S64

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

Oral Presentations Clinical trials in acute respiratory failure – 232-237 232

234 NON-INVASIVE VENTILATION TO PREVENT PROSPECTIVE RANDOMISED CLINICAL TRIAL.

EFFECTS OF PRONE POSITION VENTILATION IN ACUTE RESPIRATORY FAILURE ON THE COURSE OF CRITICAL ILLNESS Lewejohann J C1, Rieh E1, Muhl E1, Bruch H P1 1 Surgery, Universitaetsklinikum Schleswig-Holstein Campus Luebeck, Luebeck, Germany INTRODUCTION. Prone position ventilation (PPV) can improve oxygenation in acute respiratory failure (ARF). There is neither an evidence for an obvious improvement of outcome nor proved positive effects on the further course of critical illness so far. The aim of our study is to analyze the effect of PPV on the course of sepsis and acute respiratory failure. METHODS. We studied 110 consecutive patients with an ARF, n=18 with ALI and n=92 with ARDS (mean age 66±13[SE]) in a clinical follow-up design at a SICU in a university hospital, who met the criteria of the American-European consensus definition. All patients were ventilated intermittent in supine and prone position (135°left/right-side-position) for at least six hours per day for supportive treatment of ARF. Responder or non-responder each after a time interval of 8, 16, 24 or 48h after starting PPV were defined by a higher or lower oxygenation-index respectively in comparison to the median of all patients in each time interval. Data collection included individual oxygenation-index and the cause of death in deceased patients. (Statistical analysisSPSS®:Mann-Whitney-Test). RESULTS. PPV was well tolerated in all n=110 patients and showed a significant increase of PaO2/FiO2-ratio in n=106 within the first six hours of PPV (SP 149±0,52 vs. PP 230±0,73mmHg [mean±SEM]). In the remaining four cases there was a positive effect within the first 24 hours. N=67 (61%) of the patients died and n=43 (39%) survived ARF. The outcome subsequent to the defined time intervals was different: Death as a result of sepsis (S) or respiratory failure (RF) later on Time interval Responder (S) Non-Responder (S) Responder (RF) Non-Responder (RF)

after 8h 32,7% 37,5% 3,6% 5,4%

after 16h 25,0% 41,8% 1,8% 7,3%

after 24h 20,0% 49,1% 3,6% 5,5%

after 48h 24,1% 46,3% 1,9% 7,4%

EXTUBATION

FAILURE.

A

Valencia M1, Ferrer M1, Bernadich O1, Nicolas J M2, Badia J R1, Cavalcanti M1, Alarcon A1, Torres A1 1 UVIR - Respiratory Intensive Care Unit, Clinical Institute of Pneumology and Thoracic Surgery, 2 AVI - Intensive Care Area, Hospital Clìnic i Provincial de Barcelona, Barcelona, Spain INTRODUCTION. Extubation failure and reintubation is associated with increased incidence of nosocomial pneumonia and mortality in mechanically ventilated (MV) patients. Age > 65yr, APACHE II > 12 and heart failure are associated with higher risk of extubation failure. Noninvasive ventilation (NIV) is effective to facilitate extubation in patients with weaning failure, but the efficacy of NIV to prevent extubation failure has not been assessed yet. METHODS. We assessed the efficacy of NIV to prevent extubation failure in 115 ventilated patients (age 72±9 yr, 80M, 35F) who tolerated a T-piece trial after recovery from the acute episode but had increased risk for extubation failure: age> 65 yr, cardiac failure, or APACHE-II _ 12 on the day of the extubation. Patients were randomly allocated to receive NIV during 24 hours (n=60), or conventional management (control group, n=55). Both groups had similar characteristics at entry into the study. RESULTS. Extubation failure rate was significantly lower (10, 17% vs 21, 38%, p=0.01) and delayed in time (40±21 vs 23±22 hrs; p=0.04) in the NIV group. Re-intubation was avoided in 2/2 patients from the NIV group and 6/15 patients from the control group who received NIV after extubation failure; hence, re-intubation rate was no different between groups (8, 13% vs 15, 27%; p= 0.10). There was a trend to reduce the need for tracheotomy in NIV group (3, 5% vs 9, 16%; p=0.09). ICU mortality rate was lower in the NIV group (3, 5% vs 11, 20%, p=0.03), and there was a trend towards a lower hospital mortality (8, 14% vs 16, 30%; p=0.08), and ICU length of stay (10±7 vs 14±13 days, p=0.054) in the NIV group. CONCLUSION. NIV prevented extubation failure and reduced mortality in patients at increased risk. Additional benefits of NIV to prevent re-intubation were observed in patients who developed extubation failure. Grant acknowledgement: FIS 01/1505, SEPAR 2000, Red GIRA, Red Respira, 1999SGR00228, ERS Fellowship, IDIBAPS

CONCLUSION. Our results show that in ARF responders to prone position ventilation die less frequent as a result of sepsis or respiratory failure than non-responders. Therefore PPV seems to have an influence on the course of critical illness of patients with an ARF.

233

235

CURRENT USE OF NON-INVASIVE VENTILATION. RESULTS FROM A 2002 MULTICENTER FRENCH SURVEY

RANDOMIZED MULTICENTER TRIAL IN ARDS: SUPINE VS PRONE POSITION

Demoule A1, Girou E2, Taille S1, Brochard L1, The SRLF Collaborative Group on Mechanical Ventilation 3 1 Service de Reanimation Medicale, 2Unite d’Hygiene et Prevention de l’Infection, Hopital Henri Mondor and INSERM U492, Creteil, 3SRLF, Societe de Reanimation de langue Francaise, Paris, France INTRODUCTION. Non-invasive ventilation (NIV) has a major role in the management of patients with acute respiratory failure (ARF) in the intensive care unit (ICU). A prospective French survey performed in 1997 has shown that the use of NIV, while of benefit for ARF patients, was still moderate. We performed a new survey to reevaluate NIV clinical practice. METHODS. Between March 04 and 24, 2002, a prospective survey was performed in 70 ICUs. A case-report form was completed for all patients requiring any form of ventilatory support, until ICU discharge. Demographics, clinical history, institution of mechanical ventilation, and followup were recorded. RESULTS. During the study period, 1943 patients were admitted, and 1076 received any kind of mechanical ventilation. NIV was first attempted in 24 % of all patients (vs. 16 % in 1997, p<0.0001) and in 52 % of those admitted to the ICU without intubation (vs. 35 % in 1997, p<0.0001). Among patients treated with NIV, 33 (14%) had a do-not-intubate order. The conditions precipitating ARF and the respective incidences of NIV were: hypercapnic ARF in 17 % (NIV 66%), coma in 33 % (NIV 1%), cardiogenic pulmonary oedema in 8 % (NIV 47%), and hypoxemic ARF in 41 % (NIV 23%) (all increased compared to 1997). In 47 % of patients, NIV was discontinued early (vs. 48 % in 1997), and 36% required tracheal intubation. Independent risk factors for failure were SAPS II (Odds ratio 1.07, 95% CI 1.04 to 1.10), hypoxemic ARF (OR 2.22, CI 1.06 to 4.65) whereas a good NIV tolerance was associated with success (OR 0.36, CI 0.15 to 0.89). Mortality rate was 33%. Excluding patients with coma, post-operative and DNR patients, the multivariate analysis found SAPS II, immunosuppression, nosocomial pneumonia and hypoxemic ARF independently associated with mortality, whereas NIV success was associated with survival (OR 0.03, CI 0.003 to 0.20). CONCLUSION. Indications for NIV have dramatically increased among French ICUs over the last 5 years with a steady success rate. In patients admitted free of intubation, NIV is the first modality used to deliver ventilatory support. Hypoxemic ARF is associated with a lower success rate than hypercapnic ARF.

Mancebo J1, Rialp G2, Fernandez R3, Gordo F4, Blanch L3, Rodriguez F5, Garro P6, Ricart P7, Vallverdu I8, Ferrer M9, Castaño J10, Saura P11, Bonet A12, Albert R13 1 Intensive Care, Hospital Sant Pau, Barcelona, 2Intensive Care, Hosp Palma, Palma Mallorca, 3Intensive Care, Hosp Parc Tauli, Sabadell, 4Intensive Care, Hosp Alcorcon, Alcorcon, 5Intensive Care, Hosp V Camino, Pamplona, 6Intensive Care, Hosp Granollers, Granollers, 7Intensive Care, Hosp Can Ruti, Badalona, 8Intensive Care, Hosp S Joan, Reus, 9Intensive Care, Hosp Clinic, Barcelona, 10Intensive Care, Hosp V Nieves, Granada, 11Intensive Care, Hosp Manresa, Manresa, 12Intensive Care, Hosp Trueta, Girona, Spain, 13Intensive Care, Denver Health, Denver, United States INTRODUCTION. In ARDS patients, prone position (PP) improves gas exchange and may protect from ventilator induced lung injury when compared with supine position (SP). Our primary end-point was to show a relative 40% decrease in ICU mortality rate (from 50% SP to 30% PP) in ARDS patients. METHODS. Randomized multicenter clinical trial (Dec 1998-Sep 2002) with an estimated N of 200 patients; alpha=0.05, beta=0.20. ARDS patients were enrolled <48hr of diagnosis. The same ventilatory (VAC) and weaning (PSV) strategies were used and PP was maintained 20 hr/day. Statistical analysis: Student’s t Test, chi-square or Fisher exact test, Kaplan-Meier method and logrank test RESULTS. The study was interrupted after an interim analysis of 136 patients (60 SP, and 76 PP). The groups were well matched at entry (mean±SD, all p=ns, SP vs PP): age 54±17 vs 55±17 yrs; SAPSII 38±15 vs 43±15; PaO2 126±94 vs 107±65 mmHg; FiO2 0.79±0.21 vs 0.84±0.19; PaCO2 43±11 vs 45±9 mmHg; RR 19±4 vs 20±4 bpm; PEEP 12±2 vs 12±2 cmH2O; VT 8.6±1.5 vs 8.3±1.7 ml/kg; Pplat 32±4 vs 32±5 cmH2O. ICU mortality was 58% SP and 43% PP (p=0.12, two tailed). Survival curves did not show statistical differences. CONCLUSION. Although PP showed a 25% relative decrease in mortality in comparison with SP, the difference was not sufficient to reach statistical significance. Grant acknowledgement: Partially funded by ISCIII, Ministerio de Sanidad. FIS 01/0541. Spain

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

236 EFFECTS OF INCOMPLETE AND COMPLETE PRONE POSITION IN ARDS: A MULTICENTRIC STUDY Bein T1, Sabel K2, Scherer A3, Dubb R4, Hekler M4 1Anesthesia, 2Internal Medicine, University Hospital, Regensburg, 3Anesthesia, Zentralklinikum, Augsburg, 4Anesthesia, Katharinen-Hospital, Stuttgart, Germany INTRODUCTION. Prone position (PP) is an established measure to improve pulmonary gas exchange in acute lung injury or ARDS. Incomplete prone position (135o [IPP]) is often used to facilitate positioning head and abdomen, but it is unclear, whether IPP is as effective as complete prone position (180o [CPP]). We started a prospective, randomized, multicentric study to compare both variants. METHODS. From 1/03 until 3/03, 21 patients (SAPS-II 30,4 ± 13, LIS = 2,8 ± 1,1) with ARDS (definition following consensus) were investigated. Following randomization, patients were ventilated in pressure controlled mode and placed in CPP (6 hrs) and then immediately in IPP (6 hrs) ore vice versa (IPP for 6 hrs, then CPP for 6 hrs) using a special positioning system (ProNimbusTM, HNE Healthcare, Hilden, Germany). Before begin of the study (T0), 30 min (T1) and 6 hrs after positioning (T2) blood gases and static compliance (Cstat) were recorded. After end of the study period (12 hrs) patients were repositioned in supine and measurements were assessed again (T3). RESULTS. The results are presented in table 1. Both CPP and IPP improved oxygenation significantly, but did not influence Cstat. After end of the study, the benefitial effect on oxygenation continued to exist. Overall, PaCO2 worsened during CPP and IPP. Effects of IPP and CPP on pulmonary gas exchange before-T0 CPP-T1 IPP-T1 CPP-T2 IPP-T2 after-T3 PaO2/FIO2 154±51 226±78** 233±115* 273±107** 249±102* 229±107** mmHg PaCO2 40.4±5.4 44.7±8.0* 46.4±8.8* 44.1±9.0* 46.5±9.0* 45.0±12.5 mmHg Cstat 48.4±19.3 46.7±16.3 46.4±12.8 47.4±14 47.0±15.3 50.0±10.8 (ml/mbar) ** = p - 0,01 - * = p - 0,05 in comparison with T0 (Wilcoxon-test). CONCLUSION. Our first results indicate that the technique of IPP is as effective as CPP in ARDS patients. After reconfirmation by a large data base, IPP might be a suitable therapy in patients with acute lung injury or ARDS and with problems of head or neck positioning or in patients suffering from acute abdomen. Grant acknowledgement: This study is supported by HNE Healthcare, Hilden, Germany

S65

Oral Presentations Tissue oxygenation: Experimental and clinical aspects – 238-243 238 THE ROLE OF OXYGEN DERIVED FREE RADICALS FOR TISSUE DAMAGES DUE TO CONGESTION-REPERFUSION Moriwaki Y1, Sugiyama M1, Kanaya K1, Hasegawa S1, Matsuda G1, Yamagishi S1, Yoshida K1, Uchida K1, Karube N1, Toyoda H1, Kosuge T1, Yamamoto T1, Suzuki N1 1 Critical Care and Emergency Center, Yokohama City University Medical Center, Yokohama, Japan INTRODUCTION. Tissue damage due to ischemia which is brought from occlusion of supplying artery has been vigorously examined by and has been well known. However, tissue damage due to congestion has not been well examined. The aim of this study is to clarify the difference between tissue damage due to ischemia and due to congestion, and to clarify the role of oxygen derived free radicals (ODFR) for tissue damage due to congestion and reperfusion. METHODS. Small intestinal segment of Wister male rats were clamped with supplying arteries and draining veins (ischemic group, group I), or clamped with draining veins without clamping supplying arteries (congestion group, group C). In both groups, clamped segment was released 10 or 30 minutes after the clamp. Tissue lipid peroxide (LPO) and the activity to produce ODFR of neutrophils in the draining vein from involved intestinal segment (chemiluminescence, CL) were measured 10 minutes after release. Serum endotoxin (ETX) was measured 30 minutes after release. SOD and catalase were administrated before release of vessels, and LPO and CL were measured 10 minutes after release. RESULTS. The tissue damage was severer in the group C than in the group I with the same clamping time. The level of LPO in the involved bowel and CL of the vein draining from the involved bowel were also higher in the group C when the clamping time was the same. But in the group C, the level of LPO and CL was higher after the release of 10 minute’s clamping than after 30 minute’s clamping, in which the tissue damage during congestion was too severe to result in tissue necrosis. The increments in LPO and CL were suppressed by administration of radical scavengers. CONCLUSION. Tissue damage due to congestion and reperfusion is also derived from ODFRS, and is severer than that due to ischemia. Treatment with radical scavenger is useful for these tissue damages. REFERENCE(S). 1. Granger DN, et al. Superoxide radicals in feline intestinal ischemia. Gastroenterol 1981;81:22, 2. Moriwaki Y. Reactive oxygen species and tissue damage induced by ischemia-reperfusion of the rat small intestine. Yokohama Med 1990;41:275, 3. Moriwaki Y, et al. The worsening of tissue damage due to the transient release during the ischemia of rat small intestine. J Jpn Surg Soc 1993;94:1017

237

239

COULD HFPV REDUCE MORTALITY AFTER SMOKE INHALATION? A RANDOMISED STUDY.

PNEUMONIA-INDUCED SEPSIS AND GUT INJURY. EFFECTS OF A POLY-(ADPRIBOSE) SYNTHETASE INHIBITOR.

REPER P A E , G H1, Dalne E1, Van Loey C1 Critical care department, Hospital Queen Astrid, Brussels, Belgium

Lobo S M A1, De Backer D2, Preiser J C2, Cunrath G S3, Rosinha S R P O3, Chibeni G S A3, Contrin L M3, Queiroz M M3, Cury P M3, Burdmann E3, Machado A M O3, Togni P4, Szabó C5, Vincent J2 1 Intensive Care Unit, Hospital de Base - Faculdade de Medicina, Sao Jose do Rio Preto, Brazil, 2Department of Intensive Care, Erasme Hospital - Free University, Brussels, Belgium, 3Intensive Care Unit, Hospital de Base - FaculdadedeMedicina, Sao Jose do Rio Preto, 4Radiology department, FAMECA, Catanduva, Brazil, 5Inotek Corporation, Beverly, Massachussetts, United States

1

INTRODUCTION. Inhalation injury and bacterial pneumonia represent some of the most important causes of mortality in burn patients. High Frequency Percussive Ventilation (HFPV) is a ventilatory mode which combines the advantages of conventional ventilation (CV) with some of those of high frequency ventilation. METHODS. Eighty severely burned patients with smoke inhalation injury were randomised on admission for conventional (CV-control group) ventilation versus high frequency percussive ventilation (HFPV-study group). Arterial blood gases, ventilatory and haemodynamic variables were recorded during the first 5 days after injury. Mortality rate was assessed at day 5 and 28, ICU stay, ventilation days, incident complications and respiratory infections rate were also evaluated. A statistical analysis (Wilcoxon and Student t-test) was performed. RESULTS. There was no significant difference between the two groups for Burn Surface Area, age, Lung Injury Score and Apache III score. The study demonstrated a significant decrease in days under artificial ventilation, a significant decrease of ICU days and an improved survival rate as well at day five as at day 28 in the HFPV group. We also observed a significant reduction of pulmonary infectious complications and a significant improvement of gas exchanges in the HFPV group. Outcome, ICU days, days of artific and pulmonary infections in CV and HFPV groups

Days of artificial ventilation ICU days Mortality day 5 Mortality day 28

CV group

HFPV group

p

15 32 ± 24 days 8/40 ( 20% ) 12/40 ( 30% )

10 29 ± 12 days 2/40 ( 5% ) 5/40 ( 12.5% )

< 0.01 < 0.01 < 0.01 < 0.01

CONCLUSION. HFPV could improve mortality, and reduce ICU stay and respiratory infections in severely burn patients with inhalation injury. REFERENCE(S). Cioffi W. and coll: Prophylactic use of high frequency percussive ventilation in patients with inhalation injury. Ann. Surg., June 1991; 575-80.

INTRODUCTION. Pseudomonas aeruginosa (PA) is the most common gram-negative organism associated with nosocomial pneumonia. Ileal mucosa injury may be induced by severe lung infection. During septic shock, peroxynitrite-mediated DNA strand-breaks activate the nuclear enzyme poly-(ADP)-ribose synthetase (PARS) resulting in cell dysfunction (1). The aim of this study is to determine whether gut injury could be demonstrated in sepsis induced by PA and the effects of a PARS inhibitor in gut injury. METHODS. After baseline measurements 20 rabbits were randomised into 3 groups: Sham: transtracheally inoculated (TI) with 2 ml of phosphate buffer solution (PBS), n=5; PA + saline, TI with 4 x 108 CFU/mL of PA in 2 ml/kg of PBS + I.V. saline, n=8 and PA + PJ34, TI with 4 x 108 CFU/mL of PA and I.V. PJ34, 10 mg/kg bolus and 3 mg/kg/h continuous infusion. RESULTS. PA caused a hyperdynamic response characterized by a significant increase in Qao (P<0.05). QSMA was higher in PA groups. In the 3 groups, luminal gut lactate significantly increased compared to baseline measurements but PCO2-gap did not. Histological specimens showed moderate or diffuse alveolar infiltrate in-group PA+saline (6/8) and in group PA+PJ34 (6/7). Moderate to massive epithelial lifting of the epithelial layer from the lamina propria of the gut occurred in 0%, 28% and 25% (Sham, PA+saline and PA+PJ34, respectively, NS). Gut wet-todry weight ratio increased significantly in PA+saline compared to Sham (6.4+0.7 vs 7.5+0.8, respectively, p<0.05) and significantly reduced in PA+PJ34 (6.1+0.5). Haemocultures were positive in 20% (Sham), 100% (PA+saline) and 57% (PA+PJ34) (RR 0.57 CI 95% 0.30-1.08). CONCLUSION. Ileum injury occurred rapidly in normotensive sepsis secondary to bacterial pneumonia. PARS inhibitor significantly reduced ileum oedema. Intramucosal PCO2 measurements did not indicate ischemia as the cause of these alterations. REFERENCE(S). Soriano et al., Nat Med. 2001 Jan;7(1):108-13

S66

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

240

242

CEREBRAL RESPONSE TO SUSTAINED SYSTEMIC INFLAMMATION IN OVINE TRAUMATIC BRAIN INJURY

ERYTHROPOIETIN REVERSES ERYTHROID PROGENITORS

Stubbe H D1, Greiner C2, Akcocuk A1, Westphal M1, Van Aken H1, Rickert C3, Hinder F1 of Anesthesiology and Intensive Care, 2Department of Neurosurgery, 3GerhardDomagk-Institute of Pathology, University of Münster, Münster, Germany

CLAESSENS Y1, CHICHE J1, PENE F1, DHAINAUT J1, MIRA J1, CARIOU A1 1Medical Intensive Care Unit and Department of Cell Biology, Cochin Hospital, Paris, France

1Department

INTRODUCTION. Traumatic brain injury (TBI) is frequently complicated by a systemic inflammatory response secondary to multiple trauma, shock, or sepsis. We investigated the impact of systemic inflammation on cerebral haemodynamics and metabolism in a modified model of ovine closed head injury (1). METHODS. Sheep (n=15) were instrumented with arterial and sagital sinus catheters, intracranial pressure (ICP) monitoring, and internal carotid blood flow (ICBF) probes, and investigated for 14 hrs. Under anaesthesia (ketamine/diazepam) and mechanical ventilation, brain injury was induced by head impact delivered by a non-penetrating stunner. One group (TBI/ENDO, n=6) received endotoxin (S. typhi, 10 ng/kg/min), while the second group (TBI, n=6) received NaCl 0,9%. Three animals served as sham controls. Results are expressed as means ± SEM. Statistics were calculated using multiple measure ANOVA with post hoc t-tests. RESULTS. Head impact increased ICP from 9±1 to 21±6 mmHg (TBI, p<.05) and 10±2 to 24±8 mmHg (TBI/ENDO, p<.05), and decreased ICBF from 170±20 to 110±10 ml/min (TBI, p<.05) and 170±10 to 120 ml/min (TBI/ENDO, p<.05) after 1 hr. ICP remained significantly elevated for the next 12 hrs in both groups. While ICBF returned to baseline level in the TBI group 2 hrs after head impact, ICBF initially further decreased at 2 hrs and finally markedly increased after 12 hrs of endotoxemia to 250±30 ml/min compared to 170±20 ml/min in the TBI group (p<.05). This corresponded to a significant decrease of cerebral vascular resistance (CVR) in the TBI/ENDO group. Cerebral metabolic rate of oxygen (CMRO2) did not differ between these groups after 12 hrs of endotoxemia. In the Control group, all parameters remained at baseline levels throughout the experiment. CONCLUSION. Sustained systemic inflammation induced a phasic cerebral vascular response that was uncoupled to global cerebral oxygen metabolism and might thus contribute to secondary ischemic injury after brain injury.

SEPTIC

SHOCK-INDUCED

APOPTOSIS

OF

INTRODUCTION. Anaemia is common in critically ill pts with sepsis. Although several hypotheses have been proposed to explain this finding, the primary mechanism remains unclear. METHODS. After approval of our institution’s ethics committee, we prospectively studied the erythroid lineage in pts with septic shock (Bone criteria) and anaemia (Hb<100g/L). Erythroid progenitors were obtained after bone marrow separation. We studied the presence of apoptosis and the response to erythropoietin (Epo) using flow cytometry (annexin V and DiOC6 staining). We also assessed erythroid colonies formation after addition of serum from septic patients. Erythroid progenitors and serum from healthy subjects were used as control. RESULTS. Ten consecutive pts (age 50±16 y.o, SAPS2 66±20, LOD 8.2±3, mean±SD) were enrolled. All pts were mechanically ventilated and 7 required renal replacement therapy. A functional Epo receptor was detected at the membrane of erythroid progenitors. Annexin V (26.1±8.8% vs 3.2±3%, P<.05)and DiOC6 staining (56±5.4% vs 19.1±10.5%, P<.05) revealed increased apoptosis in erythroid progenitors from septic pts as compared to controls. Membrane expression of Fas was increased in septic pts (3.06±1.2 vs 1.06±0.18, P<.05). Expression of the death domain receptors to TNF and TRAIL did not differ between pts and controls. Whereas serum from septic pts decreased the number of erythroid colonies as compared to control serum (67±30 vs 272±67, P<.05), addition of Epo to serum from septic pts partially rescued erythroid development (67±30 vs 179±35, P<.05). Inhibitors of the main death domain receptors were unable to restore erythropoiesis. CONCLUSION. Increased apoptosis in the erythroid compartment contributes to anaemia in septic shock. This process can be reversed by the addition of Epo. These findings potentially explain the decreased need for blood transfusion after treatment of critically ill pts with Epo [1]. REFERENCE(S). 1. Corwin HL et al. JAMA.2002;288:2827-2835. Grant acknowledgement: This study was supported by OrthoBiotech Inc.

REFERENCE(S). 1. Lewis SB et al. (1996) A head impact model of early axonal injury in the sheep. J Neurotrauma 13:505 Grant acknowledgement: This study was supported by a grant of the Else Kröner-Fresenius Stiftung, Germany.

241

243

DOBUTAMINE REVERSES THE VASOPRESSIN-ASSOCIATED IMPAIRMENT IN OXYGEN DELIVERY IN OVINE ENDOTOXAEMIA

PERSISTENT MICROCIRCULATORY ALTERATIONS ARE ASSOCIATED WITH ORGAN FAILURE AND DEATH IN SEPTIC SHOCK

Westphal M1, Daudel F1, Van Aken H1, Stubbe H D1, Schepers R1, Bone H G1 Department of Anaesthesiology and Intensive Care, University of Muenster, Muenster, Germany

Sakr Y1, Dubois M1, De Backer D1, Creteur J1, Vincent J1 Department of Intensive Care, Erasme hospital, Free University of Brussels, Brussels, Belgium

1

1

INTRODUCTION. Arginine vasopressin (AVP) is increasingly used for haemodynamic support of septic patients [1]. Via reflex mechanisms, however, AVP increases vagal tone, thereby reducing cardiac index (CI) and in proportion oxygen delivery (DO2) [2]. This study was designed as a prospective, controlled and single-blinded laboratory experiment to determine the efficacy of dobutamine (DB) to reverse the AVP-associated impairment in systemic DO2. A further objective was to evaluate the impact of AVP, with and without concomitant DB infusion, on mixed venous oxygen saturation (SvO2).

INTRODUCTION. Microcirculatory alterations are frequent in septic patients. We investigated the possible association between the time course of these alterations and the degree of organ failure and outcome in patients with septic shock.

METHODS. Fourteen adult ewes were instrumented for chronic study. Following a baseline measurement in the healthy state, a hypotensive-hyperdynamic circulation was induced and maintained by a continuous infusion of Salmonella typhosa endotoxin (10 ng·kg-1·min-1). After 16 hours of endotoxaemia, the sheep were randomly allocated to the intervention (n = 8) or the control group, receiving only normal saline (n = 6). In the intervention group low dose AVP (2.4 U·h-1) was continuously infused. After one hour, DB was additionally administered at incrementing doses (2, 5 and 10 g·kg-1·min-1; each dose for 30 minutes). For statistical analysis, a two-way analysis of variance with appropriate post hoc correction (Student-Newman-Keuls) was used. Data are expressed as mean ± SEM. RESULTS. Following 16 hours of endotoxaemia, all sheep exhibited a hypotensivehyperdynamic circulation. AVP infusion reversed this condition but resulted in a deterioration in DO2 (treatment: 593 ± 59 vs. 1152 ± 60 mL·min-1·m-2, control: 1150 ± 50 vs. 1140 ± 39 mL·min1 -2 ·m ) and SvO2 (treatment: 55 ± 3 vs. 72 ± 2 %, control: 69 ± 1 vs. 70 ± 1 %; each p < 0.001 in and between groups) . In a dose-dependent manner DB administration improved these endpoints. CONCLUSION. During ovine endotoxaemia: 1) AVP infusion compromised key parameters of oxygen transport at the same time, when haemodynamic parameters suggested haemodynamic stabilisation; 2) Dobutamine counteracted the AVP-associated side effects on DO2 and SvO2; 3) Potentially, a simultaneous infusion of AVP and DB could represent a useful therapeutic option in the common setting of vasodilatory septic shock. REFERENCE(S). 1. Bhavesh et al.: Anesthesiology 2002; 96: 576-582. 2. Holmes CL, et al.: Chest 2001; 120: 989-1002.

METHODS. Sublingual microcirculation was investigated with an Orthogonal Polarization Spectral (OPS) imaging device in 49 patients with septic shock on the day of onset of septic shock (baseline) and each day until resolution of shock. Five sequences of 20 seconds each were recorded and analyzed off-line by a semi-quantitative method. Data were analyzed with nonparametric tests and presented as median [25th-75th percentiles]. RESULTS. Three patients died after the resolution of shock in multiorgan failure (MOF) due to the underlying disease and were excluded, 26 survived and 20 patients died: 13 due to unresolved shock and 7 due to persistent MOF after the resolution of shock. At the onset of shock survivors and non-survivors had similar vascular density (5.6 [4.7-7.0] vs. 6.2 [5.4-7.0] /ml, p=NS) and percentage of perfused small vessels (65.0 [53.1-68.9] vs. 58.4 [47.5-69.1] %, p=NS). Small vessel perfusion improved over time only in survivors (ANOVA; p<0.05 between survivors and non-survivors) but not in non-survivors. Despite similar haemodynamic and oxygenation profiles at the end of shock, patients dying after the resolution of shock in MOF had lower percentage of perfused small vessels than survivors (57.4 [46.6-64.9] vs. 79.3 [67.2-83.2] %, p=0.02). There was a significant correlation between the small vessels’ perfusion at the end of shock in survivors and patients dying in MOF and the degree of organ failure in these patients evaluated at the same day by the SOFA score (y = 16.23 – 0.1x, R2 = 0.34). CONCLUSION. Microcirculatory alterations improved rapidly in septic shock survivors but not in non-survivors. Persistent microcirculatory alterations were associated with MOF and death, even though the evolution of global haemodynamic and oxygenation parameters was similar. OPS imaging can be useful to identify patients in septic shock with persistent microcirculatory alterations in whom microvascular recruitment may be an important therapeutic target.

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

Oral Presentations Quality improvement – 244-249 244 A SYSTEMATIC REVIEW OF QUALITY IMPROVEMENT PROGRAMMES IN CRITICAL CARE Richardson A1, Turnock C2, Patterson L1, Baudouin S V1 1 Critical Care, Newcastle upon Tyne Hospitals NHS Trust, 2School of Health, Community and Education Studies, Northumbria University, Newcastle upon Tyne, United Kingdom INTRODUCTION. The successful implementation of change is acknowledged to be difficult within the field of healthcare (1). Local and national critical care improvement programmes have been initiated in many countries. However, the evidence base and cost effectiveness of the programmes is mostly unknown (2). METHODS. A systematic review was conducted searching three electronic databases – Cinahl, Medline and World Information Nursing – from 1990 to 2002, using only English language text. The search strategy used consisted of combined key terms including change, service development, innovation, models, practice development, quality improvement, process evaluation and quality management in a critical care setting. Articles were selected for detailed review if they contained (i) an explicit change intervention/s and (ii) a measure of outcome/results. Each article was reviewed by 4 independent health care professionals using a standard data extraction protocol. RESULTS. The initial research strategy identified 125 publications. However, only 27 publications contained both intervention and outcome data relevant to critical care. Most studies were cohort observational (n=25). No randomised-controlled trials were identified. The number of change interventions used ranged from 1-11. The commonly-described interventions within each study were education (n=18) and the use of a team approach (n=18). This was followed by the use of quality improvement processes (n=16), creating a shared vision (n=14) and the introduction of standards/protocols/ guidelines (n=14). The least common approaches described were site visits (n=3), a data analyst (n=3), social events (n=2) and conference calls (n=2). The number of outcome measures used to determine success of each study varied from 1-9. The most common outcome measures used were cost (n=10), patient/user satisfaction (n=10) and ICU length of stay (n=7). CONCLUSION. A small number of critical care change management studies have been published. Few are of a high quality and no RCTs were identified. Most employed multiple change interventions to achieve a successful outcome.

S67

246 IS IT WORTHILE TO IMPLENT A CQI PROGRAMM FOR PAIN MANAGEMENT AFTER MAJOR SURGERY? Diby M1, Walder B M D2, Hadorn F R N2, Gustave M R N2, Chanet C R N2, Uhlmann M2, Romand J M D2 1 Surgical IntensiveCare, 2Division of Anesthesilogy, University Hospital of Geneva, Geneva, Switzerland INTRODUCTION. Recommendations to treat pain have been published and consist in pain evaluation, establishment of medication prescription procedures, post therapy re-assessment and monitoring of both patients’ satisfaction and care-givers’ adherence to recommendation. The present study’s aim was to observe if the impact of the implementation of a Clinical Quality Improvement (CQI) program modifies caregivers practice and affects the satisfaction of patient for pain treatment after major surgery. METHODS. For patients admitted after elective cardiac surgeries, a CQI program consisting in the implementation of 1) a guideline for pain assessment consisting of a tool using Visual Analogic Scale (VAS ), 2) a protocol for intra-venous morphine prescription, 3) teaching sessions to every ICU caregivers,4) a periodic feedback. Caregivers practice was evaluated daily before (baseline period BP), during (implementation period IP) and after implementation (control period CP) of the CQI program by an external observer. Patient satisfaction was assessed using a French translated version of a questionnaire (ref) which was administrated at the end of ICU stay. Statistical analysis: Chi-squared test for proportions, Kruskall-Wallis and ANOVA tests for non parametric and parametric data respectively. RESULTS. 80 patients were included in BP, 44 in the IP and 31 in the CP. For the 3 periods 55% of patient had coronary grafting, 31% had cardiac valves surgery, and 14% had mixed surgery. Adequate drug administration if VAS > 30/100 increased from 14% to 58% and to 70% (p<0.05),controlled VAS after drug administration increased from 12%, to 33% and to 47% (p<0.05)during the 3 periods respectively. The recall of maximum pain during mobilisation during ICU stay(mean) decreased from 6/10 to 5/10 and to 4 /10 (p<0.05). The satisfaction to pain treatment during ICU stay (0: very unsatisfied to 4: very satisfied) increased from 3.02, to 3.04, and to 3.34 (NS). CONCLUSION. CQI program implementation was able 1) to change caregivers’ pain management, 2) to decrease the intensity of maximum pain recall during ICU stay after surgery and 3) patients satisfaction tended to be increased. REFERENCE(S). Reference: JAMA, 1995, 274:1874

REFERENCE(S). (1) Curtis, E, White, P Resistance to change: Causes and Solution. Nursing Management 2002; 8(10):15-20.(3) Øvretveit, J, Gustafson D, Using research to inform quality programmes BMJ 2003; 326: 759-761.

245

247

DOES AN INSERVICE TRAINING PROGRAM OUTSIDE THE ICU HELP REDUCE THE READMISSION RATE?

PRE-ARREST PATIENTS ARE NOT REFERRED TO CRITICAL CARE DESPITE PATIENT AT RISK TEAM GUIDELINES

Renuka K M1, D’Souza C M2, Kamat V N1 Anesthesiology And Critical Care, 2Nursing, Sri Ramachandra Medical College And Research Institute, Chennai, India

1

1

INTRODUCTION. Readmission (RA) to the ICU is a major cause of morbidity and mortality. Identifying potentially preventable respiratory causes of RA1 such as aspiration of stomach contents and hypoxia, we organized an in-service training program for nurses in the general wards and analysed if this helped in improving the quality of respiratory care in the wards and in reducing the RA rate. METHODS. A prospective study was conducted on all ICU admissions between August 2001 and March 2003. The in-service training program for monitoring respiratory risk for RA to ICU was implemented in June 2002. The program focused on prevention of aspiration and hypoxic episodes by a change in the method of enteral feeding , nursing in the semi-recumbent position, frequent suctioning of tracheostomy tubes, and airway hydration to prevent blockage. RA was defined as an admission to the ICU any time during the present hospital admission. Group I patients were admitted before June 2002 and Group II after June 2002.The percentage of RA, mortality rate among RA, mean ICU LOS during both the original and RA stay and the mean LOS in the ward before RA were compared.

King D J1 Intensive Care, Middlesex Hospital, London, United Kingdom

INTRODUCTION. Physiological deterioration prior to cardiorespiratory arrest is either not recognised or inadequately treated1.Patient at risk teams (PART)use physiological parameters as a trigger for referral to critical care2.This study aimed to assess the effect of introduction of a PART on treatment in the hours preceding cardiac arrest. METHODS. Retrospective case note analysis of inpatient cardiac arrests, focusing on the 6 hours prior to arrest. Where possible a patient at risk (PAR)score was recorded, based on measured physiological variables. It was noted whether patients met the PART criteria for referral to critical care, and of referrals made. Outcome of arrest was recorded. RESULTS. Criteria for referral to PART were a PAR Score of 5 or more, or a value of 3 or more for any 1 variable. In the majority of cases incomplete observations made it impossible to form a PAR Score. Of 18 patients that could be scored, 9 (50%) met the criteria for immediate referral to critical care. Of this group only 1 patient was referred. Mortality from cardiac arrest was much higher in the patients meeting criteria for referral (see table 1). No.of patients

RESULTS. Group I had a total of 1482 patients and Group II 1384 patients. The RA rate in Gp I was 2.9% compared to 1.8% in Gp II (p<0.05). Mortality rate among the RA was 58% in Gp I and 52% in Gp II (p<0.05). This was higher in both groups than the total crude mortality rate of the ICU during both periods. ( 23.76% and 20.97%). The mean ICU LOS after RA was similar in both groups (5.61 days vs 5.12 days, p>0.05). The mean LOS outside the ICU for RA patients was significantly higher in Gp II( 9.38 days vs 5.51 days in Gp I, p<0.05). The mean APACHE II and SAPS II scores at the time of RA were also significantly lower in Gp II. (12.6 vs 15.27in Gp I and 27.44 vs 34.27 in Gp I respectively p<0.05), suggesting greater awareness in early detection of complications following the training program. CONCLUSION. A training program focusing on potentially preventable respiratory causes of RA into the ICU is worthwhile. This provides better respiratory care in the general wards and helps reduce the RA rate and conserves hospital resources. REFERENCE(S). 1. Schriber. P, Frutiger. A. Intensive Care Med, 28: S1, 16, 2002. Is the readmission rate a useful indicator of ICU quality ?

PAR Score mean (median)

No .of referrals

Meeting PAR criteria 6.5(7) 1 9 Not meeting PAR criteria 1.8(2) 1 9 Mortality is related to PAR Score

Mortality 88.8% 33.3%

CONCLUSION. Physiological deterioration in the hours preceding cardiac arrest continues to be poorly treated. „Early warning“ systems fail to refer patients to critical care despite clear guidelines. Patients with high PAR Scores are more likely to arrest and have a higher mortality from cardiac arrest. Ward staff must be trained to recognise the critically ill. REFERENCE(S). 1.Schein RM, Hazday N, Pena M, Ruben BH, Sprung C. Clinical antecedents to in-hospital cardiopulmonary arrest. Chest 1990; 98:1388-92 2. Goldhill DR, Worthington L, Mulcahy A, Tarling M, Sumner A. The patient-at-risk team: identifying and managing seriously ill ward patients. Anaesthesia 1999; 54:853-60

S68

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

248 PSYCHOLOGICAL IMPACT OF INTENSIVE CARE UNIT TREATMENT Paparrigopoulos T J1, Efthymiou A A2, Kribeni G J2, Kyriakopoulou T2, Kanta A2, Kakoulli M2, Melissaki A2 1Psychiatry, Athens University Medical School , Eginition Hospital, 2ICU, Agios Savvas General Hospital, Athens, Greece INTRODUCTION. Intensive care (ICU) treatment may represent an extreme mental stressor. Despite this, the issue is seldom raised by the ICU literature. In the present study we assessed the psychological consequences of ICU hospitalization, through validated psychometric instruments. METHODS. Subjects treated in the ICU over a two-year period were contacted and asked to interview. From a total of 308 patients, 127 subjects had deceased (41%), 127 (41%) could not be found, 6 (2%) were excluded for various reasons, and 48 (16%) accepted to voluntarily participate in the study. Socio-demographic (age, gender, socioeconomic status, employment, etc.) and clinical characteristics (APACHE II, primary ICU diagnosis, duration of ICU treatment, days in hospital, pre-ICU medical record) of the subjects were recorded. The participants were assessed through the Center for Epidemiologic Studies for Depression (CES-D), a 20-item scale for the assessment of depression and the Davidson Trauma Scale (DTS), a 17-item scale measuring symptoms of posttraumatic stress disorder (PTSD). For data handling, parametric and nonparametric statistics were used as appropriate. RESULTS. The majority of subjects were males (N=33, 68.8%). Mean age ± SD was 52.7 ± 19.4 years, mean APACHE II score was 18,79 ± 6,15, mean duration of ICU stay was 13.19 ± 18.04 days, mean duration of stay in hospital was 63.0 ± 109.8 days, mean time elapsed since ICU discharge at assessment was 20.9 ± 5.5 months. Primary ICU admitting diagnosis was as follows: 10 medical (20.8%), 26 surgical (54.1%), 9 multi-trauma (18.7%), 2 brain injuries (4.2%), and one poisoning (2.1%). Mean CES-D score ± SD was 13.3 ± 13.8; in 31.25% of the sample (N=15) CES-D scores were higher than 15, which is the cutoff for the presence of clinical depression. Mean DTS score was 24.3 ± 24.3. In 25% of the sample (N=12) DTS scores were equal or higher than 40, which is an indication of the presence of PTSD. Co-morbid depression and PTSD were present in 8 individuals. Also, a significant correlation was observed between CES-D score and age (Spearman’s rho=.343, p=.017). CONCLUSION. ICU treatment may have a considerable impact on the survivor’s mental health as indicated by the high prevalence of depressive and PTSD symptoms. This should be seriously considered for adequate management and prevention of these adverse effects of ICU treatment.

Oral Presentations Antibiotic strategies – 250-255 250 ANTIBIOTIC DE-ESCALATION THERAPY (DT) FOR NOSOCOMIAL PNEUMONIA (NP) IN ICU (ADANN STUDY) Alvarez-Lerma F1, Alvarez B2, Rello J3, Ruiz-Ferron F4, Luque P5, Dominguez J M6, SanzRodriguez C7, For the ADANN Study Group 1 ICU, H del Mar, Barcelona, 2ICU, H General, Alicante, 3ICU, H Joan XXIII, Tarragona, 4ICU, H Universitario, Jaen, 5ICU, H Clínico, Zaragoza, 6ICU, H V Rocío, Sevilla, 7Med Dpt, MSD, Madrid, Spain INTRODUCTION. In DT, following an initial broad-spectrum empiric regimen, patients are switched to a spectrum-adjusted regimen based on culture results. METHODS. Prospective, observational study in 24 ICUs. CDC criteria for NP were used. Respiratory sampling was performed prior to empiric imipenem (I/C) ± aminoglucoside/ glycopeptide. Therapy was re-evaluated 3-5 d later based on microbiology data. NP were grouped as: Group 1, treatment adjustment based on microbiology data; Group 2, treatment maintained/broadened based on microbiology data; Group 3, treatment maintained despite microbiology data; Group 4, treatment maintained due to unknown aetiology; Group 5: inadequate empiric regimen. Clinical characteristics and outcome were compared between groups. RESULTS. 244 valuable patients (258 enrolled). Average APACHE II, 17.2 (SD 7.2); 90.6% had late onset, 83.5% were ventilated, and 44.0% had vasopressors. Distribution of NP: group 1, 56 (23.0%); group 2, 14 (5.7%); group 3, 38 (15.6%); group 4, 113 (46.3%); group 5, 23 (9.4%). There were no differences in clinical characteristics between groups. The most frequent isolates were S. aureus (12.8%) and Haemophylus influenzae (11.1%). Pseudomonas spp., Enterobacter spp., and Acinetobacter spp. were more common among patients with previous antibiotic therapy; 23 isolates were not susceptible to I/C, including 7 Acinetobacter, 7 SAMR, and 4 Pseudomonas. Clinical response: cure, 131 (55.0%); improvement, 53 (22.3%); failure, 54 (22.7%); lost, 6 (2.5%). While differences between groups were not statistically significant, the failure rate was higher in inadequately treated patients (38.1%) vs. the other groups (15.4-29.7%). Overall mortality was 20.2% and NP-related mortality 13.6%, with no differences between groups. In logistic regression analysis, age (p=0.033), baseline APACHE II (p=0.004), and inadequate empiric antibiotic therapy (p=0.014) were associated with greater mortality. CONCLUSION. In 82.4% (108/131) of NP with microbiology data, empiric therapy with I/C was adequate. DT was applied properly in 71.0% (93/131). No differences in clinical outcome were found between treatment groups. Yet, age, baseline APACHE II, and inadequate empiric antibiotic therapy were significantly associated with greater mortality. Grant acknowledgement: MSD Spain.

249

251

THE BENCHMARK OF INFECTION RATES AS A QUALITY INDICATOR FOR A GENERAL INTENSIVE CARE UNIT.

INITIAL ANTIMICROBIAL THERAPY IN SPANISH ICU-S: MODALITIES AND REASSESSMENT. A COLLABORATIVE STUDY

Freire M D i C .1, Knibel M F1, Regueira M2, Trote Neto J1, Mota A M1 1 Intensive Care Unit, 2Infection Control Comitee, Hospital Cardio Trauma Ipanema, Rio de Janeiro, Brazil

Insausti J1, Álvarez-Lerma F2, Palomar M3, Olaechea P4, Cerdá E5, Bermejo B6, Sancho H7, Martínez-Pellús A8, ENVIN-Group 9 1 ICU, Hospital of Navarra, Pamplona-Iruña, 2ICU, Hospital del Mar, 3ICU, Hospital ValleHebrón, Barcelona, 4ICU, Hospital de Galdakao, Vizcaya, 5ICU, Hospital de Getafe, Madrid, 6Epidemiology, Hospital de Navarra, Pamplona-Iruña, 7ICU, Hospital Reina Sofía, Córdoba, 8ICU, Hospital de la Arrixaca, Murcia, 9ICU-Infection, Nosocomial, Study-Group, Spain

INTRODUCTION. In the past few years there has been increasing concern about indicators of quality for the Intensive Care Units (ICU),and some of them have been largely used around the world. These numbers have become a great data base generated at ICUs throughout the world, sometimes not shared and not adjusted to other countries realities, some of them dealing with diverse social and economic problems. METHODS. Since October 2001 to December 2002 we attempted to measure widely used quality indicators on our 10-bed general ICU, assessing the length of stay, APACHE II scores, ICU mortality, UD of invasive devices and mechanical ventilation, non-invasive mechanical ventilation use rates and cost evaluation. That was achieved and facilitated by a computer system of data collection and data comparison among 34 ICUs throughout Brazil, the Quati Program, and totalizing 12.485 patients at January 2003. RESULTS. We found at our ICU from October 2001 to December 2002 an UD of mechanical ventilation of 21,23%,with 28,39 cases/1000 of ventilator related pneumonia; an UD of indwelling catheter of 42.33%,with 3,13 cases/1000 of catheter related bloodstream infection ;an UD of bladder catheterization of 40,24%,with 17,26 cases/1000 of catheter related urinary infection, being these infection rates above the preconized NNISS rates. On the other hand, we also found our observed mortality rates statistically equal to estimated mortality rates by APACHE II, and a medium length of stay of 11,29 days, the same observed at Pronovost study(1). CONCLUSION. Despite the attempts to reduce the UD, and the policies established to prevent device-related infections, we could not reach NNISS infection rates for the ventilator-related pneumonia and for urinary infection, but, we could not observe negative impact of these rates at the mortality rate, length of stay and ICU costs. For that, we suggest further investigation with larger populations in order to review the role of NNISS nosocomial infection rates as quality indicators for general ICUs. REFERENCE(S). 1) Pronovost, PJ et al. Physician Staffing patterns and clinical outcomes in critically ill patients: a systematic review. JAMA nov,6,2002-vol 288,nº 17 2) National Nosocomial Infection Surveillance (NNIS) System Report, data summary from january 1992 to june 2002, issued august 2002. Am J Infect Control 2002; 30:458-75

INTRODUCTION. A regular feedback of antibiotic usage in the ICU may play a significant role in the success of policies dealing with antimicrobial resistance. The main purpose of this study was to describe the initial antibiotic strategy and modifications of this therapy when the microbiological data become available. METHODS. Prospective and collaborative study involving all patients admitted in two crosssectional periods in 2001 and 2002. The following parameters were analyzed: initial antimicrobial modality (empiric or specific excluding prophylactic purposes), microbiological confirmation (obtained, not obtained or no more results), reassessment of the initial antibiotic therapy and changes (change of initial treatment or no change) and reason for changing (uncovered pathogen, narrowing of spectrum, development of resistance, unfavourable clinical course and others). Missing data were considered as not available. A statistical analysis using the chi square test was performed. RESULTS. In 2001 there were 5045 patients included (67 ICUs) and in 2002 a total of 6868 patients (85 ICUs). Initial antimicrobial therapy (n/%) in 2001: empiric 2795 (73.7) and specific 996 (26.3); in 2002: empiric 4105 (76.1) and specific 1292 (23.9) (p=0.01). Microbiological confirmation (%) (2001/2002): obtained (26.5/30.2), no confirmation (40.8/48.6), no more results (19.4/13.5), not available (13.3/7.7) (p<0.001). Reassessment and changes (%) (2001/2002): initial therapy changed (18.1/19.3), no change (63.8/63.2), not available (18.1/17.5) (p=0.34). Reason to change (%) (2001/2002): uncovered microorganism (23.8/21.3), narrowing of spectrum (20.5/21.2), development of resistance (3.9/2.1), unfavourable clinical course (36.1/39.1), other reasons (15.3/14.5), not available (0.4/1.8) (p=0.02) CONCLUSION. We have observed a clear prevalence of the empirical modality as initial antibiotic treatment in this ICU population coexisting with a quite low rate of changes after microbiological information. Grant acknowledgement: Aventis

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

S69

252

254

EMPIRICAL ANTIBIOTIC TREATMENT FOR VENTILATOR-ASSOCIATED PNEUMONIA AND RESISTANCE OF PSEUDOMONAS

LINEZOLID VS. GLYCOPEPTIDES FOR THE TREATMENT OF SUSPECTED OR KNOW GRAMPOSITIVE COCCI INFECTIONS

Gonzalez de Molina F J1, Sirvent J M1, Barbadillo S1, De Batlle J2, Motje M2, Gavalda L3, Vidaur L1, Bonet A1 1 Intensive Care Unit, 2Microbiology, 3Epidemiology, Hospital Universitari de Girona Dr. Josep Trueta, Girona, Spain

Alvarez-Lerma F1, Ucio P2, Gonzalez V3, Soler J4, Blanco A5, Leon C6, Infecciones por CGP G7 1Intensive Care Medicine, Hospital del Mar, Barcelona, 2Intensive Care Medicine, Hospital Clinico, Valladolid, 3Intensive Care Medicine, Hospital Miguel Servet, Zaragoza, 4Intensive Care Medicine, Hospital Insular Gran Canaria, Las Palmas, 5Intensive Care Medicine, Hospital Central, Oviedo, 6Intensive Care Medicine, Hospital de Valme, Sevilla, 7, Spain

INTRODUCTION. Pseudomonas aeruginosa (PA) is a nosocomial pathogen associated with increased morbidity and mortality, also the antibiotic resistance complicates the treatment of infections among ICU patients. The aim of this study was to evaluate the effect of empirical antibiotic treatment for ventilator-associated pneumonia (VAP) on antimicrobial resistance patterns of PA. METHODS. The strategy consisted in prescribing empirical antibiotic therapy with amoxicillin/clavulanate for early VAP, and cefepime plus amikacin for late VAP. This empirical antibiotic policy was based on results from a study made in the year 2000. The study analysed the density of incidence of VAP, aetiologic agents and its antimicrobial susceptibility patterns. Prospective resistance patterns of isolates of PA were collected over two years (2001-2002). We applied an Antibiotic Resistance Score (ARS) to each PA isolate, considering the resistance rate to amikacin (AMK), tobramycin (TOB), ceftazidime (CAZ), cefepime (CPE), ciprofloxacin (CP), imipenem (IPM), meropenem (MER), and piperacillin/tazobactam (PT). The PA which were susceptible to all anti-pseudomonal antibiotics were scored as ARS=0 (minimum score) while ARS=8 corresponded to the PA which were resistant to all of them (maximum score). RESULTS. Demographic data, severity scores, co-morbid conditions and mortality in 591 patients admitted in 2001 were similar to that found in 592 patients admitted in 2002. There were no differences in nosocomial infections or in VAP incidence rates or aetiological agents. Over a period of two years, 101 isolates of PA were identified. Resistance rates (2001 vs 2002) of PA isolates to each anti-pseudomonal antibiotic was: AMK (6% vs 2%), TOB (15% vs 13%), CAZ (20% vs 29%), CPE (9% vs 11%), CP (35% vs 22%), IPM (18% vs 13%), MER (8 vs 4%), and PT [37 vs 16%, p<0.05)]. The overall resistance rate scored by ARS expressed in mean ± SD (2001 vs 2002) was: 1,71 ± 1,79 vs 1,09 ± 1,37 (p<0,05). CONCLUSION. Two years after the implantation of an empirical antibiotic treatment policy for VAP, a decrease in the antibiotic resistance rates of PA was observed. This decrease was significant if all anti-pseudomonal agents were considered globally by the Antibiotic Resistance Score, and individually for piperacillin/tazobactam. Grant acknowledgement: Catalan Agency for Health Technology Assessment and Research (AATM 095/02/2000).

INTRODUCTION. Linezolid is the first of a new antimicrobial class active against all GPC, including methicilin and vancomycin resistant GPC. OBJECTIVE. To evaluate the efficacy and tolerability of Linezolid (LZD) compared to glycopeptides, Vancomycin (VCN) or Teicoplanin (TPN), in ICU patients METHODS. Phase III, multicenter, randomized (2:1:1), open-label and controlled clinical trial. Inclusion criteria: known or suspected infection caused by a multiresistant GPC in ICU patients; APACHE II score between 8 and 24. Analyzed populations: Intention to treat (IT), and clinically and microbiologically evaluable (CME). Analysis schedule: End of treatment. Statistical analysis: a Chi-square test for qualitative variables, and a Student t-test for quantitative variables. A logistic regression analysis was included in order to identify the factors influencing a successful outcome. RESULTS. 252 patients were included, (127 LZD, 63 VCN, 62 TPN). 86 patients were excluded of the efficacy analysis due to the absence of microorganisms or GPC. No significant changes were identified for the following variables: demographics, seriousness, infection appearance, infection type, and microorganisms involved, either on the IT as on the CME population. Main findings: Withdrawn from study due to treatment failure: LZD, 2,7%, VCN, 10%, TPN, 15,8% (p=0,02); Treatment with three or more antibiotics: LZD, 58,1%, VCN, 72,7%, TPN, 84,2% (p=0,02); days of treatment: LZD, 11,9, VCN, 13,8, TPN, 13,2 (p=0,08). The factors individually associated to a favorable outcome were: absence of diabetes mellitus, (OR 0,21 IC95% 0,0660,677); absence of abdominal infection, (OR 0,11 IC95% 0,016-0,781); days of treatment, (OR 0,1,36 IC95% 1,177-1,579); treatment with Linezolid, (OR 4,84 IC95% 1,593-14,688). CONCLUSION. In GPC infections, Linezolid was associated to a lower number of treatment failures. Linezolid constitutes a favorable and independent response variable comparing to glycopeptides Grant acknowledgement: Pharmacia Spain Laboratory

253

255

ANTIBIOTIC USE IN PATIENTS REQUIRING INVASIVE MECHANICAL VENTILATION

SHORT COURSE THERAPY FOR GRAM-NEGATIVE ICU-ACQUIRED BACTERAEMIA IN A MIXED MEDICAL-SURGICAL ICU

Palomar M1, Alvarez Lerma F2, Diaz E3, Sa Borges M4, Martinez Pellus A5, Burgueño M1, Grupo de estudio NAV G1 1 ICU, H Vall d´Hebron, 2ICU, H del Mar, Barcelona, 3ICU, H Joan XXIII, Tarragona, 4ICU, H Palma, Palma de Mallorca, 5ICU, H Arreixaca, Murcia, Spain

Corona A1, Wilson P1, Singer M 1 Bloomsbury Institute ICM, UCL, London, United Kingdom

INTRODUCTION. Antibiotic overuse is an important problem in the ICUs. One of the major reasons for prescribing antibiotics (ATB) to critical patients is respiratory infection (RI). Invasive mechanical ventilation (MV) is the most important risk factor for the developing of RI. METHODS. The objective of our work was to assess the characteristics of the patients requiring MV and the pattern of ATB´s use. A prospective multicenter cohort study was carried out in 17 ICUs during 12 months. All patients ventilated for more than 12 hours were included. RESULTS. There were 1704 patients of mean age 57.3 (18.1) years, male sex 66.5%, APACHE II at admission 18.2 (7.6); diagnostic category: medical 57%, scheduled surgery 14.9%, emergency surgery 15%, trauma 12.9%; 353 patients developed 403 episodes of ventilator associated pneumonia (VAP). The ICU LOS was 16.5 (1.6) days and mortality rate 38.2%. In 1520 patients (89.2%) a total of 4410 ATB were administered (2.9 ATB/ patient).The most frequently prescribed ATB were N (%): Amoxicilline-Clavunalate 353 (8); Piperacilline-Tz 333 (7.5); Cefepime 317 (7.1); Imipenem 333 (7.5); Vancomicine 288 (6.5); Amicacine 249 (5.6); Cefotaxime 240 (5.4) SDD 201 (4.7) and Ciprofloxacine 189 (4.2). The mean duration of treatment ranged between 7.2 and 13.3 days. The reasons for ATB treatment were recorded in 3592 cases, expressed in n (%): Documented infection in 2061(57,2); surgical prophylaxis 505 (14); VAP prophylaxis 414 (11.5), manoeuvres prophylaxis 20 (0.5), not defined 592 (16.4). ATB prescription was previous to intubation in 830 (19%) cases; after intubation and during MV for others reasons than VAP treatment in 2812 (84.5%) ATB, and finally to treat VAP in 713 (16,3%) ATB. CONCLUSION. Patients requiring invasive MV received ATB treatment in a very high proportion. Treatment of documented infections was the main reason to use ATB in these patients. Only 16% of the prescriptions were for the treatment of VAP. Grant acknowledgement: Bristol Myer Squibb

INTRODUCTION. No PRCT have been conducted to define optimal antibiotic therapy (Rx) for ICU-acquired bacteraemia (ICU-AB), though (Rx) may influence patient outcome. ICUs have evolved their (Rx) strategy using short (4-7 days) or long (10-14 days) courses of mono- or combination (Rx). We conducted a 6 month prospective observational audit to assess: (i) the incidence of Gram-negative (Gram-) ICU-AB and their antibiotic sensitivity pattern; (ii) the efficacy of our routine practice short courses mono-(Rx) in treating Gram- bacteraemia as judged by clinical response (CR), incidence of relapses and patient outcome. METHODS. From February 2000 over a 6 month period, all patients developing a Gram- ICUAB, had (i) the microorganisms identified, with their antibiotic sensitivity patterns; (ii) the collection of (Rx) type and duration, relapses, patient (CR) and outcome. Blood was taken for culture when clinically indicated. Microorganisms isolation and identification were made using standard techniques. Significance of isolates was assessed according to laboratory results and clinical findings. RESULTS. Among 713 patients admitted over the study period, 26 (3.2%) developed 27 GramICU-AB, with a median onset of 9 days (IQR 7-14); Klebsiella spp. (14) was the main microoganism isolated while central venous catheter the most frequent (8) bacteraemia source; septic shock was recorded in 15 (55%) episodes. Two (6.5%) Gram- isolates were multi-drug resistant (MDR) and no ESBL producing microorganisms were found. Five day (IQR 4-7) median course of mono-(Rx) was used for most (78%) of patients and was followed by clinical response in 67% of cases; 4 patients received combination (Rx), because of severe deep-seated infections and 2 were not treated for therapy withdrawal decision. Crude and bacteraemia related mortality were respectively 40% and 23%. CONCLUSION. Strategy of short course mono-Rx appears to provide a satisfactory (CR), with a low relapses rate. Moreover, MDR Gram- ICU-AB incidence was low. Multicentre PRCTs are warranted to compare short versus long courses, and mono-(Rx) versus combination (Rx).

S70

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

Oral Presentations Cardiovascular dynamics: The role of therapy (I) – 256-261 256 SEVERE THROMBOCYTOPENIA UNDER TREATMENT WITH GPIIB/IIIA-ANTAGONISTS, MORE OFTEN THAN EXPECTED? Wettstein T1, Schuiki E1, Straumann E H2, Bertel O1 1 Internal Medicine, Division of Cardiology, 2Internal Medicine, Intensive Care Unit, Zürich, Switzerland INTRODUCTION. GPIIb/IIIa antagonists (GPA) are part of standard treatment in patients (pts) with acute coronary syndromes (ACS), as well as in pts with complex percutaneous coronary interventions (PCI). Thrombocytopenia (TCP) with and without blee-ding are severe adverse effects of those drugs. Based on prospective randomised studies in selected pts severe TCP < 20’000 has to be expected in 0.3%. Is the incidence similarly low in every day routine? METHODS. We prospectively evaluated the incidence of GPA induced TCP in non selected pts under „real world“ conditions. Data of an ongoing registry were used. Pseudo-TCP was excluded by manual count. RESULTS. 632 consecutive pts were treated with PCI and GPA. In 11 pts (incidence 1.7%, 69.5+/-8.4 years, 81% men) a severe TCP occurred after a median of 6 (2-33) hours with a minimal TC count of 7000+/-4147/mm3, starting from 211’090+/-54119/mm3 before PCI. In 10 pts the indication for PCI and GPA was an ACS, in 1 pt stable disease with a complex PCI was present. 8(73%) were treated with Abciximab (Abc), 2(18%) with tirofiban (Tiro), 1 pt received Tiro after a previous allergic reaction of Abc. In 2 other pts there was an Abc re-exposition. In 6 pts unfractionated full dose heparin, and in 5 other pts full dose low molecular weight heparin was given. Heparin induced TCPwas very unlikely in all pts because of the temporal relationship and a re-exposition in 2. There was no intracranial or haemodynamically compromising bleeding. However 7 pts suffered from minor to moderate bleedings (6 at punction site, 1 haematemesis, 1 epistaxis, 1 macrohematuria). In 6 pts transfusions were necessary. All 11 pts received 1 to 2 units of erythrocyte in-fusions after discontinuation of GPA therapy. Therapy with Tc-aggregation inhibitors (asirin or combination of aspirin and clopidogrel) was continued as planned before.

258 BLOOD PRESSURE REGULATION: ROLE OF ENDOTHELIN, VASOPRESSIN AND ANGIOTENSIN Picker O1, Schwarte L A1, Schindler A W1, Scheeren T W L1 Anaesthesiology, Duesseldorf, Germany

1Anaesthesiology,

INTRODUCTION. To clarify the role of endothelin, vasopressin, and angiotensin on blood pressure regulation during anaesthesia with increasing concentrations of sevoflurane (1-3 MAC). METHODS. On different days, six awake dogs (equipped with flowprobes around the pulmonary artery) underwent each of the following four interventions: sevoflurane anaesthesia alone (1-3 MAC), or sevoflurane after blockade of either endothelin receptors using tezosentan (3 mg/kg followed by 3 mg·kg-1·h-1), or vasopressin V1a receptors using [d(CH2)5Tyr(Me2)]AVP, (40g/kg), or angiotensin receptor blockade using losartan (6 mg·kg-1·h-1). Baroreflex sensitivity was calculated as the quotient of change in heart rate and carotid sinus blood pressure after bilateral carotid occlusion (45s) (1). Plasma concentrations of endothelin, big-endothelin, vasopressin, and renin were measured. Effects of sevoflurane in the presence and absence of respective receptor blockade were analyzed (means ± SEM) and compared by an analysis of variance for repeated measures (ANOVA, followed by Fisher’s PLSD and paired t-test, p<0.05). RESULTS. Mean arterial pressure decreased concentration dependently during all interventions. At 1 MAC, this decrease was largest during angiotensin receptor blockade (-41±3 mmHg), intermediate during vasopressin and endothelin receptor blockade (-31±4 mmHg and -30±2 mmHg), and least during sevoflurane alone (-24±3 mmHg). The course of systemic vascular resistance mirrors the course of blood pressure, while cardiac output did not differ between groups. Plasma concentrations of endothelin, big-endothelin, and renin did not change during any intervention, while vasopressin concentrations increased from ~1 ng·l-1 to ~40 ng·l-1 as blood pressure decreased. Sevoflurane suppressed baroreflex sensitivity without a difference between groups. CONCLUSION. Endothelin, vasopressin, and angiotensin contribute to different amounts to the maintenance of blood pressure during sevoflurane anesthesia. However, only vasopressin is specifically activated to counterbalance further blood pressure decreases. This effect is independent of the arterial baroreflex. REFERENCE(S). 1. Ohsumi H, Sakamoto M et al. Am J Physiol 1989;625-31.

CONCLUSION. Acute severe TCP under treatment with GPA was much more common in non selected patients as it was demonstrated in randomised studies. However, life threatening and intracranial bleeding did not occur, despite continuation of treatment with aspirin. Because Tcpenia may occur very soon after starting GPA treatment, Tc counts should be performed as early as 4 hours after initiation of treatment and again after 16 hours.

257

259

EFFECTS OF NAC AND PRECONDITIONING ON HAEMODYNAMICS AND INDOCYANINE GREEN CLEARANCE

DECREASED MESENTERIC AND PORTAL BLOOD FLOW DURING ADMINISTRATION OF VASOPRESSIN IN SEPTIC SHOCK

Szakmany T1, Ghosh S1, Baumann J1, Ferencz A2, Jancsó G2 1 Anaesthesiology and Intensive Care, 2Experimental Surgery, University of Pecs, Pecs, Hungary

Krejci V1, Hiltebrand L B1, Ten Hoevel M1, Sigurdsson G H2 1 Department of Anesthesiology, Inselspital, University Hospital, Bern, Switzerland, 2Department of Anesthesia and Intensive Care Medicine, Landspitali University Hospital, Reykjavik, Iceland

INTRODUCTION. The aim of the study was to investigate whether repeated ischaemic preconditioning or N-acetylcysteine (NAC) prevent ischaemic-reperfusion injury as determined by indocyanine green plasma disappearance rate (ICG-PDR) or have favourable haemodynamic effects during reperfusion in canine livers (1,2). METHODS. In a prospective, randomized experimental study 15 mongrel dogs were randomized into three groups. The control group (n=5) underwent 60 minutes of hepatic ischemia followed by 180 minutes reperfusion. In the NAC group (n=5) 150 mg kg-1 of NAC was administered intravenously before inducing ischemia. In the preconditioned group (n=5) animals received ischemic preconditioning (10 minutes of ischemia followed by 10 minutes of reperfusion repeated three times) before clamping the portal triad. Haemodynamic parameters, ICG-PDR and serum levels of aspartate transferase (AST) and alanine transferase (ALT) were recorded at baseline, 5 minutes before reperfusion, and 5, 30, 60, 180 minutes after reperfusion. RESULTS. 13 dogs survived the study period. 2 dogs in the NAC group died due to circulatory failure unresponsive to inotropic drugs. The cardiac index and the intrathoracic blood volume index were significantly higher in the preconditioning group compared to the controls throughout the study period. ICG-PDR was similar in each group throughout the study. Serum levels of AST and ALT did not show significantly different values between the groups. CONCLUSION. Repeated ischemic preconditioning might improve haemodynamic parameters, whereas we were unable to find any significant differences between the groups regarding hepatic cell functions as monitored by ICG-PDR and AST, ALT levels. REFERENCE(S). 1. Clavien PA, Yadav S, Sindram D, Bentley RC. Protective effects of ischemic preconditioning for liver resection performed under inflow occlusion in humans. Ann Surg 2000; 232:155-162 2. Okochi O, Kaneko T, Sugimoto H, Inoue S, Takeda S, Nakao A. ICG pulse spectrophotometry for perioperative liver function in hepatectomy. J Surg Res 2002; 103:109-113 Grant acknowledgement: This work was supported by NKFP 1A/0026 research grant, Ministry of Education, Hungary

INTRODUCTION. Vasopressin has been proposed for treatment of catecholamine-resistant septic shock (1), however, very little is known about its effects on the distribution of splanchinic blood flow. The aim of this study was to measure splanchnic blood flow during administration of ornithin-8-vasopressin in septic shock. METHODS. Pigs (20 - 25 kg, n = 16) were anaesthetised and ventilated. Blood flow was measured in the superior mesenteric artery, portal vein and hepatic artery using ultrasonic transit time flowmetry. Peritonitis and septic shock was induced by instillation of autologous feces in the peritoneal cavity. After 240 min, intravenous (i.v.) colloids were administered to transform hypodynamic shock into hyperdynamic septic shock. After 300 min, group V (n=8) received 0.06 IU/kg/h of ornithin-8-vasopressin continuously i.v. during 180 min while group S (n=8) received isotonic saline. RESULTS. Baseline measurements were taken at T = 300 min. Mean arterial blood pressure increased by 25% (p<0.05) during infusion of vasopressin in the group V (P<0.05), while it remained unchanged in group S. Cardiac output decreased by 30% in group V (p<0.05) but remained unchanged in group S. Blood flow in the superior mesenteric artery and in the portal vein decreased by 50% in group V (p<0.05), which was significantly more than in group S. The maximal decrease in portal and mesenteric blood flow occurred within 60 min of starting the vasopressin infusion. Blood flow in the hepatic artery increased by 50% (p<0.05) in group V while there was a 20% decrease in group S. CONCLUSION. In this porcine model of fluid resuscitated septic shock, administration of vasopressin resulted in significantly decreased superior mesenteric artery- and portal venous blood flow. In contrast, blood flow in the hepatic artery increased, suggesting that the hepatic arterial buffer response was involved. REFERENCE(S). 1. O’Brien et Al. Terlipressin for norepinephrine-resistant septic shock. Lancet, 2002. 359(9313): p. 1209-10. Grant acknowledgement: Supported in part by the IFA Research Foundation, Berne, Switzerland.

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

260 PERFUSION PRESSURE MANIPULATION IN PORCINE SEPSIS: EFFECTS ON INTESTINAL HEMODYNAMICS Krouzecky A1, Matejovic M1, Rokyta R1, Novak I1 1ICU, Dept of Internal Medicine I, Charles University Hospital Plzen, Plzen, Czech Republic INTRODUCTION. Sepsis-induced vasoplegia may result in a wider range of pressure-dependent tissue perfusion. It is still unknown, whether the conventional trigger for the initiation of vasopressor therapy (mean arterial pressure < 65 mmHg) is adequate, particularly for the gut. In this pilot study we investigated whether increasing perfusion pressure in a „normotensive“, fluidresuscitated, hyperdynamic porcine bacteremia affects intestinal macro- and microcirculation. METHODS. Five hydroxyethylstarch-resuscitated pigs received continuous i.v. pseudomonas aerugionsa to develop hyperdynamic, normotensive (MAP > 65mmHg) sepsis. Noradrenaline (NA) was used to achieve 10-15% increase in MAP. Mesenteric arterial blood flow (Qgut, Doppler ultrasound probe), ileal microvascular perfusion (LDFgut, laser doppler) and ileal-endtidal PCO2 gap (air tonometry) were measured before NA (TP1), after 60 min of NA infusion (TP2) and 60 min after NA cessation (TP3). RESULTS. Data are medians and interquartile range, P<0.05 * vs. TP1, § vs.TP2 (ANOVA on Ranks).

MAP mmHg Qgut ml/min/kg LDFgut units PCO2gap mmHg

TP1

TP2

TP3

82 (65;85)

94 (72;96) *

73 (53;82) **

9 (8;15)

11 (9;15)

8 (5;13)

108 (66;170)

116 (58;150)

106 (37;134)

23 (21;34)

20 (18;36)

21 (18;42)

CONCLUSION. In this hyperdynamic, normotensive porcine bacteraemia, noradrenaline-induced 10-15% increase in perfusion pressure affected neither intestinal macrocirculatory blood flow nor ileal mucosal microcirculation. It seems, at least in this model, that the gut perfusion is not pressure dependent over the range of 65-95 mmHg. Grant acknowledgement: Supported by a research grants IGA MZ CR ND 6989-3/2002 and ND 6837-3/2001

S71

Oral Presentations Aspects of emergency and trauma medicine – 262-267 262 EVALUATING TRANSPORT - SEVERITY SCALE IN INTER-HOSPITAL TRANSPORTS. Samiotis I1, Dalezios M1, Markakis C1, Tsitsiridis M1, Xalkidaki A1, Koukou E1, Politi K1, Agouridakis P1 1 Intensive care unit, General Hospital, Rethymno, Greece INTRODUCTION. Inter-hospital transport of critically ill patients is common even in countries with well developed health care systems. There is an ongoing debate about risk factors and the safety of the transfers. The aim of this study was to evaluate a recently suggested scale of risk of transport1. METHODS. The study was of a descriptive prospective design and included all the patients transferred during one year from a 200 bed Community Hospital (referring hospital) to a University Hospital (receiving hospital). Transported patients were classified into two groups of severity according to the suggested scale of risk of transport: Patients summed less than 7 points in that scale were defined as Group I (non critically ill patients) and those summed 3 7 points were defined as Group II ( critically ill patients). Major complications during transport, ICU admission and ICU mortality rate were the endpoints of the comparison between these Groups of patients. The data were analyzed using the t- test and x2. RESULTS. Overall 124 patients were transported during the study period, 65 of them (52.4%) belonged to Group I and 59 (47.6%) belonged to Group II. Men outnumbered women in both groups. The transport-severity score (mean ± SD) significantly differed between two groups: 11.2 ± 3.4 for Group II vs 4.5 ± 1.02 for Group I (p < 0.005). 73% of Group II patients were intubated prior to transfer. Major en route complications were reported in 18.9% of Group II patients but in none of Group I (p<0.000). The most common complication was cardiovascular instability (67.5%). No patients died during transport. 15.4% of Group I patients were admitted into the ICU of the receiving hospital on the day of transport vs 100% of Group II (p< 0.005). ICU mortality rate was 4.7% for Group I vs 30% of Group II (p< 0.005). ICU mortality could not be directly related to major en route complications. CONCLUSION. Inter-hospital transport of critically ill patients is a rather safe procedure in the presence of expertise personnel and advanced life support devices. Scales of severity could help in transport team selection. REFERENCE(S). Etxebarria MJ,et al,Prospective application of risk scores in the interhospital transport of patients. Eur Journal of Emerg Med 1998, 5, 13 -17.

261

263

EARLY GOAL-DIRECTED THERAPY IN PATIENTS WITH END-STAGE RENAL DISEASE ON HEMODIALYSIS

IS ALBUMIN ADMINISTRATION ASSOCIATED WITH WORSE OUTCOME ?; RESULTS OF THE SOAP STUDY

Donnino M W1, Nguyen B1, Bawja R1, Jacobsen G1, Tomlanovich M1, Rivers E1 Emergency Medicine, Henry Ford Hospital, Detroit, United States

Vincent J L1, Sakr Y1, Payen D1, Carlet J1, Gerlach H1, Renieri V M1, Moreno R1, Reinhart K1, Sprung C1, Le Gall J R1 1 Department of Intensive Care, Erasme hospital, Free University of Brussels, Brussels, Belgium

1

INTRODUCTION. Patients with End-Stage Renal Disease (ESRD) on dialysis are at increased risk for infection, yet few studies in sepsis focus on this population. The current study evaluates early, goal-directed therapy (EGDT) for ESRD patients with severe sepsis and septic shock.

INTRODUCTION. Albumin administration is a highly debated topic. We investigated the possible association of albumin administration and outcome.

METHODS. This is a post-hoc analysis of a randomized, prospective trial of patients presenting to an urban ED with severe sepsis and septic shock. Patients were enrolled in the study if they had two or more SIRS criteria plus hypotension or lactic acidosis (> 4 mmol/liter). Patients were randomized to either conventional care (inclusive of central venous pressure [CVP] monitoring) or to EGDT which included resuscitation to goals of CVP between 8-12, mean arterial pressure (MAP) between 65 and 90, and central venous oxymetry (ScvO2) greater than 70%.

METHODS. This cohort, multicentric, observational study included all 3147 adult patients admitted to 198 European ICUs between May 1 and May 15, 2002. Patients were followed up until death, hospital discharge, or 60 days. Patients were classified according to administration of albumin at any time during ICU stay into albumin group and others who had never received albumin. Propensity score case-matching was performed and matched pairs were examined for baseline characters and outcome.

RESULTS. Ten patients with ESRD were in the control group and 7 in the treatment group. Baseline haemodynamic (HR, MAP, CVP, ScvO2) and severity of illness scores (APACHE, MODS, SAPS) were not different between groups. 7/10 patients in the control group died compared to 1/7 in the treatment group (p < .05). At six hours, HR and CVP remained equal between groups, but a statistically significant rise in ScvO2 and MAP was found between groups (p < .05). Additionally the treatment group displayed a significant fall in lactate, APACHE, MODS, and SAPS scores (p < .05) and was given more intravenous fluid than the control group (p < .05). The difference in ScvO2, lactate, and severity of illness scores remained significant (p < .05) at 72 hours.

RESULTS. Of 3147 patients, 354 (11.2 %) received albumin and 2793 (88.8 %) had never received albumin. Patients who received albumin had a higher incidence of cancer and liver cirrhosis, more commonly medical admissions, had higher SAPS II and SOFA scores, and higher incidence of sepsis syndromes on admission. As expected, both ICU and hospital mortality rates were higher in the albumin group (35.3 vs. 16.4, and 41.5 vs. 21.3, %, p<0.001) than in the other patients. In a multivariate analysis, albumin administration was associated with an independent risk of ICU death (relative hazard = 1.4, 95% confidence interval: 1.2-1.8, p<0.001) at 60 days. Moreover, in a 343 matched-pairs according to a propensity score, ICU and hospital mortality rates were higher in patients who received albumin than others (34.7 vs. 20.7 and 41.1 vs. 25.7 %, p<0.001). Survival at 30 days were lower in the albumin group than its matched group (Log Rank = 6.9; p < 0.001).

CONCLUSION. EGDT is both safe and efficacious for patients with ESRD resulting in reduction of both morbidity and mortality. The morbidity and mortality can be explained by eradication of global tissue hypoxia via aggressive resuscitation guided by goal-directed care. Although EGDT was performed for only six hours in the ED, the physiologic benefits (increased ScvO2/reduced lactate/reduced severity of illness) persisted between groups at 72 hours.

CONCLUSION. In this observational study, albumin administration was associated with increased mortality in a heterogeneous ICU population. Prospective randomized trials are warranted to study the effect of albumin administration in ICU patients. Grant acknowledgement: The study is supported by unrestricted grant from: Abbott Laboratories, Baxter, Eli Lilly, GlaxoSmithKline, and Novo Nordisk

S72

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

264

266

TREATMENT OF CHEST TRAUMA, WITH NONINVASIVE POSITIVE PRESSURE VENTILATION , IN INTENSIVE CARE UNIT

PET STUDIES OF CEREBRAL BLOOD FLOW AFTER RESTORATION OF SPONTANEOUS CIRCULATION FROM CARDIAC ARREST.

Xirouchaki N1, Kondili E1, Mouloudi H1, Kondoudaki E1, Anastasaki M1, Georgopoulos D1 Heraklion University Hospital, Heraklion Crete, Greece

1Department

1ICU,

INTRODUCTION. Non-invasive positive pressure ventilation (NIPPV), has an established role in the intensive care unit in providing respiratory support in patients with acute respiratory failure. The significant advantage of NIPPV is to avoid endotracheal intubation and its complications. Aim: To investigate the efficacy of NIPPV in patients with chest trauma who were admitted in Intensive Care Unit due to hypoxemic respiratory failure. METHODS. Fourteen patients with chest trauma (ten males and four females), aged 45+22 years were studied. NIPPV was applied via a tight fitting face mask, combined with regional anaesthesia in all patients. RESULTS. Four patients required endotracheal intubation (28%), after 4,5+4 hours of NIPPV. Data (heart rate, respiratory rate, PaO2 and PaCO2), at baseline and one hour after the onset of NIPPV were statistically significant different between success (not-intubated) and treatment failure (intubated), group. Heart rate, respiratory rate, pO2, and pCO2, were significantly higher at baseline and one hour after in patients of the treatment failure group. Only one patient (7%), from treatment failure group (intubated), was died in Intensive Care Unit. CONCLUSION. NIPPV could be an effective treatment to avoid endotracheal intubation in patients with chest trauma.

Mörtberg E1, Wiklund L1, Rubertsson S1 of Surgical Sciences - Anesthesiology and Intensive Care, Uppsala University, Uppsala, Sw, University hospital, Uppsala, Sweden INTRODUCTION. The cerebral blood flow after restoration of spontaneous circulation (ROSC) from cardiac arrest shows great alterations and has previously been measured with laser Doppler technique [1]. In order to better assess cerebral blood flow in the whole brain and detect regional differences we studied the cerebral blood flow after ROSC with positron emission tomography (PET).

METHODS. Ten anesthetized piglets were subjected to 5 min of ventricular fibrillation followed by closed-chest cardiopulmonary resuscitation (CPR). Bolus doses of epinephrine (20g/kg) were given i.v. at 2, 5 and 8 min of CPR. After 9 min of CPR, external defibrillatory shocks were applied to achieve ROSC. Cerebral blood flow was measured with PET utilizing the 15O-labeled water bolus technique [2, 3]. The PET scans were performed at baseline as well as 10, 30, 60 and 240 min after ROSC. RESULTS. Restoration of spontaneous circulation was achieved in 6 piglets. The mean of the cerebral blood flow at baseline was 28±7 ml/100ml/min. The cerebral blood flow increased dramatically in three piglets, 10 min after ROSC, and reached values above 56 ml/100 ml/min. Thirty min after ROSC, cerebral blood flow decreased in all but one piglet. At 240 min, the cerebral blood flow was close to baseline values. Regional differences in cerebral blood flow were obvious but difficult to measure in a standardized way. There was a positive correlation between mean arterial pressure and cerebral blood flow (r=0.6). CONCLUSION. The variations in global cerebral blood flow as measured with PET are similar to those previously measured with the laser Doppler technique. There is a hyperemic phase directly after ROSC followed by a hypoperfusion and, thereafter a slow recovery of the cerebral blood flow to almost baseline values at 240 min after ROSC. REFERENCE(S). 1. Nozari, A., S. Rubertsson, and L. Wiklund, Differences in the pharmacodynamics of epinephrine and vasopressin during and after experimental cardiopulmonary resuscitation. Resuscitation, 2001. 49(1): p. 59-72. 2. Herscovitch, P., J. Markham, and M.E. Raichle, Brain blood flow measured with intravenous H2(15)O. I. Theory and error analysis. J Nucl Med, 1983. 24(9): p. 782-9. 3. Raichle, M.E., et al., Brain blood flow measured with intravenous H2(15)O. II. Implementation and validation. J Nucl Med, 1983. 24(9): p. 790-8.

265

267

MORTALITY REDUCTION AFTER IMPLEMENTING AN ICU PROTOCOL FOR SEVERE TRAUMATIC BRAIN INJURY

PROGNOSTIC VALUE OF UNILATERAL AND BILATERAL RECORDING OF EVOKED POTENTIALS FOLLOWING CARDIAC ARREST

Arabi Y1, Haddad S1, Al-Ferayan A2, Al-Shimemeri A1 Intensive Care Unit, 2Department of Surgery, King Fahad National Guard Hospital, Riyadh, Saudi Arabia

Bauer E1, Funk G C2, Gendo A2, Kramer L2, Zauner C2, Sterz F3, Madl C2 1 Department of Medicine III, 2Department of Medicine IV, 3Department of Emergency Medicine, General Hospital Vienna, Vienna, Austria

INTRODUCTION. The purpose of this study is to examine the impact of a standardized protocol for the ICU management of traumatic brain injury (TBI) on patients’ outcome.

INTRODUCTION. Evidence of a bilateral loss of cortical short-latency somatosensory evoked potential peaks (N20) in comatose cardiac arrest survivors has a high prognostic value in predicting a bad outcome following cardiopulmonary resuscitation. Hypoxic-ischemic brain damage in comatose cardiac arrest survivors is considered as global, as blood flow in the brain ceases altogether. We hypothesized that an unilateral assessment of cortical short-latency somatosensory evoked potentials has the same prognostic reliability as a bilateral recording.

1

METHODS. In February 2001, a protocol for the management of severe TBI was introduced to our ICU based on the guidelines published by the Brain Trauma Foundation. The main components of the protocol were maintaining euvolemia using isotonic fluids, Cerebral Perfusion Pressure (CPP) > 70 mm Hg or Mean Arterial Pressure (MAP) > 80 mm Hg, normocapnia and normothermia. Data was extracted from a prospectively collected ICU database before the protocol (March 1999 to January 2001) and after the protocol (February 2001 to November 2002). Demographics, Injury Severity Score (ISS), APACHE II Score and the use of Intracranial Pressure (ICP) monitoring were documented. The primary outcomes were ICU and hospital mortality. Secondary end points were ICU length of stay (LOS) and duration of mechanical ventilation. Ttest and Chi-square were used as appropriate. Continuous variables are expressed as means ± SEM. T-test and Kruskal-Wallis tests were used as appropriate. RESULTS. The groups before the protocol (n=136) and after the protocol (n=134) were similar in the age: (35.0 ± 1.5 vs. 31.6 ± 1.9), APACHE II: (18.1 ± 0.6 vs. 19.2 ± 0.5), and ISS Scores: (31.7 ± 1.1 vs. 30.1 ± 1.1). There was no significant difference in the duration of mechanical ventilation or ICU LOS after implementing the protocol. However, there was a significant reduction in ICU mortality from 23% to 9% (P=0.002) and hospital mortality from 30% to 13% (P<0.001). This improvement in outcome was seen whether ICP monitoring was used or not. CONCLUSION. The use of a standardized protocol for the management of severe traumatic brain injury is associated with significant improvement in patients‘ outcome. This data emphasizes the importance of standardized ICU management for the prevention of secondary brain insults.

METHODS. Somatosensory evoked potentials (SEPs)were recorded in 313 comatose cardiac arrest survivors within 72 hours following cardiopulmonary resuscitation. We measured shortlatency and long-latency SEPs bilaterally. Excluded were patients with an underlying neurological disease. Patients with a Cerebral Performance Category score > 2 at 1 year were defined as patients with hypoxic-ischemic brain damage (n=240). RESULTS. All patients with an unilateral or bilateral loss of the cortical N 20 peak (n= 108) had a poor cerebral outcome. All 9 patients who showed an unilateral loss of detectable N20 peak also had a poor neurological outcome. The cortical N 20 peak was bilaterally detected in all 73 patients with a good neurological outcome, but also in 132 patients with a CPC >2. Sensitivity and specificity of unilateral and bilateral recording of cortical SEPs

Sensitivity Specificity

Unilateral loss of N20 43 % 100 %

Bilateral loss of N20 42 % 100 %

CONCLUSION. An unilateral loss of cortical short-latency somatosensory evoked potentials in comatose cardiac arrest survivors has a comparable high prognostic value in predicting a bad outcome than a bilateral loss. Therefore, we think that in cardiac arrest survivors without a preexisting neurological disease an unilateral recording of short-latency cortical somatosensory evoked potentials is sufficient to predict a bad neurological outcome.

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

Oral Presentations Intra-abdominal pressure – 268-270 268

270 EFFECT OF INTRA-ABDOMINAL PRESSURE ON PLEURAL AND FILLING PRESSURES

VALIDATION OF A NOVEL FULLY AUTOMATED CONTINUOUS METHOD TO MEASURE INTRA-ABDOMINAL PRESSURE (IAP) Malbrain M L N G1 1 Medical ICU, ACZA Campus Stuivenberg, Antwerpen, Belgium INTRODUCTION. Definitions for intra-abdominal hypertension and abdominal compartment syndrome stand or fall with the correct and reproducible measurement of IAP. The current goldstandard intra-bladder pressure has its inherent problems with zeroing, the conductive fluid column, air bubbles, or intrinsic bladder wall tension. The aim of this study is to validate a novel fully-automated continuous technique to measure IAP via a balloon-tipped catheter connected to an IAP monitor (Spiegelberg, Hamburg, Germany). METHODS. IAP was calculated using 2 different METHODS. via the bladder (IAPves) and via a balloon-tipped gastric catheter (IAPgas) connected to an IAP monitor. In total 112 paired measurements were performed in 14 ICU patients. In addition 46 paired IAPgas and insufflator pressure (Pins) measurements were performed in 2 laparoscopic pts. The M/F ratio was 3/4, age 69.7±14, APACHE-II 26.1±9, SAPSII 59.6±13.7. RESULTS. The values for IAP (mmHg) were 11±3.6 (IAPgas) vs 10.1±3.9 (IAPves) and 13.3±4.1 (IAPgas) vs 12.4±4 (IAPins). There was a good correlation between IAPgas and IAPves: IAPves= 0.9xIAPgas +0.6 (R2=0.63, p<0.0001). Bland and Altman analysis showed good agreement: IAPgas was almost identical to IAPves with a mean bias of 0.8±2.4(SD) mmHg (95%CI 0.3 to 1.3); the limits of agreement (LA) were –4 to 5.6 mmHg (95%CI –4.8 to –3.2 for the LLA and 4.8 to 6.4 for the ULA). Correlation was better between IAPgas and IAPins: IAPgas= IAPins +0.8 (R2=0.96, p<0.0001). Bland and Altman analysis showed good agreement: IAPgas was almost identical to IAPins with a mean bias of 0.9±0.8(SD) mmHg (95%CI 0.6 to 1.2); with small LA -0.7 to 2.5 mmHg (95%CI -1.1 to –0.3 for the LLA and 2.1 to 2.9 for the ULA). CONCLUSION. This novel method has numerous advantages over the classic bladder method: it is simple and requires less manipulation hence it is less time-consuming. It is fully automated with auto-calibration every hour, hence giving 100% accuracy and reproducibility. Since a continuous trend of IAP and APP can be obtained it is cost effective. Since it is an air-filled system there are no problems related to the hydrostatic fluid column (zeroing, air-bubbles with over- and underdamping). There is no risk for needle stick injuries or infection. The technique does not interfere with urine output and is body position independent. This puts it in favour of becoming the new gold standard for selected patients and multicentre research purposes.

269 EFFECTS OF INCREASED INTRA-ABDOMINAL PRESSURE ON THE RESPIRATORY SYSTEM IN HEALTY AND SEPTIC PIGS Lattuada M1, Fredén F2, Reinius H2, Hedenstierna G1 1 Medical Sciences, Clinical Physiology, 2Anesthesiology and Intensive Care, Anesthesiology and Intensive Care, Uppsala, Sweden INTRODUCTION. Increased intra-abdominal pressure (IAP) is a common event in ICU patient. Acting through the diaphragm, high IAP can have an impact also on the respiratory system. We have tested, both in healthy and endotoxin-exposed animals, the effects of increased IAP on gas exchange, lung mechanics and ventilation/perfusion [VA/Q] relationship. METHODS. 12 anesthetized and mechanically ventilated pigs (27.9 ± 1.3 Kg, 6 each group) were studied. Septic damage was induced by continuous infusion of endotoxin (LPS E.Coli); IAP was increased to 20 mmHg for two hours by intra-peritoneal CO2 inflation (pneumoperitoneum, PP). Central haemodynamics, gas exchange, respiratory system mechanics and VA/Q (MIGET technique) were measured at different time points (before, during and after increase in IAP; in septic pigs, PP followed LPS infusion). RESULTS. Increased IAP reduces oxygenation in both healthy and LPS exposed pigs: PaO2/FiO2 decreased from 302 to 205.9 * at 120 min. and from 487 to 402.3* (* p<0.05) in LPS and healthy pigs respectively. Increase in IAP increases pulmonary shunt (table 1). A correlation was found between pulmonary shunt and PaO2/FiO2 ratio (r = -0.88 and -0.82 in LPS and healthy pigs, p<0.001). Static compliance of the respiratory system decreased significantly with IAP increase in both groups. Pulmonary Shunt (%)

LPS (SD) Healthy (SD)

Baseline LPS IAP 60 min IAP 120 min 2.3 (1.7) 9.3* (5.6) 15.9* (8.5) 16.3* (6.7) 3.7 (4.6) 8.7 (9.1) 8.8* (8.6) * p<0.05 vs baseline; * p<0.05 vs LPS

S73

Deflation 13.5** (6.1) 5.3 (2.8)

CONCLUSION. Increased IAP can have severe effects on the respiratory system, with a greater impact under septic conditions, and increased shunt seems to play a central role in oxygenation impairment. Pathophysiological changes are only partially reversible with deflation.

Malbrain M L N G1, Nieuwendijk R1, Verbrugghe W1, Deeren D1, Darquennes K1, Daelemans R1, Lins R1 1 Medical ICU, ACZA Campus Stuivenberg, Antwerpen, Belgium INTRODUCTION. Intra-abdominal hypertension (IAH) and the addition of PEEP compromises cardiovascular function (1). Correct estimation of preload is crucial. Volumetric assessment with PiCCO or use of transmural pressures=endexpiratory (ee) – pleural pressure (Ppl) might better reflect the volemic status. This study looks at the effects of increasing PEEP and IAP on Ppl. METHODS. 244 paired measurements were performed in 5 ICU pts. M/F ratio: 1/4, age 72.8±5.3, APACHE-II 30.4±4.3, SAPS-II 62.4±7.3. Correlations were studied at PEEP 0 to 20 with 2.5 increments. Transmural [tm] CVP was calculated in 4ways: CVPtm=CVPee-Ppl, CVPta=CVPee-IAP, CVPest=CVPee-ItxPEEP (where It=DeltaCVP/DeltaPalv, DeltaCVP=CVPeiCVPee and DeltaPalv=Pplat-PEEP)(2), CVPform=CVPee +/- regression formula, PAOPtm was calculated in the same way. RESULTS. The values for IAP (mmHg) were 10.1±4 vs 10.2±3.4 for Ppl with significant correlation (n=244): Ppl = 0.8xIAP+1.6 (R2=0.8, p<0.0001). Bland and Altman analysis showed that IAP was almost identical to Ppl with a bias of 0.8±1.7 (SD) mmHg (95%CI 0.6to1.1); the limits of agreement (LA) were –2.6to4.2(95% CI -3to–2.2 for the LLA and 3.8to4.6 for the ULA), these small intervals reflect good agreement. Regression analysis identified this formula for CVPform= 0.9xCVPee –0.1xPEEP –0.7xIAP –0.4, and for PAOPform=0.8xPAOPee –0.1xPEEP –0.6xIAP +0.02xCdyn +2.5. Significant (p<0.0001) correlations were found in 207 paired measurements for CVPta-CVPtm (R2=0.8), CVPest-CVPtm (R2=0.5) and CVPform-CVPtm (R2=0.9). The same was true for 90 paired measurements for PAOPta-PAOPtm (R2=0.9), PAOPest-PAOPtm (R2=0.7) and PAOPform-PAOPtm (R2=0.95). Bland and ALtman analysis showed good agreement for CVPta/PAOPta and CVPtm/PAOPtm and for CVPform/PAOPform and CVPtm/PAOPtm, but agreement was poor for CVPest/PAOPest and CVPtm/PAOPtm. CONCLUSION. IAP and Ppl are significantly correlated. This has important implications in daily clinical practise in the ICU. In patients with IAH and PEEP, transabdominal filling pressure (CVPee/PAOPee –IAP) can be used to obtain a quick idea of transmural filling pressure and thus preload. The transmural pressures calculated with the regression formula reflect even better real transmural pressures with an R2 of 0.9 for CVP and 0.95 for PAOP REFERENCE(S). (1) Malbrain M(2001) In: Vincent J(Ed) Yearbook, Springer-Verlag, Berlin, pp547. (2) Teboul et al. Crit Care Med 2000;28:3631

Oral Presentations Assessing quality of care in septic patients – 271-273 271 SITE EXPERIENCE AND SOURCES OF VARIABILITY IN A PHASE III CLINICAL TRIAL OF SEVERE SEPSIS Dhainaut J1, Macias W L2, Vallet B3 1 Medical Intensive Care Unit, Paris V University, Paris, France, 2Lilly Research Labs, Eli Lilly and Company, Indianapolis, United States, 3Department D’anesthesie et Reanimation Medicale II, University Hospital of Lille, Lille, France INTRODUCTION. Clinical studies in disease states with substantial patient heterogeneity may be subject to differential treatment effects among patient groups, reducing ability to detect a treatment signal. The successful PROWESS trial showing efficacy of drotrecogin alfa (activated)[DrotAA] in severe sepsis allows an opportunity to identify and assess unanticipated sources of variability in the estimate of treatment effect. METHODS. Prospective subgroups were examined to identify sources of variability in the treatment effect estimate. Treatment-subgroup interactions were considered significant if a subgroup had a Breslow-Day interaction P < 0.20. Interactions driven by small group size (<50) or the category Unknown (e.g. myocardial infarction YES NO UNKNOWN) were not further examined. Impact of site experience was assessed by removing the first block of randomized patients (1 block = 4 patients: 2 DrotAA, 2 placebo). RESULTS. Variability in the estimate of treatment benefit was seen with subgroups defined by: (1)presence/absence of a significant protocol violation (VIOL)(p=0.07), most frequent violations were inclusion/exclusion criteria and failure to deliver the appropriate dose of study drug; (2)original vs amended protocol (PROT)(p=0.08); (3)APACHE II quartile at baseline (AQRT)(p=0.09); and (4)home vs other locations prior to entering the study hospital (LOCN)(p=0.17). Removing the 1st block enrolled at each site reduced or eliminated the significance of all interactions (VIOL p=0.13; PROT p = 0.54; AQRT p = 0.19; LOCN p = 0.83). As enrolment at sites increased, rate of both violation types decreased and relative risk of death (RR) improved. For the 1st block (n = 559), 30% of patients experienced protocol violations and RR was 0.91(95% CI 0.70-1.17). For patients enrolled after a site 3rd block (n = 520), 17% experienced protocol violations and RR improved to 0.69 (0.52-0.91). CONCLUSION. Site experience appears more important than was realized but ‘adequate experience’ cannot be presently defined. Sites not enrolling at an acceptable rate and sites with multiple violations should be discontinued. These data suggest that an increased number of sites with few patients may reduce accuracy of treatment effect estimates. Grant acknowledgement: Research was supported by Eli Lilly and Company

S74

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

272 DIFFERENT DEFINITIONS IN SEPSIS TRIALS YIELD SIGNIFICANTLY DIFFERENT STUDY POPULATIONS Peelen L M1, De Keizer N F1, Peek N B1, De Jonge E2, Bosman R3, Scheffer G J4 Informatics, 2Intensive Care, Academic Medical Center, 3Intensive Care, OLVG, Amsterdam, 4Anaesthesiology, University Medical Centre, Nijmegen, Netherlands

1Medical

INTRODUCTION. Different definitions for severe sepsis are used in therapeutic trials. We have studied the effects of applying different definitions to identify patients with sepsis in a large dataset of ICU admissions. METHODS. We have selected six recent, large sepsis trials. In- and exclusion criteria of the studies have been applied to the Dutch National Intensive Care Evaluation1 (NICE) database, which contains data of 71,929 ICU admissions from 22 ICUs in the period from December 5, 1996 until February 4, 2003. The selection criteria of the trials were used to select populations of eligible patients. Of these populations we have determined the mean and a bootstrap estimate of the 95% confidence interval of the ICU- and hospital mortality and severity of illness (estimated by SAPS-II and APACHE-II scores). RESULTS. The number of patients in the database that fulfilled the in- and exclusion criteria of the studies and the ICU mortality (mean and 95% confidence interval) are shown in the figure. The amount of eligible patients differs remarkably, though all populations are described by ‘severe sepsis’ or ‘septic shock’. Note that overlap exists between the populations; a patient can meet the criteria of more trials. The populations show considerable differences in ICU mortality. Similar results are found for hospital mortality and severity of illness (results not shown).

Oral Presentations Quality of life after intensive care – 274-276 274 REASON FOR ICU ADMISSION INFLUENCES QUALITY OF LIFE AFTER LONGER STAY IN INTENSIVE CARE Chytra I1, Kasal E1, Lukáš J1, Pradl R1, Pelnáø P1, Židková A1 Dept. of Anaesthesiology and Intensive Care Medicine, University Hospital, Plzen, Czech Republic

1

INTRODUCTION. The objective of our study was to assess outcome of patients admitted to an intensive care unit (ICU) with special emphasis on quality of life measures by means of EuroQol 5-D (EQ-5D) questionnaire before and after admission to the ICU. METHODS. All patients admitted to the interdisciplinary ICU of University hospital for more than 7 days over a 2-years period (2000-2001) were eligible for inclusion in this study. The ED5D questionnaire was completed by proxies at the admission and 12 months after ICU admission by patients. According to the type of diagnosis patients were divided into group admitted for trauma (TR, n=31) and group admitted for other reasons (OT, n=45). Both groups were compared using following data: EQ-index before and 1-year after ICU admission, age, reason for ICU admission, APACHE II at the ICU admission, the highest SOFA during the ICU stay, length of ICU stay and duration of mechanical ventilation. Chi-squared test, Mann-Whitney and Wilcoxon matched-pairs signed-rank tests were used accordingly; p<0,05 was considered statistically significant. RESULTS. There were 236 patients who met the study criteria but only 76 (32%) of them completed the EQ-5D questionnaire. Significant differences between TR and OT groups were found only in age (41,6±18,2 vs. 52,9±15,8, p=0,007) and EQ-index before ICU admission (0,9±0,24, vs. 0,72±0,37, p=0,006). Deterioration of EQ-indexes 1-year after ICU admission was found in the whole group of admitted patients and in TR group, but not in OT group (Table 1). Reason for admission EQ-index before EQ-index after p All reasons (n=76) 0,80 ± 0,33 (1) 0,60 ± 0,36 (1) * Trauma (n=31) 0,9 ± 0,24 (1) 0,55 ± 0,36 (0,62) * Other (n=45) 0,72 ± 0,37 (0,85) 0,64 ± 0,35 (0,73) NS Data expressed as mean ± SD, (median): NS = non-significant, * = p<0,05.

CONCLUSION. The use of different definitions and in- and exclusion criteria in sepsis trials results in study populations with significantly different characteristics. The impact of these findings for e.g. meta-analysis is a subject of further investigation. REFERENCE(S). 1. NICE foundation: http://www.stichting-nice.nl Grant acknowledgement: This study has been supported by a grant of Eli Lilly.

CONCLUSION. We conclude that the quality of life after longer stay in intensive care is influenced by the reason for ICU admission and by the patient’s quality of life before ICU admission.

273

275

EARLY RESPONSE TO THERAPY PREDICTS EVENTUAL SURVIVAL IN SEVERE SEPSIS

QUALITY-ADJUSTED SURVIVAL IN MULTIPLE ORGAN DYSFUNCTION

Vincent J L1, Macias W L2, Levy M M3, Russell J A4, Trzaskoma B L2, Eliezer S5 1 Intensive Care, Erasme Univ Hosp, Brussels, Belgium, 2Specialty Care, Lilly Research Labs, Indianapolis, IN, 3Intensive Care, Rhode Island Hosp, Providence, RI, United States, 4Critical Care Medicine, St Paul’s Hosp, Vancouver, BC, Canada, 5Critical Care, Albert Einstein Hospital, Sao Paulo, Brazil INTRODUCTION. Aggressive resuscitation at diagnosis may reduce mortality in septic shock. If so, then a risk prediction model based on early response to therapy (day of diagnosis) in severe sepsis patients should be highly predictive of 28-day survival and further improvement on subsequent days should minimally impact the model’s predictive power. METHODS. Placebo databases from 2 randomized controlled sepsis trials were combined (n=1036). Patients met 1991 SCCM/ACCP criteria defining severe sepsis. Day 0 was defined as start of study drug to midnight of that calendar day with each subsequent day defined as the respective calendar day. Sequential organ failure assessment (SOFA) scores were calculated using most aberrant physiologic or laboratory parameter each day. Baseline and post-baseline dichotomous SOFA scores (improved [decreased by 1 or more] or not improved [unchanged or increased]) were used in a binary logistic regression analysis to correlate early organ dysfunction changes with 28-day mortality. RESULTS. Improvement in organ function before start of Day 1 compared to no improvement was associated with lower 28-day mortality for cardiovascular (CV) (20% vs 33%; p=0.0005), renal (22% vs 33%; p=0.0015), and hepatic (13% vs 30%; p=0.0439) functions. Improvement in respiratory and haematologic functions did not significantly influence mortality (26% vs 31%; p=0.0713 and 26% vs 30%; p=0.4932). By logistic regression, improvement in CV and/or renal function before start of Day 1 was highly predictive of improved survival; odds ratios (95% CI) = 0.62 (0.42-0.90) and 0.62 (0.43-0.88), respectively. Continued improvement in CV function before start of Day 2 showed further improvement in survival; odds ratio = 0.55 (0.37-0.82). No other organ system was retained in the model and improvement beyond Day 1 in any organ function did not add to the model’s predictive power. CONCLUSION. In patients with severe sepsis, improvement in CV and/or renal function on the initial study day was highly predictive of 28-day survival. Continued improvement in CV function the next day also improved odds of survival. Improvement in other organ systems or beyond first study day was not associated with improved survival. These data suggest that the outcome for patients with severe sepsis may be determined within the first day of therapy. Grant acknowledgement: Research supported by Eli Lilly.

Wehler M1, Geise A1, Reulbach U2, Hadziomerovic D1, Aljukic E1, Hahn E G1, Strauss R1 Medicine I, 2Medical Informatics, Biometry and Epidemiology, University Erlangen-Nuremberg, Erlangen, Germany 1

INTRODUCTION. Few data are available on quality-adjusted life-years (QALYs) after multiple organ dysfunction. METHODS. During a 1-yr period health-related quality of life (HRQOL) was assessed in 170 consecutive medical patients with multiple organ dysfunction (MOD). MOD was defined by a Sequential Organ Failure Assessment (SOFA) total maximum score (TMS) of 6 or more, which corresponds to at least two organ failures (grade 3) or three or more less severe dysfunctions present at the same time. HRQOL data were collected at admission and at 6-month follow-up using the Short Form-36 Health Survey. Baseline HRQOL data referred to the 4 weeks before admission to the intensive care unit (ICU). A preference-based utility measure, the SF-6D, was calculated from the Short Form-36 (1). Survival was assessed by the Kaplan-Meier method. RESULTS. The 170 MOD patients had a mean age of 60 ±16 yrs, 63% were male, mean Charlson comorbidity index was 2.7 ±2.2, mean APACHE II score after 24 hrs was 23 ±9, mean SOFA TMS was 12 ±4, range 6-16, cumulative TISS points of all ICU days were 362 ±609, mean ICU length of stay was 17 ±23 days. Survival estimated by Kaplan-Meier was 53 ±4% at 1 mo and decreased to 36 ±3.6% at 6 mo. Mean SF-6D of survivors was low and stable (0.59 ±0.13 at baseline and 0.60 ±0.11 at 6 mo, paired Wilcoxon p=0.95). Actual survival in the first 6 mo was 30.8 years of life per 53 MOD survivors. Adjustment by SF-6D follow-up scores produced a quality-adjusted survival in the first 6 mo of 18.5 QALYs. If the change in SF-6D scores from baseline to 6 mo was taken into calculation, 30.8 QALYs (+12.3 yrs=66%) were produced. CONCLUSION. These data demonstrate that traditional short-term outcomes, such as 28-day survival, overestimate long-term survival in medical patients with MOD. Furthermore, calculation of QALYs which depend solely on utility data at follow-up may be misleading. REFERENCE(S). 1. Brazier J et al. J Health Econ 2002;2:271 Grant acknowledgement: ELAN-Fund for Research and Teaching of the Medical Faculty of the University of Erlangen-Nuremberg, Germany

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

S75

276

278

OUTCOME AND QUALITY OF LIFE OF „OLDEST-OLD“ CRITICALLY ILL PATIENTS AFTER ICU DISCHARGE

PATTERNS OF INFECTION AND RESISTANCE IN A MULTI-DISCIPLINARY INDIAN ICU OVER TIME

Gianesello L1, Pavoni V1, Paparella L1, Pavarin P1, D’Urso G1, Gritti G2 1U.O of Anaesthesia and Intensive Care, University-Hospital, Padova, 2Dept. of clinical MedicalSurgical Area, Section of Anaesthesia and Intensive Care, Firenze, Italy

Thodur V M1, Shekar U2, Kamat V N1 And Critical Care, 2Microbiology, Sri Ramachandra Medical College And Research Centre, Chennai, India

INTRODUCTION. The ultimate goal of intensive care is to obtain survival with an acceptable quality of life (QOL). This is also important for “oldest-old” patients (greater than or equal to 90 years) who seem to be increased in the health care system and, especially, in intensive care (1). Outcome studies about short and long term survival and QOL studies for this type of patients are not well established.

INTRODUCTION. Indiscriminate use of third generation cephalosporins for serious infections causes resistant strains by expression of extended spectrum beta lactamases (ESBL). Unlike previously published Indian data on total hospital infections1, we analysed the nosocomial infections and resistance in our ICU over four years for the emergence of ESBL strains .

METHODS. Medical records of all patients greater than or equal to 90 years of age admitted to a polyvalent intensive care unit (ICU) during two years were reviewed. Data collection included gender, age, admission diagnosis, SAPS II on admission, length of ICU stay (LOS), ICU and hospital mortality. One year after ICU discharge, using the Euro-Qol questionnaire, a follow-up telephone interview with the patients or, when not possible, with their proxies was conducted by the same clinician to determine QOL and mortality rate. RESULTS. Of the 1555 patients admitted to the ICU during study period, 14 (0.9%) (4 male and 10 female) were 90 years old. The mean of their age was 92±1.6. The SAPS II on admission was 34.5±7.6. Among 14 patients studied, 4 were surgical scheduled patients, 5 were emergency surgical and 5 were medical patients. The LOS was 8.7±13.6 days. Three patients (21.4%) died in the ICU and others two (35.7%) died in the hospital. One year after ICU discharge, four patients (28.5%) were alive, whereas five (35.7%) were died after a mean of 176.8±82.4 days. All patients alive and included to the study were unable to answer (three for language difficulties and one for dementia). Their proxies reported moderate and extreme problems in five dimensions as follow: mobility (moderate 75%, extreme 25%), self-care (moderate 25%, extreme 75%), usual activities (moderate 0%, extreme 100%), pain/discomfort (moderate 25%, extreme 50%), anxiety/ depression (moderate 0%, extreme 25%). EQ VAS median was 30. Status of health of all patients was equal than one year before ICU admission. CONCLUSION. This study suggests that mortality rate of “oldest-old” patients is high both during ICU and hospital stay than after ICU discharge. Despite both physical and mental limitations, self-perceived QOL is unchanged compared with pre-ICU admission.

1Anesthesiology

METHODS. A prospective analysis of all patients admitted between May 1999 and Dec 2002. Respiratory and blood samples were analysed. Cultures from BAL with colony counts more than 104 and endotracheal secretions with colony counts more than 106 were considered. The prevalence of microorganisms and the change in the sensitivity pattern to ceftazidime, piperacillin and piperacillin-tazobactam were studied. The antibiotic susceptibility, the MIC and ESBL production were determined by standard protocols2. RESULTS. In the respiratory samples, Pseudomonas sp. was the predominant pathogen followed by Klebsiella sp. and Acinetobacter sp. In bloodstream infections, Stapylococcus aureus was more common during all four years; (26%, 34%,32% and25% respectively ). The antibiotic resistance pattern of gram-negative bacteria increased significantly over the past four years with very high MIC90 levels for piperacillin and ceftazidime (>16mg/ml and 128microgm/ml respectively) indicating the emergence of ESBL strains. Among Staphylococcal infections, the incidence of MRSA increased significantly over the last four years.[23%,24%,43%and 44% respectively] CONCLUSION. The increase in the incidence of ESBL producers obliges us to discontinue third generation cephalosporins for some time. Further molecular studies need to be done for epidemiological purposes and control strategies. A prospective study with plasmid typing is in progress REFERENCE(S). 1. Mathai D et al.Diagn Microbial Infect dis,44 (2002); 367-377. 2. National committee for clinical laboratory standards

REFERENCE(S). 1. Jacob SM, Rothen HU. Int Care Med 1997;23:1165-1170.

Oral Presentations Severe infections: A world-wide issue – 277-279+369* 277 INFECTIOUS MORBIDITY AND MORTALITY AFTER SURGERY FOR HYPOPLASTIC LEFT HEART SYNDROME Bramall P J1, VanSaene R H K F2, Taylor N3, Selby A4, Pozzi M5 Paediatric Intensive Care, 2Consultant microbiologist, RLC NHS Trust Alder Hey, 3Microbiology, University of Liverpool, 4Consultant Paediatric Intensive Care, 5Consultant Surgeon, RLC NHS Trust Alder Hey, Liverpool, United Kingdom

1

INTRODUCTION. Literature reports varying mortality rates between 35 and 60% in children requiring surgery for hypoplastic left heart syndrome. Infectious mortality rates have been reported as high as 50%. This is a prospective3 year study on mortality and infection rates in this subset with prolonged intensive care. METHODS. The study was undertaken in a 20 bedded paediatric intensive care unit [PICU]. Children with hypoplastic left heart requiring four or more days of ventilation were enrolled. The microbial carrier state of the children was monitored by weekly surveillance cultures of throat and rectum, on admission and then twice weekly. When patients were found to be carriers of abnormal flora including [potentially resistant] aerobic Gram-negative bacilli [AGNB] polymyxin E/tobramycin were given as a paste in the throat and as a suspension into the gut four times daily throughout PICU treatment. Defined primary endpoints were mortality and infectious morbidity according to the carrier state. RESULTS. A total of 22 patients were admitted 29 times during the three year study. Total mortality was 63.6% [14 patients died], PICU mortality was 50%. The median paediatric index of mortality was 0.205 [IQR 0.04-0.49] in the study group. The median day of death was day 9 [IQR 6.5-15]. Five children [22.7%] developed 16 infections 11 of which were due to micro-organisms carried on admission. They were mainly low level pathogens, including coagulase negative staphylococci and enterococci. The median day of infection was 24.5 [IQR 16-34]. Enteral antibiotics were administered during 23 of the 29 episodes [79%]. About 59% of patients were free from carriage of aerobic Gram-negative bacilli at unit discharge or death. CONCLUSION. Mortality rates were high but comparable with the literature. Infection was not the cause of death and no patient died with infection. About 70% of all infections were due to low level pathogens imported by the patient in their admisssion flora. The use of enteral nonabsorbable antimicrobials controlled infection due to the target micro-organisms, including AGNB and yeasts, in this subset of patients. REFERENCE(S). Chang RKR, Chen AY, Klitzner TS. Clinical Management of Infants With Hypolastic Left Heart Syndrome in the United States, 1988-1997. Pediatrics. 2002;110(2):292-8. Grant acknowledgement: Funding from Chris Stoutenbeek foundation * for 369 see page S98

279 SEVERE FUNGAL INFECTIONS AFTER LIVER TRANSPLANT: INCIDENCE AND RISK FACTORS De Gasperi A1, Perrone L1, Corti A1, Fantini G1, Amici O1, Garrone F1, Sciascia A1, Grugni C1 2 Servizio Anestesia e Trapianti Addominali, Ospedale Niguarda, Milan, Italy

1

INTRODUCTION. Liver transplanted pts have the highest risk(4–30%)to develop invasive fungal infection, a condition associated with increased morbidity and mortality. Aims of our study were to determine the incidence of early invasive fungal infections and to identify risk factors in a series of 258 consecutive pts undergoing OLT METHODS. From Jan 1997-June 2002,258 pts underwent 279 OLT . Antiinfective prophylaxis included cefotaxime, tobramycin, SDD, Fluconazole(Flu,3 mg/kg/day for 3 to 4 weeks)and acyclovir(15 mg/kg/day).Immunosuppressive regime included antithymocyte serum, steroids, Cyclosporin A or tacrolimus. Invasive fungal infections and fungal colonization were defined according to Singh et al (1997). Statistical analysis included t and X2 tests.p< 0.05 was considered significant RESULTS. In this series overall one year survival rate was 89%.16 pts (6.2% ) developed severe fungal infections: 9 pts(3.4%) invasive aspergillosis (IA),7 pts(2.7%) c andidiasis. Of the 9 pts with IA, six died (66%) for or with systemic aspergillosis (2 cases diagnosed at autopsy), 3 survived (1 cerebral abscess,2 invasive pulmonary aspergillosis).Of the 7 pts with candida infections, 2 had candidemia (1 died), 4 had tracheobronchial localizations (all alive),one had biliary infection (alive). Mortality associated with (but not for) Candidemia was 14% (1/7). The Candida species isolated were 5 albicans (1 from the blood, 1 from the bile, 3 from lower respiratory tract), 1 Torulopsis Glabrata (central venous catheter and blood), 1 Krusei (lower respiratory tract) . Risk factors for fungal infections in this series were age more than 50 y (p<0.0018), post OLT hyperbilirubinemia (p<0.009), post OLT thrombocytopenia (p< 0.012). Different between the groups with or without fungal infections but statistically non significant were preop Child score, UNOS status, blood loss and transfusion requirements, postop acute renal failure CONCLUSION. The incidence of fungal infection in our series (6.2%) seems to be one of the lowest reported. Very low are the incidence of Candida infections: even if debated, Flu prophylaxis should have played a relevant role in the containment of Candida infections without increasing antifungal resistance. The quality of functional recovery of the graft seems to play a major role in the development of IA. The role of thrombocytopenia, as already reported, seems to be confirmed in this study.

S76

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

Oral Presentations Cerebral dynamics – 280-282 280

282

NEUROLOGICAL DAMAGE IN HYPERACUTE LIVER FAILURE – ROLE OF NEUROEXCITATION

1Anesthesia

Bellamy M C1, Milligan L1, Harding R1, Buglass H1, Millson C E2, Murdoch S1 Intensive Care Unit, 2Department of Hepatology, St James’s University Hospital, Leeds, United Kingdom

INTRODUCTION. Cerebral vasospasm (VSP) is the major cause of mortality and morbidity after aneurysmal SAH and there is no satisfactory pharmacological treatment. Milrinone (MIL), a phosphodiesterase inhibitor has been recently proposed (1).

INTRODUCTION. Several groups have failed to show an association between raised ICP and adverse neurological outcome in hyper acute liver failure [1]. We hypothesise neuroexcitation leads to neuronal death, with altered ICP occurring late. In 106 patients with hyper acute liver failure we identified a group following a “fast” course to brain death without raised ICP, and a “slow” group with raised ICP who survived [2]. We have therefore measured neurochemistry and brain haemodynamics in “fast” and “slow” patients.

METHODS. A prospective pilot study (data in median (extremes)) was conducted in 9 consecutive patients (pts, 5M/4F), 51 y/o (37-59), SAPSII 22 (8-37), WFNS score (World Federation of Neurosurgeons) 2 (1-4), Fisher score 4 (2-4), who developed at D6 (4-10) a VSP despite Nimodipine treatment (7/9 pts). A selective intra-arterial infusion of 0.25 mg /min (i.e. 8mg MIL over 30 min per spastic territory) was performed, followed by continuous intravenous (iv) MIL at a rate of 0.5 µg/kg/min. Angiogram before and after MIL allowed quoting VSP (moderate, severe and major) and computing % variations of vascular diameter.

1

METHODS. We measured neurochemistry using a microdialysis system (CMA Inc., Stockholm), ICP (Camino Lab, San Diego), bispectral index (BIS), continuous jugular venous oxygen saturation; and cellular oxygenation by near infra-red spectroscopy (Hamamatsu Photonics KK, Japan). Cerebral blood flow was measured by transcranial Doppler (TCD). Zero flow pressure (ZFP) was derived. RESULTS. Data are presented on typical, well-matched patients. The fast patient progressed to a state of neurological damage 7 hours from onset of coma, with fixed and dilated pupils. ICP and CPP remained normal throughout. Peak brain glutamate and glycerol levels were 540 and 745 micromolar respectively. Cellular oxygenation fell sharply as the pupils become non-reactive, coinciding with a rise in glycerol, lactate/pyruvate ratio, falling SjO2, and zero BIS. TCD revealed a high velocity, low volume trace suggestive of cerebral vasospasm. ZFP was 88 mmHg. At postmortem, the brain showed autolysis but no cerebral oedema. In contrast the “slow” patient had several episodes of intracranial hypertension but with less deranged neurochemistry. The peak glutamate and glycerol levels were 65 and 27 micromolar respectively. The ZFP was 52mmHg. 4 CONCLUSION. These results are consistent with brain death by excessive neuroexcitation with cerebral oedema as a secondary phenomenon. Further observational data are required prior to an interventional study being designed. REFERENCE(S). References: 1)Davies MH, Mutimer D, Lowes J, Elias E, Neuberger J. Recovery despite impaired cerebral perfusion in fulminant hepatic failure. Lancet. 1994 May 28;343(8909):1329-30. 2) Bellamy MC, Shastri P, Binks R, Millson CE. Hyper acute liver failure: What threshold of raised ICP should we accept? British Journal of Anaesthesia 2002; 88(2): 317P. Grant acknowledgement: Association of Anaesthetists of Great Britain and Ireland

281 NEUROPROTEIN S-100 KINETIC DURING SUBARACHNOID HEMORRHAGE (SAH): PRELIMINARY DATA Losser M1, Lanceleur A2, Brun J2, Beaudeux J3, Pignataro C1, Payen D2 Anesthesia and CCM, 2Anesthesia and Critical Care Medicine, Lariboisière University Hospital, 3 Biochemistry, La Pitié University Hospital, Paris, France

1

INTRODUCTION. Astroglial neuroprotein S-100 diffuses into CSF and then to venous blood during acute cerebral injury with a leaking blood brain barrier, either ischemic or traumatic. A prospective study has been conducted to test the place of S-100 in the follow up of aneurismal SAH. METHODS. Data expressed as median (extremes). During 6 months, 17 consecutives patients (pt; 3M/14F) aged 53 years (33-66) have been included. 13 pts had a least 4 arterial measurements (immunoluminometry), 6 of them had simultaneous arterial (A) and jugular venous (V) determinations (to calculate a gradient (V-A)/V, %), and in 4 pts, S-100 was in addition measured in CSF. CT Fisher score was 3 (1-4) with a clinical WFNS grade (World Federation of Neurosurgeons) of 3 (1-5), a initial CGS of 11 (4-15), a SAPS II of 29 (14-49), and a GOS (Glasgow outcome score) at 1 month of 3 (1-5). RESULTS. 6 pts will develop cerebral vasospasm (VSP+). Initial S-100 in these VSP+ pts was 0.15 µg/l (0.06-0.44) vs 0.20 (0.06-8.25) in VSP- pts (NS, control 0.05 (0.02-0.07)). Initial S-100 was not related to gravity scores and did not allow predicting the occurrence of VSP. Arterial kinetic over 15 d (fig 1) showed a progressive decrease of S-100 in VSP- (Friedman, p=0.06), while it tended to increase in VSP+. Baseline (V-A) / V was 9% (0-20) and tended, between D3D7, to increase to 23% (0-34) (p=0.15, Friedman). CSF levels were 42.5 g/l (2.4-2160), 10 to 100fold higher then in jugular vein and were not correlated with protein levels or red cell count in CSF. 0.4

arterial S- 100ß median (µg/l)

VSP+n=5 VSP–n=6

0.3 0.2 0.1 0

D1-D2

D3-D4

D5-D7

D8-D10

D11-D15

CONCLUSION. S-100, marker of cerebral cellular lesions, showed a different kinetic whether VSP occurred or not. S-100 may thus be a marker of decompensated VSP (leading to a significant cerebral injury), but should be evaluated on a larger cohort. Arterio-venous gradient of S-100 is not fixed during the time course of HSA, its significance has to be explored.

MILRINONE FOR THE TREATMENT OF CEREBRAL VASOSPASM AFTER SUBARACHNOID HEMORRHAGE (SAH). Losser M R1, Fraticelli A1, Saint-Maurice J P2, Lukaszewicz A C1, Payen D1 and CriticalCare, 2Neuroradiology, Lariboisiere University Hospital, Paris, France

RESULTS. 13 selective intra-arterial infusions were performed with a remarkable haemodynamic stability, especially in the basilar artery. All of the 24 vasospastic vessels (2 per pts (1-6)) responded to MIL. The effect (fig 1, 2-way ANOVA) was more pronounced 1)for the most severe VSP (p < 0.001) and 2) for ACA et ACM (p= 0.08). A control angiogram was performed in 4 pts 2 to 3 days after continuous iv MIL and showed no recurrence of VSP. Figure1: Figure 1: % variation vascular diameter before and after MIL ACA: anterior cerebral artery A1, MAC: middle cerebral artery M1, ICB: Internal carotid bifurcation, basilar artery. 200 150

ACA ACM ICB basilar

100 50 0

Moderate (n=8)

Severe (n=11)

Major (n=5)

CONCLUSION. Intra-arterial MIL is an effective and safe alternate therapeutic to papaverine of cerebral vasospasm. In addition, MIL can be subsequently given as continuous iv infusion. Milrinone may thus be proposed for the treatment of vasospasm resistant to Nimodipine. REFERENCE(S). 1. Neurosurgery 2001; 48: 723-30

Oral Presentations TBI-induced metabolic and endocrine disorders – 283-285 283 HYPERGLYCEMIA AND OUTCOME HAVE A STRONGER RELATION IN TRAUMA THAN IN OTHER SURGICAL ICU PATIENTS Vogelzang M1, Dingemans A J G1, Van der Horst I C C2, Ten Duis H1, Nijboer A J M M1, Nijsten M W N1 1 Surgery, 2Internal Medicine, Groningen University Hospital, Groningen, Netherlands INTRODUCTION. In several groups of critically ill patients hyperglycemia is known to be related with an adverse outcome. Prevention and treatment of hyperglycemia has been shown to have a beneficial effect on outcome in a large series of predominantly thoracic surgery patients [1]. However in that study only 4% of the patients were trauma patients. Since trauma patients are generally in full health before the accident, they differ from most other patients at the ICU. Thus hyperglycemic responses and their impact on outcome might differ in trauma patients from other patients at the surgical ICU. METHODS. All ICU patients aged over 15 years over a 12-year period were classified as trauma or non-trauma. For each patient, mean glycemia (MG) was calculated by taking the mean of all measurements done on that patient. RESULTS. Over the study period 6409 patients were included, of which 1030 (16%) were classified as trauma. The mean age was 56±18 years. Median (IQR) length of stay was 1.7 (0.9 4.9) days. The mortality rate in all patients was 13%. In trauma patients who died mean MG was 9.0±3.2 mmol/l versus 6.9±1.4 mmol/l in those who survived (p<0.001, difference is 2.1 mmol/l). In non-trauma patients mean MG was 8.2±2.6 mmol/l versus 7.6±1.8 mmol/l, p<0.001, difference is 0.6 mmol/l). The difference in MG in trauma patients is thus much higher than in non-trauma patients (p<0.001). Analysis of patients younger than 45 did not affect these results. CONCLUSION. Dysregulation of glucose metabolism is more pronounced in non-survivors. Surprisingly the glucose regulation is worse in non-surviving trauma patients compared to other patients at the surgical ICU that do not survive. This effect is not age-related. Although the underlying mechanism is not clear, this finding suggests that intensive regulation of glucose may be of special relevance in trauma patients. REFERENCE(S). 1. van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. N Engl J Med 2001;345:1359-67

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

284 PREVALENCE OF NEUROENDOCRINE DYSFUNCTION FOLLOWING TRAUMATIC BRAIN INJURY: PRELIMINARY RESULTS Flores-Cordero J M1, Rincón-Ferrari M D1, Pujol-Congregado M2, Pérez-Alé M1, MurilloCabezas F1, Leal-Cerro A3 1 Intensive Care Unit, 2Division of Neurophysiology, 3Division of Endocrinology, Hospital Virgen del Rocío, Seville, Spain INTRODUCTION. Hypopituitarism after traumatic brain injury (TBI) may be underrecognized due to its subtle clinical manifestations. We are conducting a study in order to assess the prevalence of neuroendocrine abnormalities in patients suffering from TBI at least 6 months ago. We present the preliminary results obtained from a series of standard endocrine tests. METHODS. From 1998, we prospectively include in a data base all patients with TBI admitted to our Intensive Care Unit (ICU). For the purpose of this study, we have selected 238 adult patients (range: 16-65 years) with TBI admitted to our ICU and who were discharged from hospital with a Glasgow outcome scale score of 3, 4 or 5. A clinical survey including symptoms of hypopituitarism was mailed to all patients selected, and inviting them to participe in this study. Those who accepted have undergone a series of standard endocrine tests, including serum levels of TSH, PRL, insulin-like growth factor I (IGF-I), cortisol, testosterone (males), and estradiol (females). Patients with abnormal results of these tests are being examined with dynamic tests of gonadotropin, TSH, and GH secretion. RESULTS. A total number of 188 (78.9%) patients responded the survey and 160 (67.2%) accepted to participe in the study. We have examined 110 patients (96 males and 14 females) and we have found at least one abnormal result of standard endocrine tests in 40 (36.3%) of them. The distribution of abnormal results found is the following: IGF-I 25.4% (28 of the 110 patients), cortisol 7.2% (8 patients), testosterone 7.2% (7 patients of the 96 males), estradiol 7.1% (1 patient of the 14 females), PRL 5.4% (6 patients), and TSH 0.9% (1 patient). The mean age of the 40 patients with abnormal findings was 32.2±9.4 years. The mean Glasgow coma scale score after initial reanimation was 7.6±3, and the more frequent categories of TBI, according to the computerised tomography findings, were type V (15 patients) and type II (14 patients). CONCLUSION. Our preliminary results suggest that neuroendocrine dysfunction may be identified in a substantial proportion of patients with previous TBI. Although confirmation by dynamic test is pending, GH deficiency may be more frequent as previously suspected. Grant acknowledgement: This study is supported by Consejería de Salud, Junta de Andalucía grant 06/01.

S77

Oral Presentations Nursing: Improving the patient experience of intensive care – 286-291 286 CONTINUOUS EDUCATION WITH A STANDARDIZED PROTOCOL IMPROVE SEDATION PRACTICES IN ICU Hawes R1, Moore T1, Haddad S1, Arabi Y1, Al-Shimemeri A1 1 Intensive Care Unit, King Fahad National Guard Hospital, Riyadh, Saudi Arabia INTRODUCTION. The purpose of this study is to examine the impact of a standardized inservice program along with protocol in changing sedation practices in the ICU. METHODS. The study was conducted in a 21-bed medical surgical tertiary care ICU. A standardized protocol was established based on published recommendations by a joint committee of ICU physicians and nurses. The protocol used the Sedation Agitation Scale (SAS) to assess sedation and the Numeric Rating Scale (NRS) to assess pain. Several in-services were given to the medical and nursing staff before and throughout the study period. One member of this group was routinely available to answer queries about the protocol and to provide feedback about its implementation. We collected data on patients placed on the protocol for the first 4 months. We compared sedation practices in the first 2 months to those in the following 2 months to see if there is a change in the practice with continuous education. Values were expressed as mean ± SEM. Ttest and Kruskal-Wallis tests were used. RESULTS. After 2 months of implementing the protocol and continuous education, physicians ordered significantly lighter levels of sedation (SAS 2.3 ± 0.3 to 3.6 ± 0.1, P = 0.0003). There was significant decrease in the average daily doses of midazolam (55 ± 12 vs. 27 ± 8 mg, P = 0.06), fentanyl (1113 ± 216 vs. 612 ± 116 mcg, P = 0.05) and morphine (21 ± 8 vs. 3 ± 2 mg, P = 0.04). There was no significant decrease in the daily doses of propofol. There was a trend towards lower ICU length of stay (median 8 vs. 5 days, P = 0.08). CONCLUSION. Continuous education along with sedation protocol was effective in changing sedation practices in the ICU and in shortening ICU length of stay.

285

287

BIOCHEMICAL MARKERS AND JUGULAR BULB-ARTERIAL CARBON DIOXIDE TENSION GRADIENT AS OUTCOME INDICATORS

COMBINING INTERVENTIONAL RESEARCH WITH PATIENT CONTROLLED ANALGESIA (PCA) USE

Anastasiou E1, Tsaousi G1, Geka E1, Yannakou M1, Argiriadou E1, Efthimiou A1, Nagy E1, Sofianos E1 1 Anaesthesiology and ICU, AHEPA University Hospital, Thessaloniki, Greece

1

INTRODUCTION. Hyperlactatemia can be a sign of inadequate tissue perfusion. Serum neuronspecific enolase (NSE) and S-100 protein are established as markers of brain damage, whereas DeltaPCO2 (venous-arterial PCO2 difference) is related to systemic hypoperfusion in various forms of low-flow state. In the present study these indices were evaluated as predictive of disability upon ICU discharge.

INTRODUCTION. Balancing analgesia and sedation to facilitate early extubation after coronary revascularisation (CABG) remains a challenge to cardiac recovery unit (CRU) nurses. We describe an innovative approach to increasing PCA use by seconding a CRU nurse to the Pain Management Service (PMS) to facilitate a continuous audit of practice in the CRU and lead an interventional research pilot project.

METHODS. Thirty seven patients with traumatic brain injury (TBI), associated or not with multiple trauma, admitted in the ICU were included in the study. Paired arterial (art) and jugular bulb (jb) blood concentrations of lactate, NSE, S100 protein and blood gases were measured daily for 3 consecutive days. Jugular bulb -arterial PCO2 gradient (dPCO2) was estimated. The patients were divided in 2 subgroups according to Glasgow Outcome Scale (GOS): group A (GOS 1-3, n=19) and group B (GOS 4-5, n=18). Statistical analysis was performed using Kruskal-Wallis and Mann-Whitney tests.

METHODS. With Ethics Committee approval and written informed consent consecutive patients were enrolled in the audit project from Nov 2002 to present. A total of 20 patients were recruited to the pilot study and received a standardized anaesthetic and were then randomized to receive standard (S)or remifentanil (R) therapy. After stabilisation in the CRU and cessation of propofol sedation, Group S were converted to nurse controlled analgesia (NCA) with continuous intravenous morphine, while Group R followed a 4-step reducing remifentanil infusion prior to conversion to PCA. Weaning and extubation in both groups occurred when criteria were met.

RESULTS. Lactate remained above normal levels in both groups, however no statistically significant difference was noticed. There was a statistically significant reduction in dPCO2 in group A (p=0,02) during the study period. NSE, S-100 protein (mug/L) concentrations and dPCO2 (mmHg) between the two groups are presented in table 1.

RESULTS. Patient demography were similar in each study group. Mean pain intensity scores immediately after extubation were comparable in group R and S (3.5 and 3.1 respectively), but the scores at 4 hours were statistically significant (1.5 v 3.3). The number of patients with mean pain scores greater than 3/10 was higher in the S group (40% v 20%). Audit data revealed that PCA use increased from a baseline of 9% to 68%, whilst mean pain intensity scores have reduced since the commencement of the project.

NSEart NSEjb S-100art S-100jb dPCO2

A B A B A day 1 day 1 day 2 day 2 day 3 54.6±34.3* 32.2±30.8* 47.4±31.4 29.2±26 47.1±31.8 52.3±32.4* 31.5±27.7* 52.7±30.6 36±24.5 59.2±38 0.9±1.1* 0.4±0.3* 0.8±0.6* 0.3±0.3* 1.4±0.8** 1.2±0.7* 0.5±0.4* 0.7±0.8 0.6±0.7 1.2±0.9* 7.3±4 6.7±3.5 6.5±3.1 5.7±3.1 2.8±2.4* * p<0.05 **p<0.01 (Values are expressed as mean±SD)

B day 3 32.8±23 33.2±24.7 0.3±0.2** 0.4±0.5* 5.7±4.8*

CONCLUSION. Following TBI increased NSE and S-100 protein are indices of adverse ICU functional outcome. S-100 protein seems to be a promising marker for assessing the extent of primary brain injury and ongoing secondary damage. Interpretation of dPCO2 changes needs further investigation.

AUDIT

TO

INCREASE

Cox F J1, Margetts M2, Carbery C2, Gifford-Moore A1 Pain Management Service, 2Department of Critical Care, Royal Brompton and Harefield NHS Trust, London, United Kingdom

CONCLUSION. Nurse-led audit and research have reduced pain intensity at 4 hours postextubation and has increased PCA use in the CRU. CRU nurses demonstrated greater understanding of sedation and analgesia in patients who have undergone CABG and this is supported by the audit results. The number of patients experiencing moderate to severe pain (pain higher than 3/10) was lower in the group receiving PCA. This projects highlights that a combination of a project nurse, audit, research and targeted education can improve sedation and analgesia practices.

S78

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

288

290

THE PATIENTS EXPERIENCES OF A DIARY AND A FOLLOW UP VISIT AFTER ICU TREATMENT.

IMPROVING FAMILIES’ TRANSFER EXPERIENCE FROM INTENSIVE CARE BY USING A COMMUNICATION INTERVENTION

Akerman E1 care unit, University hospital in malmoe, malmoe, Sweden

Mitchell M L1, Courtney M2, Coyer F2 1School of Nursing, Griffith University, Technology, Brisbane, Australia

1Intensive

INTRODUCTION. Patients treated at Intensive Care Unit (ICU) are often suffering from amnesia regarding their stay in ICU. Instead they remember hallucinations, fictions and nightmares. They do not realise how serious their sickness has been. These could lead to both psychological and physical problems after the treatment (1). To help these patients recover after the hospitalisation the staff is writing a diary, with photographs, that describe what happened to the patient during the time in the ICU. They are also offered a follow up visit to the ICU to get a better understanding of what have happened to them. METHODS. The study is carried out with a qualitative methodology, with partly structured questions. Interviews have been carried out with eight patients that all have been nursed at ICU during yare 2002. Each interview was recorded on tape then analysed and categorized. The purpose of the interviews was to find out the patients reactions to:1. Receiving a diary regarding their time spent at ICU.2. The follow up visit to ICU after the treatment.3. The importance of a follow up visit to ICU. RESULTS. The patients only had partial memory of the time spent in ICU. The diary gave the patient an understanding of how sick they had been.The photographs together with the written information in the diary helped the patients to over come the time spent in ICU. The diary helped them to fill in their memory gaps. All patients have had an unreal experiences from the hospital time. None of the patients could find any explanation to these experiences in their diary. During the follow up visit to ICU they experience that they got an understanding and an explanation too how sick they had been and that they got a possibility to discuss their unreal experiences during their time at ICU. They experienced that they had a need for discussing their unreal experiences with the staff from ICU and they found it meaningful and important to meet and discuss with the staff that had nursed them. CONCLUSION. The diary and the follow up visit can give the patients the possibility to fill in the memory gap they have and to get an explanation concerning their sickness and the symptoms they had after their discharge from the hospital. Without the follow up visit and the diary the patient can have problem or a hard time to understand what have happened to them and how sick they where and why the recovery takes so long time.

2School

of Nursing, Queensland University of

INTRODUCTION. The study aimed to improve families’ transfer from ICU. An educational method of patient transfer was evaluated. METHODS. The study used a quasi-experimental design. Families in an ICU were allocated to a control (n=80) and intervention group (n=82). This group had improved communication and the control had existing transfer methods. They were surveyed before and after transfer using Spielberger’s state anxiety inventory and Mishel’s uncertainty scale. Nurses (n=40) were surveyed to assess the efficacy of the intervention. RESULTS. Factors significantly effecting uncertainty included state anxiety scores (F=50.9, p=.000), the relationship of the family member to the patient (F=2.9, p=.022), and the nature of the admission (F=23.09, p=.000). State anxiety levels were significantly effected by the degree of family social support (F=10.0, p=.002). State anxiety reduced significantly following transfer for both groups and uncertainty reduced significantly for the intervention group (t=2.21, p=.03). When controlled for pretransfer levels there was no significant reduction in the intervention group when compared with the control group. Uncertainty for the intervention group reduced whereas scores for the control group did not. The intervention group experienced significantly higher levels of satisfaction with transfer (Z=-2.43, p=.015) and felt significantly better prepared for transfer (Z=-3.26, p=.001). 90.6% of nurses thought the intervention provided a framework for discussion and 94% thought it should be introduced for all transfers. CONCLUSION. Uncertainty in illness and state anxiety levels are significantly related in families moving from ICU. One’s coping ability is affected by both these factors and limits their adaptation to transfer resulting in relationship disturbances and psychological distress1 at a time when patients rely on family support. This cost effective intervention reduced uncertainty. Uncertainty is reported for the first time with families in ICU and its link to the way people cope supports further investigation of ways nurses can reduce its levels. REFERENCE(S). 1. Mishel, M. (1981) The measurement of uncertainty in illness, Nursing Research, 30,5,258-263. Grant acknowledgement: Queensland Nursing Council, Royal College of Nursing Australia

REFERENCE(S). 1.Griffiths RD,Jones,C, Intensive Care Medicine 2001;27:344-346.

289

291

EXPERIENCES EXPRESSED BY ARTIFICIALLY VENTILATED PATIENTS A1

1

FACILITATING FAMILY FOCUSED INTENSIVE CARE NURSING THROUGH ACTION RESEARCH

2

Claesson , Mattsson H , E. Idvall 1 Department of intensive Care, Oskarshamn Hospital, Oskarshamn, 2Department of research and development, County Council, Kalmar, Sweden INTRODUCTION. It has been reported that patients have difficulties to recall their stay in an intensive care unit (ICU) and that they might develop different clinical signs defined as the ICU syndrome e.g. confusion, unreal experiences and difficulties in sleeping. In our department, working with a family nursing perspective and patient diaries, it seemed to be important to increase the knowledge of what our patients remember and could tell us from their ICU stay as a basis for quality development. Thus, the aim of the present study was to describe the experiences expressed by artificially ventilated patients in an ICU in a county hospital in Sweden.

Coyer F M1, Courtney M1, Macpherson I2 School of Nursing, 2School of Professional Studies, Queensland University of Technology, Brisbane, Australia 1

INTRODUCTION. This study sought to facilitate an improved understanding by intensive care nurses about their care of family members of critically ill patients.

METHODS. Because the research question was explorative a qualitative approach using interviews was chosen. The sample consisted of seven males and one female, consecutively during spring 2002. The mean age was 66 years and the length of ICU stay ranged between 21 and 28 days. All patients asked, gave informed consent to participate in the study, and the interviews were undertaken 6 to 12 weeks after ICU discharge. Pre-selected and open-ended questions were used to collect data in the interviews. The questions concerned the patients’ general experiences in the ICU and specific e.g. about pain, fear and dreams. The interviews were audio tape-recorded and transcribed verbatim. The text was read several times and analysed by categorising in themes.

METHODS. The study utilised a collaborative action research methodology in four phases. Phase 1: The establishment of the clinical collaborative action research group (CARG). Phase 2: A prospective survey design was utilised to survey family members of critically ill patients and staff in the ICU to determine family member’s needs. Sixty-three family members and 54 staff were surveyed utilising the Critical Care Family Needs Index (CCFNI). Data were analysed using rank ordering of means and t test. Phase 3: Intervention testing in two practice areas. A) Education seminars were offered to staff to raise awareness of family needs. Thirty-four ICU registered nurses attended. Data were collected using the CCFNI. Staff responses in phase 3 were compared with family members responses from phase 2. Data were analysed using rank ordering of means and t test. B) A family assessment tool was developed for the ICU computer information system. Content was evaluated through focus group interviews with ICU registered nurses. Phase 4: Evaluation of the interventions.

RESULTS. Only a few patients remembered that they had been artificially ventilated during their ICU stay. Some patients could recall real experiences e.g. unpleasant airway suctions. The patients also remembered several unreal experiences and their narratives included e.g. illusions, hallucinations and dreams, both pleasant, crazy and nightmares. The patients expressed e.g. „…I remember talking with two black-dressed nurses with black faces …“ and „…the staff glided on rails around the bed all the time, wearing white hoods …“. It seemed to be very complicated course of events. The patients also reported that they sometimes got into a state of anxiety and felt different levels of fear for complicated investigations and the experience of pain.

RESULTS. Phase 1: The CARG determined the direction for the study. Phase 2: Family members and staff ranked the CCFNI assurance category highest (t=5.513, p=0.00). Proximity (t=5.530, p=0.00) and information (t=6.475, p=0.00) needs were ranked highly by family members whereas staff ranked comfort (t=0.511, p=0.61) and support (t=-2.446, p=0.016) needs highly. Phase 3: A) Staff ranked CCFNI assurance needs as the highest category, followed by information and proximity. B) Family assessment tool was refined and validated by focus group discussions. Phase 4: A) Education seminars are an ongoing feature of ICU staff education programs and B) family assessment is incorporated on the computer information system for all staff.

CONCLUSION. The patients could recall several experiences from their ICU stay. These narratives gave valuable knowledge for quality development e.g. to continue the work with a family nursing perspective and patient diaries, to find new and better routines for nursing care, patient sedation, information and follow-up talks after discharge.

CONCLUSION. Integral to the development of family focused intensive care nursing is ongoing education for staff concerning family needs and regular family nursing assessment. Early detection and understanding of family needs will enable staff to anticipate, assess and intervene in potential problems for the family. This is the cornerstone of a partnership in care in ICU with the family. Grant acknowledgement: This work has been supported by an award from Royal College of Nursing, Australia.