Introduction

The intensive care unit remains a large user of hospital resources, accounting for 25% to 30% of hospital costs, representing 1% to 2% of the gross national product (between $40 and $80 billion per year) despite the fact that intensive care unit beds represent only 5 to 10% of hospital beds [₁,₂,₃,₄,₅,₆]. It is well known that a small percentage of high-risk patients account for the largest proportion of surgical intensive care costs [₇]. In 1984, Li and his associates demonstrated that the presence of on-site physicians, who were not intensivists, improved utilization and decreased mortality in an intensive care unit in a community hospital [₈]. More recently, Brown and Sullivan have shown, in a retrospective analysis of two matched groups of patients, that a full-time critical care specialist makes a significant impact on the management of critically ill patients in a mixed intensive care unit by reducing overall intensive care unit mortality by as much as 52% and hospital mortality by 31% [₉].

Increased pressure on hospitals for fiscal accountability and an expanding elderly population in addition to the continued growth of life-sustaining technology demand improved management of intensive care services to reduce costs and to improve effectiveness. While there is some evidence that expenditures for intensive care services can be reduced by improving effectiveness and efficiency using an interdisciplinary team approach [₁₀], there are no data on the effect of full-time, on-site surgical intensivists on ICU operating costs, utilization, and risk-adjusted ICU mortality.

The hiring of two surgical intensivists in our Institution created a unique opportunity to study the effect of their presence on resources utilization and patient outcome in an ICU setting. In essence, to compare closed vs. open ICU models. More specifically, a study was designed to investigate the impact of the presence of these two full-time surgical intensivists on the length of stay in the SICU, the number of patients' ventilator days, the number of consultants per patient, and the ICU mortality in an 8-bed surgical intensive care unit of an 800-bed teaching hospital.

Methods

Before the arrival of the two full-time surgical intensivists, the surgical intensive care unit (SICU) was directed by a physician who had only administrative functions. Upon arrival at the institution on August 1, 1993, the first Author of this paper assumed the responsibility of director of the unit. Before enforcing a permanent closed SICU, a decision was made to undertake a study designed to examine utilization of the unit's resources, operating costs, and patients' outcome during three consecutive intervals. August 1st to September 30th 1993 constituted interval 1 (INT1); October 1st to December 31st 1993 and January 1st to March 31st 1994 constituted intervals 2 and 3 (INT2 and INT3), respectively.

During interval 1, the role of the two full-time surgical intensivists was to conduct teaching rounds and undertake administrative functions, but not to change any existing pattern of medical practice, thus mimicking the existing open unit model where patients were admitted often without triage and were cared for by their primary physician. During the two following intervals, the SICU was closed and all patients were managed exclusively by the surgical intensivists, thus allowing the introduction, as a standard of care, of the end-point type of protocols presented in Table 1.

Table 1: Changes implemented in the SICU

Statistical Analysis

Results

Patients' characteristics with regard to age, gender, and mean Apache II score did not differ during the three intervals, as shown in Table 2. There was no difference in the distribution of patients with an Apache II score ≤ 15, between 16 and 30, and greater than 30, as demonstrated in Table 3. The Apache II score was 16.0 ± 8.7 during INT1, and 15.4 ± 8.1 and 14.9 ± 6.9 during INT2 and INT3, respectively (p > 0.05 vs. INT1).

Figure 1

Table II: Patients' demographics and resource utilization data

* p < 0.05 vs INT1

Days on mechanical ventilation, VENTDAYS

The number of VENTDAYS decreased from 5.9 ± 12.5 during INT1 to 2.1 ± 4.0 and 2.6 ± 5.3 during INT2 and INT3, respectively (p < 0.05 vs. INT 1). Similarly, the SICULOS decreased from 9.7 ± 13.5 days during INT1 to 5.4 ± 4.4 and 5.6 ± 4.8 during INT2 and INT3, respectively (p < 0.05 vs. INT1).

Figure 2

Table III: Distribution of patients stratified by Apache II scores

The analysis of VENTDAYS stratified by Apache II scores for all patients is shown in Table 4.

Figure 3

Table IV: Resource utilization and outcome stratified by Apache II scores

* p <0.05 vs INT1

The presence of full-time surgical intensivists significantly reduced the number of VENTDAYS in low-risk patients (Apache II score ≤ 15) and in patients with intermediate acuity of disease (Apache II scores between 16 and 30). In patients with APACHE II scores ≤ 15, VENTDAYS decreased from 2.5 ± 6.5 during INT1 to 1.3 ± 3.5 and 1.3 ± 2.3 during INT2 and INT3, respectively (p < 0.05 vs INT1). Similarly, the number of VENTDAYS in patients with an Apache II score between 16 and 30 decreased from 10.8 ± 15.8 days during INT1 to 3.2 ± 4.4 and 4.6 ± 7.2 during INT2 and INT3, respectively (p <0.05 vs. INT1). In contrast, due to the large standard deviation, the presence of full-time surgical intensivists did not affect the number of VENTDAYS in patients with Apache II score greater than 30. In this group, the number of VENTDAYS was 15.0 ± 24.0 during INT1 and 4.1 ± 4.7 and 11.0 ± 13.0 during INT2 and INT3, respectively (p>0.05 INT3 vs INT1; p <0.05 INT2 vs. INT1).

Length of ICU stay, SICULOS

Mortality

Resources utilization

Tables 5 and 6 summarize resource utilization during each interval for survivors and non-survivors, respectively. A statistically significant improvement in resource utilization, defined as fewer VENTDAYS and shorter SICULOS, was observed in ICU survivors during INT2 and INT3. While there was a trend toward fewer VENTDAYS and shorter SICULOS in non-survivors during INT2 and INT3, the difference did not achieve statistical significance. This indicates that the reduction in VENTDAYS and SICULOS observed during INT2 and INT3 resulted mostly from improved performance in survivors.

Figure 4

Table V: Comparison of resource utilization among survivors

p < 0.05 vs INT1

Figure 5

Table VI: Comparison of resource utilization among nonsurvivors

Comparison of INT1 vs. INT2 and INT3 grouped together

Notwithstanding and fully compatibly with the results presented in the previous sub-sections, a comparison between INT1 vs. INT2 and INT3 grouped as one interval is presented in Tables 7 and 8. Table 7 presents patients SICU and hospital mortality as well as resource utilization data for these 2 groups. Table 8 presents resource utilization and outcome stratified by Apache II scores, again for these 2 groups.

Figure 6

Table VII: Patients' demographics and resource utilization data

Figure 7

Table VIII: Resource utilization and outcome stratified by Apache II scores

Discussion

Despite the emphasis on the importance of education and training in critical care, there is substantial variation in the mode of delivery of intensive care among units in the United States because of significant differences in professional staffing patterns and degree of supervision, whether by part-time or full-time intensivists. There is a growing debate on the impact of full-time critical care specialists on the efficacy, costs, and outcomes in critical care units. Weichsman et al. reported that a multi disciplinary critical care team working in a medical-surgical ICU in a university-affiliated hospital decreases financial costs without an impact on mortality [₁₂]. Similar results were presented by Lima et al.; they found that a full-time critical care specialist decreases ICU costs with no adverse effect on outcome [₁₃]. In contrast, Rafkin et al. showed that 24-hour multi-disciplinary care services improve outcome and length of stay for surgical patients compared with that predicted by Apache II scores [₁₄]. However, until now there was no clear evidence derived from the comparison of actual ICU mortality before and after the presence of full-time surgical intensivists to support the contention that mortality rates will uniformly fall if intensivists devote dedicated time to on-site staffing. Similarly, there was little evidence to support the superiority of the closed unit model over the transitional, or open model with regard to improved patients' ICU outcome.

The results of our study show that on-site staffing of six hours per day clearly improves SICU utilization while decreasing ICU mortality. In contrast to previous work [₁₅], our study showed that full-time intensivists can reduce VENTDAYS and SICULOS, even in low-risk patients. However, as one would expect, ICU mortality in low risk patients is not affected by the presence of full-time surgical intensivists. Most of the improvement in VENTDAYS, SICULOS, and ICU survival, associated with the presence of full-time surgical intensivists, occurred among patients with intermediate likelihood of death. One of the most important aspects of our study probably is the impact of staffing by full-time intensivists on the number of consultants per patient in the unit. During interval 1 (open unit model), for every patient on ventilatory support greater than 48 hours with a pulmonary artery catheter, with high temperature and with a rising creatinine there would be a pulmonologist, cardiologist, nephrologist, and an infectious disease consultant. The same type of patients had no consultants during the subsequent intervals. Of interest is the impact of the full-time surgical intensivists on nonsurvivors; the trend toward a decreased number of VENTDAYS and SICULOS before death reflects the ability of full-time intensivists, communicating daily with families, to forego futile life extending measures such as dialysis in patients with multi-system organ dysfunction syndrome, which only procrastinate the state of dying. Clearly, some of the impact on overall SICULOS was a consequence of this approach to those patients who would have been previously sustained for a prolonged period of time despite an irreversible prognosis.

The impact of full-time intensivists is multifactorial. Obviously, in our study the presence of two full-time surgical intensivists provided improved managerial and organizational structure to the SICU. Additionally, we identified one of the factors associated with superior SICU effectiveness namely, full-time surgical-ICU attendings who dedicate their time to the ICU. The presence of full-time surgical intensivists provided a strong team-oriented culture where nurses and physicians set high standards and support each other. Furthermore, the implementation of written policies and protocols to standardize patients' care had a profound impact of quality of care and outcome. Whether the physical presence of surgical intensivists is necessary for an extended time each day to obtain the results shown in this study is debatable. We believe that those SICUs that have surgical critical care fellows may forego the extended daily presence of surgical intensivists without compromising patients' care or affecting outcome.

Our study corroborates the results of Simms and associates showing that ICU mortality of critically ill surgical patients can be reduced by implementing a closed unit model [₁₆.] Obviously, while the data strongly suggest that the decreased number of ventilator days, length of stay in the SICU, and ICU mortality are the direct result of the surgical intensivists, it is possible that alternate causes may have had an impact. The implementation of specific protocols driving patients'care, rather than the actual presence of intensivists, could explain improved patients' outcome. Obviously, while protocols designed by intensivists can minimize errors and improve patients' care, they cannot by themselves change the organizational and managerial structure of the ICU; the latter have been shown to impact profoundly on unit performance. Why did ICU utilization and ICU mortality improve during intervals 2 and 3? Were the observed improvements the result of purely personal physicians skills and styles rather than of their training and dedication to surgical critical care? Would the effect on ICU utilization and mortality persist, in a post-intensivists experiment? We propose that most of the effects observed in this study were the direct result of the daily hands-on involvement of the full-time surgical intensivists in patients' care, as well as of the administrative and organizational changes that resulted from them.