16th Annual Congress – Amsterdam, Netherlands – 5–8 October 2003

Oral Presentations Mechanisms in acute respiratory failure 001-005 001 NON-INVASIVE VENTILATION AND OXYGEN AS ALTERNATIVE INITIAL THERAPIES IN ACUTE RESPIRATORY FAILURE

S5

003 LUNG PRODUCTION OF CRP IN CRITICALLY ILL PATIENTS Peres Bota D1, Lobo S M1, Carvalo F B1, De Backer D1, Vincent J1 1 Intensive Care, Erasme Hospital, Brussels, Belgium

Bersten A D1, Hunt T1, Davies A R2, O’Callaghan G1, For the ANZICS Clinical Trials Group 3 1 Department of Critical Care Medicine, Flinders Medical Centre, Adelaide, 2Intensive Care Unit, Alfred Hospital, Melbourne, 310 Ievers Terrace, Carlton, Australia

INTRODUCTION: C-reactive protein (CRP) is an important sepsis marker whose site of production is not well known. The main source was thought to be the liver but new body of evidence shows that lung, kidney and even neurons can produce CRP in different conditions. The aim of our study was to measure the arterio-venous difference of CRP in patients with ARDS, broncho-pneumonia, extrapulmonary sepsis and non-septic patients

INTRODUCTION: Non-invasive positive pressure ventilation (NIPPV) has been shown to reduce intubation rate and improve outcome in patients with acute respiratory failure (ARF) due to chronic obstructive pulmonary disease (COPD) and cardiogenic pulmonary edema (CPE). However, the use of NIPPV in critically ill patients with hypoxemic ARF is controversial, particularly in patients with acute lung injury (ALI). By delaying intubation, NIPPV could contribute to worse outcome. Consequently, we examined the use of NIPPV and oxygen therapy without positive airway pressure (O2), with the aim of exploring factors associated with hospital death.

METHODS: After informed consent was obtained, 70 patients admitted to our intensive care department with a pulmonary artery catheter in place, were included in the study. Patients were divided into 5 groups as follows: non-ventilated non-septic (NVNS, n=6), ventilated non-septic (VNS, n=9), extrapulmonary sepsis (EPS, n=12), broncho-pneumonia (BPN, n=20) and ARDS (n=23). The measurements of CRP were performed simultaneously from mixed venous and arterial. CRP was measured by immunoturbidimetry method and the coefficient of variability between two measurements was <1%.

METHODS: This multi-centre, prospective, observational study screened all patients admitted to 16 ICU’s over a 2 month period in mid-2003. All patients with ARF at-risk of developing ALI, that is at least one of sepsis, pneumonia, aspiration, pulmonary contusion, multiple transfusions and multiple trauma, and who initially received O2 therapy or NIPPV were enrolled. These patients were followed until hospital discharge, and demographics including APACHE II score, and clinical outcomes were collected. Exclusion criteria were COPD, CPE, post-operative ARF, and factors such as upper airway compromise that would contraindicate NIPPV. Patients with ARF who initially received O2 therapy or NIPPV were compared, with predefined subgroups including immunocompromised versus nonimmunocompromised patients, and intubated versus never intubated patients.

RESULTS: In all patients CRP levels ranged between 1.3 to 40.6mg/dl (mean 12.6 ± 9.8, median 9.8). The arterial CRP concentrations were significantly higher in patients with sepsis and ARDS than in non-septic patients (20.2 ± 9.1 vs. 4.3 ± 2.9 mg/dl, p=0.01). Lung CRP production was significantly higher in patients with BPN and ARDS than in NVNS, VNS and EPS patients (Fig. 1).

RESULTS: Of 3359 admissions to the 16 ICU’s, 1156 were not intubated on arrival, but 956 were excluded, leaving 200 patients with ARF at-risk of developing ALI, of whom 32 (16%) were immunocompromised. O2 was initially administered to 123 (62%) and NIPPV to 77 (39%). All ARF patients initially receiving NIPPV had increased hospital mortality (35% vs 13%, p < 0.001) when compared to those initially receiving O2 therapy, especially in the never intubated group (27% vs 8%, p = 0.007). Non-immunocompromised patients receiving NIPPV had increased hospital mortality whilst immunocompromised patients did not. Using a general linear model, the use of NIPPV (RR 2.16, CI 1.04 - 4.47), APACHE II score (RR 1.05, CI 1.00 – 1.09) and intubation (RR 2.27, CI 1.14 – 4.50) were positive predictors of hospital mortality in non-immunocompromised patients with ARF. When delayed intubation was added to the model, NIPPV became much less significant.

Lung production of CRP. CRP pro mg/dl minimum IQ 25% median IQ 75% maximum

NVNS n=6 -0.31 -0.08 0.02 0.04 1.1

VNS

EPS

BPN

ARDS

-0.05 -0.02 0.01 0.11 1.22

-0.06 0.01 0.02 0.31 0.52

-0.02 1.41 5.69 8.19 14.89

-0.04 0.85 3.32 4.34 6.89

CONCLUSION: CRP can be produced by the lung in inflammatory (ARDS) and infectious lung diseases. The cells involved in lung CRP production remains to be defined.

CONCLUSION: Although NIPPV is commonly used in the initial management of ARF that is not due to COPD or CPE in Australasian ICU’s, this is associated with increased mortality, particularly in those patients who are non-immunocompromised and/or never intubated. Grant acknowledgement: Australian and New Zealand Intensive Care Foundation

002

004

ANALYSIS OF SINGLE-BREATH TEST OF CO2 DURING LUNG RECRUITMENT IN EDEMATOUS LUNGS

LONG-TERM USE OF DEXMEDETOMIDINE DURING MECHANICAL VENTILATION: A VIEW FROM PATIENTS PERSPECTIVE

Tusman G1, Suarez Sipmann F2, Bohm S H3, Acosta C4, Turchetto E5

Pontes-Arruda A1 Unidade de Terapia Intensiva, Hospital Antônio Prudente, Fortaleza, Brazil

1 1

2Intensive

Dept. of Anesthesiology, Hospital Privado de Comunidad, Mar del Plata, Argentina, Care Medicine, Fundacion Jimenez Diaz, Madrid, Spain, 3Pulmonary Division, Hospital das Clinicas, University of SP, Sao Paulo, Brazil, 4Intensive Care Medicine, Hospital Privado de Comunidad, Mar del Plata, 5Intensive Care Medicine, Hospital Privado de Comunidad, Madrid, Argentina INTRODUCTION: The single breath test of CO2 (SBT-CO2) is a useful tool for dead space analysis. Previous reports in injured lungs have shown that all SBT-CO2 variables change after applying different PEEP levels. We hypothesized that an alveolar recruitment strategy (ARS) affects SBT-CO2 variables due to an improvement in the distribution of ventilation and gas exchange. In this study this hypothesis was tested in edematous lungs after cardiopulmonary bypass (CPB).

METHODS: We prospectively randomized 22 patients after CPB into two groups: 1) PEEP group (n = 11) VCV: VT 8 ml kg-1, respiratory rate of 12-14 bpm, PEEP of 10 cmH2O and FIO2 0.5. 2) ARS group (n = 11) ARS (i.e. 10 breaths at 45/20 cmH2O plateau pressure/PEEP in PCV mode) followed by the same ventilator settings as the PEEP group. ARS was performed 20 min post CPB after chest closure. SBT-CO2 and arterial blood gases were obtained at the end of surgery (approx. 30’ after CPB). RESULTS: PaO2 was higher after ARS (224 ± 34 mmHg) compared with the PEEP group (101 ± 22 mmHg, p < 0.05). Phase II slope was steeper after ARS (448 ± 68 mmHg/L) compared to PEEP (376 ± 57 mmHg/L, p <0.05) and the normalized phase III slope decreased after lung recruitment (3.12 ± 0.2 1/L) compared to PEEP (6.16 ± 0.3 1/L, p <0.05). All SBT-CO2 variables related to ventilatory efficiency as VD/VT ratio, Pa-etCO2 and VTCO2, br improved in after ARS (see table). Variables for efficiency of ventilation variable VD aw ml VD alv ml VD phys ml VD/VT Vol III / VT VTCO2, br ml Pa-etCO2 mmHg

PEEP 156 +- 50 119 +- 36 274 +- 50 0.49 +- 0.06 0.40 +- 0.09 21 +- 4.8 11 +- 4.9

ARS 137 +- 43 100 +- 25 237 +- 55 * 0.43 +- 0.11 * 0.49 +- 0.07 * 29 +- 5.8 * 7 +- 2.3

CONCLUSION: ARS improved all SBT-CO2 variables of lung function after cardiopulmonary bypass compared with PEEP. The decrease in phase III-slope and the increase in phase II-slope after ARS can be interpreted as an improvement in the distribution of ventilation within the lungs. REFERENCE(S): Blanch LL, Lucangelo U, Lopez-Aguilar J, Fernandez R, Romero PV. Volumetric capnography in patients with acute lung injury: effect of positive end-expiratory pressure. Eur Respir J 1999; 13: 1048-1054.

INTRODUCTION: Dexmedetomidine hydrocloride is a potent alpha2-adrenergic agonist used as a sedative and analgesic agent in ICU during the weaning of mechanical ventilation. The clinical safe and effectiveness of Dexmedetomidine HCl is now well stablished for continuous IV sedation for periods longer than 72 hours (1). The scope of this work is to determine the patients experiences during the use of this drug for long-term sedation. METHODS: We perform an analysis of 76 patients using Dexmedetomidine during the weaning of mechanical ventilation. All patients in this work used Dexmedetomidine hydrocloride during 48 hours with extubation purpose only and with no co-administration of other sedatives, opioids, hypnotics and anesthetics. The loading dose was 0.5 mg/kg/h and the maintenance dose was adjusted to maintain the sedation between 3-4 in accordance with the Ramsay sedation scale (2). After the extubation, the drug was suspended and the patients submitted to a questionnaire based on that of Venn & Grounds (3). RESULTS: The patients experiences can be observed in Table 1. Accurate recording of the length of ICU stay Accurate recording of duration of MV ICU Stay described as pleasant overall Perception of tracheal suctioning Perception of handling and movement of various tubes and lines Being on a ventilator Disconfort from ventilator Pain 1

58 (77.33%) 28 (36.84%) 68 (89.47%) 12 (15.78%) 12 (15.78%) 33 (43.42%) 12 (15.78%) (1.31%)

CONCLUSION: Dexmedetomidine appears to be safe, effective and acceptable as a sedative and analgesic agent for long-term use during the weaning of mechanical ventilation in ICU, not just from clinician´s point of view, but also from patients perspective. REFERENCE(S): 1-Pontes-Arruda, A. 24th ISICEM (Abstracts), 2004. 2-Ramsay, MAG et al. BMJ 2:656-659, 1974. 3-Venn, RM and Grounds, RM. Br J Anaesth 87:684-690, 2001.

S6

17th Annual Congress – Berlin, Germany – 10–13 October 2004

005

007

EXTUBATION FOLLOWING A BREATHING TRIAL WITH AUTOMATIC TUBE COMPENSATION (ATC) VS. T-PIECE

ROLE OF EPITHELIAL CELLS IN THE UPPER AIRWAYS IN ENDOTOXIN-INDUCED LUNG INJURY

Cohen J1, Shapiro M1, Grozovski E1, Lev S1, Fisher H1, Singer P1 General Intensive Care, Rabin Medical Center, Petah Tikva, Israel

Roth Z’Graggen B1, Neff S2, Booy C3, Schimmer R C4, Pasch T5, Beck-Schimmer B3 1 Institute of Physiology, University of Zurich, Zurich, Switzerland, 2Department of Pathology, University of Michigan, Ann Arbor, United States, 3Institute of Anaesthesiology, Institute of Physiology, 4Department of Surgery, 5Institute of Anaesthesiology, University of Zurich, Zurich, Switzerland

1

INTRODUCTION: A spontaneous breathing trial (SBT) is typically performed prior to extubation in patients requiring prolonged mechanical ventilation. No differences in outcome have been reported when the SBT is performed with various modes including T-piece, SIMV or PSV. We proposed that ATC, which compensates for the resistive load of the ETT, may more closely predict the success of extubation. METHODS: All consecutive patients who had required prolonged ventilation (> 24 hours) and were considered ready for extubation (underlying disease adequately treated, no need for vasopressors, PEEP < 8 cm H2O, paO2/FiO2 >200, Temperature < 38.5C, RVR < 105) were included in this prospective study. Patients were randomly assigned to undergo a 1-hour SBT prior to extubation with either ATC alone or T-piece. Extubation was performed immediately after successful completion of the SBT. The outcome measure was successful extubation (ability to maintain spontaneous breathing for >48 hours). Statistical analysis was performed using the ChiSqare test. RESULTS: The overall reintubation rate was 19.4% (13.5% for the ATC group and 25.7% for the T-piece group). ATC (n=40) Age (yrs) APACHE II Ventilation (days) Completed SBT Successful outcome

61.8 16.4

T-piece (n=42) 67.5 16.6

p value (n=42) NS NS

6.12 37

7.34 35

NS NS

30

23

0.06

CONCLUSION: In this study, we have shown that there was a trend for more successful outcomes when a SBT preceeding extubation was performed with ATC as compared to a T-piece trial in critically ill patients following prolonged ventilation.

INTRODUCTION: The airway compartment with epithelial cells as the predominant cell type is a physiological barrier to a variety of environmental agents. Several studies have shown that airway epithelial cells express and secrete various immune molecules such as adhesion molecules, cytokine and chemokines. While many studies have been performed investigating the distal airways with alveolar epithelial cells (1), only scarce information is available about tracheobronchial epithelial cells (TBEC) and their inflammatory response to lipopolysaccharide (LPS) as a predisposing cause of the acute respiratory distress syndrome (ARDS). Therefore, the aim of the present study was to evaluate the expression pattern of the adhesion molecules intercellular adhesion molecule-1 (ICAM-1), vascular cell adhesion molecule-1 (VCAM-1), the cytokine tumor necrosis factor-\alpha (TNF-\alpha) and the chemokine macrophage inflammatory protein-1\beta (MIP-1\beta) in TBEC after LPS stimulation. METHODS: TBEC from Wistar rats were isolated using 0.01% proteinase. The cells were grown to confluence over a 3-day period. Purity was verified using periodic acid-Schiff staining. Monolayers of TBEC were stimulated overnight with LPS (100mg/ml). Cell-based ELISA was performed for detection of ICAM-1 and VCAM-1 as well as Sandwich-ELISA for secreted TNF\alpha (PharMingen, San Diego, CA). MIP-1\beta protein in the supernatant was assess with Western blot. All experiments were performed at least three times. ANOVA was applied to determine significance of differences. RESULTS: After overnight stimulation with LPS a 52% increase of ICAM-1 protein was detected at the apical part of TBEC (p< 0.001), while VCAM-1 expression was enhanced by 101% (p< 0.0001). TNF-\alpha protein concentration in the supernatant from control cells was 244.3pg/ml, while LPS-stimulated TBEC increased their secretion by 21% (295pg/ml) (p< 0.05). Determination of MIP-1\beta protein showed a 77% increase in LPS-treated cells (p< 0.01). CONCLUSION: These data suggest that TBEC play an important role in endotoxin-induced lung injury. Further studies have to be performed to determine functional activity of these proteins. REFERENCE(S): (1) Beck-Schimmer B et al. Role of alveolar epithelial ICAM-1 in lipopolysaccharide-induced lung inflammation Eur Respir J 2002, 19 (6), 1142-1150. Grant acknowledgement: This study was supported by Swiss Life Rentenanstalt, Zurich, Switzerland and the Swiss National Science Foundation grant No. 31-55702.98.

Oral Presentations New insights in the pathogenesis of sepsis from basic research – 006-010 006 THE R753Q POLYMORPHISM IN THE TOLL-LIKE RECEPTOR 2 GENE INHIBITS TRANSACTIVATION OF NF-KB Chiche J D1, Choukroun G2, Roget K1, Rousseau C1, Dhainaut J F2, Cariou A2, Mira J P1 1 Cell Biology, Institut Cochin, 2Medical Intensive Care, Hopital Cochin, Paris, France INTRODUCTION: Recognition of Gram positive bacteria by Toll-like receptor 2 (TLR2) initiates signaling pathways that leads to nuclear translocation and transactivation of NF-kB. A mutation in the cytosolic domain of TLR2 results in an arginine (R) to glutamine (Q) substitution at residue 753. This R753Q polymorphism is associated with severe staphylococcal infections (1). We analyzed consequences of the R753Q polymorphism on the molecular mechanisms of NF-kB activation. METHODS: After transfection of plasmids encoding TLR2 or TLR2-R753Q, we investigated NF-kB activity (luciferase gene reporter system), nuclear translocation (EMSA), and transactivation (Western blots, immunoprecipitation) after stimulation of HEK293 cells with heatkilled S.aureus (HKSA). RESULTS: Stimulation of TLR2-R753Q with HKSA inhibits NF-kB activity (A). The R753Q mutation does not affect nuclear translocation of NF-kB. Conversely, this mutation inhibits phosphorylation of p65 by preventing recruitment and activation of PI-3 kinase (B).

008 ENDOTOXEMIA-INDUCED MITOCHONDRIAL DYSFUNCTION: EFFECTS OF DOPAMINE AND DOBUTAMINE Porta F M1, Weickert C1, Bracht H1, Lauterburg B H2, Kraehenbuehl S3, Takala J1, Jakob S M1 1 Departement of Intensive Care Unit, 2Departement of Clinical Pharmacology, University Hospital Bern, Bern, 3Departement of Clinical Pharmacology, University Hospital Basel, Basel, Switzerland INTRODUCTION: Dopamine but not dobutamine, decreases hepato-splanchnic oxygen consumption in sepsis and liver failure1,2, while both drugs similarly improve systemic and regional hemodynamics We therefore assessed the effects of dopamine and dobutamine on muscle mitochondrial oxygen utilization in-vitro with and without endotoxin. METHODS: Sternocleidomastoid muscle biopsy was taken from 6 anesthetized pigs. Mitochondria from each pig were immediately isolated, put on ice, and divided in six samples. Lipopolysaccharide (E) (100microgr/mg mitochondrial protein) or placebo (P) (isolation buffer) was added to the samples (n=18 for each). After one hour, each group of three samples was incubated with either dopamine, dobutamine (final concentration 100 uM, each) or P for one hour. ADP-dependent (state 3) and ADP-independent (state 4) mitochondrial respiration and their ratio were determined polarographically using glutamate (complex I), succinate (complex II) and ascorbate/TMPD (complex IV) as substrates. RESULTS: Data are expressed as median (range) a p<0.01 vs P alone bp <0.05 vs E alone. State 3 and State 4 are in nanomol/min/mg mitochondrial protein. complex I State 3 Placebo E 59.0 (33.4-72.3) Placebo P 45.6 (26.7-55.6) Dopamine E 46.7(24.5-64.5) Dopamine P 51.7 (26.7-82.3) Dobutamine E 46.7 (20.0-77.9) Dobutamine P 37.8 (22.3-89.0)

complex I complex I State 4 RCR a 3.6 (3.1-4.7)a 16.7 (8.9-22.3) 4.5 (2.2-7.8) 10.3 (7.1-12.0) 7.8(4.5-13.4)b 5.5 (3.8-8.3)b 6.1 (4.5-8.9) 8.5 (6.0-12.0) 8.3 (4.5-20.0) 4.9 (3.9-7.2)b 4.5 (3.3-17.8) 6.6 (4.3-12.0)

complex II RCR 3.6 (1.7-2.3) 5.0 (4.2-6.7) 4.5 (2.3-6.8) 5.5 (2.7-6.1) 4.1 (2.8-5.9) 5.3 (2.2-7.7)

complex IV RCR 1.9 (1.8-2.8) 1.9 (1.7-2.3) 1.5 (1.5-2.1) 2.5 (2.1-4.8) 2.0 (1.9-2.1) 2.1 (1.7-3.1)

CONCLUSION: In-vitro administration of endotoxin induced an increase in non-energy producing oxygen consumption (futile cycling), while the energy-producing oxygen consumption did not change. In the presence of endotoxin, dopamine and dobutamine both enhanced the efficiency of oxygen utilization, mainly due to reduced futile cycling. CONCLUSION: The TLR2-R753Q polymorphism prevents HKSA-induced activation of NF-kB through inhibition of the recruitment and activation of PI-3 kinase. These results might explain why the R753Q polymorphismsincreases susceptibility to G+ infection. REFERENCE(S): Infection & Immunity,2000:6398-6401

REFERENCE(S): 1Clemmesen, JO et al., Scand J Gastroenterol 1999, 2Jakob S M et al., SHOCK 2002

17th Annual Congress – Berlin, Germany – 10–13 October 2004

009 MUSCLE GLYCOLYTIC ENZYME ACTIVITY CHANGES DURING SEPSIS IN A LONG-TERM RODENT MODEL Karyampudi S1, Brealey D1, Frost M T1, Stidwill R1, Taylor V1, Singer M1 1 Department of Medicine, Wolfson Institute for Biomedical Research, London, United Kingdom INTRODUCTION: Mitochondrial dysfunction appears implicated in the development of sepsisinduced multi-organ failure. Glycolytic ATP may thus be an important alternative energy source if aerobic respiration is compromised. Little is known about changes in glycolysis during sepsis, though both up- and down- regulation are reported1,2. We have previously reported changes in liver glycolytic activity in a 3-day rat sepsis model with initial upregulation, followed by downregulation at 48h, in hexokinase and pyruvate kinase3. In the present study we assessed concurrent changes in skeletal muscle glycolytic activity. METHODS: An instrumented, fluid-resuscitated, faecal peritonitis rat model (with a 72-hour mortality rate of approximately 40%) was used. Septic (n=57) and sham (n=35) rats were sacrificed at various time points (0, 4, 24, 48, 72h). Muscle samples were harvested and assayed for maximal activity of the rate-limiting glycolytic enzymes, hexokinase (HK), phosphofructokinase (PFK), and pyruvate kinase (PK), plus glyceraldehyde 3-phosphate dehydrogenase (GAPDH) which is inhibited by nitric oxide. RESULTS: Using repeated measures 2-factor ANOVA, significant changes were seen over time in the septic group (vs. sham controls) for HK (p<0.011) and PFK (p<0.0001). Time (Hours) 0 4 24 48 HK sham 0.20 (0.02) 0.21 (0.03) 0.20 (0.02) 0.20 (0.01) HK septic 0.26 (0.02) 0.19 (0.01) 0.21 (0.01) PFK sham 12.7 (1.1) 10.7 (1.8) 9.7 (1.0) 9.2 (1.1) PFK septic 17.5 (2.8) 11.2 (0.6) 7.2 (0.5) Activity shown in pmoles/min/µg of protein. Data shown as mean (SE).

S7

Oral Presentations Cardiovascular biology – 011-015 011 BETA-2 ADRENERGIC STIMULATION PROTECTS ENDOTHELIAL CELLS DURING HYPOXIA/REOXYGENATION IN VITRO Pottecher J1, Cheisson G1, Huet O1, Laplace C1, Benhamou D1, Duranteau J1 1 Department of Anesthesiology and Critical Care, Bicêtre Hospital, Le Kremlin-Bicêtre, France INTRODUCTION: In case of shock, hypoxia/reoxygenation (H/R) is one of the mechanisms leading to multi-organ failure. The aim of our study was to demonstrate a NO-synthase (NOS) dependent protective role of eta2-adrenergic (eta2-A) stimulation on endothelial cells (EC) submitted to H/R in vitro. METHODS: EC were exposed to H/R in a flow-through chamber mounted on a epifluorescent microscope. Cellular mortality was assessed using a fluorescent probe. Increasing concentrations of the selective eta2-A agonist formoterol (Fo) were used to obtain a concentration-response curve and EC95, the Fo concentration that reduces cellular mortality by 95%. In a last group, NOS inhibitor (L-NNA) was added to Fo at EC95. Values are reported as mean ± SEM. Statistical analysis is performed by ANOVA followed by Fisher test. RESULTS: Fo produces a concentration-dependent reduction in cellular mortality (EC95=10mol/l). This protective effect is totally abrogated when adding L-NNA.

7

72 0.24 (0.01) 0.25 (0.01) 12.7 (1.0) 6.8 (1.0)

CONCLUSION: We found an initial rise in HK and PFK, followed by a down-regulation in PFK by 72 hours. PK and GAPDH activity were unaffected. Compared to the changes previously found in liver, there was much less change seen in muscle glycolytic activity. This could reflect the lesser metabolic role that skeletal muscle plays during the acute sepsis process. REFERENCE(S): 1. Ardawi et al. J.Lab Clin.Med. 1990; 115: 660-68 2. Arnold J et al. Clin Sci 1991; 80: 213-7 3. Karyampudi S et al. Intensive Care Med. 2002; 28(134):S96. Grant acknowledgement: This work is supported by the Wellcome Trust CONCLUSION: In ou in vitro H/R model, eta2-A stimulation confers EC protection involving NOS.

010

012

LYMPHOCYTE APOPTOSIS IS AUGMENTED BY A HYPERINSULINAEMIC EUGLYCAEMIC CLAMP DURING ACUTE ENDOTOXEMIA

PEROXYNITRITE BLOCKS THE ACTIVATION OF NFKAPPA B INDUCED BY TNFALPHA AND LPS IN CARDIOMYOCYTES

Nielsen J S1, Christensen V B1, Andersen S K1, Larson A2, Ledet T3, Tønnesen E1 1 Dep. of Anesthesiology and Intensive Care Med., Århus University Hospital, Århus C, 2Gentofte University Hospital, Copenhagen, 3Dep. of Pathological Biochemistry, Århus University Hospital, Århus, Denmark

Liaudet L1, Levrand S1, Pesse B1, Schaller M1, Feihl F2, Waeber B2 1 Division of medical critical care, 2Division of Clinical Pathophysiology, University Hospital, Lausanne, Switzerland

INTRODUCTION: Lymphocyte apoptosis is associated to both endotoxemia and sepsis. In the septic patient lymphocyte apoptosis has been suggested a negative factor due to the immunosuppressive effects of apoptotic cells on already compromised patients. Strict blood glucose control in critically ill surgical patients is associated with decreased morbidity and mortality rates. We have previously found that a hyperinsulinaemic euglycemic clamp(HEC) attenuates plasma levels of TNF-α, glucagon and free fatty acids during acute lipopolyssacarid (LPS) endotoxemia in pigs(1). The aim of this study is to investigate and estimate whether increased apoptosis during acute endotoxemia can be seen in both B- and T-lymphocytes. Furthermore we wanted to examine and estimate the possible effect of a HEC on lymphocyte apoptosis. METHODS: 40 pigs where randomly selected into four groups. After 1h of stabilization the pigs were anaesthetised, intubated and ventilated for another 81/2 hrs. Group I had no further intervention. Group II received a HEC (p-glucose 5mM, insulin infusion rate 0.6 mU kg-1min-1) for 81/2 hrs. Group III received a 11/2 h LPS infusion. Group IV received a combination of a HEC for 81/2 hrs and 11/2 hrs of LPS. After the 91/2 hrs the pigs where killed. Tissue samples were taken from the spleen and frozen. Four sections were cut from each sample. By stereological methods the number of apoptotic B- and T-Lymphocytes was estimated by fluorescence immunohistrochemistry with antiactive-caspase-3 and either anti-CD21 (B-lymphocytes) or anti-CD3ε (T-cells). The apoptotic Band T-lymphocyte data were analysed using a two-way ANOVA model. RESULTS: Our results showed that endotoxin infusion, independently of the clamp, induces apoptosis in B- and in T-lymphocytes (p<0.001, p=0.008). The ratio between non-endotoxin-infused and endotoxin-infused is estimated to be 0.40(CI 0.29;0.56) for the B- and 0.63(CI(0.45;0.88) for the T-lymphocytes. We have also shown that hyperinsulinaemia and euglycaemia independently of endotoxin infusion, increases the number of apoptotic lymphocytes. The ratio between the non-clamped and the clamped pigs is estimated to be 0.69(CI(0.50;0.97) for the B- and 0.69(CI(0.50;0.96) for the Tlymphocytes. CONCLUSION: We have shown that the number of apoptotic B- and T-lymphocytes increases during acute endotoxemia in pigs. A novel finding was that the HEC itself increases the number of apoptotic lymphocytes and furthermore augments the number of apoptotic B- and T-lymphocytes during acute endotoxemia. REFERENCE(S): (1)Christensen VB, Anesthesiology 100(4):861-870,04-2004. Grant acknowledgement: SSVF /The AP Møller Foundation for the Advancement of Medical Science /Politimester JPN Colind Mindelegat

INTRODUCTION: Peroxynitrite (PXN), formed from the reaction between nitric oxide and the superoxide radical, is a potent oxidant involved in myocardial reperfusion injury. In addition to direct toxicity, PXN may also indirectly modulate redox sensitive cell signaling pathways, such as that involving the transcription factor nuclear factor kappa B (NFKB). NFKB is activated by many extracellular signals and results into the transcription of pro-inflammatory and anti-apoptotic genes. The potential role of PXN in the process of NFKB activation is not known. Thus, we evaluated the influence of PXN on NFKB activation in vitro and we determined whether PXN might modulate NFKB activation induced by TNF-alpha(TNFa) and lipopolysaccharide (LPS) METHODS: Rat cardiomyocytes (H9C2 cells) were exposed to PXN (10-250 microM) for 20min, then replaced in culture medium for 10min to 4h. In parallel experiments, cells were stimulated with TNFa (20ng/ml) or LPS (1microg/ml), with or without PXN pretreatment. After stimulation, NFKB activation was determined by: (a) the degradation and phosphorylation of its cytoplasmic inhibitor IkappaB (western blotting), (b) the nuclear translocation of the p65 subunit of NFKB (western, immunocytochemistry) (c) DNA fixation of the NFKB complex (electromobility shift assay) and (d) the transcription of a transfected reporter gene (NFKBluciferase gene reporter assay). RESULTS: PXN did not activate NFKB, in contrast to a strong activation triggered by TNFa and LPS. Alternatively, PXN (250microM) suppressed NFKB activation induced by TNFa and LPS, by blocking IkappaB phosphorylation, p65 nuclear translocation and DNA fixation of NFKB. The suppressed NFKB activation by PXN resulted in a marked reduction of reporter gene transcription, which was already evident at 10microM PXN CONCLUSION: PXN blocks NFKB activation induced by TNFa and LPS in cultured cardiomyocytes. This discovery offers new perspectives in the understanding of the interactions between oxidative/nitrosative stress and inflammation Grant acknowledgement: Supported by grants 3100-65005.01 and PPOOB-68882/1 from the Swiss National Fund for Scientific Research to LL

S8

17th Annual Congress – Berlin, Germany – 10–13 October 2004

013

015

THE HEPATIC ARTERIAL BUFFER RESPONSE: EFFECTS OF LONG-TERM ENDOTOXEMIA

DOBUTAMINE DOES NOT INCREASE TISSUE OXYGENATION IN THE SKIN, THE GUT AND LIVER IN A PIG MODEL

Bracht H1, Porta F1, Takala J1, Ma Y1, Feng H1, Jakob S M1 Department of Intensive Care Medicine, Inselspital, Bern, Switzerland

1

Hager H1, Hiltebrand L2, Mandadi G2, Pestel G2, Talcott M2, Kurz A2 Anesthesiology, Medical University, Vienna, Austria, 2Anesthesiology, Washington University, St. Louis, United States

1

INTRODUCTION: The hepatic arterial buffer response (HABR) plays an important role as a hydrodynamic compensatory mechanism in states of reduced portal venous blood flow. The HABR is impaired during short-term endotoxemia (<6h) [1]. Our aim was to evaluate liver hemodynamics in a model of long term porcine endotoxemia. METHODS: Endotoxemia was induced in 8 anesthetized and mechanically ventilated pigs, 9 pigs served as controls. Blood flows were measured continuously with ultrasound flowprobes. To test the HABR, the superior mesenteric artery (SMA) was occluded and the hepatic arterial pressure was adjusted to pre-occlusion values by subsequent partial celiac trunk (CT) occlusion . The resulting blood flows were analyzed for at least 10 sec. The HABR was calculated in 2 different ways: as fractional hepatic arterial flow changes in response to the HABR manoeuvre, and as ratio of hepatic arterial (HA) and portal venous (PV) fractional flow changes. Statistical analysis was done using ANOVA for repeated measurements. Data are shown as mean (SD). RESULTS: During 18 hours of endotoxemia, cardiac output and PV blood flow increased significantly over time without intergroup difference, while HA flow remained unchanged. The HABR decreased over time in both groups without significant time-group interaction (Table 1.).

QHA CON [ml/min/kg] ETX QPV CON [ml/min/kg] ETX Frac. QHA CON [%] ETX Frac.QHA/Frac.QPV CON [%] ETX

baseline

6 hrs

12 hrs

18 hrs

3,2±1,3 4,9±4,1 17±4 21±5 47±28 39±21 74±41 58±81

3,6±1,7 4,4±4,0 20±6 27±9 40±42 57±83 72±99 88±112

3,9±2,1 4,5±3,1 22±9 29±6 49±40 36±29 108±87 47±64

4,1±1,8 3,0±1,4 26±8 28±8 33±23 14±34 51±37 4±50

time time-groupeffect P interaction P

INTRODUCTION: Oxidative killing is the primary defense against surgical pathogens. It has been shown that factors that improve subcutaneous tissue oxygenation reduce infection risk.(1) Dobutamine is an inotropic drug commonly used in patients with heart failure, which improves heart function and relaxes peripheral blood vessels. Its influence on heart and peripheral blood vessels may improve tissue perfusion and thus tissue oxygenation. We, therefore, tested the hypothesis that administration of dobutamine improves subcutaneous as well as intestinal or liver oxygenation. METHODS: 9 pigs, were studied after laparotomy under general anesthesia. The pigs underwent three treatments in randomized order: 1) no dobutamine. 2) 2.5 and 3) 5 mcg/kg/min dobutamine. FiO2 was maintained at 30% and endtidal CO2 was maintained at 35 mmHg. Oxygen tension was measured with 5 Licox tonometers (GMS, Germany) placed subcutanously at the left shoulder and in the wall of the small and large intestines, on the liver surface and in the liver parenchyma. The data was analyzed using repeated measures of ANOVA. P<0.05 was considered significant. Data are presented as means±SDs. RESULTS: Dobutamine at both doses significantly increased cardiac index, mixed venous saturation and mixed venous oxygen partial pressure. (Table 1) In contrast, tissue oxygenation in the small and the large intestines in the skin and in the liver remained unchanged during administration of both doses of dobutamine. Blood Oxygenation

0,849

0,069

0,019

0,489

0,028

0,725

30%FiO2 + Dob 2.5 mg/kg/min + Dob 5 mg/kg/min signific.

0,070

0,934

Tissue Oxygenation

CONCLUSION: We found no specific effect of prolonged endotoxemia on the HABR. In contrast, the HABR was impaired in both groups. This suggests that either major surgery, prolonged anesthesia, repeated HABR testing or their combination can impair the HABR, and that the exposure to endotoxin has minimal further effects on this. This indicates a partial loss of hepatic blood flow defence. REFERENCE(S): 1.Ayuse et al. Am J Physiol. 1995 Grant acknowledgement: Hendrik Bracht is supported by the young investigators award of the ESICM 2003

014 PEROXYNITRITE ACTIVATES ERK 1/2 VIA MEK1 BUT NOT EGF RECEPTOR AND RAF-1 KINASE IN CARDIOMYOCYTES Liaudet L1, Pesse B1, Levrand S1, Schaller M1, Feihl F2, Waeber B2 1 Division of medical critical care, 2Division of Clinical Pathophysiology, University Hospital, Lausanne, Switzerland INTRODUCTION: Peroxynitrite (PXN), formed from the reaction of nitric oxide with the superoxide anion, is a potent oxidant involved in myocardial reperfusion injury. In addition to direct toxicity, PXN may also indirectly modulate redox sensitive signaling pathways, such as the «mitogen-activated protein kinases» (MAPK), which are interconnected cellular proteins activated by various extracellular stimuli and which modulate gene expression. The 3 main MAPK include p42/p44 MAPK (ERK1/2), c-Jun NH2-terminal kinase (JNK) and p38. The potential role of PXN in the process of MAPK activation in the heart is not known. We have thus investigated the role of PXN on the activation of p42/p44 MAPK in vitro, and tried to dissect the effect of PXN on the pathways upstream of p42/p44 MAPK, represented by the epithelial growth factor receptor (EFGR), Raf-1 kinase, and MEK1 METHODS: Rat cardiomyocytes (H9C2 cells) were exposed to PXN (10-500microM) for 260min. In a second set of experiments, cells were stimulated with PXN (500microM, 15min) with or without inhibitors: AG1478 (EGFR inhibitor), ZM336372 (Raf-1 kinase inhibitor), and PD98059 (MEK1 inhibitor). p42/p44 activation was evaluated by the phosphorylation of ERK1/2 (western) and the kinase activity of ERK1/2 on the transcription factor Elk-1 (kinase assay). MEK1 activation was determined by the phosphorylation of MEK1 (western) RESULTS: PXN strongly activated ERK1/2, starting at 2min, maximal at 15min, and returning to baseline at 30min. ERK1/2 activation was secondary to MEK1 activation, being prevented by PD98059. In contrast, the activation of MEK1-ERK1/2 was not influenced by AG1478, or ZM336372, indicating that EGFR and Raf-1 are not involved in PXN-dependent ERK1/2 activation CONCLUSION: PXN activates ERK1/2 in cardiomyocytes through a pathway involving MEK1, but independently from the «classical pathway» using EGFR and Raf-1 kinase. This observation might contribute to better understand the myocardial response to oxidative/nitrosative stress during reperfusion Grant acknowledgement: Supported by Grants 3100-65005.01 and PPOOB-68882/1 from the Swiss National Fund for Scientific Research to LL

30%FiO2 +Dob 2.5 mg/kg/m + Dob 5 mg/kg/m signific

CO 2.34±0.69 4.47±1.03 5.52±1.12 p<0.001

vPO2 37.34±3.32 47.53±2.65 48.93±3.17 p<0.001

vSO2 72.02±7.30 82.37±5.39 86.18±2.48 p<0.001

artPO2 108.11±20. 98.59±22.3 105.03±21. ns

Skin PsqO2 Large intest. Small intest. Liver surface Liver parench. 91.88±40 50.72±17 34.85±18 51.58±21 37.86±7 90.11±38 60.81±11 39.27±15 59.53±17 37.19±5 84.73±41 62.72±12 39.22±15 62.12±19 38.38±6 ns ns ns ns ns

CONCLUSION: Dobutamine improves cardiac output and mixed venous oxygenation, but does not show significant improvement in gut, skin and liver tissue oxygenation in the pig model. This data suggests that although dobutamine increases systemic blood flow it does not effect splanchnic oxygenation. REFERENCE(S): (1) Haley, R.W. et al., Am J Epidemiol, 1985. 121: p. 206-215.

Oral Presentations Evaluation of prognosis and outcome – 016-020 016 TREATMENT INTENSITY AND OUTCOME OF PATIENTS AGED 80 AND OVER IN INTENSIVE CARE UNIT Boumendil A1, Aegerter P2, Guidet B3, CUB-Réa 2 U444, INSERM, 2Department of Biostatistics, Hôpital Ambroise Paré, AP-HP, 3Medical Intensive Care Unit, Hôpital Saint Antoine, AP-HP, Paris, France

1

INTRODUCTION: This work was undertaken to determine whether patients older than 80 have similar treatment intensity than younger patients in Intensive Care Unit (ICU). METHODS: In order to compare similar population we choose to identify differences in treatment intensity, hospital costs and outcome between patients aged 80 and over (oldest old) and patients aged between 65 and 79 years old (young old). Data were extracted from a multicenter database including 36 ICUs in the Paris area (France). 2,299 oldest-old patients were retrospectively matched to 2,299 young old patients admitted in one of the participating ICUs during a 4-year period. The matching criteria were: severity status on admission (± 5) (assessed by SAPS II corrected from the points of age), organ failure, type of stay (surgical vs medical), location before admission, sex, ICU, year of ICU admission, Charlson comorbidity index. The underlying condition was classified using the Mac Cabe classification. The functional status was assessed using the Knaus classification. The ICU workload was assessed by using the OMEGA scoring system. RESULTS: Oldest old patients had lower length of ICU stay and greater hospital mortality (adjusted OR: 1.41 ; 95% CI: 1.17-1.71) than matched young old patients. Total and daily workload were lower in oldest old than in matched young old patients group. Estimated mean direct medical cost per stay was approximately 900 lower for oldest old patients. Older patients were less often mechanically ventilated (adjusted OR: 0.66; 95% CI: 0.56-0.77), had less tracheotomy (adjusted OR: 0.34; 95% CI: 0.23-0.48) and renal support (adjusted OR: 0.66; 95% CI: 0.47-0.93) than matched young old patients. CONCLUSION: Oldest old patients are probably “discriminated” in ICU, since their ICU stay characteristics are different from those of younger patients: they receive less treatment in ICU even after adjustment on severity and illness.

17th Annual Congress – Berlin, Germany – 10–13 October 2004

017

019

RISK FACTORS FOR MORTALITY AMONG PATIENTS ADMITTED WITH HEATSTROKE IN FRENCH ICUs IN AUGUST 2003

SELECTION CRITERIA FOR GENERAL ICU-REPRESENTATIVE CASE-MIX

Misset B1, Gattolliat O1, Bastuji-Garin S2, Boughrara E1, De Jonghe B3, Garrouste-Orgeas M1, Annane D4, Hausfater P5, Carlet J1 1 ICU, Hopital Saint-Joseph, Paris, 2Statistics, Hopital H Mondor, Creteil, 3ICU, Hopital Poissy-St Germain, Poissy, 4ICU, Hopital Raymond Poincaré, Garches, 5Emergency, Hop Pitie-Salpetriere, Paris, France INTRODUCTION: The heatwave in August 2003 was responsible for an extimated 15,000 death in France. The aim of this study was to assess the outcome and risk factors for mortality among those patients admitted with heatstroke (HS) to French ICUs. METHODS: A questionnaire was sent to all 380 French ICUs which have at least 6 beds. A diagnosis of HS required the presence of fever over 40.5°C, neurological signs (confusion, convulsions, or coma) on the day of ICU admission, and the absence of any other cause of fever. The questionnaire assessed: factors predisposing to HS, Saps 2 score and other measures of acute severity, therapeutic modalities, and outcome data. An assessement of mortality risk factors was performed using uni and multivariate analyses. All units which did not return a questionnaire were contacted by telephone to determine their HS case load. RESULTS: Fourty-two percent of the surveyed ICUs responded, providing data on 348 cases of HS. Telephone follow-up showed this represented 80 % of all the cases admitted to French ICUs. Their age was 65 +- 17; 75% were living at home ; 67 % had an underlying disease predisposing to HS; their Glasgow coma score, prothrombin time, platelet count, and blood creatinin were 5.7 +- 3.2 pts, 26 +- 13 sec., 99,000 +- 108,000 /mL, and 207 +- 171 µMol/L respectively. Their Saps 2 score was 70 +- 21 and hospital mortality 65 %. Fourteen items were significantly different between survivors and non-survivors through an univariate analysis. A logistic regression showed that mortality was independently associated with a high Saps 2 score (OR 1.09 [95%CI 1.06 – 1.12]), a high prothrombin time (1.02 [1.01 – 1.04]), the presence of prior cardiac insufficiency (5 [1.25 - 10]), and admission to an intensive care unit which did not have air-conditioning (2.1 [1.1 – 4.6]). CONCLUSION: In this population, the overall mortality was higher than in the previous reported cohorts, but coherent with the level of severity measured with the Saps 2 score at ICU admission. Apart from items measuring severity of the acute disease, the presence of prior cardiac failure and the absence of ICU air-conditionning were independant risk factors for hospital mortality. The latter suggests that cooling plays a major role in the therapy of heatstroke.

S9

Mistraletti G1, Giacomini M1, Cigada M1, Corbella D1, Ciarloni A1, Morabito A2, Miranda D3, Iapichino G1 1 Istituto di Anestesia e Rianimazione, 2Istituto di Statistica Medica, dell’Università degli Studi di Milano, Azienda Ospedaliera - Polo Universitario San Paolo, Milano, Italy, 3EURICUS, Coordinating Centre, Bruges, Belgium INTRODUCTION: The case-mix of all patients (pts) admitted to general-ICU is rather different when considering the severity of illness and the complexity of treatment. We sought and applied the best criteria to select patients critical enough to become eligible for prospective studies or to monitor the quality of care in ICU. METHODS: All pts admitted in 89 ICUs from 12 European countries over a 4 month period in a prospective, multicentre, observational study (EURICUS-I), were enrolled. Data on demographic/clinical statistics, severity at admission and a daily score of nursing complexity/workload, were collected. Hospital mortality (Hdead) and lenght of stay (Hlos) were used as bench marks for the selection of the “critical” population. Determinants of Hdead were studied with logistic regression; determinants of Hlos were studied with Cox regression model. Then, pts were selected by two METHODS: a) SAPSII > 32 points (case-mix median) and ICU lenght of stay > 24 hours (to exclude pts too sick or not sick enough) (severity group); b) high treatment1 (HT) > 2 days (case-mix median) (complexity group). RESULTS: 12,615 pts were enrolled. The parameters significantly related to an increase in Hdead were: acute physiological score of SAPS II, age, surgical unscheduled or medical admission, consecutive HT days, NEMS score and level of care (HT or low treatment, LT) of the last day in ICU, admission from ward or other ICU. Age, surgical unscheduled or medical admission and admission from ward were significantly related to an increase in Hlos. Pts selected by severity method were 3,882, those by complexity were 2,885 and through both methods a third group of 2,055 pts was selected. All significant variables related to Hdead and Hlos were more prevalent in the third group. Moreover, adding the variable of belonging to the severity group (ptsSAPS) or to the complexity group (ptsHT), the model sensitivity increased, being these two variables both related to Hdead and Hlos. Using log-likelihood ratio statistics, we demonstrated that ptsSAPS and ptsHT were significantly and independently related in describing Hdead (ptsSAPS: p = 0.0016 and ptsHT: p = 0.0096) and Hlos (ptsSAPS: p < 0.0001 and ptsHT: p < 0.0001). CONCLUSION: A significantly more critical population can be identified combining the two methods rather than using only one. Then, this allows the selection of the typical “case-mix” population to be enrolled in prospective or retrospective clinical or audit studies. REFERENCE(S): 1 Iapichino G. et al. (2001) Daily classification of the level of care... Intensive Care Med 27:131–136.

018

020

USING THE VARIABLE LIFE ADJUSTED DISPLAY (VLAD) TO ASSESS ICU PERFORMANCE IN MANAGING BACTERAEMIA

BODY MASS INDEX (BMI) AND HOSPITAL MORTALITY

Corona A1, Wilson P1, Singer M1 1 Bloomsbury Institute ICM, University College, London, London, United Kingdom INTRODUCTION: Several methods exist for estimating the mortality risk of critically ill patients with bloodstream infections (BSI). The Variable Life Adjusted Display (VLAD) can be used to examine whether cumulative differences exist between observed and expected patient mortality, i.e., the adjusted predictive death risk (PDR) derived from the 1st 24 hour APACHE II score (1). METHODS: We performed this analysis using data collected from a prospective 6 month observational study in a 22 bedded medical-surgical ICU. The VLAD plot shows the difference between expected and actual cumulative mortality of all patients with BSI [(community (CA-B), hospital (HA-B) and ICU-acquired (ICUA-B)]. This mortality-scoring system accumulates ‘penalties’ for each death and ‘rewards’ for every survivor based on the inherent PDR of each case concerned. SPSS (SPSS inc. Chicago Ill) was used for statistical analyses. RESULTS: Of the 713 patients admitted, 13 (1.8%) were admitted with CA-B and 22 (3.1%) with HA-B. Thirty five (4.9%) had ≥1 ICUA-B during their ICU stay. The mean (SD) APACHE II score and PDR were significantly higher (p<0.05) both for CA-B [21.6 (5.3) & 0.427 (0.172)] and for HA-B [24.7 (7.2) & 0.482 (0.23)] compared to ICUA-B [20.2 (7.1) & 0.324 (0.231)], and for patients without BSI [16.2 (7.7) & 0.181 (0.224)]. The hospital overall crude mortality was 38.5% for CA-B, 63.6% for HA-B, and 49% for (ICUA-B) compared to 20.7% for non-BSI patients (p<0.0001, chi2 test). VLAD plots showed a net mean (SD)/total positive life gain of 0.43 (0.496)/0.553 for CA-B and 0.159 (0.606)/0.321 for ICUA-B over the predicted number of survivors, whereas a net mean (SD)/total [-0.518(0.23)/-3.285] negative life gain was recorded for HA-B (p<0.05, ANOVA test). On the other hand, for non-BSI patients we found a higher, albeit non- significant (p=0.13), positive mean(SD)/total [0.08(0.374)/2.526] life gain. CONCLUSION: VLAD provides a graphical display of risk-adjusted survival figures for individual type of BSI over time and could be modified to monitor performance over a range of treatments and outcomes. It confirmed that critically ill patients developing BSI are generally more severely ill, particularly those with CA-B and/or HA-B. REFERENCE(S): 1.Lovegrove J, Valencia O, Treasure T, Sherlaw-Johnson C, Gallivan S. Monitoring the results of cardiac surgery by variable life-adjusted display. Lancet. 1997 Oct 18;350(9085):1128-30.

Bosman R J1, Peelen L M2, De Jonge E3 1 ICU OLVG, On behalf of the NICE foundation, 2KIK, 3ICU, AMC, Amsterdam, Netherlands INTRODUCTION: Although there is discussion of the protective value of overweight, most studies are in agreement that lower BMI is related to higher mortality. METHODS: In order to investigate of the relation between BMI and mortality also existed in our population, we used the Dutch National Intensive Care Evaluation (NICE) database. All for APACHE II eligible patients (39662)from 1997 to 2003 were included. RESULTS: The lower BMI groups had significantly higher percentages of neoplasms (p = 0.007) and haematological malignancies (p < 0.0001).

CONCLUSION: Low BMI is associated with higher mortality, even if corrected for severity of illness. This phenomena can partly be explained by the higher incidence of neoplasm and haematological malignancies. REFERENCE(S): Garrouste-Orgeas M et al Intensive Care Med 2004; 30(3):437-443.

S10

17th Annual Congress – Berlin, Germany – 10–13 October 2004

Oral Presentations Head injury 021-025 021

023

USEFULNESS OF INTRACRANIAL PRESSURE MONITORING IN PATIENTS WITH SEVERE BRAIN TRAUMA AND ISCHAEMIA

1

Wolowicka L1 1 Intensive Care and Pain Clinic, University of Medical Sciences, Poznan, Poland

INTRODUCTION: Dramatically improved results by modern neurointensive care (NIC) and increasing demands on cost-benefit aspects strengthen the need to optimise early prediction of outcome for clinical decision making and for the clinical audit process. The aim is to develop a simple and cost-effective model for prediction of outcome after moderate and severe traumatic brain injury (TBI).

RANKING OF OUTCOME PREDICTORS IN NEUROINTENSIVE-CARE BRAIN TRAUMA PATIENTS BY BAYESIAN STATISTICS

INTRODUCTION: Pathophysiological state of severe head injury or brain ischaemia results in oedema and increased intracranial pressure (ICP). Additionally the course of intensive care play important role in ICP changes deteriorating the disturbances of brain. The availability of monitoring of ICP and CPP in intensive care units is continually limited (1). The aim of the study was to assess the changes of ICP and CPP during osmotherapy and analgosedation and to estimate the influence of the nursing and physiotherapeutical procedures to ICP trends. METHODS: Investigation was prospectively carried out on 20 patients (age 18-65) after head trauma or brain ischaemia with Glasgow Coma Scale below 8 p. All patients were control ventilated and diagnosed TC, dominant position was supine and axial head elevated 35 degrees. Investigation started after family approval and insertion of epidural device connecting with the monitoring system. For analgosedation we used thiopental (0,5-1,5mg/kg/h) or midazolam (0,030,2mg/kh/h) with fentanil (100-200ug/h). For the brain protection mannitol (0,24-0,5G/kg every 46 hours) and lidocaine (1,0-2,0mg/kg/h) were used. RESULTS: Monitoring of ICP and CPP confirmed dynamic changes after brain injury or ischaemia and important usefulness for modification of intensive care and nursing. Infusion of sedativa diminished fluctuations of ICP and increases of CPP but without any differences between used drugs (p=ns). During 30 minutes after mannitol we observed significant ICP decreasing and CPP increasing (Pearson corr. 0,96-0,99). Procedures as tracheal suction, position changes etc. caused ICP increases and CPP decreases in patients without any sedation in comparison with stabilised ICP after proper sedation (p less 0,05). Pearson corr. index between performed procedures and ICP was 0,91 while between these procedures and MAP was 0,98. CONCLUSION: 1. Continous ICP, CPP monitoring is useful method for estimation of osmotherapy and analgosedation after severe brain trauma and brain ischaemia. 2. Nursing and physiotherapeutic procedures induce the significant increases of ICP and should be performed with ICP monitoring or deepen sedation.

Stalhammar D A1, Ljungqvist J C1, Skoglund T1, Lindstrom L2, Nylén K3 Div of Neurosurgery, 2Dept of Clinical Physiology, 3Div of Neurology, Inst of Clinical Neuroscience, Goteborg, Sweden

METHODS: Data for 57 NIC TBI patients included: GOS at one year dichotomised in bad (D/V/SD) and good (MD/GR) outcomes. Outcome predicting variables were: age, sex, alcohol abuse, previous diseases, ISS, RLS, trauma energy, pupil reaction, the APACHE III, ICP elevation and two CT-scoring systems. The data were split randomly into two groups. One group was used to determine the conditional probabilities of having certain predictor values given the outcome. The other group was used to test the predictions through iterative use of Bayesian statistics to give the probability of the outcome. This prediction was then compared with the true outcome. The two groups were interchanged and the same procedures were repeated. The final result, taken as the mean of the two runs, was described as the distance from the identity line in the ROC plot, which is equal to the sensitivity plus the specificity minus one. Linear univariate and multivariate logistic regression modelling was also applied as was correlation measurements based on entropy calculations. RESULTS: Single prediction variables yieleded: earlier diseases (0.51), age (0.44), alcohol abuse (0.33) APACE III (0.27), TCDB-CT-scale (0.26), pupil reaction (0.25), Edinburgh-CT-scale (0.20), trauma energy (0.19), sex (0.18), RLS (0.11), ICP elevation (0.09), and ISS (0.04). When all predictors were used we found the ROC distance to be 0.61, which corresponds to a sensitivity of 0.75 and a specificity of 0.86. Stepwise logistic regression gave age, earlier disease, pupil reactivity and alcohol abuse with a significance level less than 0.05. The entropy based correlation gave essentially the same ranking as listed above. CONCLUSION: Reasonably accurate and simple prediction models were derived for further validation and clinical tests. Grant acknowledgement: The Health and Medical Care Executive Board of the Västra Götaland Region, Sweden. BIOMED (EU), BMH4-CT98-3406 (ERESMUS in COPD)

REFERENCE(S): 1. Rosner MJ, Daughton S, Cerebral perfusion pressure management in head injury, JTrauma 1990, 30, 933-940

022

024

QUANTITATIVE ANALYSIS OF INTRACRANIAL PRESSURE AND ITS TREATMENT IN HEAD INJURY

DETERMINATION OF EARLY PREDICTORS OF DEATH IN TRAUMATIC BRAININJURED (TBI) PATIENTS

Longhi L1, RoncatiZanier E1, Colombo A1, Ortolano F1, Ferrari C1, Stocchetti N1 Anesthesia and Critical Care Medicine, Ospedale Maggiore Policlinico IRCCS, Milano, Italy

Welschbillig S1, Yassine H1, Benlolo S1, Mateo J1, Rossignol M1, Vicaut E2, Payen D1 1 Department of Anesthesiology and Intensive Care, 2Clinical Research Unit, Lariboisière Hospital, Paris, France

1

INTRODUCTION: It has been shown that the proportion of hourly intracranial pressure (ICP) > 20 mmHg is a sensitive outcome predictor[1], and it is commonly recommended that treatment starts at an ICP = 20-25 mmHg[2]. To carefully evaluate the incidence of raised ICP we have measured the percentage of time in which ICP was > 20 mmHg using a computerized method. METHODS: In 171 severe TBI patients (median post-resuscitation motor GCS 4) ICP was monitored through subdural, ventricular or intraparenchimal probes. ICP analog data, sampled at 20 Hz, were digitized and sent to a computer. After manual filtering (to exclude artifactual data) the recorded files were analyzed using a visual-basic program (ICP analyzer, by A. Colombo). Raised ICP was treated with first tier therapies (sedation, analgesia, cerebrospinal fluid withdrawal, Mannitol and moderate hyperventilation); when these therapies were ineffective second tier interventions (Barbiturates, profound hyperventilation) were used. RESULTS: A total of 18090 hours was processed; filtering lead to the exclusion of approximately 9% of recording time. Intracranial hypertension, of various duration, occurred in 96% of the patients. Application of second tier therapies increased linearly with the proportion of the time of raised ICP as shown in table. Legend following table: Pts = patients; C: 2 Pts 2 tier Thp = patients with second tier therapies. D = % of column A referred to column C % time ICP>20 0 1-10 11-25 26-50 51-75 >75

A Pts 7 79 23 37 17 8 171

B% 4 46 13 22 10 5 100

C Pts 2 tier Thp 0 8 7 22 12 6 55

D% 0 10 30 59 71 75 32

CONCLUSION: A computerized system of ICP recording, linked to a coded recording of therapeutic interventions, leads to more than 90% of good data points. Analysis of such data shows that raised ICP is extremely frequent in TBI patients and requires aggressive management. REFERENCE(S): 1.Marmarou A., Eisenberg H.M., Foulkes M.A., Marshall L.F. Impact of ICP instability and hypotension on outcome in patients with severe head trauma. J Neurosurg 1991;75:S59-S66. 2.The Brain Trauma Foundation. The American Association of Neurological Surgeons. The Joint Section on Neurotrauma and Critical Care. J Neurotrauma 2000;17:479-511.

INTRODUCTION: Prognosis of TBI depends partially on the intensity of resuscitation and multimodal monitoring. 25 to 30 % of severe TBI continue to die despite aggressive treatment (1). The aim of this study was to determine if early clinical, radiological or biological classic parameters may predict mortality in TBI patients in one surgical intensive care unit, applying a homogenous therapeutic strategy. METHODS: We have collected all consecutive TBI patients admitted between 1994 and 2003. Recorded data: a) age, gender, initial Coma Glasgow Score (CGS), initial CT Scan, occurence of a mydriasis, blood pressure, heart rate, tidal volume, SaO2; b) multimodal parameters: intracranial pressure (ICP), jugular venous oxygen saturation (SvJO2), middle cerebral artery blood velocity for systole (SysABV) and diastole (DiasABV)(transcranial Doppler); c) biological parameters: initial glycemia, natremia at 24hrs; d) scores: ISS, Revisited Trauma Score (RTS), SAPS II, SOFA. For each patient, mortality and morbidity was evaluated at 6-month post injury and scored with the Glasgow Outcome Score (GOS). Step by step univariate and multivariate analysis were performed. RESULTS: Over ten years, 210 TBI patients (mean age = 37yrs ± 16SD, median CGS = 7) were hospitalized. Outcome was: 20% died, 8% had severe disability (GOS 2 to 3). The univariate analysis showed that mortality was associated with: initial CGS (p=0.0002), presence of a mydriasis (p<0.0001) and/or brainstem injury (p<0.0001), low diasABV (p=0.0001), arterial hypotension, norepinephrine or blood transfusion requirements (p<0.0001), initial hyperglycemia (p<0.0001), the first day natremia (p<0.0001) and scores (SAPS II, SOFA, ISS, RTS)(p<0.0001). The multivariate analysis showed that mortality was significantly associated with brainstem injury (odds ratio OR=24 confidence interval [CI]3.24-179 ; p=0.0013), the diasABV (p=0.0046), the first day natremia (p=0.0313) and SOFA score (p=0.0052). CONCLUSION: Most of the outcome related parameters depends on the type and the severity of brain injury. Few items can be modified by therapeutic strategy such as hypotension, cerebral edema, low blood flow velocity and hyperglycemia or hypernatremia. Based on these findings, prospective trial on control of these parameters could be performed to improve TBI patients’ outcome. REFERENCE(S): (1) Bulger EM, Crit Care Med. 2002;30(8):1870-6

17th Annual Congress – Berlin, Germany – 10–13 October 2004

S11

025

027

RISK FACTORS AND THE IMPACT ON THE FUNCTIONAL OUTCOME IN PATIENTS WITH SEVERE HEAD INJURY

HAND DISINFECTION COMPLIANCE ACCORDING TO WORK EXPERIENCE

Rincón-Ferrari M1, Amaya-Villar R1, Flores-Cordero J1, Valencia J2, Campanario A1, MuñozSanchez M1, Murillo-Cabezas F1 1 Intensive Care Unit, 2Department of Neurosurgery, Hospital Universitario Virgen del Rocío, Sevilla, Spain INTRODUCTION: To investigate the risk factors associated with poor neurological outcome at six months in patients with severe head injury (HI). METHODS: From January 1998, we prospectively include in a data base all patients with HI admitted to our neurocritical ICU, collecting a series of clinical and monitoring data as well as the outcome of these patients. We have selected 437 patients with severe HI and with the 6-months post-injury neurological outcome recorded. Severe HI was considered a Glasgow Coma Scale (GCS) score of 8 or less occurring on admission (postresuscitation) or during the ensuring 48 hours. Neurological functional outcome was assessed by the Glasgow Outcome Scale (GOS) at six months of injury (good neurological outcome: 4 and 5; and poor neurological outcome: 1, 2, and 3). The whole group 437 patients were divided into 2 groups. Group 1: patients with severe HI and poor neurological outcome, and group 2: patients with severe HI and good neurological outcome. Univariate and multivariate analysis were carried out with SPSS 11.0. The level of significance was p<0.05. RESULTS: The mean age was 33,9 (18, 13) years. The mean GCS at ICU admission was 5.7 (1.6), and the mortality rate was 22,9%. One hundred and fifty-four patients (35,2%) showed poor neurological outcome (Group 1) and 283 (64,8%) good neurological outcome (Group 2). Univariate analysis of risk factors for poor outcome showed the following variables: age [43,2 (20,8) vs 28,9 (14)] years, GCS score [ 4,8 (1,7) vs 6,2 (1,4)], Acute Physiology And Chronic Health Evaluation (APACHE II) score [18,5 (4,8) vs 14,2 (4,4)], hypoxemia in the first 24 hours (29,9% vs 20,8%), arterial hypotension in the first 24 hours (48,1% vs 25,1%), shock (13% vs 5,7%), anaemia (51,9% vs 31,8%) and the development of intracranial hypertension -IH- (76,6% vs 48,4%). In the multivariate analysis, risk factors to develop poor neurological outcome were age, GCS score at admission, type of lesion on head computerized tomography scan [according to Marshall’ Scale -Traumatic Coma Data Bank- (TCDB)], arterial hypotension in the first 24 hours, and the development of IH. CONCLUSION: Age, GCS score, and greater severity of lesions on head computerized tomography scan (TCDB) are risk factors associated with poor neurological outcome in patients with severe head injury. Importantly, arterial hypotension in the first 24 hours and the development of intracranial hipertensión are the only two of all risk factors associated to poor neurological outcome that could be avoidable potentially.

Oral Presentations Prevention and RCT – 026-030 026 STRATEGY FOR PREVENTING ANTIMICROBIAL RESISTANCE Brahmi N1, Blel Y1, Kouraichi N1, Thabet H1, Amamou M1 1 intensive Care Medicine, Camur, Tunis, Tunisia INTRODUCTION: Antibiotic resistance with some micro-organism has become a worldwide concern. The aim of this study is to reduce antimicrobial resistance (enterobacteria, acinetobacter baumanii, pseudomonas aeruginosa) in our ICU. METHODS: We compare two periods of clinical practice between 2001 and 2003 after new strategy of antibiotic prescription: Modification of behavior of physicians (application of guidelines, restriction of certains antibiotic classes : quinolones, optimization of choice and duration of empiric therapy and desescalade until receiving antibiogram), controlled of all prescriptions of high potential antibiotic by staffed doctors RESULTS: 391 BGN were cultured during the period of study: 280 in 2001 and 211 in 2003.We noted a significant decrease of resistance of Acinetobacter Baumannii and Pseudomonas Aeruginosa for Ticarcilline (76% vs 63%* and 59% vs 39%*), Pipperacilline (97% vs 85%* and 58% vs 29%*), Imipenem (37% vs 30% and 61% vs 41%*), ceftazidime (95% vs 84% and 47% vs 33%*), Ofloxacin(95% vs 87% and 68% vs 63%),Amikacin (85% vs 79% and 63% vs 15%*). The principal mechanism of resistance was acquired cephalosporinse and the loose of Porin D2 in 2001.We noted more sensible phenotypes of Acinetobacter in 2003(11/76) than in 2001(1/89*). Resistance of Enterobacteria was significantly decreased for aminoside: Amikacin(30% vs 13%*), Tobramycin(60% vs 29%*), Netilmycin8% vs 25%).The mechanism of resistance(BLSE)was decreased from 72% to 29%*.* p<005 CONCLUSION: Revision of strategy of antibiotic prescription would be likely to have a significant impact on reducing antimicrobial resistance.

Chierego M1, Noritomi D1, Byl B2, Vincent J L1 1 Intensive Care, 2Hygiene, Erasme Hospital, Bruxelles, Belgium INTRODUCTION: Despite being widely known as the most important action to control spread of nosocomial infection, hand-washing compliance among health care workers still remains low. In some studies it appears that the duration of work experience may interfere with the compliance rate, but no study specifically addressed this question. The objective of this study was to evaluate if hand disinfection compliance rates in health care workers in the ICU vary according to their years of work experience. METHODS: We conducted a prospective observational study during a 3-week period (May-June 2003). The Department contains 31 beds divided in 4 sub-units with individual bedrooms. Elbow operated dispensers of antiseptic solutions for hand rubbing, and hand washing facilities are located inside each bedroom. Four trained observers recorded potential opportunities for, and actual performance of, hand hygiene during 20-min observation periods (most (97%) during the day on weekdays) over the 3 weeks. Each observer monitored 3 or 4 bedrooms during each session. The health care workers studied were not informed of the study details and the observers remained as unobtrusive as possible. The level of experience was classified as high, medium and low.

RESULTS: A total of 727 opportunities for hand disinfections were observed during 97 sessions (32 hours). The opportunities were 593 (81.6%) for nurses, 70 (9.6%) for physicians, and 64 (8.8%) for physiotherapists. The average number of disinfection opportunities was 7.05 opportunities per bed per hour. Overall the compliance with hand disinfection recommendations was 27.9% (203/727). The average hand disinfection compliance rates were not different among the 3 defined groups (p=0.56) and were the same within each professional category. Experience Low Medium High

No opportunities (%) 75 (10.3) 431 (59,3) 222 (30.4)

No hand disinfections (compliance %) 23 (30.7) 114 (26.4) 66 (29.9)

CONCLUSION: Overall compliance rates were poor and the level of work experience was not correlated with hand disinfection compliance rates. This result is important as it suggests that, even in a teaching hospital, poor hygiene practice learnt early does not change with time. Adequate hand hygiene needs to be stressed early in training and reinforced regularly.

028 PROSPECTIVE RANDOMISED CONTROLLED TRIAL ON THE USE OF PRESSURE DECREASING INTERFACES IN ICU PATIENTS Malbrain M L N G1, Hendriks B1, Wijnands P1, Denie D1, De Keulenaer B1, Jans A1, Van Pellicom J1 1 Intensive Care Unit, Ziekenhuisnetwerk Antwerpen, site Stuivenberg, Antwerpen, Belgium INTRODUCTION: Little is known about the efficacy of different pressure decreasing interfaces (PDI) to prevent pressure ulcers in the ICU (1). This study was designed to look for possible differences between a static (Roho, USA) and a dynamic (Nimbus III, Huntleigh Healthcare, UK) PDI in medical ICU patients. METHODS: A prospective randomised controlled clinical trial looking at differences between 2 groups of 8 medical ICU patients. Patients were included in the study either on ICU admission if the pressure ulcer risk was very high (Norton scale less than 8) or after the development of pressure ulcers on a standard hospital mattress. Patients were randomised in a blinded fashion between the 2 PDI’s. Pressure ulcers were scored on surface area, standard grading (I to IV) and the National Pressure Ulcer Advisory Panel (NPUAP) score (www.npuap.org). Statistical analysis was done with unpaired student’s t test via Excell software. Mean APACHE-II score was 25±8.2, NORTON score 7.2±0.7, SOFA 11.4±3.3, BMI 23.1±4.9, BSA 1.75±0.19, age 64.7±15.6, M/F ratio 1/1. Mean study duration was 11.3±4.7 days. The two groups had similar patient characteristics, severity scores and pressure ulcer risk factors except for age (56.9±16.3 in the Nimbus versus 71.6±11.9 in the Roho group, p=0.02) and prealbumine (6.7±3.6 in the Nimbus vs 20.3±12.4 in the Roho group, p=0.02). RESULTS: The evolution (∆) of the ulcer characteristics between admission and end of study is listed in the Table. On admission 62.5% of patients in the Nimbus group had pressure ulcers (9 ulcers in 5 patients: 20%I, 60%II, 20%III) versus 50% in the Roho group (4 ulcers in 4 patients: 50%I, 50%II). 25% of patients developed de novo ulcers in both groups (100% grade I in the Nimbus group and equally divided grades I and II in the Roho group). In the Nimbus group 82% of the ulcers ameliorated and 18% remained stable while in the Roho group none of the ulcers ameliorated and 33% remained stable. Significant ulcers (grade III or higher) were present in 20% of the Nimbus versus 0% of the Roho group on admission and in 0% versus 44.4% respectively at the end of the study. Ulcer characteristics evolution (∆) ∆Surface ∆NPUAP ∆Grade

Nimbus -2.1±2.3 -1±1.6 0±0.6

Roho 25.8±46.1 3.4±4.8 0.8±1

p-value 0.05 0.01 0.03

CONCLUSION: The use of a dynamic PDI was superior than a static PDI in preventing the development of new significant pressure ulcers and in the reduction of the number of significant ulcers. The dynamic PDI resulted in a significant decrease in pressure ulcer surface, NPUAP score and grade. We advocate the use of a dynamic rather than a static PDI in high risk medical ICU patients. REFERENCE(S): Hofman A et al. Lancet 1994; 343: 568-571. Grant acknowledgement: The PDI’s were provided free of charge for the study purpose

S12

17th Annual Congress – Berlin, Germany – 10–13 October 2004

029 A RANDOMISED CONTROLLED TRIAL ASSESSING THE EFFICACY OF ORAL CHLORHEXIDINE IN VENTILATED PATIENTS Macnaughton P D1, Bailey J1, Donlin N1, Branfield P1, Williams A1, Rowswell H1 Care Unit, Derriford Hospital, Plymouth, United Kingdom

1Critical

INTRODUCTION: An important factor in the pathogenesis of ventilator associated pneumonia (VAP) is aspiration into the lower respiratory tract of oro-pharyngeal secretions which have become infected or colonised with pathogenic bacteria. Chlorhexidine is bacteriostatic to the gram positive and gram negative bacteria which cause VAP and its topical administration to the oropharynx has been shown to be effective in reducing nosocomial pneumonia in patients undergoing cardiac surgery (1). We have investigated whether topical chlorhexidine is effective in influencing the colonisation of the lower respiratory tract and reducing the impact of VAP in a general ICU population. METHODS: Patients admitted to a mixed surgical-medical ICU of a University Hospital who were predicted to require ventilatory support for at least 48 hours were eligible for recruitment. Patients were randomised to the twice daily topical oropharyngeal administration of 0.2% chlorhexidine or a placebo of identical appearance and smell. Investigators and carers were blinded to treatment group. Non-directed broncho-alveolar lavage (BAL) with semi quantitative culture was performed at study entry and 3 times weekly whilst the patient remained intubated. A modified clinical pulmonary infection score was recorded daily in each patient. The study was approved by the Local Research and Ethics Committee. Delayed consent was obtained in patients who ultimately survived. RESULTS: Table 1 summarises demographic and outcome data in the 179 patients that have been randomised to date (91 chlorhexidine, 88 chlorhexidine). BAL results were not significantly different between the two groups with 37.1% of specimens from the chlorhexidine and 32.6% from the placebo groups respectively developing significant microbiological growth. The acquisition of methicillin resistant staphylococcus aureus (MRSA) was similar between the 2 groups with 14.5% and 19.5% of patients acquiring MRSA in the chlorhexidine and placebo groups respectively. There was no difference in the impact of VAP or the overall use of antibiotics between chlorhexidine and placebo treated groups. Chlorhex n=91 Placebo n=88

APACHE II score 16.2 16.4

Age years 57.6 56.3

IPPV days 8 (4-11.5) 5 (3-10.75 37.5

Mortality % 39.5

CONCLUSION: We conclude that the topical administration of 0.2% chlorhexidine to the oropharynx in ventilated patients is not effective in influencing the colonisation of the lower respiratory tract with pathogenic bacteria or in reducing the impact of VAP in the patient group studied. REFERENCE(S): 1) DeRiso AJ et al. Chlorhexidine gluconate 0.12% oral rinse reduces the incidence of total nosocomial respiratory infection and non prophylactic antibiotic use in patients undergoing heart surgery. Chest 1996. 109. 1556-1561. Grant acknowledgement: Study supported by Hospital Infection Society

Oral Presentations Improving quality – 031-035 031 EFFECT OF NURSE WORKLOAD ON THE OUTCOMES OF ICU PATIENTSPRELIMINARY RESULTS Dyk D1, Cudak - Bañska E2 1Department of Anethesiology and Intensive Care Nursing, Faculty of Health Sciences, University of Medical Sciences, Pozna´n, 2Department of Anethesiology and Intensive Care Nursing, Faculty of Health Sciences, University of Medical Sciences, Poznañ, Poland INTRODUCTION: Intensive Care Unit (ICU) nurses provide highly concentrated patient care and form a vital part of the ICU team. Good nursing staff at sufficient levels for the number of patients makes a real difference to the quality of ICU care (1). Studies started in 1996 by Jack Needelmen on the patients’ mortality and the incidence of traumas concluded that an insufficient number of nurses was the causative factor in 24% of cases. The aim of this paper is the presentation of the preliminary results referring to the evaluation of nursing care in Polish ICUs and its effect on the patients’ outcomes. METHODS: The studies were carried out in the period 5.01. – 31.03.2004 in all patients admitted to the ICU of a university teaching hospital. In prospective studies, the TISS -28 scale (2) was used for the estimation of the nurses work load and APACHE II test (3) evaluating the severity condition of the patients during admission and discharge. In the 7-bed ICU, 18 qualified nurses are employed on full time basis. RESULTS: The studied group included 60 patients (28 mean and 32 women). Their median age was 58 (21-86). The median time of hospitalization was 3,5 (2-47). 45 patients were discharged, and 10 (16,6%) died. Median APACHE II score for all admissions was 20(6-33) and 7 (0-16) at discharge. There was a significant difference between APACHE II results during admission and the patients who survived and those who died (p< 0,005). The mean value of TISS scoring for all patients was 32±7,7. There was a difference in the mean TISS scores of the discharged patients 31,8±-7,8 vs. the mean TISS scores 36,7±4,8 of patients who died (p<0,0001). The mean of nursing staff in the 7-bed unit per 24 hours was 3,6±0,4 with a mean number of patients 5,2±1,4. The TISS mean value for one nurse was 46,2±-11,6. Different TISS scores per one nurse were recorded in the care for the discharged patients: median: 40 (23,7-85,6) vs. patients, who died 52,6 (30,6-70,5), (p<0,0001). CONCLUSION: The clinical condition of patients exerts a significant effect on the therapeutic outcomes. The workload of nurses is significantly different depending on the outcomes of patients treated in ICU. REFERENCE(S): 1. Depasse B., Pauwels D., Komers Y., Vincent J.L.: A profile of European ICU nursing. Intensive Care Med. (1998) 24:939-945. 2. Moreno R., Morais P.: Validation of the simplified therapeutic scoring system on an independent database. Intensive Care Med. (1997) 23: 640-644. 3. Knaus WA. et al.: APACHE II: a severity of disease classification system. Crit. Care Med. (1985) 13: 818-829.

030

032

ACQUIRED COLONIZATION BY RESISTANT MICRO-ORGANISMS IN AN ICU THAT USES SDD

A PROSPECTIVE AUDIT ON THE USE OF NON-INVASIVE VENTILATION (NIV)

Alía I1, De la Cal M A1, Cerdá E1, García-Hierro P2, Abella A1, Colomer I1, Aranguren A3 1 Intensive care unit, 2Department of microbiology, 3Department of pharmacy, Hospital de Getafe, Getafe, Spain

1

INTRODUCTION: The use of selective decontamination of the digestive tract (SDD) remains controversial, largely because of concerns that, in the long term, it may promote the emergence of antibiotic-resistant strains. We have analyzed the frequency of acquired colonization (surveillance samples or diagnostic samples) by resistant micro-organisms (RM) in an ICU that uses SDD in patients with expected duration of mechanical ventilation longer than 48 hours. METHODS: Prospective study of patients consecutively admitted to the ICU from March 1, 2002 through February 29, 2004 that received SDD from the start of mechanical ventilation and up to 48 hours after extubation. SDD regimen consisted of oral paste and enteral solution containing polymyxin E, tobramycin, amphotericin B and vancomycin that were applied to the oropharynx and by nasogastric tube respectively. For the three first days, systemic cefotaxime was provided. Microbiological surveillance specimens which included swabs of the oropharynx, nose and rectum were obtained after tracheal intubation and weekly thereafter for semiquantitative culture. Additional samples for culture were taken when clinically indicated. RESULTS: 431 patients were sumitted to SDD: male 66%, age (median and intercuartile range) 68 (56-77), SAPS II 42 (32-52), admission type (medical 70%, urgent surgery 19%, elective surgery 6%, trauma 4%), length of ICU stay 11 (7-20) days, ICU mortality 26%. 113 patients (26%, 16´7/1000 ICU days) had acquired colonization with one or more resistant strains of the following micro-organisms: MRSA (8, 1´8%, 1´2/1000 days), Pseudomonas sp (68, 15´8%, 10´1/1000 days), Acinetobacter sp (3, 0´7%, 0´44/1000 days), Burkholderia cepacea (12, 2´8%, 1´8/1000 days), Stenotrophomonas maltophilia (12, 2´8%, 1´8/1000 days), Alcaligenes xylosoxidans (3, 0´7%, 0´44/1000 days), Sphingomonas paucimobilis (9, 2%, 1´33/1000 days), other gram-negative aerobic bacilli (39, 9%, 5´8/1000 days). No vancomycin-resistant enterococcus was isolated during the study period. 16 patients developed 20 episodes of secondary endogenous infection caused by any of the aforementioned RM (P. aeruginosa 13, B. cepacea 3. A. xylosoxidans 1, Enterobacteriaceae 3). CONCLUSION: The frequency of acquisition of RM is similar to that reported in ICUs with use of SDD during prolonged periods. It exist very low level of MRSA acquisition with the use of enteral vancomycin without appearance of VRE nor GISA, low level of acquisition of resistant Enterobacteriaceae, without clinical impact, and intermediate level of acquisition of resistant Pseudomonaceae, that are responsible for the infections caused by the flora acquired in our ICU.

Clayton J1, Downs D1, Dyson M1, McCreanor J1 Acute Intensive Care Unit, Wythenshawe Hospital, Manchester, United Kingdom

INTRODUCTION: Non-invasive ventilation (NIV) has been used increasingly in the intensive care setting but its role has not yet clearly defined METHODS: The aims of the audit were to review which patients received NIV and whether they derived any benefit from it. In addition, we looked at the grade and/or discipline of those initiating NIV, any complications and outcome. A proforma was produced and a prospective study of the first 30 patients receiving NIV from November 2002 was undertaken. Data was recorded by the doctor/nurse caring for the patient. The ICU database provided information on length of stay and APACHE II score. There were no exclusion criteria. RESULTS: The average age of the patients was 62 years (range 17 – 89). Of the 30 patients, 21 (70%) were surgical patients, the remainder medical. Twenty patients (66.6%) had a primary pulmonary injury and ten (33.3%) had a secondary pulmonary injury (sepsis, ARDS, cardiogenic pulmonary oedema). Sixteen patients (53.3%) were successfully weaned from NIV. The remaining 14 (46.6%) patients were intubated and mechanically ventilated or died on NIV. Fifteen (75%) of patients with primary respiratory failure were successfully weaned from NIV compared to one (10%) from the secondary respiratory failure group. The APACHE II score was 15.8 in the successfully weaned group and 18.5 in the failure to wean group. Length of stay in ICU was lower in the successfully weaned group, (4.75 days) as opposed to 12.57 days. All patients successfully weaned from NIV were discharged home. NIV was instigated by both medical (54% of occasions) and nursing staff (46%). There were minimal complications: one documented pressure sore; two patients were intolerant of the mask (both were intubated and invasively ventilated) CONCLUSION: NIV is clearly indicated for primary respiratory failure (75% success rate of which 100% discharged) and a reduced length of stay in ICU, but only limited success in secondary respiratory failure.

17th Annual Congress – Berlin, Germany – 10–13 October 2004

S13

033

035

INTRODUCTION OF A COLLABORATIVE REHABILITATION PATHWAY FOR DIFFICULT TO WEAN PATIENTS

THE NEEDS OF RELATIVES OF DYING PATIENTS HOSPITALISED ON AN INTENSIVE CARE UNIT

Field D1, Tilly H1 Critical Care Directorate, Ashford and St Peters NHS Trust, Surrey, United Kingdom

DELVA D G1, Bruggeman J2, Hermans G3, Wilmer A3 1 Department of Social Work, 2Department of Nursing, 3Department of Medical Intensive care, University Hospitals Leuven, Leuven, Belgium

1

INTRODUCTION: Weaning patients from mechanical ventilation using standardised criteria for readiness to wean and weaning protocols have been shown to be safe and effective in reducing mechanical ventilation (MV) time, length of stay (LOS) and overall reduction in critical care costs.1 However, for a small but significant group of patients failure to wean and dependence on mechanical ventilation continues despite following established protocols. This leads to increased mortality, morbidity, LOS and costs. The purpose of this case review is to demonstrate that difficult to wean patients who continue to undergo futile weaning attempts despite meeting all weaning criteria and following a systematic weaning protocol can be liberated from mechanical ventilation by introducing a proactive collaborative rehabilitation pathway where weaning represents only a small focus in their overall care and management. METHODS: A case review was undertaken on a patient (KAM) with adult respiratory distress syndrome (ARDS) and kyphoscoliosis admitted to the intensive care unit (ICU) of a District General Hospital within the UK. KAM was receiving mechanical ventilation through a tracheostomy via a Servo i (Siemans) ventilator. Two different approaches were used to wean KAM from mechanical ventilation. Initially, the ICU’s well-established evidence-based readiness to wean criteria and protocol of daily reduction of the level of pressure support was used. Following continued weaning failures a collaborative rehabilitation pathway was introduced. The overall goals of this approach were to stop all weaning attempts and maintain adequate ventilatory support in order to optimise KAM’s physical strength to increase respiratory function and endurance through specific pulmonary rehabilitation. This involved weekly multidisciplinary team (MDT) case reviews, long term and short-term patient and MDT goals, communication strategies, exercise programme, psychological profile, weaning plan and proactive discharge planning. RESULTS: KAM underwent 4 unsuccessful weaning episodes after passing the readiness to wean criteria. The mean duration of each weaning episode is 11.75 days. The initial weaning approach used a total of 47 bed days and resulted in failure to wean KAM. The overriding reasons for failure to wean were gross muscle weakness causing ventilatory pump failure, which in turn was further complicated by KAM’s kyphoscoliosis. Following the introduction of the collaborative rehabilitation pathway KAM was liberated from MV and decannulated in 21 days and discharged home in 14 days. CONCLUSION: The case study demonstrates the efficacy of the method in overcoming the limitations of a protocol led approach to treating the long term difficult to wean patient.

INTRODUCTION: This study explores the needs of the relatives of dying patients hospitalised on the intensive care unit (MIVE A), related to their age, gender, education level and relationship to the patient. METHODS: 101 relatives of 45 different patients have participated. The needs of the relatives were measured using the averages of the Critical Care Family Needs Inventory (CCFNI)1, a self reporting questionnaire of 45 items, scoring on a 4 point likert scale (differing between ‘not important’ to ‘very important’). Five subscales can be distinguished: the need for information (14 items), the need for comfort (9 items), the need for support (9 items), the need for reassurance (7 items), the need for accessibility (6 items). Along with the CCFNI the relatives were also asked to fill in a questionnaire of 13 items, proposed by the multidisciplinary team of the MIVE A. For the statistical processing of the data, averages, correlation coefficients and variance analyses were used. RESULTS: The need for information was the highest, followed by the need for reassurance, the need for accessibility, the need for support and finally the need for comfort. In the rating of the seperate items the highest score is for ‘getting an honest answer to your questions’, followed by ‘being reassured that the patient is not suffering’, ‘being reassured that he is getting the best medical care’, ‘ knowing the prognosis (the prospects about the condition of the patient)’ and ‘being called at home about changes in the patient’s condition’. The five items with the lowest score are ‘being encouraged to cry’, ‘having comfortable furniture in the waiting room’, ‘the possibility to have good meals in the hospital’, ‘being able to help nursing the patient’ and ‘ being offerd a drink at the bed of the patient’. The younger the patient the higher the need for support by the relatives. The older the patient the higher the need for reassurance. Female relatives experience a greater need for comfort, as well as relatives of a patient who has already hospitalised on an ICU before. CONCLUSION: The study gave the teammembers a much better insight in the needs of relatives of patients hospitalised on an ICU. This evidence allowed to develop more effective care interventions for the different team members. Implications for clinical practice will be discussed. REFERENCE(S): 1. Molter, N., & Leske, J.S., (1983) Critical Care Family Needs Inventory.

REFERENCE(S): 1. Burns SM et al (2003) Implementation of an institutional program to improve clinical and financial outcomes of mechanically ventilated patients. Crit Care Med Vol. 31, No 12 2752-2763

034 IMPROVING THE CARE OF THE DECEASED PATIENT AND BEREAVED FAMILIES IN THE INTENSIVE CARE UNIT Ben- Nun M1, Polishuk Y1, Konichezky S1 1 General Intensive Care Unit, Kaplan Medical Center, Rehovot, Israel INTRODUCTION: The death of the patient in Intensive Care Units can be seen as a failure. Dealing with the loss of a patient and caring for the bereaved family presents a challenge for the nursing staff. METHODS: To improve the nursing staff’s ability to manage the situation surrounding the patient’s death a care plan was developed. The structured approach included a written protocol detailing what should be done with the deceased patient’s body. Emphasis was placed on working together as a team to clear away the redundant tubes and machinery and allowing the family a dignified parting from the patient. Guidelines were written to use when informing families of the patient’s death and a checklist printed to ensure the bureaucratic requirements are met. The new care plan was adopted as Unit policy and was introduced via staff meetings,freely available written material and individual tutoring. Nurses were encouraged to discuss cases in order to defuse painful feelings and share insight with others. A questionnaire was taken by nurses before implementation of the care plan and repeated one year later. The bereaved families were not approached before the improvement but a year after telephone contact was made. The families were asked to score their satisfaction levels regarding the support they had received from the staff on the occasion of their loved one’s death. RESULTS: According to the first questionnaire 86% of nurses felt the need for greater knowledge to help cope with the situation. 83% felt that the lack of a structured approach denied the families a dignified parting. When questioned again a year after the improvement a reversal of attitudes emerged. Now 96% felt that information was readily available,86% knew how to help the families with the bureaucracy and 93% felt that families’ needs are being met. The families also expressed high satisfaction levels with 84% awarding maximum scores for the emotional support and guidance on bureaucracy they had received. There were no intermediate scores and the remaining 16% expressed no satisfaction at all. They complained that they received no care at all as these patients had died in the night, with the families being informed the following morning. CONCLUSION: The application of a care plan, which gives clear guidance to nursing staff, has improved the confidence of staff when dealing with the death of our patients. As a result 84% of the bereaved families felt their needs had been met in a dignified and humane manner. We feel that the care of the deceased patient should be managed by using well structured protocols, as Intensive Care staff apply to all the complex and challenging interventions used to save patients lives. The same skill and commitment will be used to care for all our patients and their families until the very end whatever the outcome.

Oral Presentations Ventilatory support – 036-038 036 PATIENT-VENTILATOR INTERACTIONS DURING PSV AT DIFFERENT LEVELS OF SEDATION IN ALI PATIENTS Grasso S1, Fanelli V2, Cafarelli A1, Dalfino L1, Ingenito G1, Ancona G1, Fiore T2 ICU, Di Venere Hospital, 2ICU, University of Bari, Bari, Italy

1

INTRODUCTION: Pressure support ventilation (PSV) has been proposed in patients with acute lung injury (ALI) in order to preserve diaphragmatic activity (1). Sedation and analgesia are required for these patients to achieve an acceptable comfort in the ICU context. We evaluated patient ventilator interactions in 8 patients with ALI submitted to PSV and sedated. METHODS: Patients were sedated through a continuous infusion of midazolam (0.07 - 0.2 mg/Kg/h) and fentanyl (0.7-1.5 mug/Kg/h). Sedation was quantified according to the Richmond agitation-sedation scale (RASS) (2). Three levels of sedation were randomly studied: RASS -1 (drowsiness), RASS -2 (light sedation) and RASS -3 (moderate sedation). For each sedation level we studied breathing pattern, inspiratory work of breathing [pressure time product of Pes per breath (PTP/b) and per minute (PTP/min)] and the coefficient of variability (SD*100/mean) of breathing pattern in a 5 min period. RESULTS: Increasing the level of sedation reduced inspiratory muscle effort and increased VT and Ti. The variability of breathing pattern was greatly reduced by sedation. Data are mean ± SD, ANOVA and paired t test*) p < 0.05 RASS –1 vs RASS –2; #) p < 0.05 RASS – 2 vs RASS – 3; †) p < 0.05 RASS – 1 vs RASS – 3 VT (ml) RR (b/min) Tinsp (sec) PTP/b (cmH2O*s) PTP/min (cmH2O*s/min) VT variability (%) RR variability (%) Ti variability (%)

RASS -1 458 ± 60 17.2 ± 4 0.84 ± 0.2 5.6 ± 1.3 102 ± 40 17.3 ± 8.4 31.8 ± 8.9 25.4 ± 9.7

RASS -2 593 ± 72 * 14.1 ± 3 * 1.29 ± 0.4 * 2.2 ± 1.2 * 30 ± 18 * 11.1 ± 5.8* 17.2 ± 5.5 * 15.2 ± 6.4 *

RASS -3 679 ± 58 #† 13.6 ± 2 † 1.35 ± 0.4 † 1.18 ± 0.8 #† 16 ± 12 #† 5.3 ± 3#† 6.2 ± 1.8 #† 9.1 ± 0.4 #†

CONCLUSION: Our data suggest that levels of sedation commonly used in clinical practice could lead to muscle unloading and monotonous breathing pattern in ALI patients ventilated with PSV and that the advantages aspected from diaphragmatic contraction may be lost with sedation. REFERENCE(S): 1) Cereda M, Foti G et al Pressure support ventilation in patients with acute lung injury. 2000, Crit Care Med 28:1269-1275 2) Ely WE,Truamn B et al. Monitoring sedation status over time in ICU patients. JAMA. 2003 22:2983-2991

S14

17th Annual Congress – Berlin, Germany – 10–13 October 2004

037 RAPID OFFSET OF EFFECTS WITH REMIFENTANIL AFTER 10 DAYS OF CONTINUOUS INFUSION Malbrain M L N G1, Karabinis A2, Morais R3, Albrecht S4, Jarnvig I5, Parkinson P A6 Intensive Care Unit, Ziekenhuisnetwerk Antwerpen, site Stuivenberg, Antwerpen, Belgium, 2Intensive Care Unit, General Hospital, Athens, Greece, 3Intensive Care Unit, Dubai Hospital, Dubai, United Arab Emirates, 4Intensive Care Unit, Universität, Erlangen-Nürnberg, Germany, 5 Intensive Care Unit, Rigshospitalet, Blegdamsvej, Denmark, 6, GlaxoSmithKline, Greenford, United Kingdom 1

INTRODUCTION: Remifentanil hydrochloride (R) is a short acting \mu-opioid agonist. R has a rapid onset of action and it can be administered for long periods of time without accumulation (14) It has a context-sensitive half-time of 3-5 min which is independent of the duration of infusion (3). This is a result of metabolism by non-specific blood and tissue esterases resulting in a very short terminal half-life of < 10 minutes. This abstract reports the offset of effects of R after a maximum of 10 days of continuous infusion, compared to a hypnotic based sedation.

Oral Presentations Pathophysiology of organ dysfunction in sepsis – 039-041 039 INTERLEUKIN-6 INHIBITS RED BLOOD CELLS PROTHROMBOTIC ACTIVITY García-Allut J L1, Nuvials X2, Ruiz-Rodriguez J C3, Santos T4, Vallés J4, Caballero J3, Monasterio J5, Bóveda J L3 Care Unit, Hosp. Clínico Universitario, Santiago de Compostela, 2ICU, Hosp. Gral.Vall d’Hebron, 3ICU, Hosp. Gral. Vall d’Hebron, Barcelona, 4Investigation Centre, Hosp. Gral. La Fe, Valencia, 5Haemostasis Research Lab, Hosp. Gral. Vall d’Hebron, Barcelona, Spain 1Intensive

Grant acknowledgement: This study (USA30226) was supported by a grant from GlaxoSmithKline.

INTRODUCTION: Thrombosis is a multicellular process, where different cells modulate platelet function. It has been described a prothrombotic role for red blood cells (RBC) in the thrombus genesis, as result of a metabolic dialogue between cells. Interleukin-1\beta (IL-1\beta),IL-6 and IL-10 have been described as sepsis-involved cytokines with pro- and antiinflammatory effects. Our interest is to find out how these cytokines modulate the RBC prothrombotic activity. METHODS: Platelet rich plasma (PRP) and RBC were obtained from citrated blood samples from healthy donors. An experimental system for the study of platelet reactivity and the interactions between platelets and RBC has been reported1. This two phases experimental system studies platelet recruitment with optical aggregometry, and platelet activation measuring released serotonin(5-HT)by RIA. In both cases, the inducer of the platelets is a released obtained from a generating system formed by PRP or PRP+RBC after stimulation with collagen. Also we applied this experimental system to washed RBC after incubation for 30 minutes with increasing doses (10 and 1000 ng/ml) of IL-1\beta, IL-6 and IL-10. Results are expressed as mean±SD from 5 different experiments. Units are % of maximum aggregation (recruitment), and % of total 5-HT (activation). Data were analized with the Student´s t. A p value <.05 was considered significant. RESULTS: RBC increased platelet reactivity, as shown in table 1, measured by significant increases in platelet recruitment and activation. When RBC are preincubated with IL-1\beta and IL-10, we didn’t find any significant change in the RBC proaggregatoty effect -data not shown-. But when RBC were preincubated with IL-6, RBC lacked their proaggregatory activity. Effect of IL6 on RBC prothombotic activity : PRP PRP + RBC PRP + RBC-IL6 PRP + RBC-IL6 [10ng/ml] [1000ng/ml] Recruitment (I max) % 8±2.5 20.4±2.3* 10±3.5 10.5±4.6 Activation (5-HT) % 50.4±9.2 77.6±13* 50±16.7 51±13.5 * p<.05 CONCLUSION: Our data support a role for RBC promoting platelet reactivity. While IL-1\beta and IL-10 did not affect RBC induced aggregation, IL-6 inhibited RBC prothrombotic activity, as measured by a decrease in platelet recruitment and activation. We describe a new way how released mediators from activated leukocytes modulate thrombus formation. REFERENCE(S): 1.Santos T, Vallés J, Marcus AJ et al. Enhancement of platelet reactivity and modulation of eicosanoid production by intact erythrocytes. J Clin Invest 1991;87:571-580.

038

040

PREVENTION OF POST-EXTUBATION LARYNGEAL EDEMA BY METHYLPREDNISOLONE: A MULTICENTER RANDOMIZED TRIAL

ABNORMALITIES OF PERIPHERAL BLOOD DENDRITICS CELLS HOMEOSTASIS IN SEPTIC SHOCK PATIENTS

François B1, Gissot V2, Desachy A3, Boulain T4, Brenet O5, Vignon P1, ARCO group 6 1 Medical-surgical ICU, Dupuytren Teaching Hospital, Limoges, 2Medical-surgical ICU, Teaching Hospital, Tours, 3Medical-surgical ICU, General Hospital, Angoulême, 4Medical-surgical ICU, Teaching Hospital, Orléans, 5Medical-surgical ICU, General Hospital, Cholet, 6None, None, None, France

Guisset O1, Blanco P2, Dilhuydy M2, Camou F1, Lefevre J1, Gabinski C1, Moreau J2 1 Medical intensive care unit, Saint Andre University Hospital, 2Immunology, Pellegrin Hospital, Bordeaux, France

METHODS: 105 critically ill patients randomised 1:1 to open treatment with either Remifentanilbased sedation (RBS) or Hypnotic-based sedation (HBS) (Midazolam (MID) plus fentanyl or morphine at investigator choice). R infusion was started at 6-9mcg/kg/h and titrated to effect to provide optimal analgesia and sedation. Supplemental MID bolus was introduced at a R rate of 1218mcg/kg/h. HBS was administered according to routine clinical practice. For patients extubated within 10 days the time from start of treatment to start of weaning process and extubation were recorded. For patients remaining on study drug after 10 days continuous infusion, RBS or HBS were discontinued and time to first signs of recovery, defined as requirement for additional analgesia or sedation, was recorded prior to introduction of alternative agents. RESULTS: 29 (51%) RBS and 16 (33%) HBS patients started weaning during the 10 day treatment period and were extubated. These results have been reported previously (3). 16(28%) RBS and 16(33%) HBS received study drug infusion for 10 days. After discontinuation of study drug 15 (93%) RBS and 14 (87%) HBS patients required alternative analgesia and/or sedation. Median times to first sign of recovery were 0.25h(RBS) and 1.167h (HBS). P = <0.001. Median weighted mean infusion rates (\mu/kg/h) for all patients over the 10-day period were 14.3 (R), 1.43 (F) and 360 (M) CONCLUSION: The rapid offset of effects after 10 days of infusion confirms the lack of accumulation of R after prolonged infusion. Remifentanil has proven to be a very useful agent for use in critically ill patients requiring analgesia and sedation for long periods of time. REFERENCE(S): (1) Breen et al. Critical Care 2004 8: R21-30. (2) Kapila et al. Anesthesiology. 1995;83:968-975. (3) Westmoreland et al. Anesthesiology. 1993; 79:893-903. (4) Malbrain et al. Critical care 2004; 8: P238.

INTRODUCTION: The incidence of post-extubation laryngeal oedema is estimated to range between 2 and 16 %. It may result in reintubation and prolonged length of ICU stay. The efficacy of corticosteroids to prevent this complication remains controversial. We hypothesized that the lack of preventive effect previously reported could be related to a short time lag between the steroid administration and the extubation. Accordingly, this study was aimed to evaluate the efficacy of a 12-hour treatment with methyl-prednisolone to prevent post-extubation laryngeal edema in ICU patients. METHODS: This was a prospective, randomized, double-blind, placebo-controlled study. Over a 10-month period, all patients ventilated for more than 36 hours with a planned extubation were eligible. Patients who received steroids for other purposes were excluded. Patients were randomly allocated to receive 20 mg of methyl-prednisolone or a placebo every 4 hours before the planned extubation for a total treatment duration of 12 hours. Clinical follow-up recorded the occurrence of a laryngeal oedema up to 24 hours after extubation. Airway obstruction was qualified as minor or major and the need and reason for re-intubation were recorded. RESULTS: Among the 757 patients enrolled in the study, 60 planned extubation were not performed (8 %) and 28 re-intubation during the first 24 hours were not attributed to a laryngeal edema (4 %). Finally, 669 patients were suitable for analysis. Steroid administration was effective to prevent post-extubation laryngeal edema (Table). Using a multivariate analysis, a short duration of mechanical ventilation (p=0.008), the female gender (p<0.0001) and a large size of endotracheal tube (p=0.04) were independent risk factors for the occurrence of post-extubation laryngeal edema. No adverse event related to steroid use was noted. Study results Mean age M/F ratio SAPS II Laryngeal edema Re-intubation

Steroid (n=343) 60 ± 18 1.6 41 ± 17 10 1

Placebo (n=326) 60 ± 19 1.9 40 ± 15 73 14

P value 0.90 0.26 0.31 <0.0001 <0.001

CONCLUSION: In this study, the administration of methyl-prednisolone initiated 12 hours before a planned extubation dramatically decreased the incidence of post-extubation laryngeal edema (including when resulting in re-intubation) in ICU patients ventilated for more than 36 hours. Grant acknowledgement: Grant from the Société de Réanimation de Langue Française

INTRODUCTION: Patients with septic shock are usually immunosuppressed as evidenced by their frequent inability to eradicate their primary infection, and their propensy to acquire secondary infections. However the underlying mechanisms explaining such observations are not fully established. In this study, we hypothetized that abormalities of the dendritic cell (DC) system (1), a key cell of the innate immune system that have an important capacity to interact with T and B cells and modulate their reponses to invading pathogens, may be involved (2). METHODS: Twenty patients with a recent septic shock (onset of sepsis 24h before inclusion) and 30 healthy individuals were included. At day 0 (time of inclusion), 1, 2, 3, 5 and 7, we evaluated the percentage as well the absolute count of peripheral blood DCs (myeloid DC, and lymphoid DC , B and T-lymphocytes, NK cells, activated CD4+ and CD8+ T-lymphocytes by flow cytometry. We also measured by ELISA the serum level of IL12, IL6, IL8, TNF-alpha, IL1, and IL10 at those time points. RESULTS: Septic shock patients had a significant decrease of DCs counts interesting both myeloid and lymphoid subsets at day 0 when compared to healthy individuals (mean values : 4590/ml vs 13120/ml and 2783/ml vs 8542/ml respectively, p<0.01). In septic shock patients, survivors had a significant higher level of myeloid and lymphoid DCs (median values : 8160/ml vs 623/ml and 5092/ml vs 595/ml respectively, p<10-6) when compared to nonsurvivors counts. Among all the other immunological parameters available at day 0, DC counts was the best to predict survival. Interestingly, DC counts were comparable at day 1 in both groups. However in survivors, DC counts reached normal levels during disease resolution comparable to those found at day 0 whereas it remained dramatically low in non survivors. Survivors were characterized by high levels of IL12 (a cytokine secreted by DCs and important to induce a Th1 response) and low levels of IL10 whereas it was exactly the opposite in non survivors (142 pg/ml and 11pg/ml vs 22 pg/ml and 102 pg/ml, respectively, p<0.01). No statistical difference was found between the two groups regarding other inflammatory cytokines. CONCLUSION: Patients with septic shock are characterized by a profound alteration of DC homeostasis that may predict their evolution. In addition to better understand mechanisms implicated in sepsis immunoparalysis, those results permit to better delineate septic shock in an immunological point of view, something that may be important to propose new therapeutical approaches. REFERENCE(S): (1) Pulendran B et al. Sensing pathogens and tuning immune responses. Science. 2001,13;293:253-6. (2) Wysocka M et al. IL-12 Suppression During Experimental Endotoxin Tolerance : Dendritic Cell Loss and Macrophage Hyporesponsiveness. J Immunol. 2001,166:7504-13.

17th Annual Congress – Berlin, Germany – 10–13 October 2004

S15

041

043

INTRAABDOMINAL HYPERTENSION IN PATIENTS WITH SEVERE ACUTE PANCREATITIS.

EFFECTS OF LEVOSIMENDAN ON THE MICROVASCULAR GASTRIC MUCOSAL HEMOGLOBIN OXYGENATION IN DOGS

De Waele J J1, Hoste E1, Blot S1, Nollet J1, Colpaert K1, Roosens C1, Decruyenaere J1, Colardyn F1 1 Intensive Care Unit, Ghent University Hospital, Ghent, Belgium

1

Schwarte L A1 Section Experimental Anesthesiology, University Hospital, Clinic of Anesthesiology, Duesseldorf, Germany

INTRODUCTION: Patients with severe acute pancreatitis (SAP) are at risk of developing intraabdominal hypertension (IAHT) and abdominal compartment syndrome (ACS). The aim of this analysis was to study the incidence of organ failure in patients with IAHT in patients with SAP, patient factors associated with the development of IAHT, and the effect on outcome. METHODS: We studied all patients admitted because of SAP to the ICU of the Ghent University Hospital between January 2000 and March 2004. Patient characteristics recorded at admission included gender, age, cause of pancreatitis, APACHE II and Ranson score. Intraabdominal pressure (IAP) measurements were obtained when multiple intraabdominal fluid collections were present on CT scan on admission, or when there was the clinical suspicion of IAHT. The incidence of IAHT (defined as IAP above 15mmHg) was recorded, as was the maximal IAP during ICU stay. Details of surgical interventions to alleviate ACS were recorded. The occurrence of organ dysfunction during ICU stay was recorded, and was defined as follows: cardiovascular: hypotension requiring vasoactive medication, renal: serum creatinine above 2.0 mg/dL, pulmonary: the need for mechanical ventilation or PaO2/FIO2 ratio < 300. Mortality was defined as in hospital mortality. Patients who were referred from other hospitals later than 7 days after the start of SAP were excluded. RESULTS: Forty one patients were included the analysis, and in 24 of them IAP measurements were obtained. In the remaining patients, IAP was not measured because of clinical improvement early after ICU admission. IAHT was found in 18 patients (44%). The incidence of organ dysfunction was high in patients with IAHT: respiratory failure 94%, cardiovascular failure 94%, and renal failure 89%. APACHE II score and Ranson score on admission were higher in patients who developed IAHT (APACHE II score 23 ± 8.9 vs. 15 ± 8.5, p=0.01 and Ranson score 6.7 ± 2.1 vs 3.5 ± 1.9, p<0.001). Mean maximum IAP in these patients was 27 ± 8.3mm Hg. Mortality in the patients with IAHT was significantly higher than in patients without IAHT (44% vs. 8%, p= 0.012). Abdominal decompression was performed surgically in four patients with IAP ranging from 25 to 45 mmHg and severe organ dysfunction. Three of these patients died early in the postoperative course, one because of further deterioration of organ dysfunction, and two of uncontrollable bleeding from the retroperitoneum.

INTRODUCTION: The effect of levosimendan, a novel inotropic vasodilator (inodilator), on the microvascular gastric mucosal hemoglobin oxygenation (µHbO2) is unknown. A possible effect could thereby be selective for the splanchnic region, or primarily reflect changes in systemic O2transport (DO2) and/or O2-consumption (VO2). Since levosimendan may modify regional or systemic responses to fluid load, a defined volume challenge was additionally performed. We compared systemic and regional effects of levosimendan with those of established inotropes, milrinone and dobutamine. METHODS: Chronically instrumented dogs with flow probes for cardiac output measurement. Anesthetized, mechanically ventilated dogs (each group n=6) randomly received levosimendan (10 µg•kg-1, followed by 4 infusion steps: 0.125-1.0 µg•kg-1•min-1), milrinone (5.0 µg•kg-1, followed by 1.25-10 µg•kg-1•min-1), or dobutamine (2.5-10.0 µg•kg-1•min-1) on different days. Subsequent fluid challenges were performed with hydroxy-ethyl-starch 6% (10 mL•kg-1). RESULTS: We measured µHbO2 (reflectance spectrophotometry), DO2, VO2, and systemic hemodynamics. Levosimendan significantly increased µHbO2 from baseline (~55% for all groups) to 64±4%, and further to 69±2% with volume challenge (mean±SEM). At the systemic level, levosimendan alone only slightly increased DO2 at a stable VO2. Milrinone elicited similar systemic effects (DO2, VO2, hemodynamics), but failed to increase µHbO2. Dobutamine, conversely, increased µHbO2 to a similar extent as levosimendan, however, this was accompanied by marked increases in DO2 and VO2. The gastric mucosa-selectivity of these interventions, expressed as slope of the µHbO2/DO2 relation, was highest for levosimendan (+1.89 and +1.14, without and with volume challenge), compared to milrinone (+0.45 and +0.47) and dobutamine (+0.48 and +0.33). CONCLUSION: Levosimendan is superior to milrinone (no significant regional effects) and dobutamine (marked systemic effects) in increasing gastric mucosal oxygenation selectively, i.e., at only moderately increased DO2 and stable VO2. If our experimental data apply to the clinical setting, levosimendan may serve as an option to selectively increase gastrointestinal mucosa oxygenation in patients at risk to develop splanchnic ischemia.

CONCLUSION: IAHT is a frequent finding in patients admitted to the ICU because of SAP, and is associated with organ dysfunction in the majority of the patients. Mortality is high in patients with IAHT, but it is not clear if routine surgical decompression in these cases of ACS is advantageous. Grant acknowledgement: This study was supported by a Clinical Doctoral Grant of the Fund for Scientific Research - Flanders (Belgium) (F.W.O.-Vlaanderen).

Oral Presentations Levosimendan – 042-044 042

044

EFFECTS OF LEVOSIMENDAN IN ENDOTOXEMIC SHEEP: A CONTROLLED STUDY

1

Dubin A1, Murias G1, Barán M1, Sottile J1, Pozo M1, Kanoore Edul V1, Canales H1, Maskin B1 1 Cátedra de Farmacología, Facultad de Ciencias Médicas, Universidad Nacional de La Plata, La Plata, Argentina

INTRODUCTION: Cardiogenic shock leads to a decrease in liver blood flow and liver function. Most of the positive inotropic substances lead to an increase of cardiac output with a distribution of blood flow preferentially to the muscle. Levosimedan is a new positiv inotropic substance, which could be shown to be more effective in longterm survival. The reason is not quite clear. Beside fewer arrhytmic actions a better effect on liver function could be responsible for this effect.

EFFECT OF LEVOSIMEDAN OF ICG PDR IN PATIENTS WITH CARDIOGENIC SHOCK

INTRODUCTION: Our hypothesis was that levosimendan, an inotropic and vasodilatory drug, might improve systemic and intestinal oxygen transport and prevent intramucosal acidosis in endotoxic shock. For this purpose, we compared three groups of sheep with endotoxic shock: control, dobutamine and levosimendan-treated animals. METHODS: In 19 anesthetized and mechanically ventilated sheep, we measured cardiac output, superior mesenteric artery blood flow, lactate, gases, hemoglobin and oxygen saturations in arterial, mixed venous and mesenteric venous blood, and ileal intramucosal PCO2 by saline tonometry. Systemic and gut oxygen transports and consumptions (DO2 and VO2), and intramucosal-arterial PCO2 difference (DeltaPCO2) were calculated. After basal measurements, Escherichia coli lipopolysaccharide (5 mug/kg followed by 2 mug/kg/hour) was injected. Then, sheep were assigned to CONTROL (n = 7), DOBUTAMINE (10 mug/kg/min, n = 6) or LEVOSIMENDAN groups (100 µg/kg followed by 100 mug/kg/hour, n = 6) during 120 minutes. Saline solution (20 ml/kg/hour) was also infused in each group. Data (mean ± SD) were analyzed with ANOVA. RESULTS: Results are shown in Table 1. * p < 0.05 vs. BASAL. § p < 0.05 vs. CONTROL. # p < 0.05 vs. CONTROL and DOBUTAMINE.

CONTROL BASAL CONTROL ENDOTOXIN DOBUTAMINE BASAL DOBUTAMINE ENDOTOXIN LEVOSIMENDAN BASAL LEVOSIMENDAN ENDOTOXIN

DO2

VO2

Gut DO2

Gut VO2

DeltaPCO2

ml/min/kg

ml/min/kg

ml/min/kg

ml/min/kg

mm Hg

24.6±5.2

8.0±1.7

105±28

29±5 9±6

1.4±0.3

15.3±3.4*

7.4±1.1

56 ±26*

26±9 23±6*

3.1±1.5*

22.3±4.7

7.5±0.9

99±30

30±5

12±5

26.8±7.0§

8.0±1.2

68±23*

Lactate mmol/l

1.6±0.3

31±12 25±11* 2.5±0.7*

25.1±3.0

7.6±0.5

111±18

27±4 7±7

1.4±0.4

24.0±6.3§

8.5±1.0

98±23#

31±7

10±8#

2.9±1.1*

CONCLUSION: Both dobutamine and levosimendan increased systemic oxygen transport, but only levosimendan preserved intestinal oxygen transport and precluded the development of intramucosal acidosis. Despite maintaining systemic oxygen transport, neither levosimendan nor dobutamine were able to correct lactic acidosis. These data suggest a role of cythopatic hupoxia in lactic acidosis.

Lenz K1, Gegenhuber A1, Firlinger F1, Kapral C1, Steinmaurer T1 Internal Medicine, Barmherzige Brueder, Linz, Austria

METHODS: Hemodynamic parameters and ICG-PDR (LimonR) were studied in 8 patients (61 +/- 10 y) admitted with cardiogenic shock due to acute on chronic heart failure to our ICU before, 1 hour and 4 hours after starting and at the end (24h) of levosimedan infusion (bolusinfusion 24 ug/kg/min over 10 min followed by 0.1 ug/kg/min continuously) RESULTS: Cardiac Index increased from 2.0+/-0.1 L/min.m-2 to 2.9+/- 0.3 L/min.m-2 after 60 min (p<0.05) and 3.25+/-0.3 L/min.m-2 after 4 h (p< 0.05) and 4.1+/- 0.4 L/min.m-2 after 24 h (p<0.05). ICG PDR (normal value 18 - 25%/min) was severely depressed before levosimedan (8.1+/-1.5%/min) and did not change after 60 min (8.4 +/-1%/min) and 4 hours (10.2+/-1.8%). After 24 hours there was a significant increase but not normalization to 11.9 +/- 2.9%/min. CONCLUSION: Levosimedan normalizes cardiac output in patients with cardiogenic shock. This is not accompanied by an acute increase of ICG PDR, which is dependent of both liver blood flow and liver function. Since liver function is dependent on liver blood flow in a severely depressed state like in the patients studied, a relevant increase in liver blood flow in the first hours seems unlikely. Nevertheless in a longterm there is an improvement in liverfunction and liverbloodflow induced by levosimedan.

S16

17th Annual Congress – Berlin, Germany – 10–13 October 2004

047

Oral Presentations Organisational issues – 045-047 045

COBATRICE SURVEY OF NATIONAL TRAINING PROGRAMMES IN INTENSIVE CARE MEDICINE

DIAGNOSTIC ACCURACY IN AN OPEN VS CLOSED INTENSIVE CARE UNIT Zwaal J W1, Baba R1 Anaesthetics, Kingston Hospital, Kingston upon Thames, United Kingdom

1

INTRODUCTION: Closed as opposed to open ICU’s are associated with lower ICU mortality. Improved diagnostic accuracy of intensivists may be responsible for this1. We compared diagnostic accuracy and patient outcome after transition from an open to a closed ICU. METHODS: Retrospective Database survey by 2 intensivists over 2 year period before and after closure of our ICU.Diagnostic accuracy was defined as the discrepancy between clinical and autopsy diagnosis.Definitions:class 1 discrepancy:contributed to death and would have altered management,class 2 discrepancy:contributed to death but would not have altered management.Both periods compared for mortality,APACHE scores,SMR’s,autopsy rate,class 1 and 2 discrepancies and association between autopsy and age,sex,APACHE-score and length of stay.Differences tested via student-test or the z-statistic for proportions. A p value<0.05 was considered significant. RESULTS: SMR’s:0.98 Vs 0.92,open Vs closed period.The decision to perform autopsy was associated with male sex in the open period and with younger age in the closed period. All patients open Vs closed mortality % mean APACHE autopsies % class1 discrepancy % class2 discrepancy %

open period n=553 35.5 16.4 7.8 13.3 46.7

closed period n=819 30.3 14.4 8.1 35.7 21.4

p <0.05 <0.05 0.76
open period n=16 66.1 57.1 22.9 6

closed period n=13 55.9 64.3 20 7.3

p p=0.12 0.69
Autopsies open Vs closed age (yrs) sex male % mean APACHE length of stay (days)

CONCLUSION: Transition from an open to a closed ICU did neither lead to an improvement in diagnostic accuracy nor in severity adjusted mortality.

Barrett H1, Bion J F1, On behalf of the CoBaTrICE Collaboration 2 Division of Medical Sciences, Anaesthetics and Intensive care, University of Birmingham, Birmingham, United Kingdom, 2CoBaTrICE project, ESICM, Brussels, Belgium

1

INTRODUCTION: CoBaTrICE (Competency Based Training in Intensive Care Medicine in Europe) is a three year project which employs consensus techniques to create an internationally acceptable minimum standard of knowledge, skills and attitudes required for a doctor to be identified as a specialist in Intensive Care Medicine (ICM). The first phase of this project involves an international survey of training programmes in ICM to determine their compatibility with competency-based methods of training and assessment. METHODS: A detailed iterative survey by questionnaire, email and direct discussion of National Co-ordinators and Reporters from seven geographical regions. These representatives were nominated with the support of their national society or organisation responsible for training in ICM. RESULTS: Responses were obtained from 29 European region countries and a further 12 countries in 6 World regions. Those with shared training structures have been grouped together for analysis which is therefore based on 37 countries/regions. Formal training programmes in ICM are available or pending in 35 (94 %) of the 37 countries/regions surveyed. Twenty-one (57%) permit multidisciplinary access to a common ICM training programme. In nine (26%) training is only available through anaesthesia. Multiple pathways for ICM training are possible in 11 (30%). Fifty-nine different training programmes have been identified, 41 in the European region; three (5%) are reported to be competency-based training programmes. The content of these programmes is standardised by national ICM curricula in 30 (86%). The minimum duration of ICM training required for recognition as a specialist varies from 3 to 38 months (mode 24 months). Work-based assessment of competence in ICM is reported to be structured and formally documented in seventeen (49%) countries/regions. An exam is mandatory in twenty-seven (77%) prior to accreditation. Specialist registration (accreditation) is possible in 32 (86%) countries / regions; in 20 this is provided as dual certification in a base speciality and in ICM. CONCLUSION: Despite considerable variation in access, duration and content of national training programmes, and in methods of assessment and accreditation, most countries favour multidisciplinary access with a common national curriculum. As competency based training is an outcome strategy rather than a didactic process, these variations in structures and processes will not impede the development of a common international competency-based training programme in ICM.

REFERENCE(S): 1.Tai DYH et al(2001).Chest 119:530-536

Grant acknowledgement: CoBaTrICE is supported by a grant from the European Union Leonardo da Vinci Programme. Additional supporter: GlaxoSmithKline

046

Oral Presentations Pseudomonas aeruginosa – 048-050 048

MORTALITY IN A CLOSED FORMAT ICU: NIGHTTIME ADMISSIONS COMPARED TO DAYTIME ADMISSIONS Enting D1, Zijlstra J G1, Tulleken J E1, Van der Werf T S1, Meijering S1, Ligtenberg J J1 1 Intensive and Respiratory Care ICB, Groningen University Medical Center, Groningen, Netherlands INTRODUCTION: Patients with serious medical conditions are more likely to die in hospital, if admitted on a weekend compared to weekdays1. It is hypothesized that hospitals function less effectively on weekends and during nighttime because of fewer personnel with less experience. However, a recent study showed no increase in mortality rate in patients admitted at night time to an intensive care unit2. We report our findings on time of admission and risk of mortality in a closed format medical ICU. METHODS: We conducted a retrospective cohort study on admissions to our medical ICU between September 2000 and August 2002. Data collected included demographic characteristics, admission source, admission diagnosis, ICU mortality and length of stay. Patients were divided in daytime admission (08.00 am to 05.00 pm) and nighttime admission (05.00 pm to 08.00 am). The ICU has a dedicated medical staff, supervising 24 hours a day, 7 days a week. Resident doctors are present 24 hours a day, with no other commitments. Rounds are made every day. RESULTS: 1239 consecutive patients were included. 45% of them were admitted in daytime. Gender and age did not differ at baseline. Significantly more direct than transfer admissions were seen after 05.00 pm. Evening- or nighttime admission was associated with a lower mortality rate: mortality rate 28,1% for patients admitted during daytime, and 22.3% for nighttime admissions, p < 0.02. No difference in ICU length of stay was detected between admission groups: 5.4 ± 6.7 for daytime admission versus 5.0 ± 6.3 for nighttime admission, p = 0.3.

MOLECULAR CHARACTERIZATION OF PSEUDOMONAS AERUGINOSA FROM A UNIT WITH HIGH ENDEMIC LEVELS Santos M1, Ferraiuoli G1, Costa-Filho R2, Gutierrez F2, Garcez M2, Pignataro L2, Moreira B M3, Pellegrino F L3 1 Infection Control, 2ICU, Hospital Pro-Cardiaco, 3Laboratorio de Epidemiologia Molecular de Infeccoes, UFRJ, Rio de Janeiro, Brazil INTRODUCTION: Pseudomonas aeruginosa(Pa) has been the most common multiresistant(MR)pathogen isolated in the intensive care units (ICU) of a private tertiary hospital in Rio de Janeiro since the year of 2000. METHODS: In the present study, 32 Pa isolates obtained from 30 patients admitted to the ICU from March 2002 to September 2003 were further genetically characterized to evaluate the possible role of cross-transmission of MR Pa . An isolate was considered to be MR if resistant to imipenem or susceptible only to this agent. Clinical and demographic data were obtained by medical chart review. Bacterial identification and antimicrobial susceptibility were evaluated by the Vitek system. Metallo\beta-lactamase (Mbla) production was assessed by the disk-approximation (DA) method. Genotyping was performed by PFGE. RESULTS: Twenty one distinct genotypes were identified. Thirteen isolates from 13 patients (43%) were clustered in 5 groups: three groups with two isolates each, one group with three isolates and one group with four isolates, demostrating cross-transmission of Pa among these patients. Among these 13 patients, 11(85%) had MR isolates, 10 of whom had taken antibiotics prior to Pa isolation. Ten(77%) patients had been at ICU during the same period. One patient presented a Mbla-positive isolate revealing an identical genotype - genotype A- to that described for other Mbla-positive isolates from other four hospitals in Rio de Janeiro.

CONCLUSION: Admission to our closed format medical ICU between 05.00 pm and 08.00 am is not associated with a higher mortality or a longer ICU length of stay. Since hospital admissions of critically ill medical patients are always unexpected, care should be organized in such a way, that quality care including adequate staffing is provided around the clock. REFERENCE(S): 1. Bell CM, Redelmeier DA. Mortality amongst patients admitted to hospitals on weekends as compared with weekdays. N Engl J Med 2001;345:663-668 2. Morales IJ, Peters SG, Afessa B. Hospital mortality rate and length of stay in patients admitted at night to the intensive care unit. Crit Care Med 2003;31:858-63

CONCLUSION: About 43% of the Pa isolates evaluated in the present study were presumably acquired at the ICU. Isolates presented high leves of resistence to antimicrobials. One Pa isolate belonging to genotype A, previously found among isolates from different hospitals in Rio de Janeiro, was detected in this study, suggesting inter-hospital transmission of this strain. This confirms the value of barrier precautions for MR Pa infected patients. REFERENCE(S): Pellegrino, F. L.et. al.2002. Occurrence of a multidrug-resistant Pseudomonas aeruginosa clone in different hospitals in Rio de Janeiro, Brazil. J. Clin. Microbiol. 40:2420-2424. Bertrand, X. et al.Endemicity, molecular diversity and colonisation routes of Pseudomonas aeruginosa in intensive care units.Intensive Care Med. 2001 Aug;27(8):1263-8

17th Annual Congress – Berlin, Germany – 10–13 October 2004

049 NOSOCOMIAL BACTEREMIA DUE TO PSEUDOMONAS AERUGINOSA IN THE ICU Pedonomos M1, Frangiskatou E1, Papakonstantinou K1, Tsirantonaki M1, Kalogeromitros A1, Pedonomou M1, Mourgos L1, Karabinis A1 1 ICU, Athens General Hospital., Athens, Greece INTRODUCTION: The aim of this clinical trial is to study the resistance of Pseudomonas aeruginosa (PA), the risk factors for mortality (M) and other characteristics of PA in ICU patients (pts). METHODS: We studied retrospectively 104 ICU pts who developed bacteremia and nosocomial infections due to PA (group A). As control cases we used 104 pts of the same 22-months period who developed bacteremia and nosocomial infection due to other microorganisms (Ac. baumannii 63, St. aureus 25, St. epidermidis 10, Kl. pneumoniae 4, S. marcescens 2). All pts were mechanically ventilated, stayed in the ICU > 96 h, had at least one positive blood culture (BC) and met clinical and laboratory criteria for infection. We studied sex, age, length of stay (LOS), use of total parenteral nutrition (TPN), time of hospitalization before bacteremia occurred (Hosp), number of positive BCs (n BCs), previous antibiotic treatment (PAT) and M rates. RESULTS: In the 3 groups: All pts, group A and group B respectively: male: 151 (72.6%), 73 (70.2%), 78 (75.0%); female: 57 (27.4%), 31(29,8%), 26 (25.0%); age (years): 44.9±21.6, 52.6±20.8, 37.2±23.6; LOS (days): 19.8±11.2, 21.0±10.4, 18.6±9.7; TPN received: 23 (11,1%), 10 (9.6%), 13 (12.5%); Hosp (days): 10.3±4.2, 9.7±4.5, 10.9±3.8; nBCs per pt: 2.8±0.9, 3.1±1.0, 2.5±0.6; PAT 206 (99.0%), 104 (100%), 102 (98.1%); Development of MODS: 38 (18.3%), 18 (17.3%), 20 (19.2%); M rates: 32 (15.4%), 17 (16.3%), 15 (14.4%). Sources of bacteremia in groups A and B were respectively: pneumonia 58 and 61, surgical infection 20 and 21, central venous catheter 15 and 13, CNS 2 and 3, unknown 9 and 6. The most active agents against PA were imipenem/cilastatin 74.0%, piperacillin/tazobactam 68.2%, ceftazidime 47.1%, aztreonam 39.4% and colistin 96.2%. CONCLUSION: 1) Group B consisted of younger pts (p<0.01) because Ac. baumannii (the predominant pathogen of the group) was isolated in younger pts. 2) Pts with PA-infection had increased number of positive BCs than pts of group B (p<0.05). 3) Pts of both groups had similar LOS, percentage of PAT and M rates. No difference was shown between pts of the groups according to the use of TPN and the length of hospitalization before the onset of the infection. 4) The sources of bacteremia were similar in both groups, as well as the frequency of MODS.

S17

Oral Presentations Acute renal failure and outcome 051-053 051 INTERIM RESULTS SHARF4 STUDY: STRATIFIED COMPARISON OF 2 TREATMENT MODALITIES IN ACUTE RENAL FAILURE Malbrain M L N G1, Elseviers M2, Van der Niepen P3, Damas P4, Hoste E5, Devriendt J6, Lins R7 1 Intensive Care Unit, Ziekenhuisnetwerk Antwerpen, site Stuivenberg, 2University of Antwerp, Antwerpen, 3Nephrology-Hypertension, Free University, Brussels, 4Intensive Care Unit, University Hospital, Liege, 5Intensive Care Unit, University Hospital, Ghent, 6Intensive Care Unit, Brugmann Hospital, Brussels, 7Nephrology-Hypertension, Ziekenhuisnetwerk Antwerpen, site Stuivenberg, Antwerpen, Belgium INTRODUCTION: It is important for ICU physicians to predict mortality of acute renal failure (ARF) in the first 24-48 hours (h). For this the SHARF II score at 0 and 48 h has been developed [1]. This study looks at the differences in short and long term morbidity and mortality between 2 different modes of renal replacemant therapy (RRT). METHODS: Prospective multicentre randomised clinical trial. Target population 1600 adult ARF patients (pts). Pts needing RRT were stratified by SHARF II0 score (<30, 30-60 and >60) before randomisation to either slow extended daily dialysis (SLEDD) or continuous venovenous hemofiltration (CVVH). RESULTS: Interim analysis after 996 pts: age 66±15.1, 62% male, SHARF II0: 64.9±32.6, SHARF II48: 64.1 ± 32.2, SOFA0 8.4±4.1, SOFA24: 6.5±5.1, APACHE II: 23.5±10.6. There were 67.9% medical and 32.1% surgical pts. Cause of ARF was prerenal in 43.2% and renal in 51.5% (ATN 90.9%, AGN 5%). About 13.8% of pts were in SHARF category1 (cat) (<30), 24.8% in cat2 (30-60) and 52.2% in cat3 (>60). The higher the SHARF score the higher APACHE II and SOFA score. RRT was needed in 59% (2/3 randomised), 39% received CRRT and 61% SLEDD. Main reason for choosing another treatment was haemodynamic instability (24%), coagulation disturbance (25%) or lack of time to perform randomisation (37%). Outcome analysis (on 481 pts) showed a mortality of 56.5%, 2.9% developed ESRD, 10.2% had partial recovery and 30.4% complete recovery of ARF. Mortality was 46% in conservative treatment versus 64% in CVVH and 68% in SLEDD. The ICU and hospital LOS was higher in pts with RRT. Observed mortality was lower than expected in nonRRT but paralleled the expected in RRT. CONCLUSION: In pts with ARF, stratified according to severity of disease with the SHARF score, remarkable differences in outcome were observed between patients with and without RRT. The SHARF score, with parameters at 0 and 48h has good predictive value in estimating prognosis in ARF-patients. Overall mortality was 56.5% and 59% needed RRT. Mortality was the same regardless of RRT type used. Major problems lay in recruting centres using both techniques equally. Randomisation rate is lower than expected. For pts with RRT, the final „intention-to-treat“ analysis will be performed after completion of data collection (april 2004). This multicentre study, the first in stratifying ARF pts before RRT randomisation, the largest so far will add to our knowledge about short and longterm outcome of different treatment modalities in ARF. REFERENCE(S): (1) Lins R. Clin Nephrol 2000;53:10-7

050

052

LUNG INJURY AND VENTILATION DURATION ARE LINKED TO THE LEVEL OF P AERUGINOSA COLONISATION

RENAL DYSFUNCTION AFTER CARDIAC SURGERY PREDICTS IN-HOSPITAL MORTALITY AND LONG-TERM SURVIVAL

Zhuo H1, Wiener-Kronish J P1, Shimaburkuro D1, Brown R1, Garcia O1, Gropper M1, Kohn M A1, Misset B1 1 Anesthesia and Postoperative Care, Moffit-Long Hospital, UCSF, San Francisco, United States

1

INTRODUCTION: The diagnosis of ventilator aquired pneumonia in ICU patients is based on clinical and bacteriological signs. The use of quantitative or qualitative assessement of the bacteriological samples remains heterogeneous and controversial among ICU teams. METHODS: We have collected daily endotracheal aspirates from all intubated patients who would consent in the Medical-Surgical ICUs at Moffitt-Long Hospital for the last year. All aspirates were cultured on selective agar to identify P.aeruginosa. All patients positive for P.aeruginosa had medical data collected for 28 days after the identification of the first positive culture and were asked to undergo a bronchoalveolar lavage (BAL) with a blind sterile catheter. Daily quantitative cultures of endotracheal aspirates were done. RESULTS: Of 56 patients with complete data to date, there were no differences between the patients with positive cultures with >10^6 CFU compared to those with cultures of <10^6 cfu in terms of age, gender, diagnosis, APACHE II scores, SAPS II scores, MODS, SOFA, LODS or mortality. However, among survivors at hospital discharge, patients who had any quantitative culture with >10^6 cfu were ventilated about 6.8 days more than the patients who had less P.aeruginosa (P=.04 by simple linear regression) and the patients with any culture >10^6cfu also were in the ICU 21.2 days more than the patients with less P.aeruginosa(p=.02 by simple linear regression). Finally, using the 29 BAL results, we found a similar association between a quantitative culture of > 10^4 cfu in the surviving patients with 9.6 more days of mechanical ventilation compared to the patients who survived with < 10^4 cfu of P.aeruginosa (P= .07) . Furthermore, the surviving patients with BAL quantitative cultures of > 10^4 cfu also tended to stay in the ICU almost 27 days longer than patients with smaller quantities of P.aeruginosa in their BAL (p=.03). CONCLUSION: Just as in experiments with P.aeruginosa in animals, larger quantities of P.aeruginosa in endotracheal aspirates or in BAL are associated with more lung injury and require longer periods of mechanical ventilation and ICU stays. Quantitative cultures may be useful not only for diagnosis, but may help document the success or lack there of antibiotic treatment, and help predict length of ventilation and ICU stay. Grant acknowledgement: NIH PH50HL74005;AI44101; HL69809

Loef B G1, Epema A H2, Smilde T D3, Navis G4, Stegeman C A4 Cardiothoracic intensive care unit, 2Anesthesiology, 3Cardiology, Hospital Groningen, Groningen, Netherlands

4Nephrology,

University

INTRODUCTION: Postoperative renal function deterioration after cardiac surgery with cardiopulmonary bypass(CPB)is associated with increased in-hospital mortality. The long-term prognosis of patients with this renal function deterioration is not determined yet. We studied not only in-hospital mortality but also long-term survival in patients with postoperative renal function deterioration. METHODS: Included were 843 patients who underwent cardiac surgery with CPB in 1991. Postoperative renal function deterioration (increase in serum creatinine in the first postoperative week of at least 25%) occurred in 145 patients. Association of in-hospital mortality with different variables was tested with logistic regression and relative risk (RR) and 95%confidence intervals (95% CI) were estimated. The vital status as of April 1999 of the patients was evaluated by data from the outpatient clinic and by contacting patients. For the long-term outcome Kaplan-Meier survival analysis was used. Cox regression analysis was used to test the independent association of long-term survival with different variables. Statistical significance was accepted at p<0.05 RESULTS: Postoperative renal function deterioration carried a RR of 12.6(95% CI 15.7-27.0)for in-hospital mortality. Significant independent predictive factors for in-hospital mortality were postoperative renal function deterioration (adjusted RR 7.8;95% CI 3.1-20.0),re-exploration, duration of operation, diabetes and age. Survival analysis with a follow-up time of 100 months revealed significant differences in survival between patients with and without postoperative renal function deterioration (p=0.005 by logrank test). Significant independent predictive factors for long-term mortality were postoperative renal function deterioration (RR1.63; 95%CI 1.15-2.32), peripheral vascular disease, Cockroft-clearance, duration of operation, and age. CONCLUSION: Postoperative renal function deterioration in cardiac surgical patients predicts not only a short-term high mortality risk, but also adversely affects long-term survival.

S18

17th Annual Congress – Berlin, Germany – 10–13 October 2004

053 PREOPERATIVE RENAL IMPAIRMENT AND MORTALITY AFTER CARDIAC SURGERY: A PROSPECTIVE COHORT STUDY Lassnigg A1, Schmidlin D2, Hiesmayr M3 Anaesthesia and Critical Care, University Hospital, 2Anaesthesia and Critical Care, Hirslanden Klinik Im Park, Zurich, Switzerland, 3Cardiothoracic and Vascular Anaesthesia, Vienna General Hospital, Vienna, Austria

1Cardiovascular

INTRODUCTION: Mild-to-moderate elevation of preoperative serum creatinine (Screa) levels significantly increases mortality after heart surgery1. The aim of our study was to evaluate the relationship between elevated Screa and 30-day or late mortality. METHODS: Patients undergoing cardiac or thoracic aortic surgery between 1997 and 2001 were included in this prospective cohort study. The primary outcome was 30-day mortality and its relation to baseline serum creatinine. First we determined the relation between Screa and mortality, when divided by steps of 0.1 mg.dL-1 and Screa were divided into 5 groups2. Screa-normal [1.1, 1.3); mg.dL-1 served as the reference group. Multivariate analysis was done with the addition of established risk factors for mortality after cardiac and thoracic aortic surgery, according to the EuroSCORE and additional risk factors as weight, stable angina, congestive heart failure, diabetes mellitus, therapy with diuretics and/or angiotensin-converting enzyme inhibitors and the need for preoperative therapy with inotrops. Cox regression was used for statistical analyses. RESULTS: The cohort consisted of 4197 patients with a EuroSCORE of 5.7 (SD 3.2). Mean follow up period was 28 month (2 days-61 month, SD 18 month). Overall, early, 30–day mortality was 224 (5.3%) and mortality after 30 days was 357 patients (9.1%). After adjusting for preoperative risk factors, the hazard ratios (HR) were for Screa-low HR 1.11 (95%CI 0.48-2.5), HR 1.69 (1.23-2.34) for Screa-high, HR 1.64 (1.05-2.57) for Screa-higher, HR 2.85 (1.86-4.38) for Screa-very high. The additional significant prognostic factors in the multivariate analysis were a higher EuroScore, EF between 30-50% and urgent surgery, older age, therapy with diuretics, the use of inotrops as well as diabetes mellitus.

CONCLUSION: Our findings show that even a minimal elevation in serum creatinine is associated with a higher mortality and may be a marker of chronic organ injury.. REFERENCE(S): 1. Weerasinghe A et al. J Thorac Cardiovasc Surg. 2001;121:1083-9. 2. Fried LF et al. J Am Coll Cardiol. 2003;16;41:1364-72.