Alcohol and Alcoholism Vol. 34, No. 5, pp. 786-794, 1999
© 1999 Medical Council on Alcoholism
ESTIMATING PREVALENCE OF ALCOHOL ABUSE AND DEPENDENCE IN ONE GENERAL HOSPITAL: AN APPROACH TO REDUCE SAMPLE SELECTION BIAS
University Greifswald, Medical Faculty, Institute of Epidemiology and Social Medicine, Walther-Rathenau-Str. 48, D-17487 Greifswald and
1 Medical University of Lübeck, Department of Psychiatry, Ratzeburger Allee 160, D-23538 Lübeck, Germany
Received 20 August 1998; first review notified 4 February 1999; accepted 22 March 1999
 |
ABSTRACT |
|---|
 TOPFOOTNOTES ABSTRACT INTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Prevalence estimates of alcohol abuse or dependence in general hospitals are often limited to single wards, small data collecting periods or insufficient diagnostic procedures. Therefore, the present study aimed to ascertain alcohol abuse or dependence in one general hospital, to compare prevalence data for all the 11 wards and 6 intake months, to establish if screening is sufficient or if a two-step diagnostic procedure is needed, and to determine whether information for an alcohol diagnosis on suspicion is available. A sample of 1309 medical or surgical in-patients were screened by questionnaires or medication for withdrawal, and, if screening-positive, were interviewed with the alcohol section of a standardized psychiatric interview. In screening-negative patients, a diagnosis on suspicion was given if medication to treat withdrawal had been used, or if there was evidence of single criteria of alcohol dependence, somatic disorders from alcohol drinking, raised laboratory parameters on grounds of alcohol drinking or of self-reported high alcohol consumption. Of the medical and surgical in-patients, 20.7 and 16.0% respectively were alcohol abusers or dependants, with a range of prevalence rates of alcohol abuse or dependence among wards of 11.1–32.9% and among intake months between 11.3 and 28.7%. Of the medical department in-patients, 1.9%, and of the surgical in-patients, 2.1%, were screened as false-positive cases. In addition, 5.5% of the medical and 12.0% of the surgical patients were given a diagnosis on suspicion. It is concluded that all general wards and different intake months should be taken into account when estimating prevalence of alcohol abuse or dependence in a general hospital.
|
INTRODUCTION |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Evidence shows that 10–20% of in-patients admitted for somatic complaints in general hospitals are alcohol-dependent according to a screening diagnosis (McIntosh, 1982
). However, only a few more recent studies are based on samples representative of at least one general hospital (Moore et al., 1989; Seppä and Mäkelä, 1993; Dongier et al., 1994; Rambaldi et al., 1995). Earlier studies differed in diagnostic criteria (Barchha et al., 1968; Lloyd et al., 1982). In particular, published studies predominantly show the following limitations, which might bias estimates: (1) not all wards have been screened; (2) only one screening test was used, and not a comprehensive diagnostic procedure; (3) data have not been collected over a period longer than one month. The present study has several purposes: (1) to ascertain the prevalence of alcohol dependence and abuse on grounds of screening measures in the two departments of one general hospital; (2) to determine if wards differ; (3) to find out whether intake months of the general hospital differ in prevalence of alcohol abuse or dependence; (4) to decide whether a screening mainly based on questionnaire is sufficient, or a two-step diagnostic procedure with a diagnostic interview in screening positive cases is necessary; (5) to question whether diagnostic information can be provided for cases in which alcohol abuse or dependence is only suspected.
|
PATIENTS AND METHODS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Patients for the study were all admitted to one general hospital in a city with 217 000 inhabitants in northern Germany. The patients were aged 18–64 years (n = 2651). The region is characterized by its situation at the former border to East Germany and the Baltic Sea. There is a low level of industrialization and an unemployment rate (9.9%) slightly above the national average (8.8%). General hospital care in the region is provided by the medical school, including different in-patient treatment units, a municipal (407 beds) and two smaller hospitals. The municipal hospital was chosen as typical for a regional hospital with a surgical and medical department. In Germany, medical and surgical departments cover 66.5% of all general hospital beds (without departments for paediatrics or psychiatry) (Statistisches Bundesamt, 1994
). All first-admissions staying in hospital for at least 24 h were included in the sample during the data-gathering period of the study. They showed a large variety of somatic diagnoses published in another paper (Gerke et al., 1997). Of the original sample, 366 persons stayed less than 24 h, and there were 552 readmissions, mainly due to an oncological treatment plan. Data were collected from January to June on all of the six medical, and from July to December on all of the five surgical wards. The medical department consists of two general medical wards (M1, M2), one ward for the treatment of predominantly gastroenterological and infectious diseases (M3), one mainly for cardiovascular diseases (M4), and one for diabetes (M5). One ward (M6) is specialized in oncological therapy: the patients are admitted several times according to a treatment plan. Usually alcohol-dependent or -abusing subjects are not admitted. However, the patients on this ward underwent the same diagnostic procedure in the study as the others. The surgical department comprises two abdominal (S1, S2) and two traumatological wards (S3, S4) as well as one with a mixed abdominal and traumatological case-load (S5). Of the total eligible sample (n = 1736), 50.9% were in the medical, and 49.1% in the surgical departments; 43.4% were female. We excluded the intensive care ward, because most of its patients could not be interviewed. Of the total sample, 3.6% were too ill to participate, e.g. unable to speak, or deceased before being investigated, 7.0% refused the screening, 3.5% were foreigners unable to speak German sufficiently, 3.1% could not be investigated because of different reasons (highly infectious disease, patients in radiotherapy). Of the patients, 7.4% had been discharged before the diagnostic procedure could be finished. The final sample size was therefore 1309 in-patients. It included 41.9% females. In the medical department, the mean ± SD age was 48.6 ± 12.1; the mean age was lowest on the gastroenterology ward (M3) (43.4 ± 12.7). The medical wards differed significantly in age (analysis of variance, P < 0.000). In the surgical department the mean age was 41.5 ± 14.2); the mean age was lowest on the trauma wards (S3, S4) (39.0 ± 14.0 and 38.9 ± 14.4 respectively). These wards too differed significantly in age (analysis of variance, P < 0.000).
The sample patients gave informed consent to take part in a study with respect to alcohol and other health-related consumption behaviours. The diagnostic procedure included two steps: a screening and an ascertainment of a diagnosis. The screening comprised the CAGE (Ewing, 1984) as well as the MAST (Selzer, 1971) instruments, proof of withdrawal on grounds of a corresponding medication, and chart review. In screening-negative cases, a further part of the first diagnostic step was to ascertain if any medication to treat withdrawal had been given. If this was not the case, then patient records were analysed with respect to the following criteria: withdrawal or other criteria of alcohol dependence, somatic disorders resulting from alcohol drinking, one or more raised laboratory measures on grounds of alcohol drinking (gamma-glutamyl transpeptidase, aspartate aminotransferase, alanine aminotransferase, mean corpuscular volume), self-reported high alcohol consumption (
40 g of pure alcohol/day in females, 60 g in males). These criteria were used for a diagnosis on suspicion. It seemed useful because diagnostic information based on self-report, such as CAGE and MAST, does not detect all alcohol-dependent or alcohol-abusing subjects in the general hospital due to denial of symptoms (Rumpf et al., 1997). The laboratory parameters were included on the grounds of their diagnostic accuracy and as far as they were used in the diagnostic routine on the wards (Watson et al., 1986). To take into consideration a diagnosis on suspicion was useful as diseases typically following from alcohol abuse (definitely alcohol-related diagnoses) were relatively rare among patients with a diagnosis on suspicion (21.2%) and relatively frequent among patients with alcohol dependence (68.5%), whereas diseases other than alcohol-related ones showed a higher rate among patients with a diagnosis on suspicion (32.7%), compared to those with an alcohol dependence (8.7%) (Gerke et al., 1997).
The screening questionnaires were used after translation and back-translation with the help of an English native speaking co-worker. To make use of the CAGE as well as the MAST proved to be worthwhile, as only 57.0% of the screening questionnaire-positive cases could be detected by both of them. Most of the patients filled in the questionnaires by themselves. If this was not possible, for instance in persons illiterate or with disturbed reading capabilities, it was presented as an interview. This was the case in 35.6% of the screenings. A detailed analysis of the screening questionnaires is described elsewhere (Rumpf et al., 1997, 1998).
In the second diagnostic step, in the case of a positive screening questionnaire (CAGE >1 or MAST >4) or if medication to treat withdrawal was prescribed, the diagnostic interview according to the Alcohol Section of the Schedules of Clinical Assessment in Neuropsychiatry (SCAN) (World Health Organization, 1992) was conducted by two trained psychologists and one physician experienced in psychiatry (Hapke et al., 1998). The SCAN provided an ICD-10 as well as a DSM-III-R diagnosis of alcohol dependence or abuse. Systematic chart review led to additional cases of dependence or abuse when the patient records gave evidence of dependence or abuse according to ICD-10 or DSM-III-R. If this was not the case, criteria for a diagnosis on suspicion were checked on grounds of the case report file review: if two criteria were fulfilled, this diagnosis was given. The diagnostic procedure differentiates five alcohol-related diagnostic groups: (1) patients without evidence of alcohol abuse or dependence (no alcohol-related diagnosis); (2) a positive abuse but negative dependence diagnosis (abuse); (3) dependence diagnosis fulfilled (dependence); (4) a positive screening result, which, however, was not confirmed by a diagnosis of dependence or abuse (screening false positive), (5) patients exclusively with a diagnosis on suspicion (suspicion).
A total of 0.4% of the final sample could not be detected by the screening questionnaires, but by prescribed medication to treat withdrawal. All of them were confirmed as dependants. According to patient records, an additional 0.2% fulfilled the criteria for alcohol dependence and 1.3% for abuse. Of the sample, 2.0% showed a positive result in CAGE or MAST but could not be confirmed by SCAN. Of these 26 screening false-positive patients, three were positive by both CAGE and MAST, 14 by CAGE alone, and nine by MAST only. We did not conduct the SCAN systematically in screening negative cases, since, in another study, we showed that 1.0% of alcohol-dependent and 5.8% of alcohol-abusing patients according to a diagnosis on grounds of SCAN were not detected by screening questionnaires, including the CAGE (Rumpf et al., 1996).
|
RESULTS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
The six medical wards differed in the proportion of their content of alcohol abusers or dependants between 3.5 and 32.1%, or, if the oncological ward was excluded, between 17.8 and 32.1%. The corresponding range was 11.1–20.4% in the surgical department (Table 1
). The differences between wards were statistically significant, even if the oncological ward was excluded or if patients with any alcohol-related diagnoses were compared to those without (
2, P < 0.001).
|
Male alcohol abusers or dependants differed in prevalence rates from 7.9 to 49.2%, or, without the oncological ward, from 21.1 to 49.2% in the medical, and between 18.3 and 25.2% in the surgical department (Table 2
). Again, wards differed significantly, even if the oncological ward was excluded, or if patients with any alcohol-related diagnoses were compared to those without (2, P < 0.001).
|
According to month of admission, prevalence rates varied between 15.5 and 28.7% in the medical and 11.3 and 20.2% in the surgical department (female and male abusers and dependants combined) (Table 3
). The respective rates for males (Table 4 ) were 18.5–40.0% in the medical, and 16.5–26.5% in the surgical department. In the latter, the lowest intake rate of alcohol abusers or dependants was in October and the highest in August, for both sexes. In the medical department, fewest alcohol abusers or dependants were admitted in March and most in May.
|
|
Of the screening-positive cases, 1.9% in the medical, and 2.5% in the surgical, department could not be confirmed by the diagnosis. This rate differed between 0.0 and 4.9% on the medical and 0.0 and 3.1% on the surgical wards (Table 1
). There was a similar distribution in male in-patients (Table 2) as well as by month (Tables 3 and 4). A further 5.5% on the medical and 12.0% on the surgical wards had a diagnosis on suspicion. A particularly high rate on the gastroenterological ward (12.3%) coincides with a high rate of alcohol dependence. The two traumatological wards showed rates above 12%, whereas, according to dependence, they were near to the mean of the surgical department. The rate of patients with a diagnosis on suspicion in the surgical department was nearly twice as high as the rate in the medical department (Tables 1 and 2). According to the month of admission there was a particularly high rate of a diagnosis on suspicion in male in-patients in June (Table 4). |
DISCUSSION |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
The results show that a bias is produced in prevalence estimates in general hospitals if the sample is restricted to a single ward or an intake month. Although there are limitations which might lead to an underestimation of the prevalence rate in this study too, namely (1) there were refusers among the patients, (2) those staying less than 24 h could not be included, (3) we could not estimate the rate of screening false negatives in this investigation, four issues are raised by the results. First, there was considerable variation between different wards with a rate at least nearly twice on one ward compared to another in the medical department. In male medical in-patients, proportions of dependence or abuse ranged between 18 and 49%. One ward showed just 28.1% of the male case-load without any signs of alcohol abuse. The differences between wards may be explained by different intake traditions. Different teams had different priorities in treating patients with alcohol-related diseases, while it seemed natural that the gastrointestinal ward possessed a high rate of patients with alcohol-related diseases. Therefore, prevalence estimates should include all wards of at least one department or hospital.
Second, there was substantial variation between intake months, from 11 to 29%, although the intake load was rather evenly distributed over the months. The data collection period included winter, spring, and summer on the medical, and summer, autumn and winter on the surgical ward. However, the variation by month did not confirm a pattern ac-cording to seasonal variation. Other reasons were considered and excluded, e.g. possible changes in hospital routine, holiday times, or weather periods. We cannot state that the variation is typical for a whole year. This would require data collection for a longer time period. Furthermore it is not feasible to relate monthly variations to drinking levels in the population. In Germany, no data about alcohol consumption in the population by month are available.
Third, a two-step diagnosis gives proof of valid diagnosis. Many studies are restricted to screening measures. According to our results this restriction leads to an overestimation of only 2%. On two of the eleven wards, there was no overestimation. This is a very small proportion which leads to the conclusion that pure prevalence estimates do not need a time-consuming measure in addition. For practical reasons, however, the diagnostic interview is important because it carries aspects of supporting motivation to change. Although we used all information available about alcohol-related problems, only a few cases screened as false negative could be proven otherwise by evidence of a prescribed medication to treat withdrawal or by chart review. As we showed in another study (Rumpf et al., 1996) there would be a gain of just 1% of detected alcoholics, if the diagnostic interview were to be conducted with all patients. This contributes to the assumption that those who deny alcohol problems in the screening questionnaire would do so too in an interview.
Fourth, there was a substantial rate of patients with a diagnosis on suspicion, which was of only moderate validity, but of high practical relevance. The diagnosis on suspicion could not be checked with respect to validity. Patients with suspected dependence or abuse might preferably be cases for early intervention. There were considerable differences in the rate of a diagnosis on suspicion. On the gastroenterological ward, there was an overall high rate of patients with an alcohol-related diagnosis, whereas on one of the traumatological wards the high rate of patients with a diagnosis on suspicion suggests denial by young patients in the questionnaires. The latter point receives support from the low mean age of the trauma patients. On the whole, however, age differences did not explain variations in prevalence rates between wards. With respect to false positives on screening, no specific subgroups, e.g. subjects with previous but not recent heavy drinking episodes, could be identified.
We conclude that data should be collected on all wards of a hospital and a time span including different intake months over at least one year, with a two-step diagnostic procedure. Further studies should concentrate on the development of criteria for a diagnosis on suspicion. Practical diagnostic and intervention routine measures should follow from the present confirmation that all wards in a general hospital have a considerable case-load of alcohol-dependent patients.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
This study was supported by a grant of the German Ministry of Health (326-4914-8/38).
|
FOOTNOTES |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
* Author to whom correspondence should be addressed.
|
REFERENCES |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Barchha, R., Stewart, M. A. and Guze, S. B. (1968) The prevalence of alcoholism among general hospital ward patients. American Journal of Psychiatry 125, 681–684.
Dongier, M., Hill, J. M., Kealey, S. A. and Joseph, L. (1994) Screening for alcoholism in general hospitals. Canadian Journal of Psychiatry 39, 12–20.[Web of Science][Medline]
Ewing, J. A. (1984) Detecting alcoholism. The CAGE questionnaire. Journal of the American Medical Association 252, 1905–1907.
Gerke, P., Hapke, U., Rumpf, H.-J. and John, U. (1997) Alcohol-related diseases in general hospital patients. Alcohol and Alcoholism 32, 179–184.
Hapke, U., Rumpf, H.-J. and John, U. (1998) Differences between hospital patients with alcohol problems referred for counselling by physicians' routine clinical practice versus screening questionnaires. Addiction 39, 1777–1786.
Lloyd, G., Chick, J. and Crombie, E. (1982) Screening for problem drinkers among medical inpatients. Drug and Alcohol Dependence 10, 355–359.[Web of Science][Medline]
McIntosh, I. D. (1982) Alcohol-related disabilities in general hospital patients: A critical assessment of the evidence. International Journal of the Addictions 17, 609–639.[Web of Science][Medline]
Moore, A. D., Bone, L. R., Geller, G., Mamon, J. A., Stokes, E. J. and Levine, D. M. (1989) Prevalence, detection, and treatment of alcoholism in hospitalized patients. Journal of the American Medical Association 261, 403–407.
Rambaldi, A., Gluud, C., Belli, A., Nielsen, S., Storgaard, H. and Moesgaard, F. (1995) Prevalence of alcohol problems among adult somatic inpatients in Naples. Alcohol and Alcoholism 30, 441–448.
Rumpf, H.-J., Hapke, U. and John, U. (1996) Empirical evidence of methodological standards in estimating the prevalence of alcohol dependence or abuse in the general hospital. Paper presented at the 10th International Conference on Alcohol, Liverpool, April 14–17.
Rumpf, H.-J., Hapke, U., Hill, A. and John, U. (1997) Development of a screening questionnaire for the general hospital and general practices. Alcoholism: Clinical and Experimental Research 21, 894–898.[Web of Science][Medline]
Rumpf, H.-J., Hapke, U., Erfurth, A. and John, U. (1998) Screening questionnaires in the detection of hazardous alcohol consumption in the general hospital—direct or disguised assessment. Journal of Studies on Alcohol 59, 698–703.[Web of Science][Medline]
Selzer, M. L. (1971) The Michigan Alcoholism Screening Test: the quest for a new diagnostic instrument. American Journal of Psychiatry 127, 1653–1658.
Seppä, K. and Mäkelä, R. (1993) Heavy drinking in hospital patients. Addiction 88, 1377–1382.[Web of Science][Medline]
Statistisches Bundesamt (1994) Statistisches Jahrbuch. Metzler-Poeschel, Wiesbaden.
Watson, R. R., Mohs, M. E., Eskelson, C., Sampliner, R. E. and Hartmann, B. (1986) Identification of alcohol abuse and alcoholism with biological parameters. Alcoholism: Clinical and Experimental Research 10, 364–385.[Web of Science][Medline]
World Health Organization (1992) Schedules for Clinical Assessment in Neuropsychiatry. World Health Organization, Division on Mental Health, Geneva.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  E. M. Proude, K. M. Conigrave, A. Britton, and P. S. Haber Improving Alcohol and Tobacco History Taking by Junior Medical Officers Alcohol Alcohol., February 29, 2008; (2008) agm182v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
U. John, H.-J. Rumpf, M. Hanke, P. Gerke, and U. Hapke ESTIMATION OF TOBACCO- OR ALCOHOL-ATTRIBUTABLE DISEASE RATES IN NATIONAL HOSPITAL CARE: AN APPROACH BASED ON ROUTINE IN-PATIENT DISEASE REGISTER DATA AND SYSTEMATIC DIAGNOSIS OF ALCOHOL USE DISORDERS Alcohol Alcohol., July 1, 2003; 38(4): 339 - 346. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||