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

S 188

Oral Presentations Non-invasive ventilation ± 208±212

210

208

Carlucci A1, Delmastro M1, Rubini F2, Fracchia C2, Nava S1. 1Respiratory Intensive Care Unit, Fondazione S.Maugeri, Pavia, 2Respiratory Intensive Care Unit, Fondazione S.Maugeri, Montescano, Italy

NON-INVASIVE MECHANICAL VENTILATION (NIMV): AN ANALYSIS OF THE TRAINING EFFECT

AIR VERSUS HELIUM NON-INVASIVE PRESSURE SUPPORT IN ACUTE COPD DECOMPENSATION: A PROSPECTIVE, RANDOMIZED, STUDY Jolliet P1, Tassaux D1, Roeseler J2, Burdet L3, Broccard A3, D'Hoore W2, Borst F1, Raeynaert M2, Schaller MD3, Chevrolet JC1. 1Medical ICU, University Hospital, Geneva, Switzerland, 3Medical ICU, University Hospital, Lausanne, Switzerland, 2ICU Dpt., St.-Luc Hospital, Brussels, Belgium INTRODUCTION. In decompensated COPD, noninvasive pressure support ventilation (NPSV) with helium/O2 (He/O2) can reduce dyspnea, PaCO2, and work of breathing more than NPSV with Air/ O21,2. However, it is unknown whether these effects can have beneficial consequences on outcome, and what the financial implications in routine practice are, due to He/O2's high cost. This study aimed to explore these aspects. METHODS. All patients. with COPD, admitted to the ICU for NPSV over a 24 month period were included. At ICU admission, patients. were randomized to Air/O2 or He/O2. Initial settings: pressure support 15 cmH20, FiO2 0.25, PEEP 5 cmH2O. Subsequent setting changes, number of daily trials performed, decision to intubate, and ICU discharge criteria followed our standard practice guidelines. RESULTS. (mean  SD): 123 patients. (M/F 71/52, age 71  10 yr, Apache II 17  4) admission values: respiratory rate 28  8 /min., pH 7.31  0.07, PaO2/FiO2 232  102, PaCO2 9  2 kPa. No baseline difference nor different initial response to NPSV, complications or mortality was noted between Air/O2 and He/O2 groups. NPSV n trials/ duration min. Intubation n patients/% LOS ICU days LOS hospital days Cost/pt. total NPSV gases eur. Cost/pt. (incl. NPSV gas) ICU stay eur. Cost/pt. hospital stay eur. Cost/pt. (incl. NPSV gas) entire stay eur.

Air/O2 (n = 64)

He/O2 (n = 59)

9.2 (7.6)/100 (182) 13/20.3 6.2 (5.6) 19 (12) 1.8 (2.4) 11,373 (10,367) 14,649 (9,457) 26,023 (16,485)

9.7 (7.4)/102 (228) 8/13.5 5.1 (4) 13 (6)* 78 (56)* 9,530 (7,463) 10,320 (5,089)* 19,851 (10,322)*

INTRODUCTION. The opening in '92 of a Respiratory Intermediate Care Unit (5 beds) specifically dedicated to NIMV offers us the unique opportunity to evaluate the time effect of the progressive training with NIMV of medical and paramedical staff, on the patient's outcome. METHODS. To avoid possible confounding effects, our analysis was focused only on COPD patients. 208 episodes of Acute Respiratory Failure (ARF) were analysed. RESULTS. The rate of NIMV failure (death and intubation) progressively declined in the first 5 years from 29 % to 9 %. From '96 this rate was stabilised at around 17 %. A statistical analysis (change point test) allowed us to locate in '97 a significant change in the severity of admission pH (7.25  0.07 vs 7.20  0.08 for the years 92±96 and 97±99, respectively; p < 0.0001), while no differences were observed in the rate of failure (18 % vs 16 %). In the first 5 years period there was a significant difference in pH at admission between failure and success (7.21  0.06 vs 7.26  0.07; p < 0.003), but this was not the case after 96' (7.18  0.10 vs 7.21  0.08). CONCLUSION. We conclude that the progressive learning and familiarity with NIMV of a dedicated team may 1) improve with time the patients' outcome 2) progressively allow to successfully treat more severe patients.

Mean (SD) * p < 0.002 vs. Air/O2 CONCLUSION. With a similar NPSV approach, He/O2 did not significantly reduce the intubation rate. However, hospital stay was shorter with He/O2. NPSV gas cost was higher with He/O2, but represents a small fraction of ICU costs. Hence, hospital stay and total costs were lower with He/O2. Further studies in decompensated COPD should be conducted to better define the place of He/O2 NPSV, as it can be safely administered and could prove to be a cost-effective strategy. REFERENCES. 1.Crit Care Med 1999;27: 2422±2429; 2. AJRCCM 2000;161: 1191±1200.

209

211

HELIOX DURING NON-INVASIVE MECHANICAL VENTILATION IN HYPERCAPNIC COPD EXACERBATION

NON-INVASIVE PRESSURE SUPPORT VENTILATION DELIVERED BY HELMET AS A TREATMENT OF ACUTE HYPOXEMIC RESPIRATORY FAILURE

Esquinas AA1, Gonzalez G1, Carrillo A1, Pµrraga MJ1, Gil B1, Gil JI1. 1Intensive Care Unit, H. Universitario Jose Maria Morales Meseguer., Murcia, Spain INTRODUCTION. Non-invasive mechanical ventilation (NIMV) with gas heliox (HE) have been recommended in acute COPD hypercapnic (1). METHODS. Aims: to compare effects during NIMV with two gases: heliox (HE) and oxygenair (O2-Air) in hypercapnic COPD exacerbation. Method: Face NIMV with BiPAP Vision (Respir, Inc.). Randomizated controlled method to apply gas (HE) or (O2-Air). RESULTS. Fifty COPD were enrolled. Dates at baseline (t = 0), 30 minutes (t = 1) and end therapy (t = 2) were analysed in 25 in two groups. Gas heliox mixtures: 72:28; UCI stay: 3.56  1 days (Heliox group) and 6.41  9.9 days (Air group) (p < 0.8). Success NIMV: Heliox 25/25, Air-Oxygen = 22/25. Skin no lesion 2/25 (heliox group) y 3/25 (Air group).

Age IPAP mmHg pH-t = 0 pH-t = 1 pH-t = 2 PaCO2-t = 0 PaCO2-t = 1 PaCO2-t = 2 SAPS II EPAP mmHg PaO2-t = 0 PaO2-t = 1 RR-t = 0 RR-t = 1 RR-t = 2 Hours NIMV

1 = HELIOX

2 = O-AIR

p ( < 0.05)

72.610 173 7.260.1 7.300.02 7.350.1 79.6527 77.112 60.920

74.48 163 7.220.2 7.210.03 7.370.2 78.616.2 96.3546 64.820

ns ns ns 0.08 ns ns ns ns

1 = HELIOX

2 = O-AIR

p ( < 0.05)

31.454.12 81 61.3222 91.924 348 293 242 17.1232

35.28.7 81 7038 10831.7 3212 302 231 17.326

0.02 ns ns ns ns ns ns ns

CONCLUSION. Preliminary dates of a open pilot randomizated controlled study are: short UCI stay (.08) and improve acidosis/hipercapnia at t = 1 (.08) in NIMV- HE group were found at not significant levels. REFERENCE. Jaber S et al. Noninvasive mechanical ventilation with helium: oxygen in acute exaceration of chronic obstructive pulmonary disease. AJRCCM. 161: 1191±1200.

Antonelli M1, Conti G1, Pelosi P2, Gregoretti C3, Pennisi MA1, Costa R1, Recchioni G1, D'Amato L1, Munafò C1, Conti M1, Severgnini P2, Chiaranda M2. 1Istituto di Anestesiologia e Rianimazione, Università Cattolica del Sacro Cuore, Rome, Italy, 2Dipartimento di Scienze Cliniche e Biologiche, Università dell'Insubria, Varese, Italy, 3Servizio di Anestesia e Rianimazione, Ospedale CTO, Torino, Italy INTRODUCTION. To assess the efficacy of noninvasive pressure support ventilation (NPSV) using a new special helmet as first line intervention to treat patients with hypoxemic acute respiratory failure (ARF), in comparison to NPSV using standard facial mask. Results of a prospective clinical pilot investigation with matched control group in three intensive care units of university hospitals are reported. METHODS. Thirty-three consecutive non-COPD patients with hypoxemic ARF (defined as severe dyspnea at rest, respiratory rate > 30 breaths/min, PaO2:FiO2 < 200, and active contraction of the accessory muscles of respiration) were enrolled. Each patient treated with NPSV by helmet was matched with two controls with ARF treated with NPSV via a facial mask, selected by SAPS II, age, PaO2/FiO2, and arterial pH on admission. Primary end points were the improvement of gas exchanges, the need for endotracheal intubation and the complications related to NPSV. RESULTS. The 33 patients and the 66 controls had similar characteristics at baseline. Both groups improved oxygenation after NPSV. Eight patients (33 %) in the helmet group and 21 patients (32 %) in the facial mask group (P = 0.3) failed NPSV and were intubated. No patients failed NPSV due to intolerance of the technique in the helmet group in comparison with 8 patients (38 %) in the mask group (P = 0.047). Complications related to the technique (skin necrosis and gastric distension) were fewer in the helmet group compared to the mask group (no patients versus 10 patients(15 %), P = 0.013). Helmet allowed the continuous application of NPSV for a longer period of time (P = 0.05). Length of stay in the ICU, intensive care and hospital mortality were not different. CONCLUSION. NPSV by helmet successfully treated hypoxemic ARF, with better tolerance and fewer complications than facial mask NPSV. REFERENCE. Antonelli M; Conti G; Rocco M; Bufi M; De Blasi RA; Vivino G; Gasparetto A; Meduri GU: A comparison of noninvasive positive-pressure ventilation and conventional mechanical ventilation in patients with acute respiratory failure. N Engl J Med. 1998 Aug 13; 339(7): 429±35

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

S 189

212

214

CPAP VERSUS NON-INVASIVE PRESSURE SUPPORT VENTILATION IN THE TREATMENT OF POSTOPERATIVE ATELECTASIS

HEPARIN VERSUS RECOMBINANT HIRUDIN FOR ANTICOAGULATION IN CONTINUOUS RENAL REPLACEMENT THERAPY

Pasquina PH, Granier JM, Merlani PG*, Ricou B*. Respiratory Therapy and *Div. of Surgical Intensive Care, Dept. of Anesthesiology, Pharmacology and Surgical Intensive Care, University Hospital, Geneva, Switzerland

Vargas Hein O, Von Heymann H, Diehl T, Kox W, Spies C. Anesthesiology and Intensive Care, University Hospital Charite, Campus Mitte, Berlin, Germany

INTRODUCTION. Incidence of atelectasis (ATLS) after cardiovascular or abdominal surgery is as high as 47±92 %. Continuous positive airway pressure (CPAP) is widely used to prevent or treat postoperative ATLS. It is reported to improve oxygenation but its effect on the course of ATLS is not clear. Non invasive pressure support ventilation (NIPSV) improves tidal volume (Vt), gas exchange, respiratory rate and diaphragmatic activity [1]. The efficacy of NIPSV to resolve postoperative ATLS has never been studied. We hypothesized that it could be more effective than CPAP in this indication. METHODS. 182 patients(pat)presenting a X-ray ATLS score > 1 (0 = clear lungfields, 1 = plate-like ATLS or slight infiltration, 2 = partial ATLS, 3 = lobar ATLS, 4 = bilateral lobar ATLS)[2] after cardiovascular or major abdominal surgery in our surgical intensive care unit (SICU) were randomly assigned to CPAP or NIPSV. Pat.were treated 4 times a day during 30 min. with either CPAP with a positive end expiratory pressure (PEEP) of 5 cm H2O or NIPSV with a pressure support to reach a Vt of 8±10 ml/kg body weight and a PEEP of 5 cm H2O. The ATLS score, arterial blood gases, pulmonary function tests, adverse events and the degree of comfort by visual analogue scale were assessed by two SICU specialists daily from study entry (T0) to SICU discharge (TD) at 9 am, 4 hours after the last respiratory treatment. RESULTS. 147 pat.were analyzed after secondary exclusion of 35 pat. (19 in the CPAP and 16 in the NIPSV group). The reasons of exclusion and characteristics of these pat.were comparable. At T0, demographic, preoperatory and surgery data were similar in the two groups. At TD, no difference in pulmonary function tests, degree of comfort, adverse events and SICU length of stay was observed between the groups.

INTRODUCTION. Heparin as standard for anticoagulation in continuous renal replacement therapy (CRRT)has a bleeding incidence of 20 % and is contraindicated in patients with heparin induced thrombocytopenia type II (1). Recently, continuously applicated recombinant hirudin (rhirudin) has been reported to be as effective as heparin for CRRT (2). However, bleeding complications were observed only in the rhirudin group. The aim of the study was to compare continuous application of heparin and intermittent application of rhirudin as anticoagulants in CRRT with respect to haemofiltration efficacy, described as filter run time, and possible bleeding complications. METHODS. After ethical committe approval and written informed consent from the legal representatives/relatives 27 critically ill patients with an indication for CRRT were randomly allocated to 2 groups: heparin: 15 patients, initial dose 250 IU/h; aim activated clotting time (ACT) 180±210 s, 125 IU/h stepwise heparin dose change; rhirudin: 12 patients, initial dose 5 mg/kg bolus and consequently 2 mg/kg bolus to achieve an aimed ecarin clotting time (ECT) of 80±100 s. Every four hours prothrombin time (PT), thromboplastin time (PTT), hemoglobin (Hb) and thrombocytes were determined. A bleeding complication was defined as an Hb decrease of more than 2 g/dL with clinical bleeding signs. The observation time was 96 hours. the filter run time of clotted filters were recorded. Statistical analysis: Mann-Whitney-U-Test and Fisher exact test. RESULTS. The results are presented in table 1

Radiological response (T0 to TD) D Radiological score (Difference T0 vs TD) PaO2/FiO2

Number of patients n(%) Score median (25th-75th) (mmHg mean  SD)

CPAP (n = 72)

NIPSV (n = 75)

Worse

6(8)

1(1)

No Change Improvement

37(52) 29(40)

30(40) 44(59)*

0(0±1)

1(0±1)#

Preoperatory

34164

35079

median (range)

T0 TD

28458 28140

29564 30740§

CONCLUSION. Rhirudin is as effective as heparin regarding filter run time in CRRT. Regarding bleeding complications, the intermittent application of rhirudin seems to be safer than the continuous application in CRRT (2). REFERENCES. 1. Vanholder R, et al; Kidney Int 1994, Vol. 45; 1754±1759. 2. Vargas Hein O, et al; Int Care Med 2001, Vol. 27, 4; 673±679

#

§

CPAP vs NIPSV: * p < 0.05 by Fisher's exact test, p < 0.05 by Mann Withney's test, p < 0.05 by ANOVA CONCLUSION. NIPSV is a safe and well tolerated treatment, superior to CPAP regarding improvement of ATLS based on X-ray score and PaO2/FiO2 after cardiovascular or abdominal surgery.The clinical impact of such treatment should be further investigated. REFERENCE. Appendini L, Am J Res Crit Care Med,'94 [2]Richter Larsen K, Intensive Care Med, 95

APACHE III Age (y) ICU-stay (d) Filter run time (h) Bleeding complications (n)

Heparin

RHirudin

p

71 72 11 13 (4±63) 2

84 72 10 11 (4±30) 0

0.1 0.78 0.83 0.18 0.49

Oral Presentations Nephrology ± 213±217

215

213

Rotelli C, Maggiore U, Melfa L, Capp› G, Sagripanti S, Cardarelli F, Lombardi M, Fiaccadori E. Internal Medicine & Nephrology, Parma University Medical School, Parma, ITALY

IMPACT OF MALNUTRITION ON THE PERFORMANCE OF THE APACHE II SEVERITY-OF-ILLNESS SCORING SYSTEM IN ACUTE RENAL FAILURE

ISCHAEMIC HEPATITIS IN A MEDICAL-SURGICAL ICU Cuadra Madrid S1, Suarez F1, Gonzµlez P1, PØrez C1, L. SASTRE1, M. A. Alcala1, M. Corrales1, F. Picouto1. 1intensive Care Unit, Fundacion Jimenez Diaz, Madrid, Spain INTRODUCTION. Ischaemic Hepatitis (IH) is an acute liver perfusion disturbance associated to shock states that causes a marked and transient elevation in the plasma aminotransferases activity following a typical pattern. Generally not related to previous hepatic disease, predisposing conditions such as cardiac or chronic respiratory disease are almost invariably present. METHODS. Retrospective review of the medical records of patients with a clinical history compatible with IH, admitted to the medical-surgical Intensive Care Unit between March 97 and March 2000. The diagnosis was established according to clinical and laboratory criteria: a) Transient elevation of serum Aspartate Aminotransferase (AST) and Alanine Aminotransferase (ALT) more than ten times the upper normal limit, b) normalization of enzymatic levels within the following days and c) exclusion of other known causes of acute liver injury in patients with an episode of shock or cardiac arrest or severe cardiac-respiratory failure. RESULTS. We identified 65 patients (38 males and 27 females) with IH among the 3735 admissions during the analyzed period (1.7 % incidence). Median age was 67 years (range 30±84), the APACHE II score on admission was 18  8 with a SAPS II of 41  22 (both expressed as mean  SD). Median Length of stay in the ICU was 4 days (range 1±41) and 33 (51 %) patients died. In 50 (77 %) patients a shock episode was identified,15 (23 %) suffered a cardiac arrest, 11(17 %) had an acute exacerbation of a chronic obstructive pulmonary disease (COPD) and in 33(51 %) hypoxemia was documented. Laboratory findings and predisposing chronic underlying conditions are shown in table. Laboratory data

Median (Range)

ALT (U/L) AST (U/L) Bilirubin T(mg/dL)

1603 (527±5889) 1691 (280±9901) 2.56 (0.37±20)

INR

2.33 (1.24±7.54)

Creatinine (mg/dL) BUN (mg/dL)

2.2 (0.6±9.3) 55,5 (20±130)

Predisposing condition CHF NYHA IV Cardiac Surgery Pulmonary Hypertension Previous MI COPD

N of patients(%)

INTRODUCTION. Although preexisting severe malnutrition has been shown to be an independent predictor of mortality in acute renal failure (ARF) (1), currently used prognostic scoring systems do not take into account this factor. We tested the hypothesis that poor nutritional status impairs the predictive ability of the APACHE II model (Acute Physiology and Chronic Health Evaluation, version II)(2) in ARF. METHODS. We prospectively studied all patients admitted for ARF to the Internal Medicine and Nephrology Department over a seven-year period (Jan. 1994 to Dec. 2000): 614 patients (399 males), median age 71.5 years (interquartile range 60  78); APACHE II 22 (18  28); serum creatinine 5.1 mg/dl (3.3  7.1); blood urea nitrogen 80 mg/dL, 58  109); acute tubular necrosis in 421/614 (68.6 %); hemodialysis or continous venovenous hemofiltration in 374/614, (60.9 %). Nutritional status was evaluated at admission by the Subjective Global Assessment (SGA) method (3), and patients were divided into three classes: normal nutritional status (SGA Class A: 247/614, 40.2 %), at risk of malnutrition or moderately malnourished (SGA Class B, 146/614, 23.8 %), severely malnourished (SGA Class C, 221/614, 36 %). Risk of death was predicted for each patient using the original APACHE II equations (2). Calibration of the models was analyzed by the Hosmer-Lemeshow (L-H) ªCº statistic (4), discrimination by the Receiver Operating Characteristic (ROC) area (5). RESULTS. Data on outcome prediction performance are illustrated in the Table

31(48 %) 21 (32 %) 28 (43 %)

Observed mortality Predicted mortality ROC curve area L ± H ªCº statistic

21 (32 %)

* p .034 vs SGA Class A + B

13 (20 %) 23 (35 %)

Laboratory data represent peak values. BUN: Blood urea nitrogen. Predisposing condition. CHF: Chronic heart failure, MI: Myocardial infarction. NYHA:New York Heart Association functional class.COPD: Chronic pulmonary obstructive disease . CONCLUSION. IH was related to previous severe cardiac or respiratory disease in all patients of our series. It seems that chronic hepatic venous congestion is an important predisposing condition to develop IH during severe hypotension, hypoxemia or shock episodes. The intensivist should always be aware of this complication when these conditions coexist in a patient. . Mortality was in agreement with the predicted mortality of the APACHE II and SAPS II scores. REFERENCE. S. Fuchs et al. Isquemic hepatitis clinical and laboratory observations of 34 patients. J Clin Gastroenterol 1998; 26 (3) 183±6.

SGA Classes A + B

SGA Class C

27.5 % (108/393) 31.7 % 0.79 (95 % CI .74  0.84) 9.04 (p 0.54)

53.3 % (118/221) 40.5 % 0.70* (95 % CI 0.63  0.77) 27.6 (p 0.002)

CONCLUSION. Predictive performance of the APACHE II model is significantly made worse in the presence of severe malnutrition. Further studies are needed to verify whether the predictive performance of prognostic scoring systems improves when severe malnutrition is included in the model by customization. REFERENCES. 1. Fiaccadori E et al., J Am Soc Nephrol 1999; 10: 581±93; 2. Knaus WA et al., Crit Care Med 1985; 13: 818±29; 3. Baker JP et al., New Engl J Med 1982; 306: 969±72; 4. Lemeshow L et al. Am J Epidemiol 1982; 115: 92±106; 5. Hanley J et al. Radiology 1982 143: 29±36.

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

S 190

216 LACTATE METABOLISM DURING HAEMOFILTRATION WITH EITHER BICARBONATE OR LACATE AS REPLACEMENT FLUID

Leverve XM1, Novak I2, Krouzicky A2, Roth H1, Rokyta R2, Matìjovi› M2. 1LBFA, University J. Fourier, Grenoble, France, 2ICU, 1 st Medical Department,, Charles-University-Hosp k, Plzen, Czech-republic INTRODUCTION. Lactate is commonly used as fluid for replacement during hemofiltration since its metabolism results to an alkalinization. But lactate is also a useful substrate permitting to carry energy and reducing equivalent among cells and organs. The use of exogenous lactate as replacement fluid may interfere with its metabolism, therefore the aim of this study was to investigate lactate metabolism in acutely ill patients undergoing hemofiltration. METHODS. Seven patients (4M/3F, age = 51.7  7 y, 18±68) hospitalized in intensive care unit (APACHE II = 25  2.6, 16±37) and were assigned to receive at one day interval and in a random order bicarbonate (BICAR) or lactate (LAC) as fluid replacement during hemofiltration. Lactate metabolism was investigated in both situations by using an exogenous lactate challenge test (1): 2.5 mmol/kg body weight were infused during 15 min and plasma lactate concentrations (L) in the blood and in the ultrafiltrate were followed at T = 0, 4, 8, 12, 16, 20, 24, 30, 60, 90, 120 and 150 min after the end of the perfusion. Lactate clearance (LC), total lactate turnover (endogenous plus exogenous rates, TLT), lactate release in the ultrafiltrate (LR), endogenous lactate production (ELP), metabolized lactate (ML) were calculated from basal lactate and from the area under the curve calculated graphically from the unadjusted experimental points by using Kaleidagraph (Abelbeck Software, Reading, PA, USA). Results are given as means  sem, statistical comparisons were achieved by using a paired student's t test, * indicating significant difference, p < 0.01. RESULTS. Ultrafiltration rate was similar in BICAR and LAC (1418  133 vs 1448  137 ml/ hour). During BICAR, bicarbonate infusion rate was 17  12 mmol/kg/min while patients received 10  1.3 mmol/kg/min of lactate during LAC.

BICAR LAC

Basal L mM

Peak L mM

LC ml/kg. min

LTL mmol/kg. min

LR mmol/kg. min

ELP mmol/kg. min

ML mmol/kg. min

1.50.1 2.40.1*

7.50.2 8.20.3

10.31.2 10.40.9

15.01.9 24.61.2*

1.70.2 4.10.6*

15.01.9 14.41.4

13.32.0 20.62.4*

Oral Presentations Cardiovascular dynamics ± 218±222 218 FENOLDOPAM, BUT NOT DOPAMINE SELECTIVELY INCREASES THE MICROVASCULAR OXYGENATION OF THE GASTRIC MUCOSA IN DOGS

Schwarte LA, Schindler AW, Picker O, Fournell A, Scheeren TW. Department of Anaesthesiology, Heinrich-Heine-University, Duesseldorf, Duesseldorf, Germany INTRODUCTION. Dopamine (DOP) is used to selectively increase O2-delivery to the splanchnic region, but identical, low DOP-dosages result in inconsistent effects on gastric mucosal oxygenation (1). This might be explained by unpredictable interaction between DA1, b1, and a1-receptor-mediated effects. Thereby an a1-mediated vasoconstriction may counteract the desired DA1-mediated vasodilatation and thus prevent an increase of the microvascular oxygenation. Therefore, we examined the role of this a1-agonism for the inconsistent effects of DOP on the gastric microvascular oxygenation and compared its effects with those of fenoldopam (FEN), a novel selective DA1-receptor-agonist without a1-activity. METHODS. 6 chronically instrumented dogs (flowprobes around the pulmonary artery for continuous cardiac output measurement) were repeatedly anaesthetised with permission of the local district government to study the relation between microvascular oxygenation of the gastric mucosa (mHbO2, tissue spectrophotometry) and systemic O2-delivery (DO2). The following interventions were performed: DOP 2.5/5.0 mg/kg/min alone or after a1-blockade (prazosin), FEN 0.1/1.0 mg/kg/min alone or after DA1-blockade (SCH-23390, to exclude DA1-independent FEN-effects), and a control group (saline). The results are given as means  sem. Statistics: ANOVA for repeated measurements, p < 0.05. RESULTS. DOP (alone and under a1-blockade) had on average no effect on mHbO2 (56.1 vs. 58.2 %, p = 0.22 and 57.8 vs. 57.6 %, p = 0.75), but increased markedly and dose-dependently DO2 (from 14.4 to 18.7 ml/kg/min, p < 0.05 and from 14.5 to 18.4 ml/kg/min, p < 0.05). In contrast, FEN increased dose-dependently mHbO2 (from 49.7 to 59.8 %, p < 0.05) without changes of DO2 (19.0 vs. 20.3 ml/kg/min, p = 0.091). The effect of FEN on mHbO2 was prevented by DA1-blockade.

CONCLUSION. Basal plasma lactate concentration was significantly increased during lactate replacement as compared to bicarbonate but lactate clearance and endogenous lactate production were not affected. In addition, despite a significant increase in lactate release in the ultrafiltrate, total lactate turnover and metabolized lactate were significantly increased by lactate therapy as compared to bicarbonate period. These results indicate that exogenous lactate supplied by the fluid replacement is metabolized on top of endogenous lactate, which is not affected. REFERENCE. Ann Surg 1999;229: 505±13.

CONCLUSION. Despite marked increase of systemic O2-transport dopamine did not improve oxygenation of the gastric mucosa. This lacking effect of dopamine was not caused by an a1-mediated vasoconstriction. In contrast, fenoldopam improved the oxygenation of the gastric mucosa without marked increase of the systemic O2-transport, why this constellation might be termed as ªsplanchnic-selectiveº.

217

219

EFFECT OF THE MODE OF DELIVERY ON THE EFFICACY OF PROSTACYCLIN AS AN ANTICOAGULANT CONTINUOUS VENO-VENOUS HAEMOFILTRATION

ENDOTHELIN-1 PRODUCTION IN HUMAN AORTIC ENDOTHELIAL CELLS STIMULATED BY CEREBROSPINAL FLUID OF PATIENTS WITH SUBARACHNOID HAEMORRHAGE

O'Callaghan G, Slater M, Auzinger G, Wendon J. Liver Intensive Care Unit, Kings College Hospital, London, UK

Luciana M1, Wener A2, Ward M2, Ranieri V M3, Stewart D J4, Wallace M C5. 11DETO, Sezione di Anestesiologia e Rianimazione, University of Bari, Policlinico Hospital, Bari, Bari, 3Dept Anesthesia, University of Pisa, Pisa, Italy, 2Dept of Respirology, 5Dept Neurosurgery, 4Dept of Cardiology, University of Toronto, Toronto, Canada

INTRODUCTION. Prostacyclin is a short-acting inhibitor of surface and shear stress induced platelet activation and is frequently used as a method of anticoagulation in continuous renal replacement therapies[1]. Where there is a coexistent coagulopathy or thrombocytopenia it is often the sole anticoagulant. Prostacyclin is absorbed onto the fibres of biocompatible membranes and therefore pre-filter administration may result in less platelet inactivation than if the drug were administered as a systemic intravenous infusion. This may in turn lead to shorter filter life and more pronounced thrombocytopenia. In order to address this question we have prospectively compared filter life and platelet count for both pre-filter and systemic intravenous administration of prostacyclin. METHODS. Over a four month period we studied a total of 142 haemofiltration episodes in 16 critically ill patients ( median Apache II score 13, interquartile range 14.5) with hepatic dysfunction and platelet counts of less than 50 x 109/L . Anticoagulation was provided with an infusion of epoprostenol 5 ng/kg/min (Glaxo Wellcome, Middlesex, UK) and each filter episode was randomised for the epoprostenol to be delivered either pre-filter or systemically. Patients did not receive any other anticoagulants or antithrombotics during the course of the study. Patients with proven heparin induced thrombocytopenic syndrome were not included in the study. Both groups were treated with hollow fibre polysulfone haemofilters (Baxter Healthcare Corporation, Irvine, CA, USA), predilutional fluid replacement, 2 L/h exchange volume with a blood flow rate of 180 mls/min. Comparisons between groups were made using the Mann-Whitney U Test. RESULTS. There were 79 filter episodes in the pre-filter group and 63 in the systemic intravenous group. The median platelet count per filter episode was 34 x 109 /L in the pre-filter group and 45 x 109/L in the systemic group, P = 0.1495. The median filter life was 780 minutes in both groups, P = 1.00. CONCLUSION. Systemic administration of prostacyclin in the setting of continuous venovenous haemofiltration does not result in prolonged filter life or a significant improvement in thrombocytopenia in critically ill patients REFERENCE. Kozek-Langenecker S, Kettner S, Oismueller C, Gonano C, Speiser W, Zimpfer M. Anticoagulation with prostaglandin E1 and unfractionated heparin during continuous venovenous haemofiltration Crit Care Med 1998;26:1208±1212

REFERENCE. Scheeren et al. (1999) Intensive Care Med 25 S1: 21 A68

INTRODUCTION. Endothelin-1 (ET-1) has been showed to be increased in cerebrospinal fluid (CSF) of patients which develop cerebral vasospasm or cerebral ischaemia. However it is not clear if it represents a marker or a mediator for vasospasm. We investigated the effects of CSF collected from 20 patients with subarachnoid hemorrhage (SAH) due to rupture of cerebral aneurysm on ET-1 production from human aortic endothelial cells (HAEC). METHODS. Patients were classified according to the angiographic evidence for vasospasm (day 7 post-hemorrhage) and a daily neurological evaluation (Glasgow Coma Score = GCS): no vasospasm (NV), angiographic vasospasm (AV = mild), clinical vasospasm (CV = severe), low GCS (neurological deterioration not due to vasospasm). CSF samples were collected daily, centrifuged at 3000 G for 5 minutes, and stored at ±80 C. HAEC (80±90 % confluence) were incubated for 6 hours with CSF from patients on day 4 post-hemorrhage (when vasospasm started to occur) or from normal subjects (control group). ET-1 was measured in the medium (ELISA) and normalised for total cell protein. RESULTS. Patients were equally distributed according to the presence of vasospasm and other neurological sequelae. ET-1 levels in HAEC supernatants were 1.67  0.32, 1.66  0.54, 1.88  0.49, 1.76  0.45 and 2.165  0.79 fmol/mg (ET-1/tot prot, mean  SD) in cells treated with CSF from control, NV, AV, CV and low GCS group respectively. Only the low GCS group showed a significant increase in ET-1 production compared to control (ANOVA p < 0.05). CONCLUSION. ET-1 production was significantly increased only in HAEC treated with CSF from low GCS patients who had already developed cerebral ischaemia on day 4 post-hemorrhage. However, there was no significant difference in ET-1 production in HAEC stimulated with CSF from patients with various degrees of vasospasm. These results suggest that ET-1 production may play a role in the early stages of neurological deterioration, and may be not responsible for the development of cerebral vasospasm in SAH patients. REFERENCE(S). 1- Ziv I, Fleminger G, Djaldetti R, Achiron A, Melamed E, Sokolovsky M Increased plasma endothelin-1 in acute ischaemic stroke Stroke 1992 Jul;23(7): 1014±6. 2- Seifert V, Loffler BM, Zimmermann M, Roux S, Stolke D Endothelin concentrations in patients with aneurysmal subarachnoid hemorrhage. Correlation with cerebral vasospasm, delayed ischaemic neurological deficits, and volume of hematoma. J Neurosurg 1995 Jan;82(1): 55±62. 3L Mascia, DJ Stewart, F Mohamed, L. Fedorko, K. terBrugge, V. M.Ranieri, M .C. Wallace Temporal relationship between endothelin-1 expression and cerebral vasospasm in patients with aneurysmal subarachnoid hemorrhage Stroke 2001; 32: 1185±1190

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

S 191

220

222

IS THE CARDIORESPIRATORY INTERPLAY IN PATIENTS WITH MODS DIMINISHED IN COMPARISON WITH YOUNG HEALTHY CONTROLS?

PRIMARY SUCCESS RATE OF DIRECT CURRENT CARDIOVERSION IN SURGICAL INTENSIVE CARE PATIENTS

Schmidt HB, Nuding S, Hoffmann T, Kuhn C, Loppnow H, Rauchhaus M, Werdan K. Department of Medicine III, Martin-Luther-University Halle-Wittenberg, Halle, Germany

Mayr A, Ritsch N, H. Knotzer, M. Dünser, W. Pajk, W. Schobersberger, W. Hasibeder. Department of Anaesthesia and Critical Care Medicine, Division of General and Surgical Intensive Care Medicine, Innsbruck, Austria

INTRODUCTION. According to the hypothesis of ªUncoupling of biological oscillatorsº (1), autonomic dysfunction is one of the main determinants exaggerating MODS. We have recently shown that the cardiorespiratory interplay (cardiac chemoreflex sensitivity [CCRS]) is the more blunted the more MODS is pronounced, but to date, there is no data available regarding the normal range values of this parameter. Our aim was to characterise CCRS in young healthy controls and to compare it with values obtained in MODS patients. METHODS. We included 46 consecutive patients with septic (Sepsis Score according to Elebute & Stoner [SepSco] > or = 12) and nonseptic MODS. APACHE II Score (AP II) > or = 20 and < 30 was defined as moderate MODS, AP II > or = 30 as severe MODS. Additionally, we enrolled 13 young volunteers (4 female/9 male) without any known cardiopulmonary diseases as controls. The CCRS was calculated as the regression slope of heart interval (HI) and PaO2. HI and PaO2 were assessed at baseline, after a 1/3 increase of FiO2, and after returning to baseline (2). RESULTS. The table summarises the results of our study (mean  SD). $p = 0.001 (3 vs. 2),*p = 0.002 (3 vs. 1), + p = 0.007 (3 vs. 2), §p = 0.001 (2 vs.1 / 3 vs. 1)

1 Controls 2 Moderate MODS 3 Severe MODS

AP II

SepSco

244 365$

115 164$

CCRS (ms/mmHg)

Age (years)

Weight (kg)

n

1.00.4 0.90.6 0.40.3* +

263 6212§ 6611§

709 8415 7826

13 22 24

CONCLUSION. The cardiorespiratory interaction is depressed in MODS patients compared with young healthy controls. Restoring of autonomic function may be a promising option for future therapeutic intervention in MODS according to the hypothesis of ªUncoupling of biological oscillatorsº (1). REFERENCES. 1. Godin PJ, Buchmann TG (1996) Crit Care Med 24: 1107±16 . 2. Schmidt et al. (2000) Int Care Med 26 (Suppl.3): Abst106

221 THE IMPACT OF THE DEGREE OF TRICUSPID REGURGITATION ON UTILIZATION OF THE PULMONARY ARTERY CATHETER

Balik M1, Pachl J1, Hendl J2.1Anaesth. and Intensive Care, University Hospital Kralovske Vinohrady, Prague 10, 2Anaesth. and Intensive Care, Fac.of Phys.Educ. and Sport, Prague, Czechia INTRODUCTION. The aim of the study is to find the relationship between the grade of tricuspid regurgitation (TR) and accuracy of cardiac output (CO) measurement by thermodilution in mechanically ventilated patients. METHODS. 27 non-cardiac surgery patients were separated into three groups. All patients were initially investigated by transesophageal echocardiography (TEE) (multiplanar probe, Omniplane, Hewlett-Packard) and later the pulmonary artery catheter (PAC) was inserted for continuing haemodynamic instability. All patients with higher than the 1 st degree of aortic regurgitation were excluded. There were 8 patients with no or the 1 st degree of TR graded according to color doppler criteria, the second group consisted of 9 patients with the 2 nd degree of TR. The third group included 10 patients with the 3 rd degree of TR. All patients were measured twice simultaneously by TEE and PAC for cardiac output. At least three pulsions were measured using continuous doppler for velocity-time integral (VTI) at the level of aortic valve and at least six VTI were averaged in case of stroke volume (SV) variation in atrial fibrilation. Aortic valve area (AVA) was measured by planimetry twice and results were averaged. SV was calculated multiplying VTI with AVA and heart rate (1,2). Simultaneous PAC measurement was carried out applying three 10 ccm boluses of iced saline. RESULTS. The mean difference between TEE measurement and PAC measurement was 514.1  541.3 ml/min in the first group of patients (r = 0.96 p < 0.0001). The mean difference of 837.8  976.1 ml/min was found in the second group of patients (r = 0.92 p < 0.0001). The difference between the two modes of CO measurement was 1893.0  1143.9 ml/min in the third group (r = 0.69 p < 0.001). CONCLUSION. The difference in the third group is probably caused by inadequately low values of CO measured by thermodilution. The inaccuracy of CO measurement in the group of patients with the 3 rd degree of TR can be misleading for further therapy. It can cause more significant inaccuracy in another calculated parameters like pulmonary and systemic vascular resistances and stroke work indexes of left and right ventricle. REFERENCE. Perrino AC, Harris SN, Luther MA: Intraoperative Determination of Cardiac Output Using Multiplane Transesophageal Echocardiography, Anesthesiology, 1998, 89: 350±357

INTRODUCTION. Postoperative supraventricular tachyarrhythmias (SVTA) are a frequent complication in surgical intensive care patients. Although direct current cardioversion (DCC) is known to be a highly effective therapy for terminating SVTA in medical patients its effectiveness in the treatment of SVTA in surgical intensive care patients has never been investigated. Therefore we studied effectiveness of DCC in 37 postoperative critically ill patients with newonset SVTA without any history of previous tachyarrhythmias. METHODS. SVTA were defined as narrow-complex, non-sinus tachycardias with heart rates > 100 bpm for at least 15 minutes. In all patients a twelve lead ECG, arterial blood gas and serum electrolyte analysis were obtained before DCC. Serial creatine kinase MB isoenzyme determinations (three times within 24 hours) were performed. Demographic data including age, sex, weight, height, premorbidity factors, and admission SAPS-score as well as presence of SIRS with or without systemic infection were recorded from all patients. DCC was performed using a monophasic, damped-sinus-wave defibrillator. Energy levels used were 50, 100, 200 and 300 Joule (J) for regular SVTA and 100, 200, 360 J for irregular SVTA. Therapeutic effectiveness of DCC was followed for 48 hours. Demographic, premorbidity, laboratory data, presence or absence of SIRS with and without systemic infection were compared between patients responding to DCC and nonresponders using chi2-test, Fisher¢s exact test or Kruskal-Wallis one-Way Analysis of Variance. P-values < 0.05 were considered significant. RESULTS. None of the patients experienced clinically relevant hypoxia (paO2 < 60 mmHg), hypokalemia or hypomagnesemia at onset of SVTA. DCC restored sinus rhythm in 13 out of 37 patients (35 % primary responders). However, at 24 hours and 48 hours, only 6 (16 %) and 5 (13,5 %) patients remained in sinus rhythm, respectively. Primary responders were significantly younger and demonstrated significant differences in arterial pO2 values and ionized calcium concentrations at onset of SVTA when compared with nonresponders. In the present study 24 patients (65 %) fulfilled the criteria of SIRS with and without infection before onset of SVTA. In addition 29 patients (78 %) received catecholamine and/or phosphodiesterase inhibitor therapy. CONCLUSION. In surgical intensive care patients DCC is an ineffective therapeutic method for successful long-term termination of new-onset SVTA. We speculate that differences in effectiveness of DCC between medical patients and surgical intensive care patients may result from major differences in the primary trigger mechanisms responsible for SVTA development.

Oral Presentations Pathophysiology and epidemiology of nosocomial pneumonia ± 223±227 223 CHEST INFECTION FOLLOWING INTENSIVE CARE: THE EFFECT OF DYSPHAGIA AND TRACHEOSTOMY

Conway DH, Dawson S, MacJannet L, Eddleston J. Critical Care Medicine Department, Manchester Royal Infirmary, Manchester, UK INTRODUCTION. Poor residual health has been reported by patients following intensive care (ICU) discharge (1). Dysphagia can be a major risk factor for aspiration pneumonia (2). Tracheostomy is frequently performed on ICU. The aim of this study was to evaluate the effect of tracheostomy and dysphagia on the incidence of chest infection in ICU survivors. METHODS. A prospective observational study based in the follow up clinic of a teaching hospital. All patients were invited to attend clinic three months after ICU discharge. We reviewed consecutive patients attending clinic between February 1999 and 2001. Patients were asked if they had noticed any change in swallow or experienced a chest infection since ICU discharge. All tracheostomy scars were examined for healing. RESULTS. 171 patients were reviewed. 63 (37 %) reported chest infection since ICU discharge. 17 (10 %) of all patients reported dysphagia and one noticed improved swallow. 10 of the dysphagic patientshad suffered a chest infection, p < 0.05 chi squared. ICU tracheostomy was performed on 51 patients. Of these 21 had chest infection following discharge. Four patients had a poorly healed tracheostomy and two of these reported chest infection. 72 patients expressed satisfaction with their general health, even though 29 had suffered chest infection.

Chest infection N chest infection

n

Swallow worse

Swallow same/ better

Trache.

No trache.

63 108

10 7

53 101

21 30

42 78

Satisfied Diswith satisfied health 29 43

34 65

Incidence of chest infection in ICU survivors CONCLUSION. ICU survivors are at high risk of chest infection in the three months following discharge. Difficulty swallowing may contribute to this risk. The incidence of chest infection may be an underestimate as clinic non-attenders have more health problems (1). We could detect no association between tracheostomy and chest infection following ICU. REFERENCE. Eddleston J, White P, Guthrie E. Survival, morbidity and quality of life after discharge from intensive care. Crit Care Med 2000;28: 2293±9 2. Lundy DS et al. Aspiration: cause and implications. Otolaryngol Head Neck Surg 1999;120: 474±8

S 192

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

224

226

BACTERIAL COLONISATION OF LOWER AIRWAYS IN MECHANICALLY VENTILATED INTENSIVE CARE UNIT PATIENTS

NON-MICROBIOLOGICALLY DOCUMENTED ICU-ACQUIRED PNEUMONIA: IS IT A VALID ENTITY AND WHAT IS ITS RELEVANCE?

Agvald-Ohman C1, Edlund C2, Nord CE2, Wernerman J1. 1Dept of Anesthesiology and Intensive Care, 2Dept of Clinical Bacteriology, Huddinge University Hospital, Karolinska Institutet, Stockholm, Sweden

Pereira JM, Paiva JA, Aguiar J, Barbosa S, Gomes JP, Honrado T, Maia I, Massada S, Rios M, Sousa-Dias C, Mota AM, Duarte J. Serviço De Cuidados Intensivos, Hospital De S. Joˆo, Porto, Portugal

INTRODUCTION. The pathogenesis behind nosocomial infections in the intensive care unit (ICU) in particular ventilator-associated pneumonia (VAP) is still controversial. The aim of the present study were to assess the development over time and impact of oropharyngeal and gastric colonisation on the emergence of tracheal colonisation possibly leading to VAP in mechanically ventilated patients in a Swedish multidisiplinary ICU. Special emphasis was made on elucidating the role of anaerobic bacteria in the lower respiratory tract. METHODS. Consecutive patients (n = 41) requiring mechanical ventilation for more than 3 days were included. Patient characteristic were: median age 59 years (range 20±82), 15/41 were females. Median APACHE II score was 19, range 0±44, ICU mortality was 11/41. Ten of the patients had undergone transplantation, 32/41 patients had got at least one antibiotic before admission, 12/41 patients had clinical pneumonia at the time of admission. Samples were collected from oropharynx, subglottic space, trachea and stomach. The samples were collected after intubation and then every third day until day 33 or extubation. All samples were immediatly transported to the laboratory and frozen to ±70 C until analysed. RESULTS. Patients were heavy colonised ( > 60 % of samples) with microorganisms not considered belonging to a healthy normal gastric and oropharyngeal flora already at admission to the ICU. A majority of harboured enterococci, coagulase-negative staphylococci and candida spp. in at least one site on day 1. The patients were colonised with rather constant levels of microorganisms over time. Only in 4/41 an increasing concentration and in 9/41 a decreasing concentration over time in at least 50 % of the isolated species were found. An increasing load over time was associated with enterococci, gram-negative anaerobes and enterobacteriaceae. Decreasing levels occurred mainly in candida spp. We found anaerobes the majority were gram-negative, prevotella spp in subglottical and tracheal secret from day 12 to day 23 in a range between 40±100 % of all samples. Interestingly only one patient had pneumococci and one hemophilus influenzae species detected. One of the 41 patients developed VAP according to the criteria in the study. CONCLUSION. Patients in the ICU subjected to prolonged intubation showed a different pattern of colonisation as compared to the non-hospitalized population already at the time of admission to the unit. In some cases a colonisation pattern usually associated with long-stayers in the ICU was present from the start. In addition the presence of anaerobic species was high with a distribution also in the lower airways. This finding may be related to improved methodology in cultivating these strains. Obviously the colonisation pattern of ICU patients is variable over time as well as geographically. Therefore the relevance and applicability of literature reports for the individual unit must be interpreted with some caution.

INTRODUCTION. There is considerable debate about the relevance and even the existence of non-microbiologically documented (CD) ICU-acquired pneumonia (VAP). The goal of our study was to assess the incidence, relevance, impact and mortality of CD-VAP, compared to microbiologically documented (MD) ICU-acquired pneumonia. METHODS. Retrospective analysis of prospectively collected data of all VAP occurring in a 12-bedded Intensive Care Unit of an Emergency Department of an University Hospital in Porto, from the 1 st January 1999 to the 31 st December 2000. The agreement of at least two senior intensivists in face of the clinical situation and of one senior intensivist acting as a reviewer of the clinical file was required for the diagnosis of VAP. MD-VAP was a pneumonia acquired more than 48 hours after admission to the ICU in which a plausible pathogen was isolated in bronchial secretions and/or blood cultures. In CD-VAP no plausible pathogen was found. Goal variables used to compare MD- and CD-VAP were: hospital mortality, ICU-mortality, mortality caused by VAP, day of VAP diagnosis, days of antibiotherapy, mechanical ventilation and ICU stay after VAP, success of the first empiric regimen and concomitant existence of another infection. Concomitant infections were defined as those diagnosed from day ± 2 to day 10 relating to VAP diagnosis. Statistical tests used were Chi-square and Mann-Whitney. RESULTS. The results were as follows:

225

227

THE PATHOGENESIS OF VENTILATOR-ASSOCIATED PNEUMONIA(VAP) IN SURGICAL PATIENTS: ROLE OF AERODIGESTIVE COLONIZATION

EPIDEMIOLOGY AND OUTCOMES OF VENTILATOR-ASSOCIATED PNEUMONIA (VAP) IN A LARGE US DATABASE

Nulens M1, Verwaest C1, Verhaegen J2, Van den Berghe G1, Wouters P1, Vlasselaers D1, Demedts M3, Lauwers P1. 1Department of Intensive Care Medicine, 2Department of Microbiology, 3 Department of Respiratory Medicine, University Hospital Gasthuisberg, Leuven, Belgium

Ollendorf D1, Rello J2, Kollef M3, Vera-Llonch M1, Bellm L4, Redman R4, Oster G1. For the VAP Outcomes Scientific Advisory Group5. 1No Department, Policy Analysis Inc., Brookline, 3 Medical Critical Care and Respiratory Care Services, Barnes-Jewish Hospital, St. Louis, 4No Department, IntraBiotics Pharmaceuticals, Inc., Mountain View, USA, 2Critical Care Dept., University Hospital Joan XXIII, Tarragona, Spain, 5

INTRODUCTION. Bacteria may gain entrance into the lung by several mechanisms.The major mechanism underlying the development of VAP seems to be (micro-) aspiration of potentially pathogenic microorganisms (PPM) (1). The aim of this prospective observational study was to describe and assess the relationship between oropharyngeal (OC), gastric (GC) and tracheal colonization (TC) patterns preceding VAP in surgical patients. METHODS. During a two years period,consecutive adult surgical patients at high risk for VAP and not receiving antimicrobials were studied.Quantitative cultures of endotracheal aspirates(ETA) reflecting TC were investigated daily, whereas oropharyngeal swabs and gastric aspirates were cultured in a qualitative manner, on admission and then thrice weekly until extubation. In case of clinical suspicion of VAP (clinical VAP), quantitative invasive diagnostic techniques (ID) including PSB and BAL, were performed to confirm the diagnosis (probable VAP). Organisms of the same genus and species were considered identical on the basis of antibiotyping. Proportions were compared with the chi-square test, with Bonferroni's adjustment for multiple comparisons. RESULTS. A total of 178 patients (cardiac surgery 46 %,trauma 29 %,neurosurgery 16 %,miscellaneous 9 %)were studied.After 7 respectively 14 days of intubation,80.6 % respectively 94.3 % of patients acquired TC.ID were used to study 127 episodes of clinical VAP in 95(53 %)patients.Probable VAP was documented in 69 (38.9 %) patients. Patients with higher age (P = 0.039) and cardiosurgical patients (P < 0.007) suffered more VAP and patients developing VAP had a significantly longer ICU-stay (P < 0.0001). A total of 101 PPM causing VAP were isolated. A strong correlation was found between ETA and ID samples (Rho = 0.655,P < 0.0001) and TC consistently preceded the development of VAP. For all PPM causing VAP and considering only OC an GC prior to or concurrent with TC,OC (78 %)was significantly more frequent than GC (40 %) (P < 0.0003). Considering both OC and GC,the oropharynx was the most frequent(93.5 %)initial site to be colonized and the most frequent (65 %) single site carrying PPM that were cultured concurrently or afterwards in ETA and caused VAP. For inducible enterobacteriaceae, P.aeruginosa and S.aureus, OC and GC, prior to or concurrent with TC, were respectively 100 % (OC) and 65 % (GC) (P < 0.0002),75 % (OC) and 20 % (GC) (P < 0.0032), 83 % (OC) and 14 % (GC) (P < 0.0001). CONCLUSION. In surgical patients without previous antimicrobial therapy, the tracheobronchial tree appears to be the main source of most microorganisms causing VAP. The oropharynx seems the pivotal source of a majority of PPM causing TC, while the stomach may be less important in the pathogenesis of VAP in this category of patients. REFERENCE. D.George et al.Epidemiology of Ventilator-acquired Pneumonia based on Protected Bronchoscopic sampling.Am.J.Respir.Crit.Care Med.1998;158: 1839±1847

Global hospital/ICU mortality Mortality caused by VAP Day of VAP diagnosis Days of antibiotherapy after VAP Days of mechanical ventilation after VAP Days of ICU stay after VAP Success of 1 st empiric regimen Concomitant existence of another infection

Total VAP n = 165

MD-VAP n = 111

CD-VAP n = 54

p=

32/25 % 7% 6 8 6,5 11 68 % 30 %

31/24 % 8% 7 8 8 11 65 % 36 %

35/26 % 6% 5 7,5 4 10 74 % 19 %

NS/NS NS NS NS NS NS NS 0,018

CONCLUSION. MD- and CD- VAP seem to be similar in terms of all goal-variables studied and the existence of another infection concomitant with VAP is even more frequent in MDthan in CD-VAP. These two facts lead us to admit the existence and the clinical relevance of CD-VAP.

INTRODUCTION. Risk factors for VAP vary from study to study and its impact on outcomes is controversial. Using data from a large, geographically diverse US inpatient database, we examined the incidence of and risk factors for VAP, as well as its effect on selected outcomes. METHODS. Data were obtained for all patients admitted to an ICU from January 1998 to June 1999 who received mechanical ventilation for more than 24 hours. Patients were stratified based on whether they developed pneumonia at least 24 hours after intubation. The importance of potential risk factors for VAP was examined using crude and adjusted odds ratios. Cases of VAP were matched on age, duration of ventilation, type of admission, and severity of illness to up to 3 controls. Mortality, resource utilization, and billed charges were then compared between cases and controls. RESULTS. A total of 9,080 patients were identified; 842 (9.3 %) had VAP. On average, VAP developed approximately 5 days after hospital/ICU admission. Patients with VAP were significantly (p < .05) younger and more likely to be male and admitted for trauma than those without VAP. Trauma admissions were nearly twice as likely to develop VAP as medical admissions (adjusted OR: 1.75; 95 % CI: 1.41, 2.18). Compared to controls, patients with VAP had significantly (p < .0001) greater numbers of days of mechanical ventilation (14.3 vs 4.7 for controls), ICU days (11.7 vs 5.6), and hospital days (25.5 vs 14.0); their mean (  SD) billed charges were $40,000 higher ($104,983 [  91,080] vs $63,689 [  $75,030] for controls). Mortality did not differ between cases and controls. CONCLUSION. This study, the largest of its kind, suggests that VAP is common and associated with adverse health and economic outcomes. Strategies to prevent this complication may yield significant benefits to the patient and the healthcare system.

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

S 193

Oral Presentations Liver failure and nutrition ± 228±232

230

228

Geisser W1, Asam C1, Wiedeck H1, Vogt J1, Wachter U1, Schricker T2, Radermacher P1, Georgieff M1. 1 Universitätsklinik für Anästhesiologie, University Medical School, Ulm, Germany, 2Department of Anesthesia, McGill University, Montreal, Canada

GLUTAMINE ENRICHED TOTAL PARENTERAL NUTRITION DOES NOT IMPROVE PROTEIN CATABOLISM IN PATIENTS WITH SEPTIC SHOCK

INCIDENCE, PREVALENCE AND PATTERN OF REFEEDING SYNDROMEASSOCIATED BIOCHEMICAL ABNORMALITIES FOLLOWING COMMENCEMENT OF TOTAL PARENTERAL NUTRITION O'Hanlon D1, Corcoran N1, Bourke J2, Woolfe C2, O'Connell PR1, Phelan D2. 1Department of General Surgery, 2Department of Intensive Care Medicine, Mater Misericordiae Hospital, Dublin 7, Ireland INTRODUCTION. The refeeding syndrome refers to a complex metabolic phenomenon, which occurs following the re-introduction of carbohydrate-calories to acutely starved or chronically malnourished patients. Classically, it is marked by severe hypophosphatemia, but changes in potassium and magnesium may also be important.1 The aim of this study was to define the prevalence and incidence of refeeding syndrome-associated abnormalities in a large cohort of hospital based patients commenced on TPN. METHODS. The Mater Misericordiae Hospital is a 510 bedded University Teaching Hospital, serving a population of 200,000 and it has a large tertiary referral practice. Between December 1989 and December 1997, 916 patients were commenced on in-hospital total parenteral nutrition. The set-up, delivery and monitoring of TPN was performed by a specialist Intravenous Nutrition Service consisting of a dedicated IV nutrition nurse and a consultant intensive care physician. Referral for TPN was made through the appropriate team caring for the patient, or directly as part of the multidisciplinary team in the intensive care unit. The Intravenous Nutrition Service reviewed the biochemical profile of patients individually and prescribed TPN (including electrolytes) electively five times per week. The database of the Intravenous Nutrition Service was interrogated. Serum phosphate, magnesium and potassium levels were recorded daily prior to and following commencement of TPN. A detailed record was kept of the daily TPN prescription and this included phosphate, magnesium and potassium levels in the bag and supplemental phosphate, magnesium or potassium received. Statistical analysis was performed using ANOVA and Chi square on SPSS for Windows (Version 10). Significance was assumed at the five-percent level. RESULTS. The mean phosphate level decreased significantly (P < 0.001) after commencement of TPN. The prevalence of hypophosphatemia was 18.9 % prior to commencing TPN. The incidence on day 1 was 9.3 % (P < 0.001). The mean potassium level decreased significantly (P < 0.001) after commencement of tpn. The prevalence of hypokalemia was 20 % prior to commencing TPN. The incidence on day 1 was 8.7 % (P < 0.001). The mean magnesium level was low prior to commencing TPN and increased significantly (P < 0.001) following the introduction of TPN. The prevalence of hypomagnesemia was 39.7 % prior to commencing TPN. The percent of patients with a low magnesium level decreased to 25.4 % on day 1 and to 6.0 % on day 2 after commencing TPN. CONCLUSION. The prevalence of refeeding syndrome-associated biochemical abnormalities is high in hospitalised patients referred for TPN, but the incidence following the commencement of TPN is low. Hypomagnesemia was not a part of the refeeding syndrome following the introduction of TPN in the present study. This contasts with the pattern of refeeding hypophosphatemia and hypokalemia that were observed. REFERENCE. Solomon S, Kirby D: The Refeeding Syndrome: A Review. JPEN 14: 90±97, 1990

INTRODUCTION. In critical ill patients glutamine becomes an essential amino acid with increased turnover and release from peripheral organs. This mechanism is supposed to play a major role in regulating catabolism in patients with septic shock. The aim of this study was to investigate the influence of a glutamine enriched total parenteral nutrition (TPN) on whole body catabolism in these patients. METHODS. A prospective randomised trial on 30 Patients who fullfilled the criteria of septic shock (ACCP/SCC) was performed. Patients were randomly assigned to receive 5 days of TPN with either a glutamine enriched amino acid solution for parenteral nutrition or an isonitrogenic isoenergetic standard amino acid solution (Glamin 13 % [G] and Intrafusin 10 %[I], respectively). TPN was adjusted to the individual energy expenditure determined by indirect calorimetry. Using the stable isotope dilution technic urea production [15N2-urea] and endogenous glucose production rate [6,6±2H2-glucose] was measured. Protein breakdown was calculated using the equation by Shaw [1]. Patients were excluded if renal failure was present or developed throughout the investigation. RESULTS. 6 patients had to be excluded because of death or renal failure (2 G and 4 I). There were no differences concerning gender, age, body mass index and severity of their disease (sepsis score). A ANOVA with subsequent Tukey was performed for statistical testing. G pretreat Urea production [mmol/kg/min] Net protein breakdown [g/d](isotopic) Net protein breakdown [g/d](urine) Endogenous glucose production [mmol/kg/min]

I pretreat

G after Day 3

I after Day 3

G after Day 5

I After Day 5

115

94

129

114

106

104

21675

18483

234110

22388

19761

19778

16584

14497

15568

17664

16447

18075

236

217

168*

165*

1310*

106*

Metabolic Parameters (Values are means  SD, * designates p < 0.05 vs. pretreatment) CONCLUSION. In contrast to previous data (2) neither of the two TPN regimens was able to reduce protein catabolism in our patients with septic shock. The time dependent reduction in endogenous glucose production rate is probably caused by the nutritional support with glucose and the exogenous insulin supplementation. REFERENCES. 1. Shaw JH, Klein S., Wolfe R. R.; Surgery 1985 97 (5): 557±568. 2. Jian ZM, Cao JD, Zhu XD et al.; JPEN 1999 23 S62±66

229

231

EVALUATION OF ENTERAL, PARENTERAL AND COMBINED NUTRITION IN PATIENTS WITH SEVERE HEAD INJURY

ENTERAL NUTRITION-RELATED INTRA-ABDOMINAL PRESSURE IN CRITICALLY ILL PATIENTS

Tugrul S, Demirel I, Ergin Ozcan P, Cakar N, Esen F. Anesthesiology & Intensive Care, University of Istanbul, Medical Faculty of Istanbul, Istanbul, Turkey

Vazquez-Sanchez A, Navarro S, Artiaga M, Matias M, Abad V, Lopez R. Intensive Care Unit, Hospital Universitari Del Mar, Barcelona, Spain

INTRODUCTION. The hypermetabolism in patients with severe head injuries results in malnutrition, impaired healing and an increased tendency to multiple organ failure (1). Thus, early and adequate nutritional support plays a crucial role in clinical outcome. The objectives of our study were to compare the ability of three different nutrition strategies to meet the caloric and protein requirements in severe head injury, and to study their effects on nitrogen balance. METHODS. 30 patients with GCS < 8 were randomly placed in enteral (EN) (n: 10), parenteral (PN) (n: 10) and combined enteral-parenteral nutrition (CN) (n: 10) groups. Targeted caloric intake was 30±35 kcal/kg/day nonprotein and 1.5 gr/kg/day protein. Nitrogen balance, biochemical laboratory analysis and the amount of caloric intake were evaluated daily. RESULTS. The demographic data of the patients were similar between the groups. The mean targeted caloric intake of the groups were 2195  400 kcal/day in enteral, 2280  175 kcal/day in parenteral and 2225  270 kcal/day in combined nutrition groups (p: 0.9) (Table 1.). In enteral nutrition group, it took 5.3  1.6 days to reach targeted caloric amount, whereas balance periods of the targeted intake were shorter in parenteral (3.3  1 days) and combined nutrition (3  1 days) groups (p: 0.0003 when EN compared to PN and p: 0.001 when EN compared to CN). Nitrogen balance of enteral nutrition group on day 4 was lower than parenteral (p < 0.01) and combined (p < 0.01) nutrition groups. On day 10, there was no statistically significant difference between the nitrogen balance of three groups (Table 2.). Outcome GCS of patients, duration of mechanical ventilation and ICU stay of groups were found similar.

INTRODUCTION. Enteral feeding is often used in critically ill patients, gastrointestinal intolerance is probably the most important factor that limits its use. Our objective was to evaluate the relationship between enteral feeding tolerance and intra-abdominal pressure (IAP). METHODS. A total of 42 patients admitted to the intensive care unit (ICU) and receiving enteral nutrition were enrolled prospectively in the study, 32 had medical diseases and 10 had intraperitoneal abdominal surgery, all patients were followed until discharge from, or death in the ICU. Acute Physiology and Chronic Health Evaluation II (APACHE II) score, IAP and sagittal abdominal diameter (SAD) were recorded at admission. Every six hours we recorded: IAP,SAD, mL/h of administered diet,diarrhea,vomiting enteral formula,high gastric residuals ( > 200 mL/6 h), temporary stop of the infusion by intolerance, use of benzodiadiazepines, opiates and prokinetic agents .The patients outcome were also registered . In 37 patients the nasogastric tube was the method of feeding in 5 was a needle-catheter jejunostomy tube. Significant differences were assessed by Student's and chi-square tests . RESULTS. The first IAP before the beginning of the enteral feeding was 7.10  3.4 mmHg, the patiens with medical disease had an initial IAP of 6,1  2,9 mmHg, in the surgical patientes was 10,1  3,1 mmHg (p = 0,001).Admission APACHE II was 18.2  6.7 in medical patients and 13.7  4.0 in surgical patients ( p = 0,05).The SAD was 99  14 cm in medical patients and 103.1  13 cm in surgical patients (NS). High gastric residuals (58 %), diarrhea (28 %) and vomiting (12 %) are the main gastrointestinal complications (intolerance) . The patients (23) with everytime successful enteral feeding tolerance had a mean IAP of 5.1  2.3 mmHg, those with any episode of intolerance (19) had a mean IAP of 8.8  2.1 mmHg (p = 0.000). When the infusion was stopped a mean IAP of 10,4  1.8 mmHg was recorded, when wasn't the mean IAP was 5.1  2.3 mmHg (p = 0,000). In patients with any episode of intolerance (19), obviously there were episodes of right tolerance, in these episodes the mean IAP was of 6.7  1.9 mmHg, in patients with always successful tolerance the mean IAP was 5.1  2.3 mmHg (p = 0,02).The mL /hr of infused volume of diet were 39.9  18,9 mL in patients witn some determination of IAP over 10 mmHg and 61.9  17,6 mL in those with any under this IAP pressure (p = 0,000) The observed mortality rate was 28.6 % (12 patients), the mean IAP was 8.3  2.8 mmHg in deads and 6.2  2.6 mmHg in survivors (p = 0,03) .The enteral nutrition intolerance was higher in patients who died: 11 patients died among the 19 who intolerate diet compared with 1 dead among the 23 whit successful enteral feeding tolerance (chi2 test, p = 0,000). CONCLUSION. In critically ill patients there is a relationship between enteral feeding tolerance and IAP (the higher IAP the worse tolerance) .Enteral nutrition intolerance has a prognostic value in these patients. IAP could be an easy parameter to control the tolerance to enteral feeding .

EN PN CN

Day 1

Day 2

Day 3

Day 4

Day 5

ANOVA (p)

144286 252313 520618

786615 1162789 1337709

1426843 1930705 2196704

1819766 2223602 2200290

1993657 2281146 2225271

0.008 < 0.0001 0.012

Mean amount of daily caloric intake (kcal) EN PN CN

Day 1

Day 4

Day 10

ANOVA (p)

±12.32.7 ±11.83.8 ±12.45.6

±15.15.0 ±7.64.6 ±8.44.0

±5.93.5 ±6.02.4 ±6.13.0

0.001 0.046 0.049

Nitrogen balance CONCLUSION. In severe head injury, when early enteral nutrition support is not sufficient due to diarrhea or late gastric tolerance, combination of enteral and parenteral nutrition can be considered to improve the nutritional condition of severe head-injured patients. REFERENCE. Wilson RF, Tyburski JG, J Head Trauma Rehabil 13(1); 11±27; 1998.

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

S 194

232

234

TREATMENT EFFECT OF EARLY ENTERAL VERSUS PARENTERAL NUTRITION: A META-ANALYSIS

MORTALITY, MORBIDITY, AND WORKLOAD IN ICU-ACQUIRED SEVERE SEPSIS

Peter JV, Moran JL, Phillips-Hughes J. Intensive Care Unit, The Queen Elizabeth Hospital, Woodville, Australia

Alberti C1, Adrie C1, Chaix C1, Cheval C1, Fosse JP1, Bornstain C1, Thuong M1, Troche G1, Moine P1, Meshaka P2, Pinton P2, Timsit JF1. 1OUTCOMEREA study group, DBIM Hôpital Saint Louis, Paris, France, 2, Eli & Lilly Company, Saint Cloud, France

INTRODUCTION. Review studies have suggested that early enteral nutrition (EN) may be of advantage compared with parenteral nutrition (PN) [1]. Meta-analytic techniques are used to address this issue. METHODS. Prospective randomised controlled trials (RCT) comparing early EN with PN were identified by Medline search [1966±2001] and bibliographic review of relevant articles. Primary outcomes addressed were mortality, hospital length of stay (HLOS) and patient complications (all complications, infections and diarrhoea). Exclusion criteria included non-RCT, use of immuno-nutrition, PN compared with standard care and studies reporting only physiological variables. Treatment effects were expressed as risk difference (RD as percent, PN vs EN) for binary events and weighted mean difference for HLOS. Publication bias was identified by funnel plot and the quantitative methods of Begg, Egger and ªtrim and fillº. Modification of treatment effect was assessed by meta-regression for pre-defined confounders: age, gender, prenutritional albumin concentration and time to initiate nutrition (Ti: hours). RESULTS. Of 47 potential controlled trials, 26 fulfilled inclusion criteria. Study size ranged from 18 to 257 total patients, over the years 1980±2001. Mean(SD) Ti did not differ between EN and PN (28(22) vs 27(15), p = 0.91). Funnel plot inspection suggested publication bias, not identified by the Begg or Egger test. ªTrim and fillº methodology suggested one potential missing study, but point estimate for mortality was not changed. No modification of mortality treatment effect was suggested by meta-regression (p > 0.5) or medical vs surgical sub-group analysis.

Mortality (%) HLOS (days) Complications (%) Infections (%) Diarrhoea (%)

Treatment Effect

Confidence Interval 95 %

Significance p value

Heterogeneity p value

0 1.7 5.8 8.1 ±13.5

±2.2 to 3.0 0.34 to 3.1 0 to 11.7 3.4 to 12.7 ±22 to ± 5.0

0.76 0.01 0.05 0.001 0.002

0.65 0.73 0.30 0.16 0.001

Primary Outcome analysis CONCLUSION. EN was associated with a statistically and clinically significant reduction in complications and HLOS compared with PN. No mortality effect was observed. Diarrhoeal episodes significantly increase with EN. REFERENCE. Lipman T. JPEN 1998; 22: 167±82.

Oral Presentations Evaluating outcome in sepsis and ARDS ± 233±237

INTRODUCTION. Severe Sepsis (SS) is the first cause of mortality in ICU 1 (Intensive Cardiology Care Units excluded). Severe Sepsis epidemiology evaluation is a priority to better understand its incidence, its physiopathology and new therapeutics strategies consequences.The purpose is to evaluate the SS epidemiology including mortality and morbidity over a large population of patients representing the Parisian area. METHODS. We prospectively recorded more than 100 000 ICU consecutive stays over a 7 year-period in 34 ICU (452 beds) of the Parisian area (65.7 % were university hospitals and 34.3 % other hospitals). Our study included all patients with a SS defined according to Bone's criteria 2, with a length of stay > 2 days and admitted during a 3-year period (1997, 98, 99). It represents a population of 55047 patients. RESULTS. 10 459 patients fulfilled our criteria: 19 % of the global population admitted during the same period. It gives incidences equivocal to 8 cases/bed/year in university hospitals and 7.1 cases/bed/year in other hospitals. The patient's characteristics are mean age 61.0  16.4 years and 64 % of male. SAPS II score is 45.5  18.6 predicting a 36 % hospital mortality. Mortality observed in ICU is 31.7 % and 37.8 % in hospital.

Emergency room 40 % Pulmonary disease 14 % Pulmonary 89 % Mechanical ventilation 78 %

Origin of patients Comorbidity Organ Failure Organ support Length of stay

Total 33 days

Unscheduled surgery 7.4 %

Internal transfert 43 % Cardiac disease 6 % Circulatory 65 % Vaso active drugs 60 % Prior to ICU 4 days

External transfert 17 % Immuno compromise 21 % Renal 34 % Hemodialysis (contin. or intermittent) 19 % ICU 18 days after ICU 11 days

Patients description

Infection site

Lung56 % Bacteria54 %

Cocci Gram +

Staphy 35 %

Baciles Gram-

Pseudo 23 %

Abdo 10 % No docs 43 % S Pneum 21 % E Coli 18 %

D. On admission and > 48 hours

857 (51) 118 (14) 176 (21) 6 (4±8) 16 (9±27) 150 (18) 62 (7) 59 (32±105)

502 (30) 102 (20) 146 (29) 8 (5±13) 20 (11±34) 95 (19) 54 (11) 123 (59±220)

128 (8) 57 (45) 75 (59) 18 (10±34) 23 (14±50) 68 (53) 34 (27) 335 (256±699)

211 (12) 105 (50) 129 (61) 22 (14±37) 32 (17±55) 77 (37) 57 (27) 430 (256±772)

N (%) ICU mortality* Hospital Mortality* ICU stay (d)* Hospital stay * Catecholamine* HD-HF* Omega score*

CONCLUSION. Patients exhibiting features of secondary sepsis after the first 48 hours have a higher mortality rate and a higher length of stay in both ICU and hospital. It is related to a higher workload and cost as assessed by the total Omega score. REFERENCES. 1. Bone RC et al- Chest 1992;101: 1644±55. 2. Sznadjder M et al- Intensive Care Med 1998; 24: 582±9.

INTRODUCTION. Severe sepsis is one of the most challenging diseases requiring expensive procedures and drugs but its actual workload and total direct costs that may be evaluated by the Omega score are poorly defined. METHODS. We evaluated severe sepsis prognosis and resource consumption accordingly to the source of infection. Data were prospectively collected in 6 intensive care units in France from 1997 to 2000. Sepsis, severe sepsis, and septic shock were defined according to Bone's criteria (1). The intensity of care was measured accordingly to the daily use of invasive procedures and by the Omega score. This score describes 47 diagnostic and therapeutic weighted items measured either daily or once during the stay. It has been closely related to both workload and ICU costs (2). Data are expressed as percent or as median [25 %±75 %] quartile. Variables were analyzed either by chi-square or by Kruskal Wallis test as needed. *, P < 0.05 RESULTS. 713 (42 %) patients out of 1698 ICU patients hospitalized for more than 48 hours had severe sepsis on admission. Amongst them, 172 were in septic shock (24 %). They were 429 males (60.2 %) with a median age of 68 [55±77], a SAPS II score of 47 [35±60], a LOD score of 6 [4±8] and a SOFA score of 7 [4±9]. The ICU and hospital mortality was 29 % and 39 %, respectively. Relationships were explored between source of infection and outcome. Pneumonia, peritonitis and severe sepsis from multiple sources were associated with a higher workload as assessed by the total Omega score. All 713 ICU mortality* n (%) ICU stay (d)* Mechanical ventilation* Inotropic Agents Omega score*

Organism type

C. Severe sepsis > 48 hours

Adrie C1, Alberti C1, De Lassence A1, Azoulay E1, Garrouste M1, Cohen Y1, Thuong M1, Poitrinal P2, Bouhassira M2, Timsit JF1. 1OUTCOMEREA study group, Hôpital Delafontaine, Saint Denis, France, 2Eli & Lilly Company,, Saint Cloud, France

Guidet BG1, Aegerter PA1, Gauzit RG1, Dreyfuss DD1, 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

Scheduled surgery 14 %

B. Severe sepsis on admission

MORTALITY AND WORKLOAD OF SEVERE SEPSIS IN INTENSIVE CARE UNIT DEPENDING ON THE SOURCE OF INFECTION

SEVERE SEPSIS EPIDEMIOLOGY IN THE PARISIAN AREA: ANALYSIS OF THE CUB-REA DATA BASE

Medical 78.6 %

A. No severe sepsis

235

233

Type of patients

INTRODUCTION. Stringent criteria for severe sepsis and sepsis shock have been defined to standardize studies. However they all lack of specificity and particularly in patients hospitalized in ICU for more than 48 hours since organ dysfunction and other criteria may occur independently of infectious processes. METHODS. We evaluated both morbidity and mortality associated with ICU-acquired severe sepsis (i. e. after the 48 th hour).Data were prospectively collected in 6 intensive care units (ICUs) in France from 1997 to 2000. Sepsis, severe sepsis, and septic shock were defined according to Bone's criteria (1). The intensity of care was measured accordingly to the daily use of invasive procedures and by the Omega score. This score describes 47 diagnostic and therapeutic weighted items measured either daily or once during the ICU stay (2). We distinguished 4 groups: (A) No severe sepsis (B) Severe sepsis on admission (C) Severe sepsis occurring beyond the first 48 hours (D) Patients presenting a severe sepsis episode on both admission and after the first 48 hours. Data are expressed as percent or as median [25 %±75 %] quartile. Variables were analyzed either by chi-square or by Kruskal Wallis test as needed. *, P < 0.0001 RESULTS. Secondary severe sepsis more frequently occurred in patients with severe sepsis on admission: 211/713 = 30 % vs. 128/985 = 13 % (P < 0.0001). The occurrence of severe sepsis after 48 hours was significantly associated with a higher ICU or hospital mortality and a higher resource consumption. This was observed whether or not the patient was initially admitted for severe sepsis. The total Omega score, which has been shown to closely correlate with the workload and direct cost of each stay(2), was also significantly higher in those patients.

Urinary tract 5%

+ blood culture 31 %

Unknown 15 %

Fungus 4 %

Virus 2 %

Parasites 1 %

Other Strepto 9 % Haemophilus 6%

Enterococ 4% Klebsiella 5%

Other Gram17 %

Infections description CONCLUSION. We provide a more complete picture of the SS epidemiology than the highly selected patients included in clinical trials. Our study highlights the need for new therapies in order to reduce the mortality of this frequent disease. REFERENCE. 1. Sands K. E. JAMA 1997, 278: 234±40. 2. Bone R. C. Chest 1992, 101: 1644±55

Pneumonia Peritonitis 306 (43) 77 (25)

UTI 66 (9)

Primary Multiple Bronchitis bacteremia sources 120 (17) 58 (8) 56 (8)

209 (29)

107 (35)

22 (29)

11 (17)

22 (18)

16 (28)

20 (36)

10 (6±20)

10 (6±21)

11 (7±21)

8 (5±18)

11 (6±17)

9 (5±19)

11 (5±22)

548 (77)

237 (78)

72 (94)

43 (66)

83 (69)

43 (74)

48 (86)

172 (24) 170 (90±367)

71 (23) 190 (90±381)

30 (39) 210 (120±402)

15 (23) 133 (58±252)

21 (17) 158 (77±296)

18 (31) 160 (58±336)

11 (20) 160 (58±336)

CONCLUSION. Severe sepsis and septic shock are frequent in ICU patients hospitalised for more than 48 hours. Patients with pneumonia, peritonitis, and source of infection sites represent 76 % of the whole population and were associated with highest mortality and workload. REFERENCES. 1. Bone RC et al ± Chest 1992;101: 1644±55. 2. Sznadjder M et al- Intensive Care Med 1998; 24: 582±9.

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

S 195

236

237

28-DAY MORTALITY IN PRIMARY AND SECONDARY ACUTE RESPIRATORY DISTRESS SYNDROME

SEVERITY STRATIFICATION IN THE ACUTE RESPIRATORY DISTRESS SYNDROME: SPECIFIC VERSUS GENERAL INSTRUMENTS

Vaz Pinto I, Fevereiro T, Matos R, Moreno R. Intensive Care, Hospital de St. António dos Capuchos, Lisboa, Portugal

Vaz Pinto I, Fevereiro T, Matos R, Moreno R. Intensive Care, Hospital de St. António dos Capuchos, Lisboa, Portugal

INTRODUCTION. Until now, in most studies, all patients with the acute respiratory distress syndrome (ARDS) have been analysed together. Although some studies have suggested that primary and secondary ARDS have different outcomes, it is usually considered that those different outcomes are depend from different severity of illness in both groups. The objective of this study is to compare the 28-days all cause mortality in patients with primary and secondary ARDS, controlling for the severity of illness. METHODS. The study used the database of the Unidade de Cuidados Intensivos Polivalente, Hospital de St. António dos Capuchos. All patients admitted from July 1, 1991 to June 30, 2000 with a diagnosis of ARDS were studied. ARDS was defined according to the AmericanEuropean Consensus Conference definitions (1). The ARDS was considered primary when resulted from primary lung injury (e. g. bacterial pneumonia, aspiration, near-drowning) and secondary when the lung was affected in the context of a non-pulmonary disease (e. g. abdominal sepsis, pancreatitis). Severity of illness as analysed by the APACHE II score (2) and the SAPS II score (3). The outcome measure used was vital status at 28 days. The Kaplan-Meier method was used to plot the survival curves for primary and secondary ARDS, and the log-rank test to compare the survival curves in the two populations. All data is presented as mean  SD, except when indicated otherwise. RESULTS. In the analysed cohort (n = 142), no differences where apparent in the baseline severity of illness, as measured by the APACHE II score (primary ARDS 24.53  6.76, secondary ARDS 25.98  11.94, p = 0.365) or by the SAPS II score (primary ARDS 52.71  16.25, secondary ARDS 57.08  25.33, p = 0.217). However, 28-days all cause mortality was significantly different in both groups (primary ARDS 55.24 %, secondary ARDS 69.23 %, p < 0.01). Both survival curves began to separate early in the course of the disease and this difference was still clear at 28-days (p < 0.01). CONCLUSION. At least in this sample, primary and secondary ARDS present a very different mortality at the 28-days in the ICU, which is not captured by general outcome prediction models such as the APACHE II and SAPS II. These results seems to indicate the need for other methods of risk evaluation and stratification in this population, when we need an unbiased evaluation of the underlying patient severity. REFERENCES. 1. Bernard GR, Artigas A, Brigham KL, et al. Report of the American-European Consensus Conference on ARDS: definitions, mechanisms, relevant outcomes and clinical trial coordination. Intensive Care Med 1994;20: 225±32. 2. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985;13: 818±29. 3. Le Gall JR, Lemeshow S, Saulnier F. A new simplified acute physiology score (SAPS II) based on a European / North American multicenter study. JAMA 1993;270: 2957±63.

INTRODUCTION. Several methods have been proposed for severity stratification in patients with the acute respiratory distress syndrome (ARDS). Of these, the Lung Injury Score (LIS), proposed by Murray et al. is the most commonly used (1). The objective of this work is to compare the LIS with two general outcome prediction models, the APACHE II (2) and the SAPS II (3) for risk stratification in patients with ARDS. METHODS. The study used the database of the Unidade de Cuidados Intensivos Polivalente, Hospital de St. António dos Capuchos. All patients admitted from July 1, 1991 to June 30, 2000 with a diagnosis of ARDS were studied (n = 142). The three instruments (LIS, APACHE II and SAPS II were computed according the original descriptions, using the worst values during the first 24 hours in the ICU. ARDS was defined according to the American-European Consensus Conference definitions (4). Outcome measure used was vital status at hospital discharge. All data is presented as mean  SD, except when indicated otherwise. The discriminative capability of the scores was evaluated through the use of the Area under the Receiver Operating Characteristics (ROC) curve. RESULTS. In the analysed patients, mean APACHE II was 23.17  7.50, Mean SAPS II 50.22  16.54 and mean LIS 2.59  0.45. Overall hospital mortality rate was 71.1 %). APACHE II and SAPS II presented significantly higher values in non-survivors than in survivors (APACHE II: 25.42  6.66 versus 17.66  6.58, p < 0.001; SAPS II: 54.40  15.70 versus 39.97  14.05, p < 0.001). This was not true for LIS (2.63  0.42 versus 2.49  0.51, p = 0.15). The area under the ROC curve was significantly higher for SAPS II and APACHE II than for LIS (APACHE II: 0.800  0.062; SAPS II: 0.737  0.047; LIS: 0.581  0.062). CONCLUSION. At least in this sample, the performance of the general outcome prediction models (APACHE II and SAPS II) was significantly higher than the LIS. These results suggest that they should be used when we need to perform severity stratification in patients with ARDS. The exact role for disease-specific scores, such as the LIS, should be object of further research. REFERENCES. 1. Murray JF, Matthay MA, Luce JM, Flick MR. An expanded definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988;138: 720±3. 2. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985;13: 818±29. 3. Le Gall JR, Lemeshow S, Saulnier F. A new simplified acute physiology score (SAPS II) based on a European / North American multicenter study. JAMA 1993;270: 2957±63. 4. Bernard GR, Artigas A, Brigham KL, et al. Report of the American-European Consensus Conference on ARDS: definitions, mechanisms, relevant outcomes and clinical trial coordination. Intensive Care Med 1994;20: 225±32.