referred for treatment of alcohol abuse than for treatment of depression (Kushner et al. 1990). Thus, the practice of assessing comorbidity among recently withdrawn alcoholics risks generating inflated comorbidity rates.

Table 2 shows the relevance of specific psychiatric diagnoses and drug use disorders for comorbidity rates reported by Ross, Glaser, and Germanson (1988). Row 1 shows the psychiatric comorbidity rate for those with an alcohol disorder, with or without another drug use disorder. Eighty-two percent of those with any alcohol diagnosis had at least one comorbid psychiatric disorder, as did 78 percent of those with an alcohol disorder but no other drug use disorder (column 2). Ninety-five percent of patients diagnosed with alcohol and other drug use disorders had at least one comorbid psychiatric diagnosis (column 3), and 79 percent of these had an ASPD diagnosis. The smallest proportion of comorbidity, 67 percent, was noted for those with a nonalcohol drug use diagnosis (column 4).

These findings emphasize that either drug or alcohol disorder carries a risk for psychiatric comorbidity. It also appears that the risk for comorbidity is greater given both an alcohol and other drug disorder than for either disorder

alone. Although not shown in table 2, about 75 percent of those with an alcohol diagnosis had no other substance abuse diagnosis. The remaining 25 percent were diagnosed with both disorders. The finding that one-fourth of alcoholdisordered treatment patients can be diagnosed with another drug disorder is consistent with ECA findings, in which an alcohol diagnosis was complicated by a nonalcohol drug disorder in about 20 percent of cases.

Beginning in table 2, row 2, comorbidity rates for specific psychiatric diagnoses are listed in descending magnitude. At 51 percent, ASPD is most commonly diagnosed for those with an alcohol disorder. Phobias, psychosexual dysfunction, and major depression are common comorbid disorders, affecting from one-fourth to one-third of those with an alcohol disorder. A reanalysis of these data (Ross, Glaser, and Stiasny 1988) found that women did not show higher rates of psychiatric disorders than men but were more likely to suffer from anxiety, psychosexual disorders, and bulimia, whereas men where more likely to be diagnosed with ASPD.

To the extent that these associations affect the course or optimal treatment for either disorder in a given comorbid pair, such high degrees of

Any alcohol

diagnosis

Drug and no alcohol diagnosis (17)

overlap have important clinical implications. Statistically, more common psychiatric disorders would be expected to overlap with alcohol disorders more often by chance alone. Because schizophrenia is relatively rare (1.4 percent in the ECA community survey as reported by Regier et al. (1990), the finding that 8.9 percent received this diagnosis in the Ross, Glaser, and Germanson (1988) sample represents a nearly sevenfold increase in risk. In comparison, for a relatively common disorder such as phobia (12.6 percent in the ECA community survey as reported by Regier et al. [1990]), the greater prevalence rate (36.5 percent) reported by Ross, Glaser, and Germanson (1988) translates into a smaller increase in risk for this disorder (the ratio of odds in the community to odds for patients is 4). Therefore, the magnitude of a given comorbidity rate can be interpreted differently depending on whether one evaluates raw proportions or the change in risk for a disorder compared with a non-comorbid normative group.

Studies suggest that having either an alcohol or psychiatric disorder increases a person's risk of having the other diagnosis. Studies that do not control for this risk may generate inflated comorbidity estimates.

Models That Link Alcohol
Disorders With Other
Psychiatric Disorders

The epidemiologic evidence reviewed thus far
indicates that alcohol disorders co-occur with
non-substance abuse psychiatric disorders at a
rate exceeding that predicted by chance alone.
Although these findings suggest that the co-
occurrence between these conditions is not due
to random or coincidental factors, they do not
otherwise clarify the nature of this association.
This section provides an overview of the evi-
dence regarding the nature of the causal connec-
tion(s) between alcohol and psychiatric disorders.

Although it is beyond the scope of this chapter to discuss complex multifactorial models, it may be useful to consider several superordinate meta-models of comorbidity (figure 4) that can subsume more complex variants (Anthony 1991; Kushner et al. 1990; Meyer 1986a). For example, the secondary alcoholism model describes comorbid relationships in which a psychiatric disorder contributes to the development of an alcohol use disorder (e.g., attempts to self-medicate psychiatric symptoms result in an alcohol use

disorder). Conversely, the secondary psychiatric disorder model describes comorbid relationships in which an alcohol disorder contributes to the development of a psychiatric condition. This model can refer to the symptoms of psychiatric disorder resulting from the various toxic effects of pathological alcohol consumption, such as withdrawal, or it can refer to the onset of a persistent psychiatric syndrome as a consequence of an alcohol use disorder. The common factor model specifies that both psychiatric and alcohol use disorders are linked through a third, unspecified factor, such as genetic predisposition, that

contributes to the development of both comorbid conditions. The bidirectional model specifies that regardless of which condition emerges initially, each can exacerbate the other over time.

In considering these causal models, the picture is complicated by complex, dynamic interplay between alcohol and psychiatric symptoms.

Several causal models may account for the association between psychiatric disorders and alcohol use disorders; depression, schizophrenia, and ASPD each may be associated with alcohol problems through different mechanisms. The same psychiatric condition may be linked to alcohol problems through several causal pathways (alcoholism may lead to depression in some people and follow from depression in others). It also is possible that one model best describes the initiation of a comorbid disorder, while another describes long-term maintenance of the same comorbid association. Despite these ambiguities, efforts to understand such mechanisms may have far-reaching implications for the etiology, treatment, and prevention of alcohol and psychiatric disorders in comorbid individuals.

The secondary alcoholism model and the secondary psychiatric disorder model make the mutually exclusive predictions that a psychiatric disorder or an alcohol disorder, respectively, will appear earlier.

This section examines empirical and conceptual issues related to comorbidity mechanisms in terms of the temporal sequence of first onset of comorbid disorders, family data implications for comorbidity mechanisms, and the self-medication view of comorbid association.

Onset Order of Comorbid
Disorders

Determining the order of onset for comorbid disorders cannot demonstrate the operation of a specific causal mechanism, but different models predict different temporal patterning of onset. The secondary alcoholism model and the secondary psychiatric disorder model make the mutually exclusive predictions that a psychiatric disorder or an alcohol disorder, respectively, will appear earlier. If either the secondary alcoholism model or the secondary psychiatric disorder

model describes most comorbid associations, this fact should be reflected in studies that assess the sequence of disorder onset for those with psychiatric and alcohol diagnoses.

Although disorders vary, several studies suggest that psychiatric conditions tend to begin before alcohol disorders for a small majority of comorbid individuals. Powell et al. (1982) reported that for a large sample of comorbid male veterans in alcoholism treatment, 59 percent reported that psychiatric problems developed first, 22 percent reported coincident onset, and 19 percent said alcohol problems started first. On average, only obsessive-compulsive disorder and depression began after alcoholism. In a mixed in- and outpatient substance abuse treatment sample that included males and females, Ross, Glaser, and Germanson (1988) reported that ASPD nearly always predated alcohol abuse, but the percentage in which the psychiatric disorder predated the alcohol disorder was somewhat smaller: 60 percent for schizophrenia, 60 percent for phobias, 58 percent for panic disorder, 46 percent for obsessive-compulsive disorder, and 43 percent for major depression.

Hesselbrock et al. (1985) reported on age of onset for male and female patients in alcoholism treatment. Although gender effects were minimal for most disorders, they found that depression preceded alcoholism in 65 percent of females but 41 percent of males. This finding is replicated in the ECA community survey, in which alcoholism followed depression in 66 percent of female respondents but preceded depression in 78 percent of males. The association between alcohol problems and depression may result from different processes for males and females.

Relying exclusively on these data, neither the secondary alcoholism model nor the secondary psychiatric disorder model alone seems to describe the causal connection between comorbid disorders for most cases. Further, the implications of these data for comorbidity mechanisms are limited by the method used to date the onset of comorbid disorders. Because they use retrospective self-reporting, the studies described above do not control for error associated with mistaken memory and misrepresentations. Some patients may report that psychiatric disorders occurred before alcohol problems to rationalize the latter with the former. In addition, years of waxing and waning alcohol problems and psychiatric symptoms may leave many patients unable to report onset sequencing accurately. Determining

the presence of earlier, and perhaps more subtle, manifestations of a given problem may provide information relevant to understanding comorbidity mechanisms. This information might be overlooked in an exclusive focus on problems only after they reach threshold for a diagnosis.

Another point relating to the order of comorbid condition onset is made by Schuckit (1983, 1985, 1986), Schuckit and Monteiro (1988), and Schuckit et al. (1990), all of whom suggest that consequences of alcoholism can precipitate transitory psychiatric symptoms. Although a minority of all comorbid patients report a negative history for psychiatric disorders before their alcoholism, research shows that anxiety and depression symptoms often are temporarily elevated immediately after detoxification (Brown and Schuckit 1988; Brown et al. 1991; Schuckit and Monteiro 1988). In some cases, however, a psychiatric problem that begins after an alcohol problem may constitute an independent psychiatric syndrome.

Although identifying the step-by-step sequence by which alcohol and comorbid psychiatric conditions arise is an important step toward better understanding comorbidity mechanisms, studies fall short of this aim. Future studies can avoid the methodological shortcomings noted above by

⚫ following populations at high risk for alcohoi and psychiatric disorders prospectively rather than relying on retrospective patient reports; ♦ evaluating psychiatric diagnoses over periods of abstinence from alcohol; and

⚫ attending to behavioral patterns, drinking motivations, and subclinical symptoms that are precursors to comorbid syndromes, in addition to documenting the presence of diagnosable disorders.

The Family/Genetic Study as a
Method for Evaluating
Comorbidity Mechanisms

In figure 4, the common factor model suggests that alcohol and psychiatric disorders share a common etiology. This model, often evaluated with regard to familial or genetic factors that may increase the risk for alcoholism and psychiatric disorders, suggests that comorbid conditions are manifestations of the same pathologic genotype, learning history, or other dysfunction. Because many psychiatric disorders, including alcoholism, independently aggregate in families

(Rieder and Kaufmann 1988), it is possible to test competing hypotheses about the nature of comorbid associations by evaluating patterns of cross-generational transmission of these conditions (Biederman et al. 1991). For example, if a shared vulnerability to depression and alcoholism is transmitted genetically or environmentally among family members, an index patient with depression only should put family members at an increased risk for three distinct conditions: depression alone, depression plus alcoholism, and alcoholism alone. The case in which the index patient with depression only transmits an increased risk to family members for alcoholism only (cross-transmission of pure forms) offers

the common factor model suggests that alcohol and psychiatric disorders share a common etiology.

the clearest evidence for a common risk factor for both disorders that is transmitted in the family.

Merikangas and Gelernter (1990) reviewed family, twin, and adoption studies that evaluated alcoholism and depression. In the absence of strong, consistent evidence of cross-transmission of pure forms of these disorders, they concluded that a shared-etiology explanation seems unlikely. Using a twin-study method, Kendler (1985) reported that alcoholism and schizophrenia appear to demonstrate independent genetic risk factors. Loranger and Tulis (1985) reported that family members of patients with borderline personality disorder (BPD) were at greater risk for alcoholism than those of schizophrenic or bipolar-affective patients. When patients were further subdivided based on their own level of alcohol consumption, family risk differences for alcoholism disappeared. These studies presented no evidence for cross-transmission of pure forms and no support for the shared-etiology model. Twin studies suggest that ASPD and alcoholism are transmitted separately (Hesselbrock 1986). However, Bohman et al. (1987) suggested that ASPD in the family can increase family members' risk for alcoholism because antisocial behavior may include drinking and a vulnerability to alcohol-related problems (see Chapter 3, Genetic and Other Risk Factors for Alcoholism).