S188

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

Oral Presentations Paediatrics (IV) – 720-724 720

722

CONGENITAL DIAPHRAGMATIC HERNIA: THE GREAT ORMOND ST HOSPITAL EXPERIENCE

1Neonatal

Cassidy J1, Pierce C1, Goldman A2, Petros A1 Paediatric Intensive Care Unit, 2Cardiac Intensive Care Unit, Great Ormond St Hospital, London, United Kingdom

1

INTRODUCTION. Neonates with congenital diaphragmatic hernias (CDH) are a challenging group of infants, with continually evolving management but minimal randomised controlled trials. In this setting audit represents a valuable tool for evaluation of new therapies. METHODS. A retrospective review was performed from January 1996 to December 2002, examining all identified cases of neonatal CDH, admitted to the intensive care units at Great Ormond St. Information gathered included demographics, presence of associated anomalies, age at referral, support required in terms of ventilation and inotropic requirements, worst oxygenation index (OI), the use of extracorporeal membrane oxygenation (ECMO), timing of surgery, post operative support and outcome. RESULTS. Over the 7 year period there were 2165 admissions to the Neonatal Intensive Care Unit. 91 infants with a diagnosis of CDH were identified (4%).12% had associated cardiac anomalies, 7% significant additional gastrointestinal anomalies and 3% renal anomalies. Of the 91 infants with CDH, 76 (84%) survived to hospital discharge. Survival in the group diagnosed postnatally was 91% as compared to 62% in those diagnosed antenatally. 77 were operated on,of whom 70 (91%) survived. In the group with an OI <20 (N=50), all survived to hospital discharge and none received ECMO. In the group with an OI >20 (N=41), 25 received ECMO, with a median OI of 65( interquartile ranges 50-82) and 64% survived to hospital discharge. 16 were managed conservatively with a median OI of 35 (interquartile ranges of 25-68) of whom 56% survived to hospital discharge. In the group with an OI >40, the traditional ECMO referral line; 55% of the ECMO group survived as compared to 11% of the non ECMO group,(p = 0.04). CONCLUSION. In CDH infants with an OI < 20 our survival is excellent. With increasing severity, as reflected by an increasing OI, patients appear to do better with ECMO. Our findings support the need for a randomised controlled trial on the use of ECMO in these infants.

CARDIAC TROPONIN T (CTNT) AND CARDIAC FUNCTION IN RESPIRATORY DISTRESS SYNDROME (RDS) Clark S J1, Newland P2, Yoxall C W3, Subhedar N V3 Unit, Jessop Wing, Sheffield, 2Biochemistry, Royal Liverpool Children’s Hospital, 3Neonatal Unit, Liverpool Women’s Hospital, Liverpool, United Kingdom

INTRODUCTION. cTnT is a highly sensitive and specific marker of myocardial injury. Cord blood cTnT is raised in neonates who develop RDS, an association that is independent of demographic variables. Aims: To investigate (1) postnatal blood levels of cTnT and (2) the relationship between cTnT and cardiac function in infants with RDS. METHODS. Blood samples were taken for cTnT levels and echocardiographic examination performed in term and preterm neonates with RDS during the first 3 days of life. A subgroup of infants had serial daily cTnT levels and/or echocardiographic examinations. Cardiac function was assessed by calculating fractional shortening (FS) and cardiac output (CO). RESULTS. Values are median (interquartile range). 46 infants with RDS were studied, gestation was 29(27-31) weeks, birth weight was 1.13(0.91-1.93)kg and cTnT was 156(73-286)pg/mL. Serial samples were obtained in 22 infants on days 1, 2 and 3. cTnT levels were higher in infants with RDS compared with samples taken from 68 healthy infants divided into the same time points (table). In infants with RDS cTnT did not change with time. Echocardiographic examination and cTnT levels were performed in 18 infants. There was correlation between cTnT and FS (rho=0.73, p=0.001) and CO (rho=-0.47, p=-0.05). cTnT(pg/mL) Healthy Infants Infants with RDS Day 1 10(10-19) 98(60-136)* Day 2 24(10-55) 102(45-231)* Day 3 45(10-87) 130(61-265)* *p<0.001 compared with healthy infants at respective time points. CONCLUSION. cTnT levels were elevated in infants with RDS compared with healthy infants. Infants with RDS had a significant negative correlation between cTnT and echocardiographic markers of cardiac performance. We speculate that cTnT may be a useful marker of cardiac dysfunction in the neonate. Grant acknowledgement: New Born Appeal

721

723

LEVELS OF INTERLEUKIN-8 AND EXPRESSION OF IL-8 RECEPTORS AFTER CARDIOPULMONARY BYPASS IN CHILDREN

PREDICTORS OF PROLONGED VENTILATION AFTER PAEDIATRIC CARDIAC SURGERY

Gessler P1, Dahinden C2 1 University Children’s Hospital, Paediatric Intensive Care Medicine, Zurich, 2University of Berne, Institute of Immunology, Berne, Switzerland

1

INTRODUCTION. Cardiopulmonary bypass induces a systemic inflammatory response that causes substantial clinical morbidity. This study sought to determine cellular and humoral variables of inflammation. We hypothesized chemokines as a major source of stimulation of neutrophils and monocytes in paediatric cardiac surgery (1). METHODS. Observational prospective clinical study in 20 pediatric patients before and after open heart surgery with cardiopulmonary bypass. Plasma levels of interleukin-6, interleukin-8, myeloperoxidase, and nitric oxide were measured by immunoassays. Expression of interleukin-8 receptors (CXCR1, CXCR2) and CD14 of circulating neutrophils and monocytes was assessed by flow cytometry. Clinical evaluations included length of inotropic support and mechanical ventilation as well as oxygenation.

Spinks J J1, Cottrell S M1, Marsh M J1 Paediatric Intensive Care, Southampton General Hospital, Southampton, United Kingdom

INTRODUCTION. There has been little research on predictors of prolonged ventilation after paediatric cardiac surgery. In adults pre-existing lung pathology, duration of bypass, type and urgency of surgery [1,2] and scoring systems [3] predict post-operative ventilatory requirement. The aim of this study was to establish predictors for prolonged ventilation to identify patients suitable for accelerated weaning. METHODS. Retrospective analysis of pre-, intra- and post-operative data on all patients aged 016 years admitted over a 1 year period undergoing cardiac surgery was performed. Outcome data included time to first and final extubation, length of stay (LOS) and number of failed extubations. Statistical analysis included descriptive statistics and Spearman’s rank correlation. The Local Ethics Committee approved this study.

RESULTS. Two hours after cardiopulmonary bypass, plasma levels of interleukin-6 and interleukin-8 were strongly increased (p = 0.0001 and p = 0.0032, respectively). Interleukin-6 and interleukin-8 concentrations correlated with the length of inotropic support as well as with the length of mechanical ventilation (r > 0.70, p <= 0.0006), and were inversely related to the ratio of arterial oxygen tension to fraction of inspired oxygen. There was a strong association between the postoperative levels of interleukin-6 and nitric oxide, as well as between interleukin-6 and CD14 expression on monocytes (r > 0.62, p <= 0.0031). The expression of CXCR2, but not CXCR1 on neutrophils and monocytes correlated negatively with the levels of interleukin-8 and of myeloperoxidase.

RESULTS. 149 subjects were identified. 30 were excluded leaving 119 eligible subjects undergoing 138 operations. All values are given as median and (range). There was an equal sex distribution with age 0.44 (0.003 -15.8) years and weight 5.65 (1.6-47.6) kgs. Time to first extubation was 12.9 (1-172) hours and time to final extubation was 13.1 (1-934) hours. LOS was 27.3 (6-984) hours. There were 15 failed extubations on 9 subjects and length of further ventilation following failed extubation was 113 (3-859) hours. Prolonged ventilation had a negative correlation with weight and age (CI -0.74 to -0.55 and -0.73 to -0.54 respectively with p<0.0001). Duration of bypass and aortic cross clamp, level of hypothermia, PRISM score and level of inotropic support were significant predictors of ventilatory requirement (p values ranging from <0.0001 to 0.002). Other less strong but significant predictors included pH, lactate and base excess on admission and in the first 6 hours.

CONCLUSION. After the end of cardiopulmonary bypass, impairment of cardiovascular and respiratory function correlated with the levels of interleukin-6 and interleukin-8 as mediators of an inflammatory response. The negative correlation of CXCR2 expression with interleukin-8 and myeloperoxidase indicates that myeloid cells were stimulated by CXC chemokines with Glu-LeuArg (ELR) motif and thereby contributed to tissue damage leading to impairment of cardiovascular and respiratory function.

CONCLUSION. Small and young babies are more likely to require prolonged ventilation following cardiac surgery. Prolonged bypass and aortic cross clamp time were associated with prolonged ventilation. We speculate this may be due to either more complex pathologies or operations or to the process of bypass itself.

REFERENCE(S). (1) Baggiolini M. Chemokines and leukocyte traffic. Nature 1998; 392:565568. Grant acknowledgement: EMDO-Stiftung Zurich, Switzerland

REFERENCE(S). [1] Thompson MJ et al Cardiovasc Surg 1997; 5(4):376-81 [2] Naughton C et al Eur J Anaesthesiol 2003; 20(3):225-33 [3] Kern H et al Intensive Care Med 2001; 27:407-15

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

S189

724

726

NUCLEATED RED BLOOD CELLS AS AN EARLY MARKER OF BRAIN DAMAGE IN ASPHYXIATED NEWBORN INFANTS

ASSOCIATION OF PLATELET COUNTS WITH SIGMOID NECROSIS AND MORTALITY AFTER RUPTURED AORTIC ANEURYSM

Szwajcowska M1, Kawczynski P P1 1Neonatal Intensive Care, Institute of Obstetrics, University School of Medicine, Lodz, Poland

1Surgery,

INTRODUCTION. Nucleated red blood cells (NRBCs)are rarely found in children, but they are commonly seen in the blood of neonates. They are produced and stored in the bone marrow as precursors to reticulocytes and erythrocytes. Many acute or chronic stimuli may increase in the circulating NRBCs from either increased erythropoietic activity or release from the marrow storage pools(1). In our study we try to investigate the validity of outcome prediction after the severe birth asphyxia using the number of NRBCs.

Haveman J W1, Verhoeven E L G1, Karliczek A1, Van den Dungen J J A M1, Nijsten M W N1 Groningen University Hospital, Groningen, Netherlands INTRODUCTION. Patients with ruptured abdominal aortic aneurysm (RAAA) have a hospital mortality of 30 to 70%. Sigmoid necrosis is an important complication that is associated with an increased mortality. Studies in general ICU patients show an association between platelet count (PC) and prognosis. The association between PC and outcome in RAAA patients is largely unknown. METHODS. All patients operated for RAAA between 1990 and 2002 were reviewed.

RESULTS. Number of NRBCs were significantly higher in neonates with brain damage symptoms in comparison to healthy controls. In the study group mean number of NRBCs was 122,2/100 white blood cells, in control group the mean number was 5,8/100 white blood cells. CONCLUSION. At birth, newborn infants who developed symptoms of brain damage had higher NRBCs counts compared to infants without neurological abnormalities. Increased NRBCs counts in umbilical blood samples may represent early marker of perinatal brain damage.

RESULTS. 289 patients (age 71+/-8 years) were operated for RAAA. Mortality was 33% (95/289). In 9% (27/289) a sigmoid resection was performed because of sigmoid necrosis; 20 of these patients died. The figure shows the mean PC (SEM). Non-survivors had lower initial platelet counts (P= 0.024), as well as a poor recovery of PC. In patients with sigmoid necrosis this pattern was even more pronounced (P=0.002). 400

Survivors No sigmoid resection Sigmoid resection Non Survivors

350 300 Platelet count

METHODS. The setting is neonatal intensive care unit in university hospital. We studied 17 neonates who suffered from perinatal asphyxia and/or severe intraventricular haemorrhage and who later developed symptoms of brain damage. Control group were 102 neonates without neurological damages, Mean gestation age was 33,1 weeks (27 - 39), mean birth weight 1468grams (920 - 3250), mean Apgar score 4,06 (1 - 7). There were no significant differences between the groups. A blood sample from ubilical artery was obtained within 5 minutes after birth and NRBCs per 100 white blood cells were counted by light microscopy.

250 200 150 100 50

REFERENCE(S). 1.Hermansen M.C. Nucleated red blood cells in the fetus and newborn. Arch Dis Child fetal Neonatal Ed. 84: F211-F215

0 0

2

4

6

8

10

12

14

16

Days post surgery

CONCLUSION. A low PC and a slow recovery in patients with RAAA are associated with a poor outcome.

727

725-729

ROLE OF INTRAVENOUS ERYTHROMYCIN IN FACILITATING BEDSIDE PLACEMENT OF NASO-JEJUNAL FEEDING TUBES Taori G C1, Patel M1, Puthran P1, Sunavala J D1 1 Critical Care Medicine, Jaslok Hospital and Research Centre, Mumbai, India INTRODUCTION. Post pyloric enteral feeding tubes are increasingly being recognised as a means to improve feeding tolerance and achieve nutritional goals in the ICU setting. The most difficult challenge is to push the tube beyond the gastric pylorus at the bedside in the absence of fluoroscopic guidance. Several techniques including the use of prokinetic agents have been proposed. In view of the documented benefit of oral or intravenous erythromycin in improving gastric motility and emptying, we tried to determine whether use of a bolus dose of intravenous erythromycin facilitates bedside placement of naso-jejunal feeding tubes in adult ICU patients who otherwise were deemed to be at high risk for gastric tube feeding intolerance. METHODS. This was a double blind randomized placebo control study. Prospective cohort of 27 consecutive adult patients requiring ICU care and enteral tube feeding for nutritional support were included. These were randomized to receive a single blinded 100 ml intravenous infusion of either normal saline alone (placebo control; n=17) or 500 mg of erythromycin mixed in normal saline (study patient; n=10).This was followed by insertion of a 12-Fr, 120 cm, Freka®feeding tube using standard bedside protocol by experienced ICU staff. Position of the feeding tube was confirmed by aspirate pH measurement as well as roentgenogram verification. Time required for insertion of tube in the two groups (max 45 min) was noted. RESULTS. Use of intravenous erythromycin significantly improved the rate of naso-jejunal feeding tube placement (Erythromycin group, 10 of 10 patients or 100% vs. the control group, 8 of 17 or 47%; p <.03). Erythromycin administration also decreased the procedure time from 30 (+/- 3 min) to 20(+/-5 min); p < .04 CONCLUSION. Intravenous erythromycin facilitates bedside placement of nasojejunal feeding tube in critically ill adult patients. Single bolus dose also decreases the procedure time required.

INTRAPERITONEAL PRODUCTION OF REPAIR FACTOR AFTER SURGERY FOR PERITONITIS Huynh D1, Paugam-Burtz C1, Dupont H1, Cohen M2, Mantz J1, Dehoux M2 1 Département d’anesthésie-réanimation, 2Biochimie A, Hôpital Bichat, Paris, France INTRODUCTION. Hepatocyte Growth Factor is a repair factor produced after tissular injury to accelerate tissue regeneration (1). Abdominal surgery represents a model of peritoneal agression. This work assessed intraperitoneal (IP) and plasmatic (P) production of HGF after scheldulded colectomy (SC) or peritonitis (PC). METHODS. Daily dosages of IP and P HGF were measured by ELISA (R&D System) from day 0 (intraoperative) to day 5. Results were expressed as median and interquartile range and analyzed by Mann-Whitney or paired test. RESULTS. 20 patients (72 years [38],12 SC and 8 PC) were studied. At day O, IP levels of SC patients were lower than those of PC patients (1189 [755] vs 7240 [8705] pg/ml respectively, p=0.01). In both groups, IP levels were 4 fold greater than P levels (6925 [7180] vs 1307 [1875] respectively, p=0.03) and remained elevated until day 5.

HGF IP (pg/ml)

Oral Presentations Major abdominal surgery 725

16000

SC

12000

PC

8000 4000 0 D0

D1 D2 D3 D4 Postoperative days

D5

CONCLUSION. These preliminary data showed an IP production of HGF at basal state, followed by an early postoperative increase.This production was enhanced in peritonitis. Pronostic value of intraperitoneal production of HGF has to be determined. REFERENCE(S). 1.Funakoshi H et al (2003) Hepatocyte growth factor: from diagnosis to clinical applications, Clin Chim Acta 327:1-23

S190

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

728 COMPARISON OF THE PROKINETIC EFFECT OF CERULEIN AND NEOSTIGMINE Fruhwald S M1, Herk E1, Holzer P2, Hammer H F3, Metzler H1 1 Anaesthesiology and Intensive Care Medicine, 2Pharmacology, 3Internal Medicine, University of Graz, Graz, Austria INTRODUCTION. Therapeutic options for prokinetic therapy of paralytic ileus are the cholecystokinin agonist cerulein and neostigmine (1,2). The aim of our study was to evaluate the prokinetic effect of both drugs on guinea pig small bowel motility in vitro. METHODS. Small bowel segments of 8 cm length were set up in parallel organ bathes. Peristalsis was elicited by luminal perfusion against an aboral resistance of 400 Pascal, peristaltic pressures were recorded at the aboral end of the segments. Perfusion of the segments resulted in an increase of the luminal pressure until a pressure threshold was reached, at which peristaltic contractions were triggered. An increase of the pressure threshold is interpreted as inhibition of peristalsis, while a decrease is interpreted as stimulation of peristalsis. Increasing concentrations of cerulein (0.1-100 nM) or neostigmine (0.03-1muM) were assessed before and after a complete block of peristalsis had been induced by administration of epinephrine (300nM) or sufentanil (1 nM) to the organ bath. Each drug or drug combination was tested on 8 different segments. One way and two way ANOVA for repeated measures were used for statistics (p< 0.05). RESULTS. Cerulein resulted in a dose dependent decrease of the pressure threshold at all tested concentrations. Neostigmine decreased the pressure threshold only at a concentration of 0.03 muM, but resulted in spasms at concentrations > 0.1 muM and a paralysis at a concentration of 1 muM. Cerulein was at all tested concentrations able to restore peristalsis after a complete block of peristalsis had been induced with sufentanil or epinephrine, while neostigmine restored peristalsis only at concentrations <0.1muM. Higher concentrations resulted in a dose dependent inhibition of peristalsis. CONCLUSION. Cerulein dose dependently stimulates peristaltic activity and antagonizes the paralytic effect of high doses of epinephrine as well as sufentanil. In contrast, neostigmine stimulates peristalsis only at lower concentrations, causes spastic activity at moderate concentrations, and results in a complete block of peristalsis at higher concentrations. REFERENCE(S). 1)van der Spoel JI et al. Intensive Care Med 2001; 27: 822-7 2)Schippers E et al. Dig Dis Sci 1991; 36: 621-6

Oral Presentations Inflammation and lung injury – 730-735 730 ACTIVATED PROTEIN C REDUCES ACUTE LUNG INJURY IN ENDOTOXEMIC SHEEP Kuklin V N1 1 Department of Anesthesiology, University of Tromsoe, Tromsoe, Norway INTRODUCTION. Acute lung injury (ALI) frequently develops after severe infections. In the lungs, sepsis-induced inflammation results in increased microvascular pressur, permeability, and derangement of gas exchange that promotes to a high mortality (1). The aim of the investigation was to evaluate the cardiovascular effects of continuously infused activated protein C (APC) on haemodynamics, extravascular lung water (EVLW) andactivation of protein kinase C (PKC)alpha and epsilon isoforms(2) in endotoxemic sheep. METHODS. Twenty two instrumented sheep were subjected to intravenous (IV) infusion of Ringer lactate 3 mL/kg/h for 24 h and randomly assigned to 3 groups: 1) an APC group (n=4) received drotrecogin alpha (Xigris, Eli Lilly & Co,USA) 24 mcg/kg/h as an IV infusion from 4 to 24 h; 2) a LPS group (n=9) received an IV infusion of E.coli lipopolysacharide (LPS) 15 ng/kg/min from 0 to 24 h; 3) an APC+LPS group (n=9) received LPS, and drotrecogin alpha 24 mcg/kg/h as an IV infusion from 4 to 24 h. Haemodynamics, EVLW assessed by a double indicator technique (Cold Z-021; Pulsion Medical Systems, Germany), blood gases, and coagulation parameters were determined every 4 h. Activation of PKC alpha, epsilon was assessed by translocation of PKC from the cytosole by Western blotting. Data were assessed by ANOVA and Scheffe’s test, p<0.05 was regarded as statistically significant. RESULTS. In the APC group, all variables were unchanged for 24 h.In both endotoxemic groups, LPS caused pulmonary hypertension, increased EVLW, and led to arterial hypoxemia. Administration of APC counteracted the LPS-induced increments in pulmonary microwedge pressure by 60%(p<0.05). APC decreased EVLW by 40% as compared to the LPS group and improved oxygenation variables (p<0.05). APC prevented significantly the decreases in protein C and fibrinogen. LPS led to 70% loss of PKC alpha and epsilon from the cytosolic fraction. The translocation of PKCalpha, epsilon was blocked by APC. CONCLUSION. In LPS-induced ALI, continuously infused APC attenuates pulmonary oedema by decreasing the microvascular pressure and EVLW. REFERENCE(S). 1. Martin GS et al. Intensive Care Med 2001;27:S63-S79 2. Siflinger-Birnboom A et al. Am J Physiol Lung Cell Mol Physiol 2003;284:L435–L451

729 DOES INAPPROPRIATE ABDOMINAL REOPERATION MORTALITY OF SEVERE PERITONITIS IN ICU?

731 INCREASE

THE

DIFFERENTIAL CYTOKINE GENE EXPRESSION IN THE DIAPHRAGM IN RESPONSE TO STRENUOUS RESISTIVE BREATHING

Audibert J1, Plantefeve G1, Dupont H1, Mognol P2, Kermarrec N1, De Vaumas C1, Paugam-Burtz C1, Marmuse J2, Desmonts J1, Mantz J1 1 ANESTHESIE-REANIMATION, 2CHIRURGIE DIGESTIVE, HOPITAL BICHAT, PARIS, France

Vassilakopoulos T1, Divangahi M2, Rallis G2, Comtois A2, Hussain S2 1 Critical care, University of Athens, Athens, Greece, 2Meakins-Christie, McGill, Montreal, Canada

INTRODUCTION. The prognosis of peritonitis with organ failure remains severe in ICU. The consensus for surgical reexploration is not established. This study evaluated the impact on mortality of surgical constatations during reexploration for persisting intra-abdominal sepsis.

INTRODUCTION. 1. Strenuous resistive breathing induces plasma cytokines that do not originate from monocytes. 2. Nitric oxide (NO) downregulates cytokine production in vitro. Hypothesis:1.Cytokines are induced in the diaphragm in response to acute resistive loading 2. Endogenously produced NO downregulates this cytokine response

METHODS. From 1994 to 2002, 375 consecutive peritonitis were admitted in surgical ICU (SAPSII : 50 ± 19). Among them, 120 patients were reoperated for suspicion of persisting intraabdominal sepsis. Three groups were compared according to the demographic data and mortality: persisting peritonitis group (PP, n=70) with anatomic etiology and positive microbiological culture of intraoperative sample of peritoneal fluid, intermediate peritonitis group (IP, n=28) without anatomic etiology but positive microbiological culture and negative laparotomy group (NL, n=22)without anatomic etiology and negative microbiological culture. Kruskall-Wallis, MannWhitney and Fisher exact tests were performed, p<0.05 significant. Results were expressed with median [IQR] and with proportion.

METHODS. Anesthetized, tracheostomized, spontaneously breathing Sprague-Dawely rats (n = 8 per group) were subjected to: - 1, 3 or 6 hours inspiratory resistive loading corresponding to 4550% of the maximum inspiratory pressure (MIP). - 1 hour loading followed by 2 hours unloaded breathing - 1 hour intermittent loading (20 minutes loading, 30 minutes unloaded breathing x 3) 1 or 3 hours inspiratory resistive loading after pretreatment with L-NAME (i.p). The diaphragm and the gastrocnemius muscles were excised at the end of the loading period and messenger RNA expression of 11 cytokines (TNF-á, TNF-â, IL-1á, IL-1â, IL-2, IL-3, IL-4, IL-5, IL-6, IL-10, IFNã) and 2 housekeeping genes (L32, GAPDH) was analyzed in these muscles using multi-probe Ribonuclease Protection Assay.

RESULTS. The table reported the comparative data of the 3 groups. Upon reoperation, antibiotic treatment was inappropriate to the pathogens found in 40% (IP) and 43% (PP).

RESULTS. IL-6 and to a lesser extent IL-1b, TNF-á, IL-10, IFN-ã and IL-4 were significantly increased in the diaphragm of animals subjected to inspiratory resistive loading in a time dependent manner. Both intermittent loading (1 hour total duration) and 1 hour loading followed by 2 hours unloaded breathing further augmented the cytokine induction secondary to 1 hour loading. No changes in cytokine expression were observed in the diaphragms of control animals, or in the gastrocnemius muscle. L-NAME significantly upregulated the cytokine induction in the diaphragm.

PP (n=70) IP (n=28) Age (year) 66 [22] 62 [28] Postoperative peritonitis n 48 (69) 12 (43) SAPS II upon admission * 44 [19] 64 [27] SOFA upon admission * 7 [4] 11 [4] SOFA upon reoperation 8 [6] 11 [5] Delay of reoperation (day) 4.5 [5] 5 [4] Length of stay in ICU (day) 18 [18] 17 [18] Mortality in ICU n (%) 39 (51) 14 (47) *p<0.05 between PP and IP and between PP and NL.

NL (n=22) 72 [22] 5 (23) 57 [28] 9 [8] 10 [6] 5 [4.7] 25 [19] 14 (58)

CONCLUSION. In this cohort, NL remain rare (18%). Their prognosis did not differ from persisting peritonitis. Despite comparable organ failures upon reoperation, an inappropriate relaparotomy may not worsen the prognosis.

CONCLUSION. Strenuous resistive breathing induces differential expression of cytokines in the respiratory muscles Endogenously produced ÍÏ suppresses this induction. Grant acknowledgement: CIHR grant to Sabah Hussain

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

S191

732

734

ENDOTHELIAL FUNCTIONAL ALTERATION IN MECHANICALLY-VENTILATED NORMAL HUMAN LUNG

„PROTECTIVE“ MECHANICAL VENTILATION DOES NOT ALTER INFLAMMATORY RESPONSE TO MAJOR SURGERY

1Critical

Athanasiou C1, Kaziani A1, Augustatou C1, Kotanidou A1, Roussos C1, Orfanos S E1 Care and M. Simou Laboratory, Evangelismos Hospital, Athens University Medical School, Athens, Greece

1Anesthesiology

INTRODUCTION. Pulmonary capillary endothelium-bound (PCEB) angiotensin converting enzyme (ACE)activity is an index of pulmonary endothelial function. We have previously shown that PCEB-ACE dysfunction occurs in mechanically ventilated (MV) patients with no acute lung injury (ALI), implying a role of mechanical ventilation per se. However, the contribution of: a) mild lung injury (1) and b) systemic inflammatory response syndrome (SIRS) could not be excluded. We now assess PCEB-ACE activity in MV patients with normal lungs.

INTRODUCTION. Abdominal and thoracic surgery is known to cause release of inflammatory mediators and pulmonary complications were associated with higher intraoperative mediator levels [1]. Mechanical ventilation with higher tidal volumes (VT) and zero or low positive endexpiratory pressure (PEEP) in contrast to lower VT and higher PEEP ventilation was observed to increase mediator release in patients with acute lung injury [2-3] but not in patients with healthy lungs [4]. We hypothesized that mechanical ventilation with lower VT and higher PEEP will also have protective effects on inflammatory responses during major thoracic or abdominal surgery.

METHODS. Applying indicator-dilution type techniques, PCEB-ACE activity was estimated as transpulmonary substrate hydrolysis (v) and percent metabolism (%M), and as the functional capillary surface area (FCSA) index Amax/Km, in 5 MV trauma patients without lung involvement. Three estimations were performed per patient 48 hrs apart each other, starting on the 1st MV day. During the whole study: a) all patients had lung injury scores = 0 (1) and never developed SIRS or sepsis (2), and b) 4 patients were on ZEEP, and one on 4 cmH2O PEEP. RESULTS. Between the 1st and 3rd estimation, all patients exhibited mild significant decreases (one way ANOVARM) in all three PCEB-ACE activity indices (%M: 68.1±6.7 to 56.9±6.8%, p = 0.004; v: 1.31±0.34 to 0.91± 0.2, p = 0.03; Amax/Km: 6199±1052 to 3560±522 ml/min, p = 0.02). Changes occurred in cardiac output (7.39±0.60 to 8.15±0.47 to 5.83±0.37 L/min; p = 0.02), and PaO2/FiO2 (411±29 to 469±14 to 367±16 mmHg; p = 0.01). No differences were observed among estimations in: a) FiO2 (0.48±0.01 to 0.43±0.03), b) tidal volumes (9.1±0.4 to 8.7±0.1 ml/Kg body weight, ranging between 8 and 9.5 ml/Kg), c) APACHE II scores (7.6±2.6 to 5±2.9), d) patients´ body temperatures (37.1±0.6 to 37.3±0.4 0C), e) heart rates (91±9 to 88±11/min), and f) leukocytes (10428±801 to 8134±441/mm3). CONCLUSION. PCEB-ACE activity, a direct index of pulmonary capillary endothelial function, is reduced with time in MV humans with normal lungs (i.e. no history of lung disease, no lung injury, no SIRS or sepsis). This implies the presence of pulmonary endothelial functional alteration probably related, at least in part, to mechanical ventilation.

Wrigge H1, Uhlig U2, Zinserling J1, Behrends-Callsen E1, Uhlig S2, Putensen C1 and Intensive Care Medicine, University of Bonn, Bonn, 2Pulmonary Pharmacology, Research Center Borstel, Borstel, Germany

METHODS. 64 patients undergoing general anaesthesia and elective thoracotomy (n=34) or laparotomy (n=30) were randomized to receive either mechanical ventilation with VT=12 or 15 ml/kg ideal body weight, respectively, and zero PEEP, or VT=6 ml/kg ideal body weight with PEEP of 10 cm H2O. In addition to arterial blood gas and spirometric measurements, tumour necrosis factor, interleukin 1 (IL-1), IL-6, IL-8, IL-10 and IL-12 were determined by cytometric bead array in plasma after 0, 1, 2, and 3 h and in tracheal aspirates after 3 h of mechanical ventilation. Data were log-transformed and analyzed using parametric or non-parametric tests, as indicated. RESULTS. In 62 patients who completed the study, arterial oxygenation was not different between groups. All plasma mediators increased more during abdominal than during thoracic surgery, although the differences were small. However, neither time course nor concentrations of pulmonary or systemic mediators differed between the two ventilation strategies. CONCLUSION. In uninjured normal lungs of patients during major abdominal or thoracic surgery, mechanical ventilation with high VT and zero PEEP does not increase the pulmonary or systemic inflammatory responses to surgery. This observation is indirect evidence for the two-hit hypothesis suggesting that conventional mechanical ventilation may induce lung inflammation to clinically important levels in pre-injured or infected lungs as previously shown, but not in normal lungs even during major surgery.

REFERENCE(S). 1. Murray et al. Am Rev Respir Dis 1988;138:720-723 2. Brun-Buisson. Int Care Med 2000;26:S64-S74 Grant acknowledgement: The Thorax Foundation

REFERENCE(S). 1. Yamada T et al. World J Surgery (1998). 2. Ranieri VM et al. JAMA (1999). 3. Stüber F et al. Intensive Care Med. (2001). 4. Wrigge H et al. Anesthesiology (2000). Grant acknowledgement: BONFOR (O-117.0006), University of Bonn, and DFG (Uh 88/4-1), Bonn, Germany.

733

735

COMPARISON BETWEEN MECHANICAL ALTERATIONS AND MARKERS OF LUNG REMODELING IN THE ARDS

THROMBIN AND IL-8 ACTIVITY IN ARDS IS DEPENDENT ON THE VENTILATION STRATEGY

1

Demoule A1, Decaillot F2, Johnson B1, Maitre B3, Harf A4, Lemaire F1, Brochard L1, Delclaux C4 Service de Reanimation Medicale, 2Departement d’Anesthesie et Reanimation Chirurgicale, 3Antenne de Pneumologie, 4Service des Explorations Fonctionnelles, Hopital Henri Mondor and INSERM U492, Creteil, France

Haitsma J J1, Van Kaam A H L C2, Dik W A3, Kok J H2, Lachmann B1 1 Anesthesiology, Erasmus MC-Faculty, Rotterdam, 2Neonatology, Emma Children Hospital AMC, Amsterdam, 3Immunology, Erasmus MC-Faculty, Rotterdam, Netherlands

INTRODUCTION. The acute respiratory distress syndrome (ARDS) is characterized by a protein-rich oedema associated with lung remodelling involving collagen synthesis/accumulation as well as degradation by collagenases. The aerated lung can be explored by respiratory mechanics or bronchoalveolar lavage (BAL). No study has compared mechanical alterations with biological markers of lung remodelling. The aim of the present study was to compare mechanical alterations of the respiratory system studied through the analysis of pressure volume curves (PVcurves), with the concentration of markers of lung remodelling in BAL. Preliminary results are presented.

INTRODUCTION. High permeability oedema, resulting in inactivation of surfactant is the main characteristic of ARDS. There is increasing evidence that besides its critical role in haemostatis, thrombin is also involved in the process of pulmonary inflammation and fibrosis. Thrombin is known as an important enzyme in the coagulation cascade where it acts upon soluble fibrinogen to form insoluble fibrin. It has been shown that thrombin increases vascular permeability and the production of cytokines like IL-8, thus further enhancing the inflammatory process in acute lung injury. Therefore, we investigated whether a ventilation strategy that reduces ventilator-induced lung injury by actively recruiting and stabilizing collapsed alveoli (the open lung concept) can reduce the increased thrombin activity in ARDS.

METHODS. In patients with ARDS or acute lung injury (ALI), PVcurves were performed in ZEEP and PEEP 10 cmH2O, followed by a BAL. Linear compliance in ZEEP (ClinZEEP) and PEEP (ClinPEEP), and the volume recruited by PEEP at a pressure of 15 cmH2O (Vrecr) were measured. The concentration of Procollagen Peptide III (PCP III, an index of matrix deposition) and type 2 Matrix Metalloprotease (MMP-2, an index of matrix degradation) were measured in BAL. Comparisons were achieved through the Spearman’s rank correlation coefficient. RESULTS. 33 procedures were achieved in 23 patients (age = 54 ± 17 years, SAPSII = 52 ± 19). 21 were ARDS while 2 were ALI: mean (±SD) PaO2/Fi02 = 133 ± 50 mmHg, Lung Injury Score = 2.7 ± 0.3. In early and intermediate ARDS, a negative logarithmic correlation was found between the number of total cells and ClinZEEP (p<0.01). In the whole group, a negative correlation was also shown between ClinZEEP and ClinPEEP and PCP III (p<0.01, Rho=-0.53). MMP-2 was negatively correlated with Clin ZEEP (p<0.01, Rho=-0.49) and Clin PEEP (p<0.01, Rho=-0.56). Vrecr was negatively correlated with PCP III and MMP-2 (p<0.01, Rho=-0.52 for both). CONCLUSION. In ARDS patients, correlations exist between indexes of mechanical abnormalities and biological markers of lung remodelling. This suggests that in case of small aerated lung (as indicated by low compliance), the inflammatory process is present in the lung explored by BAL.

METHODS. 16 piglets were anesthetized, tracheotomized and connected to a ventilator. ARDS was induced by repeated whole lung lavage until PaO2 < 100 mm Hg (FiO2 =1). Subsequently animals were assigned to conventional ventilation PEEP <6 cm H2O and PIP <35 cm H2O to prevent critical hypoxemia. In the other group (OLC) collapsed alveoli were actively recruited and stabilized with minimal pressures according to the open lung concept. Animals were ventilated for 5 hours followed by broncho-alveolar lavage. Thrombin activity was determined using thrombin specific chromogenic substrate as described by Abilgaard en Lottenberg. IL-8 levels were determined by ELISA. RESULTS. Arterial oxygenation improved to prelavage values after application of the OLC, while in the conventional group the PaO2 values did not differ from the post-lavage values. Thrombin activity was significant lower in the OLC group compared to the conventional ventilated group after 5 hours. IL-8 levels were significantly lower in the OLC group CONCLUSION. Application of the OLC optimizes gas exchange while reducing thrombin and IL-8 activity in the lung, demonstrating that ventilation strategy directly influence the progress of lung injury and lung inflammation.

S192

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

Oral Presentations Severe airway infections in ventilated patients – 736-741 736

738 TRANSNASAL PUNCTURE BASED ON ECHOGRAPHIC MAXILLARY SINUSITIS EVIDENCE IN VENTILATED PATIENTS Vargas F1, Gruson D1, Bui N H1, Herbland A1, Bebear C M2, Gbikpi-Benissan G1, Hilbert G1 Médicale, 2Laboratoire de Bactériologie, Hopital Pellegrin Tripode, Bordeaux, France

1Réanimation

RISK FACTORS FOR ICU-ACQUIRED PNEUMONIA CAUSED BY METHICILLINRESISTANT STAPHYLOCOCCUS AUREUS Pereira J1, Alba M1, Cardoso A1, Correia R1, Silva M1, Paiva J1, Sousa-Dias C1, Aguiar L1, Barbosa S1, Gomes J1, Honrado T1, Maia I1, Massada S1, Rios M1, Mota A1 1 UNIDADE DE CUIDADOS INTENSIVOS POLIVALENTE DA URGÊNCIA, HOSPITAL S. JOÃO, PORTO, Portugal INTRODUCTION. Methicillin-resistant Staphylococcus aureus (MRSA) is a frequent cause of ICU-acquired pneumonia (IAP) in Portugal, requiring specific antibiotherapy, that must be early and adequate to achieve success. METHODS. All IAP that have occurred during 24 consecutive months in our ICU were retrospectively studied in order to identify risk factors for IAP caused by MRSA. Mann-Whitney, Chi-square and logistic regression tests were used. RESULTS. There were 111 microbiologically documented IAP, with a median length of stay of 22 days and ICU mortality of 24%. SAPS II on admission was 44. Agents of IAP were: Enterobactereacae 28%, Pseudomonas aeruginosa 21%, MRSA 15%, Hemophilus influenzae 13%, MSSA 11%, Streptococcus pneumoniae 5%, Acinetobacter spp. 4% and Stenotrophomonas maltophila 3%. Comparing MRSA IAP and those caused by other agents, there was no significant difference in terms of sex, age and SAPS II. MRSA IAP occurred significantly later than the others (p<0,001) and the initial empiric antibiotic regimen was significantly more often inadequate (p=0,001). ICU and ventilator days and ICU and hospital mortality was similar in both groups. ¡_7 hospital days (p<0,001), ¡_7 ventilation days (p=0,001), previous use of antibiotics (p=0,001) and COPD (p=0,048) were risk factors for MRSA IAP, by univariate analysis. Multiple logistic regression analysis showed that ¡_7 hospital days was the only factor significantly and independently associated with MRSA IAP (OR=15,99; p=0,009). CONCLUSION. MRSA is a frequent cause of inadequacy of the first line antibiotic regimen for IAP. A high level of suspicion for MRSA should be maintained in all ICU patients with at least one of these factors: ¡_7 hospital days, ¡_7 ventilation days, previous use of antibiotics or COPD. 7 or more days of hospitalization seems to be the strongest risk factor for MRSA as a cause of IAP.

INTRODUCTION. The diagnosis of nosocomial sinusitis is based on the presence of radiologic maxillary sinusitis obtained from computerised tomography (CT) and the isolation of organisms from culture of purulent material obtained after transnasal puncture(TP). The information of sinus echography should minimize referral of critically ill patient for CT. The aim of this prospective study was to evaluate the interest to perform a TP based on the sinus echography results, in intubated patients with suspicion of nosocomial sinusitis. METHODS. 60 patients were included and the ultrasonic procedure was performed. The image defined as normal was an acoustic shadow arising from the front wall. Two levels of positive echography were described: a) a moderate lesion was defined as the visualization of the hyperechogenic posterior wall of the sinus; and b) an important lesion was defined as the hyperechogenic visualization of posterior wall and the extension by the internal wall of the sinus(1). When sinus ultrasound was positive, a TP was performed the same day. The TP was positive if a fluid was obtained from sinus aspiration. The TP was negative if there was no aspirated material. RESULTS. 120 sinus were examined. B-mode ultrasound was positive in 84 cases (respectively 54 important and 30 moderate ultrasound lesion). 78/84 (93%) TP were positive. All TP based on important ultrasound lesion were positive. The only 6 negative TP were performed in patients with moderate ultrasound lesion. CONCLUSION. This study shows that TP could be performed based on echographic maxillary sinusitis evidence REFERENCE(S). Hilbert G., Vargas F., Valentino R. et al: Comparison of B-mode ultrasound and computed tomography in the diagnosis of maxillary sinusitis in mechanically ventilated patients. Crit Care Med 2001; 29:1337–1342

737

739

EFFECT OF ANTIBIOTICS ON THE INFECTED CELLS COUNT AND CULTURES FROM BRONCHOALVEOLAR LAVAGE

INVASIVE ASPERGILLOSIS IN ICU PATIENTS: ANALYSIS FOR RISK FACTORS AND MORTALITY

Da Silva N B1, Tonietto T T T1, Anflor JR L1, Koefender C1, Cardoso P1, Martins L1, Martins F1 1 Laboratório Experimental de Pós-Graduação, FFFCMPA, Porto Alegre, Brazil

Vandewoude K H1, Blot S I M1, Depuydt P1, Colardyn F1 1 Intensive Care Dpt, Ghent University Hospital, GHENT, Belgium

INTRODUCTION. We evaluate the accuracy and usefulness of percentage of infected cells (%ICO) and quantitative cultures of BAL for the diagnosis of pneumonia using a experimental model submitted to treatment or not with antibiotics.

INTRODUCTION. Invasive aspergillosis (IA) is an emerging infection in the ICU. Commonly described specific predisposing risk factors (RF) for IA are prolonged profound neutropenia due to haematologic disease, allogeneic bone marrow and stem cell transplantation, and immunosuppressive therapy. IA has also been reported in apparently immunocompetent patients. Outcome of IA patients with specific RF and apparently immunocompetent patients was compared.

METHODS. We used experimental models of pneumonia in 6 groups of Wistar imunocompetent rats caused by direct intratraqueal innoculation of S. pneumoniae or P. aeruginosa . Animals were randomized for treatment with penicilin (n=20)or not (n=18)in pneumococal model, and amikacin (n=13)or ceftazidime (n=9)or no treatment ( n=13)in Pseudomonal models. After 48h animals were sacrified and immediate BAL were done for citology and microbiology, and the lungs sent for pathology. Results were read blindly for %ICO at 2% and cultures at 1000 ufc/ml thresholds and pneumonia diagnosis considered the grades 2 and 3 of Johanson*. Differences in groups results were compared by t test,Chi2 and Bayes equation, p<0,05 were significant. RESULTS. Comparing to controls BAL tests had 100% specificity to infection. In those with pneumococal pneumonia without treatment %ICO had higher sensitivity comparing to cultures for diagnosis, and in those with treatment cultures had a more drastic drop than %ICO. On Pseudomonal pneumonia there were no differences in sensitivity of %ICO with or without treatment with amikacin, but cultures showed a very low sensitivity, in animals treated with ceftazidime cultures were negatives while %ICO maintain intermediate sensitivity BAL sensitivity % ICO Cultures ufc/ml

S.pneumoniae no antibiotic 94%

S.pneumoniae penicilin 58%

P.aeruginosa no antibiotic 69%

P.aeruginosa amikacin 69%

P.aeruginosa ceftazidime 56%

78%

21%

39%

32%

0

CONCLUSION. Direct examination of infected cells on BAL permits a fast and specific diagnosis for pneumonia. When BAL was done during effective antibiotic treatment there was a more accentuated loss of sensitivity with cultures than with %ICO, but if antibiotics were not effective %ICO was not affected and cultures could have reduction on their thresholds. REFERENCE(S). * Johanson, WG and al. Am Rev Respir Dis,1988;137:259-64

METHODS. In a retrospective cohort study (July 1997 – Dec 2000), IA was diagnosed when proven on histology or in case of an abnormal chest radiography or CT thorax with predisposing specific RF or positive findings (microscopy or culture) on a bronchoalveolar lavage. Haematological malignancy, neutropenia and immunosuppressive therapy were considered as specific RF. Previous lung damage (COPD, viral pneumonia, ARDS), liver failure, burns, severe bacterial infection and malnutrition were categorised as non-specific RF in previously immunocompetent patients. RESULTS. All patients with IA were mechanically ventilated. Patients with specific RF had higher APACHE II scores and related expected mortality. No difference was found between patients with specific and non-specific RF in in-hospital mortality. When ICU patients with specific RF were considered, no difference in observed vs. expected mortality was noted (P=0.940). In patients with non-specific RF, the observed mortality significantly exceeded the expected mortality (P<0.001). Patient characteristics and outcome comparison IA patients with specific RF (n=17)

IA patients with non specific RF (n=21) APACHE II score 31 (7.6) 22 (8.4) Expected mortality 70 (20.7) 40 (23.9) In hosp. mortality 14 (82.4) 15 (71.4) Variables are described as n (%) or mean (SD)

P value 0.003 0.003 0.0476

CONCLUSION. No difference in observed in-hospital mortality was found. Yet, the prognosis of ICU patients with non-specific risk factors worsens by the development of this infection, whereas in ICU patients with specific risk factors, the observed outcome matches the expected mortality as based on APACHE II.

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

740 WHEN SHOULD POTENTIALLY RESISTANT MICROORGANISMS BE COVERED IN ICU-ACQUIRED PNEUMONIA? Pereira J1, Paiva J1, Sousa-Dias C1, Aguiar L1, Barbosa S1, Gomes J1, Honrado T1, Maia I1, Massada S1, Rios M1, Mota A1 1 UNIDADE DE CUIDADOS INTENSIVOS POLIVALENTE DA URGÊNCIA, HOSPITAL S. JOÃO, PORTO, Portugal INTRODUCTION. Among risk factors for mortality from ICU-acquired pneumonia (IAP) only inappropriate antibiotherapy is amenable to modification by clinicians. The incidence of potentially resistant microorganisms (PRMo)is increasing. It is therefore essential to identify risk factors for infection by these pathogens to guide antibiotherapy METHODS. All IAP that have occurred during 24 consecutive months in our ICU were retrospectively studied to identify a simple methodology to guide the selection of empiric antibiotherapy, namely when should PRMo be covered. Methicillin-resistant Staphylococcus aureus (MRSA) and non-fermentative Gram-negative bacilli (NFGNB) were considered the PRMo.Kruskal-Wallis and logistic regression were used. RESULTS. There were 111 microbiologically documented IAP. Median day of diagnosis was day 6 after admission and mortality was 25%. PRMo caused 45%: MRSA 16% and NFGNB 29%. Previous use of antibiotics (OR=6,8) and duration of hospitalization (OR=5,7) ( and not duration of ventilation ) were the only risk factors for IAP caused by PRMo, by multiple regression analysis. Episodes of IAP were divided into four groups: Group 1- <7 days after admission and no previous use of antibiotics; Group 2- <7 days after admission and previous use of antibiotics; Group 3- »7 days after admission and no previous use of antibiotics; Group 4- »7 days after admission and previous use of antibiotics. 80% of IAP caused by PRMo were Group 4. 94% of MRSA IAP and 79% of NFGNB IAP occurred in this group. Only one MRSA IAP occurred outside this group and only 6% of Group 1 IAP were caused by NFGNB. NFGNB caused 6%, 30%, 25% and 20% of Group 1, 2, 3 and 4 IAP, respectively. Susceptibility of all pathogens causing Group 4 IAP was studied and the regimen Vancomycin + Carbapenem ± Aminoglycoside proved to be the most adequate one, covering 92% of the episodes. CONCLUSION. The intersection of two variables – previous use of antibiotics and duration of hospitalization – is the best methodology to select empiric antibiotherapy for IAP. MRSA as an agent of IAP is clearly associated with Group 4 and should only be covered in that context. NFGNB are only unimportant as pathogens in Group 1 and should be covered in the other three groups. Vancomycin + Carbapenem ± Aminoglycoside seems to be the best regimen for Group 4.

S193

Oral Presentations Glucose-metabolism – 742-747 742 CHANGES IN SKELETAL MUSCLE GLYCOLYTIC ENZYME ACTIVITY DURING HUMAN SEPSIS Karyampudi S1, Brealey D1, Smolenski R2, Singer M1 Department of Medicine, Bloomsbury Institute of Intensive Care Medicine, 2Heart Science Centre, Harefield Hospital, London, United Kingdom 1

INTRODUCTION. A correlation exists between low skeletal muscle ATP, mitochondrial dysfunction and poor eventual outcome in human septic shock.1 Glycolytic ATP production may increase to partially compensate for decreased mitochondrial function. We recently reported an increase in activity of the glycolytic rate-limiting enzymes hexokinase (HK), phosphofructokinase (PFK) and pyruvate kinase (PK), with a decrease for glyceraldehyde phosphate dehydrogenase (GAPDH), in a septic rat model at 24 hours.2 We report preliminary studies of glycolytic enzyme activities in human septic shock. METHODS. Vastus lateralis skeletal muscle biopsies were collected from septic shock patients within the first 24 hrs of ICU admission and controls from patients undergoing elective hip replacement. Maximal activity was measured spectrophotometrically for HK, PFK and PK, and for GAPDH, the only glycolytic enzyme sensitive to NO. ATP was measured by HPLC. ANOVA (post hoc LSD) and Pearson correlation test were done using SPSS 11. RESULTS. Samples from 24 septic patients (14 eventual survivors) and 6 controls were analysed. Hexokinase showed increased activity at 24 hours compared to controls (p<0.01), though activity was similar in eventual survivors and non-survivors. PFK activity was significantly decreased in survivors (p<0.04). Strong positive correlations were seen between both ATP levels and PK activity and PFK and GAPDH (table 1). Pearson Correlation ATP PK

HK PFK -0.199 -0.269 -0.144 0.523 * * p<0.01 ** p<0.001

GAPDH 0.668** 0.872**

PK 0.516*

CONCLUSION. High ATP levels, shown to be linked to survival in human sepsis, strongly correlated with increased activities of GAPDH and PK. While these results provide only a snapshot of enzyme activity, they suggest there may be perturbations in the glycolytic pathway which may have important ramifications upon ATP generation in sepsis with impaired mitochondria. This encourages further study, in particular the measurement of glycolytic flux. REFERENCE(S). 1. Brealey D, et al. Lancet 2002;360(9328):219-23. 2. Karyampudi S, et al. Intensive Care Med. 2002;28(134):S96. Grant acknowledgement: Funded by the Wellcome Trust

741

743

RELEVANCE OF CLINICAL DIAGNOSIS OF LUNG INFECTION IN ICU PATIENTS WITH ACUTE BRAIN INJURY

MEAN HYPERGLYCEMIA AS A MEASURE OF GLUCOSE REGULATION

INTRODUCTION. Lung infections(LI) may lengthen ICU stay and worsen secondary damage in patients with acute brain injury. Incidence varies widely according to applied clinical and radiographic criteria, supported by microbiological diagnosis or not. Aim of the study was to evaluate the relevance of clinical suspicious of LI with and without positive microbiological diagnosis in acute brain injury. METHODS. We prospectively investigated the incidence of the clinical suspicious of LI in all patients admitted to our ICU (in 2 years) with acute brain injury (traumatic-TBI; vascular-SAH, post-anoxic and post-operative) in CMV>48 hours. The clinically suspected LI (SLI) were divided in confirmed (CLI: positive microbiological diagnosis) and clinical, non confirmed LI (N-CLI: without microbiological evidence). In addition, we analysed the major clinical variables associated. Data were analysed by Chi-square test analysis. RESULTS. We enrolled 161 patients. SLI was identified in 93 patients (58%). SLI patients compared to those without the suspicious showed a significant difference in the length of ICU stay (12±7 vs 8±4 days, p<0.05), not related to SAPS. SLI incidence was correlated to the severity of neurological damage (GCS<8 in 84% of SLI vs 63% without SLI, p<0.05) and to cerebral diagnosis (more frequent in TBI and SAH; p<0.05). CLI was identified (BAL in 54% of the samples) in 50 patients (31% of the entire population-63% of SLI). There was not a significant difference between CLI and N-CLI, neither in number or type of clinical criteria of pneumonia, ICU stay, GCS at admission time, SAPS, age and outcome. Bacterial airways colonization observed before clinical signs of LI was comparable between CLI and N-CLI in terms of colony count and species of isolated pathogens. There was a difference in antibiotic therapy, preceding the diagnostic sample: 33% in CLI and 66% in the N-CLI (p<0.01). CONCLUSION. Incidence of LI in neuro-ICU was higher in SAH and TBI. The clinical suspicious of pneumonia, by leading to empirical antibiotic therapy, may affect microbiological confirmation. Clinical criteria of LI seem to identify a serious pulmonary pathology, independently from any microbiological result. SLI were associated to increased length ICU stay; moreover, LI associated fever and secretions may, per se, exacerbate secondary brain damage and endorse further and more serious infections. Grant acknowledgement: MIUR

Vogelzang M1, Van der Horst I C C2, Nijsten M W N1 1 Surgical ICU, 2Internal Medicine, Groningen University Hospital, Groningen, Netherlands INTRODUCTION. Now that algorithms are implemented to achieve normoglycemia in ICU patients, the choice of the parameter that such algorithms should optimize becomes critical. Among others, mean glucose, mean morning glucose or glucose at admission all have specific disadvantages. Theoretically, mean hyperglycemia (MHG) is a very attractive parameter to assess glucose regulation. METHODS. MHG is defined as the area under the glucose-curve above 6.0 mmol/l divided by the total length of stay (figure 1, left). We compared MHG with other measures as a predictor of 30-day mortality in a large set of patients from our surgical ICU from 1990 to 2001 with a prolonged stay (> 4 days). To assess the difference we calculated receiver-operator characteristics (ROCs) and used a binary logistic model including age, length of stay, sex and reason of admission. RESULTS. A total of 1938 admissions were included (16% mortality). The mean age was 51±23 year. The median (IQR) length of stay was 10 (6-19) days. The median (IQR) MHG was 7.0 (6.48.3) and differed significantly between survivors and non-survivors (6.9 (6.3-8.0) versus 7.8 (6.89.3), p < 0.001). The MHG had the highest area under the ROC curve. The p-values of the glucose-related parameters of the binary logistic model are shown in figure 1 (right).

Mean hyperglycemia (MHG) Glucose

Maggioni E1, Cormio M1, Lampati L1, Patruno A1, Citerio G1, Pesenti A1 1 Anaesthesia and Intensive Care, University of Milano-Bicocca, S.Gerardo Hospital, Monza, Italy

MHG: 7.2 mmol/l

Time

Variable MHG Arithmetic mean Maximum glucose during stay Admission glucose Average morning glucose

p <0.001 0.03 0.07 0.14 0.21

CONCLUSION. MHG is the best way to quantify the quality of glucose regulation.

S194

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

744

746

EVALUATION OF LACTATE AND GLUCOSE METABOLISM IN SEVERE SEPSIS OR CARDIOGENIC SHOCK

A NURSE DRIVEN INSULIN PROTOCOL: SAFETY AND GLUCOSE RESULTS ON A MEDICAL /SURGICAL ICU

Revelly J P1, Martinez A1, Tappy L2, Cayeux M C1, Berger M M1, Chioléro R1 1Surgical intensive care unit, 2Physiology, University Hospital - CHUV, Lausanne, Switzerland

1Intensive

INTRODUCTION. In critical illness, measurements of plasma lactate concentration do not allow to determine the relative importance of increased lactate production (related to dysoxia or factors increasing Na/K ATPase activity), as opposed to decreased utilization. The aim of the present study was to evaluate the relative importance of these mechanisms, as well as their relation to glucose metabolism (the major precursor of lactate), in patients suffering form either severe sepsis or cardiogenic shock. METHODS. Patients with severe sepsis (SS), cardiogenic shock (CS) and healthy volunteers (HV) were studied. Labelled sodium 13C-lactate was infused at a rate of 10 mumol/kg/min/ for 2 hours. Lactate clearance (LC) was computed in a pharmacokinetic model, as the steady state concentration minus the initial concentration, divided by the dose. Endogenous lactate production (LP) was determined as the initial lactate concentration multiplied by LC. Glucose turnover was assessed by a 2H-glucose infusion. Comparisons between groups were done by one way ANOVA (mean + SD). RESULTS. In the two patient groups, increased lactate was related to an increased production, while LC was not different from the healthy volunteers. Both glucose plasma level and turnover were increased in the two groups of patients. Results HV SS Plasma Lactate (mmol/l) 0.9 ± 0.20 3.2 ± 2.6* LC (ml/kg/min) 12.0 ± 2.6 10.8 ± 5.4 LP (mumol/kg/min) 11.2 ± 2.7 26.2 ± 10.5* Plasma Glucose (mmol/l) 5.2 ± 0.2 9.3 ± 1.1* Glucose turnover (mumol/kg 10.1 ± 1.6 20.8 ± 3.2* *: different from HV, p < 0.05

CS 2.8 ± 0.4* 9.6 ± 2.1 26.6 ± 5.1* 10.0 ± 0.4* 20.4 ± 2.3*

CONCLUSION. In patients suffering from severe sepsis or cardiogenic shock, hyperlactatemia was related to increased production rather than impaired utilisation, while glucose flow was also increased.

Kingma W P1, Viertelhauzen S1, Nijkamp H2, Van der Voort P H3 Care, 2Clinical Chemistry Department, Groene Hart Ziekenhuis, Gouda, 3Intensive Care, Medical Center Leeuwarden, Leeuwarden, Netherlands INTRODUCTION. Recently, Van den Berghe et al [1] showed that strict glycemic control reduces mortality and morbidity on a surgical ICU. We implemented a nurse-driven adjusted Leuven protocol in standard daily care and questioned whether we could control blood glucose safely and effectively on our 8-bed medical/surgical ICU. METHODS. Retrospectively all the measured blood glucoses from all the patients who were treated more than 24 hours during Sept-Nov 2001 (episode 1)were taken. In episode 1 doctors targeted blood glucose levels 8-10 mmol/l. We compared these data with Sept-Nov 2002 (episode 2) where ICU nurses, by protocol, targeted glucose levels 4.4-6.1 mmol/l. All plasmaglucose levels were measured by POCT, Rapidlab. Hypoglycaemia was defined as blood glucose <2.2 mmol/l. RESULTS. Episode 1: 35 patients (19 medical/ 16 surgical). Episode 2: 39 patients (17 medical/ 22 surgical). Mean number glucose measurements/patient was 42 in episode 1 and 89 in episode 2 (total number of measurements 1487 versus 3476). Mean glucose measurements/patient/day is 3.6 versus 7.6. The distribution is shown in table 1.(* p<0.05, Chi square) Glucose level mmol/l 0.0-2.2 2.3-3.7 3.8-4.3 4.4-6.1 6.2-9.9 10-20 > 20

Episode 1 % of measurements 0 0.1 0.3 8.8 57.7 32.1 1.0

Episode 2 % of measurements 0.1 3.0 * 6.6 * 42.2 * 39.2 * 8.2 * 0.8

CONCLUSION. The improved glycaemic control in the treatment group without associated hypoglycaemia shows that a nurse-driven protocol for glucose control with insulin is effective and safe on a mixed medical and surgical ICU. REFERENCE(S). [1] Van den Berghe et al. Intensive insulin therapy in critically ill patients. NEJM 2001;345:1359

745

747

THE DIURNAL VARIATION IN PLASMINOGEN ACTIVATOR INHIBITOR-1 IS IMPAIRED DURING HYPERGLYCAEMIA

LONGITUDINAL ASSESSMENT OF THE HYPERMETABOLIC RESPONSE TO THERMAL INJURY IN CHILDREN

Mullan B A1, McCance D R2 1 Regional Intensive Care Unit, 2Regional Metabolic Unit, The Royal Victoria Hospital, Belfast, United Kingdom

Mlcak R P1, Suman O E2, Cortiella J3, Herndon D N4 1 Respiratory Care, Shriners Hospital for Children, Galveston, 2Surgery, University of Texas Medical Branch, Galveston, Texas, 3Anesthesiology, University of Texas Medical Branch, 4Surgery, Shriners Hospital for Children and the University of Texas Medical Branch, Galveston, United States

INTRODUCTION. Stress-induced hyperglycaemia increases morbidity and mortality in critical illness. High levels of plasminogen activator inhibitor-1 (PAI-1) may contribute to the pathophysiology of organ dysfunction. We investigated the effects of acute hyperglycaemia on plasma PAI-1 concentrations. METHODS. 12 healthy male volunteers were studied in the morning on 2 separate occasions. They were randomised in a single-blind, cross-over manner to hyperglycaemia, in which plasma glucose concentrations were maintained at 14 mmol.l-1 using an intravenous dextrose infusion, or to euglycaemia, in which equivalent volumes of saline were infused instead of dextrose. Octreotide was administered on both occasions to maintain plasma insulin at basal concentrations. Blood samples were obtained at 09:00 hrs and at 15:00 hrs under each glycaemic condition. RESULTS. During euglycaemia, plasma PAI-1 concentrations displayed normal diurnal variation (17.2 ± 6.2 IU.ml-1, 09:00 hrs versus 10.4 ± 2.8 IU.ml-1, 15:00 hrs; p<0.001). However, during hyperglycaemia PAI-1 concentrations remained elevated at 15:00 hrs (16.9 ± 7.2 IU.ml-1, 09:00 hrs versus 16.7 ± 8.4 IU.ml-1, 15:00 hrs; p=0.921). This difference in response between hyperglycaemia and euglycaemia was significant (p=0.011, two-way repeated measures ANOVA). CONCLUSION. The normal diurnal variation in plasma PAI-1 concentration is impaired during acute hyperglycaemia. This can result in abnormal fibrinolysis and may explain some of the adverse clinical effects of hyperglycaemia in intensive care.

INTRODUCTION. The catecholamine-mediated hypermetabolic response to severe burn trauma is associated with increased resting energy expenditure (REE) and catabolism, which persist for up to 9 months post-burn. However, it is presently unknown if this burn induced elevated REE persist longer than 9 months. Therefore the purpose of this study was to assess REE for up to 2 years post-burn in children. METHODS. Fifty-nine patients with >40% TBSA burns were enrolled in a prospective, longitudinal study. REE was measured by indirect calorimetry and compared to predicted values according to the Harris-Benedict equation. Data were collected at discharge, 6, 9, 12, 18 and 24 months’ post-burn. Statistical analysis was completed by using one way repeated measures ANOVA followed by Tukey’s test for multiple comparisons. A p value of <0.05 was considered statistically significant. RESULTS. Mean TBSA burned was 63±16% and mean age was 7±4 years. The actual REE was significantly lower at 18 months (p<0.001) and 24 months(p<0.006) compared to discharge. In addition, the % predicted REE was significantly lower at 12 months (p<0.007), 18 months (p<0.001) and 24 months (p<0.001) compared to discharge. By 18 months post-burn, REE had returned to normal levels. CONCLUSION. In severely burned children, REE remains elevated above predicted levels for 12 months and returns to normal by 18 months post-burn. This suggest that therapeutic attempts to manipulate the hypermetabolic response to severe burn should continue long after injury.

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

Oral Presentations Organ support – 748-753 748 EFFECT OF ALBUMIN DIALYSIS ON PLATELET FUNCTION IN PATIENTS WITH LIVER FAILURE. Faybik P1, Unger S1, Baker A1, Krenn C G1, Kozek-Langenecker S A1, Werba A1, Steltzer H1, Hetz H1 1 Department of Anaesthesia and Critical Care Medicine, University Hospital, Vienna, Austria INTRODUCTION. Acute liver failure (ALF) is associated with impaired synthesis and clearance of the coagulation factors and inhibitory proteins, enhanced fibrinolysis and quantitative and qualitative platelet defects. Patients with ALF may benefit from temporary extracorporeal liver support utilizing excretion and detoxification such as Molecular Adsorbent Recirculating System (MARS). In this study, the effects of MARS on changes in platelet-related haemostatic parameters assessed by thromboelastography (TEG) have been evaluated in patients with ALF. METHODS. 40 MARS treatments in 21 patients with ALF due to liver resection (2), viral hepatitis (2), autoimmune hepatitis (2), mushroom intoxication (5), paracetamol intoxication (1), sepsis (3), graft dysfunction after liver transplantation (5), and liver failure in cystic fibrosis (1) have been evaluated. Citrated whole blood was analysed using conventional celite TEG and abciximab-fab-modified TEG. The difference in maximal amplitude (dMA) between these two TEGs allows the quantification of platelet contribution to clot firmness. All measurements were performed immediately before start (TP1), 30 min after start (TP2) and after MARS treatment (TP3). Patients were anticoagulated with prostacyclin and heparin to achieve an activating clotting time of 140-160s. RESULTS. We observed a decrease in dMA from TP1 (21+10mm) to TP2 (16+15mm) and a significant increase in dMA in TP3 (30+22mm) compared to TP1 (p<0,034) and TP2 (p<0,006). The first was caused by anticoagulation with prostacyclin and shows an adequate protection from platelet activation during the MARS treatments. The latter may result from the increase of free albumin binding sites and consecutive decrease of free platelet inhibitory factors bound to albumin, such as nitric oxide. Mean loss of platelets/MARS treatment was 14+14 G/l.

S195

750 IMPROVEMENT IN SYSTEMIC HEMODYNAMICS AFTER MARS IS ASSOCIATED WITH A DECREASE IN VASOACTIVE AGENTS Escorsell A1, Pares A1, Cisneros L1, Salmeron J M1, Mas A1, Jienez W1, Torras A2, Caballeria J1, Rodes J1 1Liver Unit, 2Nephrology Unit, IMD, Hospital Clínic, Barcelona, Spain INTRODUCTION. The effects of albumin dialysis on systemic haemodynamics and vasoactive agents has been evaluated in 26 molecular adsorbent recirculating system (MARS) procedures performed in 10 patients (7M/3F, age: 48.4 ± 2.0 years) with clinical and histologic diagnosis of severe alcoholic hepatitis defined by a total bilirubin higher than 10 mg/dl, and prothrombin index lower than 50%. METHODS. Besides standard haematological and liver function tests, systemic haemodynamics and vasoactive agents such as plasma renin activity (PRA), angiotensin II and the catabolic products of nitric oxide (NOx) were determined before and after each treatment. None of the patients had infection or upper gastrointestinal bleeding. RESULTS. Besides the favourable effects on serum bilirubin, MARS treatment resulted in increased mean arterial pressure (MAP) and systemic vascular resistance index (SVRI). These circulatory effects were associated with significant decreases in PRA (5.6 ± 0.9 to 3.1 ± 0.5 ng/mL/h, p<0.01), angiotensin II (61.0 ± 8.8 to 44.8 ± 6.1 pg/mL, p<0.05) and NOx (55.4 ± 4.0 to 45.9 ± 2.9 nM/mL, p<0.01). MAP increased in 19 procedures (group I) and decreased in 7 (group II). PRA, angiotensin II and NOx diminished particularly in group I, which were the patients with a significantly lower leukocyte count at baseline (11.7 ± 1.4 versus 17.3 ± 2.4 x 109/l, p<0.05). No other baseline differences were observed between the two groups. CONCLUSION. Albumin dialysis results in conspicuous changes in systemic haemodynamics, which are associated with marked decreases in vasoactive agents. These changes may explain, in part, the favourable effects of this procedure in patients with severe liver failure.

CONCLUSION. According to our results, MARS treatment significantly improves platelet function assessed with TEG in patients with ALF. Neither adverse effects associated with hypotension nor appearance of platelets aggregates in the extracorporeal circulation complicated the MARS treatments. Due to the high biocompatibility of membrane used, MARS appears to be safe in patients with ALF and increased risk of bleeding.

749

751

FILTER-RUN-TIME IN CVVH: REGIONAL VS SYSTEMIC ANTI-COAGULATION, PRE- VS POSTDILUTION, BIC VS LACTATE

APPROPRIATENESS OF RESOURCE USE IN INTENSIVE CARE UNIT

Van der Voort P H J1, Gerritsen R T H1, Kuiper M A1, Egbers P H M1 1 Intensive Care, Medical Centre Leeuwarden, Leeuwarden, Netherlands

1

INTRODUCTION. Filter-run-time(FRT)during Continuous Veno-Venous Haemofiltration (CVVH) may depend on anti-coagulation (regional or systemic), pre- or postdilution and maybe the substitution fluid itself (in pre-dilution mode). METHODS. We performed 3 similar studies: prospective, randomised and crossover. Study A compared pre- versus post-dilution; study B, in pre-dilution mode, bicarbonate buffered substitution fluids versus lactate and study C, regional anticoagulation with heparin-protamin (HP) versus nadroparin (NP) pre-filter. We standardised ultrafiltration rate (3 l/hr), blood flow (200 ml/min) and filter (cellulose-tri-acetate, 1.6m2 and cut-off 60.000 D). Always a bolus of nadroparin (2850IE) was given at the start of the CVVH before the filter. A session was ended when the transmembrane pressure reached 250 mmHg. We performed non-parametric tests and multiple linear regression analysis. RESULTS. Study A: In 16 different patients, 32 CVVH sessions were performed. 11 patients were male; the mean age was 69.6 years. The median APACHE II score was 27. The mean creatinine clearance during pre-dilution was significantly lower (p=0.001) compared to postdilution: 33 ml/min (range 28 to 39), versus 45 ml/min (range 30 to 48). During pre-dilution the median FRT was 45.7 hours (IQR 16.2-48.0) compared to 16.1 hours (IQR 10.5-36.5) in postdilution CVVH (p=0.005). Per session, the median total amount of blood cleared by post-dilution was 37.8 litres, compared to 80.9 litres for pre-dilution (p=0.53). All other variables were not significantly different between pre-dilution and post-dilution and not related to FRT in pre- or post-dilution. Study B: In 10 patients (20 CVVH sessions) the FRT, using the bicarbonate buffered fluid, was 25.5 hours (IQR 22-42) versus 26.5 hours using lactate (IQR 18-40), p=0.62. Study C: 15 patients (30 CVVH sessions) were eligible. During NP, median FRT was 39.5 hours (IQR 8.548) compared to 12.3 hours (IQR 7.5-27) during HP CVVH (p=0.045). None of the other variables differed significantly between NP and HP. None of the tested variables was significantly related to filter survival time in NP or HP. CONCLUSION. FRT in pre-dilution is greater compared to post-dilution but results in lower clearance. FRT in pre-dilution is independent of substitution fluid composition. Regional anticoagulation with heparin-protamin resulted in a significantly shorter filter-run-time compared to nadroparin anticoagulation pre-filter.

Pezzi A1, Noto A1, Bassi G1, Iapichino G1 Istituto di Anestesia e Rianimazione, Università degli Studi-Ospedale S.Paolo, Milano, Italy INTRODUCTION. We describe a method to measure the appropriateness of ICU resources use (available personnel & technology), applying the level of care classification that defines high (HT) or low (LT) treatment (Iapichino, Intensive Care Med 27:131). METHODS. We assumed as able to provide high level of care the bed supplied with tech and at least 0.5 nurse on 24 hours. For each of 32 ICUs, we calculated the number of high level beds available (number of nurses in the poorest daily shift x 2) and the overall number of predicted HT days in the study-period. We classified as HT or LT each of the 32 days (4 per week) studied in a 2 months period. The predicted capacity of the ICU in delivering LT days was computed as: (not delivered HT days/2)x3, where: 2=HT patients cared by one nurse, 3=LT patients cared by one nurse. Two occupancy rates (ORs) per ICU were calculated: OR-HT beds = (actually provided HT days/ predicted HT days) x 100, OR-LT beds = (actually provided LT days / predicted LT days) x 100. Each ICU was classified according to the combination of OR-HT (<80%, 81-100%, >100%) and OR-LT(<50%, 50-100%, >100%).

RESULTS. Treatment-days provided in the study-period were 5.246 (3.956 HT and 1.290 LT). The occupancy rates (mean (SD)) for all 32 ICUs resulted: OR-HT = 77.6 (32.5), OR-LT = 65.4 (78.9). Distribution of ICUs according to ORs: OR-HT>100% (1; 6 ICUs), OR-HT=81-100% and OR-LT>100% (2a; 6 ICUs) or OR-LT=50-100% (2b; 2 ICUs), OR-HT<80% and OR-LT>100% (3a; 2 ICUs) or OR-LT=50-100% (3b; 8 ICUs) or OR-LT<50% (3c; 8 ICUs). CONCLUSION. The following ICU scenarios are identified: (1) Overcrowded, too small ICUs. Unsafe condition for HT patients receiving less nurse assistance than needed. (2a) ICUs with reasonable HT activity treating too many LT patients. Possible unsafe scenario for HT/LT patients. (2b) ICU management able to adjust a HT and LT mix to appropriately use resources. The scenario (3) shows scarce HT activity with respect to facilities for all subgroups. In addition: (3a) Too many LT patients for residual nurse availability. Possible unsafe scenario as in (2a). (3b) Appropriate use of available resources. (3c) Oversized ICUs. The proposed method is easy to perform and gives to the ICU manager a reliable tool to check resource allocation in ICU.

S196

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

752 AN EVALUATION OF METHODS USED TO DETERMINE TEMPERATURE IN CRITICALLY ILL ADULTS Farnell S J1, Maxwell L1, Tan S1, Rhodes A1, Philips B1 1ICU, St George’s Hospital, London, United Kingdom INTRODUCTION. A number of techniques are available for measuring temperature in critically ill adults. This study assesses accuracy and reliability of two non-invasive methods, Tempa.DOT (TD) and Tympanic Thermometer (TT), against the gold standard Pulmonary Artery Catheter (PAC), and aims to determine the clinical significance of any discrepancies. METHODS. Twenty-five adult patients were recruited to the study. Patient temperatures were recorded with each of the three methods and an expert panel used to identify whether any temperature differences were clinically significant. RESULTS. 153 datasets were obtained from 25 patients. Of these 75.2% (n=115) of TD and 50.9% (n=78) of TT readings were within a +/-0.4oC range of the PAC. Both the TD and TT readings were significantly correlated with the temperature derived from the PAC (r2 = 0.81, p<0.0001 and r2 =0.59, p<0.0001) and this was supported by Bland Altman plots (Table 1). Clinically significant temperature differences, which might have resulted in delayed or unnecessary interventions, were noted (Table 2). Temperature Bias and Precision Temperature Method Tempa.DOT Tympanic Thermometer

Bias (oC) 0.2 0

1.96xSD +/-0.67 +/-1.16

Precision (oC) -0.5-0.8 -1.1-1.2

Clinical Significance of Temperature Discrepancy Temperature Method Tempa.DOT Tympanic Thermometer

Intervention Required 15.3% (n=26) 21.1% (n=35)

Unnecessary Intervention 28.8% (n=44) 37.8% (n=58)

CONCLUSION. Despite the TD having an inherent bias of 0.2oC this method was more accurate, reliable, and associated with fewer clinically significant temperature differences. However, a limitation of this method is the temperature range (35.5-40.4oC); as such it may be acceptable to use the TT in these instances.

Oral Presentations Ethical challenges in clinics and research – 754-759 754 PUBLIC PERCEPTION OF EMERGENCY RESEARCH - A QUESTIONNAIRE Booth M G1, Lind A2, Read E2, Kinsella J3 Department of Anaesthesia and Intensive Care, Royal Infirmary, 2Faculty of Medicine, 3 Department of Anaesthesia and Intensive Care, University of Glasgow, Glasgow, United Kingdom

1

INTRODUCTION. Emergency research (e.g. into cardiac arrest or head injury) needs to start immediately, often before the patient, or relative can give consent. A European Directive will, from, May 2004 prevent or severely limit emergency research (1). Little is known of the publics’ view on emergency research. METHODS. In order to ascertain the publics’ perception of emergency research patients attending the outpatient department of a university hospital were invited to complete a selfadministered questionnaire. Research Ethics Committee approval was obtained and all participants gave written informed consent. RESULTS. Three hundred and five of 362 respondents (84%) thought emergency research should start in the absence of consent but consent should be obtained as soon as possible from the nearest relative (82%) or the patient (90%). If consent is refused, 62% felt the data could still be used, as did 81% if the patient died. Despite 62% approving of public meetings to publicise emergency research activity only 35% would attend one. A list of preconditions for emergency research was generally approved (2). These were: no other volunteers group (47%), advance consent impossible (55%), unable to delay treatment (73%), consent to be obtained as soon as possible (88%), an adequately designed protocol (74%), ethics committee approval (71%), patient may benefit (85%), future patients may benefit (92%), the treatment is necessary and cannot be delayed (91%). Most (92%) were willing to be recruited if there were minimal risks involved and 67% if the risks were moderate. CONCLUSION. Emergency research must occur to improve the outcome from life-threatening illness or injury. This study suggests that the majority of people are aware of the importance of this research and accept that the normal rules of consent are not applicable. Deferred consent would appear to be an acceptable alternative. REFERENCE(S). 1. European Union Directive 2001/20/EC on Good Clinical Practice in Clinical Trials. Official Journal of the European Communities 2001; L121: 34-44. 2. Lötjönen S. Medical research in clinical emergency settings in Europe. Journal of Medical Ethics 2002; 28: 183-187.

753

755

INFLUENCE OF AMBIENT AIR TEMPERATURE ON PERFORMANCES OF DIFFERENT GENERATIONS OF HEATED HUMIDIFIERS

TRIAGING PATIENTS TO THE ICU: A MULTICENTER STUDY OF FACTORS INFLUENCING ADMISSION DECISIONS

Lellouche F1, Qader S1, Taillé S1, Brochard L1 1 Medical ICU, INSERM U 492, Henri Mondor Hospital, University Paris XII, CRETEIL, France

Garrouste-Orgeas M M G O1, Montesino L L M2, Moreau D D M3, Reignier J J R4, Desmettre T T D5, Boussat S S B6, Karoubi P P K7, Duguet A A D8, Montuclard L L M9 1 Intensive Care, Hopital Saint Joseph, 2Intensive Care, Hopital Bichat, 3Intensive Care, Hopital Saint Louis, Paris, 4Intensive Care, Hopital les Audaries, La Roche sur Yon, 5Intensive Care, Hopital Germont Gauthier, Béthune, 6Intensive Care, Hopital Central, Nancy, 7Intensive Care, Hopital Avicenne, Bobigny, 8Intensive Care, Louis Mourier hospital, Colombes, 9Ethical Department, Necker University, Paris, France

INTRODUCTION. We recently shown that external conditions and particularly ambient air temperature greatly influence the performances of recent heated humidifiers (HH). This effect is explained by the specific regulation of HH with heated wire on inspiratory line. The aim of the present study was to evaluate the impact of ambient air temperature on heated humidifiers with different working principles. METHODS. Several HH have been tested on bench with different settings : MR 730 (F&P) (37/40, 40/40), MR 850 with and without compensation, MR 450 (F&P), Conchatherm III (Hudson) (37/39, 39/39), Aerodyne (Kendall) (37/39, 39/39), Ultratherm (Kendall) (32, 35 et 37°C), Aquapor (Drager). A T-Bird ventilator set in volume control ventilation was used and connected to a balloon. Hygrometry of inspired gas (absolute and relative humidity) was measured at steady state with psychrometric method. Measurements were performed at two levels of ambient air temperature: « normal », between 22 and 24°C, « high », between 28 and 30°C. RESULTS. With oldest generation of heated humidifier, performances were not influenced by ambient air temperature. Performances were greatly influenced by ambient air temperature only for heated humidification with heated wire as previously shown. With high ambient air temperature, in usual settings (37°C at humidification chamber, 40°C at Y-piece), absolute humidity was between 22 and 25 mg H2O/L, while with „normal“ ambient air temperature absolute humidity was between 36.4 and 37 mg H2O/L( p<0.001). CONCLUSION. First generations of heated humidifiers are not influenced by ambient air temperature with tested settings with high levels of absolute humidity reached. We confirmed in this study previous results concerning heated humidifiers with heated wires : high ambient air temperature can result in very low performances of these humidification devices.

INTRODUCTION. Current criteria used to grant or refuse ICU admission are not based on scientific data and may vary from one institution to another, depending of the triaging physician. Few studies have shown the variability of the rate of ICU admission. The objective of the study was to assess the appropriateness of ICU triage decisions in 11 ICUs. METHODS. Prospective study (June 2001) In France. Age, underlying diseases, admission diagnosis, Mortality Probability Model (MPM0) score, and mortality were recorded. No ICU had written admission or discharge criteria. RESULTS. 572 admissions decisions were analyzed: 433 (75.5%) patients were admitted including 49 transferred to other ICUs. 139 (24.3%) were refused ICU admission. 77 pts were considered „too well to benefit“ and 51 „too sick to benefit“. One family refused ICU admission and in 9 cases the ICU was too busy to admit the patient. The rate of refusal varied from 7.1% to 64% with 5 centres above 20%. 170/538 (31.5%) pts died in the hospital. The MPM0 was more discriminating for the refused vs the admitted pts (AUC: 0.803 vs 0.686, p=0.01). Factors significant in the univariate analysis were: age, medical status, McCabe score, self dependency, bed availability, ICU physician experience, metastatic cancer, hours 8AM-6 PM. These variables were introduced in a multivariable analysis using stepwise forward logistic regression in which centres were separated using their rate of refusal (> 20%). At the last step, factors significantly associated with ICU refusal were (OR, 95% CI, p): pt totally dependent at home (18.1, 6.87-48.1, <0.0001), metastatic cancer (5.72, 2.10-15.5, 0.0001), the centre (4.91, 2.88-8.40, 0.0001), no bed available (4.82, 2.86-8.12, 0.0001), able to walk with help (2.84, 1.49-5.40, 0.003), McCabe score 2 (0.55, 0.31-0.96, 0.03), hours 8H AM-6 PM (0.46, 0.28-0.75, 0.0001). CONCLUSION. In addition to the patients related factors, this study suggests that decision to refuse ICU admission is dependent on centres. The organizational process in the ICU and hospital is likely to explain these results. Grant acknowledgement: SRLF

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

S197

756

758

WITHHOLDING AND WITHDRAWING LIFE-SAVING TREATMENTS: CAN WE DO IT BETTER?

END-OF-LIFE (EOL) CARE: POSITION OF INTENSIVE CARE SOCIETIES FROM EUROPEAN COUNTRIES

Martínez K1, Osés I1, Villanueva N1, Loinaz M1, Ansotegui A1, Díaz A1 1ICU, Nafarroako Ospitalea, Iruña-Pamplona, Spain

1Medical

INTRODUCTION. One of the most difficult challenges in an ICU is how to make the transition from an intensive treatment to a palliative one with optimal care. There is an extreme variability in these decisions. This study was carried out to know the opinions of the beloved ones of these patients and to analyze its ethical problems. METHODS. A telephonic interview was done to the closest person for the patient, 6-9 months after the decision to withhold/withdraw LST was taken. Part of the interview was designed to ask for informed consent. Statistical analysis performed was X2 test. RESULTS. We talked to 40 siblings of 43 patients. 61% thinks that the patient had a good death. 67,5% does not know if the patient had ever stated her preferences about death. 15% does not know that LST has been withheld/withdrawn. 13% does not remember whose the decision was. 9% thinks that the decision was taken too soon and 12%, too late. Even though, 87% considers it a correct decision. 79% would make the same decision anyhow. Nobody says they would not make it again. No significant relationship was found between any of the personal characteristics of the interviewees and the evaluation of the studied process. CONCLUSION. The interview was very well accepted. Information was well qualified but communication did not score so high. Relationship with families was not empathic enough in 17% of the cases. More than a third did not know the patients’ wishes as far as death was concerned. One fifth qualified their siblings death as bad. The withholding/withdrawing was not known or denied by 22,5%. Even though the vast majority considered it a correct decision. We still need more accurate indicators for deciding. Anyhow, procedural excellence will always be of great value. REFERENCE(S). Rocker GM, Shemie SD, Lacroix J. End-of-life issues in the ICU: A need for acute palliative care? J Palliat Care 2000; 16S: 5-6

Boles J1, L’Her E1, Prat G1, Boumedienne A1, Goetghebeur D1, Renault A1 Intensive Care Unit, University Hospital « Cavale Blanche », Brest, France INTRODUCTION. Limitation of life-sustaining treatment (LST) and EOL care are major issues in ICUs. We reviewed European countries professional society statements to analyze and compare their recommendations. METHODS. an e-mail questionnaire was sent to 19 intensive care societies (ICS) from 16 European countries asking whether they had adopted an official statement on EOL care and what recommendations they formulated on limitation of LST. A query was conducted on ICS websites and position papers were analyzed when possible. RESULTS. 16 societies (84%) from 13 countries (81%) answered the questionnaire. Seven out of 16 societies (44%) have an ethics committee and 4 other (25%) either a special working group or another committee addressing ethical issues. Nine out of 16 (56%) have adopted an official statement on EOL care which was approved by the board of administrators: Belgium, Finland, the 2 French societies, Germany, Great Britain, Italy, Spain and Switzerland; 4 did so before 2000 (44%). The Swiss and British national medical associations also adopted an official statement before 2000. Seven of the 9 societies published the statement in their national journal, one in the French journal Réanimation ; all 9 have put it on their website where it is available to the public. All statements recognize the necessity of limitation of LST when the clinical situation is hopeless; state that cost or medical convenience are not decision criteria ; state that there is no ethical difference between withholding and withdrawing life-sustaining treatment and that a treatment may be withheld if it appears inadvisable or withdrawn if it has proven ineffective. Recommendations all underline the importance of the decision-making process ; describe different settings ; advocate to take into account the patient’s previous and present wishes, if patient is competent (advanced decisions are accepted in Great Britain and Switzerland) ; stress the need to keep the family informed and to take into account its opinion (a surrogate is legal in Scotland and France) ; specify the decision is never that of the family ; advocate a medical staff discussion, involving the nurses (except for one ICS), to reach a general consensus ; specify that the decision is always the doctor’s one ; strongly advise to record all decisions in the patient’s file ; promote a palliative care strategy. All societies have taken a clear position against euthanasia (except the Spanish) and do not consider limitation of LST as such. CONCLUSION. only 56% of ICS from 16 European countries have adopted an offical statement on EOL care. Recommendations are all similar and advocate a palliative care strategy when limitation of LST is decided.

757

759

PHYSICIANS’ AND NURSES’ ATTITUDES TOWARDS LIFE-SUSTAINING TREATMENT FOR THEIR PATIENTS

AUDIT OF FAMILY WITNESSED RESUSCIATION IN UK EMERGENCY DEPARTMENTS

Svantesson M1, Sjokvist P2 1 Department of Anaesthesia and Intensive Care, Orebro University Hospital, Orebro, 2Department of Anaesthesia and Intensive Care, Huddinge University Hospital, Stockholm, Sweden

1

INTRODUCTION. There exist several postal surveys about physicians’ and nurses’ attitudes towards foregoing of life-sustaining treatment in hypothetical cases. What health-care professionnals view as the ideal and what actually happens in clinical reality may not be in concordance. The objective of this study was to describe physicians’ and nurses’ attitudes towards determining factors for real clinical decisions of life-sustaining treatment. METHODS. We conducted 97 structured interviews with physicians and 132 interviews with nurses regarding 806 patients at 7 medical/surgical units at a university hospital. They were asked if they thought each patient should receive full or limited life-support (or being uncertain) and were also asked to give a motivation for their choice. Both quantitative and qualitative analysis (content analysis) was performed. RESULTS. Three perspectives emerged that motivated the choice of level of life support: medical-, quality of life- and autonomy perspective. For the majority of patients (84%), both the physicians and the nurses had medical reasons. The nurses also used the quality of life perspective for 40 % of the patients and the physicians in 27%. As for the autonomy perspective, the physicians related to knowledge of the patient’s wishes in 2% and the nurses in 4%. Important determining medical factors were age (27%); prognosis (34%) and medical condition 47%. The quality of life factor most frequently mentioned was functional ability (20%). Other also mentioned were comfort, social and existential factors. Regarding 14 % of all the patients the physicians and nurses disagreed about the level of life-support. For patients over the age of 80 there were disagreements for 27% of the patients. CONCLUSION. In contrast with previous results from postal surveys, the physicians and nurses interviewed in this study of real cases only rarely spontaneously mentioned patient wishes as important. Instead the medical perspective dominated. As for the quality of life perspective, functional ability a determinant that seems to be more easily assessed was more often mentioned than other more subjective factors.

Booth M G1, Woolrich L2, Kinsella J3 Department of Anaesthesia and Intensive Care, 2Accident and Emergency, Royal Infirmary, 3Department of Anaesthesia and Intensive Care, University of Glasgow, Glasgow, United Kingdom INTRODUCTION. The America Heart Association guidelines from 2000 recommend that family be allowed to witness cardiopulmonary resuscitation (FWR)(1). There is controversy whether the family benefits and opponents fear litigation and family interference during FWR. The extent of FWR in UK emergency departments (ED)is unknown. METHODS. A telephone survey of a selection of UK EDs was performed asking about experience with FWR. RESULTS. 162 UK EDs with an average attendance of 47000 patients per year participated. FWR was allowed by 128(79%)for an adult patient(93% if a child).Of these 50%invited relatives to witness. Only 21% did not permit FWR. The perceived benefits are; accepting that all possible has been done (48%), accepting the death (48%) and help with grieving (38%). 2% did not think FWR was of help. Few had had any problems such as the family interfering during FWR. Never being asked was the commonest reason not allowing FWR followed by staff reluctance. Most respondents would wish to be present if their child (85%0, Spouse/partner (64%0 or elderly relative (52%) was being resuscitated. CONCLUSION. FWR is common in UK EDs. FWR is more common when children are being resuscitated than adults. Further research is needed to demonstrate whether it is of benefit to the relatives. REFERENCE(S). 1.American Heart Association. Guidelines 2000 for Cardiopulmonary Resuscitation: part2. Ethical aspects of CPR and ECC. Circulation 2000; 102 (suppl):I12-I21. 2.McClenathan BM, Torrington KG and Uyehara CFT. Family member presence during cardiopulmonary resuscitaion. Chest 2002; 122:2204-11.

S198

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

Oral Presentations TBI: What else should we monitor to predict outcome? – 760-765 760 ARTERIAL COMPLIANCE IN ACUTE BRAIN INJURED PATIENTS. Isnardi D1, Andrews P J D2 1 Intensive Care Unit, 2Intensive Care, University of Edinburgh, Edinburgh, United Kingdom INTRODUCTION. Evidence suggests that acute brain injury may lead to sympathetic hyperactivity resulting in vasoconstriction. Arterial compliance (C) varies with changes in vasomotor tone and age. Changes in C were measured with the non-invasive monitor Wideband External Pulse (WEP) to monitor vasomotor tone in acute brain injured patients. METHODS. 37 patients (14 sub-arachnoid hemorrhage, 15 traumatic brain injury, 8 stroke), aged between 19 and 76 (group I 19-40(n=8), group II 41-55 (n=16), group III 56-76 years (n=13)), were studied prospectively for 9 days from the injury. A post resuscitation GCS was collected in all patients; they were assessed clinically and with ICP when available (17 patients). Inotropic support was used when indicated (24 patients). 6 patients were diagnosed brain dead. Compliance measurements (ml/mmHg) were made by suprasystolic signals provided from a piezoelectrical sensor placed beneath the distal edge of a blood pressure cuff over the brachial artery. C with the WEP device is determined by analysis of the ratio of the incident to reflected waves. C measurements were repeated on consecutive days. Means of each group were compared with the Student’s t-test. RESULTS. The variation of C over time in 13 patients, having more than 3 days of measurement, is: day 1 C=1.93±0.72, day 2 C=1.64±0.46, day 3 C=1.75±0.55, day 4 C=1.40±0.16, day 5 C=1.17±0.29, day 6 C=1.74±0.66, day 7 C=1.76±0.63, day 8 C=1.82±0.53, day 9 C=1.39±0.21. C is reduced on days 4 and 5 after acute brain injury and variance also decreased for this measurement. C is significantly lower in patients aged more than 45 years: group I C=2.05±0.45, group II C=1.70±0.49 (P=0.009), group III C=1.67±0.51. Differences in initial GCS, level of intracranial pressure and dose of inotropic support do not cause significant changes in C. In brain death patients C=2.11±0.44. CONCLUSION. C decreases over 45 years in brain injured patients, like in the normal population. C decreases in the fourth and fifth days after brain injury, coinciding with increased incidence of vasospasm and may reflect hypotension in other critical illnesses that cause vasoparesis. Initial GCS, level of ICP and inotropic support do not influence C, probably because of the synchronized action of different factors such as temperature, sepsis or use of vasoactive drugs. Brain death causes loss of vasomotor tone and consequently increases C in all ages.

762 TRANSCRANIAL DOPPLER ULTRASONOGRAPHY AT THE ADMISSION OF MILD HEAD-INJURED PATIENTS Jaffres P1, Brun J1, Declety P1, Anglade D1, Fauvage B1, Kaddour A2, Payen J1, Jacquot C1 and Intensive Care, 2Emergency Unit, Michallon’s Hospital, Grenoble, France

1Anesthesia

INTRODUCTION. Mild-head injured (MHI) patients might be at risk of secondary neurological aggravation in relation with brain ischemia (Lobato,J Neurosurg 1991). But, accurate selection of high-risk MHI patients at their admission is still lacking.Brain haemodynamics can be noninvasively assessed using transcranial Doppler ultrasonography (TCD), we prospectively investigated the interest of TCD at the admission of such patients. Initial TCD measurements were compared to the neurological status at 7 days. METHODS. Thirty nine MHI patients (Glasgow Coma scale, GCS > 8) were included in the study, in the absence of systemic hypotension. All patients were included into class II or III, according to the Masters’ classification (Masters,J Med. 1987). TCD right and left middle cerebral artery systolic (SV), diastolic (DV), and mean (MV) flow velocities and pulsatility index (PI) were measured within the first 6 hours after the trauma. Neurological status was studied up to day 7. Patients were separated into two groups according to their neurological status 7 days following the trauma: non-worsening (group 1) and worsening (group 2). Neurological worsening was defined as a decrease of GCS score by > 1 point. Statistical analysis was performed by Chi2 test and by ANOVA. RESULTS. Eight patients had neurological worsening at 7 days (group 2). * p < 0.05 Age (yrs) Sex (f/m) SAPSII GCS admission Masters II/III (n) Delay DTC/Trauma(min) Right/Left IP

Gr.1 (n = 31) 34±16 10/21 16±3 13±2 18/13 205±27 1.05±0.05/1.1±0.3

Gr.2 (n = 8) 33±15 2/6 23±5 14±1 4/4 266±77 1.35±0.2 */1.54±0.5 *

CONCLUSION. Noninvasive measurements of brain haemodynamics using TCD at the admission of MHI patients can be related to their neurological status at 7 days. This approach should be useful in selecting high-risk MHI patients.

761

763

POST-TRAUMATIC CEREBRAL EDEMA AND BLOOD-BRAIN BARRIER MODIFICATIONS Ter Minassian A1, Pasco A2, Darabi D2, Tanguy J2, Lemaire L3, Le Jeune J4, Caron C2, Beydon L1 1 Anesthésie-réanimation, 2Radiologie, CHU, 3Faculté des sciences, Université, 4Médecine nucléaire, CHU, Angers, France

BLUNTING THE EFFECTS OF ETS IN HEAD TRAUMA PATIENTS: A DOSERESPONSE STUDY WITH REMIFENTANIL

INTRODUCTION. Intra and extracellular cerebral oedema are observed in severely head injured patients. The latest is assumed to be due to the alteration of the blood-brain barrier (BBB). In case of post-traumatic alteration of the BBB, elevated cerebral perfusion pressure could increase vasogenic extra cellular oedema. The aim of this NMR study was to characterize post-traumatic cerebral oedema by the measurement of the apparent diffusion coefficient of water (ADC) and to assess BBB permeability to DOTA-Gd.

INTRODUCTION. Nociceptive stimulation, such as endotracheal suction (ETS), may induce an increase in intracranial pressure in head trauma patients, which can be deleterious. The first endpoint of the study was to blunt the cough reflex during endotracheal suction (ETS) in more than 50% of patients with severe head trauma by using three incremental doses of remifentanil. The second endpoint was to assess the effects of remifentanil on intracranial pressure (ICP), cerebral blood flow estimated by transcranial Doppler sonography, and bispectral index (BIS).

METHODS. After agreement of the local ethic committee and informed consent of the next of kins, 10 SHI were prospectively studied. (GCS ≤ 8, 9 male, age : 28 ± 15 years) Inbetween 12 hours and day20 post-trauma, three NMR were performed in each patient respecting at least 4 days between each examination. RESULTS. Abnormal ADC areas were distributed as follows : decreased ADC in 4 patients (preponderant intracellular oedema : ADC < 0,70 x10-3 mm2/s) and both decreased and increased ADC in 6 (preponderant extra-cellular oedema ADC > 0,82 x10-3 mm2/s). Enhancement after DOTA-Gd injection was never observed before day 10 post-trauma. After day 10 massively increased permeability of BBB in areas of previously decreased ADC was observed in all patients. CONCLUSION. Severe head injury does not grossly affect BBB impermeability to DOTA-Gd between 12 hours and 9 days post-trauma whatever the type of cerebral oedema.

Albanese J1, Leone M1, Viviand X1, Martin C1 1 Anesthesiology and ICU, CHU Nord, Marseilles, France

METHODS. After gaining control of ICP (ICP <25 mmHg), hemodynamic values, and blood gas pressures (PaCO2 between 33 and 37 mmHg), 20 consecutive patients, each under tracheal intubation, received three successive doses of remifentanil as follows: first dose: 1 µg/kg over 1 min followed by continuous infusion of 0.25 µg/kg/min; second dose: 2 µ/kg over 1 min followed by continuous infusion of 0.5 µg/kg/min; and third dose: 4 µg/kg over 1 min followed by continuous infusion of 1 µg/kg/min. Cough reflex was assessed by performing endotracheal suction 20 min after the beginning of infusion. Heart rate, ICP, mean arterial pressure (MAP), cerebral perfusion pressure (CPP), middle cerebral artery mean flow velocity (VMCA), and BIS index were monitored at 1-min intervals throughout the 30-min study period. Patients received intravenous ephedrine or norepinephrine to maintain CPP >70 mmHg. RESULTS. Bolus infusion involved a significant increase in ICP in patients receiving the second and third doses when compared to the first dose. Remifentanil made it possible to blunt the cough reflex in more than 50% of the patients but only in those receiving the third dose. Patients receiving first, second, and third doses required an administration of vasopressor in 12, 18 and 20 cases, respectively, to maintain an adequate CPP. CONCLUSION. The results show that only the highest dose of remifentanil made it possible to blunt the cough reflex linked to endotracheal suction. This high-dose of remifentanil was followed by an increase in ICP and the need for vasopressor to maintain an adequate mean arterial pressure.

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

764 MECHANISM OF CHANGE IN INTRACRANIAL PRESSURE WITH HEAD ELEVATION IN PATIENTS WITH BRAIN INJURY. MELOT C1, RAPIN C1, BONNIER F1, MORAINE J2, BERRE J1 1Department of Intensive Care, Erasme University Hospital, 2High Institute for Physical Training and Physical Therapy, Free University of Brussels, Brussels, Belgium INTRODUCTION. The mechanism of the change in intracranial pressure (ICP) with head elevation in patients with acute brain injury is not well understood. The decrease in ICP with moderate head elevation is explained by the hydrostatic displacement of cerebrospinal fluid (CSF) from the cranium to the perimedullary space. However, an elevation higher than 30O from the horizontal plane leads usually to an increase in ICP for which the mechanism remains unclear. We tested the hypothesis that, with head elevation above 30O, ICP increased in parallel with the rise in intra-abdominal pressure (IAP) with a shift of CSF cranially. METHODS. Seven patients (2 M / 5 F; age 56 ± 5, mean ± SE) in coma due to brain injury (Glasgow 6 ± 1) were studied. All patients had an intraventricular (ICP)and a central venous catheters placed for routine care. IAP was estimated by measuring the intra-vesical pressure. The patients were randomly assigned to the following positions: 0, 15, 30, 45, and 60O of elevation of the head. RESULTS. Between 0 and 60O, IAP increased from 7 ± 1 to 17 ± 2 mmHg (p < 0.001) and central venous pressure did not change from 8 ± 1 to 8 ± 1 mmHg (p = NS). ICP decreased between 0 and 30O from 16 ± 1 to 13 ± 1 mmHg (p = 0.08) and increased between 30 and 60O from 13 ± 1 to 15 ± 2 mmHg (p = 0.12).

CONCLUSION. With head elevation from 0 to 30O, ICP decreased probably due to a shift of CSF from the cranium to the lumbar sac. With further head elevation from 30 to 60O, ICP increased in parallel with the rise in IAP suggesting a redistribution of CSF from the perimedullary space to the cranium.

765 CORRELATION BETWEEN ANEMIA AND OUTCOME IN SEVERE TBI Badr A E1, Esposito D P1, Golanov E1, Quin D1, Tullis J1 1 Neurosurgery, University of Mississippi Medical Center, Jackson, United States INTRODUCTION. More than half of brain-injured patients sustained anemia, hypoxemia, and hypotension (6). Sixty percent of head-injured patients have evidence of ischemic brain damage at autopsy (1, 2). In the TBI CBF decreases from 50 ml to 30 ml/ 100 g/ min within the first 8 hours and continues to decrease to less than 20 ml/ 100 g/ min in the worst injured patients (3, 4, 5). Maintaining of normal hematocrit may decrease ischemia following head injury by improvement of cerebral oxygen delivery. METHODS. A retrospective chart review was done on 71 patients from 1999-2001 with severe TBI (GCS<=8),ages ranged for 18-75. The outcome FIM Scores from admission to discharge were compared with relation to GCS, age of the patient and patient’s haematocrit (HCT) lowest level, Hct level at admission and Hct level at discharge from the intensive care. Pearson bivariate correlation coefficients with 2 tailed significance was used. p <0.05 was significant RESULTS. 50 M and 21 F. GCS ranged from 3-10 (5.8±1.9). Overall, there was clear tendency for correlation between higher Hct and FIM score improvement. The strength of correlation dependent upon age and sex and was strongest in males in 45-65 age group (0.687, p=0.041). In younger males (<26 yo) higher Hct at admission was associated with higher GCS (0.517, p=0.024). CONCLUSION. Results indicate that degree of anemia may be a significant factor that affects brain condition after traumatic brain injury. REFERENCE(S). 1.Graham DI, Ford I, et al: Ischaemic brain damage is still common in fatal non-missile head injury. J Neurol Neurosurg Psychiatry 52:346-350, 1989. 3.Fieschi C, Battistini N, et al: Regional cerebral blood flow and intraventricular pressure in acute head injuries. J Neurol Neurosurg Psychiatry 37:1378-1388, 1974 4.Jaggi JL, Obrist WD, et al: Relationship of early cerebral blood flow and metabolism to outcome in acute head injury. J Neurosurg 72:176-182, 1990 5.Obrist WD, Langfitt TW, et al: Cerebral blood flow and metabolism in comatose patients with acute head injury. J Neurosurg 61:241-253, 1984 6.Miller JD, Sweet RC, et al: Early insults to the injured brain. JAMA 240:439-442, 1978

S199