14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

S 280

Oral Presentations Acute lung injury (III) ± 564±569

566

564

Akinci IO, Atalan HK, Yilmazbayhan D, Tugrul S, Ergin Ozcan P, Esen F, Telci L, Akpir K, Cakar N. Anesthesiology and Intensive Care, University of Istanbul, Medical Faculty of Istanbul, Istanbul, Turkey

EFFECT OF MECHANICAL VENTILATION STRATEGIES INCLUDING RECRUITMENT MANEUVER ON VENTILATOR-INDUCED LUNG INJURY

IMPACT OF DIFFERENT LUNG RECRUITMENT MANEUVERS ON OXYGENATION AS WELL AS CENTRAL AND REGIONAL CIRCULATION: EFFECT OF VOLUME EXPANSION Odenstedt H, Šneman A, Kµrason S, Lindgren S, Stenqvist O, Lundin S. Anaesthesia and Intensive Care, Sahlgrenska University Hospital, Gothenburg, Sweden INTRODUCTION. Lung recruitment is an important part of the open lung concept but may have hazardous effects. This study examines the acute effects of different recruitment manoeuvres on oxygenation, circulation and thus oxygen delivery (DO2) as well as the impact of volume expansion (VE). METHODS. ALI was induced in ten anaesthetised pigs by repeated bronchoalveolar lavage. Descending aortic blood flow (ABF), portal and renal blood flow as well as PaO2 and SvO2 were monitored continuously. Vital capacity manoeuvres (VC) using 30 or 40 cm H2O constant airway pressure for 1 min or pressure control (PC) ventilation using respiratory rate 40, I:E 1:1, pressure 30/PEEP 15 or pressure 40/ PEEP 20 for 1 min, were evaluated in random order before and after volume expansion with Dextran solution 8 ml/kg. RESULTS. Table shows ABF, PaO2 and DO2 at baseline and at 1 minute of recruitment before and after volume expansion. The oxygenation response to recruitment manoeuvres was more rapid and markedly enhanced at higher airway pressures (40 vs 30 cm H2O, p < 0,05) irrespective of VC or PC. In contrast circulatory depression with a more marked decrease occurred at these pressures (p < 0,05, vs 30 cm H2O) and could only partly be counteracted by volume expansion. The haemodynamic response to recruitment manoeuvres was different in different vascular beds, with a more marked decrease in mesenteric vs renal blood flow. Volume expansion could completely restitute renal but not mesenteric flow and oxygen delivery during recruitment. These changes could not be assessed from global circulatory parameters such as ABF.

ABF

Time

VC 30

VC 40

PC 30/15

PC 40/20

Baseline

3.60.8 2.40.9/ 3.21.0 6.71.7 1414/118 34070 22060/ 300120

3.60.5 1.50.9/ 2.10.5 6.72.2 3026/4819 33090 16060/ 26070

3.60.5 2.50.5/ 3.51.4 6.92.1 1411/1512 33090 27050/ 35090

3.90.5 1.60.4/ 2.50.7 6.91.9 4924/5721 370100 21050/ 30090

(l/min)

1 min/1 minVE

PaO2 (kPa) DO2

Baseline 1 min/1 minVE Baseline

(ml/min)

1 min/1 minVE

INTRODUCTION. Ventilator-induced lung injury (VILI) is one of the most important complications of mechanical ventilation and directly associated with high ventilatory pressures (1). In this study we aimed to determine the effect of recruitment maneuver (RM) on VILI. METHODS. 24 Sprague-Dowley rats weighting between 280±320 were anesthetized with ethrane 4±5 % and intraperitoneal ketamine administered (50±70 mg/kg). The rats were tracheostomized and 24-G cannula inserted into carotid artery. Rats ventilated with Servo 900C ventilator on pressure control ventilation mode (FiO2: 1.0, frequency: 30 /min, I/E: 1/2). Then rats were allocated into the following study groups randomly: PEEP Group = Peak pressure (PP): 30 cmH2O PEEP: 10 cmH2O, LPV Group = PP: 14 cmH2O, PEEP: 0 cmH2O, HPV Group = PP: 45 cmH2O, PEEP: 0 cmH2O, RM Group = PP: 30 cmH2O, PEEP: 5 cmH2O, and RM (sustained inflation of 45 cmH2O CPAP for 20 seconds). Blood gas samples were taken before randomization and after 1-hour ventilation prior sacrifice. Lungs extripated unblock with heart and fixed with 10 % formalin. Coronal sections taken from lung for pathologic evaluation and examined as presence (1)(pres) or absence (0)(abs) of nine different lesions (Table 1). Results summed for each group and qui square test used for statistical analysis. Kruskall-Wallis ANOVA test was used for blood gas parameters. RESULTS. Oxygenation parameters between the groups were similar throughout the study. Pathologically more damage was observed in HPV and RM groups compared to LPV and PEEP groups.

Microscopic atelectasis Perivascular heamorrhage Alveolar heamorrhage Alveolar oedema Alveolar PNL Hyaline memb. Perivas.int.oedema Congestion Interstitial PNL Total

PEEP group Pres-abs

LPV group pres-abs

HPV group pres-abs

RM group pres-abs

p

0±6 1±5 0±6 1±5 2±4 0±6 2±4 4±2 1±5 10±44

1±5 0±6 0±6 1±5 3±3 0±6 2±4 1±5 1±5 8±46

1±5 3±3 5±1 4±2 3±3 5±1 6±0 3±3 0±6 30±24

1±5 0±6 4±2 6±0 5±1 0±6 5±1 0±6 2±4 23±31

0.76 0.065 0.002 0.003 0.36 0.0003 0.963 0.957 0.998 < 0.0001

Values are mean and SD (n = 10). CONCLUSION. There is a price to be paid for rapid recruitment of collapsed lungs in ALI with an acute impairment of central and regional blood flows as well as decreased oxygen delivery. The haemodynamic response to recruitment manoeuvres was not uniform in all vascular beds.

CONCLUSION. In this model, high pressures used for RM caused significant injury compared to LPV, and in addition to low airway pressure, PEEP is considered to be an important factor for protection of lung injury. REFERENCE. Dreyfuss D, Saumon G (1998) Ventilator-induced lung injury. Am J Respir Crit Care Med 157: 294±323

565

567

RECRUITMENT MANEUVER: DOES IT PROMOTE BACTERIAL TRANSLOCATION?

EFFECTS OF SPONTANEOUS BREATHING WITH AIRWAY PRESSURE RELEASE VENTILATION ON INTESTINAL BLOOD FLOW IN EXPERIMENTAL LUNG INJURY

Cakar N1, Akinci O1, Tugrul S1, Ergin Ozcan P1, Esen F1, Eraksoy H2, Cagatay A2, Telci L1, Nahum A3. 1Anesthesiology & Intensive Care, 2Infectious Diseases, University of Istanbul, Medical Faculty of Istanbul, Istanbul, Turkey, 3Pulmonary and Critical Care, Regions Hospital, St Paul Minesota, USA INTRODUCTION. High peak airway opening pressures are used during recruitment maneuvers (RM) to open collapsed lung units (1). High peak pressures, however, can cause lung injury as evidenced by translocation of the intratracheally inoculated bacteria (2). In this study we explored if recruitment maneuvers with high pressures cause translocation of the intratracheally inoculated Pseudomonas aureginosa. METHODS. Eighteen male Sprague Dawley rats were anesthetized, tracheostomized were ventilated with 14 cmH2O peak pressure and 0 cmH2O PEEP in pressure-controlled ventilation (frequency 30 bpm; I:E = 1:2; FiO2: 1). Intratracheal inoculation of 500 microliter of saline containing 1 x 105 P.aeruginosa was performed prior to randomization into three groups (n = 6 in each); a low pressure group (14 cmH2O peak pressure, 0 cmH2O PEEP) a high pressure group (45 cmH2O peak pressure, 0 cmH2O PEEP) and a RM group (14 cmH2O peak pressure, 0 cmH2O PEEP and RM (sustained inflation of 45 cmH2O CPAP for 30 seconds) every 15 minutes). Blood samples for blood gas analysis were obtained prior to intratracheal instillation of bacteria and at the end of the experimental protocol. Blood cultures were obtained before and after bacterial instillation at 30 minute intervals during the experiment. Blood samples were cultured directly in blood-sheep and MacConkey agar and observed on the second day. Bacteremia was defined as the presence of one or more colonies of P.aeruginosa in 1 ml blood. The isolated strains were identified by standard microbiological methods. RESULTS. The blood cultures were only positive for P.aeruginosa in 6 rats in the high pressure group and remained negative throughout the study period in the low pressure and RM groups. Oxygenation deteriorated in all groups after intratracheal instillation of bacteria. In the high pressure group it decreased from 417  67 to 79  20 mmHg (p = 0.004), where as in the low pressure and RM groups PaO2 decreased from 410  98 and 383  78 mmHg to 288  105 mmHg (p = 0.031) and 249  59 mmHg (p = 0.11) respectively. CONCLUSION. RM applied as sustained inflation does not cause translocation of intratracheally inoculated bacteria and RM superimposed on low pressure ventilation prevented development of hypoxia. REFERENCES. 1. Amato MBP, Barbas CSV, Medeiros CM, et al. Effect of protective ventilation strategy on mortality in the acute respiratory distress syndrome. N Enj J Med 1998; 338: 347±354. 2. Verbrugge SJC, Sorm V, van't Veen A, Monton JW, Gommers D, Lachmann B. Lungoverinflation without positive end-expiratory pressure promotes bacteremia after experimental Klebsiella pneumonia inoculation. Int Care Med 1998 24: 172±177

Hering R1, Viehöfer A1, Zinserling J1, Thai MK1, Stetter J1, Wrigge H1, Minor T2, Putensen C1. 1 Dept. of Anaesthesiology and Intensive Care Medicine, 2Dept. of Experimental Surgery, University of Bonn, Bonn, Germany INTRODUCTION. Spontaneous breathing with airway pressure release ventilation (APRV) has been shown to improve gas exchange, cardiac output and renal perfusion and function in patients with acute lung injury (1, 2). This study was designed to investigate in an animal model if spontaneous breathing with APRV provides better intestinal blood flow than conventional pressure-limited ventilation. METHODS. After approval by the Animal Care Committee of the university of Bonn juvenile domestic pigs received in random order APRV with spontaneous breathing (APRV + SB) and APRV without spontaneous breathing maintaining equal minute ventilation (APRV-SB). Lung injury was induced with oleic acid. Apnoe was induced with succinylcholine. Cardiac output was estimated by the transpulmonary double indicator dilution method. Blood flow to the mucosal/submucosal and muscularis/serosal layers of the stomach, the small intestines and colon was measured with coloured microspheres. RESULTS. Preliminary results of 13 pigs (mean  SD) are presented. Cardiac index increased from 2.8  0.7 l/min/m2 during APRV-SB to 3.7  0.9 l/min/m2 during APRV + SB (p < 0.01). Intestinal blood flow data are presented in the tables.

APRV + SB APRV ± SB

Stomach

Jejunum

Ileum

Colon

0.340.24 0.230.12 p < 0.05

0.670.27 0.430.15 p < 0.05

0.620.40 0.350.23 p < 0.05

0.720.49 0.430.24 p < 0.05

Mucosal/submucosal blood flow (ml/g/min)

APRV + SB APRV ± SB

Stomach

Jejunum

Ileum

Colon

0.100.07 0.070.04 p < 0.05

0.110.08 0.080.06 NS

0.100.05 0.060.04 p < 0.05

0.070.07 0.040.04 p < 0.05

Muscularis/serosal blood flow (ml/g/min) CONCLUSION. In experimental lung injury spontaneous breathing with APRV has beneficial effects on systemic blood flow and intestinal mucosal/submucosal and muscularis/serosal perfusion compared to conventional pressure limited ventilation maintaining equal minute ventilation. REFERENCE. Am. J. Respir. Crit. Care Med. 1999. 159: 1241 2. Am. J. Respir. Crit. Care Med. 2000. 161: A549

14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

568 POSSIBLE ALTERNATIVES TO THE ULTRAFILTRATION COEFFICIENT FOR ASSESSING THE INTENSITY OF VENTILATOR-INDUCED LUNG INUJURY

Broccard AF1, Vannay C1, Feihl F2, Markert M3, Schaller MD1. 1Division of Intensive Care Medicine, 2Division of Physiopathology, 3Central Laboratory for Clinical Chemistry, University Hospital of Lausanne (CHUV), Lausanne, Switzerland INTRODUCTION. The ultrafiltration coefficient (Kf) has been used extensively to assess ventilator-induced lung injury (VILI) in ex-vivo isolated perfused lung models (1,2). The objective of the present study was to look for possible substitutes of Kf in in-vivo animal models in which the latter measurement is not feasible. METHODS. We analysed the data from a previous experiment in which different intensities of lung injury were induced by varying the peak alveolar pressure and the concentration of inhaled CO2. In brief, 21 isolated sets of normal rabbit lungs were perfused [constant flow 0.3 l/min, left atrial pressure (LAP) 6 mmHg], ventilated for 20 min [pressure controlled ventilation (PCV): 15 cmH2O] with an inspired CO2 fraction adjusted for perfusate PCO2 = 40 mmHg and then randomised into 3 groups (n = 7 each). Group 1 was perfused and ventilated as above during 3 consecutive 20 min periods. In Group 2, PCV was 20, 25 and 30 cmH2O during each period of ventilation and targeted PCO2 similar to group 1. Group 3 was ventilated as Group B but targeted PCO2 was 80±100 mmHg. Regression analysis between the changes (from start to end of study) in the ultrafiltration coefficient (DKf: permeability index), the concentrations of protein and hemoglobin in the broncho-alveolar lavage (BAL) and a developed lung injury score. The latter was computed as the product of BAL protein concentration (tracking altered permeability to protein), hemoglobin lung concentration normalized to lung protein concentration measured in the supernatant of the homogenized lung samples (tracking lung hemorrhage) and the wet weight/dry weight ratio (tracking oedema) of the corresponding tissue sample. RESULTS. Regression analysis BAL protein DKf

p < 0.001

BAL Hemoglobin R 0.91

p < 0.001

Injury Score R 0.69

p < 0.001

R 0.94

CONCLUSION. BAL protein concentration and the score of injury correlated tightly with Kf alteration. Both indices could be used as alternatives to Kf measurement to assess VILI. REFERENCES. 1. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998;157: 294±323. 2. Parker JC, Hernandez LA, Peevy KJ. Mechanisms of ventilator-induced lung injury. Crit Care Med 1993;21: 131±143.

S 281

Oral Presentations Outcome of mechanical ventilation ± 570±575 570 HOSPITAL OUTCOMES AND EARLY PROGNOSTIC FACTORS IN CRITICAL CARE PATIENTS UNDERGOING MECHANICAL VENTILATION: A PROSPECTIVE STUDY Revuelta P, Naranjo C, JimØnez J J, Prieto F, Mora M L. Intensive Care Unit, Hospital Universitario de Canarias, La Laguna, Spain INTRODUCTION. Critical care patients requiring mechanical ventilation (MV)have a high mortality and associated costs. Prevalence of MV and mortality vary, depending on case-mixed and centers in concern.(1) The objectives of this study were to describe outcomes of patients undergoing MV and to identify early prognostic variables associated. METHODS. This prospective study was located in a 20-bed ICU of a university referral hospital. Of 591 consecutive patients admitted to the ICU during a 10-month follow up period, we enrolled 205 patients (35 %) who received MV for at least 12 hours. Predictive variables were age, sex, comorbidities, functional status, nutrition assessment, hospital days before ICU, admission category, SAPS II, APACHE II, organ dysfunction index (ODIN) and the indication to initiate MV. Primary outcomes were ventilation time, liberation from MV, ICU and hospital lengths of stay (LOS), and ICU and hospital survivals. The incremental impact of prognostic variables on outcomes were tested with univariate tests, survival analyses and multivariable predictive models. A logistic regression model was applied to identify independent risk factors of hospital mortality. The multivariable Cox proportional hazards model was applied to identify independent predictors of the probability over time of being liberated from MV. RESULTS. ICU survival rate was 75 % and hospital survival rate was 66 %. Median duration of MV was longer in nonsurvivors with 6 days versus 5 days in survivors (log-Rank, p < 0.00005). Median weaning duration was 2 days, 62 % of the total duration of MV. Of the 153 ICU survivors, 150 (98 %) were completely liberated from MV. Median ICU and hospital LOS were 9 days and 30 days respectively. Hospital LOS was longer in survivors, with a median value of 35 days, whereas hospital LOS in nonsurvivors was 18 days (Mann-Whitney, p < 0.00005). Chronic renal disease (adjusted odds ratio aO = 3.15; 95 % CI = 1.07 to 9.26), one unit increase in SAPS II (aOR = 1.05; 95 % CI = 1.02 to 1.08), one aditional day of hospitalization before ICU admission (aOR = 1.03; 95 % CI = 1.01 to 1.06) and cardiac arrest at arrival (aOR = 9.06; 95 % CI = 2.22 to 36.93) were independently associated with an increase in hospital mortality. One unit increase in APACHE II (adjusted hazard ratio aHR = 0.93; 95 % CI = 0.91 to 0.96) was independently associated with a reduced liberation rate from MV. Postoperative respiratory failure (aHR = 1.22; 95 % CI = 1.17 to 2.52) and chronic obstructive lung disease exacerbation (aHR = 2.46; 95 % CI = 1.05 to 5.76), were independently associated with an increased liberation rate from MV. CONCLUSION. Patients undergoing MV more than 12 hours represent an important subset of the whole population admitted to our ICU (35 %) and have a substancial short-term mortality (34 %). Severity scoring indexes and some of the indications for instituting MV are good predictors of both the rate of liberation from MV and of short-term survival. Previous hospitalization time and the presence of chronic renal failure are associated with a decrease in hospital survival rate. REFERENCE. Stauffer JL, Fayter NA, Graves B, et al. Survival following mechanical ventilation for acute respiratory failure in adult men. Chest 1993; 104: 1222±1229.

569

571

OXYGEN TENSION OF PERFLUOROCARBON DURING PARTIAL LIQUID VENTILATION: MEASUREMENT OF DISTRIBUTION BY MRI

PROGNOSTIC VALUE OF CARDIAC TROPONIN I IN PATIENT HOSPITALIZED FOR ACUTELY EXACERBATED CHRONIC OBSTRUCTIVE PULMONARY DISEASE

Scholz A 1, Heussel CP 2, Schmittner M 3, Buerger K 1, Laukemper-Ostendorf S 2, Schreiber W 2 , Kauczor HU 2, Quintel M 3, Weiler N 1. 1Department of Anaesthesiology, 2Department of Radiology, Johannes Gutenberg University, Mainz, 3Department of Anaesthesia and Intensive Care, University Hospitals, Mannheim, Germany INTRODUCTION. Partial liquid ventilation (PLV) improves oxygenation in ARDS, but the physiological mechanism remains unclear. Therefore, we analysed the distribution of alveolar oxygen tension (pO2) in the lung during PLV by 19F-MRI [1]. METHODS. Three anaesthetized pigs (16±20 kg weight) received tracheal instillation of 20 ml / kg perflubron (perfluoroctylbromide). The animals underwent pressure-controlled ventilation in supine position with various inspiratory oxygen fractions (40 % to 100 %). After stabilisation of ventilation, 19F-MRI measurements (snapshot flash sequences) were performed during expiratory holds. As oxygen has paramagnetic properties influencing 19F-T1, the distribution of local alveolar perflubron pO2 can be calculated by a T1 fitting technique [1]. Three transversal planes (basis, mid, apex of the lung) were scanned. RESULTS. We found a ventrodorsal pO2 gradient in all 3 pigs and all 3 planes. The gradient increased with higher FiO2. FiO2 40 % Ventral

FiO2 40 % Dorsal

FiO2 100 % Ventral

Baillard C1, Boussarsar M2, Fosse JF1, Cracco C2, Le Toumelin P1, Jaber S2, Cohen Y1, Brochard L2. 1 Service de rØanimation, Hôpital Avicenne, Bobigny, 2Service de rØanimation, Hôpital H Mondor, CrØteil, France INTRODUCTION.: Heart dysfunction, an important determinant of the risk of death, is frequently associated in patients with acutely exacerbated COPD. This study aimed to evaluate the prognostic value of cardiac troponin I (cTnI) in this population. METHODS. This prospective study included 125 consecutive patients hospitalized in two ICU with acute respiratory failure (ARF). Patients were separated according to the presence or not of COPD. Blood samples for cTnI were obtained on admission and within 24 hours (stratus II Dade International) and was considered positive above 0,5 ng/ml. Student's test or Mann Whitney test were used for continuous variables and chi 2 test for discret variables. Multiple logistic regression was performed for prediction of death. Mean  SD or median (extreme). Probability value of < 0.05 was considered significant. RESULTS. A positive cTnI was detected in both groups (Table1)but was only discriminative of mortality in COPD patients (p = 0.0004 and 0.21 respectively). In COPD patients: (1) cTnI was positive in 11 (15 %) and negative in 60 patients. (2) They were no different in term of clinical presentation, cardiac comorbidity, primary diagnosis or ECG. In opposite, COPD patients with positive cTni had higher SAPS II, mortality and lenght of stay in ICU. Among all data recorded, in COPD patients, cTnI, SAPS II and mechanical ventilation were associated with mortality which remained independent factors in logistic regression analysis (Table2).

FiO2 100 % Dorsal

Base plane 20084 7720 592179 29438 Mid plane 21686 8116 557152 38280 Apex plane 15959 9835 472121 36791 Regional alveolar oxygen tension (mean  standard deviation) [mm Hg] during PLV CONCLUSION. During PLV, alveolar pO2 is inhomogeneously distributed. This can be explained by inhomogeneous contact of perfluorocarbon with inspired gases and by inhomogenities in perfusion and diffusion. REFERENCE. Laukemper-Ostendorf S., et al. (2000): 19F-MRT von Perflubron in der Lunge zur quantitativen Bestimmung des Sauerstoffpartialdruckes bei partieller Fluessigkeitsbeatmung. RoeFo 172: 192S

Age (years) SAPS II Lenght of stay (days) Mortality (n) cTnI > 0.5 ng/ml (n)

COPD n = 71

non-COPD n = 54

p

6811 4118 8 (1±90) 16 (22 %) 11 (15 %)

68.411 4115 6 (1±64) 10 (18 %) 6 (11 %)

ns ns 0.03 ns ns

Table 1: Comparison between COPD and non-COPD patients

cTnI > 0.5 ng/ml Mechanical ventilation SAPS II

Odds ratio

CI 95

p

7.3 5.9 1.06

1.2±46 1.3±27 1.01±1.1

0.03 0.02 0.01

Table 2: Predictors of mortility in COPD patients (n = 71) CONCLUSION. cTnI is a strong and independent predictor of ICU mortality in patients admitted for acutely exacerbated COPD. REFERENCE. Leonello Fuso, Raffaele Antonelle Incalzi, Riccardo Pistelli et al. Predicting Mortality of Patients Hospitalized for Acutely Exacerbated Chronic Obstructive Pulmonary Disease. Am J Med 98(3): 272±277, 1995.

14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

S 282

572

574

PROGNOSIS OF ALCOHOLIC PATIENTS WITH SEVERE COMMUNITYACQUIRED PNEUMONIA

ACUTE CARDIOGENIC PULMONARY OEDEMA: IN-HOSPITAL AND LONGTERM OUTCOME AFTER ICU ADMISSION

Cuny J, Chagnon J L, Leroy O, Coffinier C. Reanimation Polyvalente, C. H. Valenciennes, Valenciennes Cedex, France

Peake SL, Moran JL, Fox V, Lorchirachoonkul T. Intensive Care Unit, The Queen Elizabeth Hospital, Woodville, Australia

INTRODUCTION. Many alcoholic patients are admitted in intesive Care Unit (ICU). Mortality prognosis factors of these patients are poorly defined.

INTRODUCTION. Variables predicting the outcomes of patients admitted to the ICU with acute cardiogenic pulmonary oedema (APO) have not been fully characterised; in particular, the utility of severity of illness scores has been questioned [1]. METHODS. Review of prospective computerised data base (1995±1998), case notes and hospital information systems to identify and characterise consecutive patients admitted with APO, which was categorised as ischaemia-related or not. Variables predicting hospital and out-of-hospital were identified using logistic regression modelling (odds ratio; OR(SE)), Homer-Lemeshow (C) statistic and area under ROC curve (Az); long term survival was assessed by Kaplan-Meier and Cox model estimates. RESULTS. The cohort of 103 patients was of mean(SD) age 75.5(14.5) years and 55 % were male. Pre-ICU, mask CPAP was administered in 50 %; patient source was emergency service in 43 %, other wards 37 % and hospital-transfer 20 %. First day APACHE II and SAPS II scores were 24(8) and 43(16) respectively. APO was ischaemia related in 49 %. Patients were mechanically ventilated (84 %) for a median of 1.5 days (range 0.1±19); intravenous nitrates, heparin and inotropic agents were administered in 92 %, 61 % and 31 % respectively. Median ICU and hospital length of stay were 1.8(0.3±19) and 9(1±64) days. Mortality rates were 19 % (ICU) and 26 % (hospital); median (Kaplan-Meier) survival was 26 months (95 % CI, 7.4±44.6). Hospital outcome was effectively modelled using (i) SAPS II score (OR, 1.07(0.02)) and (ii) APO diagnosis (ischaemic vs non-ischaemic; OR, 2.61(1.38)); C = 0.66 and Az = 0.79 (95 % CI, 0.70±0.89). SAPS II score better predicted hospital outcome compared with APACHE II score (Az(SE), 0.77(0.05) vs 0.70(0.06); p = 0.04). Predictors (p < 0.05, likelihood-ratio) of long term survival were: SAPS II score, gender (male vs female), APO diagnosis, patient source (hospital-transferred vs not), and coronary artery bypass surgery (CABVG; yes vs no). Residual analysis and decile goodness-of-fit test [2] confirmed appropriateness of the Cox model. Subject heterogeneity (frailty), suggesting non-modelled covariates, was not detected (p = 0.94); timevarying covariate effects were not evident.

METHODS. From January to December 2000, alcoholic patients admitted for severe community-acquired pneumonia in 7 ICU of non teaching hospitals were enrolled in the study. Different parameters were recorded: age, SAPS II, sex, Child score, number fo organ failure, vasoactive drugs. RESULTS. 50 patients were studied (34 male, 16 female).

patients alive (26) patients deceased (24)

ALIVE

sex (34M/16F)

AGE

SAPS II

17 M / 9F 17M / 7 F

58.113.7 57.2  13

36.213 51.7  19

D0 to D2

D2 to D15

D 15 TO D30

D30 to D45

D45 to D60

52 %

42 %

40 %

36 %

32 %

CONCLUSION. Mortality rate of alcoholic patients with severe community-acquired pneumonia in ICU was 48 %, and 68 % at two months. In this study, mortality is associated with six specific patient characteristics:SAPS II, Child and pugh score, P. pneumoniae pneumonia, dialysis, vasoactive drugs, and more than 3 organ failure.

Hazard ratio SE

SAPS II

Gender

APO

Source

CABVG

1.04(0.01)

1.65(0.49)

2.04(0.65)

2.09(0.79)

2.57(1.09)

Cox model parameter estimates CONCLUSION. Severity of illness scores, in particular SAPS II, appropriately predict outcomes in APO. Demographic and patient cardiac-specific variables influence post-hospital survival. ICU treatment-covariates appeared not to be determinant. REFERENCES. 1. Fedullo AJ, et al CCM 1988;16: 1218±1221. 2. May S, Hosmer D. Lifetime Data Analysis 1998;4: 109±120

573

575

BREATHING THEIR LAST? CHARACTERISTICS AND OUTCOME OF COPD PATIENTS IN A GENERAL INTENSIVE CARE UNIT

WEANING FAILURE DOES NOT INCREASE MORTALITY IN MECHANICALLY VENTILATED PATIENTS

Deyermond RE1, Ferguson A1, Mcleod N1. 1Anaesthetics, Antrim Hospital, Antrim, N. Ireland

Bugedo G, Bruhn A, Apablaza F, Bernucci F, Hernandez G, Castillo L. Anesthesiology, Universidad Catolica de Chile, Santiago, Chile

INTRODUCTION. COPD is increasing in prevalence especially among women [1]. The cost of care for individuals with COPD is greater on a day to day basis than other ventilated ICU patients [2]. Patients with acute exacerbation present frequently for admission to intensive care and appear, on the surface, to have protracted stays with a poor outcome. The aim of our study is to ascertain the prevalence, duration of stay and outcome of patients with COPD in intensive care and to determine the appropriateness of intensive care for such patients. METHODS. Retrospective chart review of charts of all patients admitted to an Intensive Care Unit in a district general hospital over a 2-year period, who had a diagnosis of COPD at discharge. We excluded those transferred to other hospitals and those with untraceable charts. Data pertaining to the individuals past medical history, ward data and ICU data was recorded and analysed using appropriate statistical methods. RESULTS. N = 52. Total number of ICU days was 369, 27 % of the ICU days over the period studied. 6 month mortality was 62 %, 78 % died whilst in ICU. 67 % patients had a known history of COPD, 38 % coexisting documented cardiac disease, 9 % were on home oxygen and 75 % smoked. 61 % were referred to ICU with a presumptive diagnosis of chest infection, 23 % had had a cardiac or respiratory arrest. Chest X-rays were abnormal in 30 % of patients, white cell counts elevated in 45 % and pyrexia and positive sputum cultures were present in 25 % and 26 % respectively. Apache II scores were higher in the population that died, 28.1 as compared with 19.8. 84.6 % had invasive ventilation (mean duration 6.4 days) and 18 % had a tracheostomy (mean duration ventilation 14.3 days). Mortality was 68 % in those ventilated. Reduced mortality was predicted by lower Apache II Scores (p = 0.003) and lack of evidence of infection (p = 0.006). Trends toward increased mortality were found with increasing age (p = 0.19), cardiac disease (p = 0.142) and pre-admission cardiac arrest (p = 0.6). There was increased mortality in those ventilated and exsmokers (n = 6) had 100 % mortality. Mortality was unrelated to duration of stay. CONCLUSION. As six-month survival following admission to ICU with COPD is 38 % in this study, and as patients with no evidence of infection have a better prognosis, we conclude that Intensive Care is appropriate for patients with COPD. Larger studies are needed to look at all admission parameters as to their value as prognostic indicators at ward level. REFERENCES. 1. Soriano JB et al. Thorax 2000;55: 789±94. 2. Ely EW, Baker AM et al. Crit Care Med 2000; 28: 408±13

INTRODUCTION. Weaning period is critical in the evolution of patients with acute respiratory failure (ARF) and mechanical ventilation (MV). Weaning failure has been associated with increased morbidity and mortality. We evaluated the impact of weaning failure on mortality, and MV and ICU length of stay. METHODS. Patients who were admitted to our 8-bed surgical-ICU and stayed more than 24 hours on MV were prospectively evaluated from June 1999 to June 2000. Demographics, ARF etiology, APACHE II and gas exchange and mechanical parameters were assessed. Weaning failure was defined as reintubation within 48 hours after extubation. Weaning failure (WF) patients were compared with those who were successfully extubated (SE) and the total group (TG = SE + WF). Outcome measures were mortality, MV and ICU length of stay and MV free days. RESULTS. 155 patients required MV for more than 24 hours, of which 103 (66 %) were successfully extubated, 19 (12 %) had weaning failure, and 33 (21 %) died before weaning could be attempted. There were no differences in age, sex, APACHE II scores, or etiology between WF and other groups. However, WF patients had longer ICU and MV length of stay than TG (14  6.2 vs. 9.2  7.4 days, p = 0.005; and 9.1  4.8 vs. 5.5  5.6, p = 0.002, respectively) and SE patients (8.8  6.4, p = 0,003; and 4.6  4.3, p = 0.001). WF patients had also less MV free days than SE patients (15  11.2 vs. 25  5.5, p = 0.001). There was no difference in mortality between WF and GT patients (32 % vs 26 %, p = NS). However, SE patients had 2 % mortality, which was lower than WF and GT patients (p < 0.05). CONCLUSION. Weaning failure is associated with longer MV and ICU length of stay, but does not increase mortality compared with the total group of patients. Weaning failure patients had the same risk of death as patients being connected for the first time to mechanical ventilation. REFERENCE. Epstein SK, Ciubotaru RL, Wong JB. Effect of failed extubation on the outcome of mechanical ventilation. Chest 1997; 112: 186±192.

14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

Oral Presentations Costing the septic patient ± 576±581

S 283

578 COSTS OF SEVERE SEPSIS IN A MIXED ICU IN THE NETHERLANDS

Bakker J. Intensive care, Gelre lukas hospital, Apeldoorn, The Netherlands

576 RESOURCE UTILISATION BY SURVIVORS AND NON-SURVIVORS IN SEPSIS

Barron NM1, Edbrooke DL2. 1Anaesthesia, 2Critical Care Medicine, Royal Hallamshire Hospital, Sheffield, England INTRODUCTION. Patients suffering from sepsis on ITU are known to cost more to treat and involve longer stays than many other groups. The aim of this study was to investigate this and what strategies may be targeted to provide more cost-effective care of this subgroup of patients. METHODS. The medical records for all patients admitted to the adult general intensive care unit (ICU) at Sheffield's Royal Hallamshire Hospital between 1 st April 1996 and 31 st March 1999 were screened for sepsis (diagnosed at any point during their ICU admission) by two experienced consultant medical staff. Diagnosis was based on clinical and laboratory evidence. (1) This identified a subgroup of 156 patients. This subgroup was further examined. The group was divided into survivors and non-survivors from ICU. The length of stay and cost (both average and average daily) of these groups was determined. The non-survivors were subsequently divided into those that died despite aggressive therapy and those that had treatment withdrawn and subsequently died. RESULTS. There were 93 survivors and 63 non-survivors from ICU. The survivors had a mean length of stay of 11.47 days (SD = 12.39). The non-survivors had a mean length of stay of 7.77 days (SD = 9.36). The mean total cost of survivors was £9409 (SD = £12809) and non-survivors £7701 (SD = £9404). Mean cost per day was calculated as £741 (SD = £335) in the survivors and £1037 (SD = £374) in the non survivors. When breaking the non-survivors into groups consisting fo those that died despite aggressive therapy and those that had aggressive therapy withdrawn the following results were obtained. The length of stay for the deaths despite therapy was 3.06 days (SD = 3.89). Those that had treatment withdrawn and died had a mean length of stay of 9.26 days (SD = 10.12). The mean overall costs for the groups were £9094 (SD = £10273) for the treatment withdrawal group and £3351 (SD = £3278) for the group that died despite aggressive therapy. CONCLUSION. Patients on ICU suffering from sepsis form a subgroup who are both lengthy and expensive to treat. This is particularly evident with patients that end up having treatment withdrawn after a period of time. There is a place for investigating patients whose continuing treatment of sepsis is futile and investigating the timescale of events leading to their treatment withdrawal. REFERENCE. Edbrooke DL. Hibbert CL. Kinglsey JM. Smith S. Bright NM. Quinn JM: The patient-related costs of care for sepsis patients in a United Kingdom adult general intensive care unit. Critical Care Medicine. 27(9): 1760±7 1999

INTRODUCTION. The incidence of severe sepsis is increasing and is now an important cause of morbidity and mortality in intensive care [1]. The impact of severe sepsis on costs and workload has however not been well studied. METHODS. All patients admitted to the 10-bed mixed ICU between 1±1-`99 and 30±6-`00 were included. Severe sepsis on admission was defined as 3 or more SIRS criteria in the presence of a proven/suspected source of infection and dysfunction of one or more organ systems. Direct costs (all costs related to the treatment of the patient excluding costs of physiotherapy and charges of medical specialists) were calculated on a daily basis using relational databases and the actual costs of the items used. Nursing workload was measured by use of the TISS28 scoring system. All data are expressed as mean  SE. Mann-Whitney test was used to calculate statistical significance. Currency used: Euro's RESULTS. During the study period 848 patients (525 male) were admitted (245 following elective surgery, 140 emergency surgery, 463 medical patients). 100 patients were admitted with severe sepsis. The sources were: 52 % pulmonary, 31 % abdominal, 7 % urogenital, 10 % other. Differences between patients with and without severe sepsis are given in the table. A significant correlation was found between the total TISS28 score per patient and the total direct costs per patient (all patients R2 = 0.76, p < 0.001) for both sepsis and non-sepsis patients (R2 = 0.77 and R2 = 0.71 respectively, p < 0.001). Patients with severe sepsis (12 %) were responsible for 32 % of the nursing workload and accounted for 28 % of the ICU occupancy. Severe sepsis patients used 35 % of the total direct costs budget.

Age (yr) APACHE II Predicted Mortality (%) Length of ICU stay (days) ICU mortality (%) TISS-28 (total per admission) Direct costs (per admission) Direct costs (per survivor)

No Sepsis (n = 748)

Severe Sepsis (n = 100)

p-value

641 14.00.3 190.8 3.90.4 10 1329 1265112 1409

691 14.31.9 462 14.31.9 22 46956 5199748 6665

0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001

Admissions with and without severe sepsis CONCLUSION. Severe sepsis accounts for 12 % of all admissions to this mixed ICU. These patients require a disproportionate allocation of ICU occupancy, nursing workload and direct costs budget for their treatment. REFERENCE. Parrillo JE. N Engl J Med 1993;328: 1471±1477

577

579

SEVERE SEPSIS COST IN THE PARISIAN AREA: ANALYSIS OF THE CUBREA DATA BASE

DIRECT COSTS OF SEVERE SEPSIS PATIENTS IN THREE GERMAN INTENSIVE CARE UNITS BASED ON RETROSPECTIVE ELECTRONIC PATIENT RECORD ANALYSIS OF RESOURCE USE

Guidet BG1, Aegerter PA1, Dreyfuss DD1, Gauzit RG1, Poitrinal PP2, Bouhassira MB2, Meshaka PM2, Pinton PP2. 1For the Cub Rea Study Group, Hopital Saint Antoine, Paris, 2, Eli Lilly & Compagny, Saint Cloud, France INTRODUCTION. The CUB-REA data base prospectively recorded 55047 consecutive patients over a 3-year period (1997±99) in 34 ICU (452 beds) of the Parisian area (65.7 % university, 34.3 % other). Severe sepsis (SS) according to Bone's criteria, with a length of stay (LOS) > 2 days, accounts for almost 20 % of all ICU stays. The purpose is to evaluate the economic burden of SS over the same period. METHODS. The cost assessment equations used in this study were previously described1,2. Methods 1 estimates the cost by applying a formula using LOS, work load assessed with the Omega system, vital status and SAPS II. The direct variable cost covers: clinical support services (lab, radiology), consumables (disposable, blood and blood products, drugs and fluids) and clinical nursing staff. The fixed direct cost covers: medical staff and head nurses. The indirect cost covers: administration, hotel cost, and laundry. The total ICU cost is the sum of the 3 costs. Method 2 estimates the total hospital cost by classifying3 patients in 16 groups according to the number and the duration of organ support and applying to each group the modified diagnosis related group (DRGs). The method was then valued with a national cost per DRG. Patients costs were stratified on survivors and non-survivors. All cost are given in Euro currency. RESULTS. 10,459 (19 %) patients fulfilled our criteria. The costs are shown in table 1 and 2. Method 2: Among the 16 groups, 3 groups accounted for almost half of the patients. Group 10 (Respiratory and circulatory support with a LOS > 10 days: 23.9 % with a cost per stay of A 54,844. Group 15 (Respiratory, circulatory and renal support with a LOS > 10 days: 11 % with a cost per stay of A 60,121. Group 7 (single respiratory support for more than 10 days): 9.7 % with a cost per stay of A 23,248.

Direct ICU variable cost Total ICU cost Total hospital cost

Mean

1 sd

Median

19,534 24,476 33,377

25,521 33,934 36,088

11,112 15,284 21,985

Method 1

Total hospital cost Survivors Non survivors

Mean

1 sd

Median

33,647 33,185 38,488

20,277 19,474 21,190

25,572 22,189 54,844

Method 2 CONCLUSION. Our method provides a tool for medico economic evaluation of new therapy in SS, which is a costly disease. REFERENCES. 1. Sznajder M. Intensive Care Med 2001, 27: 146±53. 2.Snajder M. Intensive Care Med 1998, 24: 582±89. 3. Misset B. RØan Urg 1998, 7: 367±74.

Moerer O1, Hofmann M2, Herklotz A3, Schmid A4, Schneider H4, Burchardi H1. 1Dept Anaesthesiology, Univ. Hospital Goettingen, Goettingen, 2Dept Anaesthesiology, Univ. Hospital Jena, Jena, 3Dept Internal Medicine, Univ. Hospital Halle, Halle, Germany, 4HealthEcon Ltd., Basel, Switzerland INTRODUCTION. Literature reflects the continuous progression in knowledge about pathology, incidence and risk factors of sepsis and its sequelae. It fails, however, to provide any information relating to the actual cost associated with sepsis. This study aimed at calculating the direct costs caused by severely septic patients in the ICU. METHODS. A bottom up approach [1] was used to determine the direct ICU cost on actual resource use (medication, laboratory tests, microbiological analysis, disposables, and clinical procedures) for patients with severe sepsis. A retrospective analysis was done using 385 electronic patient records from three adult ICU's in three university hospitals in Germany. To get the total direct costs, centre-specific personnel and basic bed (ªhotelº) cost were added to total resources consumed. RESULTS. ICU mortality of severely septic patients was 36.6 %. Mean ICU LOS was 16.6 days. Survivors stayed 4 days longer than non-survivors (18.4 vs 14.4). The mean direct ICU cost of care were 25696 (  20893) Euros per patient and 1454 Euro per day. In comparison, average daily charges for an ICU patient in Germany amount to 851 Euro (based on official statistics). Nonsurvivors were more expensive than survivors in total costs 28503 vs 24012 Euro and in per day cost (1848 vs 1269). Medication makes up the largest part of the direct costs, followed by expenses for personnel.

Distribution of total direct ICU cost

Hotel

Staff

Disposables

Micobiology

Laboratory

Medication

6%

33 %

3%

3%

18 %

37 %

CONCLUSION. Patients with severe sepsis have a high ICU mortality rate, prolonged ICU LOS, and are substantially more expensive to treat than non-septic ICU patients. Non surviving septic patients are more costly than survivors despite shorter ICU LOS. This is due to higher medication costs indicating increased efforts to keep patients alive. REFERENCE. Schuergers D et al.: Bottom up costing in the ICU. Intensive Care Med 2000; 26(Supp 3): S329

14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

S 284

580 SEPSIS COSTS IN ITALY

Lucioni C1, Mazzi S1, Currado I2, Langer M2. 1Health Economics, Institute of Health Economics, Milano, Italy, 2Intensive Care Service, San Matteo Hosp., Pavia, Italy INTRODUCTION. Sepsis is a disease for which the treatment requires special units providing intensive care in terms of personnel, equipment and drugs. Stay in those units consequently implies high costs that are not yet extensively studied and published in Italy. METHODS. The aim of this study is to evaluate additional hospitalization costs and intangible costs (mortality) in patients with ªsevere sepsis or septic shockº (SS) in Italy. The evaluation is based on clinical data from the Italian Sepsis Study, a prospective, multicentre study conducted in 99 Intensive Care Units (ICUs) across Italy. In particular, data collected included the Average Length Of Stay (ALOS) in ICU and later in the regular ward, and the mortality within four weeks and in hospital. Out of the 2,946 patients enrolled, 2,641 never developed SS and were considered as the control group (comparability was confirmed based on gender, age, and comorbidity). The additional (respective to the control group) ALOSs of the patients with SS were valued in monetary terms using per diem full costs, inflated to 2000: 1,033.43 Euros for 1 day in ICU (ref.1) and 299.54 Euros for 1 day in the regular ward (ref.2). Statistical significance was tested with Student t test. RESULTS. The ALOS in ICU was 7.1 days in the control group and 18.7 days in patients with SS (p < 0.001); ALOSs in the general ward were respectively 14.2 days and 7.5 days (p < 0.05). Taking the 11.6 days more in ICU and the 6.7 days less in the general ward spent by patients with SS, and applying daily costs, an average additional hospitalization cost per SS case was obtained of 9,981 Euros. Mortality within four weeks was 23.3 % in the control group and 65.3 % in patients with SS (p < 0.001); mortality in hospital were 25.8 % and 80.3 % respectively (p < 0.001). A more in-depth analysis was conducted making a distinction between patients with SS on admission and patients developing SS after admission (155 and 150 respectively). The average additional cost per case of SS developed after admission resulted 15,200.35 Euros, versus 5,373.73 for a case of SS on admission (p < 0.001 for different ALOSs in ICU and p < 0.2 for different ALOSs in the regular ward). Also mortality in hospital was higher (p < 0.01) in patients with SS developed after admission: 87.3 % versus 73.6 %. Other factors possibly impacting on stay and mortality (such as age, diagnosis on admission, etc.) were also investigated. CONCLUSION. The hospitalization cost of a patient with sepsis (21,571.88 Euros) is significantly higher ( + 86 %) than that of a patient without sepsis (11,590.84 Euros), due to a longer ( + 163 %) stay in the expensive ICU, not off-set by a shorter stay in the regular ward. Also intangible costs are significantly higher: the risk for a patient with sepsis to die in hospital is 3 times higher than that of a patient without sepsis. In particular, those patients developing sepsis after admission are more costly and have a higher mortality risk. REFERENCES. 1. Cavallo MC et al., Il costo del reparto di terapia intensiva in Italia. Minerva Anestesiologica, in press. 2. Langiano T, DRG: strategie, valutazione, monitoraggio; Il Pensiero Scientifico, Roma, 1997.

Oral Presentations Preventing nosocomial pneumonia ± 582±588 582 A RANDOMISED CLINICAL TRIAL OF INTERMITTENT SUBGLOTTIC SECRETION DRAINAGE IN MECHANICALLY VENTILATED PATIENTS

Hoeven JV, Smulders CM. Intensive Care Unit, Bosch Medicentrum, Den Bosch, Netherlands INTRODUCTION. Mechanical ventilation is frequently complicated by the development of nosocomial pneumonia. Aspiration of subglottic secretions may be an important contributing factor. This study describes the effect of subglottic secretion drainage on the incidence of ventilator-associated pneumonia (VAP) in mechanically ventilated patients. METHODS. A randomized clinical trial in a 12-beds general ICU in which we included 150 patients with an expected duration of mechanical ventilation of longer than 72 hours. Patients were randomly assigned to receive either a tracheal tube for intermittent subglottic secretion drainage (study group) or a standard endotracheal tube (control group). Outcome measurements were the incidence of VAP, duration of mechanical ventilation, length of ICU stay, length of hospital stay and mortality. RESULTS. Seventy-five patients were assigned to the study group and seventy-five to the control group. The two groups were similar at the time of randomization with regard to demographic characteristics and severity of illness. VAP was seen in 3 patients (4 %) in the study group and in 12 (16 %) patients in the control group (RR, 0.22; 95 % CI, 0.06; to 0.81 p = 0.014). The other outcome measures were not significantly different between the two groups. Patients who developed a ventilator associated pneumonia, had a statistically longer duration of mechanical ventilation and a longer total hospital stay (both p < 0.05) than patients without VAP. CONCLUSION. Intermittent subglottic secretion drainage reduces the incidence of VAP in patients with an expected duration of mechanical ventilation of longer than 72 hours.

581

583

SEVERE SEPSIS: A EUROPEAN ESTIMATE OF THE BURDEN OF DISEASE IN ICU

NON-PHARMACOLOGIC STRATEGIES TO PREVENT VENTILATORASSOCIATED PNEUMONIA: ADHERENCE AND EVIDENCE-BASED GUIDELINES

Davies A1, Green C1, Hutton J1, Chinn C2. 1European Office, Medtap International, London, UK, 2 Health Outcomes Res, Eli Lilly and Company Ltd, Surrey, UK INTRODUCTION. Severe sepsis and septic shock are a major cause of morbidity and mortality associated with patients in intensive care units. The total burden of the disease in Europe is unknown. As a proportion of ICU burden the costs and mortality associated with severe sepsis are likely to vary within Europe, and be much greater than the proportion of patients with the disease in ICUs. METHODS. We sought to estimate the European burden of health care costs (in ICU) and hospital mortality with information obtained from published international literature, official statistics and correspondence with researchers in the field of intensive care. A literature search for incidence, costs, mortality and other outcomes associated with the disease was performed. The literature identified was reviewed for data that could inform estimates of burden. Supplementary sources were consulted as appropriate. The table shows those countries for which the burden could be estimated. Rates for incidence of severe sepsis in ICU, and hospital mortality, were applied to best estimates of total annual ICU admissions. Costs are presented in Euros. RESULTS. The aggregate ICU burden associated with severe sepsis and septic shock in the 6 countries studied (representing 83 % of the European Union's population), is estimated at A 5.2 bn to A 6.9 bn, and mortality between 99,000 and 128,000 annually. Applying average incidence rates to the rest of the European Union, the total burden is estimated at A 6.2 bn to A 8.3 bn, and 118,000 to 153,000 deaths. Despite the present uncertainties, this clearly shows the significant burden associated with severe sepsis. We estimate the 10±14 % of European ICU patients who develop this life threatening disease are associated with approximately 40 % of the total costs in the ICU.

UK France Germany Italy Spain Holland Total

ICU '000 s admission

Incid. Low

incid. high

cases '000 s low-high

deaths '000 s low-high

A bn low

A bn high

159.2 451.2 1038.0 186.1 189.6 100.0 2124.1

21.4 % 11.9 % 7.8 % 10.4 % 10.4 % 11.0 % 10.3 %

24.7 % 19.8 % 9.4 % 12.7 % 10.4 % 20.4 % 13.6 %

34.0±39.4 53.7±89.2 80.6±97.6 19.3±23.6 19.7 11.0±20.4 218.2±289.8

15.2±17.6 20.3±33.7 28.7±34.7 15.4±18.9 15.1 4.1±7.5 98.9±127.6

0.871 1.421 1.853 0.372 0.468 0.177 5.163

1.007 2.360 2.245 0.455 0.468 0.329 6.865

CONCLUSION. There are ranges of possible values for many key parameters, and thus uncertainty around the estimates of deaths and costs associated with severe sepsis. We have nevertheless quantified this on the basis of European data. Our results are conservative estimates of the total burden, as they incorporate neither non ICU and non health care costs, nor the disease's associated morbidity. This additional burden would further emphasise the need to focus attention on the treatment of this patient group. REFERENCE. Angus C. A. Sirio C. A. Clermont G. and Bion J. International comparisons of critical care outcome and resource consumption. Critical Care Clinics 1997, 13(2): 389±407

Lorente C, Diaz E, Bodí M, Ricart M, Rello J. Critical Care, Hospital Universitari Joan XXIII de Tarragona, Tarragona, Spain INTRODUCTION. There are somes recomendations to prevent the development of ventilator-associated pneumonia. The objective of our study was to assess the rate of adherence and the most important barriers to physicians adherence to evidence-based guidelines for preventing VAP. METHODS. We used a survey about 20 non pharmacologic strategies to prevent ventilator-associated pneumonia recomended by Kollef . We asked 110 opinion leaders on VAP to answer this survey. They had to answer if they implemented each of the preventive strategie on their daily clinical practice, and if not, they were asked to answer why not, choosing only one reason for non-adherence. We have assessed the adherence to these strategies and the degree of adherence in evidence-based: strategies A (supported by at least two randomized, controled investigations), B (supported by at least one randomized, controlled investigation), C (supported by nonrandomized, concurrent-cohort investigations, historical-cohort investigations, or case series), D (supported by randomized, controlled investigations of other nosocomial infections) and U (undetermined or not yet studied in clinical investigations). We too assessed the adherence to strategies considered as effective and unproven/uncertain. Statistical analysis: Fisher exact test. RESULTS. Sixty-two physicians representing 22 countries responded the survey. Adherence was 80.4 %. The most important reasosns for non adherence were: no availability (7.4 %), disagree with the interpretation of the results of the clinical trials (3.7 %) and excesive costs (2.8 %). Adherence with recomendations graded A, B, C, D and U was 65.9, 85.4, 89.8, 91.8 and 70 % repectively. Adherence with effective strategies was 83.6 % and 76.6 % with unproven / uncertain strategies. CONCLUSION. Adherence to non pharmacologic strategies for preventing VAP is low. Non adherence was associated with a variety of factors but adherence is independent of the degree of evidence. REFERENCE. Kollef MH. The prevention of ventilator-associated pneumonia. N Eng J Med 1999;340: 627±634.

14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

S 285

584

586

FEASIBILITY OF THE SEMI-RECUMBENT POSITION (SRP) IN THREE MIXED INTENSIVE CARE UNITS

ANALYSIS OF RESPIRATORY AND NON-RESPIRATORY INFECTIONS IN PUBLISHED TRIALS OF SELECTIVE DIGESTIVE DECONTAMINATION

Van Nieuwenhoven CA1, Van Tiel F2, Vandenbroucke-Grauls C3, Strack van Schijndel R4, Joore H5, Ramsay G1, Van Tweel I6, Bonten M5. 1Intensive Care Medicine, 2Medical Microbiology, University Hospital Maastricht, Maastricht, 3Medical Microbiology, 4Internal Medicine, Free University Hospital Amsterdam, Amsterdam, 5Internal Medicine, University Hospital Utrecht, 6Statistical service, Julius Center, Utrecht, The Netherlands

Redman R1, Ludington E2, Crocker M2, Wittes J2, Bellm L1, Carlet J3, and the VAP Advisory Group4. 1Medical Affairs, IntraBiotics Pharmaceuticals, Inc., Mountain View, 2None, Statistics Collaborative, Washington (DC), USA, 3Service de Reanimation Polyvalente, Hopital St. Joseph, Paris, France, 4

INTRODUCTION. SRP position has been proposed as a measure to prevent Ventilator-associated Pneumonia (VAP). We determined the feasibility of such a treatment policy in mechanically ventilated intensive care patients. METHODS. Patients, able to be cared for in 45 , were randomized to SRP (45 ) or supine position (SP;10 ). Backrest elevation was measured continuously every 60 seconds, using a computerized potential meter with pendulum connected to the monitor of the patient. Only backrest elevation measurements of > 1 were used for analysis. Missing periods of measurement, due to transport or diagnostic procedures, were recorded as SP in both groups. Mean backrest elevations per day were calculated for the first 7 days of ventilation. Variability was recorded as number of times deviated > -5 from initial backrest elevation. RESULTS. Measurements were available from 84 patients in SRP and 90 in SP, with mean duration of measurement of 5.0  2.0) and 5.4  1.9 days, respectively. Mean daily levels of elevation during the first week are presented in table 1 (p < 0.05). Mean number of deviations was 6,3 and 2,8 times per day (p < 0,001) in the SRP and SP group respectively.

SRP (in degrees) Supine (in degrees)

Day 1

Day 2

Day 3

Day 4

Day 5

Day 6

Day 7

29.2

26.8

25.5

23.9

23.1

24.7

26.5

9.9

10.8

11.9

12.4

13.6

14.3

14.7

INTRODUCTION. Selective Digestive Decontamination (SDD) is designed to reduce bacterial colonization by applying a mixture of several topical antibiotic agents to the mouth, oropharynx and stomach; theoretically, the reduction in colonization decreases a patient's risk of developing ventilator-associated pneumonia (VAP). Most clinical studies and meta-analyses evaluating SDD have demonstrated decreased rates of VAP. However protection against early- vs lateonset VAP, the impact of concomitant prophylactic systemic antimicrobial agents and the development of non-respiratory infections has not previously been systematically evaluated. METHODS. A hypothesis-driven meta-analysis was conducted of all published randomized controlled trials of SDD for the prevention of VAP performed in an intensive care unit in which 80 % or more of the study patients were ventilated. Rates of early- and late-onset VAP and nonrespiratory infections were evaluated. Mantel-Haenszel methods were used to calculate odds ratios. RESULTS. For all studies evaluated, patients receiving SDD had overall reduced risk of VAP (odds ratio (OR) = 0.36; 95 % confidence interval (CI):0.28±0.46), including reduced risk of VAP within 48 hours (OR = 0.46, CI:0.30±0.71, p < 0.001), after 48 hours (OR = 0.43, CI:0.34±0.55, p < 0.001), within 4 days (OR = 0.29, CI:0.18±0.47, p < 0.001) and after 4 days (OR = 0.60, CI:0.36±1.00, p = 0.063). Protection against VAP was seen in studies that did and did not include administration of intravenous antibiotics as part of the SDD regimen (OR = 0.31, CI:0.20±0.46, p < 0.001 and OR = 0.40, CI:0.29±0.55, p < 0.001, respectively). For studies including and excluding patients diagnosed with pneumonia at the time of enrollment, SDD protected against VAP occurring after 48 hours (OR = 0.59, CI:0.44±0.80, p < 0.001 and OR = 0.21, CI:0.13±0.33, p < 0.001, respectively). SDD also showed protection against bacteremia (p = 0.004) and urinary tract infections (p = 0.001).

CONCLUSION. The aimed semi-recumbent position of 45 , was hardly feasible in ICU patients. With maximal efforts, backrest elevation levels of 25±30  were achieved, which still was a significant difference as compared to standard care. However, changes of the patients position occurred 2.25 times more frequently in patients in SRP.

CONCLUSION. Patients treated with topical antibiotics for SDD, with and without concomitant use of intravenous antibiotics, have reduced risk of VAP. SDD appears to protect against bacteremia and urinary tract infections as well.

585

587

THE EFFECT OF SEMI-RECUMBENT POSITION ON DEVELOPMENT OF VENTILATOR-ASSOCIATED PNEUMONIA (VAP)

A PHASE IIA SAFETY AND MICROBIAL KINETIC STUDY OF ISEGANAN (IB-367) ORAL SOLUTION IN MECHANICALLY VENTILATED PATIENTS

Van Nieuwenhoven CA1, Van Tie Fl2, Vandenbroucke-Grauls C3, Strack van Schijndel R4, Joore H5, Ramsay G1, Van Tweel I6, Bonten M7. Intensive Care Medicine, 2Medical Microbiology, University Hospital Maastricht, Maastricht, 7Medical Microbiology, 5Internal Medicine, University Hospital Utrecht, 6Statistical Service, Julius Center, Utrecht, 3Medical Microbiology, 4 Internal Medicine, Free University Hospital Amsterdam, Amsterdam, The Netherlands

Redman R1, Kollef M2, Jensen K1, Mertens R3. 1Medical Affairs, 3Clinical Research, IntraBiotics Pharmaceuticals, Inc., Mountain View, 2Pulmonary and Critical Care Division, Washington University School of Medicine, St. Louis, USA

INTRODUCTION. Aspiration of oropharyngeal and gastric contents is a risk factor for VAP. The semi-recumbent treatment position may reduce aspiration and development of VAP. We performed a prospective, randomized, controlled, multi-center trial on the effects of body position on the development of VAP. METHODS. Patients with an expected duration of ventilation 48 hours and able to be positioned in a backrest elevation of 45  were randomized to the semi-recumbent group (SRP; backrest elevation of 45 ) or the supine group (backrest elevation of 10 ). Backrest elevation was continuously monitored during the first week of ventilation. VAP was diagnosed on bronchoscopic techniques. Statistical analysis was performed using sequential interim analysis and on intention-to-treat principle. RESULTS. 221 patients were included; 109 patients to supine and 112 to SRP. Demographics and baseline clinical characteristics were similar in both study groups. Mean backrest elevations in the first week were 25,7 for SRP and 12,5 for supine (p < 0.05). The incidence of VAP was 7,3 % (8 patients) in supine and in 11,6 % (13 patients) in SRP (NS). Time of diagnosis of VAP was 5,9(  2,4) and 6,7(  2,9) days in the supine and SRP, respectively (NS). Duration of ventilation, number of days in study, and ICU-stay were comparable for both groups. Use of enteral feeding was equal for both groups (87 % vs. 82 %). ICU-mortality was equal in both groups (30 % vs. 29 %; NS). CONCLUSION. The semi-recumbent patient positioning with an aimed backrest elevation of 45  does not reduce the incidence of VAP.

INTRODUCTION. Iseganan, previously known as IB-367, is a synthetic analog of the protegrin antimicrobial peptides with broad activity against gram-positive and gram-negative bacteria and yeast and low potential for resistance. We conducted a Phase IIa, multi-center, placebo-controlled study evaluating the potential of iseganan oral solution as an oral decontaminant for prevention of ventilator associated pneumonia (VAP). METHODS. Sixteen orally intubated patients were randomized to receive 9 mg iseganan (n = 6) or placebo solution (n = 2) every (q) 4 hours (h), or 9 mg iseganan (n = 6) or placebo solution (n = 2) q 6 h for up to 5 days. For each administration, 3 mL of study drug were applied to all surfaces of the oral cavity and retained for at least 5 minutes. Oral secretions were collected pre- and post-dose for quantitative microbial analysis q 24 h. RESULTS. Iseganan oral solution was well tolerated. Immediate decreases in mean oral microbial burden were seen after administration of the first dose of each day: overall mean decrease for each day patients were intubated were 1.0 log colony forming units (CFU) for iseganan and 0.1 log CFU for placebo (p = 0.01). A cumulative decrease was seen over 5 days of administration for iseganan q 4 h (n = 3; 3.2 log CFU) vs placebo (n = 2; 0.9 log CFU). Data where oral CFUs were below the limit of detection and thus suspect to technical error were excluded from statistical analyses. CONCLUSION. Oral-topical administration of iseganan solution safely and rapidly reduces the oral microbial burden of orally intubated and mechanically ventilated patients. Decreases in oral microbial burden are observed daily with both q 4 h and q 6 h dosing; cumulative decreases over 5 days of treatment are seen with q 4 h administration. Iseganan solution is a promising broad spectrum single agent candidate for the prevention of VAP.

S 286

14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

588

590

EARLY ONSET VENTILATOR-ASSOCIATED PNEUMONIA (EOP) IN ICU PATIENTS: IMPACT OF INITIAL ANTIMICROBIAL THERAPY

CEREBRAL BLOOD FLOW IN TRAUMATIC CONTUSIONS

Bornstain C1, Azoulay E1, De Lassence A1, Cohen Y1, Moine P1, Garrouste M1, Thuong M1, Schlemmer B1, Timsit JF1. For The outcomerea study group1. 1Reanimation medicale, Hopital Saint Louis, Paris, France INTRODUCTION. Assessing risk factors for EOP, the most frequent nosocomial pneumonia (1) could help to improve prognosis in ICU-patients and to define preventive strategies. METHODS. Objective of the study was to define the baseline and time-dependent factors associated with the occurrence of EOP among ICU patients. A prospective multicenter (7 ICUs) study was conducted in Paris (France) over a 3-y period. Data from 747 patients mechanically ventilated within 12 hours of ICU admission for more than 48 hours were prospectively recorded: severity of illness, ventilation, nutrition, pattern of stress ulcer prevention, use of invasive devices, antibiotics and vasopressors. EOP was defined on both clinical and bacteriological grounds, using Andrews' criteria (2) occurring 2 to 6 days after ICU admission, associated with significant quantitative results of distal bronchial sampling (3) and culture. RESULTS. 80 out of 747 patients (10.7 %) developed EOP. Univariate analysis: patients with EOP had a higher Glasgow coma score on admission than patients without EOP (7.72  4 vs. 6.7  4 respectively p = 0.04). Patients with EOP received less antibiotics on admission and at Day 1 than the others (74.4 vs. 50 % p = 0.0071 and 81 vs. 55 % p = 0.0025 respectively). EOP patients were more likely to receive enteral nutrition between admission and Day 2 ( 40.5 vs. 52.5 % p = 0.04). Overall mortality and hospital stay were similar in both groups. ICU stay was significantly longer for EOP patients (median 16 [10±25] days vs. 11 [6±21] days p = 0.0006). Multivariate analysis: patients who received aminoglycosides (OR 0.36 [95 % CI: 0.15±0.85] p = 0.02) and third-generation cephalosporins (OR 0.51 [95 % CI 0.04±1.1] were less likely to develop EOP. Glasgow coma score between 7 and 13 on admission was associated with an increased risk of EOP (OR 2.29 [95 % CI: 1.38±3.79] p = 0.001), as well as enteral nutrition at Day 1, although not significant (OR: 1.5 [0.93±2.39], p = 0.099). CONCLUSION. In our population of mechanically ventilated patients, those admitted with a Glasgow coma score between 7 and 13 are more likely to develop EOP. Antibiotic exposure on admission decreases the risk for EOP, especially with aminoglycosides. Further studies are needed to assess the protective effect of such an ªantibioprophylaxisº, although this strategy is challenged by the risk for selection of resistant bacteria.

Chieregato A1, Fainardi E2, Tanfani A1, Bocchino M1, Bucci S1, Cocciolo F1, Servadei F3, Targa L1. 1NeuroICU, 2Neuroradiology, 3Neurosurgery, Bufalini Hospital, Cesena, Italy INTRODUCTION. Cerebral contusions seems to be associated to a reduction of regional cerebral blood flow (rCBF), measured by means of Xenon-enhanced computerized tomography (1). We aimed to evaluate, over time, the topography of rCBF in contusion and to observe if any relationship between rCBF and baseline paCO2 and CPP occurs. METHODS. We performed 52 studies in 28 contusions, larger than 2 cm, from 19 patients with severe head injury. The patient were managed with a stair case protocol to maintain ICP < 20 mmHg and CPP > 70 mmHg. During the CBF studies, ICP, CPP and arteriovenous differences were measured. Any effort was made to maintain paCO2 and CPP stable regarding previous values obtained in ICU. Six different ROIs were obtained freehand on the Ct scan: 1) the hemorrhagic core, 2) the pericontusional hypodensity, 3) within 1 cm rim of normal appearing brain tissue, 4) the hemisphere of the same side (excluding the contusion), 5) controlateral, in an area symmetric to the contusion, 6) the whole brain. RESULTS. Median age of the patient was 28 yrs (30), sex (12 males), median mGCS 3 (3), pupillary abnormalities in 9 cases. The worst Ct included DI II (6), DI III (2), EML (11, 1 contusion, 8 SDH, 2 EDH). All the patients survived the acute phase. During the CBF studies the following median values were found: CPP 76 mmHg (23), ICP 16 mmHg (10), paCO2 36 mmHg (6), hb 10 gr/dl (2), AVDO2 3.9 ml/dl (2.3), AVDL ±0.1 mmol/l (0.1), CMRO2 1.2 ml/100 gr/ min (0.6). CBF studies were separated in those before (n = 26)(median 39 h, 46 IQR) or after 96 h post injury (n = 26) )(median 203 h, 150 IQR) (tab). No differences between CPP, paCO2 and CMRO2 were found. In 23 cases on 26 measurements obtained 96 h post injury, rCBF in the pericontusional hypodense area was below 20 ml/100 gr/min (the ischaemia threshold). No clear relationship was found between baseline CPP and paCO2 and rCBF values. core CBF < 96 h ml/100 gr/min CBF > 96 h ml/100 gr/min MannWhitney

Peri

normal

hemisphere

contra

global

ANOVA

16 (29)

36 (33)

47 (28)

43 (26)

40 (22)

42 (21)

0.0004

9 (22)

15 (16)

27 (19)

27 (17)

23 (21)

26 (18)

0.0003

0.06

< 0.01

< 0.01

< 0.05

< 0.05

< 0.01

REFERENCES. 1. Cook DJ et al. Incidence and risk factors for ventilator asociated Pneumonia in critically ill patients. Ann Int Med 1998; 129 (6) 433±40. 2. Andrews CP et al. Diagnosis of nosocomial pneumonia in acute, diffuse lung injury. Chest 1981; 80: 254±58. 3. Chastre J et al. Invasive diagnostic testing should be routinely used to manage ventilated patients with suspected pneumonia. Am J Resp Dis Crit Care Med 1994; 150: 570±74

CONCLUSION. The values of rCBF in the contusion reduce over time. Initially rCBF in pericontusional and normal appearing areas is only slightly reduced and sometimes is abnormally high (even respect contralateral brain). Later, all CBFs reduce without any biochemical and clinical sign of global ischaemia, but with values in pericontusional area below the threshold for ischaemia. CPP and paCO2 values, in a range in accord to guidelines, do not seem affect rCBF in pericontusional hypodense area. REFERENCE. Schröder ML et al. J Neurosurg 1995; 82: 966±971

Oral Presentations CNS trauma ± 589±594

591

589 RELATIONSHIP BETWEEN GLOBAL CEREBRAL METABOLISM AND CEREBROVASCULAR PRESSURE REACTIVITY AFTER HEAD INJURY

Steiner LA, Coles JP, Czosnyka M, Minhas PS, Menon DK, Pickard JD. Wolfson Brain Imaging Centre, University of Cambridge, Cambridge, UK INTRODUCTION. Intact cerebrovascular autoregulation (AR) after head injury is a powerful protective mechanism against secondary ischaemic insults. However, AR is frequently impaired following head injury due to endothelial or metabolic dysfunction or management at an inappropriate cerebral perfusion pressure (CPP). We have investigated the relationship between an index of cerebrovascular pressure-reactivity and global cerebral metabolism. METHODS. Twenty-seven Positron Emission Tomography (PET) examinations were performed in 23 patients with severe closed head injury to determine mean global oxygen extraction fraction (OEF), cerebral metabolic rate for oxygen (CMRO2) and cerebral blood flow (CBF) using 15O and standard kinetic models. Pressure-reactivity was monitored continuously using an index of cerebrovascular pressure-reactivity (PRx) (1) and averaged over the two hours preceding PET. Only patients with CPP comparable to that measured during PET were included. Respiratory parameters were not changed between determination of PRx and PET. RESULTS. PRx correlated to mean global OEF by a ªU-shapedº relationship (second degree polynomial; r2 = 0.53, p = 0.0001) i. e. an inappropriately low or high OEF was correlated with disturbed pressure-reactivity. The correlation to mean global CMRO2 followed an inverse function (r2 = 0.32, p = 0.002) i. e. low CMRO2 correlated with disturbed pressure-reactivity. There was no significant correlation between PRx and mean global CBF. CONCLUSION. Disturbed pressure-reactivity, which implies disturbed pressure AR correlates with metabolic failure which may be seen during both ischaemia (high OEF) or hyperaemia (low OEF). There are three possible explanations for the correlation between PRx and metabolism: a) disturbed metabolism and AR are a consequence of the severity of the trauma. b) the disturbed AR is the consequence of disturbed metabolism and c) metabolism is disturbed as a consequence of inadequate AR. Studies using PET before and after a step change in CPP are needed to further characterize this relationship. The lack of a correlation between PRx and CBF was possibly due to confounding variables that were not controlled (e. g. temperature) and the fact that most of our patients had adequate CBF and functional or only moderately disturbed pressure-reactivity. REFERENCE. Czosnyka M. et al.: Continuous assessment of the cerebral vasomotor reactivity in head injury. Neurosurgery 1997 (41) p. 11±17

THE USE OF INDOMETHACIN IN THE TREATMENT OF PLATEAU WAVES

Imberti R, Fuardo M, Bellinzona G, Langer M. II Servizio di Anestesia e Rianimazione, IRCCS Policlinico S. Matteo, Pavia, Italy INTRODUCTION. Acute rises of intracranial pressure (ªplateau wavesº) are due to sudden cerebral vasodilatation occurring in patients with reduced cerebral compliance(1,2). Under these conditions, the administration of indomethacin, a potent cerebral arteriolar vasoconstrictor, could be appropriate. The purpose of the study was to evaluate this hypothesis. METHODS. Indomethacin (Chiesi, Italy) was administered as a bolus i. v. (15±20 mg) in 5 patients affected by severe traumatic brain injury (GCS < 8) during episodes of plateau waves. The following parameters were continuously monitored in all patients: MABP, ICP (Camino Laboratories), CPP, ETCO2, cerebral tissue PO2 (PbrO2) (Licox system, GMS, Germany). Data were recorded by a multiparametric monitor and stored at 15-second intervals in a dedicated personal computer provided with adapted software (LabVIEW, National Instruments, USA). SvjO2 and veno-arterial PCO2 (v-a PCO2) difference was evaluated by intermittent blood sampling from the jugular bulb 5 minutes after indomethacin administration. FiO2 was adjusted to obtain PaO2 values ranging between 100 and 150 mm Hg. Hemoglobin levels were maintained above 10 g/dl. During 3 episodes, mean cerebral artery flow velocities were evaluated by transcranial doppler sonography (Explorer DMS, France). RESULTS. During plateau waves mean ICP was 60.1  18 mm Hg. Indomethacin administration resulted in a steep and wide ICP reduction in all episodes. After 5 minutes, mean ICP was 21.4  8 mm Hg, p < 0.001). As a consequence CPP increased from 39.7  17 to 75.2  12 mm Hg (p < 0.001). During plateau waves mean SvjO2 was 47.2  8.2 % and PbrO2 was 16.9  13 mm Hg. Indomethacin administration increased both SvjO2 and PbrO2 (61.5  7.5 % and 26.5  15.5 mm Hg, respectively; p < 0.05) and reduced the veno-arterial PCO2 difference from 8.5  2.8 to 5.9  1.7 (p < 0.01). Indomethacin increased the mean flow velocity in the mean cerebral artery from 45.1  7.6 to 60.5  9.2 (p < 0.01), mainly by increasing the diastolic velocity. The effects of indomethacin on ICP lasted from 10 to 60 minutes. Decompressive craniotomy was necessary in 2 patients. CONCLUSION. This study indirectly confirms that plateau waves are due to cerebral vasodilatation and shows that indomethacin is effective in the extinguishing plateau waves, ultimately improving cerebral perfusion and oxygenation. REFERENCES. 1. Rosner MJ, Becker DP. J Neurosurg 1984, 60, 312±324. 2. Czosnyka M, Smielewski P, Piechnik S. J Neurosurg 1999, 91, 11±19.

14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

S 287

592

594

CLINICAL EVALUATION OF SERUM S-100B PROTEIN AS A ROUTINE LABORATORY PARAMETER FOR BRAIN CELL DAMAGE IN CRITICAL CARE PATIENTS

NITROTYROSINE AND SUPEROXIDE DISMUTASE INDUCTION AFTER TRANSIENT SPINAL ISCHAEMIA UNDER HYPERBARIC OXYGENATION

Raabe A, Kopetsch O, Lang J, Tsimpas A, Zimmermann M, Seifert V. Department of Neurosurgery, Johann Wolfgang Goethe University, Frankfurt am Main, Germany

Murakami N1, Horinouchi T1, Sakurai M1, Suzuki H2, Hashimoto Y2, Kato M1, Matsukawa S1.1department Of Anesthesiology, Tohoku University School of Medicine, Sendai, 2Department of Anesthesiology, Sen-en Hospital, Tagajyo, Japan

INTRODUCTION. Several biochemical substances have been studied to find a specific marker for the brain that indicates cell damage equivalent to creatine kinase-MB or troponin-T for myocardial cells. S-100B protein is regarded as one of the most promising biochemical markers of glial cell damage. Serum S-100 protein was mainly investigated in stroke and head injury (1,2,3). The objective of our study was to investigate whether serum S-100B is useful as a biochemical monitoring of critical care patients who suffer from a secondary brain insult. METHODS. Two-hundred-twenty neurocritical care patients with different intracranial and spinal pathologies were included in our study. Serum S-100B protein was measured daily using a immunoluminometric assay (LIAISON, Byk-Sangtec Diagnostica, Dietzenbach, Germany). Patients with intrinsic brain tumors or melanoma metastasis were excluded. Time course and peak values of serum S-100 protein were analysed in two groups of patients: 1) patients with severe neurological complications (postoperative stroke, hypoxia, intracerebral hemorrhage, hydrocephalus, brain oedema, hermiation) and 2) patients without neurological complications. Values were compared to investigate sensitivity, specificity, positive and negative predictive value of serum S-100 protein in objectifying the extent and the time course of secondary brain damage. RESULTS. 28 of the 220 neurocritical care patients (13 %) suffered from a severe neurological complication. The most frequent complication were stroke, intracranial hypertension and postoperative hemorrhage. Patients suffering from neurological complications showed significantly higher values of serum S-100 protein compared to patients without complications (median 1.33 mg/l vs. 0.19 mg/l, p < 0.001, Mann-Whitney U test). All patients with neurological complications showed increased values of serum S-100B protein. Peak value and time course were closely associated with the infarcion or hemorrhage volume. CONCLUSION. Diagnosis and monitoring of neurological complications in critical care diseases is still based on clinical signs, radiological findings and, less frequent, electrophysiological parameters. Serum S-100 protein is a sensitive and specific measure of secondary brain cell damage. It offers potential value mainly in analgo-sedated critical care patients were diagnosis of neurological complication is difficult to establish. REFERENCES. 1. Buttner T, Weyers S, Postert T, Sprengelmeyer R, Kuhn W: S-100 protein: serum marker of focal brain damage after ischaemic territorial MCA infarction. Stroke 28: 1961±1965, 1997. 2. Ingebrigtsen T, Waterloo K, Jacobsen EA, Langbakk B, Romner B: Traumatic brain damage in minor head injury: relation of serum S-100 protein measurements to magnetic resonance imaging and neurobehavioral outcome. Neurosurgery 45: 468±475, 1999. 3. Raabe A, Grolms C, Sorge O, Zimmermann M, Seifert V: Serum S-100B protein in severe head injury. Neurosurgery 45: 477±483, 1999.

INTRODUCTION. The purpose of this study was to investigate how immunoreactivity of nitric oxide(NO)-relating factors were modified under hyperbaric oxygenation (HBO). NO reacts with superoxide radical (O2-) at a very high rate constant to give peroxynitrite, ONOO-. Peroxynitrite, a stronger pro-oxidant, can react with superoxide dismutase (SOD) to form a powerful nitrating agent, resulting in the nitration of tyrosine-residue of cellular proteins and initiation of cellular dysfunction and death. These reactions suggest that induction of neuronal NO synthase and SOD may give rise to formation of nitrotyrosin (NT). The nitration of tyrosine is suggested to be a maker of NO dependent oxidative damage. We examined the effect of HBO on these factors with our reproducible rabbit spinal ischaemia model.

593

Oral Presentations Impact of new technology ± 595±600

ELEVATED INTRACRANIAL INTERLEUKIN-18 LEVELS AFTER CLOSED HEAD INJURY IN HUMANS AND MICE: EVIDENCE OF NEUROPROTECTION BY INTERLEUKIN-18-BINDING-PROTEIN AFTER EXPERIMENTAL HEAD INJURY

Yatsiv I1, Stahel PF2, Novick D3, Rubinstein M3, Dinarello CA4, Morganti-Kossmann MC2, Morganti-Kossmann T2, Shohami E5. 1Pediatric ICU and Pharmacology, 5Department of Pharmacology, Hadassah Hebrew University Medical Center, Jerusalem, 3Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel, 2Division of Research and Division of Trauma Surgery, University Hospital, Zurich, Switzerland, 4Department of Medicine, University of Colorado Health Sciences Center, Denver, USA INTRODUCTION. Pro-inflammatory cytokines such as IL-1*, TNF* and IL-6 are upregulated in the brain within hours after traumatic brain injury (TBI), and were shown to be harmful, mainly at the early post-injury period. IL-18, a recently identified cytokine related to the IL-1 family, is pro-inflammatory cytokine that rises after TBI. IL-18 binding protein (IL-18 bp) is a naturally occurring specific IL-18 inhibitor (1). The present study investigates whether closed head injury (CHI) triggers the release of IL-18 in the brain of experimental animals, in the CSF of neurotrauma patients, and whether the administration of exogenous IL-18 bp confers neuroprotection in a mouse model of CHI. METHODS. CHI was induced in mice using a weight-drop device (2). Tissue and CSF levels of IL-18 were evaluated up to 10 d after trauma using ELISA kit. Clinical evaluation of traumatized mice was done using the Neurological Severity Score (NSS), testing 10 different tasks. One point is given for failing to perform a task. NSS was evaluated at 1 h post CHI (reflecting severity of trauma) and after 1, 3 & 7 days. Once the 1 h NSS was evaluated treated animals were injected with 50 mcg huIL-18 bp i. p. and controls with vehicle Rate of recovery at time (t) is defined by DNSS(t) = NSS(1 h)±NSS(t) (3). Edema formation was evaluated as percent tissue water content. RESULTS. The intrathecal IL-18 levels were significantly elevated in 9/10 CHI patients, as compared to control CSF from 5 patients without trauma or inflammatory neurological disease (p < 0.05; repeated measures ANOVA). In brain-injured mice elevated IL-18 were detected in tissue homogenates, however, only at 7 d post CHI the levels in the contused hemisphere were significantly higher than sham (67.6  5.1 ng/ml vs. 54.3  2.7 ng/ml, respectively p < 0.01). We then tested the effect of IL-18 bp on clinical recovery after CHI in mice. Whereas the initial NSS was similar in treated and control groups, the IL-18 bp-treated mice showed faster recovery as compared to control (DNSS at 24 h 2.0  0.35 vs 1.06  0.21 p < 0.05 and at 7 d: 3.56  0.36 vs 1.55  0.50, p < 0.001, respectively, Mann-Whitney test). No significant effect was observed on the formation of oedema at 24 h after CHI. CONCLUSION. The present findings show that IL-18 levels are elevated both in the CSF of patients and in brain tissue of mice after trauma. We show for the first time a beneficial effect of IL-18 binding protein in traumatic brain injury. We believe that this is a result of inhibition of the early pro-inflammatory effect of IL-18 after CHI. REFERENCES. 1. Novick et al. Immunity 1999;10(1): 127±36. 2. Chen et al. J Neurotrauma 1996;13: 557. 3. Stahel et al. JCBF 2000;20: 369.

METHODS. Thirty Japanese white rabbits weighing 2 to 3 kg were involved in this study. A spinal cord ischaemia model of infrarenal aortic occlusion for 15 min was employed. A balloon-tipped catheter was placed in the abdominal aorta without inflation of the balloon in six out of 30 rabbits as sham control. The other rabbits, randomized into one of two groups, underwent ischaemic insults with the balloon inflation for 15 minutes. One hour HBO at 3 atmospheres absolute with 100 % oxygen at 30 minutes after reperfusion were employed in the rabbits in group A (n = 12) and no treatment in group B (n = 12). All the 24 rabbits were sacrificed at 8, 12, 24, and 48 h after reperfusion in order to harvest the spinal cords. After we made the cross sections of 10 mM thickness at a level between L2 and L3 of the spinal cords, we applied monoclonal anti-SOD and NT antibodies to the cross sections for immunohistochemical studies. RESULTS. Motor neurons in group B were fully labeled for SOD at 8, 12, and 24 h after reperfusion, while SOD-like immunoreactivity of the motor neurons in group A was no more than those in sham group. Motor neurons in group B were fully labeled for NT at 8 and 12 h after reperfusion, whereas motor neurons in group A were slightly labeled. CONCLUSION. HBO reduced the degree of induction of SOD and NT immunoreactivity after transient spinal ischaemia. According to our results, it may be speculated that HBO, if applied at an early stage after ischaemia, has beneficial effect on spinal cord ischaemia to some extent. REFERENCES. 1. Kauer H, Halliwell B: Evidence for nitric oxide-mediated oxidative damage in chronic inflamation. Nitrotyrosine in serium and synovial fluid from rheumaoid atients. 1994; FEBS Lett 350: 9±12 2. Murakami N, Horinouchi T, Sakurai M, et al: Hyperbaric oxygen therapy given 30 minutes after spinal cord ischaemia attenuates selective motor neuron death in rabbits. 2001; Crit Care Med 29: 814±818

595 PREDICTING AND MONITORING FLUID RESPONSIVENESS IN PATIENTS AFTER CARDIAC SURGERY BY ASSESSMENT OF LEFT VENTRICULAR STROKE VOLUME VARIATIONS Reuter DA1, Felbinger TW1, Schmidt C2, Kilger E1, Goetz AE1. 1Department of Anesthesiology, Ludwig-Maximilians-University, Grosshadern University Hospital, 2Technical University,, Munich, Germany INTRODUCTION. Mechanical ventilation causes changes in left ventricular preload leading to distinct variations in left ventricular stroke volume (SVV) and systolic arterial pressure (SPV). Retrospective off-line quantification of SPV has been validated as a sensitive method to predict left ventricular response to volume administration (1). We report of the real-time measurement of SVV by continuous arterial pulse contour analysis to predict and to monitor volume responsiveness in patients after cardiac surgery. METHODS. 20 mechanically ventilated patients were studied immediately after cardiac surgery with cardiopulmonary bypass. Left ventricular stroke volume variations (SVV) were obtained by continuous arterial pulse contour analysis (PiCCO, Pulsion Medical Systems, Germany) before and after volume loading (VL) with oxypolygelatine 3.5 % 20 ml times body mass index. In parallel, SVI (transpulmonary thermodilution) and CVP were measured. Respective values of SPV were analyzed off-line. RESULTS. Hemodynamic data are shown in table 1. Overall, SVV and SPV decreased significantly due to VL and correlated strongly (r = 0.89; p < 0.001). DSVV and DSPV correlated strongly (r = 0.85; p < 0.005). DSVV correlated significantly to changes in stroke volume index (DSVI) (r = 0.67; p < 0.005) as did DSPV (r = 0.56; p < 0.05). Pre-VL SVV correlated significantly with DSVI (R = 0.67; p < 0.005). 13 patients responded to VL with an increase of SVI > 15 % (responder), whereas 7 patients did not (non-responder). Using Receiver Operating Characteristic Curves (ROC), the area under the curve was statistically greater for SVV (0.82; 95 % confidence interval (CI): 0.64±1.0) and SPV (0.81; 95 % CI:0.62±1.0) than for CVP (0.451; CI: 0,17±0,74).

before VL after VL

SVI [ml/m2]

SVV [%]

SPV [mmHg]

CVP [mmHg]

4010 509 p < 0.01

237 113 p < 0.01

167 63 p < 0.01

72 103 p < 0.01

CONCLUSION. Monitoring of SVV enables a real-time prediction and analysis of the left ventricular response to preload enhancement in patients after cardiac surgery. SVV therefore is extremely helpful for guiding volume therapy. REFERENCE. Anesthesiology 1987, 67: 498±502

S 288

14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

596

598

PERFORMANCE OF NEURAL NETWORKS AND MULTILINEAR FITTING METHODS FOR EXTRACTING RESPIRATORY SYSTEM COMPLIANCE

A HEATED SPO 2-PCO 2 SINGLE SENSOR: IMPROVEMENTS COMPARED TO CONVENTIONAL PULSE OXIMETRY AND CAPNOMETRY

Perchiazzi G 1, Vena A 1, Ruggiero L 1, Fiore T 1, Giuliani R 1, Hedenstierna G 2. 1Emergency and Transplantation, Bari University, Bari, ITALY, 2Clinical Physiology, Uppsala University, Uppsala, Sweden

Gisiger PA 1, Marschall A 1, Eberhard P 1, Männle C 2. 1Linde, Medical Sensors AG, Basel, Switzerland, 2Anesthesiology, Thoraxklinik, Heidelberg, Germany

INTRODUCTION. Artificial Neural Networks (ANNs) are computer-based connectionist systems able to extract information from signals, once having been trained to perform this specific task. Multilinear Curve Fitting (MCF), by means of least squares fitting algorithm, can yield the parameters of multiple line equations that can reproduce the behaviour of a system to be fitted. Aim of the present contribution is to compare the performances of ANNs and MCF based methods in assessing the Respiratory System Compliance (CRS) in an animal model of Acute Lung Injury (ALI). METHODS. Eight intubated and anesthetized pigs were ventilated using Volume ControlledConstant Flow Mechanical Ventilation (VC-CFMV). Airways Pressure, Flow and delivered Volume were recorded together at fixed intervals after the induction of ALI, in duplicate recordings. Each tracing consisted of ten or more consecutive breaths in VC-CFMV, followed by a Breath with an Inspiratory Pause (BIP) of more than 2.5 seconds. The BIP was used by the investigators to calculate CRS according to the standard interrupter technique (IT). From the breath preceding the BIP, CRS was extracted by applying separately ANN and MCF methods on the recorded tracings. The ANN had been previously trained on a different pool of animals to perform this task (1,2). Using the IT method as the gold standard, the error of measurement of ANN and MCF methods were separately assessed according to Bland and Altman. Then the Wilcoxon Signed Rank (WSR) test was applied to the two populations of errors in order to define the significance probability that the two error populations had different median (i. e. bias). RESULTS. By using MCF, CRS was assessed with an error of (bias  standard deviation) 2.32  2.43 [ml/cmH2O]; by using ANN, CRS was extracted with an error of ±0.56  1.57 [ml/ cmH2O]. WSR test confirmed that the two biases were significantly different with a p < 0.01. CONCLUSION. These data show that in this experimental setting, ANNs extract CRS with a significantly lower bias and less scatter than MCF-based method. These results indicate new options in the development of monitoring tools for the on-line measurement of respiratory mechanics. REFERENCES. 1. Perchiazzi G, Högman M, Rylander C, Giuliani R, Fiore T, Hedenstierna G ªAssessment of respiratory system mechanics by artificial neural networks: an exploratory studyº J Appl Physiol; 90: 1817±1824, 2001. 2. Perchiazzi G, Ruggiero L, Högman M, Giuliani R, Fiore T, Hedenstierna G ªNeural networks extract static respiratory system compliance without needing to stop respiratory flowº Int Care Med; 26 (S 3): S294, 2000.

INTRODUCTION. Continuous non-invasive measurement of SaO2 and CO2 in adults is conventionally performed by means of pulse oximetry and capnometry. Limitations of these techniques are related, a. o., to low perfusion and movement artifacts for SpO2, resp. to influences of respiratory function and pulmonary circulation for EtCO2. We report here about a new sensor which combines the measurement of arterial oxygen saturation, pulse rate and transcutaneous PCO2: the OxiCarbo sensor. It comprises the elements of an optical pulse oximetry sensor and those of an electrochemical CO2 sensor. The OxiCarbo sensor is heated to 42 C to create local arterialisation of the cutaneous tissue as required for the transcutaneous measurement. Such heating also increases the SpO2 signal strength, thus, improving the performance of the sensor during low perfusion and movement artifacts [1]. METHODS. All measurements were performed in adults. The OxiCarbo sensor was applied to the ear lobe and connected to a pulse oximeter (Kontron 7840) for the SpO2 and pulse rate measurement, and to a transcutaneous blood gas monitor (Linde MicroGas 7650) for the PCO2 measurement (OxiCarboPCO2). Comparative measurements were performed by using conventional pulse oximetry at the finger, conventional TcPCO2 measurement at the arm, arterial blood gas analysis for PaCO2 and capnometry for EtCO2. RESULTS. SpO2: Measurements performed in patients with sleep apnea syndrome showed that the OxiCarbo sensor detects rapid SpO2 variations with a significantly higher sensitivity than a standard nonheated finger sensor. Typical results from one patient were: SpO2 variation = 28 O2%/min with OxiCarbo, = 18 O2%/min with a nonheated finger sensor using the same signal averaging period. Also, the SpO2 measured with the finger sensor is delayed by 20 to 30 s compared to the OxiCarboSpO2 determined at the ear lobe. PCO2: Measurements with 24 patients: N = 24, PaCO2 range 30 to 55 mmHg, TcPCO2 = 1.00PaCO2 + 0.96, SE = 2.98 mmHg, r = 0.91. EtCO2 = 0.61PaCO2 + 8.44, SE = 3.63 mmHg, r = 0.75. The good correlation between TcPCO2 and PaCO2 suggests to use TcPCO2 as reference to evaluate the PCO2 function of the OxiCarbo sensor. Measurements with 6 volunteers gave following RESULTS. N = 24, TcPCO2 range 26 to 43 mmHg, OxiCarboPCO2 = 1.01TcPCO2-0.35, SE = 1.09 mmHg, r = 0.98. CONCLUSION. The integration of a SpO2 sensor element in a heated transcutaneous PCO2 sensor reduces the number of sensors needed to monitor oxygenation and ventilation. The results obtained with the OxiCarbo sensor show that the performance of the SpO2 measurement is significantly improved, especially in situations where the signal-to-noise ratio is low, such as during low perfusion and movements artefacts. The OxiCarbo sensor allows, in general, a more reliable CO2 measurement than capnometry. It can also be applied in situations where capnometry cannot be used. A further advantage is the use of the ear lobe as a centrally located measuring site. REFERENCE. OxiCarbo, a single sensor for the non-invasive measurement of arterial oxygen saturation and CO2 partial pressure at the ear lobe. Gisiger P. A. et al., Sensors and Actuators B, 76/1±3 (2001) 526±529

597

599

INTRA-AORTIC BALLOON PUMPS AND CARDIAC ARRHYTHMIA SPECIFICALLY REDUCE THE PERFORMANCE OF THIRD-GENERATION PULSE OXIMETERS

CONTINUOUS OESOPHAGEAL DOPPLER CARDIAC OUTPUT MONITORING AS A TOOL TO OPTIMIZE VASOACTIVE SUPPORT IN SEPTIC SHOCK

Kroeber S, Urankar S, Lutter N. Department of Anesthesiology, University of ErlangenNuremberg, Erlangen, Germany

De Vaumas C, Lafanech›re A, Kermarrec N, Paugam-Burtz C, Mantz J, Desmonts J M, Beloucif S. Dept of Anesthesiology & Critical Care Medicine, Bichat Hospital, Paris, France

INTRODUCTION. About 100,000 intra-aortic balloon pumps (IABP) are used per year [3] as mechanical supports capable of augmenting the output of a failing heart. Aside from the interference of low perfusion [1] and motion, these cardiac assist devices as well as cardiac arrhythmia affect the arterial pulse waveform and thereby the measurement of the pulsatile absorbance of arterial blood. For this reason, this study was designed in order to determine the reliability of pulse oximeters in presence of IABP and cardiac arrhythmia. METHODS. After informed consent, 42 ICU patients (ASA physical status II-IV), 21 treated with IABP and 21 suffering from cardiac arrhythmia, were connected to three third-generation pulse oximeters simultaneously, each utilizing a specific signal processing method: Agilent Viridia CMS 2000, Nellcor Symphony N-3000 and Masimo SET. Functional oxyhemoglobin saturation (SpO2) and pulse rate were recorded digitally, the alarms were classified by a clinically experienced anesthesiologist into technical/physiological and false/correct [2]. Sensitivity and specificity were calculated as follows: Specificity = TN/[TN + FP], sensitivity = TP/[TP + FN] (TN true negative, TP true positive, FP false positive, FN false negative) RESULTS. With IABP alarms were registered every 3.1 minutes (under conditions of cardiac arrhythmia every 4.3 minutes), and CMS generated the smallest percentage of false positive alarms (35/7.2 %), compared to SET (188/38.9 %) and N-3000 (229/47.4 %). Additionally, false negative events were most frequent with N-3000 (10.0 %; CMS: 8.5 %, SET: 3.7 %). If cardiac arrhythmia was present, however, the highest degree of false positive alarms was found with CMS (96/27.8 %; SET: 78/22.5 %, N-3000: 43/12.4 %). During a total measuring time of 50 h, the time in which data were not available for technical reasons was shortest for SET in both groups (1.0 % with IABP, 1.8 % with cardiac arrhythmia), followed by CMS (6.7 %/5.1 %) and N-3000 (7.9 %/6.7 %). In presence of IABP, the highest specificity was calculated for CMS with 88.7 % (cardiac arrhythmia: 43.6 %), followed by SET with 42.0 % (52.4 %) and N-3000 with 26.1 % (75.0 %). Sensitivity was calculated with 26.8 % (95.2 %) for CMS, 51.9 % (98.0 %) for SET and 14.6 % (100 %) for N-3000. The majority of false positive alarms was associated with ASA physical status IV patients in both study groups. The highest percentage of false positive alarms was found in patients aged 70±79 years in the IABP group and in patients aged 60±69 in presence of cardiac arrhythmia. CONCLUSION. False positive alarms were caused by interference of IABP in 16 (76.2 %) of 21 patients, and by cardiac arrhythmia in 10 (47.6 %) of 21 patients. In the remaining 4, respectively 11 patients, false positive alarms were the result of low perfusion and motion. IABP and cardiac arrhythmia considerabely decrease the overall reliability of third-generation pulse oximeters. However, IABP reduces the clinical performance of all tested pulse oximeters to a significantly larger extent than cardiac arrhythmia. REFERENCES. 1. Ahlborn V et al. (2000) Acta Paediatr, 89(5): 571±6. 2. Lutter N et al. (2001), STA Proceedings: 11. 3. Mehlborn U et al. (1999), Thorac Cardiovasc Surg 47, Suppl 2: 298±303.

INTRODUCTION. The obtention of a reliable and easy-to-use bedside assessement of preload and contractile status would be useful to optimize the indications of vasopressor and inotropic support during septic shock. Continuous cardiac output monitoring using esophageal Doppler (EDM) also provides an insight in preload, via measurements of systolic flow time (FTc), and in contractile status, via measurements of maximum acceleration of blood in descending thoracic aorta (MA, the first derivative of velocity over time: dV/dt) (1). We hypothesized that monitoring of these indexes could be used as a tool to optimize the indications of vasoactive support in the therapeutic management of severe septic shock. METHODS. The aims of this study were to optimize the therapeutic strategy (dose and nature of vasoactive drugs and fluid loading) in 7 patients (54  15 years) receiving vasopressors and inotropic support (targeted according to their blood pressure [BP]) for severe septic shock. Upon ICU arrival, usng Doppler indexes (CardioQ Esophageal Doppler Monitoring, Deltex, Fance) were used to modify the initial therapy with: a) fluid loading when FTc was low, [n = 330±360 msec] b) reduction of inotropic drugs when MA was elevated [n = 10±12 m/sec2] c) reduction of vasoconstrictors and/or increased inotropic support when MA was low. Data (mean  SD) before and after therapeutic optimization were compared using Wilcoxon test. RESULTS. The initial therapeutic support consisted in dobutamine: 17  18 mcg/kg/min, epinephrine: 2  1 mg/hr and norepinephrine: 1.5  0.6 mg/hr. Following optimization these doses were respectively modified by ±50 %, ±61 % and + 3 %, while CO was significantly increased at a constant BP. The associated fluid loading (750  380 ml) induced a significant increase in FTc. Heart rate (HR) was decreased, resulting in an increased stroke volume (SV). (Table: *: p < 0.05 vs ªbeforeº)

Mean blood pressure (mmHg) Heart rate (beat/min) Cardiac output (l/min) Stroke volume (ml) FTc (msec) MA (m/sec2)

Before

After

6813 11216 5.31.7 4913 30133 8.54.6

655 10123 * 72.1 * 7619 * 33628 * 9.14.5

CONCLUSION. The evaluation of preload and of inotropic status using EDM allowed to significantly reduce initially elevated catecholamine doses. This resulted in an increased CO without any deleterious effects on BP,suggesting the existence of an initially inappropriately elevated afterload. EDM appeared useful in the therapeutic decision strategy during septic shock, not only because of cardiac output determination, but also because it could provide a continuous monitoring of cardiac filling and contractility. REFERENCE. Singer. Critical Care Medecine. 1989;17: 447±52

14th Annual Congress ± Geneva, Switzerland ± 30 September±3 October 2001

600 FILTERSURVIVAL TIME DURING PRE- AND POST-DILUTION CONTINOUS VENO-VENOUS HAEMOFILTRATION

Van der Voort PH, Gerritsen RT, Egbers PH, Kuiper MA. Intensive Care, Medical Centre Leeuwarden-Zuid, Leeuwarden, Netherlands INTRODUCTION. Substitution of water and solutes during Continuous Veno-Venous Haemofiltration (CVVH) is mandatory to prevent dehydration. These fluids can be substituted before (pre dilution) or after (post dilution) the passage of blood through the filter. There is a general believe that pre dilution, by haemodilution, prevents clotting in the filter compared to post dilution, which was the subject of this study. METHODS. The study design was prospective and cross-over. After randomisation every patient was consecutively studied in pre- and post dilution CVVH or visa versa. All CVVH sessions were standardized by ultrafiltration rate (3 litres per hour), blood flow (200 ml/min) and filter (cellulose-tri-acetate, 1.6 m2 and cut-off 60.000 D). In all CVVH sessions a bolus of nadroparine (2850IE) was given at the start of the CVVH before the filter and 475 IE per hour by continuous infusion before the filter during the entire CVVH session. A session was ended when the transmembranepressure reached 250 mmHg. The mean filtersurvival time and creatinine clearance in pre- and post dilution CVVH were compared by paired samples T-test. Multiple regression analysis on filtersurvival time was performed with thrombocyte count, leukocyte count, haemoglobin and albumin level, number of transfused blood products, blood pressure and central venous pressure as independent variables. RESULTS. In 16 different patients 32 CVVH sessions were performed. 11 patients were male; the mean age was 69.6 years. The mean APACHE II score was 20. The mean creatinine clearance during pre dilution was significantly lower (p = 0.001) compared to post dilution: 33 ml/ min (range 28 to 39), versus 43 ml/min (range 30 to 48). During pre-dilution the mean filter survival time was 35.1 hours compared to 22.5 hours in post dilution CVVH (p = 0.005). Per session the mean total amount of blood cleared by post dilution was 58.1 litres, compared to 69.3 litres for pre dilution. All other variables were not significantly different between pre dilution and post dilution. All tested variables were not significantly related to filter survival time in pre- or post dilution. CONCLUSION. Pre dilution CVVH resulted in a significantly lower creatinine clearance and a longer filtersurvival time compared to post dilution that could not be explained by other factors.

S 289