Alcohol & Alcoholism Vol. 38, No. 4, pp. 369375, 2003 doi:10.1093/alcalc/agg091, available online at www.alcalc.oupjournals.

org

ACUPUNCTURE FOR ALCOHOL WITHDRAWAL: A RANDOMIZED CONTROLLED TRIAL

FRANOIS TRMPLER1, SUZAN OEZ1, PETER STHLI2, HANS DIETER BRENNER3 and PETER JNI4,5*
1 Psychiatric Hospital Mnsingen, 3110 Mnsingen, 2Department of Complementary Medicine (KIKOM), Inselspital, University of Berne, 3010 Berne, 3Department of Social and Community Psychiatry, University of Berne, Laupenstrasse 49, 3010 Berne, Switzerland, 4MRC Health Services Research Collaboration, Department of Social Medicine, University of Bristol, Bristol, UK and 5Department of Social and Preventive Medicine, University of Berne, Finkenhubelweg 11, 3012 Berne, Switzerland

(Received 1 February 2002; first review notified 12 March 2002; in revised form 11 March 2003; accepted 27 March 2003) Abstract Background and Aims: Previous trials on acupuncture in alcohol addiction were in outpatients and focused on relapse prevention. Rates of dropout were high and interpretation of results difficult. We compared auricular laser and needle acupuncture with sham laser stimulation in reducing the duration of alcohol withdrawal. Methods: Inpatients undergoing alcohol withdrawal were randomly allocated to laser acupuncture (n = 17), needle acupuncture (n = 15) or sham laser stimulation (n = 16). Attempts were made to blind patients, therapists and outcome assessors, but this was not feasible for needle acupuncture. The duration of withdrawal symptoms (as assessed using a nurse-rated scale) was the primary outcome; the duration of sedative prescription was the secondary outcome. Results: Patients randomized to laser and sham laser had identical withdrawal symptom durations (median 4 days). Patients randomized to needle stimulation had a shorter duration of withdrawal symptoms (median 3 days; P = 0.019 versus sham intervention), and tended to have a shorter duration of sedative use, but these differences diminished after adjustment for baseline differences. Conclusions: The data from this pilot trial do not suggest a relevant benefit of auricular laser acupuncture for alcohol withdrawal. A larger trial including adequate sham interventions is needed, however, to reliably determine the effectiveness of any type of auricular acupuncture in this condition.

Downloaded from http://alcalc.oxfordjournals.org/ by guest on May 9, 2013

INTRODUCTION In the 1970s, a number of enthusiastic reports were published about the effectiveness of acupuncture in combating all forms of addiction (Wen and Cheung, 1973; Kao and Lu, 1974; Sacks, 1975; Shuaib, 1976; Leung, 1977). Speculations focused on acupuncture-triggered release of endogenous opiates, supposedly blocking the mechanisms involved in withdrawal and craving (Wen et al., 1979). However, randomized controlled trials to support these speculations are scarce, particularly in alcohol addiction: up to 1990, ter Riet et al. (1990) identified two trials (Bullock et al., 1987, 1989). A Medline search up to May 2001 yielded another two (Worner et al., 1992; SapirWeise et al., 1999). All these trials had outpatient settings and focused on relapse prevention. As in conventional research into alcohol addiction (Whitworth et al., 1996; Anton et al., 1999), rates of drop-out and losses to follow-up were high, and the possibility of attrition bias (Jni et al., 2001) made interpretation of results difficult (ter Riet et al., 1990). Including inpatients instead of outpatients, and using acupuncture to reduce the duration of alcohol withdrawal states rather than to prevent relapses, could minimize patient attrition, and thus be a more appropriate clinical model to determine the efficacy of acupuncture for alcohol addiction. In this randomized controlled pilot trial we therefore set out to determine the feasibility of this approach, and to perform a preliminary comparison of laser and needle acupuncture with a sham intervention for alcohol withdrawal.

SUBJECTS AND METHODS Participants The trial was conducted at the Psychiatric Hospital Mnsingen, Switzerland, a comprehensive psychiatric facility with 417 beds. Patients were recruited from consecutive admissions to an alcohol withdrawal unit between October 1999 and February 2000. On admission, all patients received a standardized medical and psychiatric evaluation by a psychiatrist unrelated to the trial, and routine laboratory tests, including breath alcohol and drug urine screening. To be entered into the trial patients had to meet DSM-IV criteria for alcohol dependence (American Psychiatric Association, 1994). Exclusion criteria were current drug abuse, pregnancy, clinically evident cognitive impairment, current medical conditions requiring treatment, a history of schizophrenia, epilepsy or coagulation disturbances, current anticoagulation, and age <18 or >65 years. The relevant ethics committees approved the trial. All patients gave written informed consent. Randomization and interventions Within 24 h after admission, patients were randomized to one of three groups receiving daily sessions of: (1) low level laser stimulation, (2) needling or (3) sham laser stimulation at the ear. The randomization schedule was determined independently by one of us (P.S.) at a site remote from the hospital prior to the start of the study. The schedule was unblocked, generated by drawing lots, and concealed using sealed, opaque, sequentially numbered envelopes, which were opened only after a patient had definitely been registered in the trial. Patients were treated by one of two acupuncturists (F.T., S.O.), who chose between two and 10 (median eight) out of 24 prespecified ear points, regarded as specific for chemical dependency (Nogier, 1969; Wen and Cheung, 1973; Smith et al., 1982). Points were chosen individually in each session
369

*Author to whom correspondence should be addressed at: Department of Social and Preventive Medicine, University of Berne, Finkenhubelweg 11, 3012 Berne, Switzerland.

Alcohol & Alcoholism 38 Medical Council on Alcohol 2003; all rights reserved

370

F. TRMPLER et al.

according to principles of ear acupuncture, using pulse techniques (Nogier, 1982) and an electronic detection device (Punctoscope DT, Sedaletec, Inc., Irigny, France). Most frequently used points were Diaphragm [point #0 according to Olesons nomenclature (Oleson, 1995); 66% of sessions], Cheerfulness (#190; 62%), Insomnia (#195 F2, 58%), Sympathetic (#2; 53%), Spleen (#163; 51%), Laterality (#199; 50%), Lung (#70; 47%) and Shen Men (#1; 46%). Sessions lasted between 30 and 45 min and were performed daily until the end of withdrawal. Sessions were held in a quiet room detached from the unit, with hospital staff remaining uninvolved. Needle stimulation was performed after alcohol disinfection by inserting sterilized ear acupuncture needles (B type, stainless steel, 0.2 15 mm; Seirin, Inc., Neu-Isenburg, Germany) to a depth of 13 mm at ear points considered appropriate. Needles were twirled 180 degrees during insertion and left in place for up to 40 min. Because the local ethics committees were reluctant to accept invasive sham procedures, we were unable to include sham needling in this pilot trial. Therefore, patients could not be blinded as to the allocation to needle acupuncture. Low-level laser stimulation was performed using a 830 nm infrared semiconductor laser device (Modulas; schwa-medico, Inc., Ehringshausen, Germany), stimulating each point for the duration of 1 min. For sham stimulation we used a device of identical appearance and function, apart from an inactivated laser beam. Active and sham devices were coded with letters A and B, with acupuncturists unaware of the key, and codes changed twice during the trial. Acupuncturists, patients and hospital staff responsible for co-interventions and outcome assessment were therefore blinded for the laser versus sham laser comparison. Routine treatment of alcohol withdrawal in our hospital includes prescription of clomethiazole rather than benzodiazepines (Mayo-Smith, 1997), and all patients were prescribed clomethiazole on an individual basis, with symptom-guided dosages (Banger et al., 1997). In addition, patients could be prescribed benzodiazepines against withdrawal symptoms if it was considered appropriate by the treating psychiatrists. As was the case with clomethiazole, reduction of benzodiazepines was also symptom guided (Banger et al., 1997). Other drugs taken before study entry had to be kept constant until the end of the trial. Acupuncturists were not involved in routine care, and their interaction with patients was limited to the time required for the stimulation procedures. Assessments Severity of withdrawal symptoms was assessed using the Mainz Alcohol Withdrawal Scale (MAWS), a validated 12-item composite scale developed for mild to moderate alcohol withdrawal states with scores ranging from 0 to 34, the higher scores reflecting more severe symptoms (Banger et al., 1992). Each patient was rated daily at 16.00 h by trained nurses using this scale. Nurses and other hospital staff were required not to communicate with the patients about the allocated treatment. Since they were involved in routine care, we cannot exclude the possibility, however, that they intentionally or unintentionally undermined blinding regarding needle stimulation. Therefore, only the laser versus sham laser comparison was considered to have adequate blinding of outcome assessors. The pre-specified primary outcome was

the time from baseline to end of withdrawal, with end of withdrawal defined by scoring below two points on the MAWS (a priori). Time from baseline to end of pharmacological treatment of withdrawal symptoms (clomethiazole and benzodiazepines) was the secondary outcome. Because differential sedative intake across groups could potentially lead to performance bias (Jni et al., 2001), we also recorded daily clomethiazole doses, and accounted for additional benzodiazepine intake using equivalence doses (10 mg diazepam equivalent to 576 mg clomethiazole, a priori) (McGrath, 1975; Lapierre et al., 1983; Burroughs et al., 1985; MayoSmith, 1997). Assessments of side-effects of experimental treatments were performed daily until the end of inpatient care. Statistical analysis We chose a sample size of 48 (16 in each group) based on our experience that >90% of withdrawals in our unit are completed after 7 days. This sample size allowed us to detect a reduction of withdrawal duration of 1.5 days in active intervention groups as compared with the sham intervention group with 80% power at P = 0.05 (two-sided), assuming a standard deviation of 1.5 days and normally distributed data. However, outcome data were not normally distributed (P < 0.05 by Shapiro-Wilk test; Shapiro and Wilk, 1965). Therefore, we calculated medians and used KruskalWallis and Mann Whitney tests (Mann and Whitney, 1947; Kruskal and Wallis, 1952) for comparisons between groups. In exploratory analyses, we used the Cox proportional hazards model (Cox, 1972) to determine associations of pre-specified potential confounding factors with success of withdrawal in univariable and multivariable models, including: (1) disease severity at baseline as defined by MAWS; (2) sedative intake at baseline (clomethiazole and benzodiazepine) using equivalence doses; (3) average daily reduction of sedatives from baseline to end of withdrawal using equivalence doses; and (4) the presence of psychiatric comorbidity at baseline (depression, anxiety or personality disorder). Assumptions of the Cox model were assessed graphically and found to be approximately satisfied for all variables. We then calculated hazard ratios for intergroup comparisons as estimates of the relative risk for successful withdrawal, adjusting for these potential confounders. Finally, to determine whether results could be explained by confounding by differences related to the two acupuncturists or to the choice of acupuncture points, we included identifiers for acupuncturists and for most frequently used points into the Cox proportional hazards model and found results to be unchanged. All analyses were based on the intention-to-treat principle, with all subjects analysed in the originally assigned groups. One patient allocated to sham laser was withdrawn and therefore lost to follow-up on the sixth day of withdrawal, after a clinical decision to stop the allocated intervention because of a lack of efficacy. For the calculation of hazard ratios, we censored the patient after 6 days. Since the patient was allocated to the sham intervention, we assigned a withdrawal duration of 7 days for non-parametric comparison purposes. Considering that the last observed MAWS score was 4 on the sixth day of withdrawal, 7 days represent the shortest withdrawal duration the patient could theoretically have achieved. All analyses were performed in Stata (Stata, Inc., College Station, TX, USA). No adjustments were made for multiple comparisons; P-values are two-sided.

Downloaded from http://alcalc.oxfordjournals.org/ by guest on May 9, 2013

ACUPUNCTURE FOR ALCOHOL WITHDRAWAL

371

RESULTS Eighty-nine consecutive patients were screened; of those 37 did not meet the selection criteria, and four refused to participate. The remaining 48 patients entered the trial and were randomized (Fig. 1). All 48 patients were actively drinking at the time they were admitted, and satisfied at least five of the seven DSM-IV criteria for alcohol dependence. Baseline characteristics of randomized patients are presented in Table 1. Owing to the small number of patients, there were baseline differences between comparison groups. Patients in the sham intervention group tended to have more severe disease in terms of a longer duration of alcohol misuse, more withdrawals before entry, higher psychiatric comorbidity, and more frequent use of psychotropic medications other than sedatives. Presumably related to these indications of greater severity of alcohol dependence, in the sham intervention group the symptom-guided prescription of clomethiazole resulted in a higher prescription at commencement of withdrawal (median 1920 mg/day) than in the other groups (median 1540 mg/day in laser and needle acupuncture groups). We also found differences between treatment groups regarding the pharmacological management of withdrawal symptoms. Four patients took benzodiazepines in addition to clomethiazole (two allocated to laser, two allocated to sham laser). Because of intolerance to clomethiazole, another four patients exclusively took benzodiazepines (all allocated to sham laser). Finally, two patients did not take any of the prescribed drugs (one allocated to needle acupuncture, one allocated

to sham laser). The rate of decrease of clomethiazole dose (or equivalent) in the sham intervention group tended to be higher (median decrement 280 mg/day) than in laser (260 mg/day) and needle acupuncture groups (190 mg/day; P = 0.17 by KruskalWallis test). This difference may be due, in part, to the higher starting dose of clomethiazole in the sham intervention group. No relevant differences were observed for clomethiazole doses (or equivalent) at the end of withdrawal (median 670 mg/day in sham intervention group, 770 mg/day in laser and needle groups; P = 0.70). Figure 2 presents withdrawal durations of individual patients for each comparison group separately, whereas Fig. 3 shows KaplanMeier survival estimates for withdrawal. No differences were observed for the laser versus sham laser comparison, with identical median withdrawal durations and similar durations of clomethiazole use in the two groups (Table 2). Patients allocated to needle acupuncture were found to have a significantly shorter duration of withdrawal symptoms than patients allocated to the sham intervention, and also tended to have a shorter withdrawal symptom duration than patients allocated to laser acupuncture. However, hazard ratios corresponding to these withdrawal durations, which were calculated for exploratory purposes, decreased considerably and ceased to be statistically significant after adjusting for four potential confounders pre-specified in the protocol (Tables 3 and 4). Adjustment for sedative intake at baseline had a particular impact on effect estimates. We did not observe any local side-effects related to the experimental interventions. One patient in the needle acupuncture

Downloaded from http://alcalc.oxfordjournals.org/ by guest on May 9, 2013

Fig. 1. Trial flow diagram.

372

F. TRMPLER et al.
Table 1. Baseline characteristics of randomized patients Group

Parameter Age [years (IQR)] Females [n (%)] Duration of alcohol abuse [n (%)] Less than 5 years 515 years More than 15 years MAWS [score points (IQR)] Number of withdrawals before entry Positive alcohol breathing test [n (%)] Breath alcohol [ (IQR)] MCV [fl (IQR)] GGT [U/l (IQR)] Psychiatric comorbidity [n (%)] Depression Personality disorder Anxiety disorder Somatic comorbidity [n (%)] Cardiovascular Pulmonary Neurological Other Psychotropic medication [n (%)] Benzodiazepines Other sedatives Antidepressants Neuroleptics

Laser (n = 17) 43 (3746) 7 (41) 4 (24) 8 (47) 5 (29) 8 (510) 5 (210) 14 (82) 2.1 (0.43.1) 96 (92102) 75 (36184) 4 (24) 10 (59) 1 (6) 3 (18) 1 (6) 0 (0) 4 (24) 2 (12) 0 (0) 4 (24) 1 (6)

Needle (n = 15) 45 (3254) 5 (33) 3 (20) 5 (33) 7 (47) 8 (510) 6 (311) 13 (87) 1.7 (1.43.1) 95 (8797) 68 (44207) 5 (33) 4 (27) 2 (13) 3 (20) 1 (7) 1 (7) 2 (13) 0 (0) 2 (13) 5 (33) 1 (7)

Sham laser (n = 16) 49 (4055) 8 (50) 1 (6) 8 (50) 7 (44) 9 (612) 11 (515) 13 (81) 2.2 (0.52.9) 96 (9199) 74 (42121) 6 (38) 9 (56) 6 (38) 1 (6) 2 (13) 2 (13) 3 (19) 4 (25) 1 (6) 5 (31) 1 (6)

Downloaded from http://alcalc.oxfordjournals.org/ by guest on May 9, 2013

Data are presented as medians with interquartile ranges, or numbers with percentages. IQR, interquartile range; n, number of patients; , g/l serum alcohol equivalent; MCV, mean corpuscular volume (reference range 78105 fl); GGT, gamma glutamyl transferase (reference range <50 U/l for men, <32 U/l for women); MAWS: Mainz Alcohol Withdrawal Scale.

group, however, had self-limiting generalized convulsions of 5 min duration on the fifth day of withdrawal while she was sleeping, judged to be a withdrawal-related epileptic seizure on clinical grounds; no epilepsy-specific potentials could be detected in a subsequent electroencephalogram. None of the patients included developed delirium tremens. As reported in Fig. 1, one patient allocated to sham laser was withdrawn and therefore lost to follow-up on the sixth day of withdrawal, after a clinical decision to stop the allocated intervention because of a lack of efficacy. DISCUSSION We found no evidence to suggest a beneficial effect of auricular laser acupuncture in alcohol withdrawal. Our pilot trial was designed to have an 80% power to detect a minimally important difference in withdrawal duration of 1.5 days between the active intervention and sham intervention groups using parametric tests. Outcome data were not normally distributed, however, and we had to rely on non-parametric tests for comparisons between groups, compromising the trials power. The wide confidence intervals of median withdrawal durations and hazard ratios highlight that statistical precision was limited and that the absence of statistically significant differences between the laser and sham laser groups may be a type II error. Yet, median withdrawal durations in the laser and sham laser groups were identical, suggesting that the likely difference in withdrawal duration between the two

Fig. 2. Histogram presenting withdrawal durations of individual patients separately for each comparison group.

ACUPUNCTURE FOR ALCOHOL WITHDRAWAL

373

Downloaded from http://alcalc.oxfordjournals.org/ by guest on May 9, 2013

Fig. 3. KaplanMeier survival estimates for withdrawal. *One patient in sham intervention group censored after day 6.

Table 2. Duration of withdrawal and of clomethiazole treatment in the three study groups Medians with 95% confidence intervals Parameter Time to end of withdrawal (days) Time to end of clomethiazole use (days) Laser (n = 17) 4 (35) 6 (57) Needle (n = 15) 3 (23) 5 (46) Sham laser (n = 16) 4 (36) 6.5 (57) P-value from intergroup comparisons Laser vs sham 0.44 0.51 Needle vs sham 0.019 0.071

Duration of withdrawal was considered as the primary outcome, and duration of clomethiazole treatment as secondary outcome. P-values from MannWhitney test.

Table 3. Association of potential confounding factors with success of withdrawal Hazard ratios (95% confidence intervals) Parameter Baseline severity Sedative intake at baseline Daily sedative reduction Psychiatric comorbidity Basis Per 10 MAWS points Per 1000 mg/daya Per 1000 mg/daya Versus none Univariable analysis 0.23 (0.090.55) 0.59 (0.400.86) 0.52 (0.055.02) 0.84 (0.391.81) Multivariable analysis 0.30 (0.130.73) 0.53 (0.310.90) 3.74 (0.3145.90) 0.80 (0.371.75)

A hazard ratio below one indicates that the probability of a successful withdrawal is decreased, a hazard ratio above one indicates the opposite. Baseline severity and sedative intake at baseline had a significant inverse association with success of withdrawal (P < 0.05). a Expressed as clomethiazole equivalence dose, accounting for clomethiazole and benzodiazepine intakes (10 mg diazepam equivalent to 576 mg clomethiazole).

groups might be rather small. This notion is also supported by the lack of relevant differences in the duration of clomethiazole use. The small sample size not only led to limited power, but also to potentially important baseline differences between groups regarding disease severity. In exploratory analyses, we adjusted for these differences, but found the hazard ratio for successful withdrawal and the corresponding confidence intervals comparing laser and sham stimulation to be unaffected, suggesting that confounding is an unlikely

explanation for the lack of difference between laser and sham intervention. Conversely, adjusting for baseline differences considerably influenced the hazard ratio comparing needle with sham stimulation: after adjustment for sedative intake at baseline, the effect of needle acupuncture became less beneficial and non-significant. Hence, confounding by disease severity is one potential explanation for the observed differences between needle and sham interventions. Another potential explanation for the observed differences relates to the lack of a comparison group receiving sham

374

F. TRMPLER et al.
Table 4. Hazard ratios for successful withdrawal according to comparison groups, unadjusted and adjusted for four potential confounding factors Hazard ratios (95% confidence intervals)

Parameter No adjustment Adjustment for Baseline severity Sedative intake at baselinea Daily sedative reductiona Psychiatric comorbidity Full adjustment

Laser vs sham 1.41 (0.682.91) 1.25 (0.602.61) 1.55 (0.753.24) 1.41 (0.682.91) 1.42 (0.692.93) 1.37 (0.652.87)

Needle vs sham 2.21 (1.034.76) 2.36 (1.065.25) 1.79 (0.794.04) 2.48 (1.105.60) 2.18 (1.014.74) 2.16 (0.915.15)

Full adjustment relates to the exploratory inclusion of all four variables into the model. A hazard ratio above one indicates that the probability of a successful withdrawal is increased as compared with the comparison group. a Expressed as clomethiazole equivalence dose, accounting for clomethiazole and benzodiazepine intakes (10 mg diazepam equivalent to 576 mg clomethiazole).

to follow-up and drop-outs were high, the analysis could not be performed according to the intention-to-treat principle, and the possibility of attrition bias made interpretation of results difficult. To avoid this, we chose a different approach and focused on inpatients in acute alcohol withdrawal states. In this setting, the primary outcome could be ascertained for all except one patient (98%), and potential attrition bias therefore appears to be minimal. No problems were encountered in patient recruitment: 58% of screened patients qualified for the trial, an extraordinary low proportion refused consent (8% of those who qualified), and the necessary number of patients needed for our pilot trial could be recruited in our 20-bed withdrawal unit within 5 months. In conclusion, our preliminary results do not suggest a relevant benefit of auricular laser acupuncture in alcohol withdrawal. A large parallel group trial comparing auricular laser and needle acupuncture with adequate sham interventions is needed, however, to reliably determine their effectiveness in this condition. Unless results of such a trial indicate otherwise, the routine use of auricular acupuncture for alcohol withdrawal cannot be recommended. Contributorship Franois Trmpler conceived the trial, was responsible for protocol development and fundraising, performed acupuncture treatments, and was involved in data management. Suzan Oez participated in protocol development, performed acupuncture treatments, and was involved in data entry and management. Peter Sthli was involved in protocol development, was responsible for randomization procedures and advised on acupuncture treatment. Hans Dieter Brenner advised on the protocol and supervised the trial. Peter Jni advised on the protocol, supervised the trial, had main responsibility for data checking and management, planned and performed data analyses, and wrote the first draft of the paper. All authors were involved in writing the final draft. Franois Trmpler and Peter Jni are the guarantors of the trial.
Acknowledgements We are grateful to Brigitte Ausfeld-Hafter and Urs Lenggenhager, who advised on acupuncture treatment, and to Clive Adams, Shanon Collett, Terry Oleson and Anne Rutjes, who commented on the manuscript. This work was supported by the Hans Eggenberger Foundation, Zurich, the Novartis Foundation, Basle, the Mobiliar Foundation, Zurich, the Josephine Clark Foundation, Berne, and the Foundation for Integrative Medicine, Solothurn, Switzerland. P.J. was supported by the Swiss National Science Foundation and the UK Medical Research Councils Health Services Research Collaboration.

Downloaded from http://alcalc.oxfordjournals.org/ by guest on May 9, 2013

needling with patients only kept unaware of the allocations to laser and sham intervention. We attempted to blind the hospital staff who provided routine care and assessed the outcomes for all three comparison arms, but cannot exclude the possibility that blinding was undermined regarding the allocation of patients to the needle stimulation. Observed effects of needle acupuncture may therefore be entirely due to bias (Jni et al., 2001). The unequal provision of care apart from randomized interventions could have led to performance bias. We partially addressed this by adjusting for daily decreases in sedative dosages during withdrawal, and found no evidence to support this notion. The control of co-interventions was not optimal, however: we found differences in the pharmacological management of withdrawal, with more patients allocated to sham intervention taking benzodiazepines. We used equivalence doses to account for benzodiazepine intake, but this may be problematic, because trials comparing benzodiazepines and clomethiazole (McGrath, 1975; Lapierre et al., 1983; Burroughs et al., 1985) are scarce, small and methodologically inferior. Other potential biases related to the non-inclusion of sham needling are detection bias through unequal assessment of outcomes across groups, and differential placebo responses because of unequal patient expectations (Jni et al., 2001). These two potential sources of bias will only be properly addressed through the inclusion of a comparison group receiving an adequate sham intervention. A recently introduced telescopic sham needle, which does not penetrate the skin, could be used for this purpose (Park et al., 2001). High rates of losses to follow-up and drop-outs are a frequently encountered problem in clinical trials on alcohol addiction. Since patients excluded after randomization are unlikely to be representative of all patients included in a trial, this may introduce attrition bias (Jni et al., 2001). All randomized patients should therefore be included in the analysis and kept in the originally assigned groups, regardless of their adherence to the study protocol (intention-to-treat analysis). All four previous acupuncture trials on alcohol addiction had outpatient settings with some measure of relapse as the primary outcome; consequently, rates of losses

REFERENCES
American Psychiatric Association (1994) Diagnostic and statistical manual of mental disorders, 4th edn. American Psychiatric Association, Washington, DC. Anton, R. F., Moak, D. H., Waid, R., Latham, P. K., Malcolm, R. J. and Dias, J. K. (1999) Naltrexone and cognitive behavioral therapy for the treatment of outpatient alcoholics: results of a placebo-controlled trial. American Journal of Psychiatry 156, 17581764. Banger, M., Philipp, M., Herth, T., Hebenstreit, M. and Aldenhoff, J. (1992) Development of a rating scale for quantitative measurement of the alcohol withdrawal syndrome. European Archives of Psychiatry and Clinical Neuroscience 241, 241246. Banger, M., von Wilmsdorff, M. and Baier, M. (1997) Score-gesteuerte Behandlung des akuten Alkoholentzugssyndroms. Krankenhauspsychiatrie 8, 5661.

ACUPUNCTURE FOR ALCOHOL WITHDRAWAL

Bullock, M. L., Umen, A. J., Culliton, P. D. and Olander, R. T. (1987) Acupuncture treatment of alcohol recidivism: a pilot study. Alcoholism: Clinical and Experimental Research 11, 292295. Bullock, M. L., Culliton, P. D. and Olander, R. T. (1989) Controlled trial of acupuncture for severe recidivist alcoholism. Lancet i, 14351439. Burroughs, A. K., Morgan, M. Y. and Sherlock, S. (1985) Double-blind controlled trial of bromocriptine, chlordiazepoxide and chlormethiazole for alcohol withdrawal symptoms. Alcohol and Alcoholism 20, 263271. Cox, D. R. (1972) Regression models and life-tables. Journal of the Royal Statistical Society 34, 187220. Jni, P., Altman, D. G. and Egger, M. (2001) Systematic reviews in health care: assessing the quality of controlled clinical trials. British Medical Journal 323, 4246. Kao, A. H. and Lu, L. Y. (1974) Acupuncture procedure for treating drug addiction. American Journal of Acupuncture 2, 201207. Kruskal, W. H. and Wallis, W. A. (1952) Use of ranks in one-criterion variance analysis. Journal of the American Statistical Association 47, 583621. Lapierre, Y. D., Bulmer, D. R., Oyewumi, L. K., Mauguin, M. L. and Knott, V. J. (1983) Comparison of chlormethiazole (Heminevrin) and chlordiazepoxide (Librium) in the treatment of acute alcohol withdrawal. Neuropsychobiology 10, 127130. Leung, A. (1977) Acupuncture treatment of withdrawal symptoms. American Journal of Acupuncture 5, 43. Mann, H. B. and Whitney, D. R. (1947) On a test of whether one of two random variables is stochastically larger than the other. Annals of Mathematical Statistics 18, 5060. Mayo-Smith, M. F. (1997) Pharmacological management of alcohol withdrawal. A meta-analysis and evidence-based practice guideline. Journal of the American Medical Association 278, 144151. McGrath, S. D. (1975) A controlled trial of chlormethiazole and chlordiazepoxide in the treatment of the acute withdrawal phase of alcoholism. British Journal of Addiction to Alcohol and Other Drugs 70, (Suppl 1), 8190. Nogier, P. F (1969) Trait de lAuriculothrapie. Ed. Maisonneuve, Sainte-Ruffine.

375

Nogier, P. (1982) La perception du V.A.S. Auricular Medicine 28, 2529. Oleson, T. (1995) International Handbook of Ear Reflex Points. Health Care Alternatives, Los Angeles, CA. Park, J., White, A. R. and Ernst, E. (2001) New sham method in auricular acupuncture. Archives of Internal Medicine 161, 894. Sacks, L. (1975) Drug addiction, alcoholism, smoking, obesity, treated by auricular staplepuncture. American Journal of Acupuncture 3, 147150. Sapir-Weise, R., Berglund, M., Frank, A. and Kristenson, H. (1999) Acupuncture in alcoholism treatment: a randomized out-patient study. Alcohol and Alcoholism 34, 629635. Shapiro, S. S. and Wilk, M. B. (1965) An analysis of variance test for normality. Biometrika 52, 591611. Shuaib, B. (1976) Acupuncture treatment of drug dependence in Pakistan. American Journal of Chinese Medicine 4, 403407. Smith, M. O., Squires, R., Aponte, J., Rabinowitz, N. and Rodriquez, R. B. (1982) Acupuncture treatment of drug addiction and alcohol abuse. American Journal of Acupuncture 10, 161163. ter Riet, G., Kleijnen, J. and Knipschild, P. (1990) A meta-analysis of studies into the effect of acupuncture on addiction. British Journal of General Practice 40, 379382. Wen, H. L. and Cheung, S. Y. (1973) Treatment of drug addiction by acupuncture and electrical stimulation. Asian Journal of Medicine 9, 138141. Wen, H., Ho, W., Ling, N., Ma, L. and Choa, G. (1979) The influence of electroacupuncture on naloxone-induced morphine withdrawal II. Elevation of immuno-assayable beta-endorphin activity in brain but not in blood. American Journal of Chinese Medicine 7, 237239. Whitworth, A. B., Fischer, F., Lesch, O. M., Nimmerrichter, A., Oberbauer, H. and Platz, T. (1996) Comparison of acamprosate and placebo in long-term treatment of alcohol dependence. Lancet 347, 14381442. Worner, T. M., Zeller, B., Schwarz, H., Zwas, F. and Lyon, D. (1992) Acupuncture fails to improve treatment outcome in alcoholics. Drug and Alcohol Dependence 30, 169173.

Downloaded from http://alcalc.oxfordjournals.org/ by guest on May 9, 2013