 11; OR, 0.63; 95% CI, 0.48 to 0.84). There was a concomitant reduction in mean stays in the ICU (from 5.0 ± 0.3 days to 3.9 ± 0.3 days; p < 0.05) and in the hospital (from 22.6 ± 1.4 days to 17.7 ± 1.0 days), along with an improvement in standard of knowledge of residents.

Mallick and et al examined a 1991 survey by the Society of Critical Care Medicine of nearly 3000 ICUs to determine the effectiveness of the role of the ICU director. They concluded that significant involvement of the ICU director in the day-to-day operation of the unit reduced inappropriate bed occupancy, thus improving efficiency. Strosberg and colleagues questioned nurse managers from 137 ICUs on the involvement of ICU directors in bed management at their hospitals. This revealed a perception of limited nocturnal availability, even though many hospitals had ICU directors.

Zimmerman and colleagues looked at organizational issues in nine ICUs and determined that superior organization was characterized by a patient-centered culture, strong medical and nursing leadership, effective communication and coordination, and open, collaborative approaches to solving problems and managing conflict. They failed to equate superior organization to improved risk-adjusted survival rates.

Shortell and colleagues examined risk-adjusted mortality rates in 42 ICUs involving 17,440 patients using APACHE III. They found that high-quality organization was associated with a lower risk-associated mortality rate, lower risk-adjusted LOS, lower nurse turnover, and higher patient and family member satisfaction. Examples of organizational excellence included technological availability, lack of diagnostic diversity, and caregiver interaction comprising the culture, leadership, coordination, communication, and conflict management abilities of the unit.

A large European study of ICU organization, EURICUS-1, published in 1998, looked at the organizational characteristics of 89 ICUs in 12 European countries. It was determined that the optimal model of ICU organization—where the strategic apex of shared medical-nursing administration lies within the ICU—existed in only 12% of ICUs studied. Further, there was no clear concept of “intensive care,” little planning or purposeful organization, and few defined objectives.

In the pediatric ICU setting, Nishisaki and colleagues conducted a retrospective study to monitor the impact of a transition from a 12-hour ( n = 10,182) to a 24-hour ( n = 8520) attending physician coverage model of in-hospital pediatric critical care. They found that implementation of 24-hour in-hospital pediatric critical care attending coverage was associated with a shorter duration of mechanical ventilation (median, 42 hours versus 56 hours; p < 0.001) and a shorter length of ICU stay (median, 2 days [interquartile range, 1 to 4] versus 2 days [interquartile range, 1 to 5]; adjusted p < 0.001). However, there was no difference in unit mortality (2.2% versus 2.5%; p = 0.23).

The Leapfrog group has proposed that intensive care services provided by telemedicine, involving an intensive care specialist covering several ICUs from a remote location, constitute a reasonable surrogate for a full-time intensivist. This has been a widely embraced approach to alternative intensivist staffing, and some outcome benefit has been demonstrated. Breslow and colleagues showed that tele-ICU services improve outcomes (reduced hospital mortality rate, 9.4% versus 12.9%; RR, 0.73; 95% CI, 0.55 to 0.95) and reduce LOS (3.63 days [95% CI, 3.21 to 4.04] versus 4.35 days [95% CI, 3.93 to 4.78]). This approach should be envisioned as complementing and extending organized ICU services rather than manifesting an alternative model for critical care service delivery.

Telemedicine has been touted as a viable option to alleviate the increased demand for intensivist presence in ICUs. The data that exist on the impact of telemedicine indicate decreased mortality rates and ICU LOS. However, some studies report conflicting results.

Willmitch and colleagues examined the institution of a telemedicine service in five separate hospitals and 10 ICUs. Charts of 24,566 patients were reviewed retrospectively for the baseline year and 3 years after telemedicine implementation. The results demonstrated statistically significant decreases in severity-adjusted hospital LOS of 14.2%, ICU LOS of 12.6%, and relative risk of hospital mortality of 23% in a multihospital health care system.

Young and colleagues conducted a meta-analysis on the impact of telemedicine ICU coverage on in-hospital mortality rates, ICU LOS, and hospital LOS. A total of 41,374 patients were included in the meta-analysis, and tele-ICU coverage was associated with a reduction in the ICU mortality rate (OR, 0.80; 95% CI, 0.66 to 0.97; p = 0.02). There was no change in the overall in-hospital mortality rate. Similarly, tele-ICU coverage was associated with a reduction in ICU LOS (mean difference –1.26 days; 95% CI, –2.21 to –0.30; p = 0.01) but not in-hospital LOS.

Lilly and colleagues performed a prospective stepped-wedge clinical practice study of 6290 adults admitted to both MICUs and SICUs. These patients were then monitored before and after the institution of an adult telemedicine unit. The hospital mortality rate was 13.6% during the preintervention period compared with 11.8% during the tele-ICU intervention period. The tele-ICU intervention period compared with the preintervention period was associated with higher rates of best clinical practice adherence as well as shorter hospital LOS (9.8 versus 13.3 days).

Evidence-based literature increasingly supports the value of telemedicine on ICU outcomes, but the actual volume of data supporting claims of lower mortality rates and decreased LOS is limited. Some fear that telemedicine will draw intensivists away from rural settings and toward more academic centers that are capable of supporting such programs. This change may exacerbate ICU staffing issues in rural areas and in smaller community hospitals.

In conclusion, the conversion of ICUs from open to closed formats and the appointment of an ICU medical director appears to confer modest benefits in terms of mortality rate, morbidity, resource utilization, and LOS. At least in part, these outcome benefits relate to more advanced critical care built on the intensivist model. Although telemedicine’s fate remains unknown, it may well be a feasible option to offset the work hour burden of the 24-hour intensivist model.