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

Oral Presentations Mechanisms in acute lung injury: Experimental findings – 807-812 807 ELEVATION OF PEEP ACTIVATES NFKAPPAB IN ALVEOLAR CELLS AND INDUCES IL-8 RELEASE Meier T1, Rose H1, Lange A1, Uhlig U2, Ziemann M1, Schmucker P1, Uhlig S2, Stamme C1 1Anesthesiology, University of Lübeck, Lübeck, 2Division of Pulmonary Pharmacology, Research Center Borstel, Borstel, Germany INTRODUCTION: In patients with acute lung injury (ALI) mechanical ventilation with low endexpiratory pressure (PEEP) and high tidal volumes can induce cytokine release (1). To clarify the effect of PEEP elevation (10 cm H2O) on the activation of the nuclear factor (NF) KappaB in alveolar cells and the release of cytokines we performed a randomized controlled study in operative patients with normal lung function. METHODS: After approval of the local ethic committee 30 patients (ASA I-II) with normal lung function, scheduled for elective ENT surgery, were enrolled in a prospective, randomized controlled study. After induction of anaesthesia (propofol 2 mg/kg, sufentanil 0.5 mug/kg, rocuronium 0.4 mg/kg) and tracheal intubation, the patients received either volume controlled ventilation (8 ml/kg ideal body weight; anaesthesia ventilator: Cicero and Sulla 808; Dräger, Lübeck, Germany; pulmonary monitoring (Capnolog Ultima, Datex-Ohmeda, Germany)) with zero end-expiratory pressure (ZEEP) or with PEEP (10 cmH2O). Anesthesia was maintained with propofol (0.1 mg/kg/min.). Arterial blood gases were analyzed in 30 min. time intervals; venous blood samples were taken before induction of anaesthesia and after surgery. After surgery a bronchoalveolar lavage (BAL, 6 x 20 ml NaCl 0.9%, right middle lobe) was performed. Microbiological analyzes, cell differentiation and nuclear extraction were performed immediately. In the nuclear extracts NFKappaB translocation was analyzed by electro mobility shift assay (EMSA). NFKappaB activation was calculated as the relative intensity to in vitro cell stimulation by lipopolysaccharide (10 ng LPS for 4h). Concentration of cytokines (IL2, IL4, IL6, IL8, IL10, GCMSF, IFNg, TNFa, IL1b, IL5, IL7, IL12, IL13, IL17, GCSF, MCP1b, MIP1b) in plasma and BAL supernatants were simultaneously detected with a BioPlex immunoassay (BioPlex, Biorad laboratories, Hercules, CA). Statistical analyses were performed using t-tests in case of normality or otherwise by Mann-Whitney test. RESULTS: We observed a small, but significant increase in nuclear translocation of NFKappaB in alveolar cells of patients ventilated with a PEEP of 10 cm H2O (p=0.048) in comparison to the ZEEP group. In line with this, IL-8 levels were increased in BAL supernatants in the PEEP group (p=0.011). There were no further differences in other cytokines in plasma or BAL supernatants between both study groups. CONCLUSION: During volume controlled mechanical ventilation PEEP can activate NFKappaB in alveolar cells and induce release of IL-8 in healthy lungs. These results corroborate the hypothesis (2,3) that stretching of lung cells can cause inflammatory cell activation. REFERENCES: 1. Ranieri et al. JAMA 1999; 282: 54-61; 2. Pugin et al. Am J Physiol 1998 ; 275: L1040-L1050; 3. Stamme et al. Pulm Pharmacol Ther 2002 ; 15: 445-61.

S209

809 PROTECTIVE ROLE OF PEEP IN AN ISOLATED LUNG MODEL OF VILI PERFUSED AT HIGH VASCULAR FLOW López-Aguilar J1, Piacentini E1, Villagrá A1, Pascotto S1, Murias G1, Blanch L1 1 Critical Care Center, Hospital of Sabadell. Institut Universitari Parc Taulí. UAB. Sabadell, Sabadell, Spain INTRODUCTION: High vascular flow (Q) and high transmural pressure (TMP= capillary alveolar pressure) favor ventilator induced lung injury (VILI). The particular contribution of each factor on VILI development remains unknown. OBJECTIVE: To determine the interaction between Q and TMP in the development of VILI. To evaluate the protective role of PEEP in VILI development by reducing TMP and flow in an isolated lung model. METHODS: 32 isolated, perfused, and ventilated rabbit heart-lung blocks were randomized to 4 groups (n=8), depending on levels of PEEP, pulmonary artery pressure (PAP) and left atrial pressure (LAP) as described in Table 1. Lungs were perfused with constant Q required to obtain the PAP assigned by randomization, and subsequently submitted to mechanical ventilation (MV) during 30 minutes in pressure-controlled mode (Ppeak = 30 cmH2O, RR = 20 rpm and PEEP 5 or 15). Pulmonary edema was assessed by weight (W) gain during MV. Data are expressed as mean ± SD. Statistical analysis: one-way ANOVA considering p<0.05 significant. RESULTS: See Table Group

PEEP PAP-LAP cmH2O mmHg I 5 20-10 II 5 35-10 III 5 20-20 IV 15 35-10 * p < 0.05 compared with group II

Q ml/min 33±17 529±181 40±0 329±213

TMP cmH2O 12 22 22 12

Wgain g/g lung 0±0.1* 4.4±2.9 0.7±0.8* 1.1±1.4*

CONCLUSION: Under these experimental conditions high Q and a high TMP are associated to development of VILI. The use of PEEP reduces edema formation after the MV period, probably by decreasing intrapulmonary Q and TMP. High Q was the main determinant of edema formation. The combination of high TMP and low Q produced minimal lung injury during MV. Grant acknowledgement: FIS 01/0947, 99/3091 and 01/F015, Red GIRA G03/063, Marató TV3 and Mútua Sabadellenca Fundació Privada.

808

810

ROLE OF PI3K-GAMMA IN VENTILATOR INDUCED LUNG INJURY

LUNG LAVAGE IS NOT A VALID EXPERIMENTAL MODEL OF ARDS

Lionetti V1, Lisi A1, Patrucco E2, Rivalta L1, Rossi A1, Ceci S1, Hirsch E2, Ranieri V M1 1 Anestesia e Rianimazione, 2Genetica, Università di Torino, Torino, Italy

Cunha-Goncalves D1, Perez-de-Sa V2, Ingimarsson J3, Curstedt T4, Johansson J5, Robertson B6, Tonnesen E7, Werner O2, Larsson A8 1 Thorax Anesth, 2Ped Anesth, USIL, Lund, 3Anesth, MAS, Malmo, 4Clin Chem, 5Med Bioch, 6Exp Pathol, KS, Stockholm, Sweden, 7Anesth, AKH, Arhus, 8ICU, Aalborg, Aalborg, Denmark

INTRODUCTION: Ventilator-induced lung injury is characterized by release of inflammatory mediators and pulmonary damage in a stretch-dependent manner. Phosphatidylinositol 3-kinase isoform gamma (PI3K-?) belongs to the family of phosphoinositide 3-kinases that regulates fundamental cellular functions. We investigated the role of PI3K-? in proinflammatory signalling mechanisms triggered by stretch during mechanical ventilation. METHODS: Ex vivo, isolated ventilated and not perfused lungs with ARDS (Lung lavage) obtained from wild type and PI3K?-null male mice were randomised to be ventilated for 3 hours, in a humidified chamber, with 3 ventilatory strategy: a) No Stretch Ventilation (n=8;CPAP: PEEP=5cmH2O); b) Low Stretch Ventilation (n=8; Vt=7ml/Kg; PEEP=3cmH2O;RR=40 bpm); c) High Stretch Ventilation (n=8;Vt=40ml/Kg; PEEP=0;RR=40 bpm). Pressure-volume curves were determined before and after the ventilation; one lung was frozen for biological analysis and the other was formalin-fixed for histological analysis. We measured changes in proinflammatory cytokines (IL-1, IL-6, TNF-alpha and MIP-2) by ELISA on lung tissue and phospho-Akt, a marker of PI3K-? activity and phospho-IkB-alpha, a marker of transcription factor nuclear factor kappa B(NF-kB) activation were measured (western blott, lung tissue). RESULTS: A significant reduction of compliance occurred after 3hours of High Stretch Ventilation in wild type, but not in PI3Kgamma-null; histological analysis showed hyaline membranes and detachment of epithelial cells in wild type, but not in PI3K? -null. Cytokines increased with the ventilatory strategy in both wild type and PI3K?-null, CPAP and Low Stretch Ventilation did not cause functional and morphological alterations no related with inflammatory mediators in both group. Western blot analysis revealed an increased phosphorylation of Akt after 3h of High Stretch Ventilation in wild type, but not in PI3K?-null. The phosphorylation of IkBalpha increased with stretch in both wild type and PI3K? -null. CONCLUSION: In this experimental model, mechanical stretch may induce functional and morphological alterations through a pathway that is PI3K-? mediated, but is probably independent of cytokines release and NF-kB activation. Grant acknowledgement: Supported by Cofin 2002

INTRODUCTION: Lung lavage, which causes surfactant depletion and lung collapse, is a commonly used experimental model of ARDS. We hypothesized that lung lavage per se does not elicit an ARDS-injury, but that the injury observed in some studies is rather a consequence of different degrees of ventilator induced injury. METHODS: 14 anesthetized and mechanically ventilated piglets (5.4 - 7.1 kg) were submitted to 8 consecutive lung lavages with saline 30 ml/kg. After recruitment, 7 piglets (lavage only or LOgroup) continued to receive VCV (TV 8 ml/kg, PEEP 10 cmH2O, PIP < 35 cmH2O, I:E 1:1), while the remaining 7 were submitted to 45 min of lung injurious ventilation (LIV = PCV, PIP 40 cmH2O, NEEP –5 cmH2O, I:E 1:1). After a new recruitment, PEEP was increased to 15 cmH2O in both groups. FRC (SF6-washout), PV-loops (for measurements of maximal static compliance from the deflation limb _Crs, and inpiratory capacity _IC), arterial blood-gases, plasma and BAL cytokines, and protein in the lavage fluid were measured before lung lavage (stage1), 30 min (stage 2), and 4 h (stage3) after LIV (or at equivalent time in LO). After death, the lungs were sent to histological and cytokine analysis. RESULTS: There were no differences between the groups at baseline (Mann-Whitney), but at 4 hours all measured variables showed extensive lung injury in the LIV-group, but not in the LOgroup. Measurements at stage 3 (Mean ± SD) Crs PaO2/FiO2 mL/kg LO 3.00±0.90 526±44 LIV 1.55±0.32 176±73 p (p=0.001) (p=0.001)

PaCO2*plat pressure 168±16 236±25 (p=0.017)

Protein g/L 0.91±0.41 1.57±0.52 (p=0.035)

CONCLUSION: Lung lavage by itself did not induce a lung injury consistent with ARDS, and this model is therefore unsuitable for evaluation of new treatment modalities for this syndrome. On the other hand, lung lavage followed by standardized injurious ventilation elicited a stable ARDS-like lung injury. Grant acknowledgement: Danish Medical Research Council, grant no 22-03-0299

S210

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

811 ACTIVATED PROTEIN C ATTENUATES OLEIC ACID-INDUCED LUNG INJURY IN SHEEP Waerhaug K1, Kirov M Y1, Kuzkov V1, Kuklin V1, Bjertnaes L J1 Department of anesthesiology, Faculty of Medicine, University of Tromsø, Tromsø, Norway

1

INTRODUCTION: Acute lung injury may evolve from fat embolism after trauma to the skeleton. Oleic acid (OA), when injected into the pulmonary circulation, produces a lung oedema reminding on non-septic ALI (1). We have recently reported that activated protein C (APC) counteracts endotoxin-induced lung injury in sheep (2). However, it still remains unsettled whether APC exerts a similar effect on ALI subsequent to fat embolism. Our aim was to find out whether APC ameliorates OA-induced ALI in sheep. METHODS: Twenty-two yearling sheep underwent instrumentation for a study awake. The day of experimentation, sheep received a continuous intravenous (I.V.) infusion of saline (5 ml/kg/hr) and were randomized to three groups: 1) An OA+APC group (n=8) receiving OA 0.06 ml/kg (Sigma Chemical, St. Louis, MO) dissolved in 50 ml of the animals own blood and infused over a 30 min in parallel with a continuous I.V. infusion of APC (drotrecogin alpha, Eli Lilly & CO) 24 mg/kg/h during 2 hrs; 2) an OA group received OA as above (Control group, n=9); 3) a shamoperated group (n=6) was observed throughout. At 2 hrs, sheep were sacrificed with an I.V. injection of thiopental sodium 15 mg/kg (Abbott, North Chicago, IL) and KCl 1 mmol/kg. Haemodynamics and gas exchange were determined at 30 min intervals. Extravascular lung water index (EVLWI) was assessed by transpulmonary single thermodilution (PiCCO plus; Pulsion Medical Systems, Münich, Germany). Data were analyzed by ANOVA and Scheffe’s test; p< .05 was regarded as statistically significant. RESULTS: Preliminary data show that APC significantly improves oxygenation, as evaluated by the prevention of PaO2 from falling in the OA+APC group (p<.05). APC also precludes lung microvascular pressure and EVLWI from rising (p<.03). CONCLUSION: APC constitutes an effective treatment of OA-induced ALI.

Oral Presentations Experimental evaluation of therapeutic approaches for systemic inflammation and sepsis – 813-818 813 HYPOXEMIC RESUSCITATION OF HEMORRHAGIC SHOCK IN RABBIT ATTENUATES PULMONARY INFLAMMATORY RESPONSE Augustatou K1, Livaditi O1, Kolintza A1, Krania M1, Orfanos S E1, Roussos C1, Douzinas E E1 Critical Care Dpt EvangelismosHospital,M Simou Lab, University of Athens, Athens, Greece

1

INTRODUCTION: Shock resuscitation has been associated with the development of acute lung injury. Hypoxemic reperfusion has been shown to exert a favourable effect resuscitating ischemic states (1,2. In the current study the effect of hypoxemic resuscitation of hemorrhagic shock on the pulmonary inflammatory response was investigated in the rabbit. METHODS: Shock was induced within 30 minutes by exsanguinations to mean arterial pressure (MAP) of 35 mmHg that was maintaining for 60 min. Ten adult male New Zealand white rabbits were randomly allocated to receive normoxemic [group control (C), n=6, PO2=95-105mmHg), or hypoxemic (group HR, n=4, PO2=35-40mmHg with gradual return to normoxemia) resuscitation for the next 120 min. Bronchoalveolar lavage (BAL) was performed in the right lung; left upper lobe was removed for wet/dry (W/D) weight ratio. Cells in BAL fluid (BALF) were differentiated (WrightGiemsa staining) and counted (Coulter Electronics) and cytokines were measured by flow cytometry with a FACS-Calibur using cytometric bead array. For the statistical analysis Student-t test was performed. RESULTS: The W/D weight ratio was 7,294 ± 0,580 and 3,285 ± 0,826 in C and HR groups respectively (p=0,006). Total cell count per ml of BALF was 2,26x106 ± 7,8x106 and 0,96x106 ± 0,25x106 (p=0,022), neutrophils 4923 ± 1392 and 2901 ± 700 (p=0,09) and macrophages 2,09x106 ± 0,86x106 and 0,79x106 ± 0,2x106 (p=0,042) in C and HR groups respectively. Concentrations of inflammatory cytokines in BALF are presented in the figures.

REFERENCES: 1.Schuster DP. Am J Respir Crit Care Med. 1994;149(1):245-60. 2.Kuklin VN et al. Acta Anaesth Scand. 2003 ; 47(suppl 16) :p 32. (Abstract) Grant acknowledgement: Helse Nord project 4001.721.132

CONCLUSION: Hypoxemic resuscitation of hemorrhagic shock seems to attenuate the pulmonary inflammatory reaction. REFERENCES: 1. Douzinas EE et al. The effect of hypoxemic reperfusion on cerebral protection after a severe global ischemic brain insult. Intensive Care Med 2001; 27:269-275. 2. Douzinas EE et al. Myocardial ischemia in intestinal postischemic shock: the effect of hypoxemic reperfusion. Crit Care Med. 2003; 31:2183-9.

812

814

ET-1 RECEPTOR BLOCKADE PREVENTS SEPSIS-INDUCED INCREMENT IN LUNG MICROVASCULAR PERMEABILITY

PROTECTIVE ROLE OF RECOMBINANT BETA-DEFENSIN-2 PEPTIDE ON LUNG INJURY

Kuklin V N1, Sovershaev M A2, Skogen V3, Andreasen T4, Ytrehus K4, Bjertnaes L J1 1 Department of anesthesiology, 2Department of Biochemistry, 3Department of Microbiology, 4Department of Physiology, Faculty of Medicine, University of Tromsø, Tromsø, Norway

1

INTRODUCTION: We recently demonstrated in sheep that the endothelin-1 (ET-1) receptor blocker tezosentan, antagonizes acute lung injury (ALI) following infusion of endotoxin or ET-1 (1) by reducing the enhanced lung microvascular pressure. However, the influence of tezosentan on lung microvascular permeability still remains unsettled. Our aim was to evaluate whether ET-1 receptor blockade attenuates sepsis-induced changes in lung microvascular permeability. METHODS: Sepsis was induced in rats by cecum ligation and puncture (CLP). A CLP group (n=7) received saline and a CLP+Tezosentan group (n=7) additionally received tezosentan (Actelion Ltd., Switzerland) 30 mg/kg. A sham-operated group (n=5) underwent laparotomy and closure without CLP. Twelve hrs postoperatively, the lungs were isolated and perfused with whole blood from 38 similarily treated rats. Papaverin (0,1 mcg/kg) was added to the perfusate for vascular paralysis. Sepsis was affirmed by culturing blood for aerobic and anaerobic bacteria. Pulmonary arterial pressure (PAP), microvascular pressure (Pmv), fluid filtration coefficient (Kfc) and lung compliance (CL) were determined every 30 min for 210 min. Data were analyzed by ANOVA and Scheffe’s test; p<.05 was regarded as statistically significant. RESULTS: At baseline, no intergroup differences were found. Due to vascular paralysis, there were no intra- or intergroup differences in PAP and Pmv throughout the experiment. In the shamoperated rats, Kfc and CL remained unchanged. In the CLP group, Kfc increased 3-fold and CL decreased 4-fold in parallel with increasing pulmonary edema from 30 to 90 min. Marked differences in Kfc between the CLP and the CLP+Tezosentan groups were found from 30 to 90 min (p<.05). In the CLP groups, blood culturing revealed polymicrobial Gram-positive and Gramnegative infection while no growth was found in the sham-operated rats. CONCLUSION: In isolated blood perfused rat lungs, blockade of ET-1 receptors inhibits the sepsis-induced increase in microvascular permeability. REFERENCE: 1. Kuklin VN, et al. Critical Care Medicine 2004; 32:766-773. Grant acknowledgement: Helse Nord project 4001.721.132

Fang X M1, Chen Z2, Shi Z1, Shu Q3, Yao H P1, Stuber F4 Anaesthesiology, 2Infection, 3Cardiovascular surgery, Zhejiang University, HangZhou, China, 4Anaesthesiology, University of Bonn, Bonn, Germany

INTRODUCTION: Human beta-defensin 2 (HBD2)is a cationic antimicrobial peptide with little development of microbial resistance. In vitro, HBD-2 has also been demonstrated to be a potent chemoattractant for human neutrophils. The purpose of this study was to investigate the antimicrobial activity of recombinant b-defensin-2 peptide in P. aeruginosa-induced lung injury in vivo, furthermore, to investigate the anti-inflammatory role of recombinant b-defensin-2 peptide in CLP-induced lung injury in vivo. METHODS: Reconstruct adenoviral genome by homologous recombination and insert rat bdefensin-2 (rBD2) gene, which can express mature rBD2 efficiently in eukaryotic cell. Twenty SD male rats weighting 250-300g were randomly divided into four groups of six animals each: in group A & C, control adenovirus without the rBD2 gene was titrated into trachea via intubations, in group B & D, adenovirus with the rBD2 gene was titrated into trachea via intubations, 48 hrs later, 200ul of 6?108CFU/ml Pseudomonas aeruginosa ATCC27853 was titrated into trachea in group A & B, while cecal ligation and puncture (CLP) was performed in group C & D. Twentyfour hours following the titration of bacteria or CLP, lungs, blood and bronchoalveolar lavage fluid (BALF) were harvested. The expression of rBD-2 mRNA & peptide were determined by RT-PCR and western blotting respectively. Immunohistochemistry was performed to detect the pathological alteration of pulmonary tissue, expression of rBD2 and ICAM-1. Leukocyte count of blood and BALF, and bacteria clearance were examined via culture and cytometer. RESULTS: Group B & D expressed rBD2 while group A & C didn’t express rBD2. The damage degree of lung in group B & D was significantly lighter than that in group A & C. The ICAM-1 expression and the count of bacteria CFU in BALF in group B & D was significantly less. The leukocyte count of blood had no significant difference between groups, but the leukocyte count of BALF in group B & D was significantly less than that in group A & C. CONCLUSION: The recombinant adenovirus vector can express rBD2 efficiently in vivo. The expressing recombinant rBD2 is of antimicrobial activity in vivo, and of anti-inflammatory role in pulmonary tissue. The mechanism of beta-defensin against the lung injury needed to be elucidated. This study suggested that beta-defensin 2 could be serve as templates for the development of pulmonary pharmaceuticals. REFERENCES: 1. J Immunol. 2003; 171(12): 6820-6826. 2. Immunology. 2004; 111(3): 273281.3. Inflamm Res. 2002; 51(1): 8-15. Grant acknowledgement: This study has been supported by National Natural Science Foundation of China (NSFC), No.A30070854

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

815

817

POTASSIUM CHANNEL BLOCKADE RESTORES THE ATTENUATED NORADRENALINE SENSITIVITY IN HUMAN ENDOTOXEMIA

ANTITHROMBIN III-MEDIATED EFFECTS ON LEUKOCYTES INVOLVE SYNDECAN-4-LINKED SRC TYROSINE KINASES

Pickkers P1, Sprenger R1, Smits P2, Van der Hoeven H1 1 Intensive Care, 2Pharmacology-Toxicology, University Medical Centre Nijmegen, Nijmegen, Netherlands

1

INTRODUCTION: Sepsis-induced vasodilation is characterized by an attenuated sensitivity to vasoconstrictor substances like noradrenaline. Upregulation and activation of vascular potassium (K) channels may play a role, as in animal experiments K channel blockers are able to restore sepsis-induced decrease in vascular tone and blood pressure. Aim of present study was to determine whether the attenuated noradrenaline sensitivity could be restored by K-channel blockers quinine and tetra-ethyl ammonium (TEA) in humans volunteers that received endotoxin. METHODS: The brachial artery was cannulated for infusion of drugs. Forearm blood flow (FBF) was measured in both arms using strain-gauge venous occlusion plethysmography. Endotoxin (2 ng/kg) was administrated i.v., noradrenaline (1-3-10-30 ng/min/dl), quinine (n=6, 50 ?g/min/dl), TEA (n=5, 1 mg/min/dl), or saline (n=6, time controls) were administrated intraarterially.RESULTS: Intraarterial noradrenaline infusion decreased FBF. (Percentage of baseline ratio: 100±0, 84±4, 70±4, 55±4, 38±4%,(mean±SEM). Following endotoxin administration (2 ng/kg), noradrenaline-induced vasoconstriction was attenuated: 100±0, 101±4, 92±4, 83±6, 56±7% (p<.001, pooled data, n=17). Time control experiments demonstrated excellent repeatability of the attenuated noradrenaline response. K channel blocker quinine did not significantly restore the noradrenaline response. In contrast, TEA completely restored the vasoconstrictive effect of noradrenaline (Table, n=5, p=0.05). Percentages of baseline ratio Noradre naline 0 1 3 10 30

Baseline 100±0 79±6 72±6 56±7 47±5

Endotoxin 100±0 106±5 97±7 95±15 73±15

Endotoxin + TEA 100±0 84±8 72±5 61±6 52±8

S211

Mosheimer B A1, Feistritzer C1, Tancevski I1, Wiedermann C J1 Internal Medicine, University of Innsbruck, Innsbruck, Austria INTRODUCTION: Tyrosine kinases are known to play a critical role in the regulation of leukocyte function. Antithrombin III (AT) mediates its effects via syndecan-4 which is known to be linked to SRC-kinases. In this study we investigated the role of SRC –family selective tyrosine kinase inhibitors on AT regulated leukocyte migration and SRC - tyrosine kinase phosphorylation in response to stimulation with AT. METHODS: Neutrophils and monocytes were obtained from forearm venous blood. Monocytes were purified using a MACS separation system. Chemotaxis assays were performed using a 48well Boyden microchemotaxis chamber. Neutrophils and monocytes were pre-treated with various SRC –family selective tyrosine-kinase inhibitors with or without AT followed by washing and assessment of their migratory response toward AT or interleukin-8 and RANTES, respectively. Activation status of the two major SRC – tyrosine kinase phosphorylation sides Tyr-416 and Tyr527 were tested using western blot analysis. RESULTS: Dose-dependent reversal of the AT-mediated effects on neutrophil and monocyte migration was induced by the selective SRC-kinase inhibitors PP1 and PP2, which induces highly specific inhibition of the SRC kinases Lyn, Fyn and Hck. PP3, the inactive analogue of PP1 which also blocks the EGFR, did not effect AT-induced leukocyte migration. AT increased Tyr416 and decreased Tyr527 phosphorylation of SRC-tyrosine kinases in a time and dose dependent manner. This effect could be blocked by the inhibitors PP1 and PP2 but not PP3. Moreover, co-incubation with AT lowered the level of RANTES-induced Tyr416 phosphorylation in monocytes. Stimulation of monocytes with AT led to an increased expression of the Src kinase Lyn, but not of Hck and Fgr. CONCLUSION: SRC-kinases known to be linked to syndecan-4 play a role in mediating ATinduced effects on leukocyte motility.

CONCLUSION: In experimental human endotoxemia, noradrenaline sensitivity is decreased compared to baseline. Potassium channel blocker TEA completely restores the vasoconstrictive effects of noradrenaline. Pharmacological blockade of vascular potassium channels may represent a novel intervention that deserves further elucidation in sepsis-induced potassium channel activation and attenuated noradrenaline sensitivity in humans. Grant acknowledgement: PP is a recipient of a Clinical Fellowship grant of the Netherlands Organisation for Scientific Research (ZonMw).

816

818

VITAMIN D3 DOWNREGULATES MONOCYTE TLR4 EXPRESSION: REDUCED TF ACTIVITY AFTER LPS STIMULATION

RECOMBINANT HUMAN ACTIVATED PROTEIN C UPREGULATES PRODUCTION OF PDGF-BB IN HUMAN ENDOTHELIAL CELLS

Sadeghi K1, Wessner B1, Tamandl D1, Laggner U1, Ploder M1, Roth E1, Boltz-Nitulescu G2, Spittler A2 1 Surgical Research Laboratories, 2Department of Pathophysiology, Medical University of Vienna, Vienna, Austria

Brueckmann M1, Engelhardt C1, Lang S1, Hoffmann U1, Liebetrau C1, Borggrefe M1, Haase K K1, Huhle G1 1 Department of Medicine I, Faculty of Clinical Medicine Mannheim, University Hospital of Heidelberg, Mannheim, Germany

INTRODUCTION: Toll-like receptors (TLRs) represent an ancient front-line defence system that act as crucial mediators of microbial sensing and as triggers of inflammation, respectively. Monocytes play a central role in inflammation and express among other receptors TLR4, the major endotoxin receptor. TNF-alpha (TNF) and tissue factor (TF) play a dominant role in the development of disseminated intravascular coagulation (DIC) after microbial invasation. In a rat model it has been previously shown that treatment with vitamin D3 (1alpha,25dihydroxycholecalciferol) diminishes the development of LPS-induced DIC. In the present study we therefore examined the effects of vitamin D3 on monocyte TLR4 expression and investigated cytokine secretion and functional tissue factor activity followed by stimulation with endotoxin.

INTRODUCTION: Platelet-derived growth factor has been shown to promote the structural integrity of the vessel wall, to be angiogenetic, to contribute to the regulation of blood vessel tonus, to decrease platelet aggregation and to increase wound healing capacity. Non-anticoagulant biological activities, such as anti-inflammatory and anti-apoptotic mechanisms of action, have been suggested for recombinant human activated protein C (rhAPC; drotrecogin alfa (activated)). However, little is known about the effect of rhAPC on the endothelial synthesis and release of growth factors, such as platelet-derived growth factor-BB (PDGF-BB).

METHODS: Human monocytes from healthy donors were isolated by negative depletion using magnetic beads. Subsequently, cells were cultured with various concentrations of Vit. D3 (10-6 to 10-10 M) and additionally stimulated with lipopolysaccharid (LPS). Expression of TLR4, CD14, tissue factor (CD142) and intracellular TNF production was measured by flow cytometry. TLR4 mRNA level was determined by semiquantitative RT-PCR and was related to beta-actin and 18SrRNA gene expression. TNF secreted in the supernatant was measured by chemoluminescence (Immulite). To analyse functional tissue factor activity one stage clotting assay was performed. RESULTS: Vitamin D3 induced a time- and dose-dependent downregulation of TLR4 while CD14 levels were significantly enhanced. A significant reduction of TLR4 surface protein was initially observed after 12 hours (TLR4: control 34.3±7.1% vs vit.D3 13.6±1.8%; p<0.05) and maintained after 72 hours (TLR4: control 54.8±3.5% vs vit.D3 11.6±1.0%; p<0.001). Moreover, TLR4 mRNA was found to be reduced after incubation with vitamin D3. TNF production after LPS challenge was significantly diminished in a vitamin D3 dose-dependent manner. Similarly, vitamin D3 reduced LPS-induced procoagulatory tissue factor expression. One stage clotting assays revealed reduced functional TF activity after LPS stimulation. CONCLUSION: Our data show that vitamin D3 downregulates the expression of TLR4 on protein and mRNA level leading to reduced TNF production and TF activity after LPS stimulus. Since TNF and TF are important triggers of DIC, vitamin D or analogues that exhibit hypocalcemic properties might be useful therapeutic agents in the treatment of sepsis.

METHODS: The effect of rhAPC (50ng/ml-10µg/ml) on the release of PDGF-BB into supernatants of human umbilical vein endothelial cells (HUVEC) was measured by an enzymelinked immunosorbent assay (ELISA). PDFG-BB-messenger RNA in endothelial cells was determined by reverse transcriptase-polymerase chain reaction (RT-PCR). Experiences were performed in the presence of recombinant hirudin (lepirudin, 5µg/ml) to rule out any effect from residual thrombin activity within the rhAPC preparation. Statistical analysis from at least three independent experiments was performed by student’s t-test and ANOVA. RESULTS: RhAPC stimulated PDGF-BB-mRNA-gene transcription and PDGF-BB-proteinsynthesis in a time- and dose-dependent manner: PDGF-BB was detected in HUVEC-supernatants as early as 6 h after the addition of rhAPC (1-10µg/ml) and continued to increase through the 24h incubation period (p<0.05 for rhAPC 1µg/ml; p<0.001 for rhAPC 10µg/ml). Corresponding to the increased amounts of PFGF-BB protein in supernatants, PDGF-BB-mRNA was also up-regulated as compared to untreated controls after incubation with rhAPC (10µg/ml). CONCLUSION: The ability of rhAPC to upregulate PDGF-BB production in human endothelial cells may represent a new molecular mechanism, by which rhAPC controls vessel wall homeostasis and increases wound healing capacity, thereby contributing to the efficacy of rhAPC in systemic inflammation and sepsis. Further in vitro and in vivo investigations are needed to confirm these results. Grant acknowledgement: This study was supported by a grant of the Faculty of Clinical Medicine Mannheim, University of Heidelberg, Germany.

S212

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

Oral Presentations Outcomes in community-acquired pneumonia – 819-824 819

821

MORTALITY AND OUTCOME OF COMMUNITY-ACQUIRED PNEUMONIA NEEDING MECHANICAL VENTILATION Tejerina E E1, Frutos-Vivar F1, Esteban A1, Anzueto A2, Apezteguia C3, Bugedo G4, Gonzalez M5, Abroug F6 1 Intensive Care Unit, Hospital Universitario de Getafe, Getafe, Spain, 2Intensive Care Unit, University of Texas Health Science Center, San Antonio, United States, 3Intensive Care Unit, Hospital Profesor Posadas, Buenos Aires, Argentina, 4Intensive Care Unit, Hospital Pontificia Universidad Catolica, Santiago, Chile, 5Intensive Care Unit, Hospital General de Medellin, Medellin, Colombia, 6Intensive Care Unit, Hospital Fattouma Bourguiba, Monastir, Tunisia INTRODUCTION: Approximately 10 to 36% of patients with community-acquired pneumonia (CAP) needing hospitalization require intensive care unit (ICU) treatment among whom a mortality rate of 22 to 53% has been reported. Respiratory failure develops and mechanical ventilation (MV) is required in 58 to 87% of patients with severe CAP. The objectives of this study were to estimate its attributable morbidity and mortality in a heterogeneous cohort of mechanically ventilated patients >12 hours. METHODS: We used the database of a prospective, multicenter, international cohort of 5183 adult patients who received MV >12 hours from March 1, 1998 to March 31, 1998 at 361 ICUs (1). We excluded of the analysis those patients whose reason of MV was pneumonia after the first day of hospital stay (N=527). To determine the attributable morbidity and mortality of CAP we did a matched cohort study. The matched variables were variables that we have reported to be associated with mortality in mechanically ventilated patients (1): a) variables present at the beginning of MV(age, SAPS, limited activity as previous functional status); b) variables related to patient management(use of vasoactive drugs and use of neuromuscular blockers); c)complications developing over the course of MV (ARDS, sepsis, shock, renal failure, hepatic failure, coagulophaty and ratio paO2/FiO2). RESULTS: We found 124 controls without CAP for the 124 patients with CAP. There were no statistically significant differences in hospital mortality (32.3% vs. 35.5%; p=0.59), duration of MV [median 5 days (interquartile range: 3,8) vs. 4 days (interquartile range: 2,6), p=0.23] and ICU length of stay [median 8 days (interquartile range: 4,15) vs. 6 days (interquartile range: 4,13), p=0.55] among patients with or without CAP needing MV. CONCLUSION: In conclusion, community-acquired pneumonia needing mechanical ventilation is not a factor associated with mortality, controlling for potential confounding variables. And, it does not prolong either duration of MV or ICU length of stay. REFERENCE: 1. Esteban A, Anzueto A, Frutos F, Alía I, Brochard L, Stewart TE, Benito S, Epstein SK, Apezteguía C, Nightingale P, Arroliga AC, Tobin M, for the Mechanical Ventilation International Study Group: Characteristics and outcomes in adult patients receiving mechanical ventilation. A 28-day international study. JAMA 2002; 287:345-355.

INTRODUCTION: Severe community acquired pneumonia (SCAP) is a frequent pathology with a high mortality. Early administration of antibiotics and intensive vital treatment could reduce the complications and organ failure derived from this kind of pathology. We evaluated the different mortality, length of stay (LOS) in ICU and length of mechanical ventilation.

820

822

ASSESSMENT OF COMMUNITY ACQUIRED PNEUMONIA WITH FINE’S SCORE AND A RADIOLOGICAL SCORE

ICU PNEUMOCOCCAL COMMUNITY ACQUIRED PNEUMONIA: EPIDEMIOLOGY AND PROGNOSIS FACTORS

Borsali-falfoul N1, Chargui A1, Zaghdoudi I1, Rezgui M1, Hachicha Y1, Jerbi Z1 1 Emergency and critical care, Habib thameur hospital, tunis, Tunisia

SEVERE COMMUNITY ACQUIRED PNEUMONIA: DIFFERENT EVOLUTION ACCORDING TO THEIR INITIAL MANAGEMENT Taboada R1, González P1, Pérez C1, López D1, Gamo A1, Pérez M1, Suárez F1, Alcalá M1 1 ICU, Fundación Jiménez Díaz, Madrid, Spain

METHODS: Prospective observational without intervention study. We include patients admitted to a polyvalent ICU diagnosed of SCAP. Patients were assigned into two groups according to their origin, group 1: emergency room or group 2: ward less than 48 hours. We collected data like severity scores at admission time, age, gender and co morbidity factors to compare both groups. We gather their mortality, LOS in ICU and length of mechanical ventilation. We evaluated the influence of mild and moderate acute renal failure (ARF) (creatinine >1.5 and 2 mg/dL respectively) and moderate and severe hepatic failure (bilirubine >2 and 4 mg/dL respectively). Statistical analysis was performed with the chi-square statistic for qualitative variables and Fisher’s exact test for quantitative variables; p< 0.05 was considered significant. RESULTS: We included 97 patients, 44 from the emergency room and 53 from a ward. Both groups were homogeneous looking their first severity index, age and gender like we can see in the attached. There was not either difference in co morbid factors between the groups. Mortality of patients admitted from the ward was 45% and mortality of that from emergency room was 22.7%, with a significant difference, p=0.002. There was not difference neither in length of mechanical ventilation (11.96±11.18 vs. 15.53±15.14 days, NS) nor in LOS in ICU (13.88±15.32 vs. 16.35±15.12 days, NS) for patients admitted from emergency room and from the ward respectively. Both renal and hepatic failures have no influence for a longer LOS in ICU or length of mechanical ventilation. We found as independent factors for higher mortality mild and moderate ARF (p=0.003 and p=0.005) and severe hepatic failure (p=0.004). Analyzing this factors in each group we found not difference in group 1 patients, but it was very significant in group 2 (p=0.004, p=0.0017 and p=0.0037 respectively). CONCLUSION: Patients diagnosed of SCAP admitted early in the ICU have a smaller mortality than those admitted previously in a ward. Initial intensive treatment in the second group could reduce the mortality of SCAP preventing the development of moderate acute renal failure and severe hepatic dysfunction.

Ben Romdhane K1, Ghadoun H1, Ben Khelil J1, Sonia C1, Belkodja K1, Besbes M M B1 Intensive care, A. Mami Hospital, Ariana, Tunisia

1

INTRODUCTION: Community acquired pneumonia (CAP) is a frequent cause of emergency consultation. The fine’s score seems correlated to the seriousness and useful to help emergency physicians to decide the hospitalization (1-2). To improve prediction of outcome we use the Murray’s Radiological Score (RS) (3) with the fine’s score and analyse the correlation between the 2 scores.

INTRODUCTION: Streptococcus pneumonia is the leading cause of community acquired pneumonia (CAP). Its attributable, despite the emergence of new antimicrobial agents and improvements in critical care medicine morbidity and mortality in recent, decades has remains unchanged. The Study objective was to evaluate the epidemiology profile and to study the different prognosis factors of pneumococcal CAP admitted in the ICU.

METHODS: a cohort study of patients received at an emergency department of a general teaching hospital, which the main diagnosis is a CAP. FS and RS are calculated on admission and we analyse the place of hospitalization and outcome at 30 days. Statistical analysis: chi 2 test with significant statistical difference (SSD) if p 0,05.

METHODS: Patients hospitalised between 1994 and 2004 with diagnosis of CAP, whose bacteriologic investigations (blood cultures, pulmonary samples and/or pleural effusion); obtained within 48 hours of hospitalization; demonstrated growth of Streptococcus pneumonia were retrospectively included in this studied.

RESULTS: 174 CAP, 104 men and 70 women, mean age = 57,5 17,7 years, the main results are on the table 1. There is a correlation between FS and RS (r=0.16, p=0.034). The RS in survivors versus (vs) dead in the FS categories II, III, IV and V is respectively 3.46± 1.3 vs 5±1, 3.25 ± 1 vs 5 ± 2.4, 3.7 ± 1.7 vs 4.8 ± 2.3 and 4 ± 1.3 vs 5 ± 2.5. There is a SSD only in category II.

RESULTS: Of 244 patients with CAP admitted in this period, 94 patients (38,5%) had pneumococcal CAP. Mean age was 43,9 ± 20,8 years. Twenty three (24,5%) were over 65 years old. The main underlying diseases consists on COPD (28,7%) and diabetes mellitus (10,6%). Mean IGSII gravity score was 30,6 ± 18,3 and fifty five patients (58,5%) were over class III of fine score. Principal clinical features on admission were represented by acute respiratory failure (57,4%), shock (11,7%), coma (7,4%), acute renal failure (46,8%), rhabdomyolisis (19,1%), hepatic failure (6,4%) and CIVD (3,2%). Thirty two patients (34%) had bilateral attaint on chest radiography. The Bacteriologic study that confirms pneumococcal CAP were: BAL (88,3%), blood culture (6,4%), pleural effusion (3,2%) and sputum cultures (2,2%). Thirty six patients (38,3%) underwent mechanical ventilation and fifteen (15,9%) develop multi organ failure. Mortality rate was 23,4%. History of diabetes mellitus, pH 14 mmol/l on admission were independently associated with death. Risk of death was significantly higher when IGSII score is over 25 (OR = 15,6 ; p<10-3) and when patients are over class III of Fine score (OR = 16,9 ; p<10-7).

FS N (%) RS Checking out (%) Internal medicine % ICU % dead %

I 6(3.4) 3±0 40 60 0 0

II 46(26.4) 3.6±1.4 13.9 83.3 0 2.8

III 40(23) 3.5±1.5 22.6 53.4 17.5 6.5

IV 67(38.5) 4.1±2 11.3 15.1 52.8 20.8

V 15(8.6) 4.3±2 0 16.6 41.7 41.7

CONCLUSION: RS is correlated to FS. RS may improve FS and other studies with larger cohort are necessary to show a SSD. REFERENCES: (1) Fine MJ and al. N Engl J Med 1997; 336:243-50. (2) Jerbi Z and al. Réanimation 2002;11 suppl. 3:178s (abstract). (3) Murray JF and al. Am Rev Respir Dis 1988; 138:720-3.

CONCLUSION: Pneumococcal community-acquired pneumonia in the ICU is frequent and is responsible of high morbidity and mortality. IGS II over 25 and class III of fine score are associated with a significantly high death. REFERENCE(S): 1. Community acquired pneumonia mortality, Respiratory Med(2004) 98, 1724 2. Community acquired pneumoniae, lancet 2003, 362: 1991-2001. 3. Prospective observational study of bacteremic pneumococcal pneumonia, Crit Care Med 2004; 32:625-631. 4. Pneumococcal disease in Sweden, Am J Med 1999;107(1A):44S-49s

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

823 NON INFECTIOUS ETIOLOGIES IN PATIENTS ADMITTED TO THE ICU BY SEVERE COMMUNITY-ACQUIRED PNEUMONIA Carbajal-Guerrero J1, Fernandez-García E1, Aldabó-Pallás M1, Garnacho-Montero J1, Ortiz-Leyba C1 1 Intensive Care Unit, Hospital Universitario Virgen del Rocío, Seville, Spain INTRODUCTION: Non-infectious aetiologies must be considered in all cases of non-resolving pneumonia. Our objectives were to evaluate the incidence, clinical presentation, diagnosis, treatment, and outcome of the non-infectious causes mimicking pneumonia in a of patients admitted to the intensive care unit with the diagnosis of severe community-acquired pneumonia. METHODS: : Prospective cohort study enrolling all patients admitted to the ICU with the diagnosis of severe community-acquired pneumonia from January 2001 to December 2003. Patients with HIV infection were not included in the study. The following data were collected: APACHE II at the admission the ICU, underlying diseases, blood culture, culture of respiratory sample (expectorate or endotracheal suction sputum), Legionella urinary antigen, pneumococcus urinary antigen, and serology. Patients were studied until death or hospital discharge. RESULTS: : Fifty-four patients were included in this study. Thirty-four required mechanical ventilation (57,4%). Microbiogical documentation was obtained in 23 patients (42,5%). Most frequent germens were Streptococcus pneumoniae (12) and Legionella pneumophila (7). In four cases (7.4%) an alternative diagnosis was obtained due to the absence of microbiogical documentation and a non-consistent evolution. All these patients had presented respiratory failure at the admission to the ICU.In all of these cases a thoracic CT scan was performed suggesting the alternative diagnosis. The final diagnosis was eosinophilic pneumonia in two cases and bronchiolitis obliterans organizing pneumonía (BOOP) in the other two patients. A patient suffering from BOOP died without having received corticosteroids and three of them were treated with corticosteroids intravenously (1g Metyl-prednisolone i.v per day for five days). One patient died and the necropsy confirmed the diagnosis of BOOP. CONCLUSION: Non-infectious causes mimicking a community-acquired pneunomia should be considered in patients with respiratory failure, absence of microbiological documentation and unfavourable clinical evolution. These alternative diagnosis are indeed important, for starting an early treatment with corticosteroids that may modify the evolution of disease. Our series confirms that an early CT scan is a useful tool in the management of these patients.

S213

Oral Presentations Monitoring II – 825-830 825 SAFETY AND EFFICACY OF A CLOSED LOOP OXYGEN DELIVERY VENTILATOR Modan I1, Shapiro M1, Grozovski E1, Cohen J1, Singer P1 1 General Intensive Care, Rabin Medical Center, Petah Tikva, Israel INTRODUCTION: While providing adequate oxygen for optimal cell function is a critical goal of ventilated patients, providing excess oxygen, by promoting free radical production, may be harmful. The inspiratory oxygen concentration (FiO2) delivered by conventional ventilators is determined by the attending staff and is frequently inappropriately high for prolonged periods. In order to achieve an optimal FIO2, a closed loop oxygen delivery system (iVent, Versamed, Israel) was recently developed and tested now for the first time. METHODS: Forty ventilated stable patients requiring a FiO2> 0.50 were included in the study after informed consent was obtained. All patients were ventilated with the iVent ventilator and randomized into 2 groups: group 1, a control group (n = 20), where the FiO2 was adjusted by the attending staff according to usual practice in the unit , and group 2, the intervention group, (n = 20), where FiO2 was adjusted by a closed loop system ( the afferent component comprising a pulse oximeter built into the respirator and the efferent component a mechanism decreasing FiO2 by 5% as long as SpO2 was higher than the goal value (92%) with the aim of reaching the lowest FiO2 for the goal SpO2. SpO2 values were validated by SaO2 measurements. The 2 methods were compared using ANOVA analysis of variance and Student’s t tests. RESULTS: The 2 groups were similar in age (58.8 vs 59.6 years) and initial PaO2/FiO2 (145.1 ± 40.6 vs. 188.3 ± 96.7). The APACHE II score was higher in the control group (21.7 ± 7.6 vs 14.9 ± 5.1). No decrease below the goal SpO2 was noted in the intervention group. FiO2 levels T0 T1 hour Group 1 FiO2 0.66 0.61 Group 2 FiO2 0.60 0.50* * p value vs baseline (T0) <0.05

T2hours 0.55 0.44*

T6hours 0.51* 0.40*

T24hour 0.50* 0.39*

T2hours 218 202.5

T6hours 225.7 227.7

T24hour 234.7 224.3

paO2/FiO2 levels T0 T1hour Group 1 188.3 218 Group 2 169.8 183.1 p=NS vs baseline and Gp1 vs Gp 2

CONCLUSION: We have shown that a new closed loop oxygen delivery ventilator more effectively decreased FiO2 when compared to usual clinical practice and was safe.

824

826

FATAL SEVERE COMMUNITY-ACQUIRED PNEUMONIA. A MULTICENTER SPANISH STUDY

DETECTION OF INCREASED LUNG VOLUME DURING MECHANICAL VENTILATION BY RESPIRATORY SOUNDS ANALYSIS

Bodí M1, Rodríguez A1, Gilavert C1, Vidaur L1, Blanquer J2, Solé-Violan J3, Rello J and SCAP multicenter spanish study group 1 1 Critical Care, H. U. Joan XXIII, Tarragona, 2Critical Care, H. Clínico, Valencia, 3Critical Care, H. Dr. Negrín, Gran Canaria, Spain

Vena A1, Perchiazzi G2, Fiore T2, Giuliani R2, Hedenstierna G1 1 Clinical Physiology, Uppsala University, Uppsala, Sweden, 2Emergency and Transplant, Bari University, Bari, Italy

INTRODUCTION: Impact of different variables on clinical outcome in community-acquired pneumonia has received a great deal of attention over the past decade because of its substantial mortality, morbidity and costs. The objective of the present study was to establish the association between different variables and fatal evolution of severe community-acquired pneumonia (SCAP). METHODS: Multicenter, prospective observational study during a 15 months period in 32 different ICUs in Spain. Inclusion criteria: patients with SCAP. The demographic data, APACHE II score, co-morbidities, aetiologies and complications developed during the ICU stay were evaluated, and their independent association with fatal evolution was studied. Analysis of data was performed using CIA and BMDP statistical package. RESULTS: Five hundred and thirty-three patients were included, and 27,4% died (n=146; Fatal group). Variables associated with fatal evolution (multivariate analysis) were: age (OR 1,7; IC95% 1,3-2,4); immuno-suppression (OR 2,4; IC95% 1,1-5,2), shock (OR 4,9; IC95% 2,7-9,2), acute renal failure (OR 4,2; IC95% 2,4-7,1), ventilator associated pneumonia (OR 3,3; IC95% 1,38,1)and rapid spread of radiographic infiltrates (OR 2,4; IC95% 1,4-4,1). The severity at admission time and aetiology were not associated with fatal outcome. CONCLUSION: The fatal evolution of SCAP is associated with immunosuppression and evolutive complications of illness. However, the severity at admission and the microbiology do not influence the outcome. REFERENCES: 1. Roson B, Carratalá J, Fernández-Sabé N, et al. Causes and factors associated with early failure in hospital patients with community-acquired pneumonia. Arch Intern Med 2004; 164:502-8. 2. Fine MJ, Smith MA, Carson CA, et al. Prognosis and outcome of patients with community-acquired pneumonia. A meta-analysis. JAMA 1996; 276:134-41. Grant acknowledgement: Supported in part by CIRIT SGR 2001/414, Distinció Recerca Universitaria (JR) and Red Respira-ISCIII-RTIC-03/11

INTRODUCTION: Acoustic transmission through the respiratory system has recently been investigated in a number of studies [1,2] that demonstrated the density of pulmonary parenchyma impacts on the propagation and attenuation of normal lung sounds. The aim of the present study is to evaluate whether the computerized analysis of respiratory acoustic signals can identify changes in lung volume induced by application of increasing values of positive end-expiratory pressure (PEEP). METHODS: Five healthy pigs were anesthetized and mechanically ventilated (volume-controlled constant-flow) in the supine position. Respiratory rate, tidal volume and inspiratory flow were kept constant while the level of applied PEEP was increased from 0 to 20 cmH2O in steps of 5 cmH2O in order to obtain different values of lung volume. Lung sounds were recorded over two sites from both sides of the chest (FR, DR, FL and DL; F:frontal, D:dorsal; R:right, L:left) simultaneously with pressure and flow tracings at airway opening. Duration of recording was 60 s and an average number of 12 breaths were collected from each location. Acoustic signals, acquired in digital format (sample rate: 44100 Hz; resolution 16 bit), were filtered (Butterworth band pass filter: 752000 Hz) and downsampled (sample rate: 11025 Hz; resolution: 16 bit). Three consecutive inspirations for each animal were analysed by calculating the sound amplitude in decibels (dB). The increase in end-expiratory lung volume (EELV) related to the PEEP application was also measured during exhalation manoeuvres to functional residual capacity. The correlation between changes in EELV and amplitude of lung sounds was than examined. RESULTS: The amplitude of recorded lung sounds was significantly reduced by the application of a PEEP ≥ 10 cmH2O (p < 0.05). Although the sound amplitude was constantly higher at the upper (FR and FL) than at the lower (DR and DL) sensor locations, the trend of acoustic attenuation was similar for all microphones. The increase in PEEP by 5 cmH2O steps from 0 to 20 cmH2O, reduced the acoustic energy of lung sounds recorded at ZEEP by 0.3 dB, 2 dB, 5 dB and 7 dB respectively, which corresponds to percentage values of 1%, 6%, 14% and 21% in acoustic attenuation. The observed variations in amplitude correlated with changes in lung volume (R=0.87; p<0.05). CONCLUSION: In this study, normal lung sounds were collected in different conditions of lung density (volume) during mechanical ventilation. Computerized analysis of sound amplitude was able to detect increases of lung volume by proportional changes in sound amplitude. REFERENCES: [1] Leung et al. J Appl Physiol. 2000 - [2] Pohlmann et al. Physiol Meas. 2001

S214

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

827

829

ENDOTHELIAL AND COAGULATION DYSFUNCTION IN PORCINE SEPSIS: THE EFFECTS OF SELECTIVE iNOS-INHIBITION

LOCAL TIDAL VARIATION IN FUNCTIONAL ELETRICAL IMPEDANCE TOMOGRAPHY REFLECT THE VENTILATORY CHANGES

Matejovic M1, Krouzecky A1, Martinkova V1, Radej J1, Radermacher P2, Novak I3 ICU, Dept of Internal Medicine I, Charles University Hospital Plzen, Plzen, Czech Republic, 2Sektion APV, Dept of Anesthesia, University Hospital Ulm, Ulm, 3ICU, Dept of Internal Medicine I, Charles University Hospital Plzen, Plzen, Germany

Großherr M1, Meier T1, Gehring H1, Schmucker P1, Leibecke T2, Leonhardt S3 Clinic for Anaesthesiology, 2Institute for Radiology, University Clinic Schleswig-Holstein, Campus Lübeck, Luebeck, 3Institute for Med.Informationstechnique, RWTH, Aachen, Germany

1

INTRODUCTION: In porcine long-term sepsis we recently showed that iNOS inhibiton reduced oxidative stress and improved microcirculatory and metabolic derangements (1). This study evaluated the effects of iNOS inhibition on P. aeruginosa (Pa)-induced endothelial and haemostatic stress. METHODS: 12 h after induction of shock with continuous i.v. Pa 14 pigs received either no drug (CON, n=7) or selective iNOS inhibitor L-NIL (n=7, 1-2 mg/kg/h). Hydroxyethylstarch was infused to maintain hyperdynamic circulation. Before, 12,18 and 24 hrs after the start of Pa plasma levels of markers related to endothelial function (von Willenbrand factor (vWf), hypercoagulability (trombin-antithrombin complexes (TAT), and fibrinolysis inhibition (activity of plasminogen activator inhibitor type 1 (PAI-1 act) were measured. RESULTS: Data are median and interquartile range, P<0.05. * vs. Preshock (RM ANOVA on Ranks);§ L-NIL vs CON (Mann-Whitney Rank Sum Test). Baseline vWf CON mU/gprot 8(7;31) vWF L-NIL mU/gprot 7 (7,8) TAT CON ug/gprot 1(1;3) TAT L-NIL ug/gprot 2(2;5) PAI1 CON U/ml 22(16;23) PAI1 L-NIL U/ml 17(16;22)

12 hours 15(13,48)* 11(9,14)* 3(1;5)* 3(2;4) 37(31;39)* 32(24;34)*

18 hours 22(12;57)* 11(10,13)* 6(3;13)* 3(2;5) 39(38;39)* 33(3038)*§

24 hours 25(19;99)* 13(1114)*§ 6(5;14)* 6(4;7) 37(32;39)* 31(26;37)*

CONCLUSION: Live bacteria-induced sepsis resulted in endothelial activation associated with a procoagulant state. Selective iNOS inhibition beneficially attenuated the detrimental consequences of sepsis on endothelial and haemostatic dysfunction. Suppression of oxidative stress might contribute to these results.

1

INTRODUCTION: This study should investigate wether functional eletrical impedance tomography could be used as an online monitoring device for ventilation after acute lung injury METHODS: After approval of the regional committee for protection of animals against cruelty 6 domestic pigs received premedication and anaesthesia. Propofol and rocuronium were applied for induction and maintenance of anaesthesia. After tracheotomy and intubation lungs were recruited (peak airway pressure 30 mbar for 10 seconds) and ventilated with different positive end expiratory pressures (PEEP)( 0, 10, 20, 10, 0 mbar) for 10 minutes. At the end of each ventilation sequence local tidal variation was determined in 4 different zones by functional electrical impedance tomography. Theses zones had the same width and divided the thorax from ventral (zone 1) to dorsal (zone 4). At the same time Hounsfield Units were measured for the ventral and the dorsal half. These measurements were performed before and after an induction of an acute lung injury (ALI). ALI was generated by a depletion of surfactant by a lavage with warm saline solution 6 times for 45 seconds. RESULTS: After an acute lung injury increased X-ray density of the dorsal lung is associated with a decrease of tidal variation. Results of animal #5 are presented in figure 1 and 2.

Grant acknowledgement: Supported by a research grant IGA MZ CR: ND6837-3/2001

CONCLUSION: The changes of local tidal variation reflect the radiological changes in ventilation with different PEEP levels after acute lung injury. Ventilation with 10 or 20 mmHg PEEP ventilation corrects the lung injury associated decrease of tidal variation to its reference again.

828

830

GASTRIC TONOMETRY IS NOT USEFUL IN PREDICTING WEANING OUTCOME FROM MECHANICAL VENTILATION

RELIABILITY OF PULSE OXIMETRY IN INTENSIVE CARE SETTINGS AND AFFECTING FACTORS

Forel J1, Presutti M1, Roch A1, Arnal J1, Demory D1, Loundou A2, Papazian L1, Gainnier M1. 1Medical Intensive Care Unit, Sainte Marguerite Hospital, 2Laboratory of Statistics, Faculty of Medicine, Marseille, France

Guven M1, Sungur M1 1 Intensive Care, Erciyes University, Medical Faculty, Kayseri, Turkey

REFERENCE: 1.Matejovic et al. Shock: 2004; 5:458-465

INTRODUCTION: Difficult weaning from mechanical ventilation (MV) is currently encountered in Medical Intensive Care Unit (MICU). Accurate predictors of weaning outcome should decrease MV duration while limiting re-intubation risks. An association between weaning failure and gastric mucosal hypoperfusion has been reported but remains discussed.1,2 The aim of this study was to test accuracy of tonometry parameters in predicting weaning outcome. METHODS: 47 patients mechanically ventilated in a MICU were included in a prospective study after written informed consent. Weaning procedure was a 2-hour T-tube trial (TTT). Weaning success was pronounced after 48h of spontaneous breathing. Gas tonometry (Tonocap, DatexOhmeda, Finland) measured gastric intramucosal PCO2 (PCO2im), PCO2gap (intra-mucosal PCO2 - arterial PCO2), intramucosal pH (pHim) just before weaning (T0), at 20 min, 40 min, and at the end of TTT (Tend). Their Area Under Receiver Operating Characteristics Curves (AUC), operational characteristics and Likelihood Ratios (LR) were analysed before and during TTT. RESULTS: Weaning failure was observed in 40% in the overall population, and in 65% of COPD patients. No significant difference in PCO2gap was found between weaning success and failure groups. However, PCO2gap increased significantly over time in weaning failure group (p =0.003 for time effect, RM ANOVA). All tonometric parameters showed an AUC lower than 0.68 and a LR for a Positive Test (LR +) lower than 1.96 whatever the time measurement (Table 1). The best PCO2gap sensitivity, specificity, AUC and LR + appeared at the end of TTT. Time Threshold

pHim

T0 <7.32 Tend <7.32 >43.5 PCO2im, T0 mmHg Tend >51 >17.5 PCO2gap, T0 mmHg Tend >18

Sensitivity (95% CI)

Specificity (95% CI)

Likelihood Ratio + (95% CI)

42 (20-64) 63 (41-85) 58 (36-80) 63 (41-85) 42 (20-64) 63 (41-85)

50 (31-68) 61 (43-79) 32 (15-49) 68 (51-85) 54 (35-72) 61 (43-79)

0.84 (0.44-1.6) 1.61 (0.9-2.85) 0.85 (0.54-1.35) 1.96 (1.04-3.72) 0.91 (0.47-1.76) 1.61 (0.9-2.85)

Area Under ROC Curve (95% CI) 0.38 (0.21-0.56) 0.63 (0.47-0.79) 0.46 (0.29-0.63) 0.68 (0.51-0.84) 0.41 (0.24-0.57) 0.66 (0.5-0.82)

CONCLUSION: PCO2gap and others tonometric parameters (PCO2im, pHim), measured before and during weaning trial, presented insufficient AUC and likelihood ratio to predict weaning failure. Gastric tonometry is not useful in predicting weaning from MV outcome in case-mix medical patients. REFERENCES: 1. Hurtado F, Beron M, Olivera W, et al : Gastric intramucosal pH and intraluminal PCO2 during weaning from mechanical ventilation. Crit Care Med 2001; 29: 70-6. 2. Maldonado A, Bauer T, Ferrer M, et al : Capnometric recirculation gas tonometry and weaning from mechanical ventilation. Am J Respir Crit Care Med 2000; 161: 171-6.

INTRODUCTION: Reliability of pulse oximetry (SpO2) is controversial. We wanted to determine the reliability of SpO2 in intensive care unit (ICU). METHODS: This observational study was undertaken in the ICU of Erciyes University. No arterial blood samples (ABG) were taken specifically for the study if not indicated by the clinical condition. All patients who had an arterial line for pressure monitorization, and who were being monitored by SpO2 were included in the study. Patients who had jaundice or peripheral arterial diseases were excluded. ABG samples were analyzed within 2 minutes by using a hematoxymeter. SpO2 readings, blood pressures, need for dopamine and Hb concentrations were obtained simultaneously with ABG. Four different pulse oximeters were used (Novametrix CO2SMO Plus,; Nihon Kohden Life Scope 9; Hewlett-Packard Viridia 24C SpO2/Pleth; Siemens SC9000). The agreement between SaO2 and SpO2 was assessed by Bland-Altman plot. ANCOVA test was used to compare biases in subgroups. ROC was used to find cut-off value to detect hypoxemia. RESULTS: A total of 1000 data were obtained from 115 patients. SpO2 value was 93.0±3.2% and SaO2 was 95.1±3.6%. SpO2 have a bias of 2.3±3.7(NS). Types of oximeter, pH, Hb, dopamine and mean arterial pressure (MAP) were independent factors. The results of Siemens were more reliable than others. SpO2 (93.8±2.8) and SaO2 (94.5±2.8) in conditions including low pH (<7.45), high Hb (>10) and MAP (>80 mmHg) (bias%±SD; 0.7±2.5; p=0.054)were not different. However, agreements between SpO2 (97.2±1.0) and SaO2 (94.1±2.8) were low in conditions including high pH (7.45), low Hb (<10) and MAP (<80 mmHg) (bias%±SD; 3.3±3.5; p<0.001). To detect a SaO2<90%, we found that SpO2 of 90% had sensitivity of 60.4% (CI 95%;50.2-70.0) and specifity of 95 % (CI 95%;93.4-96.3). To detect severe hypoxemia (PO2<60 torr), we found that SpO2 of 90% had sensitivity of 79.7% (CI 95%;69.2-88.0) and specifity of 95.3% (CI 95%;93.8-96.6). Area under curves of SpO2 (0.92) and SaO2 (0.98) were significantly different. CONCLUSION: SpO2 may have some biases. The range of bias is wide. False or over diagnosis of hypoxemia is possible.SpO2 has some important limitations such as hypotension, using dopamine, anaemia, acid-base disorders and type of oximeter. Under normal condition,SpO2 is more reliable. But, the results must be re-evaluated in patients who have impaired parameters. REFERENCE: 1-Intensive Care Med 2001;27:1606.3- Crit Care Med 2000;28:703.

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

Oral Presentations Perioperative cardiothoracic IV – 831-836 831 WASHING RETRANSFUSED MEDIASTINAL BLOOD DURING OPCAB SURGERY REDUCES PLASMA AND URINARY TNFSR2 Allen S J1, Armstrong M A2, Penugonda S P1, Campalani G3, Phillips A S1, McMurray T J4, McBride W T4 1 Department of Clinical Anaesthesia, Royal Victoria Hospital, 2Department of Immunobiology, Queens University Belfast, 3Department of Cardiac Surgery, Royal Victoria Hospital, 4Department of Anaesthesia, Queens University Belfast, Belfast, United Kingdom INTRODUCTION: It is thought that the plasma and urinary anti-inflammatory tumour necrosis factor soluble receptor-2 (TNFsr-2) response at coronary artery bypass graft surgery (CABG), both with cardiopulmonary bypass (CPB)and without it(OPCAB), is in response to, and thus an early indication of, significant plasma TNF-alpha elevations.1 We wondered if the practice of retransfusion of washed mediastinal shed blood in OPCAB patients would be associated with lower plasma and urinary TNFsr-2 than patients retransfused with unwashed mediastinal shed blood. METHODS: Thirty-six ASA grade 3-4 patients undergoing OPCAB were randomly allocated to one of two groups. Group 1 had mediastinal blood retransfused after completion of revascularisation. In Group 2, shed mediastinal blood was either discarded (if less than 500ml) or if greater than 500ml, was retransfused following washing in a Dideco Compact cell saver. Anaesthesia was induced with fentanyl 20-25 \mumg kg-1, pancuronium 100-150 \mug kg-1 and maintained using target controlled intravenous anaesthesia with propofol and fentanyl increments. Plasma and urine samples were obtained as follows: baseline (sample A), 10 min after completing revascularisation (sample B), 2, 6, 24 and 48 hours after revascularisation (samples C,D,E, F ). Analysis was with Wilcoxon Signed Rank or Mann Whitney U test as appropriate as well as between group area under the concentration time curve (T-test). RESULTS: Plasma TNF-alpha increased from baseline and was significantly higher in the unwashed group at time C (p < 0.05). In both groups plasma and urinary TNFsr 2 increased significantly as compared with baseline and in the washed group this was lower at time E in urine (p < 0.05. Washed: 15031+/- 9068 pg/ml. Unwashed: 19810+/- 4582 pg/ml. Mean+/- SD). Moreover, the area under the TNFsr-2 concentration time curve was lower in the washed group for both plasma (p < 0.05) and urine (p < 0.01). CONCLUSION: Since higher anti-inflammatory TNFsr-2 reflects a larger underlying proinflammatory TNF-alpha, it follows that the higher postoperative plasma and urinary TNFsr-2 in the unwashed group suggests that retransfusion of unwashed mediastinal shed blood contributes significantly to the postoperative pro-inflammatory response. REFERENCE: 1. Baker R, Allen S, Armstrong MA, McBride WT. Editorial. Role of the kidney in perioperative inflammatory responses. Br J Anaesth 2002;88(3):330-334. Grant acknowledgement: Supported by The Heart Fund Trust and the Royal Victoria Hospital Research Fellowship Fund.

S215

833 THE USE OF EPINEPHRINE AND ICU-ASSOCIATED TRAUMATIC MEMORIES IN PATIENTS AFTER CARDIAC SURGERY Schmoelz M1, Richter M1, Nollert G2, Schelling G1 1 Department of Anaesthesiology, 2Department of Cardiac Surgery, Ludwig Maximilians University, Munich, Germany INTRODUCTION: Patients undergoing cardiac surgery (CS) often require high doses of exogenously administered epinephrine. There is extensive evidence from experiments in animals, that stress hormones like epinephrine can cause consolidation of traumatic memory by directly activating amygdala-modulated brain memory networks. Stress hormone induced consolidation of emotional memory could therefore result in an increased incidence of traumatic memories in critically ill patients despite sedation. In addition, traumatic memories in patients after ICUtherapy can be associated with the development of chronic stress states like post-traumatic stress disorder (PTSD) and a decline in health-related quality of life (HRQL). We performed a prospective cohort study in CS-patients to test the hypothesis that the exposure to epinephrine in the ICU can result in traumatic memory formation, the consecutive development of PTSD and impairments in HRQL-outcomes. METHODS: 148 CS-patients completed a validated questionnaire evaluating traumatic memories and PTSD at 3 time points: at 1 day before CS and at 1 week and at 6 months after discharge from the ICU. Traumatic memories were defined as the recall of pain, respiratory distress, feelings of anxiety/panic or nightmares. HRQL was measured at 1 day and at 6 months after discharge from the ICU using the SF-36 instrument. RESULTS: The number of traumatic memories from the ICU correlated significantly with the administered total dosages of epinephrine (p<0.01) whereas the administered total dosage of propofol had no effect (p=0.38). PTSD scores correlated significantly with the number of traumatic memories from the ICU. HRQL-outcome scores at 6 months after CS were negatively associated with the number of traumatic memories and PTSD scores. Patients with newly developed PTSD at 6 months after CS (n=20) had a significantly higher number of traumatic memories from the cardiovascular ICU (p=0.02). In addition, these patients did not show improvements in HRQL after CS. In contrast to patients without evidence of PTSD, patients with new PTSD reported no gain in Physical Function HRQL scores and their Mental Health Summary Score declined significantly as compared to their preoperative status. CONCLUSION: Epinephrine given to critically ill patients can cause consolidation of emotional memory despite sedation with propofol. Because a higher number of traumatic memories were associated with a higher incidence of PTSD in our study, the facilitating effect of exogenously administered stress hormones on the formation of traumatic memories could indirectly influence the incidence and intensity of chronic stress reactions. Furthermore, our study has identified PTSD and traumatic memories from the ICU as significant determinants of HRQL-outcomes after CS.

832

834

IL-10 AND TNF-BETA GENE POLYMORPHISM AND LACTATE LEVELS AFTER CARDIAC SURGERY

CATECHOLAMINES AND INOTROPES CAN MODULATE THE EFFECTIVENESS OF VOLUME EXPANDERS

Riha H1, Hubacek J2, Poledne R2, Kellovsky P1, Brezina A1, Pirk J3 Department of Anesthesiology and Intensive Care, 2Center for Experimental Medicine, 3Department of Cardiovascular Surgery, Institute for Clinical and Experimental Medicine, Prague, Czech Republic

1

INTRODUCTION: Lactate levels after cardiac surgery are influenced by different proinflammatory (TNF, IL-6, IL-8) and antiinflammatory (IL-10) cytokines. The goal of this study was to determine relationship between polymorphism in the IL-10 (G–1082A) and TNF-beta (G–308A) gene and lactate levels in patients after cardiac surgery. METHODS: We studied 169 consecutive adult patients aged < 74 years with left ventricle ejection fraction > 40% undergoing elective coronary artery bypass grafting with hypothermic (33 ºC) cardiopulmonary bypass (CPB) and crystalloid cardioplegia. Patients with insulin dependent diabetes mellitus, history of hepatic disease and inotropic support with epinephrine were excluded. Induction and maintenance of anesthesia were standardized in all patients. Lactate levels were documented at five different time points: 10 min after beginning of CPB, 40 min after CPB termination, 30 min and 8 hrs after arrival in ICU, and 16 hrs after surgery. Genetic analysis was performed by mismatched polymerase chain reaction (PCR) and restriction analysis. The results were evaluated by ANOVA with repeated measures. RESULTS: No association was found between single polymorphism in IL-10 or TNF-beta gene and lactate levels. Nevertheless, the carriers of IL-10/TNF-beta gene haplotype +A/GG had significantly different course of lactate levels in time with decrease in lactate (in comparison with increase in other groups with different gene haplotypes) at 8 hours after arrival in ICU (mean ± SD: 1.72 ± 1.20 vs 2.44 ± 1.42, 2.09 ± 1.50, 1.73 ± 0.94 mmol/l, respectively; p < 0.05).

CONCLUSION: IL-10 (G–1082A) and TNF-beta (G–308A) gene polymorphism influences lactate levels and their course in time after cardiac surgery.

Kramer G C1, Prough D S1, Vane L A1, Connolly C M1, Kinsky M P1 Anesthesiology, University of Texas Medical Branch, Galveston, United States

1

INTRODUCTION: Intensivists and anaesthesiologists administer fluid therapy to nearly all their patients and anaesthetics, inotropes and vasoactive drugs to many patients. The primary goal of fluid therapy is vascular volume expansion with associated augmentation and maintenance of venous return, cardiac output and oxygen delivery. The cardiovascular effects of anaesthetics, inotropes and vasoactive drugs can augment or depress venous return. However, new data show that anaesthetics and catecholamines can modulate volume expansion during fluid therapy and can alter partitioning of fluid between the blood and tissue spaces. METHODS: We used chronically instrumented, normovolemic splenectomized sheep to define the time course of plasma volume expansion after a 24 mL/kg 20-min bolus of normal saline under conditions of consciousness, isoflurane anaesthesia and during infusion of adrenergic agonists of alpha, beta and dopa receptors (phenylephrine (3 mug * kg-1 * min-1), isoproterenol (0.1 mug * kg-1 * min-1) and dopamine (50 mug * kg-1 * min-1)), respectively and the beta antagonist, esmolol (50-100 mug * kg-1 * min-1). Plasma volume is measured at baseline with indocyanine green dye dilution, while the time course of volume expansion is calculated from the dilution of haemoglobin. RESULTS: Bolus infusion in conscious sheep expanded plasma volume 13.0 ± 2.7 mL/kg by the end of the bolus. Isoflurane resulted in an antidiuresis and greatly increased extravascular sequestration of fluid with only a minimal effect on plasma volume expansion compared with conscious sheep. Isoproterenol augmented peak volume expansion, 50%, while dopamine had little effect on peak expansion and phenylephrine reduced volume expansion by 32%. Two hours after the 0.9% NaCl bolus, sustained plasma volume expansion was greatest in the isoproterenol protocol (12.2 mL/kg), while the dopamine protocol (6.8 mL/kg) remained similar to the control protocol (4.1 mL/kg) and the phenylephrine protocol had a net loss of plasma volume (-1.9 mL/kg. Some of the volume expansion effect of isoproterenol is caused by an antidiuresis, while dopamine augmented plasma expansion and caused a diuresis. Phenylephrine reduced volume expansion, and caused a diuresis. As opposed to the augmented volume expansion with the Betaagonist isoproterenol, the esmolol reduced peak volume expansion by 62% compared to the control group with plasma volume returning to baseline 10-min after the bolus. Mechanisms may involve alterations in capillary permeability or lymphatic pumping. CONCLUSION: Catecholamines profoundly modulated the effectiveness of volume expansion of a bolus of crystalloid. Strategies to optimize plasma volume expansion and cardiac output may develop from better defining interactions between volume expanders, inotropes and vasoactive drugs.

S216

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

835 VOLUME MANAGEMENT BY GOAL DIRECTED THERAPY REDUCES CATECHOLAMINE-NEED IN CARDIOSURGERY PATIENTS Goepfert M S G1, Reuter D A1, Akyol D1, Kilger E1, Goetz A E1 1 Anaesthesiology, University of Munich, Munich, Germany INTRODUCTION: Central venous pressure (CVP) and pulmonary artery occlusion pressure being considered as surrogates for cardiac preload have been shown to mislead clinicians in guiding volume therapy in critcally ill patients [1]. Recent advances using transcardiopulmonary thermodilution allow the measurement of the global end-diastolic volume index (GEDVI) which comprises the volume of blood contained in the heart. GEDVI has shown to be an accurate parameter to monitor changes of cardiac preload in patients following cardiac surgery and septic patients [2,3]. Aim of this study was to guide perioperative volume therapy by a treatment algorithm based on measurement of GEDVI and assessment of extravascular lung water (EVLW) by transcardiopulmonary thermodilution in combination with on-line cardiac output by pulse contour analysis. Endpoint was the need for catecholamines in patients during and after coronary artery bypass grafting. METHODS: 40 patients undergoing elective coronary artery bypass grafting were studied prospectively and matched for relevant parameters as ejection fraction and surgical procedure with 40 historical controls. Study patients and controls did not differ regarding age (63.5 ± 7.4 vs. 64.6 ± 6.2 yrs), body mass index (25.8 ± 2.4 vs. 26.1 ± 2.2 kg m-2), pre-operative ejection fraction (0.62 ± 0.16 vs. 0.59 ± 0.13). In the study group the haemodynamic goals were a GEDVI >640 ml and EVLW <10 ml/kg BW, a cardiac index (CI) >2.5 l min-1 m-2 and a mean arterial pressure >70 mmHg. Prior to every catecholamine and vasopressor support the physician had to control and to maintain the algorithm given thresholds. The need for volume and the catecholamine/vasopressor support was quantified. Interleukin-6 (IL-6) blood levels were measured 6 hrs after surgery. RESULTS: Duration of ECC (91 ± 18 vs. 94 ± 24 min), aortic cross clamping time (58 ± 12 vs. 60 ± 14 min), temperature at the end of ECC (36.8 ± 0.54 vs. 36.8 ± 0.72°C) and total time of surgery (228 ± 34 vs. 227 ± 47 min) did not differ between groups. Volume bilance from ICUadmission till 48 hrs after surgery showed a higher volume of colloids in the study group (6917 ± 1159 vs. 4853 ± 1512 ml; p<0.05). The total amount of norepinephrine at 48 hrs. on ICU was significantly lower in the study group (0.8 ± 2.08 vs. 6.62 ± 7.56 mg; p<0.01) as it was with epinephrine (0.01 ± 0.7 vs. 0.87 ± 1.69 mg; p<0.01). Levels of blood IL-6 differed significantly between both groups (315 ± 163 pg ml-1 vs. 466 ± 403 pg ml-1). CONCLUSION: Guiding perioperative volume therapy by the goals GEDVI >640ml m-2 with an EVLWI <10ml m-2 during cardiac surgery and in postoperative ICU treatment in order to achieve adequate perfusion pressure and cardiac output resulted in a reduced need for catecholamines and lower postoperative IL-6 blood levels.

Oral Presentations Management of glycaemia – 837-842 837 MANAGEMENT OF BLOOD GLUCOSE IN THE CRITICALLY ILL IN AUSTRALIA AND NEW ZEALAND Mitchell I A1, Finfer S R2, Bellomo R3, Higlett T4 Intensive Care, The Canberra Hospital, Garran, 2Intensive Therapy Unit, Royal Northshore Hospital, St Leonard’s, 3Intensive Care, Austin and Repatriation Hospital, Heidelberg, 4Clinical Trials Group, ANZICS, Carlton, Australia 1

INTRODUCTION: Hyperglycaemia is a common finding in acutely ill patients and is associated with adverse outcomes. A recent single centre study of ventilated surgical patients reported that maintaining normoglycaemia with intensive insulin therapy (IIT) reduced morbidity and mortality .The effectiveness of IIT in a more general population of intensive care (ICU) patients is not known. As a prelude to a multi-centre trial in a general ICU population we conducted a practice survey and an inception cohort study of glycaemic control in Australian and New Zealand ICUs. METHODS: 29 of 46 ICUs affiliated to the Australian and New Zealand Intensive Care Society Clinical Trials Group accepted the invitation to take part in the study. ICU directors were surveyed on their awareness and adoption of IIT. Participating units collected data on all consecutive admissions during a two week period commencing between 1st March and 31st August. Data collected during the ICU stay included patient demographics, severity of illness, reason for admission, dextrose administration and blood glucose (BG) management. Patient outcome was collected at ICU and hospital discharge. RESULTS: All ICU directors reported being aware of the paper reporting improved outcome with IIT. 12% of ICU directors reported their ICU used IIT for all patients, 27% for some patients and 61% not at all. A total of 939 patients were studied over 3790 intensive care patient days. The patient population was predominantly male, older than 60 years, overweight and only half were admitted immediately after surgery. 165 patients (17.6%) patients had a history of diabetes and of those 46 (27.9%) were taking insulin. A target range for BG was documented for 32% of patient days; the range was consistent with IIT (4.4 – 6.1 mmol/L) in 3.6% of patient days. The commonest target range for BG was 6-10mmol/L. The median (IQR) highest BG during ICU stay was 9.9 mmol/L (7.8, 12.6) and the median amount of intravenous dextrose administered during the first 24 hours of ICU admission was 1.1g (0, 30). 287 patients (31.1%) were given iv insulin to control the BG during their ICU stay. The median BG concentration that triggered administration of intravenous insulin, was 11.9 (9.4, 14) mmol/L with no differences between ICUs.

REFERENCES: [1]Chest 2002 Jun;121(6):2000-8 - [2]J Cardiothorac Vasc Anesth 2002 Apr;16(2):191-5 - [3]Chest 2003 Nov;124(5):1900-8

CONCLUSION: Few ICUs have adopted IIT. At the time of this study, control of BG in Australian ICUs closely resembles that of the control group in the recent study. A randomised controlled study is ethically justified and necessary to determine the role of IIT in future Australian ICU practice.

836

838

CARDIAC SURGERY IN 50 JEHOVA´S PATIENTS: DIFFERENT TRANSFUSION CRITERIA

INTENSIVE INSULIN THERAPY IN THE CRITICALLY ILL: AN OBSERVATIONAL STUDY

Krapivski A1, Pizanis N1, Marggraf G1, Thielmann M1, Knipp S1, Massoudy P1, Jakob H1, Schulte-Herbrueggen J1, Piotrowsky J1 1 Department of Thoracic and Cardiovascular Surgery, West German Heart Center, University of Essen, Essen, Germany

Vancollie O1, Dive A1, Jamart J2, Pietquin J1, Brochier S1, Domecki S1, Donckier J3, Installé E1 1 Intensive care, 2Biostatistics, 3Endocrinology, Mont Godinne University Hospital, yvoir, Belgium

INTRODUCTION: Major surgical procedures are inevitably associated with blood loss, and refusal of blood transfusions is supposed to imply a substantial risk of complications and adverse clinical outcome. We analyzed the data of Jehovah´s Witness patients who underwent cardiac surgery at our institution. METHODS: From 1/2000 through 1/2004, records of all Jehova’s Witness patients (group 1) who had undergone cardiac surgery with CPB were studied. All had a haemoglobin above 13.5 g/dl. The data of the Jehova’s patients were compared to 100 patients with no restriction to blood transfusion (group 2). RESULTS: 50 consecutive Jehova’s Witness could be analyzed (31 male, 19 female, range 9 to 82 years). Procedures included first-time CABG (n=35), re-do CABG (n=4) and valve surgery with and without CABG (n=11). In both groups there was no intraoperative death. Lowest haematocrit and lowest haemoglobin concentration during CPB were similar in group 1 vs. group 2 (24.77±2.97 vs. 24,41±2,7; 8,16±1,0 vs. 7,96±0,9). Patients in group 2 received 1.6±1.04 units of red blood cells and 1.21±1.7 units of fresh frozen plasma. There was no difference with regard to the duration of ICU stay, time of mechanical ventilation, in-hospital stay, and major postoperative complications. Control patients had a greater fluid loss via chest drainage than Jehova´s patients (635,41±372,9 ml vs. 382,5±230,39 ml, p < 0,01). CONCLUSION: Our results show that different transfusion criteria were applied to patients refusing blood products and to patients accepting blood products. It may thus be argued that in non Jehova’s Witness patients the treatment with blood products is handled to liberally.

INTRODUCTION: In this study, we assessed the quality of glucose control resulting from implementation of an intensive insulin therapy (IIT) protocol in our ICU, as well as the main factors responsible for poor glucose control in individual patients. METHODS: IIT was applied in 50 consecutive patients. Quality of glucose control was identified throughout the iv insulin infusion period by the time (hrs) spent inside normoglycemic range (80110 mg/dl) and by calculation of area under curve (AUC) of glucose concentration over time outside normoglycemic range. AUC was calculated using the trapezoidal method. The possible impact of medications, haemodialysis, respiratory support, preexisting diabetes, and artificial nutrition on poor glucose control was assessed by repeated measures analysis using generalized estimating equations (multivariate analysis). RESULTS: 50 consecutive patients (age: 66 ± 12 y., body mass index: 28.8 ± 7.8, diabetes: n = 19, M/F: 34/16, Medical/Surgical: 33/17, Apache II score: 20 ± 8, length of stay: 10 ± 9 days) were administered continuous insulin infusion during a total of 335 days. 48 episodes of transient hypoglycaemia (glucose£ 60 mg/dl) were detected in 24 pts, with no sequel. Mean glucose level, daily time spent within the normoglycemic range and daily AUC outside normoglycaemic range were respectively: 114 ± 26 mg/dl, 11.3 ± 6.1 hrs, and 285 ± 350 mg.hr/dl. As shown by multivariate statistical analysis (table), infusion of catecholamines, preexisting diabetes and artificial nutrition were associated with poor glucose control. Initiation/interruption of artificial nutrition, intubation/extubation, and patient’s transfers outside the Unit were frequently responsible for transient hyperglycemia. AUC>110 p value catechol 0.011 Artificial nutrition <0.001 diabetes 0.026

AUC>150 p value 0.06 0.015 0.01

AUC<80 p value 0.017 ns 0.095

AUC<60 p value 0.099 ns 0.044

CONCLUSION: IIT implementation was safe. Identification of specific conditions associated with poor glucose control should enable to improve future ITT algorithms in ICU patients.

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

S217

839

841

TARGETTING AND MAINTAINING A TIGHT BLOOD GLUCOSE RANGE IN THE CRITICALLY ILL: FEASIBLE OR NOT?

RELIABILITY OF SIMPLE INDICES OF INSULIN RESISTANCE IN CRITICALLY ILL MEDICAL PATIENTS

Oeyen S1, Poelaert J2, Vandewoude K3, Decruyenaere J1 1 SICU, 2CSICU, 3MICU, Ghent University Hospital, Gent, Belgium INTRODUCTION: The Van Den Berghe group developed an insulin protocol (IP) to target and maintain glycaemia (G) between 4.5-6.1 mmol/L (normal range). We aimed to evaluate the compliance with the IP in our surgical (SICU) and medical ICU (MICU) and thereby the efficacy and safety of this protocol. We also tried to identify causes of poor G control. METHODS: All ICU-nurses got a 1 hour lesson and a 4 weeks initiation period on the IP before evaluation of the protocol was started. All patients with an arterial line and an expected ICU-stay (ILOS) > 48 h were included. Nurses measured G on undiluted arterial blood samples using a bedside glucometer (GlucoTouch, LifeScan, Benelux). The according rate of the continuous insulin infusion (Humulin Regular, Lilly) and the frequency of G measurements were also exclusively managed by ICU-nurses. A correct compliance with the IP was seen as a right insulin dose adjustment according to the protocol and the individual condition of the patient. Efficacy was established by comparing the number of G values within the correct range to the total number of G measurements. Protocol safety was assessed by the number of events where G was < 3.3 mmol/L, defined as hypoglycaemia. To identify factors which could influence G control we defined 2 groups adjusted to the median percentage of correct G range: (1) G values ≤ 39% and (2) > 39% in the normal range. Data are presented as mean ± SD and as percentage. Statistical analysis was performed using an independent samples T-test. Significance was accepted when p<.05. RESULTS: Out of 6016 G measurements, (30 patients - APACHE II 26±8) a compliance with the IP in 70.9% (N=4267) was found. Only 41.6% (N=2500) of all G values were regulated in the desired range despite hourly or two-hourly G measurements (17 ± 7 per day). Hypoglycaemia occurred in 1,8% (N=107). Group 1 showed significant higher G values [G>8.4 mmol/L 26±31 vs 17±17; p<.05 and G>13.9 mmol/L 1.5±1.5 vs 0.7±0.9; p<.006]. In group 2 there were significant more G measurements <2.2 mmol/L [0.1±0.4 vs 0.5±0.9; p<.01]. The 2 defined groups were comparable concerning APACHE II score [25±11 vs 26±8], number of diabetic patients [5 vs 2], ILOS [16±11 vs 21±15], compliance with the IP [66.5% vs 74.1%] and number of G measurements per day (both 17±7).

Holzinger U1, Zauner A1, Ratheiser K1, Funk G2, Madl C1, Zauner C1 Internal Medicine IV, ICU 13H1, 2Internal Medicine IV, Department of Pulmonary Medicine, Medical University Vienna, Vienna, Austria

1

INTRODUCTION: Peripheral insulin resistance (IR) – mainly due to post-aggression metabolism – is frequently observed in critical illness. IR and enhanced gluconeogenesis are responsible for hyperglycaemia, which has been associated with an adverse clinical outcome in critically ill patients. The gold standard for quantification IR is the expensive and time-consuming euglycemic-hyperinsulinemic clamp. Alternatives to the clamp comprise the Quantitative Insulin Sensitivity Check Index (QUICKI) and the Homeostasis Model Assessment (HOMA), which are calculated indices of IR. They are much cheaper and easier to apply, however their validity in critical illness is unclear. The aim of this study was to assess these indices of IR for their reliability in critically ill patients. METHODS: Severity of IR (as measured by the M-value) was determined by a euglycemichyperinsulinemic clamp in 30 critically ill patients on the day after admission and after an overnight fast. Additionally, QUICKI and HOMA were calculated using basal blood glucose concentration and blood insulin concentration. Correlations between the measured and the calculated insulin resistance were assessed by means of Spearman’s correlation coefficient (rs). RESULTS: M-value, QUICKI and HOMA were 2.07 ± 0.64, 0.319 ± 0.04 and 6.38 ± 6.43 mg?kg-1?min-1, respectively. M-value correlated neither with QUICKI (rs = 0.133, ns) nor with HOMA (rs = 0.133, ns). CONCLUSION: Both methods of calculating IR detected the increased IR, however failed to reliably quantify it. Thus, estimation of IR by QUICKI and HOMA appears to be unreliable in critical illness.

CONCLUSION: In severely ill patients a good compliance with the IP does not imply a high percentage of G values within the target range. Even with well-trained ICU-nurses the IP is rather difficult to perform. REFERENCE: Van Den Berghe et al: Intensive insulin therapy in critically ill patients. NEJM 2001, 345: 1359-1367

840

842

SEVERE HYPOGLYCAEMIA COMPLICATING SEPTIC SHOCK AND INSULIN THERAPY

INCIDENCE OF HYPERGLYCEMIA IN PAEDIATRIC INTENSIVE CARE

Pieters R C1, Hoogenberg K2, De Wit H3, Nijsten M W N4, Zwaveling J4 1 Department of Thoracic ICU, Groningen University Hospital, 2Department of Internal Medicine, Martini Hospital, 3Department of Clinical Chemistry, 4Department of Surgery, Groningen University Hospital, Groningen, Netherlands

1

INTRODUCTION: Sepsis associated severe hypoglycaemia has been reported (1), yet the incidence of severe hypoglycaemia is unknown. Since normoglycaemia is now more aggressively pursued in ICU patients (2), the incidence of hypoglycaemia may have increased. We investigated the determinants and outcome of severe hypoglycaemia in hospitalized patients with special regard to ICU patients. METHODS: From October 1998 to October 2003, all patients treated in our university hospital were screened for all glucose determinations. Severe hypoglycaemia was defined as a repeated blood glucose level <=0.5 mmol/L. As outcome measure, hospital mortality was used. RESULTS: We found 29 patients with severe hypoglycaemia. 8 of these patients were not at the ICU; 6 of them survived. Of the 21 patients with severe hypoglycaemia at the ICU, only 4 were known to have diabetes mellitus. 14 patients had severe hypoglycaemia and septic shock. These patients had high APACHE-II scores, and nearly all of them suffered from severe lactic acidosis, impaired renal and impaired hepatic function. Many patients were not receiving adequate enteral or parenteral feeding. Mortality was 100% in these 14 patients, with a time to death of only hours from the moment of glucose sampling. 7 patients developed severe hypoglycaemia at ICU without having septic shock, 3 of whom died. Of these 7 patients, only 2 had known diabetes. 4 of the 5 patients without diabetes were receiving insulin treatment. With regard to the implementation of stricter glucose regulation since November 2001 (2), we observed a remarkable recent increase (p<0.05) in the incidence of insulin treatment as a determinant of severe hypoglycaemia in ICU patients. CONCLUSION: Although the mechanism leading to hypoglycaemia is probably multifactorial in origin, hypoglycaemia complicating septic shock has a strong association with combined liver and renal failure. The short-term prognosis of severe hypoglycaemia in septic shock that we observed is very poor. There has been an alarming increase in insulin-associated severe hypoglycaemia particularly in non-diabetics - since our ICUs apply intensive insulin therapy for tight control of blood glucose levels. We think our results underscore the importance of close monitoring of glucose levels, especially if intensive glucose control is desired. Concentrated glucose infusion should be given to all ICU patients at risk for hypoglycaemia. REFERENCES: 1. Romijn JA, Sauerwein HP. Hypoglycaemia induced by septic shock. Neth J Med 1988; 33: 68-73. 2. van den Berghe G, Wouters P, Weekers F, Verwaest C, Bruyninckx F, Schetz M, Vlasselaers D, Ferdinande P, Lauwers P, Bouillon R. Intensive Insulin therapy in the critically ill patients. NEJM 2001; 345: 1359-67.

Namachivayam P1, Baines P1 Paediatric Intensive Care Unit, Alderhey Childrens Hospital NHS Trust, Liverpool, United Kingdom INTRODUCTION: Studies in adults have shown that the use of exogenous insulin to maintain blood glucose at a level no higher than 6.1 mmol/l (110 mg/dl) reduced morbidity and mortality among critically ill patients in the surgical intensive care unit1. Even a modest degree of hyperglycemia occurring after intensive care unit admission increased hospital mortality in a heterogeneous population of critically ill patients2. In-hospital hyperglycaemia is a common finding and represents an important marker of poor clinical outcome and mortality in patients with and without a history of diabetes3. METHODS: Retrospective audit of biochemistry reports. The highest Blood Glucose during PICU stay for all patients was collected over a 6-month period from Nov 2002 to April 2003

RESULTS: A total of 544 patients were admitted. Results were available for 513 patients (94%). The incidence of hyperglycaemia (blood glucose≥6.1mmol/l) was 78%. A total of 35 deaths were recorded during this period. Results were available for 34 of the children who died (97%). The mean hyperglycaemia for all admissions was 9.6mmol/l and the mean hyperglycaemia for all deaths was 12.2mmol/l. 32% of the deaths occurred in those with maximum blood glucose between 6.1-8.5mmol/l (36% of the patients). 59% of the deaths occurred in those with maximum blood glucose more than 8.5mmol/l(23% of the patients). CONCLUSION: The incidence of hyperglycaemia in Paediatric Intensive care unit is high and the mean hyperglycaemia is significantly higher for the mortality group. Mortality increased progressively as glucose values increased. More aggressive glycaemic control might improve outcome in Paediatric intensive care units (as proven in adults) and these findings justify the need for a proper trial of glycaemic management in Paediatrics. REFERENCES: 1.Greet Van Den Berghe et al. Intensive insulin therapy in critically ill patients. The New England Journal Of Medicine 2001 Nov; 345:1359-65. 2.Krinsley JS. Association between hyperglycemia and increased hospital mortality in a heterogeneous population of critically ill patients. Mayo Clin Proc. 2003 Dec; 78(12): 1471-8. 3.Guillermo E. Umpierrez et al. Hyperglycemia: An Independent Marker of In-Hospital Mortality in Patients with Undiagnosed Diabetes. The Journal Of Clinical Endocrinology & Metabolism Vol 87 (3): 978-982

S218

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

845

Oral Presentations Clinical outcomes – 843-848 843

INTENSIVE CARE PATIENTS AND LONG TERM MEDICATION (LTM) Bloomfield R1, Campbell A1, Noble D W1 Intensive Care, Aberdeen Royal Infirmary, Aberdeen, United Kingdom

1

MALE GENDER IS ASSOCIATED WITH INCREASED MORTALITY AND HIGHER RESOURCE CONSUMPTION IN ICUS Reinikainen M T1, Niskanen M M1, Uusaro A1, Ruokonen E T2 Department of Anaesthesiology and Intensive Care, Kuopio University Hospital, Kuopio, 2 Department of Intensive Care, Tampere University Hospital, Tampere, Finland

1

INTRODUCTION: Gender modifies hormonal and immunologic responses caused by severe trauma or critical illness [1]. These gender-based differences might affect outcome. We aimed to investigate the impact of gender on hospital mortality, length of intensive care unit (ICU) stay, and intensity of care of patients treated in ICUs. METHODS: This was a cohort study using a national ICU database (The Finnish Consortium of Intensive Care Data). Participating units included eighteen ICUs in university and central hospitals in Finland. The study population consisted of 27,571 ICU admissions during the period 19992001. We used logistic regression analysis to test the independent effect of gender on hospital mortality. Severity of illness as measured with Acute Physiology and Chronic Health Evaluation II (APACHE II) scores and the different diagnostic categories were controlled for. After stratification for disease severity, we compared the lengths of ICU stay and the intensity of care as measured with Therapeutic Intervention Scoring System (TISS) scores between men and women. RESULTS: Male gender was independently associated with increased hospital mortality (adjusted odds ratio 1.10 (95% confidence interval 1.01-1.18, P = 0.023)). This was due to a better prognosis of women than men particularly among the postoperative ICU admissions and among the patients with relatively low APACHE II scores (< 21). Mean length of ICU stay was 3.3 days for men and 2.7 days for women, P < 0.001. The length of ICU stay was higher for males than for females in all categories of disease severity except the most severely ill patients (APACHE II scores > 30). The mean of the average daily TISS scores was 23.6 for men and 22.2 for women, P < 0.001. Male patients comprised 61.6% of the study population but they consumed 65.6% of intensive care resources, if resource consumption is measured with TISS scores. CONCLUSION: Male gender was independently associated with increased hospital mortality in a heterogeneous population of ICU patients. In addition, male patients were treated longer and more intensively than female patients. These results imply that women may have a better ability than men to recover from trauma or critical illness.

INTRODUCTION: Patients presenting to the intensive care unit (ICU) often have pre-existing conditions for which they are prescribed LTM. Many of these drugs have disease modifying effects and have a good evidence base for improving outcome e.g. â-blockers, aspirin. With the onset of critical illness, the alterations in physiology and pharmacokinetics make it difficult or inappropriate to continue these drugs in ICU. After the resolution of the acute illness, the preexisting co-morbidities are likely to require ongoing treatment. During an intensive care admission patients are cared for and discharged by other medical teams. We wanted to ascertain if such drugs are reinstated after the acute illness. Also, new medication is frequently commenced on ICU, and may or may not be intended to be continued. We performed an observational study of the management of long term medication during and after an admission to ICU. METHODS: We conducted a retrospective review of case notes of intensive care survivors over the age of 60, admitted to Aberdeen Royal Infirmary ICU between July 2001 and January 2002. This age group was chosen to identify a patient cohort likely to be on chronic medication. To maximise clinical relevance, simple analgesics, laxatives, vitamin supplements and short-term antibiotics were excluded from the study. RESULTS: Seventy one patients were eligible, 59 were prescribed 197 LTMs. Of these, 112 were not continued upon ICU admission. For 39 drugs there was an obvious or implied temporary contraindication but only two drugs had a documented reason for this. Out of the 112 drugs discontinued, 13 were recommenced during the patient’s stay. Of the remainder, 87 drugs were not restarted by ICU discharge with no reason documented, 50 were for cardiac disease. Other medical teams recommenced 32 drugs by hospital discharge, 33 were not recommenced including 23 cardiac medications. Of 154 new medications commenced in ICU, 110 were listed in the ICU transfer letter but for 96 no rationale or long term plan was documented. Twenty eight of these continued beyond hospital discharge, 12 were H2 antagonists. CONCLUSION: This study demonstrates that LTM is not always reinstated after the resolution of the acute illness. In some cases this will be appropriate, in others this may suggest errors of omission rather than an active clinical decision. Drugs commenced in ICU lack clear documentation causing uncertainty over indications and duration of therapy. There is potential for disease recurrence or lack of symptom control from the inappropriate management of effective drug therapies. Processes in the ICU should be developed to minimise this risk and improve communication between different medical teams.

REFERENCE: 1. Knöferl MW et al. (2002) Preservation of splenic immune functions by female sex hormones after trauma-haemorrhage. Crit Care Med 30:888-893

844

846

A PROSPECTIVE STUDY INTO SEXUAL FUNCTION AFTER INTENSIVE CARE

OUTCOME IN ICU PATIENTS WITH AND WITHOUT TRACHEOSTOMY

Griffiths J A1, Quinlan J2, Gager M3, Waldmann C1 1 Anaesthetics and Intensive Care Medicine, Royal Berkshire Hospital, Reading, 2Anaesthetics, John Radcliffe Hospital, Oxford, 3ICU Follow up Clinic, Royal Berkshire Hospital, Reading, United Kingdom

Flaatten H1, Aardal S1, Gjerde S1 1 Dep. of Anaesthesia and Intensive care, Haukeland University Hospital, Bergen, Norway

INTRODUCTION: Subjective quality of life issues are now superseding objective measures of outcome after intensive care. Sexual dysfunction, as part of a full quality of life assessment, has been noted in patients following cardiac surgery1 and major trauma2. Sexual function provides a sensitive measure of outcome in all patients discharged from intensive care but has not previously been specifically investigated. METHODS: A structured, previously validated questionnaire 3 was given to 106 patients attending an intensive care follow up clinic. All patients had spent at least 4 days on the intensive care unit during the course of their illness. Patients scored their sex life before and after ICU admission: any patient who had a fall in score from pre- to post-ICU admission was defined as having sexual dysfunction. Other outcome measures included the presence of Post Traumatic Stress Disorder (PTSD) and the individual’s and their partner’s satisfaction with their current sex life. Where sexual dysfunction was identified an attempt was made to differentiate an organic or psychological cause. RESULTS: The reported incidence of sexual dysfunction at any stage during 12-month follow-up was 46.3% (50 patients). It was more common in males (50.7%) than females (37.8%) but this did not reach statistical significance (p=0.2). Categories of sexual dysfunction identified within those 50 patients were „no desire“ (48%), „nothing works“ (34%), and „physical changes“ (42%). A strong correlation was found between sexual dysfunction and the presence of PTSD (Table 1). 45% of patients and 40% of partners were not satisfied with their current sex life. No other medical practitioner had sought the presence of sexual dysfunction during the patient’s recovery. Chi-Squared Data box showing the relationship between PTSD and sexual dysfunction. Sexual Dysfunction at any stage No Yes TOTAL Pearson chi2 (1) = 6.7394 p = 0.009

PTSD No 34 16 50

PTSD Yes 24 32 56

TOTAL 58 48 106

CONCLUSION: Sexual dysfunction is common in patients recovering from a period of critical illness. Sexual dysfunction can have a significant impact on quality of life, both for the patient and their partner. The intensive care follow up clinic provides a suitable forum for the identification, referral and treatment of patients in whom sexual dysfunction is present. REFERENCES: 1. Nielsen D, Sellgren J, Ricksten S-E. Crit Care Med 1997;25:52-72. 2. Thiagarajan J, Taylor P, Hogbin E, Ridley S. Anaesthesia 1994;49:211-218. 3. Quinlan J, Gager M, Fawcett D, Waldmann CS. Sexual dysfunction after intensive care. BJA 2001; 87: 348P.

INTRODUCTION: The number of patients with tracheostomy performed in our ICU has increased considerably after the introduction of percutaneous techniques. This study was performed in order to reveal any effects from the procedure on hospital or 90-days mortality. METHODS: All major procedures in our ICU patients are prospectively registered in our ICU database. This includes tracheostomy, classified as surgical or percutaneous, with the date performed. All patients with a tracheostomy performed from 1997 to 2003, with their SAPS II score, ICU length of stay and respirator time were registered. Time of death for non-survivors were retrieved using data from the Peoples Registry of Norway. Outcome data for patients with tracheostomies were analyzed separately for those surgically and percutaneously performed, and compared with patients where no tracheostomy had been performed. RESULTS: A total of 2841 patients were treated in this period and 464 (16.3 %) had a tracheostomy performed; 364 (78.4 %) percutaneously. Tracheostomy was performed surgically most often perioperatively in patients with head and face injury or with ENT malignancies, while patients with percutaneous tracheostomy had the procedure performed because of anticipated prolonged need of an artificial airway. The mean days on ventilator was 2.5 in patients without tracheostomy, and 11.5 and 8.6 days for patients with percutaneous and surgical tracheostomies respectively. More data from each group are shown in the table (data presented as mean if not stated otherwise) Patients with and without tracheostomy Group

Age SAPSI (yrs) I No tra n=2377 48.5 37 Percut n=364 55.0 47.1 Surg n=100 47.6 30.9 *SMR = Standardized mortality ratio

ICU days 3.2 15.2 10.8

ICU mort % 18.2 10.7 4.0

Hosp mort % 27.5 28.0 13

90 d mort % 38.5 33.5 18

SMR* .94 .67 .68

CONCLUSION: Patients with a tracheostomy differ in many ways from other ICU patients in requiring more resources, and they more often die on the ward than in the ICU. The two groups with tracheostomy also differ in terms of age and severity of illness. The low SMR in both groups together with 90 days mortality data gives no reason to believe that this procedure inflicts any severe problem to this group of patients as a whole.

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

847 HAS THE MORTALITY OF ACUTE RENAL FAILURE INCREASED? Sakr Y1, Eympa Y2, Reinhart K1, Vincent J L2 1 Department of Anaesthesiology and Intensive Care, Friedrich Schiller University Hospital, Jena, Germany, 2Department of Intensive Care, Erasme Hospital, Free University of Brussels, Brussels, Belgium INTRODUCTION: We performed a systematic review of the literature to determine mortality rates in patients with acute renal failure (ARF) over the past decades and to report associated comorbidities that might have contributed to the mortality in this population. METHODS: We performed a MEDLINE search using the keywords ‘acute renal failure’ crossed with ‘outcome’, ‘mortality’, ‘ICU’, ‘critically ill’ or ‘prognosis’ in the period from January 1970 to December 2002. Abstracts and full articles were eligible if mortality rates were reported. We also reviewed the bibliographies of available studies for further potentially eligible studies. The dates of the observation period for each study and not the publication dates were considered for our analysis, so the earliest data were from 1956. Assuming a uniform distribution of admissions and deaths in each article over the study period, the total admissions and deaths were equally divided annually during the reported period. The mortality rates were computed for each interval according to the number of patients and deaths in this specific interval. RESULTS: Of 72 articles fulfilling the criteria, 5 were excluded because of duplicate publications using the same database, so that 67 were included in our review with a total of 14353 patients. The evolution of mortality in patients with ARF is shown in the Table. Aggravating factors could be classified as concomitant co-morbidities on admission, factors related to the pathophysiology and course of ARF, and associated organ dysfunction/failure during hospitalization. Among the frequently reported aggravating factors were age, prognostic scores on admission, sepsis, and cardiovascular and respiratory failure. Temporal changes in mortality Period 1956-60 1961-65 1966-70 1971-75 1976-80

n 302 420 413 581 1144

% death 42 47 49 63 61

period 1981-85 1986-90 1991-95 1996-00 Total

n 1659 1769 4008 4048 14353

% death 51 52 55 55 55

CONCLUSION: Despite progress in the management of ARF over the last fifty years, mortality rates seem to have remained unchanged at around 50%.

848 INCIDENCE AND PROGNOSIS OF EARLY HEPATIC DYSFUNCTION IN THE ICU - A PROSPECTIVE MULTICENTER STUDY Kramer L1, Metnitz P G2, Jordan B3, Druml W4, Hiesmayr M2, Steltzer H2, Bauer P4, for the Austrian Epidemiologic Study on Intensive Care 5 1 Dept. of Medicine IV, 2Dept. of Anaesthesiology, 3Dept of Medical Statistics, 4Dept. of Medicine III, Medical University Vienna, 5ASDI Study Group, Vienna, Austria INTRODUCTION: The incidence of early hepatic dysfunction in critically ill patients is largely unknown. This analysis was performed to (1) assess the frequency of early hepatic dysfunction in a large sample of critically ill patients and (2) to investigate the effects of early hepatic dysfunction on mortality. METHODS: A prospective multicentric cohort study was performed in 47 adult medical, surgical, and mixed Austrian ICUs over a 5-year period (1998-2002, ASDI Database). This database was analyzed for admission data, co-morbidities, SAPS-II and TISS-28 scores, length of ICU stay and mortality. RESULTS: A total of 31,287 consecutive admissions (patient age, 62.5±17yr; medical reason for admission, 44.5%; planned surgery, 32.9%; emergency surgery, 17.9%) were analyzed. Preexisting cirrhosis was present in 644 patients (2.1%), 106 patients (0.3%) had acute hepatic failure and 4146 further patients (13.25%) developed hepatic dysfunction (bilirubin >2mg/dl) within 48 hrs of admission. Patients with hepatic dysfunction differed from other patients in most baseline characteristics, had longer ICU stays (5 vs 2 days, medians), higher SAPS-2 (36 vs 26) and daily TISS-28 scores (35 vs 28), higher ICU mortality (23.4% vs 9.7%) and hospital mortality (30.4% vs 14.7%, p<0.001 for all comparisons). In patients with hepatic dysfunction, observed mortality exceeded SAPS-2-predicted mortality by 0.6%. In contrast, observed mortality was 2.8% below predicted mortality in other patients, suggesting potential effects of hepatic dysfunction on mortality. To explore independent effects of hepatic dysfunction, a stepwise logistic regression model was constructed. Stepwise logistic regression model : Effect Age Diagnosis SAPS II score Circulat. failure Renal failure Neurologic failure Pulmonary failure Hepatic dysfunct.

p - value <.0001 <0.0001 <0.0001 0.4941 <0.0001 <0.0001 <0.0001 <0.0001

Estimate 0.00995

OR 1.010

95% CI 1.007-1.01

0.0999 0.0493 0.2272 0.2415 0.3734 0.5718

1.105

1.099-1.11

1.255 1.273 1.453 1.771

1.159-1.35 1.166-1.39 1.303-1.62 1.365-2.30

CONCLUSION: In this large group of critically ill patients, early hepatic dysfunction occurred much more frequently than commonly considered. Logistic regression analysis demonstrated independent effects of hepatic dysfunction on mortality, exceeding those of age, sex and extrahepatic organ dysfunctions. Consequently, early hepatic dysfunction should be recognized as a major independent risk factor in critically ill patients.

S219