Symptoms of Protracted Alcohol Withdrawal in Patients with Alcohol Use Disorder: A Comprehensive Systematic Review

Abstract

Background: Alcohol withdrawal syndrome (AWS) is characterized by different phases (acute, early and protracted). Protracted alcohol withdrawal (PAW) presents some symptoms, which may persist for several weeks, months or even years after drinking cessation.

Methods: We conducted a systematic review of the literature in major scientific databases on selected AWS symptoms (craving, sleep disorders, and anhedonia) in patients with alcohol use disorder.

Results: Of the 102 eligible publications (70 RCTs and 32 cohort studies), 88 provided data on craving, 21 on sleep disorders, and 1 on anhedonia. Overall, 37 studies assessed craving using the Obsessive Compulsive Drinking Scale (OCDS). Pooled OCDS decreased from 24.2 at baseline to 18.8 at 1 week, 10.3 at 1 month and 9.7 at 3 months. The corresponding estimates for treated individuals were 23.9, 18.8, 8.7, and 8.8, and for non-treated subjects, they were 25.3, 13.9, 13.2, and 11.4, respectively. In 4 studies assessing sleep disorders using the Epworth Sleeping Scale (ESS), the scale remained stable in time, i.e., 7.3 at baseline, 7.3 at 1 week, 7.2 at 1 month, and 7.1 at 3 months.

Conclusion: This study confirms the presence of PAW after the resolution of the acute phase of AWS. The pharmacological approach to managing PAW may ensure a more rapid reduction of symptoms in three weeks. We highlight the importance of studying PAW and the ability of pharmacological treatment to reduce its symptoms.

This review protocol is registered in Prospero (registration number: CRD42020211265).

Summary: This systematic review summarizes literature on major symptoms of protracted alcohol withdrawal in patients with alcohol use disorder. The pharmacological approach to manage protracted alcohol withdrawal ensures a more rapid reduction of symptoms (craving in particular), achieving in three weeks similar results obtained only after almost 6 months without treatment.

1. INTRODUCTION

Alcohol withdrawal syndrome (AWS) is a clinical condition that appears after the abrupt cessation or the reduction of alcohol use in a patient affected by alcohol use disorder (AUD) with physical dependence (DSM-5) [1]. Major symptoms associated with AWS include anxiety and dysphoria, insomnia and sleep disturbance, tremor, hyperactivity and craving [2-6]. As is well-known, these symptoms are severest during the initial period of abstinence. As early as 2010, Heilig and colleagues postulated the presence of three different phases of AWS: acute alcohol withdrawal (AAW) lasting 2-3 days, early alcohol withdrawal (EAW) lasting 2-3 weeks, and protracted alcohol withdrawal (PAW) lasting 3-6 months or more. In particular, PAW is characterized by the resolution of symptoms, such as tremor, sweating and nausea, and the persistence of other symptoms, such as anxiety, depression, anhedonia, craving, and sleep disturbance [7-9].

Given that today, most patients undergoing a moderate to severe form of AWS can be pharmacologically treated safely and effectively to alleviate persistence [3, 10] and, consequently, maintain abstinence from alcohol [2, 11, 12], it is important to investigate if and how specific symptoms of PAW persist in patients with AUD after the resolution of AAW and EAW. To our knowledge, no review has as yet investigated the frequency of PAW symptoms, overall and separately, in patients either undergoing or not the pharmacological treatment to reduce these symptoms.

Of the potential symptoms of PAW, we decided to exclude anxiety and depression to avoid the inclusion of studies based on patients with AUD and co-morbid psychiatric conditions, which would have rendered the attribution of these symptoms to PAW rather than to a psychiatric disorder questionable. In addition, other symptoms of PAW (i.e., chronic fatigue, reduced interest in sex, irritability, etc.) have not been investigated in clinical trials. We, therefore, conducted a systematic review of the literature to elucidate the persistence of craving, anhedonia and sleep disturbance after the resolution of AAW and EAW in order to define the clinical features of PAW and the response to pharmacological treatment.

2. METHODS

The methods used in this systematic review follow Prepared Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. This review protocol is registered in Prospero (registration number: CRD42020211265).

2.1. Data Sources and Search Strategy

We conducted a systematic review of the literature in Pubmed/MEDLINE, Embase and the Cochrane Database of Systematic Reviews (CDSR). The search string included terms for selected alcohol withdrawal symptoms, including craving, insomnia or sleep disorder and anhedonia, alcohol dependence, including alcoholism and alcohol use disorder, and withdrawal or abstinence (Table S1 ^{(1.2MB, pdf)} ). We considered only publications in English and limited our research to articles published over the last 15 years (since January 1, 2006). Original articles identified in the reference list of eligible meta-analyses and systematic reviews were also screened for inclusion.

2.2. Study Selection and Eligibility Criteria

Articles were eligible if they provided information at least one outcome of interest (i.e., craving, insomnia or sleep disorder and anhedonia) after alcohol detoxification. For the aim of the present review, only publications providing data from longitudinal studies (i.e., randomized clinical trials, RCT, or prospective cohort studies, including case-cohort studies) were considered eligible. Unpublished studies, conference abstracts and proceedings, dissertations, theses and, more in general, non-peer reviewed papers were not considered.

According to this comprehensive search strategy, on February 25, 2020, there were 744 articles found on Pubmed/MEDLINE, 63 on Embase (excluding those also in MEDLINE), and 13 on CDSR. After excluding 34 duplicates, a total of 786 articles were obtained. After retrieving the full text of all these publications, two independent researchers (EB and AL) evaluated each publication based on pre-defined eligibility criteria. Overall, 102 articles met those criteria (Fig. 1).

Fig. (1).

Flowchart of the systematic review on the symptoms of protracted alcohol withdrawal in patients with alcohol use disorder.

2.3. Data Extraction

Information on the first author, year of publication, study type, country, study period, type of population, method of diagnosis assessment, a possible treatment for alcohol cessation, and number of subjects and outcomes was retrieved and described (Table S2 ^{(1.2MB, pdf)} ). Various publications were also described according to the methods of outcome assessment, type of measures provided, estimates, other outcomes described in the study and comments on the outcomes (Table S3 ^{(1.2MB, pdf)} ).

2.4. Statistical Analyses

Each eligible article dealing with craving and insomnia or sleep disorders was described and summarized. Extracted data included the mean values of various scales used to determine the presence of craving, insomnia or sleep disorders. Pooled mean estimates for craving and insomnia or sleep disorder measures, at baseline and at various follow-up times after withdrawal, were estimated by computing the inverse-variance weighted means of various studies. For the most frequently used scales of craving (i.e., OCDS) and sleep (i.e., ESS), we fitted linear splines as a random intercept to take into account the correlation between repeated measures with the same subjects. We estimated the Akaike Information Criterion (AIC) of linear splines with 0 knots (linear regression) or 1 knot set at each time point. The optimal model for our data minimized AIC. With reference to OCDS, the same procedure was used for stratified analyses by treatment. To investigate differences between the treatments, the interaction between time after withdrawal and treatment was tested using linear regression models. In a post-hoc analysis, linear splines were also fitted for measures of anxiety and depression. All statistical analyses were performed using Stata 15 (StataCorp. 2017, College Station, TX, USA).

3. RESULTS

Of the 102 publications that met the eligibility criteria, including 70 RCTs and 32 cohort studies, 88 provided data on craving (Table S4 ^{(1.2MB, pdf)} ), 21 on sleep disorders (Table S5 ^{(1.2MB, pdf)} ) and 1 on anhedonia. Additionally, 73 provided at least two estimates of the scales used to measure selected symptoms at baseline (alcohol withdrawal) and after at least one time point at follow-up. Specifically, 64 studies provided information on craving, 12 on sleep disorders and 1 on anhedonia.

Of the 64 studies focusing on craving, 39 were RCTs and 25 were observational cohort studies. Overall, 37 (57.8%) studies assessed craving using OCDS, 16 (25.0%) using PACS, 14 (21.9%) using VAS, 3 (4.7%) using ACS, and 2 (3.1%) using ACQ.

Fig. (2). plots linear spline for alcohol craving measured by OCDS over time, considering all groups of subjects (treated and non-treated groups) (n=34 studies). The optimal model for OCDS had one knot at 18 days. OCDS decreased from 24.2 at baseline to 18.8 at 1 week, 13.5 at 2 weeks, 10.3 at 1 month and 9.7 at 3 months (Table 1). The corresponding estimates considering 29 studies that provided data on OCDS among subjects treated were: 23.9, 18.8, 13.7, 8.7, and 8.8, whereas for the 16 studies on non-treated subjects, the estimates were 25.3, 13.9, 13.7, 13.2, and 11.4, respectively. No statistically significant difference was observed between treated and non-treated groups (p=0.837; Fig. 3); however, even if linear splines showed knots at a different time, the OCDS declined below 10 points more rapidly for treated than non-treated subjects. In addition, among treated patients, no statistically significant differences in the reduction of OCDS scores were observed between subjects treated with GABAergic and non-GABAergic compounds. However, within the first three weeks, craving declined more rapidly among subjects treated with GABAergic (20 days) than with non-GABAergic compounds (39 days) (Table 1).

Fig. (2).

Linear spline for craving measured by Obsessive Compulsive Drinking Scale (OCDS) over time until one year of follow-up.

The knot is the time point where the optimal linear spline model changes slope. Knot observed 18 days after withdrawal; β1 (x<18): -0.76 (p-value<0·001); β1+β2 (x≥18): -0·01 (p-value=0·163), where β1 represents the slope of the line obtained from the linear spline before the knot and β2 after the knot.

Table 1.

Mean estimates at baseline and at various follow-up times for craving and sleep disorder measures obtained using a linear spline model with random intercept.

Outcomes	Number of Studies	Time (Days) of the Possible knot#	Mean Estimates of Various Scales at Different Time Points
			Baseline	1 Week	2 Weeks	3 Weeks	1 Month	2 Months	3 Months
Craving	-	-	-	-	-	-	-	-	-
OCDS	34	18	24·2	18·8	13·5	10·4	10·3	9·9	9·7
treated	29	21	23·9	18·8	13·7	8·7	8·7	8·7	8·8
GABA-ergic agents*	21	20	24·7	19·2	13·6	8·8	8·8	8·6	8·3
non-GABA-ergic agents^	14	39	22·7	18·7	15·8	12·3	8·8	4·6	6·6
non-treated	16	4	25·3	13·9	13·7	13·4	13·2	12·3	11·4
PACS	11	4	16·1	8·6	8·6	8·7	8·8	9·1	9·3
VAS	11	28	4·4	3·5	2·7	1·9	1·1	1·3	1·5
Sleep disorders	-	-	-	-	-	-	-	-	-
ESS	3	NA	7·3	7·3	7·3	7·3	7·2	7·2	7·1

OCDS: Obsessive Compulsive Drinking Scale; PACS: Penn Alcohol Craving Scale; VAS: Visual Analogue Scale; ESS: Epworth Sleeping Scale; NA: not available;

#The knot is the time point where the optimal linear spline model changes slope. NA means that no knot has been identified.

*GABAergic drugs: sodium oxybate, baclofen, gabapentin, lorazepam, diazepam, oxazepam, chlordiazepoxide and partial GABA-ergic agents (topiramate, valproic acid, pregabalin, carbamazepine, oxacarbazepine, zonisamide, tiagabine, clometiazole).

^non-GABAergic drugs: naltrexone, nalmefene, acamprosate, disulfiram, varenicline, mecamylamine, quetiapine, memantine, aripripazole, thiapride, escitalopram, trazodone, benfotiamine, thiamine, flumazenil, prazosin, doxazosin, dextromethorphan, metadoxine, oxytocine, nitric oxide, carnitine.

Fig. (3).

Linear spline for craving measured by Obsessive Compulsive Drinking Scale (OCDS) over time, until six months of follow-up, in strata of pharmacological treatment (the black line: treated with a pharmacological drug, the gray line: untreated or treated with placebo).

Treated (black): Knot observed 21 days after withdrawal; β1 (x<21): -0·75 (p-value<0·001); β1+β2 (x≥21): 0·00 (p-value=0·710)

Placebo (grey): Knot observed 4 days after withdrawal; β1 (x<4): -2·83 (p-value<0.001); β1+β2 (x≥4): -0·03 (p-value=0·007) Interaction treatment*time: β=-0·00, p-value=0·837.

Of the 12 studies on sleep disorders, 11 were RCTs and 1 was a cohort study, 4 (33.3%) assessed sleep disorders using ESS, 4 (33.3%) using TST, 3 (25.0%) by SOL, and 3 (25.0%) by PSQI. The optimal model measuring sleep disorders had no knot, and ESS remained stable, i.e., 7.3 at baseline, 7.3 at 1 week, 7.3 at 2 weeks, 7.2 at 1 month, and 7.1 at 3 months.

Only one study [13] provided information on anhedonia using a VAS scale and a SHAPS. This study, based on 44 patients treated with two different doses of Acetyl-L-Carnitine and 20 patients in the placebo group, showed that the VAS scale for anhedonia among patients treated with Acetyl-L-Carnitine decreased from 5.2 at baseline to 3.1 after 10 days and 2.4 after 80 days. The corresponding estimates in the placebo group were 6.2 at baseline, 5.6 after 10 days and 3.7 after 80 days from alcohol cessation. Similar decreases were observed using the SHAPS scale.

In a post-hoc analysis, among our selected studies, 30 provided information on one measure of depression at baseline and at least one time point at follow-up, and 24 provided information on anxiety. The most frequent scales were BDI (10 studies) for depression and STAI (7 studies) for anxiety. The optimal models for depression (BDI) had one knot at 9 days, with BDI decreasing from 17.8 at baseline to 12.7 at 1 week, 8.4 at 2 weeks, 8.4 at 1 month, and 8.2 at 3 months. The corresponding model for anxiety (STAI) had no knots, with STAI decreasing from 52.1 at baseline, 51.5 at 1 week, 51.0 at 2 weeks, 49.9 at 1 month, and 45.5 at 3 months.

4. DISCUSSION

This is the first systematic review showing the persistence of selected symptoms of PAW, including craving, sleep disorder, and anhedonia, after the resolution of AAW and EAW. Although these symptoms gradually taper in intensity, they do not completely disappear for several months.

Craving, defined as an intense, urgent, or abnormal desire to use an addictive substance or to follow compulsive behavior, is considered one of the most important symptoms characterizing addicted subjects. Therefore, it was included among the 11 items to perform a diagnosis of substance use disorder in the latest version of the Diagnostic and Statistical Manual of Mental Disorder-5 [1]. Craving has become a crucial symptom of the compulsive and obsessive desire for an addictive substance (in this case alcohol) and is used indirectly to monitor variations in the appetite for the substance after detoxification and counselling and pharmacological treatment to maintain abstinence and/or to avoid relapse. Several scales, such as OCDS, ACS, and PACS, monitor the intensity of craving and the efficacy of specific drugs in reducing craving according to time after drinking cessation in both RCTs and observational cohort studies. It is worth noting that OCDS, VAS, and PACS are the most commonly used scales to monitor craving for alcohol in patients with AUD [14]. OCDS is used to investigate the presence of craving characterized by a minimum of 0 and a maximum of 40 points, identifying two types of craving (obsessive and compulsive) [15]. VAS is a visual scale mainly used to investigate the intensity of this symptom with a range from 0 to 10 [16], and PACS comprises five items (each on a scale of 0-6) capturing craving over the previous week (score >20 identifies people with the craving symptom; score 15-20 defines people with a ‘moderate’ and ‘sometimes’ symptom, and score <15 indicates no symptom) [17, 18]. In addition, it is well-known that craving may be one of the most frequent symptoms accounting for relapse into heavy drinking [19, 20]. Moreover, years ago, craving was considered a symptom of AWS [21, 22], and only during the last couple of decades, it has acquired different considerations, becoming an independent “condition of mind” separate from AWS [17, 23] and with specific characteristics [24, 25]. It is important to specify that the neurobiological basis and physiopathology of craving in PAW are different to the craving that occurs in the early phases of AUD. In particular, craving observed during PAW is likely a consequence of neuro-adaptation, particularly of GABAergic and glutamatergic circuits induced by chronic alcohol consumption [26-28]. On the other hand, craving in the first phases of AWS may be related mainly to the sensitivity of opioid/dopaminergic circuits in the reward system and to cognitive and motivational models frequently associated with exposure to alcohol-related cues [29-31].

In our systematic review, as shown by the innovative application of linear splines summarizing the evidence, OCDS and PACS scores fall dramatically, particularly during the first 4 weeks of treatment, while the VAS score is more linear. This might be explained by the fact that craving intensity is drastically reduced during the first weeks of treatment without being fully suppressed during subsequent weeks and months. In addition, when studies are stratified into two subgroups by treatment (pharmacological vs. non-pharmacological), craving intensity appears to be reduced more markedly in patients undergoing pharmacological treatment than in patients treated with placebo; in this latter subgroup, indeed, the reduction is linear and not pronounced, unlike the other group. Given that differences between treatment and non-treatment groups are not significant, likely because the studies analyzed were not designed with this endpoint, some considerations are necessary since this is the first review to analyze the efficacy of treatment for protracted withdrawal syndrome. Interestingly, the treatment group manifested a significant reduction in craving after three weeks and maintained similar values until the end of the sixth month, while the non-treatment group displayed a weaker reduction in craving after 1 month, continuing to decrease and reaching the level of the treated group only after almost six months. A possible interpretation of these data is that craving follows a physiological curve of reduction during the first six months after interrupting alcohol abuse and that introducing a specific pharmacological treatment can substantially bring forward the reduction that normally would be obtained only after six months. According to this hypothesis to be tested by future studies, specific pharmacological treatment is recommended in the six months following the resolution of AWS in order to reduce craving, and thereby prevent alcohol relapse.

In addition, when the subgroup of studies was further divided into those following GABAergic agents, a more significant reduction in OCDS (and craving) was observed in the three weeks after alcohol discontinuation compared to non-GABAergic compounds. Although this difference is not significant and more studies are needed to understand what type of drugs can best treat PWS symptoms, the efficacy of GABAergic agents is consistent with the pathophysiology of PWS, which involves a down-regulation of the GABAergic system, which needs several months to return to equilibrium [26].

Sleep disorder or insomnia is one of the most frequent symptoms reported by patients during the period of active drinking and during AWS, either in the early or late phase [32]. Namely, time-reversible alteration in sleep latency, total sleep time, slow wave sleep, REM sleep and REM latency may be altered and may favor drinking relapse [33-35]. In particular, a review of seven studies with data on sleep, using polysomnographic recordings of the brain while people slept, found that sleep abnormalities can persist for 1 to 3 years after stopping alcohol consumption [26, 36-40]. These abnormalities include difficulty falling asleep, decreased total sleep time, and sleep apnea. This systematic review documents sleep disorder measured with the ESS scale, and shows that it improves in a linear manner with time after drinking cessation.

Moreover, anhedonia, a symptom where a person’s ability to experience pleasure is decreased, is frequently found in patients with AUD after the detoxification period, and can last for several months [26]. Very few studies have examined this particular symptom. Pozzi and colleagues examined anhedonia in individuals abstinent from alcohol, opioids, and/or other drugs for some time, with no other identified psychiatric disorder, concluding that anhedonia appeared to be a symptom of PAW unrelated to other clinical and psychosocial features [40]. Again, Martinotti and colleagues found that signs and symptoms, including anhedonia, lasted for the duration of a year long study of people recovering from AUD [9]. Our systematic review was able to examine only the above-mentioned studies. Therefore, no summary figure was obtained in relation to this particular symptom.

Furthermore, anxiety and depression are co-morbidities frequently found in patients suffering from AUD [41, 42]. It has been estimated that 30-50% of patients with AUD are affected by co-morbid psychiatric conditions [43, 44]. Therefore, studies including patients with psychiatric disorders were excluded from the systematic review, which, although not specifically designed to investigate anxiety and depression comprehensively, demonstrated in post-hoc analysis that these symptoms decreased during the 3-6 months after the cessation of the acute phase of AWS. However, the symptoms did not completely disappear because of their persistence.

This study has some limitations. We included studies investigating the efficacy of types of medication for the treatment of the acute phase of AWS; in these studies, benzodiazepines are the most frequent drugs employed; however, these studies are not numerous and used only the CIWA-Ar scale to measure the intensity of AWS [45], not scales measuring craving. Hence, they have no influence on regression analyses. Moreover, minor symptoms of PAW have not been investigated. Among its strengths, in addition to the originality and the comprehensiveness of the systematic review, this study enables the identification and treatment of a specific syndrome during the first months of alcohol abstinence, deemed a highly vulnerable period for drinking relapse.

CONCLUSION

In conclusion, this study confirms the persistence of craving and sleep disorder after the acute phase of AWS, as emerges clearly when the OCDS scale is considered in the placebo subgroup of patients. This supports the existence of a pathological condition. In addition, the pharmacological approach might have a role in reducing these symptoms, particularly during the first six months of treatment. Albeit without a significant statistical difference between the two classes of medications, a more rapid trend in the reduction of craving for alcohol was observed using GABAergic rather than non-GABAergic compounds. However, future studies are required to statistically confirm this difference. In addition, controlled clinical trials investigating the presence of major and minor symptoms by using PAW testing specific scales should be carried out, including trials investigating specific targeted interventions with the label and off-label drugs currently used in the treatment of AUD.

Finally, the results of this systematic review are of crucial importance not only for experts in addiction treatment, but also for physicians since the investigation of the presence of symptoms of PAW may direct pharmacological therapy appropriately in order to avoid heavy drinking relapse.

LIST OF ABBREVIATIONS

ACQ

Alcohol Craving Questionnaire

ACS

Alcohol Craving Scale

AUD

Alcohol Use Disorder

AWS

Alcohol Withdrawal Syndrome

BDI

Beck Depression Inventory

ESS

Epworth Sleeping Scale

OCDS

Obsessive Compulsive Drinking Scale

PACS

Penn Alcohol Craving Scale

PSQI

Pittsburgh Sleep Quality Index

RCT

Randomized Controlled Trial

SHAPS

Snaith Hamilton Pleasure Scale

SOL

Sleep Onset Latency

STAI

State-Trait Anxiety Inventory

TST

Total Sleep Time

VAS

Visual Analogue Scale

CONSENT FOR PUBLICATION

Not applicable.

STANDARDS OF REPORTING

PRISMA guidelines were followed for the study.

FUNDING

Laboratorio Farmaceutico CT S.r.l. financed the present work.

CONFLICT OF INTEREST

For the present systematic review, Silvano Gallus, Alessandra Lugo and Elisa Borroni acted as statistical and epidemiological consultants for the Laboratorio Farmaceutico CT S.r.l.

SUPPLEMENTARY MATERIAL

Supplementary material is available on the publisher’s website along with the published article.