|Home | About | Journals | Submit | Contact Us | Français|
The goal of this cross-sectional study was to assess the relationship of alcohol craving with biopsychosocial and addiction factors that are clinically pertinent to alcoholism treatment. Alcohol craving was assessed in 315 treatment-seeking, alcohol dependent subjects using the PACS questionnaire. Standard validated questionnaires were used to evaluate a variety of biological, addiction, psychological, psychiatric, and social factors. Individual covariates of craving included age, race, problematic consequences of drinking, heavy drinking, motivation for change, mood disturbance, sleep problems, and social supports. In a multivariate analysis (R2 = .34), alcohol craving was positively associated with mood disturbance, heavy drinking, readiness for change, and negatively associated with age. The results from this study suggest that alcohol craving is a complex phenomenon influenced by multiple factors.
Reward mechanisms and craving have been identified as important aspects of alcoholism. Craving, or the urge to use alcohol, is present in 54% to 72% of alcohol dependent subjects.1,2 The importance of alcohol craving especially at the onset of treatment is underscored by its association with relapse throughout the first 12 months of recovery.3–6 Despite this commonality of craving and its associated clinical importance, there is however no consensus on the definition of craving, the clinical factors that influence craving, or the optimal assessment measure for craving.7 In order to understand the construct of craving better, previous research has examined specific clinical factors associated with craving.
In principle, treatment of alcohol dependence, similar to other psychiatric treatments, utilizes the biopsychosocial model, which consists of biological, psychological, psychiatric and social factors. In the Biological component, gender differences in craving have been seen in female subjects (in comparison to male subjects), showing, a higher alcohol urge reactivity in response to negative mood states,8 and a positive association of alcohol craving with serum leptin levels at the onset of treatment.9 A family history of alcohol dependence has been associated with higher craving.10,11 In the Addiction component, alcohol craving has been positively associated with the severity of alcohol dependence, a history of recurrent detoxifications,1,12 and nicotine dependence.13 In the Psychological/Psychiatric component, psychological and psychophysiological aspects of alcohol craving have been extensively evaluated (see reviews14,15). Alcohol craving has been shown to be positively associated with a negative mood state or stress,16,17 and an increased anxiety level and a novelty-seeking temperament.18
There are, however, other variables in the biopsychosocial addiction model that are important to addiction treatment and whose relationship with craving is currently unknown. For instance, in the Biological component, age and craving have been indirectly associated through their relationship with dopamine.19–22 The racial difference in craving is incompletely understood currently. With regards to race, there is a lower prevalence in most African Americans of the Asp40 allele of the μ-opioid receptor gene; this gene has been shown to be associated with euphoria or craving in response to alcohol23 and effectiveness of naltrexone treatment.10,24 In the Addiction component, the problematic consequence of a subject’s alcohol use has been clinically seen to have a variable relationship with craving. The problematic consequence may increase the urge to use alcohol in some patients, whereas in others it helps to desist from alcohol use. In the Psychological component, motivation for change, a clinically pertinent treatment variable, may be a product of a conflict between drinking urges versus ambivalence and coping,25,26 thereby encouraging entry into treatment. In the Psychiatric component, sleep abnormalities specifically insomnia, are highly prevalent in alcohol dependent subjects and may predict relapse after treatment.27,28 In the Social component, higher levels of social supports have predicted reductions in alcohol use during treatment, whereas lower levels of social supports have predicted dropout from treatment.29 In addition, social supports have been shown to negatively moderate the association between incidences of stressors and craving in subjects with addictions.30 In summary, the relationship of craving with different variables in the biopsychosocial and addiction components, including, age, race, problematic consequences of alcohol use, motivation for change, sleep disturbance and social support, are currently unknown.
In this current study we evaluated the association between alcohol cravings and several new variables as well as previously investigated variables in the biopsychosocial addiction models. Specifically, we assessed the relationship of alcohol craving with the following factors: (1) Biological – age, gender, race, and a family history of alcohol dependence in first degree relatives (2) Addiction – problematic consequences from drinking, proportion of days of heavy drinking over the last 90 days, and nicotine dependence (3) Psychological/Psychiatric – motivation for change, mood disturbance, and sleep disturbance, and (4) Social – subjective social supports. Our hypothesis included the following: (1) alcohol craving would be positively associated with female gender, a family history of alcoholism in first degree relatives, proportion of days of heavy drinking, problematic consequences of drinking, nicotine dependence, motivation for change, mood disturbance and sleep problems (2) alcohol craving would be greater in the Caucasian subjects as compared to the African American subjects and (3) alcohol craving would be inversely associated with age and the presence of subjective social supports.
This cross-sectional study analyzed data obtained during the baseline assessment of subjects enrolled in the “Extending the Treatment Efficacy of Naltrexone” (ExTENd) trial (N = 315), regardless of whether or not they were subsequently randomized into that clinical trial. The study was conducted at the University of Pennsylvania Medical Center and the Philadelphia Veterans Affairs Medical Center (PVAMC).
The aim of the study was to define the response to naltrexone and determine the appropriate clinical algorithm for responders and non-responders to treatment. The study participants were recruited through advertisements in the local media or from the clinical services at the Philadelphia VA Medical Center. The study was reviewed and approved by the Institutional Review Boards of the University of Pennsylvania and the Philadelphia VA Medical Center with all participants providing written informed consent prior to study participation.
Participants were 18 years of age or older, had a current DSM IV diagnosis of alcohol dependence and were able to comprehend and converse in English. They were required to have been drinking within 30 days of randomization into the study and report a minimum consumption of 48 standard alcoholic drinks in a consecutive 30-day period prior to being recruited, and have at least 2 or more days of heavy drinking (defined as over 5 drinks in males and 4 drinks in females) in the 30-day period prior to recruitment.
Individuals were excluded from the study if they met any of the following criteria: dependence on any substance with the exception of alcohol, nicotine and marijuana, as defined by the DSM IV criteria; a lifetime history of any significant psychiatric diagnosis (e.g. bipolar disorder, schizophrenia or any psychotic disorder); evidence of a significant medical illness (hepatic/hematological/pulmonary/endocrine/cardiovascular/renal or gastro-intestinal disease); use of a psychotropic medication (including diphenhydramine and disulfiram) regularly within seven days prior to recruitment; current risk of becoming pregnant or currently pregnant.
The SCID-1 was used to establish the diagnosis of alcohol dependence.31
The MINI version 5.0.0 was also used to verify the diagnosis of alcohol dependence, as well as to assess for any manic or hypomanic episode, panic disorder, PTSD, GAD and psychotic disorder.32
The PACS, a 5-item scale assessing for craving over the last seven days, was used to assess for alcohol craving in the subjects. Item scores range from 0–6 in increasing order of severity, with a total PACS score ranging from 0–30. The PACS possesses good psychometric properties (Cronbach’s α = 0.92).33 We will use the term craving in this manuscript to denote “urge to use alcohol” or “the desire to drink.”
The DSSI yields 4 subscales, namely subjective support, social network, social interaction, and instrumental support. The subjective support subscale, a distinct entity that measures support from family and friends, is inversely related to mental distress and is perhaps the most important dimension related to health outcomes.34 We used the subjective support subscale of the DSSI to assess for social support.
The SIP is a 15-item questionnaire that assesses for the severity of alcohol dependence across five alcohol related problem areas over the past 3 months (recent) and lifetime. The total SIP score has good internal consistency (Cronbach’s α = 0.79) and test-retest correlation of 0.74.35
The TLFB was used to record the quantity and frequency of drinking during the 90 days prior to entry into the study.36 The drinking variable of interest was the number of heavy drinking days (≥5 standard drinks per day in males and ≥4 standard drinks per day in females). The proportion of days of heavy drinking (PDHD) was estimated as the fraction of days of heavy drinking to the total number of drinking days.37
The FTND is a 6-item measure to assess for nicotine dependence with a total score range of 0–10.38
The URICA, a 32-item self report inventory, assesses motivation for change.39 Several studies that included alcoholics report good internal consistency of the 4 subscales.40 The composite measure, Readiness for Change, is obtained by summing the totals on the Contemplation, Action and Maintenance subscales and then subtracting the score on the Pre-contemplation subscale; this measure has shown good concurrent validity with baseline characteristics and change process variables.41
This 37-item scale was developed to assess transient, distinct mood states with 6 subscales and a global distress score, Total Mood Disturbance.42 The internal consistency (Cronbach’s α) for the subscales range from 0.76 – 0.95 and from 0.87 – 0.90 for the Total Mood Disturbance.43
This validated 12-item self report sleep measure assesses the different dimensions of subjective sleep.44 This scale has 4 subscales which include (mean ± SD, of the reference group published by the authors): sleep disturbance (initial and middle insomnia and “sleep not quiet”, 29.20 ± 23.37), quantity of sleep (duration of sleep at night, 6.93 ± 1.40 hours), sleep adequacy (amount of sleep to feel well rested upon waking in the morning 60.67 ± 25.38) and daytime somnolence (daytime drowsiness and taking naps, 26.41 ± 19.82). Higher values on the sleep disturbance and daytime somnolence denote bigger problems, whereas a lower number on the sleep adequacy scale denotes a bigger problem. This scale also generates a global score, known as the Sleep Problems Index (SPI) (9 items, 29.15 ± 18.04), which has been shown to have the strongest correlations with health measures. The central tendency (standard deviation) for the total sleep duration is 6.93 ± 1.40 hours.45 The internal consistencies (α) of the sleep disturbance, daytime somnolence, and SPI scales range from 0.63 –0.82; that for sleep adequacy ranges from 0.63 – 0.73. The reliability coefficients of these scales range from 0.75 to 0.86.45
The FISC is designed to assess family history of psychopathology and psycho-active substance use disorders in first- and second-degree relatives.46
The tests of significance included the independent sample t-tests (craving in males and females, craving in Caucasian and African American subjects, first-degree family history of alcohol dependence versus other family history). Correlation analyses using bivariate correlations (Pearson’s correlation coefficient) were used to assess for correlation between pairs of variables prior to conducting univariate analysis. Univariate analyses to assess for covariates of craving were conducted using linear regression analyses with the total craving score as the dependent variable. Independent variables were comprised of demographics (age, gender, race), subjects with family history of first-degree relatives, Subjective Support-DSSI score, SIP-recent total score, drinking scales derived from the TLFB, FTND, URICA scale scores, POMS-SF scale scores, and the MOS Sleep Scale subscales. Summary scores for the different scales were used wherever indicated (i.e. SIP-Recent total score, Proportion of Days of Heavy Drinking, Total FTND score, Readiness Scale score from the URICA scale,41 Total Mood Disturbance from the POMS-SF,42,43 and Sleep Problems Index from the MOS Sleep Scale44). A Bonferroni correction with p = .035 was applied for the multiple comparisons. A multivariate analysis was conducted using a backward regression model simultaneously entering all the significant variables from the univariate analyses, which included the following: mood disturbance, PDHD, readiness for change, age, subjective social support, race, sleep disturbance, and SIP-recent. Since the current analysis is a preliminary analysis of multiple new variables, as well as some previously investigated variables,1,5,16,21 all the variables were therefore simultaneously entered into the final multivariate model.
The SIP-recent and the PDHD assess for the severity of alcohol addiction; however, minimal colinearity exists between these two variables since they assess for different aspects of alcoholism (quality and quantity of alcohol dependence, respectively). Both variables were significant covariates of craving, hence were simultaneously entered into the final multivariate model. Mood disturbance and sleep problems represent different domains of psychiatric variables and exhibit minimal co-linearity. Since sleep has not been assessed for its relationship with craving but was individually seen to be a significant covariate, the global sleep score was also entered into the final model.
The majority of our subjects were middle-aged, male, Caucasian, and non-Hispanic, with more than 12 years of education. One-third of the subjects had a first- degree relative with alcohol dependence (see Table 1). Most of the subjects were married, living with at least one other person, and employed within the last 30 days. Six (2.1 %) of the subjects reported no craving. The mean craving score was 15.69 ± 7.16 (range of 0–30).
The age of the subjects was significantly and inversely associated with craving (p = .001), (see Table 2). Caucasian subjects had significantly higher craving 16.61 ± 6.78 (mean ± S.D.) as compared to the African American subjects 13.62 ± 7.46 p = .003. Univariate analyses showed that race was significantly associated with craving even after controlling for potential moderating variables of education and income (p = .001). No difference in the craving between male and female subjects was seen. A family history of alcoholism in the first degree relatives was not associated with craving (see Table 2).
The SIP-total score (SIP-recent, over the past 3 months) and SIP-lifetime total score were significantly associated with craving (p <.001). The SIP-recent total score was also positively associated with proportion of days of heavy drinking (r = .22, p = .001), readiness (r = .29, p <.001) and mood disturbance (r = .58, p <.001), and negatively associated with age (r = −.23, p <.001).
Subjects drank 11 ± 6 standard drinks per day, with the proportion of heavy drinking days predominating over the last ninety days at 68 ± 29%. All the drinking variables were positively associated with craving. A trend towards a relationship between nicotine dependence and alcohol craving was also seen (p = .07) (see Table 2).
As expected in these treatment-seeking subjects, their stages of change profile showed higher scores in stages of contemplation, action, and readiness for change scales (see Table 2). The readiness for change scale was significantly and positively associated with craving. Additionally, the readiness for change scale was positively associated with SIP – total lifetime (r = .26, p <.001) and SIP – total recent (r = .29, p <.001).
The five negative mood state subscales and the global distress score from the POMS-SF showed increased mean scores, in comparison to the initial data for the POMS-SF,42 (see Table 2). All of the mood scales were significantly associated with craving. An interaction analysis of mood disturbance and subjective social support did not show a significant relationship with craving.
Comparing the mean score of our subjects to the normative data,45 our subjects showed a decreased mean subjective total sleep time [6.07 ± 1.54 hours], increased sleep disturbance scale score [44.06 ± 26.67], decreased sleep adequacy scale score [48.82 ± 27.17)], increased somnolence [27.45 ± 20.50], and increased sleep problems index [39.26 ± 19.78] (see Table 2). All the sleep scales were significantly associated with craving (see Table 2). An association between the mood and sleep scales was seen on correlation analysis (r-values ranging from 0.36–0.46, with all p <.001).
Better social support was associated with more craving (p = .002). There was a significant correlation between social support and mood disturbance (r = .34, p = <.001); however, no correlation existed between social support and readiness for change (motivation). Thus social support was positively associated with craving and mood disturbance, however as mentioned above, there was no interaction of mood disturbance and social support with craving.
In the multivariate analysis (see Table 3), craving was significantly associated with the proportion of days of heavy drinking, total mood disturbance, the readiness for change scale and the subject’s age. This model explained 34% of the variance of the craving. The variables that were not seen to be significant in the final model included: race, subjective social support, SIP-recent, and sleep problems.
Prior craving studies have reported the individual relationship of alcohol craving with gender, family history, social support, intensity of alcohol dependence, nicotine dependence as well as stress or mood state. In this cross-sectional study of actively drinking alcohol dependent subjects, we evaluated several novel covariates and re-evaluated some previously investigated covariates of craving. Unique covariates of alcohol craving found in this study included the following: race, subjective social support, problematic consequences of alcohol use, motivation for change, and sleep complaints. Alcohol craving was negatively associated with age. Craving was also positively associated with alcohol use, and mood disturbance. In a multivariate analysis, 34 % of the variance of alcohol craving was predicted positively by heavy drinking, mood disturbance and motivation for change; alcohol craving was negatively predicted by age.
These results support the individual association of most of the variables with craving, as we had hypothesized, with the exception of gender, family history of alcohol dependence and nicotine dependence. The lack of an association of gender and family history of alcohol dependence with craving may be due to sample characteristics; specifically, the relatively low number of female subjects, and most of our subjects lacking a family history of alcohol dependence in the first degree relatives. Similarly, the low level of nicotine dependence may have accounted for the lack of association of nicotine dependence with alcohol craving.
Age was inversely associated with craving in our subjects. This finding may help account for the better prognosis for older adults in alcoholism treatment.47 The increased craving in our Caucasian subjects may be associated with the racial differences in the efficacy of the drug, naltrexone, a mu-opioid receptor antagonist medication that decreases craving. The greater efficacy of naltrexone in Caucasian than African-American subjects has been related to the underlying higher prevalence of the genetic polymorphism of the Asp40 allele of the mu-opioid receptor 1 gene in Caucasians.10 In our study, social support was independently associated with craving. This finding may be possibly due to our actively drinking alcohol dependent subjects being in the company of others, who are also actively involved with drinking. Social supports was however, neither a predictor in the final model, nor seen to moderate the relationship of mood and craving in our subjects, as reported in previous studies.30,48
The proportion of heavy drinking days prior to entering treatment, a sign of severity of alcohol dependence, was significantly and positively associated with craving, as previously reported.1,12 Alcohol craving was associated with the problematic consequences of alcohol use, through its relationships with mood disturbance, heavy drinking, and motivation for change. Thus the increased craving secondary to psychosocial consequences after prolonged drinking may compel the alcohol dependent patients to seek treatment.
Motivation for change was positively associated with craving, another unique finding in this study. Motivation for change has previously been associated with decreased drinking at entry into the study,49 and with decreased post-treatment drinking after one year.50 Once drinking resumes, however, factors other than stages of change appear to be associated with the drinking.51 Using the construct of self efficacy,52 we may speculate that high craving state is associated with distress in these participants. This high craving may thus be associated with an increased motivation to seek change or treatment.
The subjects also reported symptoms of mood disturbance, that was individually associated with craving as has been shown before.17,53,54 Consistent with prior literature, our actively drinking subjects reported sleep complaints in the domain of insomnia.55 The individual relationship of sleep problems with craving, is another novel finding in this study. The variance of sleep abnormalities, however, was explained by the underlying mood disturbance, consistent with previous literature.56 Intensified sleep problems and REM sleep abnormalities have been reported in subjects with secondary depressive disorder.28,57 Treatment of insomnia in recovering alcoholics may therefore be considered a therapeutic target for the treatment of the mood symptoms and craving, and thus the underlying alcohol dependence.
The relationship between craving and these multiple variables in the final model may have an underlying neuro-anatomic basis. Alcohol craving have been associated with brain activity in the nucleus accumbens, anterior cingulate cortex, orbitofrontal cortex, and the limbic system.58,59 Neuro-anatomical correlates of sleep disturbance in depressed subjects have included cingulate cortex, insula, limbic system and basal ganglia.60 Reasoning and informed decisions have been associated with the prefrontal and cingulate cortex61 and the posterior cingulate and the hippocampus62 respectively. In summary, the areas of prefrontal cortex, orbitofrontal cortex, limbic system and basal ganglia may be crucial to the understanding of the underlying interrelationship of motivation and mood disturbance, and craving in alcoholics.
The limitations of this study were threefold. First, this was a convenience sample of treatment-seeking subjects and may not be representative of all alcohol dependent subjects. Second, the study was cross-sectional in nature; therefore the temporal relationship of alcohol craving with the other variables could not be elicited. Third, this study lacked the broad inclusion of certain minority populations (including Asians and Native Americans). Longitudinal studies will be necessary to clarify the relationships between craving and motivation for change, and, the interplay of different variables including social support and sleep problems with mood disorders and craving. These relationships may have important implications for integrated treatment planning in alcohol dependence.
This study was supported by grant # 5R01AA014851 from the National Institutes of Health, Bethesda, MD (Dr. Oslin); ClinicalTrials.gov Identifier: NCT00115037.
We thank Drs. Les A. Gellis, Nirav Patel, Shahrzad Mavandadi, Kevin Lynch and Ms. Lauren Courtright for their help and advice in the preparation of this manuscript.
Declaration of Interest
The authors report no conflict of interest. The authors alone are responsible for the content and writing of the paper.