EVALUATION OF A CLINICAL SCREENING INSTRUMENT TO IDENTIFY STATES OF THIAMINE DEFICIENCY IN INPATIENTS WITH SEVERE ALCOHOL DEPENDENCE SYNDROME

X. Sgouros1,*, M. Baines1, R. N. Bloor2, R. McAuley3, L. O. Ogundipe4 and S. Willmott5

1 Department of Clinical Chemistry, Royal Liverpool University Hospital, Liverpool, UK, 2 Academic Department of Psychiatry, Harplands Hospital, Stoke-on-Trent, UK, 3 ‘Edward Myers’ Substance Misuse Treatment Unit, Harplands Hospital, Stoke-on-Trent, UK, 4 ‘Stepping Stones’ Specialist Addiction Service, St. George's Hospital, Stafford, UK and 5 Department of Mathematics, Keele University, Staffordshire, UK

* Author to whom correspondence should be addressed at: Greenfield Community Mental Health Centre, Furlong Road, Tunstall, Stoke-on-Trent, ST6 5UD, UK. Tel.: 01782 425740; Fax: 01782 425741; E-mail: Xenofon.Sgouros{at}nsch-tr.wmids.nhs.uk

(Received 21 August 2003; first review notified 3 September 2003; in revised form 21 January 2004; accepted 10 February 2004)


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 REFERENCES
 
Aims: To develop a Thiamine Deficiency Questionnaire (TDQ), and to assess its reliability in the identification of Thiamine deficiency, in patients with severe alcohol dependence. Methods: 58 severely alcohol dependent patients underwent socio-demographic, medical, psychiatric, and alcohol use assessment, including administration of the Thiamine Deficiency Questionnaire (TDQ). The Red Blood Cell Thiamine Pyrophosphate concentration provided the ‘gold standard’ to test the validity of the instrument. Univariate 2 x 2 diagnostic test tables and multivariate analysis were performed. Results: A set of eight questionnaire items had an overall predictive power of 73.7%. Two of these were highly specific: ‘missed meals due to lack of funds’, and the clinical co-occurrence of medical conditions potentially related to poor nutrition. The Michigan Alcohol Screening Test and serum gamma glutamyl transferase were moderately predictive. Conclusions: Screening that combines socio-demographic, clinical and biological factors, and/or standardized questionnaires, could improve early recognition of thiamine deficiency.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 REFERENCES
 
Thiamine is present in all natural foods (Thurnham, 2000Go) and was the first member of the vitamin B complex to be chemically identified. Thiamine is absorbed mainly by the upper-intestine (Thurnham, 2000Go), and is rapidly converted predominantly to its active form, thiamine diphosphate (pyrophosphate, TPP), which functions as a co-enzyme in carbohydrate metabolism (Chaney, 2002Go). A serious deficiency of thiamine leads to neurological and cardiovascular disturbances that give rise to Wernicke–Korsakoff Syndrome (WKS), peripheral neuritis, and beriberi heart disease (Somogyi, 1976Go).

As a water-soluble vitamin, thiamine stores in the body are limited to ~30 mg with a biological half-life of 9–18 days (Ariaey-Nejad et al., 1970Go). In underdeveloped countries thiamine deficiency is generally the result of poor dietary practices, but in the UK and other developed countries it is most often related to chronic alcohol dependence syndrome (Leevy et al., 1982Go; Thomson et al., 1987Go).

The most dramatic complication of thiamine deficiency in the population with alcohol related problems is Wernicke's Encephalopathy (WE), which in its classic form is characterized by the triad of ocular abnormalities, ataxia, and a global confusional state (Victor, 1993Go). Unfortunately the classic triad is neither consistently nor frequently encountered, and the onset of the syndrome may be acute or gradual (Victor et al., 1989Go; Blansjaar et al., 1992Go). In a recent literature review, Thomson et al. (2002)Go concluded that since only about 10% of patients present with the classical triad of signs, we might be failing to make the diagnosis of WE in up to 90% of patients. This conclusion is supported by post-mortem and other studies (Harper et al., 1986Go, 1995Go; Cook et al., 1998Go).

A number of biochemical measurements, both direct and indirect, have been used to assess thiamine status as an aid to clinical diagnosis. Earlier methods, such as the measurement of blood pyruvate concentration (Morgan et al., 1968Go) were not specific, and indirect or functional methods, such as the measurement of the activity of the erythrocyte enzyme transketolase (ETKA) or the thiamine pyrophosphate effect (TPP effect; Warnock, 1975Go) may be influenced by factors other than thiamine deficiency (Baines, 1985Go; Baines et al., 1988Go). In this study, we have used direct high performance liquid chromatographic (HPLC) measurement of the principal physiological form of thiamine, thiamine pyrophosphate, in erythrocytes, a tissue which has been shown to be a good indicator of body stores (Brin, 1964Go).

Diagnosis of WE remains a difficult medical task because it is mainly based on clinical findings, complemented by laboratory testing, and is confirmed only by post-mortem histo-pathological examination (Victor et al., 1989Go; Victor, 1993Go). However, failure to diagnose WE can result to serious amnesic syndrome (Korsakoff Syndrome, KS), or irreversible brain damage, or death (Victor et al., 1989Go; Victor, 1993Go). Early diagnosis, and treatment of WE can prevent the development of KS. Unfortunately, a standardized instrument for early detection of thiamine deficiency related to alcohol dependence is lacking. Thus, it is desirable to obtain a standardized questionnaire, which could be used as a screening tool in clinical settings for the identification of patients with alcohol-related disorders and likely associated thiamine deficiency. This instrument would enhance the clinician's ability to diagnose such a diverse, and sometimes insidious, condition as WE, promptly.

This paper reports on the development of such an instrument, as well as evaluating its predictive capability to identify the state of thiamine deficiency in patients with alcohol dependence syndrome. Thiamine deficiency is not an isolated feature of malnutrition related to alcohol dependence (Thomson et al., 1987Go; Bunout, 1999Go) and it can be hypothesized that a questionnaire detecting a generalized nutritional deficiency may be used as a proxy measure to the thiamine deficiency associated with alcohol dependence. This type of instrument could complement clinical and laboratory testing, in early diagnosis of patients at risk for WE.


    SUBJECTS AND METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 REFERENCES
 
Study population
Patients admitted to the inpatient unit of the local Substance Misuse Department for detoxification, who were between 18–65 years of age and had a history of either Harmful Use or Dependence Syndrome due to alcohol, as defined in ICD-10 (World Health Organisation, 1994Go) were invited to take part in the study. Patients who were suffering from diabetes mellitus, or other medical conditions such as end-stage cancer or chronic renal failure that have been associated with development of thiamine deficiency were excluded (Saito et al., 1987Go; Hung et al., 2001Go; Togay et al., 2001Go; Ogershock et al., 2002Go). Diagnosis of diabetes mellitus was determined by World Health Organisation criteria (Oxford Handbook of Clinical Medicine, 2001Go). We excluded patients who had a history of anaphylactic shock due to IV administration of vitamin B1, or any other drug, and also patients suffering from any medical or psychological condition which in the opinion of the investigator should preclude entry to the study.

Development of the Thiamine Deficiency Questionnaire
Literature was reviewed to identify important nutritional and clinical parameters. We developed a Thiamine Deficiency Questionnaire (TDQ), consisting of 16 items (Appendix A). The Nutritional Information (NI) questions were adopted from a nutrition-screening tool being used with clients with learning disabilities and mental health problems (Bryan et al., 1998Go).

Methodology
The Local Research Ethics Committee approved the protocol. Three psychiatrists in training and a staff grade psychiatrist, all with at least 2 years of experience in the assessment and management of psychiatric patients, completed interviews and medical examinations, supervised by the principal investigator.

Each participant signed informed consent and completed a questionnaire on socio-economic status, lifestyle, alcohol use history, and self-reported alcohol consumption during a typical week. All subjects, within 48 h of entering the study, received a full medical and psychiatric evaluation, which followed a structured format developed by the investigators, and included Medical and Psychiatric history, Mental State Examination, Physical and Neurological examination, Mini Mental State Examination (MMSE), Severity of Alcohol Dependence Questionnaire (SADQ; Stockwell et al., 1983Go), and Michigan Alcohol Screening Test (MAST; Zung et al., 1975). The TDQ was administered at that stage. Height and weight, and blood alcohol concentration, measured with a standard breath analyser, were recorded on admission to the unit. Body Mass Index (BMI) was calculated as weight (kg) divided by squared height (m2).

Fifteen subjects (25% of sample population) presented on admission with symptoms and signs of alcohol intoxication and/or had blood alcohol concentration >0.035 mg/100 ml. These patients were thought to be incapable of giving informed consent on admission to the unit. Consent was obtained when symptoms and laboratory evidence of intoxication subsided, which was within 6 h of the initiation of treatment. We routinely perform the laboratory investigations presented below, apart from measurement of Red Blood Cell Thiamine Pyrophosphate (RBCTPP), on all patients admitted to the inpatient unit for detoxification. Blood for measurement of RBCTPP from these fifteen subjects was stored until final informed consent was obtained.

Blood analyses and treatment with parenteral thiamine
Blood samples were obtained on admission, before the beginning of treatment of the withdrawal syndrome, for measurement of RBCTPP, serum electrolytes (Na+, K+ and Ca++), urea and creatinine, liver function tests (alanine aminotransferase, ALT; aspartate aminotransferase, AST; alkaline phosphatase; total bilirubin; gamma-glutamyl transferase, GGT), non-fasting blood glucose, and a full blood count. Urine samples for measurement of glucose were also collected at that stage.

The RBCTPP concentration before initiation of treatment provided the ‘gold standard’ to judge the criterion validity of the new instrument. RBCTPP was measured in washed red cells by the HPLC method of Baines (Baines, 1985Go). The reference range was 165–286 nmol/l and the inter-batch precision was 5.7% at a concentration of 223 nmol/l.

Assessment of withdrawal symptoms was carried out on admission, and subsequently 6-hourly, by a clinical interview and administration of the Addiction Research Foundation Clinical Institute Withdrawal Assessment for Alcohol (CIWA-Ar; Sullivan et al., 1989Go; Stuppaeck et al., 1994). Treatment of alcohol withdrawal, with diazepam and symptomatic medication, began for all participants within 2 h after admission and followed a standard protocol. All subjects also received two pairs of Pabrinex® intravenous (IV) high potency multi-vitamin injection, by IV infusion over 15–30 min in 100 ml of 0.9% sodium chloride, every 8 h for a period of 48 h. On completion of the Pabrinex administration, a second blood sample for measurement of RBCTPP was obtained.

Statistical Analysis
A database of 26 variables containing demographic details, laboratory measurements, and relevant questionnaire scores was created. Demographic variables were age, gender, marital and occupational status; laboratory variables were GGT, mean red cell volume (MCV), and pre- and post-treatment RBCTPP. Individual TDQ item scores, sub-category scores as well as total questionnaire scores (SADQ, MAST) were included. All statistical computations including descriptive statistics were carried out using SPSS for Windows (SPSS, 1999Go).

Subjects were identified as Cases of Thiamine Deficiency (CTD) if their pre-treatment RBCTPP values were <165 nmol/l. Non-Cases of Thiamine Deficiency (NCTD), had RBCTPP values ≥165 nmol/l. The univariate 2 x 2 diagnostic test table for each variable was computed so that individual variable sensitivity, specificity, Positive Predictive Value (PPV) and Negative Predictive Value (NPV) could be calculated. The mid-point score was used as the cut-off point to identify cases of thiamine deficiency, for all TDQ items apart from item 14 (see Appendix A).

For multivariate analysis, a logistic regression was applied to the database to select variables that were significantly associated with caseness. This type of regression has been shown by Rosenberg et al. (1998) to be successful in selecting significant groups of variables for predicting cases. The Receiver Operating Characteristic (ROC) curve (Altman, 1991Go; Cloves, 1999Go) was used to determine the predictive power of various combinations of significant variables as chosen by the regression.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 REFERENCES
 
Sixty subjects entered the study. Two were found to have diabetes mellitus and were excluded from the final analysis. There were no participants with a history of anaphylactic shock, or suffering from cancer or chronic renal failure, or any other medical or psychiatric condition that precluded entry to the study. The remaining 58 patients were between 21–65 years old, with average age of 42.36 (SD ± 11.65) years, mostly males (62.7%) and unemployed (88.1%). Only 31% of the patients were married or living with a partner, and 50% were living alone. All patients reported more than 10 years of alcohol misuse and related problems. They consumed 40–952 units (1 unit equivalent to 8 g of ethanol) of alcohol per week (mean: 228.16 ± 182.90 SD), in the last month before admission, mainly in a regular pattern of use. Fifty of our subjects (86.2%) presented with clinical and/or laboratory evidence of liver dysfunction [presence of abdominal pain and tenderness and/or hepatomegaly and/or splenomegaly and/or jaundice and/or abnormal Liver function tests (ALT > 40 u/l, AST > 40 u/l, GGT > 50 u/l, alkaline phosphatase >100 u/l, or total bilirubin > 17 µmol/l)] but none had ascites or other clinical signs of cirrhosis (skin manifestations or ecchymotic lesions).

Only 13.8% of the patients could be categorized as being ‘underweight’ (BMI < 20), with 34.5% having a BMI within recognized normal limits (20–24) for their age and height, and 51.7% a BMI > 24. The distribution of SADQ and MAST scores indicated a severe level of dependence in our study cohort with mean scores of 34.9 (SD ± 12.97) and 36.1 (SD ± 8.74) respectively. Table 1 shows some descriptive statistics of demographic, laboratory and questionnaire variables. A ranking of variables/items based on their sensitivity and specificity with corresponding Positive Predictive Value (PPV) and Negative Predictive Value (NPV) is shown in Table 2. Table 3 contains the computational detail and results of logistic regression.


View this table:
[in this window]
[in a new window]
 
Table 1. Summary statistics of variables

 

View this table:
[in this window]
[in a new window]
 
Table 2. TDQ items with highest Specificity and Sensitivity with corresponding Positive Predictive Value (PPV) and Negative Predictive Value (NPV)

 

View this table:
[in this window]
[in a new window]
 
Table 3. Logistic Regression for questionnaire items and laboratory measurements that best identify cases

 
Thirty-one subjects (53.4%) had pre-treatment RBCTPP values <165 nmol/l and hence were identified as cases of thiamine deficiency. Only 11 subjects (19.0%) were identified as cases after the treatment with Pabrinex. To confirm that the change of RBCTPP levels between pre- and post-treatment is significant the paired samples t-test was applied to the RBCTPP values for the pre- and post-treatment. The two-tailed significance was <0.001 (P < 0.001, df = 57), indicating that the change of RBCTPP levels between pre- and post-treatment is significant. However, eight of the subjects had post-treatment RBCTPP values lower than their pre-treatment values, with five actually having pre-treatment values within the reference normal range, which became ‘abnormal’ after treatment with Pabrinex (see Table 4). The latter subjects were between 23–42 years old, had a higher ratio of females (3/5) than in the total sample, and had relatively high serum GGT activity (mean 540.80 UI/l) and MCV (mean 101.14 fl) values. Two of these subjects had BMI values > 30. Mean TDQ and MAST total scores (68.80 and 35.20 respectively) were similar to the means on these scales in the total sample.


View this table:
[in this window]
[in a new window]
 
Table 4. Pre- and Post Treatment RBCTPP values in subjects with lower post-treatment values (reference range 165–286 nmol/l)

 
A comparison between Table 2 and Table 3 highlighted two main groups of variables selected by the logistic regression. The first group contains variables such as TDQ items 1, 4 and 8 that have high sensitivity while the second group contains variables such as TDQ items 5, 14 and BMI that have high specificity. The remaining items/variables were selected to enhance the predictive power. If area under the ROC curve is used to represent the predictive power of the TDQ items and variables as summarized in Table 5, it may be seen that the combination between the selected TDQ items and measured variables provides high predictive power. For example, the overall combination of all selected items and variables has predictive power of 0.837 (83.7%). We notice however, that if only the TDQ items were used to predict cases, the predictive power (0.737 or 73.7%) is higher than the predictive power (0.697 or 69.7%) of the measurement variables (BMI and GGT) combined with MAST and CE sub-score. This result implies that further development and refinement of TDQ may lead to an effective screening tool with high predictive power.


View this table:
[in this window]
[in a new window]
 
Table 5. ROC curve areas for different combinations of variables

 

    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 REFERENCES
 
The Thiamine Deficiency Questionnaire (TDQ) has been developed to identify patients with alcohol related conditions who are also nutritionally deficient, and might require treatment with parenteral thiamine in clinical settings. The importance of early treatment of thiamine deficiency was emphasised in the Royal College of Physicians recent report entitled ‘Alcohol—can the NHS afford it?’ which also contains recommendations for the treatment of WE in the Accident and Emergency department (Report of Royal College of Physicians, 2001Go, Appendix 3). The current widely accepted practice, nationally and internationally, is to administer parenteral (usually intravenous) replacement therapy to all high-risk patients undergoing treatment of alcohol withdrawal syndrome in inpatient settings, especially if they show signs of chronic malnutrition (Report of Royal College of Physicians, 2001Go, Appendix 3; Thomson et al., 2002Go).

Although this ‘treat all’ policy has many advantages, there are drawbacks. The albeit small risk of anaphylaxis is well documented and can be fatal (Stephen et al., 1992Go; British National Formulary (BNF), 2003Go). A ‘treat all’ policy can be expensive, since intravenous administration of any drug is a logistically demanding clinical procedure that also causes discomfort to many patients. The BNF, recognising the risk of anaphylaxis, advises, ‘use be restricted to patients in whom parenteral treatment is essential’. There is neither a clinical assessment protocol nor a laboratory investigation method that can, rapidly, predict the need for treatment with Pabrinex. Our study demonstrates that a standardized questionnaire instrument could be invaluable in assisting clinicians in treatment decisions.

The RBCTPP concentration before initiation of treatment provided the ‘gold standard’ to judge the criterion validity of the new instrument. Thirty-one subjects (53.4%) were identified as cases of thiamine deficiency, which is not dissimilar to the percentage that had been previously reported in the literature (30–80%; Thomson et al., 1987Go). In a recent paper, Mancinelli et al. (2003)Go described an HPLC method for the simultaneous determination of thiamine, and the mono- and diphosphates in red cells. Other such methods have been previously published (Tallaksen et al., 1991Go, 1993Go; Herve et al., 1994Go). Whilst these studies are interesting, and provide research data helpful in the understanding of thiamine utility, they are at present research methods and not generally available for routine use. Further, they confirm the pivotal role that RBCTPP has as an indicator of thiamine status, though studies using methods such as the above may provide data on the handling of thiamine by the alcoholic patient. However, we believe that RBCTPP will remain the primary biochemical diagnostic test of thiamine status.

In our results, selected items with high sensitivity and/or specificity, grouped together, can predict thiamine deficiency in ~80% of cases. Nutritional and clinical examination items had higher predictive capability than alcohol use variables (apart from frequency of alcohol use), which suggests that malnutrition contributed more to the development of vitamin deficiency in our population than their drinking per se. Item 5 (missed meals due to lack of funds) and item 14 (co-occurrence of other nutritional related conditions) had particularly high Positive Predictive Values (PPV, 100% and 73.33% respectively). These items could be useful minimal assessment tool for clinicians who assess patients in emergency or other settings, since they can fairly confidently detect a high proportion of true thiamine-deficient cases.

The high specificity of a low BMI possibly reflects its distribution in our sample (52% had a BMI > 24).

Further refinement and evaluation of the TDQ are needed to improve the instrument's validity. In particular, the generalizability of the instrument requires further investigation. Our sample came from a population with a long history of heavy drinking with severe social and occupational dysfunction. It would be important to test this instrument in a larger and more varied population such as patients attending Accident and Emergency Departments.

Logistic regression was used to regress our independent variables toward the binary outcome. We used the total MAST score as one of the independent variables, as shown in Table 3. Although MAST is a detailed instrument compared to individual TDQ items (see Table 3), it has less statistical significance than these items, when it comes to the prediction of thiamine-deficient cases. This is understandable since MAST was not intended for this. The total MAST score was selected by multivariate analysis as a moderately strong factor and it would be of interest to combine selected items of the MAST with selected items of the TDQ. Taking into account that most of the items of the latter are denoting self-report judgements, it would seem more appropriate to add or replace items with MAST items that are based on more objective evidence. Skinner (1979)Go, in his multivariate evaluation of the MAST, identified five important dimensions (subsets of items). In the study's factor analysis, the second subset of items (Factor II: legal, work and social problems) had internal consistency reliability of 0.76 and eigenvalue of 3.71, which were second only to the Factor I (recognition of alcohol problems by self or others) values. All these items denote specific, objective, alcohol-related events and consequences (e.g. arrests, fights, occupational or marital difficulties, other social events, delirium tremens events) and can be incorporated into the TDQ.

GGT was also selected by logistic regression as a moderately strong factor. GGT is a routinely used laboratory marker of liver dysfunction, and is also a widely used marker of the severity of alcohol dependence (Reynaud et al., 2000Go; Conigrave et al., 2002Go). Patients who present with high values may present with more severe liver pathology, which may lead in reduced ability to store vitamins. Reduced hepatic storage has been identified as one of the mechanisms of thiamine deficiency in chronic alcoholism (Hoyumpa, 1980Go) and therefore this result may indicate a significant effect of alcoholic liver disease on thiamine levels.

Treatment was effective in alleviating thiamine deficiency (P < 0.0001), which may reflect a clinically significant treatment effect. Relatively high doses of Pabrinex were used. Nevertheless, a small number of the subjects had post-treatment RBCTPP lower than their pre-treatment values, with five actually having post-treatment values lower than the reference normal range. There are no demographic, physical or mental health, nutritional, or alcohol use factors, in our sample that appeared to explain this. The precision of the HPLC method used for determination of RPCTPP values (inter-batch precision 5.7%) should not be a factor (see Table 4). Moreover, all participants in the study received the same dose of thiamine, other vitamins, and anhydrous glucose during parenteral treatment with Pabrinex. They also had the same daily number of meals during the treatment. This finding poses a puzzling question which needs further investigation in a larger sample, because it may be an isolated finding. Possible explanations might be a defect with the uptake of the parenteral thiamine by the red cell, or defects with the phosphorylation of the administered thiamine within the cell. Another might be that ‘rebound’ erythropoiesis occurs on instigation of treatment as the suppressive effect of alcohol is removed (Hourihane et al., 1970Go), but there is no evidence of this in our results. Finally, these subjects might be ‘fast utilizers’ of thiamine, a concept that has not been sufficiently investigated in the literature.

In conclusion, the development of a screening tool that will enable clinicians to quickly and correctly identify alcohol dependent patients at risk of thiamine deficiency, and associated WE, is certainly desirable. This evaluation of a new questionnaire demonstrates that early recognition of thiamine deficiency with a standardized tool is possible. Further evaluation and refinement of the tool is necessary in order to determine its usefulness and validity when used in different population samples and clinical settings.


    APPENDIX A
 TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 REFERENCES
 
A brief description of the TDQ items:

  1. Frequency of eating carbohydrates (1a), or fruit and vegetables (1b), or meat, fish or poultry (1c) in the past year (per day)
  2. Frequency of meal completion
  3. Weight loss in the past year (in kg)
  4. Weight gain in the past year (in kg)
  5. Frequency of missed meals because of lack of funds (per week)
  6. Frequency of reduced appetite with avoidance of meals (per month)
  7. Frequency of eating breakfast (per week)
  8. Frequency of alcohol use
  9. Units of alcohol in a typical drinking day
  10. Frequency of having 6 or more units of alcohol on one occasion
  11. Frequency of occurrence of ‘blackouts’ (defined as transient circumscribed memory deficit episodes that occur during periods of severe intoxication)
  12. Frequency of early morning drinking due to withdrawal symptoms
  13. Total number of episodes of vomiting in the past month
  14. Co-occurrence of other nutritional-related conditions (poluneuropathy, amblyopia, pellagra, anaemia or subacute combined degeneration, yes or no)
  15. General Clinical Impression about nutritional status scale (based on clinical judgement)
  16. Body Mass Index

Nutritional Information Variables (NI): Items 1–7, range 0–3, cut-off scores: Items 1 and 4 ≤2, Items 3, 5–7 ≥2).

Alcohol Use Variables (AU): Items 8–12, range 0–4, cut-off score ≥2.

Clinical Examination Variables (CE):Items 13–15 (Item 13, range 0–3, cut-off score ≥2), (item 14, rated as either 0 or 3, cut-off score = 3), (item 15, range 0–4, cut-off score ≥2). Total score = BMI plus NI, AU and CE sub-scores.


    ACKNOWLEDGEMENTS
 
The authors wish to thank Dr I. Aggel, Dr A. Boardman and Dr A. Schule for their invaluable assistance during data collection. The TDQ was developed in collaboration with Mrs F. Bryan, Department of Nutrition and Dietetics, North Staffordshire University Hospital NHS Trust. We acknowledge the invaluable assistance of Professor I. Crome who edited the first draft manuscript of our paper. We wish to thank the nursing staff of ‘Edward Myers’ Substance Misuse Unit, Harplands Hospital, Stoke-on-Trent for their support during the data collection. We thank all the patients who kindly agreed to participate, and made this study possible.

Statement—Dr X. Sgouros was responsible for the design, execution, data collection and interpretation, and preparation of the manuscript. Dr M. Baines helped with the design of the project, biochemical analysis, data interpretation and preparation of the manuscript. The study framework was derived from Dr Bloor's original idea, who also was the overall advisor of the project. Dr McAuley was involved in the protocol design, execution and data collection stage and Dr L. O. Ogundipe helped with protocol design, and also submitted the project for ethical approval. Dr Willmott performed the statistical analysis and was involved in the preparation of the manuscript.


    FOOTNOTES
 
This study was conducted in the ‘Edward Myers’ Substance Misuse Treatment Unit, Harplands Hospital, Stoke-on-Trent, UK


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 REFERENCES
 
Altman, D. G. (1991) Practical Statistics for Medical Research, Chapman & Hall, London, UK, p. 611.

Ariaey-Nejad, M. R., Balaghi, M., Baker, E. M. and Sauberlich, H. E. (1970) Thiamine Metabolism in Man. American Journal of Clinical Nutrition 23, 764–778.[ISI][Medline]

Baines, M. (1985) Improved high performance liquid chromatographic determination of thiamine diphosphate in erythrocytes. Clinical Chemistry Acta 153, 43–48.[CrossRef][ISI][Medline]

Baines, M. and Davies, G. (1988) The evaluation of erythrocyte thiamine diphosphate as an indicator of thiamine status in man, and its comparison with erythrocyte transketolase activity measurements. Annals of Clinical Biochemistry 25, 698–705.[ISI][Medline]

Blansjaar, B. A. and Van Dijk, J. C. (1992) Korsakoff–Wernicke syndrome. Alcohol and Alcoholism 27, 435–437.[Abstract]

Brin, M. (1964) Erythrocyte as a biopsy tissue for functional evaluation of thiamine adequacy. Journal of American Medical Association 187, 762–766.[ISI]

British National Formulary (BNF), 45th edition, March 2003, p. 469.

Bryan, F., Jones, J. M. and Russell, L. (1998) Reliability and validity of a nutrition-screening tool to be used with clients with learning difficulties. Journal of Human Nutrition and Dietetics 11, 41–50.[CrossRef][ISI]

Bunout, D. (1999) Nutritional and metabolic effects of alcoholism: their relationship with alcoholic liver disease. Nutrition 15, 583–589.[CrossRef][ISI][Medline]

Chaney, S. G. (2002) Energy releasing water soluble vitamins. In Textbook of Biochemistry with Clinical Correlations, 5th edn, T. M. Devlin (ed.) pp. 1148–1150, Willey–Liss, London.

Cloves, M. (1999) Receiver Operating Characteristic (ROC) analysis. Stata Technical Bulletin 52, 19–33.

Conigrave, K. M., Degenhardt, L. J., Whitfield, J. B., Saunders, J. B., Helander, A. and Tabakoff, B. (2002) CDT, GGT and AST as markers of alcohol use: the WHO/ISBRA collaborative project. Alcoholism: Clinical and Experimental Research 26, 332–339.[ISI][Medline]

Cook, C. C. H., Hallwood, P. M. and Thomson, A. D. (1998) B-vitamin deficiency and neuro-psychiatric syndromes in alcohol misuse. Alcohol and Alcoholism 33, 317–336.[Abstract]

Harper, C. G., Giles, M. and Finlay-Jones, R. (1986) Clinical signs in the Wernicke–Korsakoff complex: a retrospective analysis of 131 cases diagnosed at necropsy. Journal of Neurology, Neurosurgery and Psychiatry 49, 341–345.[Abstract]

Harper, C., Fornes, P., Duyckaerts, C., Lecomte, D. and Hauw, J.-J. (1995) An international perspective on the prevalence of the Wernicke–Korsakoff syndrome. Metabolic Brain Disease 10, 17–24.[ISI][Medline]

Herve, C., Beyne, P. and Delacoux, E. (1994) Determination of thiamine and its phosphate esters in human erythrocytes by high-performance liquid chromatography with isocratic solution. Journal of Chromatography B: Biomedical Applications 653, 217–220.[CrossRef][ISI]

Hourihane, D. O. and Weir, D. G. (1970) Suppression of erythropoiesis by alcohol. British Medical Journal 1, 86–89.[ISI][Medline]

Hoyumpa, A. M. (1980) Mechanisms of thiamine deficiency in chronic alcoholism. American Journal of Clinical Nutrition 33, 2750–2761.[Abstract]

Hung, S. C., Hung, S. H., Tarng, D. C, Yang, W. C., Chen, T. W. and Huang, T. P. (2001) Thiamine Deficiency and unexplained encephalopathy in haemodialysis and peritoneal dialysis patients. American Journal of Kidney Diseases 38, 941–947.[ISI][Medline]

Leevy, C. M. (1982) Thiamine deficiency and Alcoholism. Annals of New York Academy of Sciences 378, 316–326.[ISI][Medline]

Mancinelli, R., Ceccanti, M., Guiducci, M. S., Sasso, G. F., Sebastiani, G., Attilia, M. L. and Allen, J. P. (2003) Simultaneous liquid chromatographic assessment of thiamine, thiamine monophosphate and thiamine diphosphate in human erythrocytes: a study in alcoholics. Journal of Chromatography B 789, 355–363.[CrossRef]

Morgan, H. G. (1968) Acute neuropsychiatric complications in chronic alcoholism. British Journal of Psychiatry 114, 85–92.[ISI][Medline]

Ogershock, P. R., Rahman, A., Nestor, S., and Brick, J. (2002) Wernicke's encephalopathy in non-alcoholic patients. American Journal of Medical Sciences 323, 107–111.[ISI]

Oxford Handbook of Clinical Medicine, 5th edn (2001) Longmore M., Wilkinson I., Török E. (eds), pp 282–283, Oxford University Press, New York.

Reynaud, M. Schellenberg, F., Loisequx-Meunier, M. N., Schwan, R., Maradeix, B., Planche, F. and Gillet, C. (2000) Objective diagnosis of alcohol abuse: compared values of carbohydrate-deficient transferrin, gamma-glutamyl transferase, and mean corpuscular volume. Alcoholism: Clinical and Experimental Research 24, 1414–1419.[ISI][Medline]

Royal College of Physicians (2001) Report of a working party: Alcohol—can the NHS afford it? Recommendations for a coherent alcohol strategy for hospitals, Royal College of Physicians, London.

Saito, N., Kimura, M., Kuchiba, A. and Itokawa, Y. (1987) Blood Thiamine Levels in outpatients with Diabetes Mellitus. Journal of Nutritional Sciences and Vitaminology 33, 421–430.

Skinner, H. A. (1979) A multivariate evaluation of the MAST. Journal of Studies on Alcohol 40, 831–844.[ISI][Medline]

Somogyi, J. C. (1976) Early signs of Thiamine deficiency. Bibliotheca Nutrition and Dieta 23, 78–85.

SPSS Inc. (1999) SPSS Software: Release 9.0 Chicago, IL 60611, USA.

Stephen, J. M., Grant, R. and Yeh, C. S. (1992) Anaphylaxis from administration of intravenous thiamine. American Journal of Emergency Medicine 10, 61–63.[ISI][Medline]

Stockwell, T., Murphy, D. and Hodgson, R. (1983) The severity of alcohol dependence questionnaire: its use, reliability and validity. British Journal of Addiction 78, 145–155.[ISI][Medline]

Stuppaeck, C. H., Barnas, C., Falk, M., Guenther, V. and Hummer, M. (1992) Assessment of the alcohol withdrawal syndrome: Validity and reliability for the translated and modified Clinical Institute Withdrawal Assessment for Alcohol scale (CIWA-A). Addiction 89, 1287–1292.

Sullivan, J. T., Sykora, K., Schneiderman, J., Naranjo, C. A. and Sellers, E. M. (1989) Assessment of alcohol withdrawal: The revised Clinical Institute Withdrawal Assessment for Alcohol scale (CIWA-AR). British Journal of Addiction 84, 1353–1357.[ISI][Medline]

Tallaksen, C. M. E., Böhmer, T., Bell, H. and Karlsen, J (1991) Concomitant determination of thiamine and its phosphate esters in human blood and serum by high-performance liquid chromatography. Journal of Chromatography A 564, 127–136.[CrossRef][ISI]

Tallaksen, C. M. E., Sande, A., Böhmer, T., Bell, H. and Karlsen, J. (1993) Kinetics of thiamine and thiamine phosphate esters in human blood, plasma and urine after 50 mg intravenously or orally. European Journal of Clinical Pharmacology 44, 73–78.[ISI][Medline]

Thomson, A. D., Jeyasingham, M. D., Pratt, O. E. and Shaw, G. K. (1987) Nutrition and alcoholic encephalopathies. Acta Medica Scandinavica Suppl. 717, 55–65.

Thomson, A. D., Cook, C. C. H., Touquet, R. and Henry, J. A. (2002) The Royal College of Physicians report on alcohol: Guidelines for managing Wernicke's Encephalopathy in the Accident and Emergency Department. Alcohol and Alcoholism 37, 513–521.[Abstract/Free Full Text]

Thurnham, D. I. (2000) Water Soluble Vitamins In Human Nutrition and Dietetics, 10th edn, Ganrow, J. S., James, W. P. T. and Ralph, A. (eds) pp. 257–162, Churchill–Livingstone, Edinburgh.

Togay, I., Yi, C. and Mutluer, N. (2001) Wernicke's Encephalopathy due to hyperemesis gravidarum:an under-recognised condition. Australian and New Zealand Journal of Obstetrics and Gynaecology 41, 453–456.[Medline]

Warnock, L. G. (1975) Transketolase activity of blood haemolysate, a useful index for diagnosing thiamine deficiency. Clinical Chemistry 21, 432–436.[Free Full Text]

World Health Organisation (1994) Mental and Behavioural Disorders due to Psychoactive Substance Abuse, in the Pocket Guide to the ICD-10 classification of Mental and Behavioural disorders with Glossary and Diagnostic Criteria for Research (ICD-10: DCR-10), Cooper J. E. (ed.), pp. 63–91, American Psychiatric Press.

Victor, M., Adams, R. D., and Collins, G. H. (1989) The Wernicke–Korsakoff syndrome and related neurologic disorders due to Alcoholism and Malnutrition, 2nd edn, F.A. Davis, Philadelphia.

Victor, M. (1993) The Wernicke–Korsakoff syndrome. In Principles of Neurology, 5th Int. edn, Adams R. D. and Victor M. (eds), pp. 851–858, McGraw-Hill, New York.