Evaluation of heterogeneity in pharmacotherapy trials for drug dependence: a Bayesian approach.

AIMS: Difficulty identifying effective pharmacotherapies for cocaine dependence has led to suggestions that subgroup differences may account for some of the heterogeneity in treatment response. Well-attested methodological difficulties associated with these analyses recommend the use of Bayesian statistical reasoning for evaluation of salient interaction effects. METHODS: A secondary data analysis of a previously published, double-blind, randomized controlled trial examines the interaction of decision-making, as measured by the Iowa Gambling Task, and citalopram in increasing longest sustained abstinence from cocaine use. RESULTS: Bayesian analysis indicated that there was a 99% chance that improved decision-making enhances response to citalopram. Given the strong positive nature of this finding, a formal, quantitative Bayesian approach to evaluate the result from the perspective of a skeptic was applied. CONCLUSIONS: Bayesian statistical reasoning provides a formal means of weighing evidence for the presence of an interaction in scenarios where conventional, Frequentist analyses may be less informative. [Supplementary materials are available for this article. Go to the publisher's online edition of The American Journal of Drug and Alcohol Abuse for the following free supplemental resource: Appendix 1].

Effects of a stage-based alcohol preventive intervention for inner-city youth.

The purpose of this study was to examine first year outcomes of an alcohol preventive intervention within inner-city middle schools. Subjects consisted of 650 sixth grade students from one neighborhood inner-city school (n = 262) and one bused school (n = 388). At posttest, chi-square analyses showed that significantly fewer neighborhood intervention students initiated alcohol use, used alcohol during the past seven-day and thirty-day periods, drank heavily during the past thirty days, and drank over any period of time, compared to control students (p's < .05). Significant group x prior alcohol consequences interaction effects were found for bused students, showing those with past alcohol consequences who received the intervention had less intentions to use alcohol and less frequent use of alcohol (p's < .05). These results suggest that a brief, stage-based preventive intervention may result in significant reductions in alcohol initiation and consumption among some inner-city youth.

Using transtheoretical model profiles to differentiate levels of alcohol abstinence success.

Using data from 673 outpatient and 510 aftercare clients in Project MATCH, this study identified drinking-outcome-related differences in profiles across transtheoretical model (TTM) variables. Abstinent, moderate, and heavier drinking outcome groups were formed as a function of their 1-year posttreatment drinking behavior. Using profile analysis, group differences in the shapes of the mean profiles on TTM-related measures of stage and self-efficacy were found (a) for both the outpatient and aftercare populations and (b) at both the baseline and end-of-treatment time points. The use of cognitive and behavioral change processes during treatment was measured and the scores for each were included in the creation of the end of treatment profiles. The relation between number of TTM goals attained and probabilities of successful drinking outcome supports the usefulness of profile differences in predicting long-term drinking outcomes and can enable therapists to help clients set interim goals.

Postpartum return to smoking: staging a "suspended" behavior.

Two hundred fifty-six pregnant women who had quit smoking were classified into the precontemplation (PC), contemplation (C), preparation (PA), or action (A) stages of change for postpartum smoking cessation based on a 3-item algorithm assessing personal goals, self-efficacy, and smoking behavior. Logistic regression and event history analyses indicated group differences in return to smoking at 6 weeks and 3, 6, and 12 months postpartum. The percentage of women who returned to postpartum smoking was highest in the precontemplation stage and decreased with each subsequent stage, that is, at 6 weeks 83% of PCs, 64% of Cs, 35% of PAs, and 24% of As had returned to smoking. Results lend support for the stages of change for postpartum smoking abstinence. Stage-based interventions may be developed to assist women in maintaining abstinence postpartum.

Psychiatric severity and behavior change in alcoholism: the relation of the transtheoretical model variables to psychiatric distress in dually diagnosed patients.

Treatment programming for individuals diagnosed with a chronic mental illness and an alcohol use disorder could be enhanced by employing techniques that focus on those change process variables that are most strongly related to psychiatric distress. Prochaska and DiClemente's transtheoretical model (TTM) provides a useful framework within which to study these relations. The associations between psychiatric severity and the TTM constructs of stages and processes of change, decisional balance, temptation, and self-efficacy were measured among 132 alcohol-dependent patients in a public mental health clinic's outpatient dual diagnosis program. Participants' scores on the Temptation subscale of the Alcohol Abstinence Self-Efficacy Questionnaire are strongly related to psychiatric severity: The more psychiatric distress a person is experiencing, the more he or she is tempted to drink, particularly in situations that trigger negative affect. Decisional balance considerations are also related to psychiatric severity: The higher participants scored on the Global Severity Index of the Brief Symptom Inventory, the more importance they placed on the negative aspects, or cons, of drinking. Subjects with more psychiatric distress also scored higher on the maintenance stage of change subscale, possibly indicating an increased fear of relapse and struggle to maintain sobriety.

Causal relationships of processes of change and decisional balance: stage-specific models for smoking.

This study, a secondary analysis of prospective data of smokers, tested whether the causal relationships between the processes of change and decisional balance of the transtheoretical model of change (TTM) are stage-specific. It was expected that for smokers in the contemplation stage, higher levels of experiential processing cause the cons of smoking to become more important and the pros of smoking to become less important. In other words, the level of experiential process use was expected to causally influence decisional balance (pros minus cons) for people in the contemplation stage. For ex-smokers in the action stage, when the cons outweigh the pros (cons become more important while pros become less important), they should increase their behavioral process use: decisional balance was expected to causally influence use of behavioral processes. Cross-lagged panels were analyzed using structural equation modeling. Results indicate that experiential process use has causal predominance over decisional balance for smokers in the contemplation stage. For those in the action stage, however, neither decisional balance nor behavioral process had apparent causal predominance. Mean-level invariance indicates that the contemplation and action stages are different. Further analysis investigated smokers who progressed from contemplation to either preparation or action or from preparation to action. For these smokers who had progressed toward action, decisional balance did causally influence use of behavioral processes. This evidence provides support for the use of the TTM as the basis for planning interventions that target specific stage-dependent causal mechanisms.

Doing the right thing at the right time? The interaction of stages and processes of change in successful smoking cessation.

Research on the transtheoretical model has provided substantial support for both stages of change and processes of change (coping strategies influencing successful behavior change). This study examined whether timing of process use (using particular change processes during one stage and not during others) influences quit success. Hypothesized patterns of optimal process use were those marked by more use of experiential processes and less use of behavioral processes during contemplation and preparation and by the reverse pattern during action. Participants (N = 388) began in contemplation or preparation and took action during a 4-6 week period. Multivariate analyses of covariance examined the relationship between patterns of process use and success in staying quit both at 1-month follow-up and 5-6 months later. For the most part, results supported the hypothesis that successful stage transitions involve doing the right thing at the right time: engaging in experiential process activities during contemplation and preparation stages and shifting to behavioral process activities during action.

Pregnancy smoking cessation: a case of mistaken identity.

This study examined the process of change in pregnancy smoking cessation, using the stages and processes of change from the Transtheoretical Model, to compare women who stopped smoking during pregnancy with women who were in the process of smoking cessation, but were not pregnant. Differences in smoking cessation process activity and abstinence self-efficacy were hypothesized between the pregnant and nonpregnant groups of women. Study participants were 89 pregnant women who quit smoking, 28 nonpregnant women in the action stage of smoking cessation, and 92 nonpregnant women in the preparation stage. The Smoking Cessation Processes of Change Scale and the Smoking Abstinence Self-Efficacy Scale served as dependent measures. One-way MANOVA and follow-up Newman-Keuls comparisons indicated significant differences between pregnant and nonpregnant women in their levels of process activity and self-efficacy. Pregnancy smoking cessation differed dramatically from the process of nonpregnancy smoking cessation. Pregnant quitters were not engaging in experiential and behavioral processes at levels associated with the action stage of change. Low levels of process use and high efficacy indicated an externally (for the baby) motivated stopping rather than an internal, intentional process of change, which may account for high relapse rates postpartum.

Gender and smoking cessation: a factor structure comparison of processes of change.

This study examined gender differences in the pattern of process use for smoking cessation using the Processes of Change Questionnaire (J. O. Prochaska, W. F. Velicer, C. C. DiClemente, & J. Fava, 1988). The goals were (a) to determine the degree to which the covariance structure of the Processes of Change Questionnaire is invariant across gender, (b) to test the existence of the theoretical 2-factor process model using confirmatory factor analysis, and (c) to explore mean differences, if found, in the use of the 10 processes of change across 4 stages of change (precontemplation, contemplation, preparation, and action). The sample (N = 516) had an equal distribution of men and women across the stages of change. Results demonstrated that the structure of the measure for men and women was invariant at the level of the variance-covariance matrices and that the hypothesized 2-factor model fit the data. Only stage of change predicted the experiential and behavioral process factors.

Ensuring balanced distribution of prognostic factors in treatment outcome research.

In comparative or matching research involving two or more treatments, the equivalence of the patient groups is of critical importance. In the past, equivalence has either been imposed by matching or balancing, or has been assured statistically by randomization. Matching and balancing, while useful in many contexts, nonetheless have important limitations, as does simple randomization. In recent years, a new tool has been developed that represents a compromise between balancing and randomization. This method, urn randomization, gives clinical investigators new options for improving the credibility of studies at a relatively modest cost. Urn randomization is randomization that is systematically based in favor of balancing. It can be used with several covariates, both marginally and jointly, producing optimal multivariate equivalence of treatment groups for large sample sizes. It preserves randomization as the primary basis for assignment to treatment and is less susceptible to experimenter bias or manipulation of the allocation process by staff than is balancing. Disadvantages include the fact that it is more difficult to implement, and that it violates the simple probability model of simple randomization. A number of research studies on addictions, including client-treatment matching trials, have used urn randomization. A summary of the mechanics of urn randomization is presented, and guidelines for its use in treatment studies are discussed.

Alternative analytical methods for detecting matching effects in treatment outcomes.

Project MATCH presented a unique opportunity for a team of statisticians, data analysts and content experts to come together and explore the strengths and weaknesses of the application of various statistical models to the data of the type being collected in this large trial. The following models were evaluated: multilevel models, event history models, multiple were structural equation modeling, time series models, ordinal repeated measures designs and generalized estimating equations. No one model was found to be the perfect solution and each seemed to have something to recommend it. Future research on these methods will shed light on many issues raised. It is hoped that alcohol researchers will find useful guidelines within this chapter as they plan and carry out their studies.

Classical analytical methods for detecting matching effects on treatment outcome.

This article presents a classical approach for analyzing repeated measures designs with specific application to treatment matching studies. The generic treatment matching hypothesis is formulated under the multivariate general linear model, transforming the dependent variables to account for the repeated measures structure of the data. Issues of primary importance in the use of this approach (such as correcting for inflated Type I error and robustness of statistical tests to parametric assumptions) are discussed. The article concludes with an assessment of the strengths and weaknesses of this approach compared with alternative approaches.

The Alcohol Abstinence Self-Efficacy scale.

This study describes the development and the initial psychometric properties of a 20-item, self-report measure to assess Bandura's construct of self-efficacy applied to alcohol abstinence. Efficacy expectations are theorized to mediate behavior change and moderate effort and effective action. The sample was 174 male and 92 female subjects who came to an outpatient alcoholism treatment clinic. Abstinence efficacy was assessed with subject ratings on a 5-point Likert scale of confidence to abstain from alcohol across 20 different high-risk situations. A parallel set of items assessed subjects' temptation to drink in each situation. This Alcohol Abstinence Self-Efficacy scale (AASE) demonstrated a solid subscale structure and strong indices of reliability and validity. The four 5-item subscales measured types of relapse precipitants labeled negative affect, social positive, physical and other concerns, and withdrawal and urges. Both in structure and mean scores the AASE demonstrated no substantive gender differences. The AASE represents a brief, easily usable and psychometrically sound measure of an individual's self-efficacy to abstain from drinking.

Identifying pregnant women who drink alcoholic beverages.

This randomized 2 x 2 study compared disclosure rates of alcohol use with two response formats (multiple choice and dichotomous) and two communication channels (oral and written) in an adult prenatal population (N = 1078). The multiple choice question improved disclosure, regardless of channel, by 40% across white, African-American, and Hispanic subgroups.

Improving disclosure of smoking by pregnant women.

Smoking is a major modifiable risk factor in pregnancy, and low-cost interventions have been developed and tested in diverse populations of pregnant smokers. Successful intervention depends on identification, however, and nondisclosure can be a problem. This randomized study compared rates of disclosure with two response formats--multiple choice, in which the patient is able to describe herself as having "cut down," and the usual history question, "Do you smoke?," in which she is forced to answer simply "yes" or "no". Each format was tested in both oral and written channels with a multiethnic adult prenatal population (n = 1078) entering care in a multispecialty group. Study results indicate that the multiple choice question improved disclosure, regardless of channel (oral versus written), by 40%. This effect was observed across racial and ethnic groups. Biochemical tests of urine samples from reported nonsmokers indicated smoking in only 3%. Eleven percent of the "nonsmokers" in the experimental groups refused consent for the urine test, however, and many of these were probably smokers.

Assessing the congruence between physician behavior and expert opinion in smoking cessation counseling.

This study examines the degree of transfer of smoking cessation innovation from research to health care settings by comparing frequency-of-practice ratings by a national sample of family practice physicians (n = 903, response rate = 70%) and importance ratings by smoking cessation and prevention experts (n = 58, response rate = 84%) for 14 counseling techniques. The physician survey elicited a profile that combines traditional and behavioral techniques--discussing smoking with patients, encouraging goal setting, suggesting specific steps for quitting, and presenting pamphlets. They refer to others infrequently and rarely report planning for follow-up about smoking. The experts rated these selected techniques as moderately to highly important. They favored a behavioral approach coupled with active follow-up. The major differences between physician and expert rankings were that the experts placed higher priority on planned follow-up and a lower priority on pamphlets. The uneven quality of counseling reported by physicians suggests that weighting their responses according to expert opinion would provide a more sensitive profile. Scaled weighting produced scores that may help researchers define a composite quality-quantity measure of activity.

Pre-stimulus spectral EEG patterns and the visual evoked response.

The relationship between the latencies and amplitudes of the N1 and P2 components of the visual evoked potential (VEP) and the psychophysiological state of the brain immediately preceding the time of the stimulus has been investigated in 7 male subjects. Power spectral measures in the delta, theta, alpha and beta bands of the 1 sec pre-stimulus EEG were used to assess the brain state, and low intensity flashes, delivered randomly between 2 and 6 whole seconds, were used as the stimuli. Trials were ranked separately according to the relative amounts of pre-stimulus power in each EEG band and were partitioned into groups by an equal pre-stimulus spectral power criterion. Averaged EPs were computed from these groups and multiple regression analysis was used to relate pre-stimulus spectral power values to EP features. Five of the 7 subjects displayed consistent increases in N1-P2 amplitude as a function of increasing pre-stimulus relative alpha power. The between-subjects effect of pre-stimulus EEG on N1 latency was small, but was moderate for P2 latency (both significant). Both N1 and P2 latency were found to decrease with increasing amounts of pre-stimulus relative delta and theta power.

Health family trees: a tool for finding and helping young family members of coronary and cancer prone pedigrees in Texas and Utah.

We report on the feasibility and utility of a new approach for identifying the small percentage of families in the general population with strong familial predisposition to early coronary heart disease, strokes, and common familial cancers (breast, colon, lung), using the "Health Family Tree," a medical family history. A total of 24,332 "trees" were completed by parents and students in 37 high schools in 14 urban and rural communities in Texas and Utah during the years 1980-86. Completed "trees" were obtained from 68 per cent of all enrolled students. High-risk families, included 1,796 families with early coronary disease (7.5 per cent of all student families or 3.7 per cent of their parents' families), 870 stroke families (3.6 per cent), and 415 cancer prone families (1.7 per cent). Among these 3,081 high-risk families there were 8,245 family members already reported to have been diagnosed by a physician to have the familial disease of interest and 43,269 high risk unaffected siblings and offspring of these persons. The average cost per identified high-risk unaffected person was under $10. We conclude that the "Health Family Tree" is a feasible and cost-effective way to find high-risk families.

Application of cancer prevention knowledge: a longitudinal follow-up study of medical students.

Measuring the clinical application of knowledge and skills acquired in specific medical school courses is a critical part of curriculum evaluation. Second-year Cancer Prevention Course (CPC) students and a comparison group were surveyed by questionnaires, cognitively tested, and a subsample participated in in-depth interviews. The 6-month and 18-month follow-up interviews indicated that more CPC students practiced prevention in their clinical rotations than did the comparison group. Direct, one-on-one access to patients during clinical training was an important mediating factor in the practice of prevention skills. In addition, a behavioral intent inventory showed that CPC students, in their future careers, intend to perform certain specific cancer prevention activities more than others. Long-term retained knowledge (75% correct) stabilized at levels significantly higher (p less than .001) than baseline test scores prior to the Cancer Prevention Course. At the time of the two follow-up exams of cognitive knowledge, the comparison group of students was able to do no better than had the CPC students at baseline before the course. Positive beliefs about cancer prevention improved over the study period. More than twice as many cancer course students attended subsequent optional lectures and reported optional reading on cancer prevention topics than did the comparison group. Overall, the study showed a lasting effect of the elective course in cancer prevention, in terms of tested knowledge, and self-reported measures of attitudes, beliefs, practice of prevention in clinical rotations, and intent to apply prevention in future practice.