Treated and treatment-naive alcoholics come from different populations

1.1. The potential bias in studying clinical samples: Berkson's fallacy 

The first clues to the association between diseases and both their antecedents and their consequences often derive from the study of select samples of treated individuals, hospitalized patients, or autopsy cases. This is the case regarding our knowledge of the antecedents of alcoholism and of the effects of alcoholism on brain structure and function. Because not all alcoholics from the general population are equally likely to be in these study samples, bias may result when findings in these samples are presumed to apply to the population at large. This type of bias, known as Berkson's fallacy (after the person who first studied it in detail) (Berkson, 1946, Berkson, 1955), occurs whenever the association between the independent variable (e.g., the diagnosis of alcohol dependence) and the dependent variable (e.g., the antecedents of alcoholism, severity of alcohol use, or the consequences of alcohol dependence) differs between the population from which the sample derives (hospitalized alcoholics or alcoholics in treatment or shortly after treatment) and (alcoholics in) the general population.

Fleiss (1981) presents examples of this bias, as well as the mathematics underlying it. A classic example of Berkson's fallacy occurred when Pearl (1929) found a negative association between the presence of cancer and tuberculosis in autopsy cases. Tuberculosis was less frequent in autopsy cases with cancer than in autopsy cases without cancer. Pearl inferred (erroneously) that the same negative association should apply to live patients and proposed treating patients with terminal cancer with tuberculin to arrest their cancer. He failed to understand that extrapolating an association found in autopsy cases to live patients is a fundamental error, unless all patients who die are equally likely to undergo autopsy.

Roberts et al. (1978) were the first to publish a study whose results empirically demonstrated Berkson's fallacy. Their sample comprised 257 individuals (of a random sample of 2,784 individuals interviewed in the community) hospitalized during the prior 6 months. There was a very large positive association between the presence of respiratory disease and the presence of locomotor disease in the hospitalized individuals. However, Roberts et al. correctly ascertained that respiratory and locomotor diseases were essentially independent in the entire random sample. The spurious association between respiratory disease and locomotor disease arose because the hospitalization rate of people with both diseases (29%) was about three times the rate of people with only respiratory or locomotor disease or neither disease (7%–10%). As Fleiss (1981) succinctly stated, “unless something is known about differential hospitalization rates…, a good amount of skepticism should be applied to any generalization from associations found for hospitalized patients…to associations for people at large” (p. 13).

Parnas and Teasdale (1987) presented an example of Berkson's fallacy in schizophrenia research with direct applicability to alcoholism research. In an American–Danish prospective study (Parnas & Teasdale, 1987) of children of schizophrenic mothers, individuals psychiatrically hospitalized or untreated for schizophrenia spectrum disorders were compared on a number of characteristics. Hospitalized and untreated individuals were similar on a number of measures. However, hospitalized individuals exhibited higher levels of substance abuse, affective symptoms, and psychopathic tendencies. Parnas and Teasdale suggest that “the clinical population may not be representative of the diagnostic category in question owing to [a greater] co-existence of confounding symptomatology (Berkson's fallacy)” (p. 44).

Another instance of Berkson's fallacy in the study of alcohol dependence could be created by drawing convenience samples from treated samples. Co-existing disorders (e.g., depression or bipolar affective disorder, antisocial personality disorder, attention deficit hyperactivity disorder, posttraumatic stress disorder, and other substance abuse disorders) may be greater in the treated samples than in alcoholics in the general population. However, these co-existing disorders may not be severe enough to result in clinical diagnoses that would exclude subjects from “alcoholism” research samples. Alternatively, the bias due to Berkson's fallacy may result if the severity of alcoholism is greater in clinical versus general population samples. Once again (in either case), findings in clinical samples may not generalize to alcoholics in the general population.

1.2. How big is the potential bias in alcoholism research? 

The magnitude of the potential bias consequent to Berkson's fallacy depends on the proportion of alcoholics who are in the treated subpopulation. The most current study results available indicate that the number of alcoholics in treatment is a small proportion of alcoholics in the general population. In the 1992 National Longitudinal Alcohol Epidemiologic Survey (Grant, 1997), it was estimated that more than 27 million Americans exhibit alcohol abuse or alcohol dependence. At about the same time, Harwood et al. (1994) estimated that there were approximately 1.8 million Americans who were receiving treatment for alcohol problems in non-Federal hospital and community-based treatment settings. Grant (1997) estimates that only one in 10 individuals who need treatment for alcohol abuse problems has actually sought treatment. These estimates derive from different methods and sampling plans. However, even if one assumes that three times the 1.8 million individuals from the Harwood study received some form of treatment for alcoholism, the treatment population is still less than a quarter of the number of people who exhibit alcohol problems. Therefore, it makes sense that estimates drawn from clinical samples represent at most only one quarter of individuals who exhibit alcohol abuse or dependence. Studies, in which alcoholics in treatment are compared with alcoholics who have not sought treatment, are needed to determine whether clinical samples differ from treatment-naive samples in the antecedents, severity, and “consequences” (in the psychologic, social, legal, and biologic arenas) of alcoholism.

1.3. The bias resulting from Berkson's fallacy may differ between male and female alcoholics 

The bias inherent in studying only treated alcoholics may be different for men and women, and this difference in bias may underlie the different results reported for male versus female clinical samples. The finding in the literature, showing that women, in comparison with men, suffer more cerebral consequences from long-term alcohol dependence, is based on studies of clinical samples (Bergman, 1987, Jacobson, 1986). However, this finding may be spurious if clinical samples of treated alcoholic women differ from treatment-naive alcoholic women more than treated alcoholic men differ from samples of treatment-naive alcoholic men. This is entirely possible, because it has been claimed that women (for a variety of reasons) are less likely than men to receive treatment for alcohol problems (The National Center on Addiction and Substance Abuse at Columbia University, 1996).

1.4. Examination of Berkson's fallacy in alcoholism research 

In our laboratory, we have two ongoing studies in which alcohol-dependent samples are being compared with age-comparable light drinking/nondrinking samples. In one study, we are examining alcoholics with 6 or more months of abstinence. All those individuals (between the ages of 35 and 55 years) underwent treatment (we include Alcoholics Anonymous as one form of treatment). In the other study, we examined alcohol-dependent individuals (between the ages of 20 and 50 years) who had never been in treatment. None of the latter subjects identified themselves as alcoholic, although all met DSM–IV [Diagnostic and Statistical Manual of Mental Disorders (4th ed.); American Psychiatric Association (1994)] criteria for alcohol dependence.

For the current study, we compared alcohol-dependent subjects from the two ongoing studies in our laboratory as to the subjects' alcohol use history, examining both quantity and use trajectory. We tested the null hypothesis that the treatment-naive alcoholics and the long-term abstinent alcoholics have similar alcohol use trajectories over time (the treatment-naive sample being observed earlier in their alcohol use histories). In addition, if their trajectories differed, we tested the hypothesis that that difference is larger in female subjects than it is in male subjects.

2.1. Subjects 

As noted above, subjects reported in this article came from two different studies: one of abstinent alcoholics (age, 35–55 years) and the other of treatment-naive alcoholics (age, 20–50 years). In both studies, alcohol-dependent and control samples were recruited. For the analyses reported in this article, only the alcohol-dependent samples were used. The abstinent alcoholics needed to meet the lifetime criteria for alcohol dependence and be abstinent for at least 6 months. Treatment-naive alcoholics needed to meet lifetime criteria for alcohol dependence and to have never been in treatment. All participants for either study were informed of the study's procedures and signed a written consent form before their participation. There were a total of four sessions for each study, each lasting between 1 and 2.5 h, involving clinical, neuropsychologic, electrophysiologic, and neuroimaging assessments. All subjects who participated in any session were paid for their time and travel expenses. Subjects who completed all four aspects of either study received a completion bonus. An independent review committee (the Institutional Review Board, Independent Review Consulting, Corte Madera, CA) approved all procedures before study, and all procedures were carried out in compliance with the Helsinki declaration of 1975, as revised in 1983.

Exclusion criteria for both studies were the following: (1) history or presence of an axis I diagnosis on the Diagnostic Interview Schedule [(DIS); Washington University, St. Louis, MO; http://epi.wustl.edu/dis/dishome.htm]; (2) history of drug dependence other than caffeine or nicotine; (3) significant history of head trauma or cranial surgery; (4) history of diabetes, stroke, or hypertension that required medical intervention, or of other significant neurologic disease; (5) clinical or laboratory evidence of active hepatic disease; (6) clinical evidence of Wernicke–Korsakoff syndrome; or (7) current substance abuse other than alcohol (aside from caffeine and nicotine). Table 1 shows reasons for exclusion of subjects in both studies, as well as the total number of participants who either were excluded or met inclusion criteria and completed the study. Table 2 presents subject demographics.

Table 1. Subject exclusion/inclusion data

		Abstinent alcoholics (n)	Treatment-naive alcoholics (n)
	Reason for exclusion
	Drug use	26	123
	Did not meet alcohol use inclusion criteriaa	48	94
	Met medical exclusion criteria	33	9
	Met psychiatric exclusion criteria	32	7
	Otherb	21	30
	Total subjects
	Excluded	160	263
	Met all inclusion criteria and completed study	47	70

a For abstinent alcoholics, either did not meet requirements for length of sobriety or did not meet requirements for alcohol dependence or alcohol use. For treatment-naive alcoholics, were either formerly treated for alcohol abuse or did not meet requirements for alcohol dependence or alcohol use. b For example, did not want to commute, lost to follow-up, or refused continuation or participation.

Table 2. Subject demographics

		Abstinent alcoholics		Treatment-naive alcoholics
	Variables	Males (n=25)	Females (n=14)	Males (n=25)	Females (n=14)
	Age, mean (± S.D.)	45.6 (± 7.0)	47.0 (± 6.3)	32.2 (± 6.9)	30.3 (± 6.7)
	Years of education, mean (± S.D.)	15.5 (± 2.0)	15.0 (± 2.7)	16.3 (± 1.7)	15.9 (± 1.8)
	Ethnicitya	1 AA, 22 C, 1 H, 1 M	14 C	2 A, 18 C, 4 H, 1 M	11 C, 3 M

a A=Asian; AA=African American; C=Caucasian; H=Hispanic; M=Multiracial. S.D.=Standard deviation.

2.2. Assessment 

All subjects were assessed by using a computerized psychiatric DIS. Subjects were also interviewed to assess their drug and alcohol use by using the Lifetime Drinking History (LDH) method (Skinner and Sheu, 1982, Sobell and Sobell, 1990, Sobell et al., 1988). In addition, medical histories were reviewed, liver function tests were performed and evaluated, and Family Drinking Questionnaires were administered on the basis of the Family Tree Questionnaire by Mann et al. (1985) [see also Stoltenberg et al. (1998)].

2.3. Subject matching and computation of dependent variables 

The LDH assessment, which uses the timeline follow-back interview method (in which subjects break their drinking history into periods with consistent alcohol use) was used to gather retrospective alcohol use information (Sobell and Sobell, 1990, Sobell and Sobell, 1992, Sobell et al., 1988). We used data obtained from the LDH to match abstinent alcoholic subjects and treatment-naive alcoholic subjects on a one-to-one basis. Subjects were matched on sex and age at the onset of heavy drinking. Heavy drinking was defined as the age when a female subject first reached a monthly dose of 80 drinks per month and a male subject first reached 100 drinks per month. There were 47 abstinent alcoholic subjects, but only 39 of these could be matched to a treatment-naive subject. The remaining subjects first met the criteria for heavy drinking relatively late in life (in their mid-30s), and we did not have any same-sex treatment-naive subjects who matched them on that variable. The average difference in the age at which subjects met the criteria for heavy drinking was 1.85 months [standard deviation (S.D.)=15.4 months] across the 14 female and 25 male matched-subject pairs. Once the matches were completed, the alcohol use variables from the LDH of the abstinent alcoholic subject from each pair were computed as if that subject were the age of his or her matched treatment-naive subject. For example, if an abstinent alcoholic subject was 55 years of age, the matched treatment-naive subject was 30 years of age, and both subjects met the heavy drinking criteria at age 23 years, the alcohol use variables for the abstinent alcoholic subject were recomputed by using the drinking history up to the point when that subject reached the age of 30 years. Alcohol dose, duration of use, and duration of abstinence variables were computed from the LDH for two intervals: the time between the person's first drink and the date at which he or she met criteria for heavy drinking, and the period between that date and the current age of the treatment-naive person from each pair. This procedure is illustrated in Fig. 1.

View Large Image Download to PowerPoint

Fig. 1. Matching procedure. The top section of the figure presents the raw data for a matched pair of subjects [a 46-year-old male abstinent alcoholic (gray line) and a 34-year-old male treatment-naive alcoholic (orange line)], both of whom met criteria for heavy drinking at about 18 years of age. The middle section shows the two steps of the matching and alignment procedure. In the first step, the two subjects are aligned at their age of meeting the criteria for heavy drinking (this amounts to a shift of 4 months for the abstinent alcoholic subject), and that age is set to zero. In the second step, the data obtained for the abstinent alcoholic subject are truncated (red hash marks), so that they are followed for the same duration after criteria for heavy drinking were met as that for the treatment-naive subject. The bottom section shows the data after alignment and truncation.

2.4. Analysis 

The groups were compared on alcohol use variables by using a repeated-measures analysis of variance (ANOVA) within the Statistical Analysis System [(SAS), Release 8.02; SAS Institute, Inc., Cary, NC]. The trajectories of alcohol use were also examined visually. For that purpose, because subject pairs first met the criteria for heavy use at very different ages (range, 13 to 40 years), lifetime drinking histories were normalized with the age of criteria for heavy drinking being met set to zero. Each subject's use history was subsequently plotted as a function of time. Data obtained for subjects from the two groups were plotted by using different colors to aid visualization of differences in drinking trajectories between the two groups.

3.1. Interval from first drink to the beginning of heavy drinking 

In the interval from first drink to first heavy use, there were strong sex effects for all alcohol dose variables (average, peak, and average in most recent 6 months), with male subjects consistently having higher doses than were recorded for female subjects. All effects were large, with sex accounting for 14.9% of the variance of average dose, 22.1% of the variance of peak dose, and 16.9% of the variance of the dose in the most recent 6 months. The only group difference was on abstinence duration during this period, with alcoholics who eventually were treated having some periods of abstinence in this interval before they had even begun to drink heavily. Treatment-naive alcoholics had no abstinence periods during this interval. Of the 39 alcoholics who eventually were treated, five (four males and one female) had periods of abstinence in this interval.

3.2. Interval from beginning of heavy drinking to end of matched period 

In the interval from first heavy use to end of matched period, there were strong sex effects for average dose and peak dose, with sex accounting for 12.6% and 15.2% of the variance. There were also group effects that were larger than sex effects, accounting for 27.8% of the variance of average dose, 20.8% of the variance of peak dose, and 21.8% of the variance of the dose in the most recent 6 months. For all three variables, alcoholics who eventually were treated had much higher alcohol doses. There were no significant group by sex interaction effects. The effects for average alcohol dose for this interval are illustrated in Fig. 2. There were also significant group effects for duration of alcohol use and duration of abstinence for this interval. Alcoholics who eventually were treated had more abstinence during this interval. Because the entire interval was the same across groups, they had a smaller period of alcohol use during this interval as a consequence of their greater abstinence (Fig. 3).

View Large Image Download to PowerPoint

Fig. 2. Average dose for the period from the first heavy use to end of match period. In the matched period, the average alcohol dose was significantly lower in the treatment-naive subjects than in the treated subjects who eventually attained long-term abstinence (P<.01). In comparison with doses for male subjects, female subjects had lower doses (P<.05), but no group by sex interactions were observed. AbstAlc=Abstinent alcoholics; TxNaive=treatment-naive alcoholics.

View Large Image Download to PowerPoint

Fig. 3. Alcohol use duration for the period from the first heavy use to end of match period. In the matched period, the duration of alcohol use was lower for the abstinent alcoholics (AbstAlc) than for the treatment-naive alcoholics (TxNaive) (P<.01) owing to increased periods of abstinence in the treated subjects who eventually attained long-term abstinence. No sex or group by sex interactions were observed.

Fig. 4 presents the alcohol use trajectories of all research participants relative to the age at which the individual met the criteria for heavy drinking. The figure shows that the trajectories for treated and treatment-naive alcoholics overlap almost entirely in the period before subjects met the criteria for heavy drinking, but that the treated alcoholics have higher average doses in the period after criteria for heavy drinking were met.

View Large Image Download to PowerPoint

Fig. 4. Alcohol use trajectories. When the trajectories of an abstinent alcoholic and a treatment-naive alcoholic overlap, it is indicated in red. The trajectories for female subjects are lower than those for male subjects. (Note the difference in scale between the sections of the figure for female versus male subjects.) After criteria for heavy drinking were met, the trajectories for abstinent alcoholics (gray lines) are higher than the trajectories for treatment-naive alcoholics (orange lines).