3. How to measure relapse in humans
Editors: Spanagel, Rainer; Mann, Karl F.
Title: Drugs for Relapse Prevention of Alcoholism, 1st Edition
Copyright 2005 Springer
> Table of Contents > How to measure relapse in humans
How to measure relapse in humans
Henry R. Kranzler1
Howard Tennen2
Departments of Psychiatry1 and Community Medicine and Health Care2, University of Connecticut Health Center, 263 Farmington Avenue, Farmington CT 06030, USA
Introduction
Although the measurement of alcohol consumption and alcohol-related problems has received considerable attention in reviews of treatment research [1, 2, 3, 4 and 5], to our knowledge these reviews have not focused on issues specific to pharmacotherapy trials. Methods for the conduct of clinical trials testing medications to treat alcohol dependence have undergone substantial change over the past two decades [6, 7 and 8] and the issues specific to measurement of treatment outcome in these studies warrant consideration. Outcome measures should fit the practical requirements and hypothesized effects of the intervention being examined [9]. Because medications often have specific effects on drinking behavior, pharmacotherapy trials may impose unique requirements on the process of outcome measurement. Furthermore, to the extent that different medications have different hypothesized mechanisms of action and practical constraints, it is unlikely that uniformity of assessment can be obtained across studies [9]. In this chapter, we review developments in the methods used to collect data for the evaluation of outcomes and the specific measures used to evaluate the success of the medications.
Before focusing specifically on pharmacotherapy trials, however, we consider some of the issues that are of general relevance to the measurement of drinking behavior in treatment populations. This will provide the background for a selective review of pharmacotherapy studies that have been published over the past 30 years, a description of recent developments in this field, and speculation on what the future may hold for studies examining the effects of medications on alcohol treatment outcomes. The examples we selected to illustrate trends over time are drawn from randomized, controlled trials, since these represent the gold standard for the evaluation of treatment efficacy. However, the most rigorous methods have generally been used in more recent trials, and earlier studies often utilized methods no longer considered acceptable for use in efficacy trials. We focus this review on medications or groups of medications that have received the greatest research interest: lithium, the alcohol-sensitizing medications disulfiram and calcium carbimide, acamprosate and the opioid antagonists naltrexone and nalmefene.
P.24
Outcome measures in alcohol treatment trials
Finney et al. [3] examined 404 treatment outcome studies published between 1968 and 1998 that included two or more treatment/control conditions. In general, the number of treatment outcome studies published annually has increased in a linear fashion during this period. All of the studies that specified the method of data collection (96%) indicated that the data were obtained by self-report. The mean number of outcome variables assessed in these studies was 6.8 (median = 5). The correlation of outcome measures with year of publication gave no evidence of a significant increase in the number of outcome variables over time.
Table 1 shows the outcome measures that were used with greater frequency over the 31-year span, along with outcome indicators that have fallen out of favor. The measures that were more widely adopted over time include time-based abstinence (e.g., days of abstinence), time-based alcohol consumption (e.g., drinks per day), physiological markers [e.g., -glutamyl transpeptidase (GGT)], dependence symptoms, and drinking-related problems, all of which were more likely to be assessed in later years. In contrast, improvement in drinking behavior, measures of global functioning or improvement in global functioning, and measures of occupational functioning or participants' financial situation were less likely to be examined in more recent studies. Noteworthy for their lack of association with date of publication were categorical (i.e., complete) abstinence and psychological functioning.
Finney et al. [3] also examined the popularity of various outcome indicators in studies published between 1990 and 1998. This period is particularly relevant to the focus of this chapter. As a consequence of support from the US government (i.e., the National Institute on Alcohol Abuse and Alcoholism (NIAAA)) and the pharmaceutical industry in Europe, pharmacotherapy trials began to increase substantially in number and sophistication beginning in the 1990s. During this 9-year period, categorical abstinence and quantity-based
P.25
alcohol consumption were most commonly used as outcome measures (used in 54% of published trials), followed by psychological functioning (38%), time-based abstinence (34%), and time-based alcohol consumption (33%).
Table 1. Significant correlations between percentage of studies measuring class of outcome variable and year of publication (Adapted from [3]) | |||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| |||||||||||||||||||||||||||||||||
These findings led Finney and colleagues to the conclusion that the variety of outcome measures in alcohol treatment trials makes it difficult to compare results across studies. The large number of outcomes assessed also raises the question of how to correct for multiple comparisons [3]. This is a particularly thorny issue since the average power to detect a medium effect size in these studies at p < 0.05 was only 0.54, with only 23% of studies providing adequate power (i.e., 0.80) to detect such an effect [10].
Methods used to collect drinking data in alcohol treatment trials
A critical issue in the selection and use of outcome measures in pharmacotherapy and other treatment trials, is the feasibility of collecting complete data from each research participant at baseline (i.e., reflecting pre-treatment) and at various follow-up points during and after treatment [9].
Self-report measures
Alcohol treatment researchers have traditionally used self-report assessments to measure alcohol consumption and its consequences. Although these measures are generally considered valid, poorly designed questions, being under the influence of alcohol during the assessment, and fallible memory contribute to a lack of reliability and accuracy of these measures [11]. The risk of inaccurate information, which may be augmented by contingencies that favor misrepresentation, may be particularly pronounced during treatment or afterward, when patients may minimize the extent of their drinking to prevent embarrassment to themselves, to the researchers or to the clinicians treating them.
In response to the commonly held assumption that self-report data have questionable validity owing to patients' inclination to lie or forget, Babor et al. [12] used data from Project MATCH, a large, multi-center study of clienttreatment matching, to examine client characteristics associated with discrepancies between subject self-report and other sources of information, including biochemical markers and independent reports from collateral informants. Most participants in this trial appeared not to be lying, sociopathic, or otherwise behaving in an intentionally deceptive manner. Rather, demographic factors, alcohol severity, cognitive functioning, and social stability measures were more closely related to discrepancies among the data sources, than were measures of personality or psychiatric symptoms. Individuals who presented discrepant results had more severe drinking problems, more previous treatments, higher levels of pretreatment drinking, and more cognitive impairment, all of which could contribute to inaccurate recall. In sum, it appears that research
P.26
participants in clinical trials generally do not minimize their drinking on self-report measures and, when they do, it may be less a result of intentional deception on their part, and more a function of memory deficits associated with a more chronic and severe course of alcohol involvement.
A decade ago, Babor et al. [9] concluded that although biochemical measures and collateral informants' reports may increase an investigator's confidence in treatment outcome results, they add little beyond self-report data. Nonetheless, credibility is a major consideration in the reporting of treatment outcome data, and the added value of alternative measures may be worth their cost and inconvenience. An important consideration, when obtaining data from multiple sources, is how to best integrate the findings in meaningful ways. Moreover, resources devoted to collecting these alternative sources of outcome data might be better invested in procedures aimed at increasing the validity of self-report information, such as providing recall cues and emphasizing the importance of accurate information or, as we describe below, collecting data on a more frequent (e.g., daily) basis.
Methods for the collection of self-reported drinking data
Three approaches to the collection of self-reported drinking data are relevant to clinical trials [5]. Quantity-Frequency methods (QF), which are retrospective estimates of average or usual drinking, do not lend themselves to assessing variation in drinking intensity over time. Given some evidence that medications such as the opioid antagonists have their greatest effects on the intensity of drinking [13, 14], this is an important limitation to the use of QF methods for measurement of drinking outcomes in pharmacotherapy trials. Daily drinking estimation measures (DDE) are retrospective estimates of daily drinking, using methods such as the Timeline Followback method or Form 90. Over the past decade, DDE methods, particularly the Timeline Followback method (TLFB; [15]), have become the preferred method to measure drinking behavior in alcohol treatment trials. The Form 90, a DDE method similar to the TLFB, was developed for use in Project MATCH. Concurrent recall methods (CR) are reports provided in close temporal proximity to the drinking occurrence, using methods such as self-monitoring via paper diaries, palmtop computers, or interactive voice response (IVR) technology. CR methods have not been widely used in alcohol treatment research. However, prospective daily monitoring, a CR approach that has been examined in this context [16, 17 and 18], appears to offer a number of potential advantages over DDE methods, as described below.
Timeline followback method (TLFB)
The TLFB requires the interviewer to use a calendar of events identified by the respondent to establish anchor points such as holidays, anniversaries, major
P.27
national events, etc., which serve to facilitate the respondent's retrospective estimates of drinking on a day-to-day basis. The TLFB has been shown to be reliable and valid when administered as an in-person interview [19, 20], when administered by telephone or computer [21] and when used with Swedish-, Spanish-, or English-speaking alcohol abusers [22].
Recent studies of the TLFB have focused on comparisons of aggregate and more fine-grained comparisons with other data collection methods. Carney et al. [23] evaluated the overall correspondence and day-to-day agreement between daily reports and reports obtained using the TLFB [15]. Daily booklets were completed for 28 days by a group of problem drinkers and for 30 days using a palmtop computer by a group of moderate drinkers. Although aggregate levels of drinking were highly correlated across the two methods, the TLFB systematically underestimated drinking relative to daily assessment. There were also large differences among participants in day-to-day correspondence, suggesting that the TLFB may be less useful for measuring patterns (rather than levels) of alcohol consumption.
Prospective daily monitoring
Designs that employ prospective daily monitoring have a number of advantages over traditional cross-sectional designs: (a) rapidly changing processes can be measured closer to their real time occurrences; (b) retrospection biases are minimized; (c) powerful within-person analyses that eliminate sources of confounding can be applied to individuals' time series; and (d) temporal sequencing of variables can be established more confidently.
One approach to collecting daily data involves IVR technology, a relatively new and increasingly popular data collection method that uses the telephone to administer survey questions. In using IVR, the respondent answers each question by pressing the keys on the telephone keypad or provides a spoken response, and the responses are entered automatically in a database. IVR enjoys a number of advantages over traditional face-to-face interviews, including interview consistency, access to difficult-to-reach populations, including alcoholics [24], immediate data availability and accessibility, and convenience both for respondents and researchers. Searles and colleagues have demonstrated the feasibility, reliability and validity of daily IVR drinking reports in community and student samples [25, 26 and 27].
Searles et al. [27] compared daily reports using IVR technology for 366 days with TLFB assessments conducted in person at 13-week intervals in a group of drinkers, some of whom met criteria for an alcohol use disorder. Aggregate correlations for amount consumed, and frequency of drinking days and heavy drinking days were modest, with substantial variability among individuals. Compared to participants without an alcohol use disorder, those with such a diagnosis underreported their drinking.
P.28
We [17] recently examined the potential utility of IVR for measuring daily drinking behavior and related mood measures, as well as medication compliance in a pilot trial of naltrexone for problem drinkers. In this study, nine participants provided daily data via IVR during the 12-week treatment period. IVR was well received by study participants, who demonstrated a high degree of adherence to the call requirements. When compared across subjects, we found convergence on both the number of drinking days and mean daily consumption between daily IVR and TLFB, which is in agreement with findings reported by Bardone et al. [26] However, when compared using within-person observations and multi-level modeling (analytic methods of the kind highlighted by Stout [4] for maximizing the statistical power of treatment trials), there was evidence that subjects under-reported alcohol use on the TLFB. On average, participants reported drinking a mean of 0.58 more drinks/day, when assessed using daily IVR compared to the biweekly TLFB, and 0.78 drinks more drinks/day compared to the 12-week TLFB. These differences are consistent with the findings of Searles et al. [27, 28], who found that traditional QF indicators of alcohol use yielded lower levels of alcohol use compared to daily IVR.
In concluding that individuals under-reported their alcohol use on the TLFB, we assume that close to real time drinking reports (i.e., daily IVR) are more accurate than recalled drinking and that recall methods introduce memory decay and memory bias [29]. The extreme variation in the correlation for individual participants between daily IVR and TLFB also indicates that for some individuals, TLFB-derived drinking reports bear little resemblance to drinking reports captured close to their real-time occurrence through daily IVR.
In this study, there was substantial variation in the number of daily IVR reports provided, ranging from 12 to 84. The data missing from some individuals are potentially problematic, but no more so than in studies in which the TLFB is used retrospectively to measure drinking behavior. A common problem in alcohol treatment studies is the participants' failure to return to the treatment site and to comply with efforts to collect TLFB information via the telephone. An advantage of daily data collection via IVR is that multi-level modeling, the use of which is enhanced by daily reports, is robust with regard to missing data and that it weights each participant's data based on the amount of data supplied by that individual.
A potential limitation of this study is that some individuals may have been intoxicated at the time they provided daily reports. Using blood alcohol concentration and collateral reports, Perrine et al. [30] demonstrated the reliability of daily IVR reports. Furthermore, we found no evidence of outlying reporting days during which it took more time than usual to complete the daily protocol, which would suggest intoxication in the respondent.
Another potential problem with intensive self-monitoring using daily data collection is that it could influence the behaviors being measured [31] by increasing awareness of the temporal contingencies between behavior and internal or environmental triggers, by initiating self-focused attention, or by interrupting distraction and related coping efforts [29]. Efforts to minimize
P.29
measurement reactivity include recording more than one behavior [32] and limiting recording to once a day [33]. In a series of daily process studies, including studies of heavy drinkers, we [18, 23, 34] found no significant evidence of trending or temporal changes in within-person associations that might signal measurement reactivity. Similarly, Hufford et al. [35] reported that electronic monitoring via palmtop computers appeared to show no noticeable reactivity effects.
Daily reports also provide an opportunity to study the moderating effect of treatment on relations between drinking and subjective states, such as mood [17, 36, 37]. Daily measurement of mood or desire to drink may help to elucidate the mechanism of treatment effects on drinking behavior, including the effects of pharmacotherapies [33, 37].
Biological measures and collateral informant reports of drinking
Concerns that alcoholics may under-report or deny their drinking have sparked efforts to identify sensitive and specific biological indicators of recent alcohol use that do not depend on self-report. Available biochemical tests have the advantage of being specific and reliable, and cannot be biased by motivational or cognitive factors. In addition, biological markers of direct or indirect alcohol effects provide an indication of alcohol's physical toxicity and as such can indicate improvement or deterioration during treatment.
Irwin et al. [38] reported that parallel increases in serum GGT concentrations of 20% or greater, in serum aspartate aminotransferase (ASAT) of 40% or greater, and in serum alanine aminotransferase (ALAT) of 20% or greater differentiate recovering alcoholics who remain abstinent from those who resume drinking with a sensitivity of 100% and a specificity of 82%. In the VA disulfiram study [39], blood was obtained every two months for ASAT measurements to monitor for hepatotoxicity. When ASAT increased from baseline, other evidence indicated that the patient had resumed drinking and toxicity was not the problem.
Conigrave et al. [40] recently reviewed the literature on conventional biological measures (including ASAT, ALAT, and GGT), which despite their limited sensitivity and specificity are clinically widely available. A number of pharmacotherapy studies (e.g., [41, 42]) have used these measures to complement or validate self-report information obtained from the patient. Unfortunately, GGT and other liver function tests are subject to large individual differences (e.g., according to age and gender), and are affected by many factors other than alcohol (e.g., liver infections and medication use). In addition, many heavy drinkers do not have abnormal GGT levels, and the variability among individuals with abnormal levels is so great that the use of parametric statistical tests may not be appropriate.
CDT may be marginally better than GGT [43], though it has very poor sensitivity and specificity, when used to screen for heavy drinking in community
P.30
samples or general medical practice, and it may perform less well for women than for men [44]. Other limitations of these tests include cost, the feasibility of obtaining blood samples at the time of follow-up evaluations, and a relatively short period of maximum sensitivity (i.e., 2-14 days).
Recent approaches involve combining tests, such as CDT with GGT, to enhance their value in identifying relapse [45, 46]. Javors and Johnson [47] reviewed the findings on newer biological markers (including CDT, total serum sialic acid, sialic acid index of apolipoprotein J, and serum -hex-osaminidase). These authors concluded that of the newer tests, CDT is the only one with demonstrated value as a marker for alcohol consumption. They call for additional research on the newer markers and on combining these with more established markers to optimize the detection of heavy drinking.
The limitations of both self-report measures of drinking and biological indicators have led investigators to corroborate self-report data with information collected from collateral informants, who are knowledgeable about the patient's daily activities. Typically, parents, spouses, or friends are interviewed about the patient's drinking and general functioning. Collateral reports, when available, generally corroborate client self-reports [11, 12]. However, because of limitations of collateral reports, there has been the recognition that the high degree of effort required to obtain this information may not be justified by its limited value for validating self-reported drinking behavior by individuals participating in the treatment trial.
Collateral informants are difficult to recruit, sometimes lose contact with the client, and often lack detailed information about the quantity or frequency of drinking. And while it is generally assumed that collaterals honestly report what they observe, there is some anecdotal evidence that at times collaterals may collude with the client to deny drinking when it occurs [48] or possibly punish clients by over reporting (whether intentionally or not). Connors and Maisto [49], after reviewing the literature on the reliability and accuracy of alcoholics' self-reported alcohol consumption in relation to reports provided by collateral informants, concluded that individuals provide accurate reports about their drinking and alcohol-related consequences. When reports provided by collaterals are discrepant with participants' reports, the latter almost always show higher levels of alcohol consumption. Agreement between participant and collateral reports is most likely to occur when collaterals are in frequent contact with the participant, are spouses or partners, and are confident about the reports they provide [12].
Alcohol pharmacotherapy trials
Lithium
Patients who were enrolled in early trials of medications to treat alcoholism were treatment-seeking patients, who generally were moderate-to-severe in
P.31
their level of alcohol dependence. Early studies of the efficacy of lithium for the treatment of alcohol dependence predominantly included inpatient alcoholics. The primary outcome measure reported in these studies was complete abstinence from alcohol, though over more than a decade of efforts to examine the efficacy of this medication, a number of methodological advances appeared. Kline et al. [50] compared lithium with placebo in a sample of inpatient alcoholic veterans. The efficacy outcome employed by these investigators was the occurrence of disabling alcoholic episodes, which occurred when patients' drinking interfered with their daily life, thereby necessitating inpatient detoxification. Merry et al. [51] randomly assigned inpatient alcoholics, stratified on the basis of co-morbid depressive symptoms, to receive either sustained-release lithium or placebo. Following discharge, patients were monitored at 6-week intervals for 1 year. The self-reported outcomes on which the treatment groups were compared were the number of days on which drinking occurred and the number of days on which patients were incapacitated by alcohol, the proportion of patients who were totally abstinent, and the change in depressive symptoms. Pond et al. [52] recruited alcoholics through newspaper advertisements, in an effort to enroll individuals who were not undergoing treatment for their alcoholism. They randomized participants to receive lithium or placebo using a crossover design. Self-reported drinking behavior was used to evaluate the efficacy of treatment, with grams of alcohol consumed per day as the primary outcome measure, which was obtained through weekly interviews. Although information was obtained on the number of tablets consumed, the occurrence of alcohol-related events (e.g., hospitalization) and non-alcohol-related hospitalization or illness, as well as changes in economic, social and psychological status, findings related to these measures were not reported.
In the late 1980s, studies evaluating lithium for alcoholism treatment showed clear methodological improvements, including the use of multiple measures of treatment outcome. Fawcett et al. [53] compared lithium to placebo in a sample of individuals recruited from an inpatient setting. Drinking outcomes were obtained through interviews conducted at least monthly by a nurse with alcoholism treatment experience. For most patients, a significant other provided information at 6-month intervals, which together with the self-report information and medical records was used to make a clinical judgment as to whether the patient had been completely abstinent, whether the patient had received other treatment and whether he or she had been working regularly. In addition to a dichotomous classification as abstinent or not, the time to relapse to any drinking was used in this study as an outcome of treatment. Other outcomes that were examined were abstinence from all abused drugs, severity of depressive symptoms, number of days missed from work, and number of hospitalizations.
Dorus et al. [54] conducted the largest and most methodologically sophisticated study of lithium for alcoholism treatment. The study was a placebo-controlled, multi-center VA Cooperative Study, which enrolled 457 alcoholics
P.32
recruited as inpatients and stratified on the basis of a lifetime history of depression. The outcomes included total abstinence, number of days of drinking, number of alcohol-related hospitalizations, change in the rating of alcoholism severity and change in the severity of depression. These measures were obtained at 4-week intervals by interviewing the patient and a collateral informant. Based on the study's failure to demonstrate an advantage for lithium over placebo in either the depressed or non-depressed subgroups, it is now generally conceded that except for individuals with co-morbid bipolar disorder, lithium has no role in the treatment of alcoholism [55].
Alcohol-sensitizing medications
Early studies of disulfiram were, in many respects, similar to the later studies of lithium. Fuller and Roth [56] used total abstinence and the number of drinking days as the primary outcomes in the first, large-scale, controlled trial of disulfiram for alcoholism treatment. However, these investigators also compared treatment groups on the number of days of work attended by study participants, their family stability, and adherence to the schedule of study visits as secondary measures of efficacy. A similar approach was taken with the multi-center VA Cooperative Study of disulfiram [57], which, in addition to total abstinence and the number of drinking days as primary treatment outcome measures, included a measure of the time to first drinking episode. In both of these studies, self-reported drinking behavior was supplemented by collateral informant report. The disulfiram studies were, however, methodologically superior to the studies of lithium in that they used a riboflavin marker, which made it possible to measure medication compliance.
In a placebo-controlled study of the efficacy of the alcohol-sensitizing drug calcium carbimide, Peachey et al. [58] employed the following self-report measures of drinking behavior: drinking days, total quantity, and typical daily quantity. This study was unique, insofar as it required participants to return a urine sample daily, which made it possible to measure the concentrations of both alcohol and riboflavin (with which both the active and placebo medications were formulated), thereby providing objective measures of drinking behavior and medication compliance. The validity of daily reports of drinking behavior and medication compliance were thus evaluated against these criterion measures.
Acamprosate
A recent comprehensive meta-analysis of clinical trials of acamprosate [59] identified a total of 20 randomized, placebo-controlled studies, all but one of which are published. The studies were conducted in 12 European countries, the US, and Korea. Seventeen studies met the criteria for inclusion in the meta-analysis,
P.33
which represent a total of 4087 patients (2160 who were treated with acamprosate and 1927 who received placebo). These studies enrolled patients with a diagnosis of alcohol abuse or dependence, so that there was variation in the severity of alcoholism across the studies, though there was an abstinenceoriented treatment orientation in all. The most common primary outcome measure was cumulative abstinent days (n = 9), followed by drinking behavior (n = 3), continuous abstinence (n = 2), and time to first relapse (n = 2, both of which had cumulative abstinent days as a co-primary outcome measure). Relapse rate at each visit, serum GGT activity, and time to treatment failure were each used as a primary outcome in only one of the studies reviewed. Although continuous abstinence rates were not the primary outcome measure in the studies reviewed, Mann et al. [59] were provided access by the pharmaceutical sponsor to data for these studies, and they were able to examine the overall effect of acamprosate on this outcome measure. These authors found a significant advantage for acamprosate over placebo on continuous abstinence rates. The effect sizes, though modest, increased progressively as treatment duration increased from three to six and then to 12 months.
Chick et al. [60] conducted a meta-analysis that included data from 15 of the same studies as those included in the Mann et al. [59] meta-analysis. Although the method of collecting drinking data at the assessment points varied among the studies, data available at each assessment made it possible to examine the overall effect of acamprosate on average daily alcohol consumption, number of days drinking, and the proportion of patients drinking an average of five or more drinks per day. Chick and colleagues acknowledged, however, lack of precision in the consumption estimates for the studies that used categories for the mean reported number of drinks per drinking day and the mean number of drinking days per week.
Overall, in contrast to studies of either lithium or disulfiram, there is some consistency in the outcomes assessed in the acamprosate studies, presumably because a single pharmaceutical company sponsored all of these studies. Nonetheless, there were a number of primary outcomes examined, particularly in view of the wide diversity of the countries in which the studies were conducted and the diversity of requirements for registration in those countries.
Opioid antagonists
A recent meta-analysis of the effects of opioid antagonists on treatment of alcohol dependence [61] identified a total of 14 randomized, placebo-controlled studies of naltrexone and two studies of nalmefene. These studies varied in duration from 4 weeks to one year, with a total of 1688 subjects, 868 who received active medication (including 84 treated with nalmefene) and 820 who received placebo. As shown in Table 2, these studies employed a total of 10 different measures of treatment outcome, the most common of which (94%) was the rate of study discontinuation.
P.34
Table 2. Outcome measures reported in randomized, placebo-controlled studies of opioid antagonists (n = 16) | ||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
Both naltrexone [13, 62] and nalmefene [14, 63] have been shown to reduce the risk of heavy drinking. As a consequence of these findings, subsequent studies of the oral formulation of naltrexone in alcoholics and in problem drinkers [18] and of long-acting naltrexone formulations in alcohol-dependent individuals [64, 65] have used the frequency of heavy drinking as a primary outcome measure.
Outcome measures for alcohol treatment trials: future directions
A decade ago, in anticipation of analyzing data from Project MATCH, investigators from that study [9] addressed the measurement of outcomes in alcohol treatment research. Based on empirical findings from four data sets, the group recommended the use of measures of both frequency (days drinking) and intensity (drinks per drinking day). They found these measures to be relatively uncorrelated and hence argued that they represent independent dimensions of drinking behavior. Furthermore, they pointed out that these measures are easily and inexpensively assessed using self-report techniques and are differentially sensitive to different kinds of treatment.
Recently, in an effort to develop an optimal measure for use in alcohol treatment trials, NIAAA convened a panel of scientists with expertise in the assessment of drinking behavior, particularly as it relates to clinical studies [5]. The panel recommended that percent of days heavy drinking (defined as 4 drinks in a day for women and 6 drinks in a day for men) be used as the single outcome measure in alcohol treatment trials. The rationale for choosing this measure was that it is a good predictor of acute alcohol-related problems. Furthermore, use of abstinence as a categorical measure of treatment outcome is limited by the small number of patients who are able to achieve it over an extended period of time and by its inability to show improvement over time.
P.35
Abstinence (though traditionally viewed as the most appropriate goal of alcohol treatment [59]), is not a suitable treatment goal for harm reduction efforts or for individuals with levels of severity that do not require abstinence as a treatment goal. Harm reduction approaches, in which reduced drinking frequency and intensity, rather than abstinence from alcohol, are the goals of treatment, have assumed greater importance in both clinical practice and as the focus of alcohol treatment trials, particularly pharmacotherapy research.
The NIAAA panel also recommended use of the TLFB for collecting data on percent of days heavy drinking. The TLFB is the most psychometrically sound and most widely used method of data collection for alcohol treatment trials. Although it is acknowledged that CR methods may provide more accurate measures of drinking, they argue that under most circumstances CR methods cannot be used to estimate pretreatment drinking and thus are not the preferred methods for comparing pre- and post-treatment outcomes. However, this begs the question of whether a hybrid approach, in which subjects report pretreatment drinking using the TLFB or another DDE approach and prospectively report drinking using a CR approach, may incorporate the best aspects of both approaches. As discussed earlier, use of the TLFB or a similar DDE method also ignores the utility of the CR approach for gathering data on mood, desire to drink and other factors that vary day to day. An approach that incorporates these fine-grained data and employs modern analytic tools makes it possible to detect very specific effects of treatment through increased statistical power [4].
The search for a single, optimal treatment outcome measure may also prove to be futile. In underscoring the importance of statistical power in treatment outcome studies, Stout [4] argues that a more efficient alternative to increasing sample size is the use of improved measures of treatment outcome and analytic methods tailored to the predicted effects of treatment. Because these effects differ among studies, no single outcome measure is likely to be useful for all studies, a point that contrasts with the efforts of the NIAAA panel.
The choice of outcome measures in the Combine Study, a large multi-center study comparing naltrexone, acamprosate and their combination with placebo, with all medication conditions delivered with either medical management or a more intensive psychotherapy [66], may influence the design of pharmacotherapy research over the next decade. In this study, the primary outcome measures are percent days abstinent and the number of days to first heavy drinking episode, as measured using the Form 90. Secondary outcome measures include number of heavy drinking days, a composite outcome measure that integrates both alcohol consumption and alcohol-related problem variables, biological markers of heavy drinking (e.g., CDT), level of alcohol craving, and presence of DSM-IV alcohol dependence.
As the pharmaceutical industry has come increasingly to recognize the potentially lucrative market that medications to treat alcoholism could represent, there has been growing interest in the conduct of multi-center clinical trials.
P.36
These allow rapid recruitment of the large study samples required to provide adequate statistical power. However, the measurement of drinking outcomes in multi-center clinical trials presents a number of challenges in terms of feasibility, sensitivity, and conceptualization of the dependent variable. Traditionally, practical considerations have favored the use of self-report measures because of their lower cost and ease of use. Although alternative measures of drinking, such as biochemical tests and collateral informant reports, have been proposed, they have not gained wide acceptance as independent outcome measures. If collected at all in outcome studies, they are more likely to be used to provide circumstantial evidence for the accuracy of self-report information, rather than as a specific validity check on a given individual. Even under the optimal conditions of a large-scale clinical trial, it is more difficult to obtain blood samples and collateral informants' reports than client self-reports. All measures decline in completeness as the time from the baseline assessment increases.
Prospective daily monitoring of drinking behavior using a method such as IVR has the potential to serve as a centralized method for use in multi-center trials. We would speculate that the intersite variability in outcome measurement involving methods such as the TLFB or Form 90 represents a major source of error variance that can threaten the internal validity of multi-center trials. IVR-based administration of a common script for eliciting data on drinking behavior could reduce some of the site effects, thereby representing an important methodological advance in the conduct of multi-center trials. This approach may also provide insights into the mechanisms by which pharmacological treatment effects occur [37].
References
1 Breslin FC, Sobell SL, Sobell LC, Sobell MB (1997) Alcohol treatment outcome methodology: state of the art 1989-1993. Addict Behav 22: 145-155
2 Del Boca F, Darkes J (2003) The validity of self-reports of alcohol consumption: state of the science and challenges for research. Addiction 98 Suppl 2: 1-12
3 Finney J, Moyer A, Swearingen C (2003) Outcome variables and their assessment in alcohol treatment studies: 1968-1998. Alcohol Clin Exp Res 27: 1671-1679
4 Stout RL (2003) Methodological and statistical considerations in measuring alcohol treatment effects. Alcohol Clin Exp Res 27: 1686-1691
5 Sobell LC, Sobell MB, Connors GJ, Agrawal S (2003) Assessing drinking outcomes in alcohol treatment efficacy studies: selecting a yardstick of success. Alcohol Clin Exp Res 27: 1661-1666
6 Kranzler H, Mason B, Pettinati H, Anton R (1997) Methodological issues in pharmacotherapy trials with alcoholics. In: M Hertzman, D Feltner (eds): The Handbook of Psychopharmacology Trials. New York University Press, New York, 213-245
7 Kranzler H, McLellan A, Bohn M (1995) Pharmacotherapies for alcoholism: Theoretical and methodologic perspectives. In: H Kranzler (ed.): The Pharmacology of Alcohol Abuse. Springer-Verlag, New York, 1-10
8 Linnoila M, Fawcett J, Fuller R, Harford T, Martin P, Meyer R, Romach M, Sellers E (1994) Evaluation of pharmacologic treatments for alcohol abuse, dependence, and their complications. In: R Prien, D Robinson (eds): Clinical Evaluation of Psychotropic Drugs: Principles and Guidelines. Raven Press, New York, 625-649
P.37
9 Babor TF, Longabaugh R, Zweben A, Fuller RK, Stout RL, Anton RF, Randall CL (1994) Issues in the definition and measurement of drinking outcomes in alcoholism treatment research. J Stud Alcohol Suppl 12: 101-111
10 Moyer A, Finney JW, Swearingen CE (2002) Methodological characteristics and quality of alcohol treatment outcome studies, 1970-98: an expanded evaluation. Addiction 97: 253-263
11 Babor TF, Stephens RS, Marlatt GA (1987a) Verbal report methods in clinical research on alcoholism: response bias and its minimization. J Stud Alcohol 48: 410-424
12 Babor TF, Steinberg K, Anton R, Del Boca F (2000) Talk is cheap: measuring drinking outcomes in clinical trials. J Stud Alcohol 61: 55-63
13 O'Malley SS, Jaffe AJ, Chang G, Schottenfeld RS, Meyer RE, Rounsaville B (1992) Naltrexone and coping skills therapy for alcohol dependence. A controlled study. Arch Gen Psychiatry 49: 881-887
14 Mason BJ, Salvato FR, Williams LD, Ritvo EC, Cutler RB (1999) A double-blind, placebo-controlled study of oral nalmefene for alcohol dependence. Arch Gen Psychiatry 56: 719-724
15 Sobell L, Sobell M (1992) Timeline follow-back: a technique for assessing self-reported alcohol consumption. In: R Litten, J Allen (eds): Measuring alcohol consumption: Psychosocial and Biochemical Methods. Human Press, Clifton NJ, 41-42
16 Armeli S, Tennen H, Affleck G, Kranzler HR (2000) Does affect mediate the association between daily events and alcohol use? J Stud Alcohol 61: 862-871
17 Kranzler H, Abu-Hasaballah K, Tennen H, Feinn R, Young K (2004) Using daily interactive voice recording to measure drinking and related behaviors in a pharmacotherapy study of problem drinkers. Alcohol Clin Exp Res 28: 1060-1064
18 Kranzler H, Armeli S, Tennen H, Blomqvist O, Oncken C, Petry N, Feinn R (2003) Targeted naltrexone for early problem drinkers. J Clin Psychopharmacol 23: 294-304
19 Maisto SA, Sobell LC, Cooper AM, Sobell MB (1982) Comparison of two techniques to obtain retrospective reports of drinking behavior from alcohol abusers. Addict Behav 7: 33-38
20 Sobell MB, Sobell LC, Klajner F, Pavan D, Basian E (1986) The reliability of a timeline method for assessing normal drinker college students' recent drinking history: utility for alcohol research. Addict Behav 11: 149-161
21 Sobell LC, Brown J, Leo GI, Sobell MB (1996) The reliability of the Alcohol Timeline Followback when administered by telephone and by computer. Drug Alcohol Depend 42: 49-54
22 Sobell LC, Agrawal S, Annis H, Ayala-Velazquez H, Echeverria L, Leo GI, Rybakowski JK, Sandahl C, Saunders B, Thomas S et al. (2001) Cross-cultural evaluation of two drinking assessment instruments: alcohol timeline followback and inventory of drinking situations. Subst Use Misuse 36: 313-331
23 Carney MA, Tennen H, Affleck G, Del Boca FK, Kranzler HR (1998) Levels and patterns of alcohol consumption using timeline follow-back, daily diaries and real-time electronic interviews . J Stud Alcohol 59: 447-454
24 Alemi F, Stephens R, Parran T, Llorens S, Bhatt P, Ghadiri A, Eisenstein E (1994) Automated monitoring of outcomes: application to treatment of drug abuse. Med Decis Making 14: 180-187
25 Searles JS, Helzer JE, Rose GL, Badger GJ (2002) Concurrent and retrospective reports of alcohol consumption across 30, 90 and 366 days: interactive voice response compared with the timeline follow back. J Stud Alcohol 63: 352-362
26 Bardone AM, Krahn DD, Goodman BM, Searles JS (2000) Using interactive voice response technology and timeline follow-back methodology in studying binge eating and drinking behavior: different answers to different forms of the same question? Addict Behav 25: 1-11
27 Searles JS, Helzer JE, Walter DE (2000) Comparison of drinking patterns measured by daily reports and timeline follow back. Psychol Addict Behav 14: 277-286
28 Searles JS, Perrine MW, Mundt JC, Helzer JE (1995) Self-report of drinking using touch-tone telephone: extending the limits of reliable daily contact. J Stud Alcohol 56: 375-382
29 Tennen H, Affleck G (1996) Daily processes in coping with chronic pain: Methods and analytic strategies. In: M Zeidner, N Endler (eds): Handbook of Coping. Wiley, New York, 151-180
30 Perrine MW, Mundt JC, Searles JS, Lester LS (1995) Validation of daily self-reported alcohol consumption using interactive voice response (IVR) technology. J Stud Alcohol 56: 487-490
31 Helzer JE, Badger GJ, Rose GL, Mongeon JA, Searles JS (2002) Decline in alcohol consumption during two years of daily reporting. J Stud Alcohol 63: 551-558
32 Hayes S, Cavior N (1980) Multiple tracking and the reactivity of self-monitoring: II. Positive behaviors. Behav Assess 2: 283-296
P.38
33 Tennen H, Affleck G, Armeli S, Carney MA (2000) A daily process approach to coping. Linking theory, research, and practice. Am Psychol 55: 626-636
34 Armeli S, Carney MA, Tennen H, Affleck G, O'Neil TP (2000) Stress and alcohol use: a daily process examination of the stressor-vulnerability model. J Pers Soc Psychol 78: 979-994
35 Hufford MR, Shields AL, Shiffman S, Paty J, Balabanis M (2002) Reactivity to ecological momentary assessment: an example using undergraduate problem drinkers. Psychol Addict Behav 16: 205-211
36 Collins RL, Morsheimer ET, Shiffman S, Paty JA, Gnys M, Papandonatos GD (1998) Ecological momentary assessment in a behavioral drinking moderation training program. Exp Clin Psychopharmacology 6: 306-315
37 Kranzler HR, Armeli S, Feinn R, Tennen H (2004) Targeted naltrexone treatment moderates the relations between mood and drinking behavior among problem drinkers. J Consult Clin Psychol 72: 317-327
38 Irwin M, Baird S, Smith TL, Schuckit M (1988) Use of laboratory tests to monitor heavy drinking by alcoholic men discharged from a treatment program. Am J Psychiatry 145: 595-599
39 Iber FL, Lee K, Lacoursiere R, Fuller R (1987) Liver toxicity encountered in the Veterans Administration trial of disulfiram in alcoholics. Alcohol Clin Exp Res 11: 301-304
40 Conigrave KM, Davies P, Haber P, Whitfield JB (2003) Traditional markers of excessive alcohol use. Addiction 98 Suppl 2: 31-43
41 Kranzler HR, Burleson JA, Del Boca FK, Babor TF, Korner P, Brown J, Bohn MJ (1994) Buspirone treatment of anxious alcoholics. A placebo-controlled trial. Arch Gen Psychiatry 51: 720-731
42 Volpicelli JR, Rhines KC, Rhines JS, Volpicelli LA, Alterman AI, O'Brien CP (1997) Naltrexone and alcohol dependence. Role of subject compliance. Arch Gen Psychiatry 54: 737-742
43 Anton RF, Moak DH (1994) Carbohydrate-deficient transferrin and gamma-glutamyltransferase as markers of heavy alcohol consumption: gender differences. Alcohol Clin Exp Res 18: 747-754
44 Allen JP, Litten RZ, Anton RF, Cross GM (1994) Carbohydrate-deficient transferrin as a measure of immoderate drinking: remaining issues. Alcohol Clin Exp Res 18: 799-812
45 Sillanaukee P, Olsson U (2001) Improved diagnostic classification of alcohol abusers by combining carbohydrate-deficient transferrin and gamma-glutamyltransferase. Clin Chem 47: 681-685
46 Anton RF, Lieber C, Tabakoff B (2002) Carbohydrate-deficient transferrin and gamma-glutamyltransferase for the detection and monitoring of alcohol use: results from a multisite study. Alcohol Clin Exp Res 26: 1215-1222
47 Javors MA, Johnson BA (2003) Current status of carbohydrate deficient transferrin, total serum sialic acid, sialic acid index of apolipoprotein J and serum beta-hexosaminidase as markers for alcohol consumption. Addiction 98 Suppl 2: 45-50
48 Edwards G (1985) A later follow-up of a classic case series: D. L. Davies's 1962 report and its significance for the present. J Stud Alcohol 46: 181-190
49 Connors GJ, Maisto SA (2003) Drinking reports from collateral individuals. Addiction 98 Suppl 2: 21-29
50 Kline NS, Wren JC, Cooper TB, Varga E, Canal O (1974) Evaluation of lithium therapy in chronic and periodic alcoholism. Am J Med Sci 268: 15-22
51 Merry J, Reynolds CM, Bailey J, Coppen A (1976) Prophylactic treatment of alcoholism by lithium carbonate. A controlled study. Lancet 1: 481-482
52 Pond SM, Becker CE, Vandervoort R, Phillips M, Bowler RM, Peck CC (1981) An evaluation of the effects of lithium in the treatment of chronic alcoholism. I. Clinical results. Alcohol Clin Exp Res 5: 247-251
53 Fawcett J, Clark DC, Aagesen CA, Pisani VD, Tilkin JM, Sellers D, McGuire M, Gibbons RD (1987) A double-blind, placebo-controlled trial of lithium carbonate therapy for alcoholism. Arch Gen Psychiatry 44: 248-256
54 Dorus W, Ostrow DG, Anton R, Cushman P, Collins JF, Schaefer M, Charles HL, Desai P, Hayashida M, Malkerneker U et al. (1989) Lithium treatment of depressed and nondepressed alcoholics. JAMA 262: 1646-1652
55 Calabrese JR, Shelton MD, Bowden CL, Rapport DJ, Suppes T, Shirley ER, Kimmel SE, Caban SJ (2001) Bipolar rapid cycling: focus on depression as its hallmark. J Clin Psychiatry 62 Suppl 14: 34-41
56 Fuller RK, Roth HP (1979) Disulfiram for the treatment of alcoholism. An evaluation in 128 men. Ann Intern Med 90: 901-904
P.39
57 Fuller RK, Branchey L, Brightwell DR, Derman RM, Emrick CD, Iber FL, James KE, Lacoursiere RB, Lee KK, Lowenstam I et al. (1986) Disulfiram treatment of alcoholism. A Veterans Administration cooperative study. JAMA 256: 1449-1455
58 Peachey JE, Annis HM, Bornstein ER, Sykora K, Maglana SM, Shamai S (1989) Calcium carbimide in alcoholism treatment. Part 1: A placebo-controlled, double-blind clinical trial of short-term efficacy. Br J Addict 84: 877-887
59 Mann K, Lehert P, Morgan MY (2004) The efficacy of acamprosate in the maintenance of abstinence in alcohol-dependent individuals: results of a meta-analysis. Alcohol Clin Exp Res 28: 51-63
60 Chick J, Lehert P, Landron F (2003) Does acamprosate improve reduction of drinking as well as aiding abstinence? J Psychopharmacol 17: 397-402
61 Srisurapanont M, Jarusuraisin N (2002) Opioid antagonists for alcohol dependence. Cochrane Database Syst Rev 2: CD001867
62 Volpicelli JR, Alterman AI, Hayashida M, O'Brien CP (1992) Naltrexone in the treatment of alcohol dependence. Arch Gen Psychiatry 49: 876-880
63 Mason BJ, Ritvo EC, Morgan RO, Salvato FR, Goldberg G, Welch B, Mantero-Atienza E (1994) A double-blind, placebo-controlled pilot study to evaluate the efficacy and safety of oral nalmefene HC1 for alcohol dependence. Alcohol Clin Exp Res 18: 1162-1167
64 Kranzler HR, Modesto-Lowe V, Nuwayser ES (1998) Sustained-release naltrexone for alcoholism treatment: a preliminary study. Alcohol Clin Exp Res 22: 1074-1079
65 Kranzler H, Wesson D, Billot L (2004) For the DAS Naltrexone Depot Study Group. Naltrexone depot for treatment of alcohol dependence: A multi-center, randomized, placebo-controlled clinical trial. Alcohol Clin Exp Res 28: 1051-1059
66 CombineStudyResearchGroup (2003) Testing combined pharmacotherapies and behavioral interventions in alcohol dependence: rationale and methods. Alcohol Clin Exp Res 27: 1107-1122
EAN: 2147483647
Pages: 26