Excessive alcohol consumption and perioperative outcome

Original communications Excessive alcohol consumption and perioperative outcome Joachim Klasen, Dr med, Axel Junger, PD Dr med, Bernd Hartmann, Dr med, Lorenzo Quinzio, Dr med, Matthias Benson, PD Dr med, Rainer Ro¨hrig, Dr med, and Gunter Hempelmann, Prof Dr med, Dr hc, Giessen, Germany

Background. Excessive alcohol consumption is a well-recognized factor contributing to premature morbidity and mortality. Methods. This retrospective, matched cohort study was designed to assess the attributable effects of excessive alcohol consumption on outcome in patients undergoing noncardiac surgery. All data of 28,065 patients operated at a tertiary care university hospital were recorded with a computerized anesthesia record-keeping system. Cases were defined as patients with history of excessive alcohol consumption (>30 g alcohol per day). Controls were selected according to matching variables in a stepwise fashion. Results. In our data set, 928 patients (3.3%) were found with a history of excessive alcohol consumption. Matching was successful in 897 patients (97%). The crude mortality rates for the cases were 1.3% and 1.6%, for the matched controls (P = .084, power = 0.85). Prolonged length of hospital stay was observed in 38% versus 33% (P = .013, power = 0.50), admission to the intensive care unit was deemed necessary in 11% versus 9% (P = .027, power = 0.55), and intraoperative cardiovascular events were detected from the database in 22% versus 21% (P = .053, power = 0.61). Conclusions. In this study, history of excessive alcohol consumption alone is not a factor leading to an increased perioperative risk in noncardiac surgery. (Surgery 2004;136:988-93.) From the Department of Anesthesiology, Intensive Care Medicine, and Pain Therapy, University Hospital Giessen, Giessen, Germany

THE IMPACT OF EXCESSIVE CONSUMPTION of alcoholic beverages on morbidity and mortality is still a matter of debate. Whereas a daily intake of not more than 2 drinks (15 to 25 g of alcohol) seems to have a beneficial effect, heavy consumption of Supported in part by a grant from IMESO GmbH, Hu¨ttenberg, Germany. The founding agreement ensured the authors’ independence in designing the study, interpreting the data, and writing and publishing the report. M. Benson, PD Dr med, is a partner of IMESO GmbH (Hu¨ttenberg, Germany) and an employee of the University Hospital Giessen. None of the other authors or participants has any financial interest in the subject matter, materials, or equipment discussed in this article or in competing materials. Accepted for publication March 5, 2004. Reprint requests: Axel Junger, PD Dr med, MBA, Abteilung Anaesthesiologie, Intensivmedizin, Schmerztherapie, Universita¨tsklinikum Giessen, Rudolf-Buchheim-Str 7, D-35392 Giessen, Germany. 0039-6060/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.surg.2004.03.003

988 SURGERY

alcohol is associated with significant morbidity and mortality.1 Although there still is substantial debate about the critical threshold, consumption of alcohol of more than 60 g/d for women and 80 g/d for men should definitely be considered harmful.2 A significant proportion of impaired health is due to liver diseases, gastroesophageal malignancies, suicide, and accidents.3 Even in Western countries, shortening of life expectancy by excessive alcohol consumption is estimated to be 5 years.4 In the surgical setting, excessive alcohol consumption may lead to potentially serious pre- and postoperative problems. A major issue is recognition and treatment of withdrawal symptoms, which in the case of a delirium tremens may be associated with admission to an intensive care unit (ICU) and increased mortality.5 However, no studies have tried to distinguish the effects attributable directly to excessive alcoholism without significant comorbidities from those that often are associated with alcohol consumption and may bear an

Surgery Volume 136, Number 5

increased surgical risk by themselves (eg, high-risk esophageal surgery). The intention of this study was to investigate the effect of excessive alcohol consumption (more than 30 g of alcohol per day) on surgical risk by correcting for concomitant diseases with the matched pairs method. MATERIAL AND METHODS The study is based on data sets of 28,065 patients having undergone noncardiac surgery at a tertiary care teaching hospital. Data acquisition was performed with an online, computerized anesthesia record-keeping system (NarkoData; IMESO GmbH, Hu¨ttenberg, Germany).6,7 Patients under the age of 18 were excluded from this study. The record-keeping system collects all data relevant to anesthesia during the procedure, including demographic data, administered drugs, laboratory results, vital data, and the data set for quality assurance, according to the German Society of Anesthesiology and Intensive Care Medicine.8 Systolic, diastolic, and mean arterial blood pressure and heart rate are recorded online at least every 5 minutes with noninvasive measurement or every 3 minutes with invasive measurement. Any drugs applied are entered manually at the moment of administration. All patient-related data collected during the preoperative ward round, informed consent of the patient, results of clinical examination, and additional investigations are recorded by the anesthesiologist in the electronic anesthesia record on the day preceding an elective procedure; data of emergency procedures are assessed immediately before the operation. On termination of the anesthesia procedure, files are imported into the database (Oracle 7; Oracle Corporation, Redwood Shores, Calif) after running through plausibility and integrity checks. All data fields used for this study were mandatory providing an improved data quality. All patients undergoing elective or urgent surgery are interviewed during the ward round about their alcohol consumption. As a part of the routine preparation for anesthesia, the number of alcohol drinks per day is recorded for every patient in the patient chart. One alcohol beverage is defined as either 1 bottle of beer (330 mL, 13 g alcohol), 1 glass of wine (200 mL, 13 g alcohol) or cider (330 mL, 13 g alcohol), or 1 glass of spirits (20 mL, 6.5 g alcohol). All patients were informed that correct answering of this question might have an impact on the performance of the anesthesiology procedure. Furthermore, in all patients, values of gamma-glutamyltransferase and mean erythrocyte

Klasen et al 989

volume were derived from the preoperative laboratory evaluation. We used the matched pairs method to evaluate the impact of history of excessive alcohol consumption on hospital mortality, length of hospital stay, admission to ICU, and incidences of intraoperative cardiovascular events. Cases were defined as patients with a history of excessive alcohol consumption (as defined previously). Matched controls were automatically selected among all patients of the described data pool over the study period according to matching variables without excessive alcohol consumption. Matching criteria included the following:  ASA physical status9  High-risk surgery (intracranial, thoracic, abdominal, and major vascular surgery) as defined by the Revised Cardiac Risk Score10  Urgency of surgery (elective, urgent [surgery within 6 hours after admission], emergency [immediate surgery in the next available theater])  Age  Gender

The selection of the matched controls was performed in a stepwise manner, by first attempting to match on the American Society of Anesthesiologists (ASA) physical status, then type of surgery, then urgency of surgery, then age, and finally gender; only 1 control was matched to each case. Hospital mortality, length of hospital stay, and admission to ICU were derived from the hospital information system. Prolonged length of hospital stay was considered if the patient was not discharged home after 21 days (value of the 75% quartile of the complete study population). Crude mortality ratio was the ratio of the hospital mortality rate in cases divided by the mortality rate in matched controls. Structured query language queries were used for retrospective detection of intraoperative cardiovascular events (hypotension, hypertension, bradycardia, and tachycardia) out of the database according to the definition of the German Society of Anaesthesiology and Intensive Care Medicine.8 Relevant cardiovascular events were defined as follows:  Hypotension: Decrease of mean arterial blood pressure >30% within a 10 minutes interval and administration of a vasoconstrictor or a positive inotropic drug within 20 minutes after beginning of the decrease (epinephrine, norepinephrine, dopamine, dobutamine, dopexamine, amezinium metilsulfate [Supratonin]), cafedrine/theodrenaline [Akrinor],

990 Klasen et al

enoximone, milrinone). Additional volume administration was not considered.  Hypertension: Increase of mean arterial blood pressure >30% within a 10 minutes interval and administration of an antihypertensive drug within 20 minutes after the beginning of the increase (nifedipine, urapidile, clonidine, hydralazine, droperidol, glyceryl trinitrate, sodium nitroprusside).  Bradycardia: Heart rate < 50 minÿ1 for at least 5 minutes and intravenous drug administration to increase heart rate within 15 minutes after the beginning of bradycardia (atropine, orciprenaline, ipratropium bromide, epinephrine, or pacemaker).  Tachycardia: Heart rate >100 minÿ1 for at least 5 minutes and intravenous drug administration to decrease heart rate within 15 minutes after the beginning of tachycardia (beta blocker, calcium antagonist, cardiac glycoside, sodium channel blocker, [Vaughan Williams class I], potassium channel blocker [Vaughan Williams class III], cardioversion, defibrillation).

Statistical analyses. For statistical evaluation, data were exported from the database into the SPSS statistics program (SPSS Software GmbH, Munich, Germany). Either v2 test or Fisher exact test was used to detect statistically significant differences between case patients and matched controls in outcome variables. Metric variables were compared with the nonparametric Mann-Whitney U test. The level of significance was set at P < .05. The power (1e b-failure) was determined with the use of the software GPOWER Version 2.0 (http://www.psycho. uni-duesseldorf.de/aap/projects/gpower/index. html).11 Since the patients in the case group and the matched control group were not assigned randomly with respect to a history of excessive alcohol consumption, we developed logistic regression models for all study patients, using the enter method to predict the impact of a history of excessive alcohol consumption on hospital mortality, prolonged length of hospital stay, admission to ICU, and incidence of intraoperative cardiovascular events. Independent variables in the 4 models included all matching criteria. Independent variables were analyzed as categories by using dummy variables for ASA physical status, high-risk surgery, urgency of surgery, and gender. Age was handled as a continuous variable. RESULTS In our data set, 928 of the 28.065 patients (3.3%) gave a history of excessive alcohol consumption. Matching was successful for 97% of the cases,

Surgery November 2004

leading to 897 pairs. The distribution of age and body mass index, as well as laboratory values known to be influenced by excessive alcohol consumption (mean corposcule volume and gammaglutamyltransferase) in cases and matched controls, are presented in Table I. Both mean corposcule volume and serum gamma-glutamyltransferase activity were greater (P < .0001) in study patients compared to controls. The distribution of the categorical matching variables in cases and matched controls is shown in Table II. All matching criteria were found to fit well (ASA classification, urgency and type of surgery) apart from gender. More case patients were of male gender compared to controls (89% vs 83%). The results of the main outcome measurements are presented in Table III. The total mortality was low in both groups (1.3% in case patients and 1.6% in matched controls, respectively). Thus, the crude mortality ratio of cases to controls was 0.81, a result that was not significant (P = .84). The number of cardiovascular events reached about 20% in both groups and did not differ significantly between cases and controls (P = .53). In both groups only a small proportion of patients (11% of cases and 9% of controls, respectively) was treated in the ICU postoperatively without intergroup differences (P = .27). The length of hospital stay was similar (P = .13). The logistic regression model for all patients revealed 3 variables to be associated with increased risk of an adverse outcome: ASA physical status, high-risk surgery, and urgency of surgery (Table IV). The odds ratio (OR) for mortality was 6.55 with a higher ASA physical status (P < .01); a high-risk surgical procedure increased the mortality 2.88fold (P = .01), whereas urgency of surgery led to an 1.86-fold increase of risk (P = .05). Apart from ASA classification (OR 2.13, P < .01), male gender led to an increased risk for prolonged hospital stay (OR 1.58, P = .03). The main risk factors for admission to ICU proved to be higher ASA physical status (OR 2.53, P < .001) and high-risk surgery (OR 4.12, P < .001). Older age and high-risk surgery were associated with an increased incidence of intraoperative cardiovascular events (P < .001). In highrisk procedures, the OR was 2.44, whereas increasing age only showed a minor effect (OR 1.04). DISCUSSION The Diagnostic and Statistical Manual of Mental Disorders’ definition of excessive alcoholism is accepted internationally;12 however, the effect of daily consumption of smaller amounts of alcohol on

Klasen et al 991

Surgery Volume 136, Number 5

Table I. Distribution of demographic variables and laboratory data in cases and matched controls, presented as MV, SD, and 95% CI with P value Case patients Variables Age (y) BMI (kg
m-2) MCV (fl) GGT (U/L)

MV ± SD

n 897 897 674 581

51.7 ± 12.7 25.5 ± 6.1 93.1 ± 6.3 58.3 ± 123

Matched controls CI

[50.8; [25.1; [92.7; [48.3;

MV ± SD

n

52.5] 25.9] 93.6] 68.3]

897 897 658 555

51.6 27.2 90.1 37.5

± ± ± ±

12.7 9.9 5.9 60.4

CI [50.8; [26.6; [89.7; [32.5;

52.5] 27.9] 90.6] 42.5]

P value

Power

0.97 0.001 0.001 0.001

0.95

BMI, Body-mass-index; MCV, mean corposcule volume; GGT, gamma-glutamyltranferase.

Table II. Distribution of the categorical matching variables in cases and matched controls Case patients Variable

n

%

Matched controls n

%

ASA physical status I 46 5% 46 5% II 449 50% 449 50% III 343 38% 343 38% IV 58 6% 58 6% V 1 0% 1 0% High-risk surgery (intracranial, thoracic, abdominal, and major vascular surgery) No 742 83% 742 83% Yes 155 17% 155 17% Urgency of surgery Elective 742 83% 742 83% Urgent 128 14% 128 14% Emergency 27 3% 27 3% Gender Female 103 11% 155 17% Male 794 89% 742 83%

physical and mental health remains controversial. Whereas a diet containing regular consumption of small amounts of alcohol seems to be beneficial,2 heavy alcohol ingestion is hazardous.13 In this view, drinking of more than 2 alcoholic beverages per day may be considered potentially hazardous. Excessive ingestion of large volumes of alcohol may lead to problems in the surgical setting. For this reason, most preoperative anesthesia questionnaires ask about drinking habits. Several problems may arise from excessive alcoholism: anesthesia-relevant morbidities associated with alcoholism (cardiomyopathy,14 liver cirrhosis,15 increased susceptibility for infectious diseases,16 malignancies,17 mental disorders, increased need for analgesic and sedative drugs,18 and withdrawal symptoms after discontinuation of alcohol).19 In the last few years, a number of published studies have tried to quantify the effect of heavy alcohol consumption on surgical mortality.20,21 These studies suggested that chronic consumption of large

Table III. Results of main outcome parameters in case and control patients Outcome variable Hospital mortality Case patients Matched controls CVE Case patients Matched controls ICU Case patients Matched controls LOS Case patients Matched controls

n

%

P value

Power

12 14

1.3 1.6

0.84

0.85

198 186

22 21

0.53

0.61

98 83

11 9

0.27

0.55

184 148

38 33

0.13

0.50

CVE, Cardiovascular adverse events; ICU, admission to the intensive care unit; LOS, prolonged length of hospital stay (>21 days).

amounts of alcohol may have a detrimental effect on surgical morbidity and mortality. However, none of the published studies have tried to quantify the effect of excessive alcohol consumption on surgical risk while correcting for relevant and concomitant comorbidities. Correction for existing comorbidities is best achieved with the ASA classification,9 which is clearly defined, easily evaluated, and internationally accepted; it provides excellent predictive power,22 and thus may be more practical and thereby superior even to more sophisticated systems. As alcohol drinking has a negative public image, heavy consumption is frequently denied. Compared to the gold standard for diagnosis of alcoholism,12 the proportion of patients in a recently published study who were correctly detected as alcoholics only by an anesthesia questionnaire was only 16%. This number was increased substantially by increasing the number of preoperative visits, adding an alcohol-specific questionnaire, and laboratory values.23 As our retrospective study in this large number of patients was conducted under clinical conditions, which do not allow for filling out a dedicated questionnaire, we probably may have

992 Klasen et al

Surgery November 2004

Table IV. Results of the logistic regression models using the 4 outcome measures as dependent and all matching criteria as independent variables with P value, odds ratio, and 95% CI P value

Odds ratio

Hospital mortality Variables ASA physical status High-risk surgery Urgency of surgery Age (y) Gender (male) Chronic alcohol consumption

< .01 .01 .05 .47 .44 .75

6.55 2.88 1.86 1.01 0.69 0.88

[3.20; [1.25; [1.01; [0.98; [0.27; [0.38;

13.4] 6.63] 3.44] 1.05] 1.77] 2.01]

Prolonged length of hospital stay (>21 days) Variables ASA physical status High-risk surgery Urgency of surgery Age (y) Gender (male) Chronic alcohol consumption

< .01 .60 .46 .71 .03 .19

2.13 1.10 0.90 1.00 1.58 1.20

[1.70; [0.77; [0.66; [0.99; [1.05; [0.91;

2.66] 1.56] 1.21] 1.01] 2.38] 1.59]

ICU admission Variables ASA physical status High-risk surgery Urgency of surgery Age (y) Gender (male) Chronic alcohol consumption

< .01 < .01 .71 .14 .12 .28

2.53 4.12 1.06 1.00 1.46 1.20

[1.95; [2.94; [0.77; [0.98; [0.90; [0.86;

3.29] 5.77] 1.46] 1.00] 2.37] 1.66]

Cardiovascular events Variables ASA physical status High-risk surgery Urgency of surgery Age (y) Gender (male) Chronic alcohol consumption

.07 < .01 .41 < .01 .65 .51

1.19 2.44 1.11 1.04 1.08 1.08

[0.99; [1.85; [0.87; [1.03; [0.77; [0.86;

1.43] 3.22] 1.43] 1.05] 1.52] 1.37]

missed a substantial proportion of alcoholics. In our sample, 3.3% of all patients were detected as excessive alcohol consumers only by the preoperative questionnaire. In a recent survey in a department of general medicine, 7.3% of patients were detected with an alcohol problem. However, only 25% of them had a documentation of excessive alcohol consumption in their patient chart.24 We thus may not have detected all alcoholdependent patients in our 28,000 patients. This factor, however, would only be a problem for the clinical utility of the study if the patients correctly detected as alcoholics would substantially differ in their characteristics from those not identified correctly as alcoholics by the questionnaire. We can find no such evidence in our study population. Nearly all of these patients (97%) were successfully matched to control cases, further improving clinical significance of our results.

CI

Furthermore although the routine anesthesia questionnaire does not ask for the duration of alcohol consumption, our data of liver enzyme and mean erythrocyte volume suggest that the duration was long enough to produce biochemical alterations in case patients typical for excessive alcohol consumption. However, there is no clear cutoff date from which alterations in biological markers of alcoholism can be expected.25,26 In our matched pairs analysis, none of the observed adverse outcomes was statistically different between cases and control patients. However, mortality in our patients was very low, and possibly differing results might be observed in even larger patient numbers. Statistical power of our observations was high (85%), which suggests that no clinically significant differences exist. There may be 3 possible explanations for our surprising results. First, the attending surgeons and anesthesiologists

Surgery Volume 136, Number 5

may have been aware of possible problems related to heavy alcohol consumption in these patients, leading to enhanced attention and early treatment of emerging problems. In the second place, heavy alcohol consumption possibly may only be hazardous in the long run. For surgical risk, identification of relevant comorbidities and adaptation of anesthesiologic strategy seem to be of paramount importance for an optimal outcome. Third, modern strategies of treatment of alcohol withdrawal symptoms in the ICU seem to be highly effective without marked sequelae.27 CONCLUSION Our study results suggest that excessive alcohol consumption is not a major risk factor for an adverse outcome in surgical patients. In contrast, our results show that the most important determinants for an increased risk of mortality, admission to ICU, and prolonged length of hospital stay seem to be a higher ASA classification and high-risk surgery. We would like to thank Moredata GmbH, Giessen, Germany, for their help with data management and statistical evaluation. REFERENCES 1. Malyutina S, Bobak M, Kurilovitch S, Gafarou V, Simonova G, Nikitin Y, Marmot M. Relation between heavy and binge drinking and all-cause and cardiovascular mortality in Novosibirsk, Russia: a prospective cohort study. Lancet 2002;360:1448-54. 2. Bondy SJ, Rehm J, Ashley MJ, Walsh G, Single E, Room R. Low-risk drinking guidelines: the scientific evidence. Can J Public Health 1999;90:264-70. 3. White IR, Altmann DR, Nanchahal K. Alcohol consumption and mortality: modelling risks for men and women at different ages. BMJ 2002;325:191. 4. Mathers CD, Vos ET, Stevenson CE, Begg SJ. The burden of disease and injury in Australia. Bull World Health Organ 2001;79:1076-84. 5. Spies CD, Nordmann A, Brummer G, Marks C, Conrad C, Berger G, et al. Intensive care unit stay is prolonged in chronic alcoholic men following tumor resection of the upper digestive tract. Acta Anaesthesiol Scand 1996;40: 649-56. 6. Benson M, Junger A, Quinzio L, Fuchs C, Sciuk G, Michel A, et al. Clinical and practical requirements of online software for anesthesia documentation an experience report. Int J Med Inf 2000;57:155-64. 7. Benson M, Junger A, Quinzio L, Michel A, Sciuk G, Fuchs C, et al. Data processing at the anesthesia workstation: from data entry to data presentation. Methods Inf Med 2000;39:319-24. 8. DGAI-Kommission ‘‘Qualita¨tssicherung und Datenverarbeitung in der Ana¨sthesie’’. Kerndatensatz Qualita¨tssicherung in der Ana¨sthesie. Ana¨sth Intensivmed 1993;34:331-5. 9. American Society of Anesthesiologists (ASA). New classification of physical status. Anesthesiology 1963;24:111.

Klasen et al 993

10. Lee TH, Marcantonio ER, Mangione CM, Thomas EJ, Polanczyk CA, Cook EF, et al. Derivation and prospective validation of a simple index for prediction of cardiac risk of major noncardiac surgery. Circulation 1999;100:1043-9. 11. Erdfelder E, Faul F, Buchner A. GPOWER: A general power analysis program. Behavior Research Methods. Instruments and Computers 1996;28:1-11. 12. American Psychiatric Association. Diagnostic and Statistical manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1995. 13. Corrao G, Bagnardi V, Zambon A, Arico S. Exploring the dose-response relationship between alcohol consumption and the risk of several alcohol-related conditions: a metaanalysis. Addiction 1999;94:1551-73. 14. Lee WK, Regan TJ. Alcoholic cardiomyopathy: is it dosedependent? Congest Heart Fail 2002;8:303-6. 15. Singer MV. Effect of ethanol and alcoholic beverages on the gastrointestinal tract in humans. Rom J Gastroenterol 2002;11:197-204. 16. Nelson S, Kolls JK. Alcohol, host defence and society. Nat Rev Immunol 2002;2:205-9. 17. Badger TM, Ronis MJ, Seitz HK, Albano E, IngelmanSundberg AM, Lieber CS. Alcohol metabolism: role in toxicity and carcinogenesis. Alcohol Clin Exp Res 2003;27:336-47. 18. Fassoulaki A, Farinotti R, Servin F, Desmonts JM. Chronic alcoholism increases the induction dose of propofol in humans. Anesth Analg 1993;77:553-6. 19. Rosenbaum M, McCarty T. Alcohol prescription by surgeons in the prevention and treatment of delirium tremens: historic and current practice. Gen Hosp Psychiatry 2002;24:257-9. 20. Neuenschwander AU, Pedersen JH, Krasnik M, Tonnesen H. Impaired postoperative outcome in chronic alcohol abusers after curative resection for lung cancer. Eur J Cardiothorac Surg 2002;22:287-91. 21. Tonnesen H, Kehlet H. Preoperative alcoholism and postoperative morbidity. Br J Surg 1999;86:869-74. 22. Junger A, Engel J, Quinzio L, Banzhaf A, Jost A, Hempelmann G. Risikoindizes, Scoring-Systeme und prognostische Modelle in der Ana¨sthesie und Intensivmedizin Teil I Ana¨sthesie. Ana¨sthesiol Intensivmed Notfallmed Schmerzther 2002;37:520-7. 23. Martin MJ, Heymann C, Neumann T, Schmidt L, Soost F, Mazurek B, et al. Preoperative evaluation of chronic alcoholics assessed for surgery of the upper digestive tract. Alcohol Clin Exp Res 2002;26:836-40. 24. Isaacson JH, Nielsen C, Urbanic R, Challgren E. Markers for Patients with Alcohol Problems in an Outpatient General Medicine Clinic. Subst Abus 1999;20:141-7. 25. Bataille V, Ruidavets JB, Arveiler D, Amouyel P, Ducimetiere P, Perret B, Ferrieres J. Joint use of clinical parameters, biological markers and CAGE questionnaire for the identification of heavy drinkers in a large population-based sample. Alcohol Alcohol 2003;38:121-7. 26. Hoffmeister H, Schelp FP, Mensink GB, Dietz E, Bohning D. The relationship between alcohol consumption, health indicators and mortality in the German population. Int J Epidemiol 1999;28:1066-72. 27. Spies CD, Dubisz N, Funk W, Blum S, Muller C, Rommelspacher H, et al. Prophylaxis of alcohol withdrawal syndrome in alcohol-dependent patients admitted to the intensive care unit after tumour resection. Br J Anaesth 1995;75:734-9.