UNPLANNED VERSUS PLANNED DISCHARGES FROM IN-PATIENT ALCOHOL DETOXIFICATION: RETROSPECTIVE ANALYSIS OF 470 FIRST-EPISODE ADMISSIONS

José Martínez-Raga,*, E. Jane Marshall1, Francis Keaney1, David Ball1 and John Strang1

Unidad de Conductas Adictivas Area 9, Generalitat Valenciana Conselleria De Sanitat, Centro de Salud de San Marcelino, San Pio X, 33, 46017 Valencia, Spain and
1 National Addiction Centre (Institute of Psychiatry, King's College London and The Maudsley Hospital), 4 Windsor Walk, London SE5 8AF, UK

Received 25 June 2001; in revised form 30 October 2001; accepted 26 November 2001


    ABSTRACT
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
— Failure to complete treatment during alcohol detoxification is a major complication of effective clinical management. We determined the socio-demographic characteristics, the pattern of alcohol and drug use, the hepatitis C status and the psychiatric state of these patients, using retrospective data on 470 first admissions to a specialist alcohol in-patient unit. The 316 patients (67.2%) who had a planned discharge (PD group) were compared with the 154 patients (32.8%) who had unplanned discharges (UPD group). Patients in the UPD group were younger on admission, had higher Alcohol Problems Questionnaire scores, had started to drink heavily at a younger age, were more likely to have previously used cocaine, amphetamines and heroin, and to have smoked cannabis in the 30 days prior to admission. They were more likely to be positive for markers of hepatitis C infection, to have a borderline personality disorder, antisocial personality disorder, or to have concurrent opiate or benzodiazepine dependence. PD was associated with depressive disorder. Early identification can be made of alcohol-dependent individuals at risk of premature drop-out during in-patient treatment. Interventions to retain patients at risk of premature UPD should be developed.


    INTRODUCTION
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Identifying subgroups of individuals dependent on alcohol or other substances may help elucidate aetiology, predict clinical course, rationally assign treatment, and improve outcome (Tucker and Westermeyer, 1995Go; Weiss et al., 1997Go; Ross et al., 1998Go). Little has been published on the characteristics of alcohol-dependent patients who leave in-patient/residential treatment prematurely. Patients with a diagnosis of substance abuse are more likely to leave hospital prematurely when compared to other psychiatric patient groups (Bradley and Zaekin, 1997Go). Failure to complete treatment programmes is associated with medical or psychiatric complications and higher mortality (Schuckit, 1992Go).

Previous studies have found that substance-dependent patients who leave in-patient treatment prematurely tend to be younger (Armenian et al., 1999Go), single (De los Cobos et al., 1997Go), intoxicated on admission (Beck et al., 1983Go), as well as being more likely to have a history of leaving against medical advice (Cook et al., 1994Go). To date, no European studies have been published comparing alcohol-dependent subjects who leave in-patient treatment prematurely with those who complete the programme. The topic has added importance if in-patient facilities are scarce.

We report data from a UK in-patient specialist treatment unit for alcohol use disorders. Patients with a planned discharge (PD) from the multidisciplinary treatment programme were compared with patients who had unplanned discharges (UPD), in order to identify factors associated with premature discharges.


    PATIENTS AND METHODS
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Subjects were 482 ‘first episode’ admissions to a 16-bedded unit at The Bethlem Royal Hospital, London, UK, between September 1995 (when the unit was opened) and September 1998. Subjects who were transferred to another medical or psychiatric ward (n = 8) and four subjects whose medical records could not be localized, were excluded from the study. ‘First episode’ meant first admission to this unit. Patients were not excluded if they had an admission to another unit prior to September 1995 or in the period September 1995 to September 1998. None of the patients underwent drug detoxification or was on substitute prescribing regimes. Thus, the personal and clinical characteristics and treatment retention of 470 alcohol-dependent subjects were compared in this analysis.

No differences were found in a preliminary analysis of the study variables between subjects who self-discharged against medical advice and subjects who had other forms of UPD. Consequently, in contrast to some other recent studies that only considered subjects who self-discharged against medical advice (De los Cobos et al., 1997Go; Armenian et al., 1999Go), our UPD group included all prematurely discharged patients.

Setting
The ward is a 16-bed self-contained unit offering assessment, medically assisted withdrawal, and a relapse prevention treatment package to individuals with severe alcohol dependence. The ward is located in the grounds of the Bethlem Royal Hospital (South and Maudsley NHS Trust), in a large parkland estate in a leafy middle-class residential suburb in the South London/Kent area. It is staffed by a team of psychiatrists, psychologists, nurses, nursing assistants and occupational therapists. Patients are requested to stay on the ward during the period they receive medication for their detoxification and are only allowed off the ward during this period with a nurse escort. Patients are allowed visitors after the third day of their admission. Unit policies allow nurses discretion to search all visitors as well as patients returning unescorted to the ward.

Admissions were voluntary, and planned. There were no emergency admissions. Patients are admitted for either 14, 28 or 42 days. This is decided in advance by the referring agency, and related to the clinical needs of the patient. For our study, patients who left before the planned date, but having negotiated the prior agreement of the staff, were still included in the PD group.

During admission, patients underwent a comprehensive assessment of physical, neuropsychological and psychiatric complications of alcohol dependence, including mental state examination and a cognitive assessment. Chlordiazepoxide was used in the 5–9-day withdrawal regime (oxazepam for patients with known liver disease). Patients were expected to attend the programme, which included one-to-one and group sessions, conducted by the ward staff. Breath-alcohol level and urine toxicology were assessed on admission and then randomly repeated.

Data collection
Admission information was extracted from the medical records by J.M.-R. He was not blind to the discharge method. This included socio-demographic data; alcohol and other substance use histories, and information on co-morbid psychiatric disorders (Russell, 1987Go).

Diagnosis of alcohol dependence and co-morbid psychiatric disorders recorded in the records was based on assessment by two independent psychiatrists using ICD-10 (World Health Organization, 1992Go) and DSM-IV (American Psychiatric Association, 1994Go) criteria, discussed and recorded at the weekly multidisciplinary ward round. Patients received a psychiatric diagnosis in conjunction with an alcohol or substance use disorder only if they were currently ill or had previously fulfilled ICD-10/DSM-IV criteria for that diagnosis while alcohol-free. Based on the particular difficulties involved in distinguishing between addiction-related symptoms and enduring personality traits (Gerstley et al., 1990Go), antisocial and borderline personality disorder were the only personality disorder diagnoses made.

The Alcohol Problems Questionnaire (APQ) (Drummond, 1990Go) and the Severity of Alcohol Dependence Questionnaire (SADQ) (Stockwell et al., 1983Go) were routinely administered.

Data analysis
Analysis used the Statistical Package for the Social Sciences (SPSS) for Windows (Norussis, 1993). Statistical significance was established by the {chi}2-test for categorical data; Fisher's exact test was employed when sample size was insufficient to calculate {chi}2. The two-tailed Student's t-test was used for comparison of means with interval data. P < 0.05 was considered significant. Finally, in order to identify variables that were significantly and independently associated with UPD, a stepwise logistic regression was also conducted using PD versus UPD status as dependent variables.


    RESULTS
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Of the 470 subjects under scrutiny, 316 (67.2%) had a PD and 154 subjects (32.8%) a UPD. The UPD group included 88 subjects (18.7%) who discharged themselves against medical advice, 58 subjects (12.3%) discharged for alcohol or drug use during their admission, and eight subjects (1.7%) who were discharged from the ward because of violent or threatening behaviour. The average length of in-patient stay in days (mean ± SD) was 28.8 ± 16.1 for PD and 19.1 ± 17.4 for UPD (t = 6.0, P < 0.001).

Socio-demographic characteristics
UPD were younger on admission than PD (t = 3.0, P < 0.005). There were no significant differences between the two groups with regards to gender, marital status, ethnic background or living arrangements (Table 1Go).


View this table:
[in this window]
[in a new window]
 
Table 1. Socio-demographic characteristics and alcohol use histories of subjects who had a planned discharge (PD) and those who had an unplanned discharge (UPD)
 
Alcohol-use-related variables
Table 1Go also shows that PD and UPD subjects did not differ significantly in average daily alcohol intake in the month prior to admission, nor in their age at first drink. However, UPD subjects had started drinking heavily at a younger age (t = 4.1, P < 0.001).

APQ and SADQ scores are also shown Table 1Go. These two parameters were available from 281 (205 PD and 76 UPD) and 292 (212 PD and 80 UPD) subjects, respectively. While there were no differences in SADQ scores between the two groups, UPD had significantly higher APQ scores (t = –3.2, P < 0.005).

Reported drug use and concurrent substance use disorders
Information on cannabis use in the 30 days prior to admission was available from 444 subjects (305 PD and 139 UPD). Similarly, information on lifetime cocaine use, lifetime amphetamine use and lifetime heroin use was obtained from 431 (295 PD and 136 UPD), 446 (307 PD and 139 UPD) and from 447 (306 PD and 141 UPD) subjects, respectively. As shown in Table 2Go, UPD were more likely to have smoked cannabis in the 30 days prior to admission ({chi}2 = 14.7, df = 1, P < 0.001); whilst PD were more likely never to have used cocaine ({chi}2 = 24.4, df = 1, P < 0.001), amphetamines ({chi}2 = 16.0, df = 1, P < 0.001) or heroin ({chi}2 = 17.8, df = 1, P < 0.001) during their lifetimes. Diagnoses of other co-morbid substance use disorders were available from all 316 PD subjects (except n = 315 for opiate dependence), and from all 154 UPD subjects (Table 2Go). UPD were more likely to fulfil criteria for co-morbid opiate dependence ({chi}2 = 9.8, df = 1, P < 0.005), as well as co-morbid benzodiazepine dependence ({chi}2 = 4.6, df = 1, P < 0.05), but not for concurrent cocaine dependence.


View this table:
[in this window]
[in a new window]
 
Table 2. Reported drug use and current substance use disorder of subjects who had a planned discharge (PD) and those who had an unplanned discharge (UPD)
 
Hepatitis B and C
A total of 273 subjects (185 PD and 88 UPD) were tested for hepatitis C. Using an immunoassay, hepatitis C IgG antibody was detected in 11 PD (5.9%) and 16 UPD (18.2%) ({chi}2 = 10.0, df = 1, P < 0.005). Of the 272 subjects (185 PD and 87 UPD) tested for hepatitis B, 18 PD (9.7%) and 12 UPD (13.8%) were positive for the hepatitis anti-HBc IgG antibody; these differences were, however, not significant. A full description of variables associated with viral hepatitis infection in this sample has been published elsewhere (Martinez-Raga et al., 2001Go).

Psychiatric diagnoses
As shown in Table 3Go, PD were significantly more likely to have a co-morbid depressive disorder ({chi}2 = 6.5, df = 2, P < 0.05), although there were no significant differences between the two groups in the rates of coexistent post-traumatic stress disorder or in lifetime rates of anxiety disorders. However, UPD were significantly more likely to have a diagnosis of borderline personality disorder ({chi}2 = 10.2, df = 1, P < 0.005) and of antisocial personality disorder ({chi}2 = 24.8, df = 1, P < 0.001).


View this table:
[in this window]
[in a new window]
 
Table 3. Comparison of psychiatric diagnoses of subjects who had a planned discharge (PD) and those who had unplanned discharge (UPD)
 
Logistic regression analysis
Stepwise logistic regression was used to identify those variables which were significantly and independently associated with UPD. The independent variables in the analysis included those which were shown to be significantly associated with the dependent variable (type of discharge: PD or UPD). Age on admission, and age when starting to drink heavily were entered as continuous variables. Cannabis use in the 30 days prior to admission, lifetime cocaine, amphetamine and heroin use, current depressive disorder, co-morbid borderline and antisocial personality disorder, current opiate and benzodiazepine dependence, were included as dichotomous variables. Due to missing data in these variables, 58 subjects were excluded from the analysis, leaving an active sample for analysis of 412 subjects. Because APQ scores and hepatitis C diagnosis were only available for a limited number of subjects, these two variables were excluded from the analysis. Younger age at starting to drink heavily, having a diagnosis of borderline or antisocial personality disorder, and having previously used heroin were significantly and independently associated with UPD (Table 4Go).


View this table:
[in this window]
[in a new window]
 
Table 4. Stepwise logistic regression showing variables associated with having an unplanned discharge
 

    DISCUSSION
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
We have identified several major differences between alcohol-dependent patients who had a PD from an in-patient treatment programme and those who failed to complete the planned treatment and had a UPD. UPD were younger on admission, had higher APQ scores, had started to drink heavily at a younger age, were more likely to have previously used cocaine, amphetamines and heroin and to have smoked cannabis in the 30 days prior to admission. They were more likely to be hepatitis C-antibody positive, to have a co-morbid borderline personality disorder, antisocial personality disorder, also concurrent heroin and benzodiazepine dependence. These factors have all previously been associated with a poorer prognosis and with chronicity of drinking problems (Rounsaville et al., 1987Go; Miller et al., 1990Go; Shaw et al., 1998Go). Of these variables, the study identified that starting to drink heavily at a younger age, having had previous heroin use, and a co-morbid borderline or antisocial personality disorder were independently associated with UPD. However, PD patients were more likely to have a co-morbid depressive disorder.

The rate of UPD in this report is similar to rates reported in some previous publications (Stark, 1992Go; Armenian et al., 1999Go). Our results are consistent with other studies showing that substance-dependent individuals discharged against medical advice or through other forms of premature discharge are likely to be younger (Beck et al., 1983Go; Armenian et al., 1999Go). In addition, previous reports have shown that a history of alcohol-related problems, the use of at least one illicit drug (Leigh et al., 1984Go), and higher scores on alcohol screening instruments (Noel et al., 1987Go) led to a higher drop-out rate in alcohol-dependent patients. Similarly, in our sample, having a UPD was associated with higher APQ scores, having used cocaine, amphetamines and heroin, as well as with having a concurrent heroin and benzodiazepine dependence. Moreover, drug use prior to admission has also been related to poorer treatment retention (Beck et al., 1983Go; De los Cobos et al., 1997Go), while in our study UPD was higher among patients who had smoked cannabis in the 30 days prior to admission.

The association between psychopathology and treatment retention has only rarely been examined in this population. Early work from Beck et al. (1983) found that a diagnosis of personality disorder was associated with discharges against medical advice from alcohol and drug in-patient treatment. Elsewhere, high Minnesota Multiphasic Personality Inventory (MMPI) psychopathic deviant scores and sociopathic diagnoses have been reported as more common among drop-outs from alcohol treatment (Pekarik et al., 1986Go). Similarly, in the present study, having a diagnosis of borderline personality disorder or antisocial personality disorder was significantly and independently associated with a UPD.

Patients who completed the planned treatment were more likely to have a depressive disorder. It is possible that the unit staff may find it easier to empathize with, and therefore to engage, depressed patients, in contrast with patients with a personality disorder.

Those patients who failed to complete treatment were more likely to be infected with the hepatitis C virus. Alcohol use increases the risk of cirrhosis and hepatocellular carcinoma in patients positive for hepatitis C (Regev and Jeffers, 1999Go). It is therefore advisable to link these patients with a specialist liver service. Moreover, as hepatitis C seropositivity was diagnosed for the first time in the course of the admission, patients may require added support. By dropping out of treatment prematurely, some of these patients may also disengage from further out-patient care and fail to attend any appointments with specialist liver services.

As with any retrospective study, the present report has limitations. It depended on review of medical records, and so information was not complete in some cases. Lack of consistency in the way in which information was entered in the records could have influenced the validity of some of the data collected, although this was minimized partly by the standardized way clinical information is collected on the unit and partly by choosing variables which could be least affected by a collector's bias. The reliance largely on self-report data for alcohol and substance use histories is also a potential limitation of our study. However, self-report information is the most widely used approach to assessment in the substance use field and has been found to be broadly valid (Babor et al., 1990Go; Carroll, 1995Go; Darke, 1998Go). In addition, urine drug screens were performed on all patients on admission, which help to corroborate reports of cannabis use prior to admission. A possible additional limitation is the fact that standardized instruments were not used to reach diagnoses of co-morbid psychiatric disorders. Instead, these were based on assessment by two psychiatrists along with longitudinal clinical observation. However, diagnoses were only made when symptoms had previously been present while drug-free. A strength of this method is that it allowed us to avoid false positives, but it may have caused under-diagnosis of some disorders and increased the number of false negatives. In addition, an extended evaluation period in combination with multiple sources of information have been found specifically to increase validity in research (Helzer et al., 1985Go; Weiss et al., 1988Go).

Our study may have clinical implications. Thus, the period of alcohol withdrawal is a time when patients should start to receive support for the next phase of their recovery, because they are vulnerable to relapse (Mattson et al., 1998Go; Shaw et al., 1998Go). Simply making the staff aware of the level of treatment drop-out rate may help to reduce this (Craig, 1985Go). Identifying a sub-group liable to drop-out, as we have done in this study, may help to focus attention. However, there should be further development of more individualized interventions to help these patients remain in treatment.


    FOOTNOTES
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
* Author to whom correspondence should be addressed. Back


    REFERENCES
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
American Psychiatric Association (1994) Diagnostic and Statistical Manual of Mental Disorders, 4th edn. American Psychiatric Association, Washington, DC.

Armenian, S. H., Chutuape, M. A. and Stitzer, M. L. (1999) Predictors of discharges against medical advice from a short-term hospital detoxification unit. Drug and Alcohol Dependence 56, 1–8.[ISI][Medline]

Babor, T. F., Brown, J. and DelBoca, F. K. (1990) Validity of self-reports in applied research on addictive behaviors: fact or fiction? Behavioural Assessment 12, 5–31.

Beck, N. C., Shekom, W., Fraps, C., Borgmeyer, A. and Witt, A. (1983) Prediction of discharges against medical advice. Journal of Studies on Alcohol 44, 171–180.[ISI][Medline]

Bradley, C. J. and Zaekin, G. A. (1997) An inpatient profile of patients with a substance abuse diagnosis. Journal of Substance Abuse and Treatment 14, 155–162.

Carroll, K. M. (1995) Methodological issues and problems in the assessment of substance use. Psychological Assessment 7, 349–358.[ISI]

Cook, C. A., Booth, B. M., Blow, F. C., McAleenan, K. A. and Bunn, J. Y. (1994) Risk factors for AMA discharge from VA inpatient alcoholism treatment programs. Journal of Substance Abuse and Treatment 11, 239–245.

Craig, R. J. (1985) Reducing the treatment drop out rate in drug abuse programs. Journal of Substance Abuse and Treatment 2, 209–219.

Darke, S. (1998) Self-report among injecting drug users: a review. Drug and Alcohol Dependence 51, 253–263.[ISI][Medline]

De los Cobos, J. P., Trujols, J., Ribalta, E. and Casas, M. (1997) Cocaine use immediately prior to entry in an inpatient heroin detoxification unit as a predictor of discharges against medical advice. American Journal of Drug and Alcohol Abuse 23, 267–279.[ISI][Medline]

Drummond, D. C. (1990) The relationship between alcohol dependence and alcohol-related problems in a clinical population. British Journal of Addiction 85, 357–366.[ISI][Medline]

Gerstley, L. J., Alterman, A. I., McLellan, A. T. and Woody, G. E. (1990) Antisocial personality disorder in patients with substance abuse disorders: a problematic diagnosis? American Journal of Psychiatry 147, 173–178.[Abstract]

Helzer, J. E., Robins, L. N. and McEvoy, L. T. (1985) A comparison of clinical and Diagnostic Interview Schedule diagnoses: physician re-examination of lay-interviewed cases in the general population. Archives of General Psychiatry 42, 657–666.[Abstract]

Leigh, G., Ogborne, A. C. and Cleland, P. (1984) Factors associated with patient dropout from an outpatient alcoholism treatment service. Journal of Studies on Alcohol 45, 359–362.[ISI][Medline]

Martínez-Raga, J., Marshall, E. J., Keaney, F., Best, D., Ball, D. and Strang, J. (2001) Hepatitis B and C in alcohol-dependent patients admitted to a UK alcohol inpatient treatment unit. Addiction Biology 6, 363–372.[ISI][Medline]

Mattson, M. E., DelBoca, F. K., Carroll, K. M. et al. (1998) Compliance with treatment and follow-up protocols Project MATCH: predictors and relationship to outcome. Alcoholism: Clinical and Experimental Research 22, 1328–1339.[ISI][Medline]

Miller, N. S., Millman, R. B., Keskinnen, S. (1990) Outcome at six and twelve months post inpatient treatment for cocaine and alcohol dependence. Advances in Alcohol and Substance Abuse 9, 101–120.

Noel, N. E., McCrady, B. S., Stout, R. L. and Fischer-Nelson, H. (1987) Predictors of attrition from an outpatient alcoholism treatment program for couples. Journal of Studies on Alcohol 48, 229–235.[ISI][Medline]

Norusis, M. J./SPSS, Inc. (eds) (1993) SPSS for Windows, Release 6.0. SPSS, Chicago.

Pekarik, G., Jones, D. L. and Blodgett, C. (1986) Personality and demographic characteristics of dropouts and completers in a nonhospital residential alcohol treatment program. International Journal of the Addictions 21, 131–137.[ISI][Medline]

Regev, A. and Jeffers, L. J. (1999) Hepatitis C and alcohol. Alcoholism: Clinical and Experimental Research 23, 1543–1551.[ISI][Medline]

Ross, R., Fortney, J., Lancaster, B. and Booth, B. M. (1998) Age, ethnicity, and comorbidity in a national sample of hospitalized alcohol-dependent women veterans. Psychiatric Services 49, 663–668.[Abstract/Free Full Text]

Rounsaville, B. J., Dolinsky, Z. S., Babor, T. F. and Meyer, R. E. (1987) Psychopathology as a predictor of treatment outcome in alcoholics. Archives of General Psychiatry 44, 505–513.[Abstract]

Russell, G. F. M. (1987) Psychiatric Examination. Oxford Medical Publications, Oxford.

Schuckit, M. A. (1992) Inpatient and residential approaches to the treatment of alcoholism. In Medical Diagnosis and Treatment of Alcoholism, Mendelson, J. H. and Mello, N. K. eds, pp. 393–419. McGraw-Hill, New York.

Shaw, G. K., Waller, S., Latham, C. J., Dunn, G. and Thomson, A. D. (1998) The detoxification experience of alcoholic in-patients and predictors of outcome. Alcohol and Alcoholism 33, 291–303.[Abstract]

Stark, M. J. (1992) Dropping out of substance abuse treatment: a clinically oriented review. Clinical Psychology Review 12, 93–116.[ISI]

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]

Tucker, P. and Westermeyer, J. (1995) Substance abuse in patients with comorbid anxiety disorder. A comparative study. American Journal of Addictions 4, 226–233.[ISI]

Weiss, R. D., Mirin, S. M., Griffin, M. L. and Michael, J. L. (1988) A comparison of alcoholic and non-alcoholic drug abusers. Journal of Studies on Alcohol 49, 510–515.[ISI][Medline]

Weiss, R. D., Martínez-Raga, J., Griffin, M. L., Greenfield, S. and Hufford, C. (1997) Gender differences in cocaine dependent patients: a six-month follow-up study. Drug and Alcohol Dependence 44, 35–40.[ISI][Medline]

World Health Organization (1992) International Classification of Diseases, 10th revision. World Health Organization, Geneva.