Introduction

Hospitals and physicians are constantly searching for the most effective and efficient use of resources, while still providing exceptional patient care. Previously, the benefits of an intensivist for patients and the community have been debated [1]. Few studies have examined the effect of an intensivist in a neurocritical care (NCC) unit [2–5]. Subspecialized intensive care units (ICU) have become much more common recently; these include those which focus on neurological, cardiac, and trauma patients [6]. Having these specialized ICUs is more expensive; therefore, it is crucial to demonstrate the overall benefits that these specialized units provide to the hospital as well as to the patient [7].

Previous observational studies have been published which have evaluated the structure and staffing models of ICU’s [4, 6]. The purpose of these studies was to identify the potential benefit from intensivist staffing on a variety of patient outcomes in an ICU setting. Knopf et al. found that when observing critically ill stroke patients, neurointensivists had a positive impact on the outcomes of patients who had experienced a subarachnoid hemorrhage [6]. A study by Hanson et al. demonstrated the benefits of an intensivist by comparing cohorts that were admitted into a surgical ICU. This study compared two groups, one of which was cared for by an on-site intensivist-led team and the other by a general surgeon led team. The intensivist-led group was found to have improvements in several key metrics when compared to the general surgeon led group. These advantages included fewer patient complications, shorter length of stay (LOS), less resource utilization, and decreased hospital charges [8].

A systematic review by Pronovost et al. included twenty-six studies that focused on intensivist staffing models as well as their effect on outcomes. This meta-analysis demonstrated that ICU staffing with an intensivist improves a variety of clinical outcomes. Intensivists were shown to reduce mortality as well as decrease overall ICU and hospital LOS [9]. Additional studies by Suarez et al. have suggested that a neurocritical care team was associated with a decrease in hospital mortality and LOS without affecting readmission rates [10]. A study by Samuels et al. determined that patients with aneurysmal subarachnoid hemorrhage were more likely to be discharged home when treated by a neurointensivist-led neurocritical care team [11]. Varelas et al. identified improvements in outcomes for patients with head trauma when the unit was converted from an open to a semi-closed model, and the patients were managed by an intensivist-led team [3]. Although several studies have observed the effects of an intensivist on specific disease management and outcomes, to date there are limited data on the overall effect of an intensivist for patients in a NCC unit.

Florida Hospital Orlando has a 20 bed NCC unit that treats patients with a variety of neurological and neurosurgical disorders. Historically, over 1500 patients are admitted annually to the NCC unit and capacity averages 90 %. In November 2010, due to capacity issues and in an effort to optimize patient care and resources, the NCC unit converted to a partially closed model. This model required all patients to be managed by an intensivist. Previously, intensivist management was optional, except in the case of ventilated patients who were required to be managed by an intensivist. There were no other significant changes in staffing or personnel which occurred during this time period.

The purpose of this study was to identify patient and community benefits of a new requirement for mandatory intensivist management upon admittance into the NCC unit.

More specifically, we wanted to demonstrate that mandatory intensivist management in a NCC unit decreases LOS. Our secondary objectives were to prove that mandatory intensivist management maintains patient outcomes and benefits the community by decreasing service line closure rates. This was accomplished by comparing Acute Physiology and Chronic Health Evaluation III (APACHE III) scores, the Florida Hospital Emergency Transfer Center (ETC) decline rates, mortality, and outcomes during the two time periods.

Methods

Patients admitted into the NCC unit during the study period were included. The inclusion criteria time frame was November 1, 2009 through October 31, 2010 and January 1, 2011 through December 31, 2011. We included the 12-month time period before the establishment of mandatory intensivist management in the NCC unit and excluded November 1, 2010 through December 31, 2010 in order to allow for full implementation of the new requirement. This pre-implementation time period was then compared with 12 months during which time intensivist management was mandatory.

In order to collect a range of comparable data points, patients were included if they were established in the APACHE III Version J Methodology database at Florida Hospital Orlando. Patients are included in the APACHE III database if they meet all APACHE III criteria. The criteria require patients be in the ICU for more than 8 h for APACHE III predictive data to be calculated. A patient cannot arrive from a non-APACHE ICU or from another hospital’s ICU. APACHE data are calculated every month, starting on the 1st and ending on the last day of the month. To avoid excluding patients from the APACHE data, the report is run 15 days after the end of the month to ensure a more comprehensive dataset. The list of APACHE patients admitted during our two time periods was 1551 patients from November 1, 2009 to October 31, 2010 and 1702 patients from January 1, 2011 to December 31, 2011.

Once the patient lists were obtained, patients who did not meet age inclusion criteria (<18 years old) were excluded. During our study period, there were 0 non-predicted patients in the APACHE data. Therefore, no patients were excluded due to APACHE scores. The patient lists were sent to our patient financial services (PFS) team to collect data regarding patient LOS, discharge outcomes, demographics, and whether or not the patient was managed by an intensivist. Florida Hospital collects these data on all patients based on their discharge. Once the inclusion lists were compiled, the data were sent to the research team. Research staff manually verified all included patients to ensure that datasets from both APACHE and PFS encompassed the same patients. The APACHE patient list was established based on admissions (1551 vs. 1702), while PFS collects patient data based on discharge (1500 vs. 1751). The datasets varied by two patients overall. These two patients were lost in the data collection system with PFS for unknown reasons. Although the APACHE and PFS datasets varied in subset numbers, the two datasets were not used to compare similar data points, i.e., APACHE III data were used to analyze average APACHE III scores by month, mortality, NCC LOS, number of patients per month, ventilator days, and average acute physiology scores (APS) by month. PFS data were used to analyze hospital LOS, and discharge outcomes. Three percent of patients (49 of 1751 patients), who were defined as pre-intervention based on the APACHE groups, admitted between November 1, 2009 and October 31, 2010, were included in the post-intervention group in the PFS analysis due to their discharge date. The effect of this overlap was not significant.

Furthermore, we examined hospital LOS using data from PFS, which is based on patient discharge. For this dataset, we were only able to compare years, as monthly data were not available. We analyzed the LOS based on patients discharged in 2010 (January–December) with patients discharged in 2011 (January–December) in order to compare the impact on the same length of time (12 months). Consequently, this decreased the number of patients included in our analysis, 1266 patients in 2010 and 1720 patients in 2011.

Data were collected through several existing hospital resources due to the multiple data points being collected. In addition to APACHE and PFS, the service line closure rates were provided by the Florida Hospital ETC. This allowed us to identify the percentage of patients who were declined admission because of capacity issues. To identify infection rates, the Florida Hospital Infection Prevention Department was provided the list of patients. An infection prevention scorecard was then created based on patients that experienced infections: central line-associated blood stream infections (CLABSI), catheter-associated urinary tract infections (CAUTI), and ventilator-associated events (VAE).

Patient discharge status between the two time periods was compared, using the discharge data obtained from the PFS dataset (1500 patients before mandatory intensivist management and 1751 patients after). Discharge status was grouped into good outcome, poor outcome, and other. Good outcomes included patients discharged in one of the following categories: home, self-care, home health service, discharge/transferred to rehab facility, and against medical advice. Poor outcomes included skilled nursing facility (SNF), expired, hospice-home, and hospice-medical facility. The other category included short-term hospital, discharged/transferred to long-term care hospital, discharged/transferred to an intermediate care facility (ICF), discharged/transferred to another type of health care institution, and discharged to a psychiatric hospital or psychiatric distinct part of a hospital.

All data were compiled into several Excel spreadsheets separating the PFS data points, the APACHE data points, discharge outcomes, percentage of intensivist managed patients, ETC data, and infection rates. The data were de-identified and sent to the Florida Hospital Biostatistician. The analysis software used was SPSS version 21. A variety of statistical tests were utilized based on the type of each data element. The study sample size was selected to compare the average NCC LOS using 80 % power with a critical difference of one-half day using a one-tailed comparison with alpha set to 0.05. All tests used alpha 0.05 to define statistical significance.

Results

Prior to the new staffing model, 772 of 1500 (51.47 %) patients were managed by an intensivist. After implementation of mandatory intensivist management, this rose to 1709 of 1751 patients (97.60 %) (Fig. 1). To show that mandatory intensivist management improves efficiency while maintaining patient outcomes, we examined the differences in NCC and hospital LOS, APACHE III scores, APS scores, patient mortality, infection rates, and patient outcomes between the two time periods. The number of patients admitted each month, as well as the percentage of patients declined admission via the ETC, was also analyzed to determine the impact of mandatory intensivist management on hospital resource utilization.

Fig. 1
figure 1

Based on patient consults provided by PFS; p < 0.01

Full size image

APACHE III scores and APS were both comparable between the two time periods. No statistical differences were found for either APACHE score averages (45.0 vs. 43.4, p = .11) (Fig. 2) or APS score averages (32.3 vs. 31.0, p = .11) before or after implementation of mandatory intensivist management. This indicates that acuity levels were unchanged.

Fig. 2
figure 2

APACHE III scores were based on APACHE III monthly averages; not statistically significant. Mortality numerical difference based on APACHE III monthly averages; not statistically significant

Full size image

NCC LOS was tested using the Mann–Whitney test, which was selected as a non-parametric test due to the small sample size and non-normal distribution of values. The comparison of LOS for the two time periods showed a statistically significant decrease in LOS from 4.6 to 3.7 days, (p < 0.01) (Fig. 3). This equates to a savings of 1532 ICU days. Additionally, hospital LOS decreased by 0.9 days from 9.8 days to 8.7, (p < 0.01), which paralleled the decrease in NCC LOS (Fig. 3).

Fig. 3
figure 3

Based on the monthly averages provided by APACHE III data; p < 0.01. Based on patient hospital LOS provided by PFS; p < 0.01

Full size image

The rates of ventilator-associated events (VAE), central line-associated blood stream infections (CLABSI), and catheter-associated urinary tract infections (CAUTI) each month were totaled and compared between the two time periods. No statistical differences were found in any of the infection rates, which were based on monthly average rates (per 1000 device days). For VAE, there were no infections in either time period and device days declined, although not significantly (1795 vs. 1749 days, p = 1). For CAUTI, there were less total infections and more device days, but these were not statistically significant (infections: 27 vs. 23 and device days: 6491 vs. 6707, p = .76). For CLABSI, there was one more infection in the second time period and less device days, but these also did not reach statistical significance (infections: 2 vs. 3 and device days: 4512 vs. 4272, p = .76). Infection rates thus remained consistent and the quality of care given was maintained. There was also no statistical difference in ICU mortality based on the monthly counts reported in the APACHE III data. The average monthly deaths did, however, decrease from 8.5 to 7.2 (p = .35). For both time periods, mortality remained below the national predicted monthly totals of 9.4 for 2010 and 10.3 for 2011 (Fig. 2).

Monthly NCC patient ventilator day averages for 2010 and 2011 were also examined from the APACHE dataset using the Mann–Whitney test. There was no statistical difference found as the average ventilator days increased slightly from 4.2 to 4.3, (p = .84). This was not unexpected because during the first time period hospital policy mandated that all ventilated patients be managed by an intensivist.

No statistical differences were found in discharge outcome groups when comparing time periods before and after mandatory intensivist management. Good discharge outcomes increased from 65 to 65.3 %, poor outcomes increased from 29.7 to 30.7 %, and other outcomes decreased from 5.3 to 4.0 %, (p = .18, χ 2 test). Therefore, the outcomes remained consistent between the two time periods (Fig. 4).

Fig. 4
figure 4

Based on PFS data for each patient; not statistically significant

Full size image

Another observed impact of mandatory intensivist management was an increase in the unit’s efficiency, which resulted in an increase in admissions. The number of NCC admissions based on APACHE III counts increased from 129 patients per month to 142 patients per month, (p = .02, Mann–Whitney test analysis). This also resulted in a decrease in the percentage of patients that were declined admission due to service line closure from 12.36 to 5.66 %, (p = 0.02). A non-parametric test to compare the averages was computed.

Discussion

The primary aim of this study was to examine the impact of implementation of mandatory intensivist management on patient outcomes and the NCC unit LOS. During the initial time period with an open unit model, only 51.47 % of patients in the NCC unit were managed by an intensivist. After mandating intensivist management, the percentage of patients cared for by an intensivist rose to 97.60 %. In the literature, several studies have shown the benefits of an intensivist staffing model, which include improved patient outcomes, such as fewer complications and shorter LOS [7–9]. This study demonstrates the impact of intensivist management in a NCC unit. Our data showed that the introduction of mandatory intensivist management reduced both NCC and overall hospital LOS. This allowed for an average of nearly 13 additional monthly admissions into a 20 bed NCC unit which often has capacity issues. The NCC LOS stay was decreased by nearly a full day from 4.6 to 3.7 days, (p < 0.01). The reduced LOS was maintained in our analysis of hospital LOS for these patients—9.8 vs. 8.7 days, (p < 0.01). This suggests that the impact seen in the hospital LOS was a result of the interventions in the NCC unit and not as a result of other factors. The net result of this decrease in LOS is a savings of 1532 ICU days. Furthermore, based upon the best available data, we have (mortality rates, APACHE scores, APS, and infection rates), the decrease in LOS did not seem to be at the expense of patient care as no differences in discharge outcomes, mortality, or infection rates were noted. There was, in fact, a slight trend toward lower monthly mortality averages after the implementation of mandatory intensivist management. The average monthly deaths decreased from 8.5 to 7.2, (p = .35), remaining below the national predicted monthly deaths of 9.4 and 10.3, respectively.

Furthermore, we wanted to assess the effect that this policy had on patient admissions and service line closure rates. We have shown that implementation of mandatory intensivist management will allow for more admissions into a NCC unit and decrease the time that the service line is closed due to capacity issues. In our study, a total of 151 more patients were seen in the NCC unit during the second time period as compared to the first. This demonstrates more efficient utilization of NCC beds, a limited resource, which allowed for a decrease in patients declined admission because of a closed service line. The percentage of patients that were declined decreased from 12.36 to 5.66 % with mandatory intensivist management. The results of this study demonstrate that mandatory intensivist management can improve NCC LOS and maintain optimal patient care. In addition, intensivist management can allow a NCC unit to better serve the community, by minimizing the number of patients declined admission to a comprehensive stroke center. By increasing admissions and more efficiently utilizing NCC beds, a hospital should benefit from a financial standpoint.

Previous studies have shown intensivist management of ICU patients to be associated with decreased complications [8] and improved mortality [9]. In our study, there was no statistical difference in complications or mortality. One potential reason for this would be that a significant number of patients (51.47 %) were already managed by intensivists. As the hospital had a standing policy that intensivists manage ventilated patients, it is reasonable to surmise that the higher acuity patients were already being seen by an intensivist. Therefore, some of the benefits of intensivist management were already being realized. In addition, mortality rates were already below APACHE predicted and showed a non-significant decrease with the addition of mandatory intensivist management. Likewise, CAUTI, CLABSI, and VAE were rare events during both time periods so it proved difficult to show any statistical improvements in these parameters.

Limitations

Our study was limited in several aspects. Data collected for this study were performed retrospectively and analyzed from two different sources: APACHE III and PFS. APACHE III analyzed patients based on admissions and PFS analyzed patients based on their discharge. Both datasets examined the same set of patients, but there was a difference in the number of patients that were grouped in the pre- and post-categories, 1551/1702 (APACHE III) versus 1500/1751 (PFS). Consequently, 49 of 1751 patients (3 %) who were defined as pre-intervention based on the APACHE groups were included in the post-intervention group. Additionally, due to only 12 months in each time period for APACHE data, we needed to complete non-parametric analysis and visual comparison for several of the data points. To further demonstrate the difference in the role of the intensivist pre- and post-intervention, it would have been informative to examine the differences in APACHE scores of the patients who were and were not managed by an intensivist in the pre-intervention period. However, one of the limitations of using APACHE data is that it is based on monthly averages of patients admitted, not individual data. Therefore, we were unable to analyze the differences in APACHE scores between these two groups. This would have likely confirmed our assumption that during the first time period intensivists were primarily involved in the care of the most critically ill patients. In fact, we found no statistical difference in patient ventilator day averages between the two time periods. This was likely due to the fact that during the pre-intervention period ventilated patients were managed by an intensivist. We were also unable to assess any impact diagnosis may have had on the differences we found in LOS pre- and post-intervention, although the typical admission diagnoses in our NCC unit did not change during the study period. The most common admission diagnoses included stroke, brain tumor, seizure, spinal injury, and traumatic stupor and comas.

Conclusions

In summary, implementation of mandatory intensivist management in the NCC unit decreased LOS, allowed for an increase in NCC unit admissions, and decreased the percentage of patients declined admission because of capacity issues. Overall, quality of care remained consistent throughout the studied period. These results occurred without any significant new polices or procedures in the NCC unit. The fact that over 3000 patients were included in the study lends strength to the findings. This study did not review one specific diagnosis, rather the entire patient population admitted into the NCC unit during the inclusion dates. We demonstrated that mandatory intensivist management can improve LOS and utilization of a scarce commodity—subspecialty ICU beds. Additional studies are needed to determine the effect of intensivist management in subspecialty intensive care units.