An Examination of Factors Affecting the Length of Stay in a Palliative Care Center

Abstract

Background:

With the increase in elderly population, life-threatening chronic diseases are increasing, simultaneously increasing the need for palliative care centers (PCCs).

Objectives:

To evaluate the factors affecting the length of stay (LOS) and discharge of patients from a PCC.

Methods:

A retrospective scan was made of the records of patients followed up in the PCC between January 2013 and March 2016. A record was made of patient age, gender, diagnosis, conditions/comorbidities, Glasgow Coma Scale, Karnofsky Performance Scale, LOS, prognosis (exitus or surviving), percutaneous endoscopic gastrostomy (PEG), tracheostomy, mechanical ventilator, nutrition (total parenteral nutrition [TPN] or enteral nutrition), and the results of cultures taken during stay in PCCs (blood, tracheal aspirate, urine, rectal swab, wound). Evaluation with regression analysis was made of the data related to factors thought to have a possible effect on the LOS in PCCs.

Results:

Four hundred thirty-five patients were included in the study, comprising 58.6% men and 41.4% women with a mean age of 70.6 ± 17.2 years. The LOS was 27.2 ± 30.9 days. A total of 234 patients were discharged and 201 (46.2%) were lost to mortality in PCCs. The bacteria most isolated in cultures were Escherichia coli (28.5%) and methicillin resistant Staphylococcus aureus (MRSA) (17%). According to the results of the regression analysis, cancer, hypoxic brain, and advanced age had a negative effect on LOS and PEG, TPN, hypertension, and E. coli, Proteus, Pseudomonas, and Acinetobacter infections increased LOS.

Conclusion:

The results of this study revealed some basic factors that affect LOS in PCCs. However, there may be much variation in the data obtained with the various reasons for which this patient group is admitted to a PCC.

Introduction

The World Health Organization (WHO) defines palliative care (PC) as an approach that improves the quality of life of patients and their families facing problems associated with life-threatening illness.¹ In Turkey, life expectancy is 75.3 years for men and 80.7 years for women.² Together with an increasing aging population, life-threatening chronic diseases and the need for palliative care centers (PCCs) are also increasing.³ Just as in other countries, the elderly in Turkey have many multisystemic problems. There is an urgent need to increase the number and quality of PCC and the number of beds for patients requiring PC and their families. According to the data of the Turkish Public Hospitals Authority (October 2016), there is a total of 2220 PC beds in 197 centers in Turkey and this number is much lower than the number of beds required.⁴ PC, which has started to be seen as a human right, must be accessible, fair, and inclusive for all.⁵ It is inevitable that the spread of PC services will entail a severe economic burden to insurance companies. Defining the patient profile of those in PCC and a better definition of factors affecting the length of stay (LOS) will contribute to productive use of the existing PC beds. The aim of this study was to evaluate the factors affecting the LOS and discharge of patients admitted to PCC.

Materials and Methods

Ethics statement

The study was approved by Ankara Numune Training and Research Hospital Ethics Committee (approval No. 11.25.2015/E-15-653). All procedures were applied in accordance with the principles of the Declaration of Helsinki.

Patients

A retrospective study was made of the records of patients followed up in the PCC of Ulus State Hospital between January 2013 and March 2016.

Exclusion criteria

Patients were excluded if the records were incomplete, if the LOS in PCC was <3 days, or if they had been admitted more than once to the PCC.

Measurements

A record was made for each patient of age, gender, diagnosis, conditions/comorbidities, Glasgow Coma Scale (GCS), and Karnofsky Performance Scale (KPS) that was calculated to evaluate clinical course, LOS, prognosis (exitus or surviving), Percutaneous Endoscopic Gastrostomy (PEG), tracheostomy, mechanical ventilator (MV), and nutrition (total parenteral nutrition [TPN] or enteral nutrition [EN]), and the results of cultures taken during stay in PCC (blood, tracheal aspirate, urine, rectal swab, wound).

The KPS measures health status according to patient's self-care capacity, ability to do daily tasks, and general functioning for survival.⁶ KPS is a measure of 11 classes progressing from 0 to 100 points; On this scale, 0 points = death and 100 points = indicates normal condition without any disease.⁶ In this study, analysis was made by grouping GCS and KPS. Examination was made by separating patients into three groups; the former consisted of mild GCS (GCS: 13–15), moderate GCS (GCS: 9–12), and severe GCS; the latter consisted of mild KPS (KPS: 70–100), moderate KPS (KPS: 40–60), and severe KPS (KPS: 0–30).⁷ The data were obtained from patient records electronically from hospital information management system software (HBYS-Alpdatta Company, Ankara, Turkey).

Statistical analysis

The data recording and statistical analyses were made using IBM SPSS Statistics 21.0 (IBM Corp. Released 2012. IBM SPSS Statistics for Windows, Version 21.0. IBM Corp., Armonk, NY) and MS-Excel 2007 software. Conformity of the data to normal distribution was evaluated with the Shapiro-Wilk test. Quantitative variables were stated as median (minimum-maximum) and mean ± standard deviation. Categorical variables were stated as number (n) and percentage (%). Evaluation with regression analysis was made of the data related to factors thought to have a possible effect on the LOS in PCC, including age, gender, comorbidities, GCS, KPS, tracheostomy, TPN, EN, MV, and bacteria produced in cultures. At each stage of the regression model, the variable entry probability was taken as 0.05 and removal probability as 0.10. To assess multiple correlations, the nonparametric Spearman Rho correlation coefficient was applied in the evaluation of relationships between independent variables. Conformity to normal distribution of the model remainder was examined with the Shapiro-Wilk test. As the LOS variable had a noticeable distribution from the right, logarithmic transformation was applied including a regression model for symmetry (1n LOS). Each variable for LOS was defined by reverse transformation for the coefficients when interpreting the results (e^{Fixed coefficient+β ^(for each variable)}) (Table 5). A value of p < 0.05 was accepted as statistically significant.

Results

Five hundred eighty-three patients were admitted to the PCC of Ulus State Hospital between January 2013 and March 2016. Of these, 40 were excluded because of incomplete records, 53 because of repeated admissions, and 55 because LOS was less than 3 days. A total of 435 patients were included in the study, comprising 255 (58.6%) men and 180 (41.4%) women with a mean age of 70.6 ± 17.2 years. Mean GCS was calculated as 11.3 ± 3.3, and KPS as 39.6 ± 14.4. The LOS was 27.2 ± 30.9 days (Table 1). Two hundred thirty-four (53.8%) patients were discharged and 201 (46.2%) were lost to mortality in PCC. The distributions of the GCS and KPS scores of the patients are shown in Table 2.

Table 1.

Patient Characteristics

Parameter	Median (min; max)	Mean ± SD
Age (years)	75.0 (16.0; 99.0)	70.6 ± 17.2
LOS	17.0 (3.0; 212.0)	27.2 ± 30.9
GCS	12.0 (3.0, 15.0)	11.3 ± 3.3
KPS	40.0 (10.0; 80.0)	39.6 ± 14.4

GCS, Glasgow Coma Scale; KPS, Karnofsky Performance Status Scale; LOS, length of stay in PCC; PCC, palliative care center; SD, standard deviation.

Table 2.

The Distribution of Patients in the Specified Ranges of Glasgow Coma Scale and Karnofsky Performance Status Scale Scores

Parameter	n (%)
GCS
Severe (≤8)	104 (23.9)
Moderate (9–12)	173 (39.8)
Mild (13–15)	158 (36.3)
KPS
Severe (0–30)	165 (37.9)
Moderate (40–60)	246 (56.6)
Mild (70–100)	24 (5.5)

Values are given as number of cases (%).

The most commonly seen diagnoses and comorbidities were hypertension (HT) in 205 (47.1%) patients, cancer in 167 (38.4%), cerebrovascular event in 112 (25.7%), and diabetes mellitus in 97 (22.3%). PEG was applied to 128 (29.4%) patients, tracheostomy to 86 (19.8%), and MV support to 24 (5.5%). All patients received nutritional support, 265 (60.9%) with TPN and 128 (29.4%) with EN (PEG and nasogastric tube). The clinically descriptive data of the patients are shown in Table 3.

Table 3.

Descriptive Information of the Patients

	Present, n (%)
Conditions/comorbidities
Alzheimer's—Dementia	52 (12.0)
Parkinson's disease	20 (4.6)
DM	97 (22.3)
HT	205 (47.1)
CVS	27 (6.2)
CVE	112 (25.7)
Respiratory disease	44 (10.1)
Hypoxic brain	23 (5.3)
Traumatic brain and spinal cord injury	38 (8.7)
Cancer	167 (38.4)
Concomitant conditions
PEG	128 (29.4)
Tracheostomy	86 (19.8)
Mechanical ventilation	24 (5.5)
TPN additional nutrition	379 (87.1)
TPN	265 (60.9)
Oral nutrition	51 (11.7)

Values are given as number of cases (%).

CVE, cerebrovascular event; CVS, cardiovascular system disease; DM, diabetes mellitus; HT, hypertension; PEG, percutaneous endoscopic gastrostomy; TPN, total parenteral nutrition.

The bacteria most isolated in cultures were Escherichia coli (n: 124, 28.5%). Other bacteria were determined as MRSA (17%), Pseudomonas spp. (14.7%), and Acinetobacter spp. (12.9%) (Table 4).

Table 4.

Type of Bacteria Produced from Cultures and Amount of Each Bacteria

Bacteria type	Produced, n (%)	Not produced, n (%)	Amount of bacteria produced, median (min; max)
Escherichia coli	124 (28.5)	311 (71.5)	1 (1; 10)
Pseudomonas spp.	64 (14.7)	371 (85.3)	1 (1; 8)
Acinetobacter spp.	56 (12.9)	379 (87.1)	1 (1; 6)
MRSA	74 (17.0)	361 (83.0)	1 (1; 4)
Proteus spp.	22 (5.1)	413 (94.9)	1 (1; 5)
Others	66 (15.2)	369 (84.8)	1 (1; 4)

MRSA, methicillin-resistant Staphylococcus aureus.

In the regression model, the independent variables of age, comorbidities, PEG, TPN, and bacteria produced in cultures were found to have an effect on LOS in PCC. The factors effective on LOS and coefficients are shown in Table 5. Although the model was not greatly explanatory (R² = 0.319), the model was statistically significant (F = 18.023, p < 0.001). This showed that the model could be developed and there is a need for new factors to increase the explanatory level of the model.

Table 5.

Variables Related to Length of Stay in Palliative Care Center: Multivariable Linear Regression Analysis

Variable	\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $$\hat{ \beta}$$ \end{document}	\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $$SE ( \,\hat{ \beta} )$$ \end{document}	\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $$\hat{ \beta}$$ \end{document} (% 95 CI) (lower to upper)	t	p	LOS ^a	Coefficient for LOS ^b
Fixed coefficient	2.730	0.202	2.332 to 3.127	13.512	<0.001	15.333	—
Age	−0.009	0.002	−0.014 to −0.004	−3.560	<0.001	15.196	0.991
Hypoxic brain	−0.419	0.187	−0.786 to −0.051	−2.239	0.026	10.085	0.658
Cancer	−0.187	0.092	−0.367 to −0.007	−2.047	0.041	12.718	0.829
HT	0.176	0.082	0.015 to 0.338	2.145	0.032	18.284	1.192
Pressure injury	0.217	0.088	0.045 to 0.389	2.484	0.013	19.049	1.242
PEG	0.445	0.097	0.255 to 0.636	4.588	<0.001	23.927	1.560
TPN	0.453	0.085	0.286 to 0.620	5.343	<0.001	24.119	1.573
E. coli	0.440	0.096	0.252 to 0.628	4.603	<0.001	23.807	1.553
Proteus spp.	0.576	0.187	0.210 to 0.943	3.090	0.002	27.276	1.779
Pseudomonas spp.	0.338	0.122	0.098 to 0.577	2.774	0.006	21.499	1.402
Acinetobacter spp.	0.254	0.128	0.002 to 0.506	1.983	0.048	19.767	1.289

n = 435, p = 11, R² = 0.319, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $$\overline {R}$$ \end{document} ² = 0.301, Durbin Watson d = 1.963, F = 18.023, p < 0.001, values calculated for ln(LOS).

LOS = e^{Fixed coefficient+
\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}
$$\hat{ \beta}$$
\end{document}
.}

Coefficient for LOS = variable/fixed coefficient.

CI, confidence interval.

According to the results of the regression analysis, cancer, hypoxic brain, and advanced age have a negative effect on LOS. Nutrition with PEG was found to increase LOS 1.560-fold and nutrition with TPN, 1.573-fold. Patients with HT had LOS increased by 1.192-fold and those with pressure injury by 1.242-fold. The effect of infections on LOS was evaluated in the same model. LOS was seen to be increased in patients with isolated Proteus spp. by 1.779-fold, in those with E. coli by 1.553-fold, in those with Pseudomonas spp. by 1.402-fold, and with Acinetobacter by 1.289-fold. It was determined that gender, GCS, KPS, clinical conditions except cancer, hypoxic brain and HT, MV support, and MRSA produced from culture did not have any effect on LOS (Table 5).

Discussion

In contrast to developed countries, the concept of PC is new in Turkey. The PC capacity and clinical experience in Turkey are limited. Due to the increasing elderly population worldwide, the need for PCC is also increasing. The number of PCC beds in Turkey is well below the requirement. There are studies in literature, which have reported the necessity for the application of PC. The effect on LOS and costs of the application of PC has been evaluated in these studies and it has been shown that for PC patients with a shorter LOS, the costs are lower.^8–10 Previous studies have evaluated the effect of various factors on LOS and costs.^11–13 However, these studies have only evaluated the effect of several factors on LOS. The current study is the first on this subject to evaluate at what rate which multiple factors affect LOS in PCC.

The patients in the current study were elderly (mean age 70.6 ± 17.2 years) with multiple chronic diseases, GCS median 12, and KPS median 40. Infection is one of the factors affecting PCC LOS. In studies that have evaluated infections in PCC, the most frequently isolated bacteria in cultures have been determined to be E. coli and Staphylococcus aureus.^14–16 The bacteria isolated in cultures may vary between centers. In the current study, E. coli (28%) and MRSA (17%) were seen to be the most frequently produced bacteria, which was consistent with literature. PCC patients are elderly, with chronic diseases and limited movement, as was seen in the current study. The most commonly seen infections in this patient group are urinary system infections,¹⁷ which have severe morbidity and mortality. The bacteria most often isolated in urinary system infections is E. coli, but MRSA, Proteus spp., and Pseudomonas spp. may also be isolated.^17,18

Woodford and George showed that LOS and mortality rates were greater in cases of urinary system infection with bacteriemia.¹⁸ With the current increase in the facilities of antibiogram together with the use of bacteria-sensitive antibiotics, mortality has decreased significantly. A decrease in mortality causes an increase in LOS. In the current study, the LOS of patients was seen to be increased by 1.553-fold in the presence of E. coli, by 1.779-fold with Proteus spp. and by 1.402-fold with Pseudomonas spp. production.

In a study evaluating MRSA in PCC patients, MRSA production was determined in 11% of patients and this was reported to prolong LOS.¹² In another study, mortality rates of patients with urinary system S. aureus infection were shown to be extremely high.¹⁷ In the current study, MRSA was determined in 17% of patients, but it was not found to have any effect on LOS. The reasons for these different results in studies of the effect of MRSA production on LOS could be the state of resistance of the bacteria, and differences in laboratory cultures, antibiogram, and antibiotic facilities and these would therefore explain the variations in morbidity and mortality rates.

Acinetobacter spp., which is a gram-negative bacteria found in multidrug-resistant strains, remains alive by covering itself with a film on medical instruments and surfaces and causes bacteriemia, pneumonia, urinary system infections, deep wound infections, and osteomyelitis.¹⁹ As a cause of severe hospital-acquired infection, Acinetobacter spp. has recently become endemic in intensive care units and residential homes for the elderly.^19,20 As this patient group is in PCC, Acinetobacter infection is often seen. Several studies have reported that acinetobacter infections increase LOS and mortality.^21–23 Similarly in the current study, Acinetobacter production was determined to increase LOS by 1.289-fold but did not increase mortality (p > 0.05).

The diagnoses of patients admitted to PCC and additional clinical characteristics affect LOS. Although the concept of PC is generally associated with terminal stage oncology patients, many benign pathologies are also followed up in PCC. Integration of the treatment of critical neurology patients, such as those with head trauma, is recommended with PC.^7,24 Many benign diseases are monitored in PCC such as hypoxic brain, cerebrovascular event, neurological diseases, and immobile patients who require care following trauma. In the current study, 61.2% of the patients were being followed up in PCC for a diagnosis other than cancer. The majority of these patients had a neurological diagnosis, were attached to mechanical ventilation, were nourished with PEG, and had pressure wounds.

In a study conducted in a PCC of patients diagnosed with cancer, it was reported that the most frequent reasons for admittance to PCC were complaints of pain, nausea, weakness, and dyspnea and the LOS of patients with cancer was median 7 days.²⁵ In the current study, the PCC LOS of cancer patients was determined to be 0.171-fold shorter. A similarly shorter stay was determined for patients with hypoxic anoxic brain. In literature, mortality on the first day after cardiac arrest has been reported as 52%–90% and the 30-day mortality rate as 25%–40%.²⁴ When early death hypoxic brain cases were evaluated together with surviving and long-term cases in PCC, the LOS was found to be low.

In the current study, a negative relationship was found between age and LOS. The age of the patients lost to mortality was found to be more than that of the surviving discharged patients. The median age of the discharged patients was 73 years and the median age of the exitus patients was 78 years, which was determined as a statistically significant difference (p = 0.005). The shorter LOS of older patients was thought to be associated with mortality.

Pressure injuries are known to increase hospital costs and LOS in PCC.²⁶ Pressure injuries are often seen together with infections and this combination significantly increases LOS. In the current study, the presence of pressure injury was determined to increase LOS 1.242-fold.

While PEG can be applied to many patient groups, it is mostly preferred for patients with a poor state of consciousness, swallowing problems, and those with motility problems in the upper gastrointestinal system.²⁷ Neurology patients have a greater requirement for PEG in PC and as these patients are unconscious or at a low level of consciousness, the combination of pressure injury and infection is also frequently observed. A long LOS is an expected outcome in patients applied with PEG nutrition.

TPN is applied to patient groups where gastrointestinal intake is not possible. Complications related to the administration of TPN have been reported in literature.²⁸ Catheter infection, bacteriemia, and sepsis may be seen in patients receiving TPN. It has been concluded that because of the effect of the infection agents, the LOS of patients applied with TPN is prolonged.

The limitations of this study can be said to be that standardization could not be fully provided as the PCC patients were not a homogenous group for various reasons such as age groups, diagnoses, and different additional clinical features. However, the study has revealed some basic data in the understanding of factors that affect the LOS in PCCs.

Conclusion

The results of this study have shown some basic factors that have an effect on the LOS of patients in PC units. For many reasons, there may be many variables in the data obtained from the patient group in PCC. However, this is the most comprehensive study on the effect of infection agents and clinical features in the PCC patient group. There is a need for further studies of more extensive patient groups to determine the factors affecting LOS.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

WHO: Definition of Palliative Care. www.who.int/cancer/palliative/definition/en/. 2017. (Last accessed October 1, 2017).

TUIK: www.tuik.gov.tr/PdfGetir.do?id=18618. 2015. (Last Accessed on October 30, 2016).

Higginson

, Davies

: Better Palliative Care for Older People. World Health Organization, Copenhagen, Denmark, 2004.

TKHK: Turkish Public Hospitals Agency, Palliative Care Units Registration List [in Turkish]. 2016. www.tkhk.gov.tr/6736_-palyatif-bakim-unut. (Accessed on October 1, 2017).

Brennan

, Gwyther

, Harding

: Palliative Care as a Human Right. Open Society Institute Law and Health Initiative on Palliative Care and Human Rights, Open Society Institute (OSI) Public Health Program, OSI, Brussels, 2008.

Buccheri

, Ferrigno

, Tamburini

: Karnofsky and ECOG performance status scoring in lung cancer: A prospective, longitudinal study of 536 patients from a single institution. Eur J Cancer, 1996; 32a:1135–1141.

Kahveci

, Dinçer

, Doger

, et al.: Traumatic brain injury and palliative care: A retrospective analysis of 49 patients receiving palliative care during 2013–2016 in Turkey. Neural Regen Res, 2017; 12:77–83.

Twaddle

, Maxwell

, Cassel

, et al.: Palliative care benchmarks from academic medical centers. J Palliat Med, 2007; 10:86–98.

Penrod

, Deb

, Luhrs

, et al.: Cost and utilization outcomes of patients receiving hospital-based palliative care consultation. J Palliat Med, 2006; 9:855–860.

10.

Smith

, Coyne

, Cassel

, et al.: A high-volume specialist palliative care unit and team may reduce in-hospital end-of-life care costs. J Palliat Med, 2003; 6:699–705.

11.

Kim

, Hwang

, Ko

, et al.: Association between the emotional status of family caregivers and length of stay in a palliative care unit: A retrospective study. Palliat Support Care, 2015; 13:1695–1700.

12.

Gleeson

, Larkin

, Walsh

, et al.: Methicillin-resistant Staphylococcus aureus: Prevalence, incidence, risk factors, and effects on survival of patients in a specialist palliative care unit: A prospective observational study. Palliat Med, 2016; 30:374–381.

13.

Lin

, Wu

, Chen

, et al.: Analysis of long-stay patients in the Hospice Palliative Ward of a Medical Center. J Chin Med Assoc, 2008; 71:294–299.

14.

Pereira

, Watanabe

, Wolch

: A retrospective review of the frequency of infections and patterns of antibiotic utilization on a palliative care unit. J Pain Symptom Manage, 1998; 16:374–381.

15.

Homsi

, Walsh

, Panta

, et al.: Infectious complications of advanced cancer. Support Care Cancer, 2000; 8:487–492.

16.

Vitetta

, Kenner

, Sali

: Bacterial infections in terminally ill hospice patients. J Pain Symptom Manage, 2000; 20:326–334.

17.

Ackermann

, Monroe

: Bacteremic urinary tract infection in older people. J Am Geriatr Soc, 1996; 44:927–933.

18.

Woodford

, George

: Diagnosis and management of urinary tract infection in hospitalized older people. J Am Geriatr Soc, 2009; 57:107–114.

19.

Dijkshoorn

, Nemec

, Seifert

: An increasing threat in hospitals: Multidrug-resistant Acinetobacter baumannii. Nat Rev Microbiol, 2007; 5:939–951.

20.

Mody

, Gibson

, Horcher

, et al.: Prevalence of and risk factors for multidrug-resistant Acinetobacter baumannii colonization among high-risk nursing home residents. Infect Control Hosp Epidemiol, 2015; 36:1155–1162.

21.

Sunenshine

: Multidrug-resistant Acinetobacter infection mortality rate and length of hospitalization. Emerg Infect Dis, 2007:13:97–103.

22.

García-Garmendia

J-L

, Ortiz-Leyba

, Garnacho-Montero

, et al.: Mortality and the increase in length of stay attributable to the acquisition of Acinetobacter in critically ill patients. Crit Care Med, 1999; 27:1794–1799.

23.

Lee

N-Y

, Lee

H-C

, Ko

N-Y

, et al.: Clinical and economic impact of multidrug resistance in nosocomial Acinetobacter baumannii bacteremia. Infect Control Hosp Epidemiol, 2007; 28:713–719.

24.

Frontera

, Curtis

, Nelson

, et al.: Integrating palliative care into the care of neurocritically ill patients: A report from The IPAL-ICU (improving palliative care in the intensive care unit) project advisory board and the center to advance palliative care. Crit Care Med, 2015; 43:1964.

25.

Elsayem

, Swint

, Fisch

, et al.: Palliative care inpatient service in a comprehensive cancer center: Clinical and financial outcomes. J Clin Oncol, 2004; 22:2008–2014.

26.

Allman

, Goode

, Burst

, et al.: Pressure ulcers, hospital complications, and disease severity: Impact on hospital costs and length of stay. Adv Skin Wound Care, 1999; 12:22–30.

27.

DeLegge

: Percutaneous endoscopic gastrostomy. Am J Gastroenterol, 2007; 102:2620.

28.

Heyland

, MacDonald

, Keefe

, et al.: Total parenteral nutrition in the critically ill patient: A meta-analysis. JAMA, 1998; 280:2013–2019.