ALCOHOL RELAPSE AFTER TREATMENT: THE INFLUENCE OF GENDER AND MARITAL STATUS 1

Running Head: Alcohol Relapse after Treatment: The Influence of Marital Status and Gender

Alcohol Relapse after Treatment: The Influence of Marital Status and Gender

Stephanie L. Hood

J. Scott Tonigan, PhD

The Center on Alcoholism, Substance Abuse, and Addictions

University of New Mexico

Abstract:

Background: To date, studies have not investigated the effects of gender, marital status, and interaction in spite of its clear importance to understanding the mechanisms accounting for alcohol relapse. Such an investigation, using Project MATCH is a prerequisite to understanding how, if at all, coping is important in Marlatt’s Relapse Prevention model.

Methods:Using Reasons for Relapse Questionnaire from Project MATCH, the 1,726 participants self reportedinfluential factors of alcohol relapse. Using SPSS 18.0, all general linear models (GLM) models included gender, marital status, and the interaction between gender and marital status. All inferential tests were evaluated at p < .05.

Results:At three months, spousal support played a greater significant role for females than males (p .019) and amongst married individuals (p<.026). Conversely non-married individuals and the relationship between gender and marital status played a significant role for “people outside the family” (both at p.000). Females also reported greater significance than males for “feeling down or blue.” (p<.026). At nine months, males reported greater significance than females about “feeling good” (p.001), females reported greater significance about feeling “uptight or anxious” (p<.016), and for married individuals reported having greater significance than non-married individuals for “letting down your guard about alcohol”. Lastly at 15 months, married individuals reported a greater significance for spousal or spousal equivalent (p.034), and females continually reporting significance for “feeling uptight or anxious” (p<.002).

Conclusion: This study compared the retrospective reasons men and women provided for a relapse, with attention given to how these reasons may be influenced by patient marital status.

Key Words: Marlatt’s Relapse Prevention Model, Alcoholism, Gender, Marital Status, Project MATCH

Alcohol is the most frequently abused substance in the country, with nearly 85,000 deaths yearly. Unknown to many, alcohol is the only drug known to result in death from withdrawals (Mokdad et al., 2004; Schneider et. al., 2003). In spite of effective and evidence-base treatments relapse is common. Approximately 93% of all problem drinkers, for example, resume alcohol use in the time-span of 5 years (Emrick and Hansen, 1983). One of the prominent models of relapse (Marlatt and Gordon, 1985) arguesthat patients are poorly equipped to identify relapse prone situations and therefore patients must learn to identify relapse situations and appropriate responses to prevent relapse. In this light, Marlatt’s approach includes environmental and emotional considerations. They categorize these into two groups; one is immediate determinants (ex. high risk situations, coping skills, and outcome expectancies) and, the second, covert antecedents which include lifestyle imbalance (stress, sadness, etc.) and urges and cravings for alcohol. Further, their relapse prevention (RP) model includes therapist-patient interactions intended to identify high risk situations and the conditions under which they may or may not drink. Central to avoiding relapse is abstinence self-efficacy or the confidence one has in avoiding the use of alcohol. Marlatt and Gordon’s model therefore seeks to provide the patient with the “big picture” in trying to control these situations (Larimer, et al., 1999). Within this model therapists seek to elicit patient responsibility for their drinking, and to encourage patients to take a more immediate view of relapse triggers. Marlatt also describes that therapists must have their patients examine the myths, placebos, and misperceptions they hold about the positive effects of alcohol. This allows therapists to interpret patient’s alcohol expectancies before they drink which are often based on these myths and placebos (ex. how they will feel, the high, etc.) versus what the real side effects are, which a majority of them are negative side effects (sleepiness, emotional instability, etc.). The RP suggest that one can manage lapse or relapse by identifying triggers or high-risk situations and using cognitive restructuring techniques learned in therapy to avoid drinking.

Although the RP model has gained widespread acceptance and its use of empirical support has been mixed about its’ validity. To illustrate, the Relapse Replication & Extension Project (RREP), which was funded by the National Institute on Alcohol Abuse and Alcoholism (Larimer, et al., 1999), studied predictors of relapse every two months for a year. At each assessment, they measured five different domains: (1) the occurrence of negative life events; (2) cognitive appraisal variables including self-efficacy, alcohol expectancies, and motivation for change; (3) client coping resources; (4) craving experiences; and (5) affective/mood status. This study’s impact is important because as a multi-site trial with a large sample substantial confidence can be placed in study findings. Surprisingly, high-risk situations did not predict later relapse as suggested by Marlatt, suggesting that individuals are especially good at identifying high risk situations. In contrast, coping skills were predictive of relapse as proposed in the model, with positive approach and negative avoidance coping, predicting 85% of the cases that had relapsed at six-months.

An important question is, do men and women differ in their relapse rates? Based upon a large meta-analytic study (Vannicelli & Nash, 1984) this deceptively simple question appears to have an elusive answer. First, according to these authors women were underrepresented in alcohol research. To illustrate, women comprise about 20-25% of alcohol dependent adults in treatment but in reviewing 259 studies only 7.8% of the study participants were female. Jarvis (1992) meta-analytically analyzed Vanicelli & Nash’s study to determine if women have poorer prognoses than men. Jarvis found no support for the claim that gender moderated outcome. Similar conclusions were drawn by Annis and Liban (1980) in their review of 23 studies. Here, men and women had relatively similar treatment outcomes although these authors highlighted that predisposing factors predicting relapse (men vs. women) have not been explored. With Jarvis’ meta-analysis of qualitative and quantitative research she looked at relapse from 3-6 months, 7-12 months, and 12+ months. Jarvis concluded from her review that: (1) men have had better inpatient results then women, (2) men and women have better outcomes with different treatments, (3) women have a lower alcohol intake than men, which might result in successful/non successful treatment outcomes, and (4) women respond better to one-on-one therapy due to high stigma linked to the “role” of women with alcoholism and that they can discuss more issues in confidence rather than in a group. Unlike with men, Jarvis notes that men like to discuss their problems amongst other men who are going through the same thing and would rather do this in a group setting. This is also discussed in Pemberton’s article which 50 females were admitted to the hospital and evaluated by different factors about their drinking onset. Their findings supported Jarvis conclusions; men feel more comfortable discussing their alcohol issues amongst others, while women remained secretive about their alcoholism. Jarvis found that gender difference was very small and that women have better outcomes in the first year, while men do better past 12+ months. Regarding treatment, Krentzman, et. al’s study discovered no gender difference in one year sobriety for their study (who identified themselves as AA members). These women’s demographic information were statistically different than males which were more likely to be White, more educated, and were more employed. This goes against Rubin and colleague’s study who found gender differences that women drink to intoxication more than men, while men experienced more positive mood states during relapse than women. Walton, et al., 2003 reports that men have higher self-efficacy, which predicts lower alcohol use, and women report greater resource needs, which predicted more drug use. Furthermore, in Timko, et al.’s results found that there are gender differences in baseline status and help-seeking at their 8 year follow up.

Evidence suggests a complex relationship between relapse and gender; one that may be moderated by patient marital status. To begin, Walton, et al., (2003) followed 180 participants after treatment, and interviewed them at 1-month and two years. It should be noted that Walton, et al., 2003 studied alcohol as well as general substance abuse. He divided results into two graphs (alcohol and drugs). Within this sample, 25.4% were never married, 1.3% was widowed, and 40.8% separated or divorced. They found that alcohol relapse correlated with income, marital status, cravings, leisure activities, self-efficacy (previously mentioned with RP), and resource needs. Marital status which is our main concern, was significantly directed both in interpersonal assets (.45) and also social/environmental (.46). Adding to that, Pemberton’s research found that females find it more difficult to establish a satisfying role amongst their family due to a demanding husband, another illness they might be suffering from, or failure to adapt to the loss of their husband. Self-efficacy significantly mediated indirect effects of income, gender, marital status, and problem severity in Walton’s study. Markers of low self-efficacy included lower income, being female, greater problem severity, and being unmarried. Walton, et al. (2003) found that coping in this study did not predict post treatment alcohol as suggested by Marlatt and Gordon. Previous work (Cronkite & Moos, 1984) has found that being married is associated with more positive treatment outcome for male substance abusers, but the impact of being married on women’s relapse is not consistent and needs further research. Schneider et al, (1995), states that results indicated being married is consistently related to less drinking for men, while for women, being married contributes to relapse in the short term.

Central in the RP model is how people identify and cope with stressful situation. According to Noone et al., (1999) the lack of coping resources, poor social support, low efficacy, loss of control, and negative coping can all lead to relapse. A moderator of stress that has not been tested is cognitive hardiness, which is the motivation and high efficacy personality needed to reduce stress which could decrease the rate of relapse. The health belief model also helps reduce drinking behavior by describing in depth to patients that drinking deteriorates health and resorting to sobriety can improve these health implications like liver disease, alcoholic hepatitis, and severe abnormalities to the brain. In this study, stress was assessed by their health, work, financial stability, family, environmental, and social hassles. The only demographic variable which showed a significant multivariate effect was level of education. Of course, high efficacy and high social support decreased drinking according to Noone et al., (1999) and also Brown, et. al., 1995. Noone’s findings were that 26% remained abstinent and 28% drank at harmful levels of 6+ drinks a day. Stressors in the patients’ lives were measured a month prior before their follow-up date and was said to be a significant predictor. Social support is very important to help recovering alcoholics and concludes that treatments should include stress management techniques, encouragement for patients to utilize ongoing social support, and positive coping strategies to increase self-efficacy.

The scope of this study is to investigate the attributions people make about why they relapse. To date, studies have not investigated this topic in spite of its clear importance to understanding the mechanisms accounting for alcohol relapse. Specifically, this study will compare the retrospective reasons men and women provided for a relapse, with attention given to how these reasons may be influenced by patient marital status. Such an investigation is a prerequisite to understanding how, if at all, coping is important in the RP model. The context for this investigation is a large multi-site clinical trial (PMRG, 1998; 2002) studying the effectiveness of cognitive behavioral, motivational enhancement and 12-step outpatient treatment. In examine reasons for relapse we will adopt the practice of grouping reasons according to whether they are situation or emotional in nature. Given that relapse typically occurs rapidly after treatment, our study will focus on the first 12 months after the outpatient experience or 15 months after the initiation of treatment.

METHOD

Project MATCH aftercare (N = 774) and outpatient (N = 952) samples were used in this retrospective study. Briefly, Project MATCH was a multi-site clinical trial investigating client-treatment matching, and findings have been reported elsewhere (e.g., PMRG, 1997; 1998). Following recruitment into the study, clients were randomly assigned to one of three psychosocial treatments: Cognitive Behavioral Therapy (CBT; Kadden et al., 1992), Motivational Enhancement Therapy (MET; Miller et al., 1992) or Twelve Step Facilitation (TSF; Nowinski et al., 1992). Therapy lasted twelve weeks, and therapists were nested within therapy conditions. Follow-up assessments were conducted in three month intervals from randomization which corresponded to an end of treatment assessment and follow-up interviews 3, 6, 9, and 12 months after treatment.

A noteworthy contribution of the Project MATCH research group was the aggressive attention paid to the training of therapists and research assistants, documentation of treatment fidelity, high follow-up rate (exceeding 90% at all follow-ups), and use of an exhaustive number of instruments, nearly all of which had published psychometric data. Full descriptions of these aspects of Project MATCH have been provided elsewhere (Connors et al., 1994; Zweben et al., 1998). In addition, a test-retest exercise was conducted (N = 82) to evaluate the reliability of instruments that were developed specifically for Project MATCH, instruments which form the core assessments for this study (Tonigan et al., 1997)

Quick Screen Interview. This interview was conducted to make a final determination about eligibility and to collect basic demographic information. Information critical for this study collectedmeasure included participant gender and marital status.

The Form 90 family of instrumentswas developed for Project MATCH as the central measure of client drinking (Miller, 1996). The Form 90 is a semi-structured interview that combines grid (Miller & Marlatt, 1984) and calendar-based approaches to reconstruct day-by-day drinking and health-related activities over a 90-day period. Measures of percent days abstinent (by month) and drinks per drinking day (by month) have good reliability (Tonigan et al., 1997), as do measures of the frequency of health-related experiences, e.g., emergency room visits for medical care.

Reasons for Relapse Questionnaire. This was a self-report survey that included a total of 45 items that asked about two domains, reasons for relapse and methods for staying sober. This study examined response to section 1, which was divided into situational influences (6 items), personal influences (8 items), and (11 items) general influences. Of these, we focused on situational and personal influences (14 items) because they corresponded most closely to Marlatt's RP model.

(Reasons For Relapse and Methods For Staying Soberquestionnaire about here)

Statistical Analyses.In addition to descriptive statistics, e.g., means and standard deviation (SD), used to describe the sample we used SPSS version 18.0 to conduct General Linear Model (GLM) tests. The dependent measures in these analyses included the 14 items in the Reasons for Relapse Questionnaire, all which straddled the fence between ordinal and interval scaled data (1 through 5, Likert scale with anchors of small to great influence). Our two independent variables were marital status (married versus not married), and gender. In the analysis, these factors were considered fixed. All GLM models included the marital, gender, and marital times gender interaction. All inferential tests were evaluated at p< .05 (1 in 20 chances of being wrong). However, we planned to conduct 14 tests which inflated Type I error.

RESULTS

Displayed in Table 1 are the samples’ characteristics. The total sample of this study was 1,726 participants; 1,305 males and 421 females. Amongst this sample the average age amongst both men and women was 40 and nearly the same level of education of 13 years. The alcohol severity was recorded for males at 52.81 and 48.12 for women which is extremely severe. Anyone over 20 is considered alcohol dependent. This study had many categories of the patient’s status (married, divorced, separated, widowed, etc.), but with sole concern on the patients marital status, males had a percentage of 34.8% being married and women were at a lower rate of 26.84%. A majority of the study was Caucasian, with a variety of Hispanic background. An estimated 22% of this study was unemployed amongst males and females.