Running head: IN SICKNESS OR IN HEALTH
IN SICKNESS OR IN HEALTH:
GENDER AS A MODERATOR OF THE RELATIONSHIP
AMONG SUBSTANCE ABUSE IN THE FAMILY OF ORIGIN,
CODEPENDENT ATTITUDE AND BEHAVIOR,
AND ADULT-OFFSPRING MEDICAL PROBLEMS(1)
Daniel Harkness, Ph.D., A.C.S.W.
School of Social Work
Boise State University
1910 University Drive
Boise, Idaho 83725
Scott Curtis, M.S.W.
The Work and Learn School
750 Robbins Road
Boise, Idaho 83702
Keywords: Gender; codependency; substance abuse; medical problems.
Conventional wisdom in the treatment of addictions views substance abuse in the family of origin (SAFO) as a stressor that may lead to medical problems in codependent offspring. But stress models cannot explain the high rate of acute medical problems observed among SAFO offspring, suggesting the need for additional explanatory models. Building on preliminary findings that codependent attitude and behavior may attenuate the relationship between SAFO and acute offspring medical problems, we conducted a pilot study to explore the hypothesis that offspring gender moderates the relationship among substance abuse in the family of origin, codependent attitude and behavior, and acute medical problems, helping males more than females. Elements of our hypothesis were supported by the evidence, but our model oversimplified the complex interactions observed. Male codependent behavior moderated the relationship between SAFO and acute medical problems, as expected, but the codependent behavior of females mediated the relationship between SAFO and acute medical problems, consistent with a view that codependent behavior helps men but hurts women. We invite challenging replications of our small-sample quasi-experiment.
GENDER AS A MODERATOR OF THE RELATIONSHIP
AMONG SUBSTANCE ABUSE IN THE FAMILY OF ORIGIN,
CODEPENDENT ATTITUDE AND BEHAVIOR,
AND ADULT-OFFSPRING MEDICAL PROBLEMS
Many children are exposed to substance abuse in the family of origin (SAFO). Based on data from the 1988 National Health Interview Survey, one recent estimate asserted that 17.5 million children of alcoholics under the age of 18 lived in the United States (Eigen & Rowden 1995). Another estimate determined that 17.6% of California children lived with a parent who used illegal substances (Young 1997). And reflecting the adage that SAFO offspring grow up, a third estimate concluded that 22 million of the 28.6 million children of alcoholics who lived in the United States were age 18 and older (Russell 1990),
A body of evidence suggests that the risk of medical problems may be heightened for SAFO offspring (Woodside, Coughey & Cohen 1993). Anecdotally, addictions counselors are well aware of the problem (Cermak 1984; 1986) but, perhaps because the best empirical evidence remains buried in doctoral dissertations and technical reports with limited circulation, only six lines were devoted to the topic in a recent review of the scientific literature (Johnson and Leff 1999).
This is the second in a series of reports from an exploratory program of small-sample research designed to investigate Cermak's (1984; 1986) hypothesis that offspring codependency is an underlying mechanism linking SAFO with offspring medical problems. As our first report indicated (Harkness in press), some evidence is consistent with Cermak's hypothesis, and some
evidence is not. Extending that research, the present report has two goals: (1) to concentrate attention on the special risk of acute medical problems among SAFO offspring, and (2) to examine the interactive role of offspring codependency and gender in that linkage to address the criticism that the codependency construct is rooted in sexism.
The Mediating View: Cermak's Hypothesis
Cermak (1984: 40) contends that SAFO promotes offspring codependency, a "pattern of behavior and attitudes characteristically found in family members of an alcoholic" for "making a painful truce with people with whom we can live neither with nor without" (Cermak 1986). To manage the stress of living in a SAFO household, Cermak (1984: 41) argues that "offspring develop codependent attitude exquisitely attuned to . . . each nuance of [SAFO] interaction . . . before it gets out of control," and that offspring "hypervigilent" for stressful SAFO interactions develop codependent behavior "to control [their] own feelings [and] the feelings of actions of others."
A codependent attitude is a "relatively stable and enduring predisposition (Chaplin 1968: 42) for "a dysfunctional pattern of relating to others with an extreme external focus outside oneself, lack of expression of feelings, and personal meaning derived from relationships with others" (Fischer, Spann & Crawford 1991: 87), and codependent behavior connotes aiding and abetting an addicted person's behavior through over-zealous helping (Asher & Brissett 1988; Frank & Golden 1992; Haaken 1992; Haaken 1993; Harper & Capdevial 1990). Consider, for example, the experiment by Lyon and Greenberg (1991) that operationalized codependency as having an alcoholic parent and randomly assigned normal and codependent female college students to interact with a male playing exploitive or nurturant roles in the guise of a university researcher. Congruent with Cermak's (1986: 17) predictive assertion that codependency leads to "stress-related medical illnesses," the codependent women in Lyon and Greenberg's experiment liked the exploitive man better, described the exploitive man as more intelligent, and offered the man more help when his role was exploitive. Thus Cermak (1984; 1986) views codependency as a mediator of the relationship between SAFO and offspring stress-related medical problems, as shown in Figure 1. A mediating variable (codependency) is one that accounts for the relation between a predictor (SAFO) and criterion (medical problems) (Baron and Kenney 1986). -----------------
Insert Figure 1 about here
---------------- Cermak's hypothesis is a popular example of dynamic diathesis stress modeling (Windle 1997), an explanatory paradigm distinguished by three features.
First, it is consistent with the general model adopted in psychiatric research of the interaction between a person's constitutional predisposition to acquire a certain disorder (i.e., diathesis) and outside stressors. . . Second, the model may be viewed as . . . multivariate . . because it recognizes that many personal and environmental factors contribute to ongoing behavioral interactions and given outcomes. . . Third, the model is referred to as "dynamic, because it explicitly recognizes that the interrelationships between personal and environmental variables change over time and develop into meaningful regularities or cyclical patterns over time . . that may reciprocally influence each other . . . to produce . . . a given outcome. . ." Windle 1997: 189).
Such models are consistent with a body of circumstantial evidence that SAFO is a stressor (Beesley 1998; Fischer 1997; Lapointe 2001) associated with codependent offspring attitude (Black, Bucky, & Wilder-Padilla 1986; Crothers & Warren 1996; Fischer & Crawford 1992; Fischer, Wampler, Lyness & Thomas 1992; Fuller & Warner 2000; Potter-Efron & Potter-Efron 1989) and behavior (Burris 1999; Lyon & Greenberg 1991; Harkness, Swenson, Madsen-Hampton & Hale 2001). In turn, stress is thought to be related to (a) cardiovascular disorders associated with chronic hyper-arousal, (b) infections and cancers associated with hyperactivity of the pituitary-adrenal axis that require a healthy immune resistance, and (c) diseases of hyper-immunity (Barsky et al 1998; Cohen et al 2002; Kiecolt-Glasser et al 2002; Krantz & McCeney 2002; O'Leary et al 1997; Reynolds et al 2001; Vogt 1991). Moreover, a "strong, statistically significant" relationship has been reported between codependent attitude and self-perceptions of health and functional ability (Martsolf, Sedlak, & Doheny 2000: 156), and SAFO has been associated with elevated rates of endocrine, nutritional, metabolic, gastro-intestinal, and genito-urinary (Holder & Blose 1986; Holder & Hallan 1981; Roberts & Brent 1982; Woodside, Coughey, & Cohen 1993) and "other medical and psychological" categories of offspring health problems (Johnson and Leff 1999; Loughead, Spurlock, & Ting 1998; Meyer 1997; Plotnick et al 1982; Roehling & Gaumond 1996; Wells, Glickhauf-Hughes & Bruss 1988). Finally, codependent attitude and behavior were found to predict some adult-offspring medical problems in a recent study by Harkness (in press).
The Moderating View: A Second Hypothesis
Against the backdrop of anecdotal and circumstantial evidence, Cermak's dynamic stress- model linkage between SAFO and offspring codependency and medical problems is persuasive, but consider that SAFO offspring were found to have no more health problems overall than other children in a study of 599 preadolescents by Dobkin, Tremblay, and Desmarias-Gervais (1994). Indeed, SAFO offspring utilized less health care than the offspring of non-SAFO parents in a large-sample study by Shatkin (1990). Trying to explain why the 1,295 health-insured SAFO children in her sample utilized less health care than did the 34,236 health-insured offspring of non-SAFO parents, Shatkin (1990) discovered that SAFO offspring had more insurance claims for medical problems related to accidental injury (open wounds, poisoning by drugs, and toxic effects of nonmedicinal substances) than did their non-SAFO counterparts, as well as higher mean expenditures for open wounds to the head, neck, and trunk, and higher mean expenditures for inpatient care. Buttressed by a study of 2 million insurance subscribers and their dependents that found higher rates of accidents and hospital days among SAFO family members (Blue Cross 1987), and by a study of 1.6 million insurance subscribers and their dependents that reported higher rates of injuries, poisonings, inpatient admissions, and hospitalization days among SAFO offspring (Woodside, Coughey & Cohen 1993; See also Dube, Anda, Felitti, Croft, Edwards & Giles 2001), Shatkin's (1990) findings suggest that SAFO offspring are both less likely to receive medical care for routine childhood or chronic health problems and more likely to receive care for acute medical problems in the United States. (Similar findings have been reported for SAFO offspring in India [Rao, Begum, Venkataramana & Gangadharappa 2001]).
If SAFO offspring face a higher risk for acute medical problems but receive less care for routine childhood or chronic health problems, Harkness (in press) has reasoned that offspring with codependent attitude might acquire an important health advantage in SAFO households, because codependent attitude predicts risk-avoidant behavior (Fischer, Wampler, Lyness & Thomas 1992). Notwithstanding stresses associated with role-reversed interactions between parents and codependent caretaking children, hypervigilence for the vicissitudes of SAFO interaction coupled with risk-avoidant behavior might help offspring avoid acute medical-problem events. Moreover, role-reversed SAFO-parent interactions might improve offspring access to routine medical care. Thus, according to our hypothesis, codependency should moderate the relationship between SAFO and acute offspring medical problems. Where mediating variables catalyze the relationship between a predictor and its criterion (Baron & Kenny 1986), moderating variables influence the direction or strength of the predictor-criterion relationship (Rogosch, Chassin & Sher 1990), as shown in Figure 2.-----------------
Insert Figure 2 about here
A Feminist Perspective: The Smokescreen Hypothesis
Cermak views codependency as the catalyzing mediator of the relationship between SAFO and offspring medical problems, and our research suggests that codependency functions as a moderator of that relationship, but from an altogether different perspective the mediator-moderator controversy is a tempest in a teapot because codependency is a smokescreen for sexism in health care (Asher & Brissett 1988; Haaken 1993; Harkness & Cotrell, 1997). This logic of this feminist perspective argues that because men are more likely than women to seek treatment for addictions (Beckman & Amaro 1984; SAMHSA 1993; Weisner & Schmidt 1992), and because men are more likely to find employment in the field of addictions (NIDA 1979; NIAA 1984; Rosenqvist 1991; Weisner & Schmidt 1992), that men are diagnosing women "codependent," disproportionately and in absentia. Thus, feminist critics have charged that codependency is a slanderous label used to pathologize women for helpful interpersonal attitude and behavior, tantamount to "blaming the victims" of male substance abuse for enacting traditional gender roles (Asher 1992; Asher & Brissett 1988; Brown 1988; Collins 1993; Frank & Golden 1992; Haaken 1992; Haaken 1993; Harper & Capdevial 1990; Katz & Aimee 1991; Krestan & Bepko 1990; Mulry 1987), and emblematic of sexism in substance-abuse treatment (Vannicelli 1984a; Wilsnack, Klassen, Schur & Wilsnack 1991).
The empirical evidence has been of little help in help in clarifying such issues. Some investigators have found that addictions counselors make gender-biased treatment decisions (Pello 1987; Walitzer & Conners 1997), but others have not (Callaway 1995). And although neither Thorpe (1997), nor Harkness and Cotrell (1987), nor Harkness, Swenson, Madsen-Hampton, and Hale (2001) were able to detect gender bias in studies of how addictions counselors evaluate codependency in clinical practice, skeptics might remind us of charges that gender bias infuses addictions research (Vannicelli 1984b; Brett, Graham & Smythe 1995; Greenfield 2002).
Linking SAFO and Offspring Medical Problems with Gender
Equal access to appropriate health care is at the heart of concern that addictions counselors and research are biased for gender (Walitzer & Connors 1997; Greenfield 2002), because some evidence suggests that male and female SAFO offspring may have different medical problems. In a study of 51 families with at least one alcoholic family member and 90 matched control families, for example, Roberts and Brent (1982) found that female (but not male) SAFO offspring had higher health-care utilization rates, higher rates of certain endocrine, nutritional, metabolic, and genito-urinary diagnoses, higher rates of stress-related diseases in the gastrointestinal, nervous, and "other medical and psychological" categories (than controls) in their health-care records. Conversely, in a study of 100 children with one or two alcoholic parents and a control group of 100 children, Chafetz, Blane, & Hill (1971) found that male (but not female) offspring of SAFO families had a greater tendency to have had a serious illness or accident, and Putnam (1987) found that sons (but not daughters) of alcoholics had an injury rate 1.57 times higher than controls in a study of 126 children of alcoholics and a control group of 114 children. The global impression left by this pattern of findings is that female SAFO offspring utilize health care for medical problems that are chronic in nature, whereas male offspring utilize health care for acute medical problems, but such inferences are based on small samples. Unfortunately, the large-sample studies of health-care utilization among SAFO offspring conducted by Blue Cross (1987) and Shatkin (1990) did not report gendered analyses.
Insert Figure 3 about here
Apparently, only two investigations have directly examined the relationship among SAFO, codependency, and medical problems. Testing a latent-variable stress model of risk for and resistance to physical illness among 210 employee assistance program participants, Yoshimura (1991) reported that none of the illness variables in her study were associated with codependency. Using multiple regression to test an alternative model, Harkness (in press) found that codependent attitude and behavior moderated the relationship between SAFO and measures of acute offspring medical problems. A limitation of both studies, however, was that neither investigation reported analyses of the data by gender.
To address that limitation, as shown in Figure 3, we predicted that gender would interact with codependent attitude and behavior to moderate the relationship between SAFO and three self-report measures of acute medical problems among SAFO offspring: (1) hospitalizations, (2) days of recent medical problems, (3) perceived trouble with medical problems. Our focus on acute medical problems was guided by the alarming large-sample findings of Blue Cross (1987) and Shatkin (1990). Any interaction between codependency and gender should benefit SAFO men more than women, we reasoned, because Fischer, Wampler, Lyness, and Thomas (1992) found that codependent attitude reduce risk-taking in men more than women. If one assumes that risk-avoidance reduces accidents, injuries, and other forms of acute medical problems, then male offspring should accrue greater health benefits from codependent attitude and behavior.
MATERIALS AND METHODS
To test our hypotheses, we used a popular self-report measure of codependent attitude and independent expert observations of codependent behavior to examine the interaction between codependency and gender as moderators of the relationship between SAFO and self-reports of illness in a heterogenous sample of adult males and females. Our study was conducted as part of a counterbalanced multiple-treatment experiment (Kazdin 1992) designed to evaluate the reliability and validity of the Idaho Codependency scale (Harkness, Swenson, Madsen-Hampton & Hale 2001), an observational measure of codependent behavior. We recruited ten adults to participate as "cases" in a study of codependency as perceived "through the eyes and ears of substance-abuse counselors." Volunteer cases completed a questionnaire and participated in a semi-structured videotaped interview. The questionnaire included demographic questions, the Spann-Fischer Codependency Scale (Fischer, Spann & Crawford 1991), and questions about medical problems drawn from the Addiction Severity Index (McLellan et al 1985). Interviews were conducted to elicit behavioral signs and symptoms of codependency for independent counselor ratings by asking each subject about substance abuse in their family of origin and other interpersonal relationships. The interviews were video-taped, and the video-tapes were evaluated for codependency by 27 substance-abuse counselors trained in the use of the Idaho Scale. The materials and methods used in the study were reviewed and approved for human-subjects research by independent review boards in university and medical settings.
Male and female cases were recruited from five adult populations: (1) adult spouses of outpatients in substance-abuse treatment were recruited to obtain traditional Codependents; (2) unrelated adult outpatients in substance-abuse treatment were recruited because Cermak (1986) argued that Codependents and substance abusers share the same compulsive psychology; (3) members of Codependents Anonymous were recruited as recovering codependency cases; (4) Bureau of Land Management Smoke Jumpers were recruited as risk-taking icons of rugged individualism; and (5) university students majoring in business or economics were recruited on the assumption that distinguishing between profitable and unprofitable investments was at the heart of their training. Harkness, Swenson, Madsen-Hampton, and Hale (2001) have described our recruitment procedures in detail.
Male and female cases were recruited from diverse groups in order to maximize the variation of substance-abuse in the family of origin, codependency, and health status. Although
Smoke Jumpers, who earn their living by parachuting from airplanes to extinguish forest fires, might invite playful contrasts and comparisons with traditional views of codependent behavior, we had a hunch that codependency would be low in this group, because Fischer, Wampler, Lyness, and Thomas (1992) have found that codependency moderates adult risk-taking behavior. Business and economics majors were recruited on the assumption that goal-oriented students seeking employment in those market sectors would be less likely to manifest codependent behavior than others. A male and female case was recruited from each group to address concerns that the diagnosis and measurement of codependency may be biased for gender (Asher 1992; Asher & Brissett 1988; Collins 1993; Finagrette 1991; Fischer, Spann & Crawford 1991; Frank & Golden 1992; Haaken 1992; Haaken 1993; Harper & Capdevial 1990; Katz & Amee 1991; Krestan & Bepko 1990; Longino 1993; Nelson-Zlupko, Kauffman & Dore 1995; Swigonski 1994; Harkness & Cotrell 1997; Harkness, Swenson, Madsen-Hampton & Hale 2001). The cases who volunteered for this study reported three ethnic backgrounds and ages from 18 to 65 years (0 = 41). Additional descriptions might jeopardize the privacy of individuals who volunteered to participate in this study.
Counselors from a state list of 88 certified substance-abuse counselors were recruited to evaluate cases for codependent behavior. Counselors were recruited by offered continuing education credit for completing a workshop in the measurement of codependency. Twenty-seven counselors (31 percent of the sampling frame) enrolled in the workshops, nine males and 18 females. The ethnicities counselors used to describe themselves included Native American, Hispanic, and Caucasian. Counselors ranged from 28 to 70 years in age (0 = 49.48), and reported from one to 20 years of substance-abuse counseling experience (0 = 8.7). Thirty percent of the counselors had earned baccalaureate degrees, and 52% had earned graduate degrees. Their disciplines included counseling (N = 14), rehabilitation (N = 1), social work (N = 7), psychology (N = 2), nursing (N = 1), and other (N = 2). Six counselors endorsed the Minnesota model of treatment, two reported a cognitive orientation, two reported a family-systems orientation, and the remaining counselors described their orientation as eclectic. Seventy-five percent of the counselors agreed that they were "in recovery;" 78% acknowledged that a close family member or "significant other" was currently abusing alcohol or other drugs; and 63% were self-described adult children of alcoholics.
The Health and Employment Questionnaire developed for this study was administered to each case prior to the videotaped interview. In addition to inviting each case to self-report gender, the questionnaire included a measure of codependent attitude and asked questions about health problems and health-care utilization. The Spann-Fischer Codependency Scale (SFCS) is a 16-item paper-and-pencil self-report attitude scale that respondents use to describe their maintenance of an "extreme external focus outside oneself, lack of expression of feelings, and personal meaning derived from relationships with others" (Fischer, Spann & Crawford 1991, p. 87). SFCS items are rated on a six-point Likert scale, with scale scores ranging from a low of 16 to a high of 96. With Cronbach's alphas ranging from .73 to .80, the reliability of the SFCS is distinguished relative to other measures of codependent attitude, and a number of investigators have confirmed its reliability and validity (e.g., Crothers & Warren 1996; Fischer & Crawford 1992; Fischer, Wampler, Lyness & Thomas 1992; Hewes & Janikowski 1998; Pidcock & Fischer 1998). The three health questions posed by the questionnaire were drawn from the Addiction Severity Index (McLellan et al 1985).
• How many times in your life have you been hospitalized?
• How many days have you experienced medical problems in the past 30?
• How troubled or bothered have you been by these medical problems
in the past 30 days?
Each case participated in a semi-structured video-taped interview. A protocol was developed to standardize the interview format, using open-ended and closed-ended questions suggested by the theoretical and empirical codependency literature to elicit codependent behavior. The protocol was pilot-tested with substance-abuse counselors and revised. A copy of the protocol is available from the author.
Interviews were conducted in a small university studio designed to videotape distance-education lectures. An electronic device was used to distort the visual image of each videotaped case in order to disguise their identity. Participants were invited to read the printed interview protocol before the interview began, and seating arrangements enabled participants to control the recording equipment. During the interview, each female was called "Mary," and each male was called "John."
At a pre-determined point in each interview, the participant was asked to answer four yes-or-no questions about SAFO. The answers were recorded and used to determine whether the subject's parents had ever abused substances or been chemically-dependent "when you were growing up." On that basis, a global SAFO index was calculated for each case by summing the respondent's affirmative answers.
After reminding subjects how to turn off the recording equipment, the interviewer offered to address any unanswered questions about the research, and observed that sometimes the interview raised personal questions. Each subject was offered a list of certified substance-abuse facilities and counselors to address any residual clinical issues that the interview raised. Interviews ranged from 15 to 35 minutes in length (0 = 22.2 minutes).
Substance-abuse counselors were randomly assigned to one of four workshops to evaluate codependent behavior observed in a series of five video-taped cases. One workshop observed male cases, one workshop observed female cases, and two workshops observed cases of both genders. A coin-tossing procedure was used to assign the male or female case from each of the three codependency groups (outpatients, spouses, and Codependents Anonymous) and two control groups (Smoke Jumpers and students) to one of the two mixed-gender workshops. After case videotapes were assigned to workshops, the order of presentation was randomized within each workshop.
Each workshop included 1.5 hours of data collection and training. Workshops began with a brief overview of codependency measurement tools and their psychometric characteristics. Next, the investigators explained the development of the Idaho Codependency scale and the purpose of the present research. Finally, counselors observed a series of training videotapes for one supervised hour of rating practice.
The Idaho Codependency Scale is an example-anchored rating scale that operationalizes codependency as a continuum of behavior, as rank-ordered by a national sample of expert judges (Harkness & Cotrell 1997). The Idaho scale appears to have excellent inter-rater reliability and construct, convergent, discriminant, and concurrent validity in the hands of well-trained judges (Harkness, Swenson, Madsen-Hampton & Hale 2001), and our pilot study of this rating method found it free of gender bias (Harkness & Cotrell 1997). A copy of the scale is available from the first author.
Whether codependency and gender are moderators of the relationship between substance-abuse in the family of origin and adult-offspring illness is a multivariate question. The recommended procedure for evaluating multivariate moderation models is multiple regression (Rogosch, Chassin & Sher 1990; Jaccard, Turrisi & Wan 1990). Several dimensions of our study militated against the use of multiple regression, however. For example, multiple regression is a parametric procedure which assumes that predictor variables are quantitative or dichotomous and that dependent variables are quantitative and normally-distributed (Berry 1993), but the shape of the codependency distribution is unknown, and the Spann-Fischer and Idaho scales are not quantitative measures but ordinal. On the other hand, the Gauss-Markov assumptions of multiple regression are violated routinely to explore multivariate questions, because practical non-parametric multivariate alternatives do not exist (Bradley 1968; Cliff 1996). Another limitation may be due to our small-sample estimation procedure. We "bootstrapped"(Diciccio & Romano 1988; Efron 1981; Mooney & Duval 1993) the size of our sample by an average factor of 13.5 to regress medical problems on self-reports of SAFO, attitude, and gender, linking self-report measures of SAFO, codependent attitude, gender, and medical problems from the ten "cases"in our study with 135 independent counselor ratings of codependent behavior. With these these limitations, we invite replications that challenge our findings.
We tested the interaction between gender and codependency as a moderator of the relationship between SAFO and offspring medical problems by conducting a series of three-step regression analyses. Main effects for SAFO, codependency, and gender were entered in step one; two-way interactions between SAFO and codependency, SAFO and gender, and codependency and gender were entered in step two; and the three-way interaction among SAFO, codependency, and gender was entered in step three. Significant two and three-way interactions indicate moderation effects (Rogosch, Chassin & Sher 1990). We examined codependent attitude first, with replications for codependent behavior.
The first hypothesis predicted that the interaction between gender and codependency would moderate the relationship between SAFO and offspring illness by reducing hospitalizations for men relative to women. In the omnibus analysis, we found significant main effects for SAFO, gender, and codependent attitude (R2 Change = .301, p = .000), and significant interactions between SAFO and gender, SAFO and codependent attitude, and codependent attitude and gender (R2 Change = .627, p = .000). However, the three-way interaction among SAFO, codependent attitude, and gender was not significant. The evidence did not support the attitudinal hypothesis for hospitalization.
A somewhat different pattern of findings was observed in the replication for codependent behavior. We found significant main effects for SAFO, gender, and codependent behavior ( R2 Change = .299, p = .000), and significant interactions between SAFO and gender, SAFO and codependent behavior, and codependent behavior and gender (R2 Change = .118, p = .000). The three-way interaction among SAFO, gender, and codependent behavior was significant as well (R2 Change = .089, p = .000), supporting the behavioral hypothesis.
The significant three-way interaction among SAFO, gender, and codependent behavior invited a gendered analysis of codependent behavior as a moderator of the relationship between SAFO and adult-offspring hospitalizations. After establishing the moderating effect of codependent behavior in our omnibus analysis, we proceeded to test our hypothesis by analyzing the data for males and females separately. To simplify interpretation, we regressed hospitalizations on SAFO in the first step of each gendered analysis, and introduced codependent behavior and SAFO x codependent behavior interaction in step two.
Insert Figure 4 about here
As Figure 4 reveals, SAFO x codependent behavior interaction had a moderating effect on the relationship between SAFO and the hospitalization of male offspring, but the same interaction mediated the relationship between SAFO and the hospitalization of female offspring. Codependent behavior, which reduced the risk of hospitalization for the male offspring of substance-abusing families, catalyzed the risk of hospitalization for females.
Days of Recent Medical Problems
The second hypothesis predicted that the interaction between gender and codependency would moderate the relationship between SAFO and offspring illness by reducing days of recent medical problems for men relative to women. In the omnibus analysis, we found significant main effects for SAFO, gender, and codependent attitude (R2 Change = .172, p = .000), as well as significant interactions between SAFO and gender, and between SAFO and codependent attitude (R2 Change = .627, p = .000). Neither the two-way interaction between codependent attitude and gender, nor the three-way interaction among SAFO, codependent attitude, and gender, had a significant moderating effect on the relationship between SAFO and days of recent medical problems. The evidence disconfirmed the attitudinal hypothesis.
We observed a different pattern of findings in the replication for codependent behavior. We found significant main effects for SAFO, gender, and codependent behavior ( R2 Change = .175, p = .000), as well as significant interactions between SAFO and gender, and SAFO and codependent behavior (R2 Change = .06, p = .021). And although the two-way interaction between codependent behavior and gender was not significant, the 3-way interaction among SAFO, gender, and codependent behavior was significant (R2 Change = .120, p = .000), supporting the behavioral hypothesis.
The significant 3-way interaction among SAFO, gender, and codependent behavior encouraged another gendered analysis of codependent behavior as a moderator of the relationship between SAFO and adult-offspring illness. Having established the moderating effect of codependent behavior in an omnibus analysis, once again we analyzed the data for males and females separately, regressing days of recent medical problems on SAFO in step one, and introducing codependent behavior and SAFO x codependent behavior interaction in step two. Figure 5 shows our findings. ------------------
Insert Figure 5 about here
In Figure 5, SAFO x codependent behavior interaction had a moderating effect on the relationship between SAFO and days of recent medical problems among male offspring, but the same interaction mediated the relationship between SAFO and days of recent medical problems for female offspring. Codependent behavior, which reduced the risk of medical-problem days for men, catalyzed that risk for women.
Trouble with Medical Problems
The third hypothesis predicted that the interaction between gender and codependency would moderate the relationship between SAFO and offspring illness by reducing how much trouble with recent medical problems male offspring reported relative to females. In the omnibus analysis, we found significant main effects for gender and codependent attitude (R2 Change = .458, p = .000), but not for substance abuse in the family of origin. However, we found significant interactions between SAFO and gender, SAFO and codependent attitude, and codependent attitude and gender (R2 Change = .300, p = .000) in the second regression model, in which the main effect for SAFO became significant. The evidence did not support the attitudinal hypothesis, however, because the three-way interaction among SAFO, codependent attitude, and gender was not significant.
Once again, the replication for codependent behavior showed a different pattern of
findings. We obtained significant main effects for SAFO, gender, and codependent behavior (R2
Change = .261, p = .000), as well as significant interactions between SAFO and gender, SAFO
and codependent behavior, and codependent behavior and gender (R2 Change = .122, p = .000).
Moreover, the 3-way interaction among SAFO, gender, and codependent behavior was also
significant (R2 Change = .093, p = .000), consistent with the third behavioral hypothesis.
The significant 3-way interaction among SAFO, gender, and codependent behavior prompted a final gendered analysis. We analyzed the data for males and females separately, regressing trouble with medical problems on SAFO in the first step of the analysis, and introducing codependent behavior and SAFO x codependent behavior interaction in step two.
Insert Figure 6 about here
------------------ In Figure 6, male codependent behavior weakened the relationship between SAFO and trouble with medical problems, but the relationship between SAFO and trouble with medical problems was exacerbated by female codependent behavior. As a moderator of offspring trouble with medical problems, codependent behavior helped men but hurt women.
Substance abuse in the family of origin puts offspring at risk of acute medical problems, according to large-sample studies of health care utilization. In this study, we explored that phenomenon from three points of view:(1) on the basis of anecdotal evidence from clinical practice, Cermak (1984; 1986) has argued persuasively that the relationship between SAFO and stress-related offspring medical problems is mediated or catalyzed by offspring codependency; (2) we have argued that codependency moderates the relationship between SAFO and acute offspring medical problems, based on empirical evidence from our own small-sample research (Harkness in press); but (3) both points of view are suspect from a feminist perspective that perceives codependency as a smokescreen for sexism in addictions treatment and research.
The feminist critique of codependency is consistent with evidence that male and female SAFO offspring have different health-care problems, that male and female offspring utilize health-care differently, and that much health-care research has been biased for gender. Thus, although large-sample studies suggests that all SAFO offspring are at risk for acute medical problems, other evidence suggests that the risk of acute medical problems may be greater for male offspring, whereas the risk of chronic medical problems may be greater for female offspring. Finally, it seems reasonable to infer that females have been more likely than males to be labeled codependent, if only because females have been much less likely than males to receive treatment for addictions, biasing codependency theory and research. To address that critique, we retested our moderation model of the relationship among substance abuse in the family of origin, codependent attitude and behavior, and offspring medical problems for gender interactions.
Based largely on a report by Fischer, Wampler, Lyness & Thomas (1992) that a codependent attitude may be an adaptation that buffers offspring against some SAFO health hazards by enhancing risk avoidance, and an adaptation that helps men more than women, our model predicted that the interaction of gender and codependent attitude and behavior would moderate the relationship between SAFO and acute offspring medical problems, favoring men over women.
The model we tested oversimplified the complex relationship observed among SAFO and offspring codependency, gender, and medical problems. First, the interaction between offspring gender and codependent attitude had no discernable effect on the relationship between SAFO and offspring medical problems. Second, male codependent behavior moderated the impact of SAFO on hospitalizations and medical-problem days, reducing health risks for men, but the codependent behavior of females mediated the relationship between SAFO and health, catalyzing health risks for women. Finally, although codependent behavior moderated the relationship between substance abuse in the family of origin and trouble with medical problems for both genders, the only beneficiaries were males. Once again, codependent behavior hurt women.
Some findings of our findings are consistent with Cermak's mediation model of codependency, some are consistent with our moderation model, and some findings are consistent with a feminist critique of both models. Codependent behavior did mediate the relationship between SAFO and offspring hospitalizations and medical-problem days, as Cermak (1984; 1986) might have predicted, but only for women. Moreover, codependent behavior did moderate the relationship between SAFO and offspring hospitalizations and medical-problem days, as we predicted, but only for men. Finally, to a degree that surprised us, the relationship among SAFO, codependency, and health outcomes was gendered, as feminists have claimed.
Threats to the validity of this study may support alternative interpretations of our findings. The volunteers we recruited for this study were selected for heterogeneity, not representation, for example, and volunteer self-selection may have biased our findings. Second, statistical conclusion validity is always at issue when powerful multivariate statistics are used to test multiple hypothesis about small-sample human behavior. Finally, because our sample was too small to support generalization of our findings, we invite challenging replications of this exploratory pilot study.
Asher, R. 1992. Women with alcoholic husbands: Ambivalence and the trap of codependency. Chapel Hill: The University of North Carolina Press.
Asher, R., & Brissett, D. 1988. Codependency: A view from women married to alcoholics. The International Journal of the Addictions 23: 331-350.
Baron, R., & Kenny, D. 1986. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51: 1173-1182.
Barsky, A., Orav, J., Delamater, B., Clancey, S., & Hartley, H. 1998. Cardiorespiratory symptoms in response to physiological arousal. Psychosomatic Medicine 60: 604-609.
Beckman, L., & Amaro, H. 1984. Patterns of women's use of alcohol treatment agencies. In Wilsnack, S., & Beckman, L., Alcohol problems in women: Antecedents, consequences, and intervention (pp. 319-348). New York: Guilford.
Beesley, D. 1998. The relationship between need for control, attachment style, and relationship satisfaction among adult children of alcoholics. Doctoral dissertation, The University of Oklahoma.
Berry, W. 1993. Understanding regression assumptions. Newbury Park, CA: Sage.
Black, C., Bucky, S., & Wilder-Padilla, S. 1986. The interpersonal and emotional consequences of being an adult child of an alcoholic. International Journal of the Addictions 21: 213-231.
Blue Cross of Greater Philadelphia and Pennsylvania Blue Shield. 1987. 1987 Community Data Report. Philadelphia, Pennsylvania: Blue Cross of Greater Philadelphia.
Bradley, J. 1968. Distribution-free statistical tests. Englewood Cliff, New Jersey: Prentice-Hall.
Brett, P., Graham, K., & Smythe, C. 1995. "An analysis of specialty journals on alcohol, drugs, and addictive behaviors for sex bias in research methods and reporting." Journal of Studies on Alcohol 56(1): 24-34.
Brown, S. 1988. Treating adult children of alcoholics: A developmental perspective.
N.Y.: John Wiley & Sons.
Burris, C. 1999. Stand by your (exploitive) man: Codependency and responses to performance feedback. Journal of Social and Clinical Psychology 18 (1): 272-298.
Callaway, R. 1995. Bias in alcohol referral: A study assessing bias among mental health training programs. Ph.D. dissertation: University of Memphis.
Cermak, T. 1984. Children of alcoholics and the case for a new diagnostic category of codependency. Alcohol Health and Research World 8 (4): 38-42.
Cermak, T. 1986. Diagnostic criteria for codependency. Journal of Psychoactive Drugs 18: 15-20.
Chafetz, M., Blane, H., and Hill, M. 1971. Children of alcoholics: Observations in a child guidance clinic. Quarterly Journal of Studies on Alcoholism 32: 687 - 698.
Chaplin, J.P. 1968. Dictionary of psychology. New York: Dell.
Cliff, N. 1996. Ordinal methods for behavioral data analysis. Mahway, New Jersey: Lawrence Erlbaum Associates, Publishers.
Cohen, S., Hamrick, N., Rodriguez, M., Feldman, P., Rabin, B., & Manuck, S. 2002. Reactivity and vulnerability to stress-associated risk for upper respiratory illness. Psychosomatic Medicine 64: 302-310.
Collins, B. 1993. Reconstruing codependency using self-in-relation theory: A feminist perspective. Social Work 38: 470-476.
Crothers, M., & Warren, L. 1996. Parental antecedents of adult codependency. Journal of Clinical Psychology 52 (2): 231-239.
Diciccio, T., & Romano, J. 1988. A review of bootstrap confidence intervals. Journal of the Royal Statistical Society 50 (3): 338-354.
Dobkin, P., Tremblay, R., Desmarias-Gervais, L., & Depelteau, L. 1994. "Is having an alcoholic father hazardous for children's physical health?" Addiction 89(12):1619-1627.
Dube, S., Anda, R., Felitti, V., Croft, J., Edwards, V., & Giles, W. 2001. "Growing up with parental alcohol abuse: Exposure to childhood abuse, neglect, and household dysfunction. Child Abuse & Neglect 25(2): 1627-1640.
Efron, B. 1979. Bootstrap methods: Another look at the jacknife. Annals of Statistics 7: 1-26.
Eigen, L. D., & Rowden, D. W. 1995. A methodology and current estimate of the number of children of alcoholics in the United States." In Abbott, S. (Ed.), Children of Alcoholics: Selected Readings (pp. 77 - 97). Rockville, MD: National Association for Children of Alcoholics.
Finagrette, H. 1991. Alcoholism: The mythical disease. In D. Pittman & H. White, (Eds.). Society, culture, and drinking patterns reexamined (pp. 417-438). New Brunswick, NJ: Rutgers Center of Alcohol Studies.
Fischer, J., & Crawford, D. 1992. Codependency and parenting styles. Journal of Adolescent Research 7: 352-363.
Fischer, J., Spann, L., & Crawford, D. 1991. Measuring codependency. Alcoholism Treatment Quarterly 8: 87-99.
Fischer, J., Wampler, R., Lyness, K., & Thomas, E. 1992. Offspring codependency: Blocking the impact of the family of origin. Family Dynamics of Addiction Quarterly 2: 1-12.
Fischer, K. 1997. An investigation of the relationship of parental alcoholism and family dysfunction to stress and problem drinking among college students. Doctoral dissertation, Southern Illinois University.
Fitzpatrick, Mary A. 1995. Explaining family interactions. Thousand Oaks, CA: Sage Publications.
Frank, P., & Golden, G. 1992. Blaming by naming. Social Work 37: 5-6.
Fuller, J., & Warner, R. 2000. Family stressors as predictors of codependency. Genetic, Social, & General Psychology Monographs 126 (1): 5-22.
Gotham, H., & Sher, K. 1996. Do codependent traits involve more than basic dimensions of personality and psychopathology? Journal of Studies on Alcohol 57(1): 34-39.
Greenfield, S. 2002. "Women and alcohol use disorders." Harvard Review of Psychiatry, 10(2): 76-85.
Haaken, J. 1992. A critical analysis of the co-dependence construct. Psychiatry 53: 396- 406.
Haaken, J. 1993. From Al-anon to ACOA: Codependence and the reconstruction of caregiving. Signs 18: 321-345.
Harkness, D. in press. "To have and to hold: Codependency as a moderator of the relationship between substance abuse in the family of origin and adult-offspring medical problems." Journal of Psychoactive Drugs.
Harkness, D., & Cotrell, G. 1997. The social construction of co-dependency in the treatment of substance abuse. Journal of Substance Abuse Treatment 14 (5): 473-480.
Harkness, D., Swenson, M., Madsen-Hampton, K., & Hale, R. 2001. The development, reliability, and validity of a clinical rating scale for co-dependency. Journal of Psychoactive Drugs 33 (2): 159-171.
Harper, J., & Capdevial, C. 1990. Codependency: A critique. Journal of Psychoactive Drugs 22: 285-292.
Hewes, S., & Janikowski, R. 1998. Parental alcoholism and codependency: A comparison of female children of alcoholics in two college-age groups. College Student Journal 32 (1): 140-147.
Holder, H., & Blose, J. 1986. Alcoholism treatment and total health care utilization and costs: A four-year longitudinal analysis of federal employees. Journal of the American Medical Association 256 (1): 1456-1460.
Holder, H., & Hallan, H. 1981. Medical care and alcoholism treatment costs and utilization: A five-year analysis of the California pilot project to provide health insurance coverage for alcoholism. Prepared for NIAAA under contract No. ADM 281-79-0008.
Jaccard, J., Turrisi, R., & Wan, C. 1990. Interaction effects in multiple regression. Newbury Park, CA: Sage.
Johnson, J., & Leff, M. 1999. Children of substance abusers: Overview of research findings. Pediatrics 103 (5): 1085- 1099.
Katz, S., & Aimee, L. 1991. The codependency conspiracy. NY: Warner Books.
Kazdin, A. 1992. Research design in clinical psychology (2nd. ed). Needham Heights, Mass: Allyn & Bacon.
Kiecolt-Glasser, J., McGuire, L., Robles, T., & Glaser, R. 2002. Psychoimmunology and psychosomatic medicine: Back to the future. Psychosomatic Medicine 64: 15-28.
Krantz, D., & McCeney, M. 2002. Effects of psychological and social factors on organic disease: A critical assessment of research on coronary heart disease. Annual Review of Psychology 53: 341-69.
Krestan, J., & Bepko, C. 1990. Codependency: The social reconstruction of female
experience. Smith College Studies in Social Work 60: 217-232.
Lapointe, A. 2001. An investigation of the impact of family of origin and the role of internalized shame on adult intimate relationship satisfaction for adult children of alcoholics. Doctoral dissertation, The University of Kansas.
Lindley, N., Giordano, P., & Hammer, E. 1999. Codependency: Predictors and psychometric issues. Journal of Clinical Psychology 55(1): 559-64.
Longino, H. 1993. Feminist standpoint theory and the problems of knowledge. Signs:
Journal of Women in Culture and Society 19: 201-212.
Loughead, T., Spurlock, V., & Ting, Y. 1988. Diagnostic indicators of codependence: An investigation using the MCMI-II. Journal of Mental Health Counseling 20 (1): 64-76.
Lyon, D., & Greenberg, J. 1991. Evidence of codependency in women with an
alcoholic parent: Helping out mr. wrong. Journal of Personality and Social Psychology 61: 435-439.
Martsolf, C., Sedlak, C., & Doheny, M. 2000. Codependency and related health variables. Archives of Psychiatric Nursing 14 (3): 150-158.
McLellan, A., Luborsky, L., Cacciola, J., Griffith, J., McGahan, P., & O-Brien, C. 1985. Guide to the addiction severity index: Background, administration, and field testing results. DHHS Publication No. (ADM) 85-1419. Rockville, Maryland: NIDA.
Meyer, D. 1997. Codependency as a mediator between stressful events and eating disorders. Journal of Clinical Psychology 53 (2): 107-116.
Mooney, C., & Duval, R. 1993. Bootstrapping: A nonparametric approach to statistical inference. Newbury Park, CA: Sage.
Mulry, J. 1987. Co-dependency: A family addiction. American Family Physician 35: 215-219.
National Institute on Alcohol Abuse and Alcoholism. 1984. Development of model professional standards for counselor credentialing (DHHS Publication No. ADM 281-82-000 1). Washington, DC: U.S. Government Printing Office.
National Institute on Drug Abuse 1979. Professional and paraprofessional drug abuse counselors (DHEW Publication No. ADM 79-85 8). Washington, DC: U.S. Government Printing Office.
Nelson-Zlupko, L., Kauffman, E., & Dore, M. 1995. Gender differences in drug
addiction and treatment: Implications for social work intervention with substance-abusing women. Social Work 40: 45-54.
O'Leary, A., Brown, S., & Suarez-Al-Adam, M. 1997. Stress and immune function. In T. Miller (Ed.), Clinical Disorders and Stressful Life Events (pp. 181-215). Madison, CT: International Universities Press.
Pello, M. 1987. Gender-related hospital staff attitudes about chemical addiction. Ph.D. dissertation: Fordham University.
Pidcock, B., & Fischer, J. 1998. Parental recovery as a moderating variable of adult offspring problematic behaviors. Alcoholism Treatment Quarterly 16 (4): 45-57.
Plotnick, D. Adams, K., Hunter, H., & Rowe, J. 1982. Alcoholism treatment programs within prepaid group practice HMOs: A final report. Report prepared for the National Institute on Alcohol Abuse and Alcoholism. Washington, DC: U.S. Department of Health and Human Services.
Potter-Efron, R., & Potter-Efron, P. 1989. Assessment of co-dependency with individuals from alcoholic and chemically dependent families. Alcohol Treatment Quarterly 6 (1): 37-57.
Putnam, S. 1987. Family stress and illness behavior: The effects of alcoholism, other major illness, and role stress on family members utilization of HMO services. Ph.D. Dissertation, Brown University.
Rao, K., Begnum, S., Venkataramana, V., & Gangadharappa, N. 2001. Nutritional neglect and physical abuse in children of alcoholics. Indian Journal of Pediatrics 68 (9): 843-845.
Reynolds, L., O'Koon, J., Papademetiou, E., Szczygiel, S., & Grant, K. 2001. Stress and somatic complaints in low-income urban adolescents. Journal of Youth & Adolescence 30: 499-514.
Roberts, K., & Brent, E. 1982. Physician utilization and illness patterns in families of alcoholics. Journal of Studies on Alcohol 43 (1): 119-128.
Roehling, P., & Gaumond, E. 1996. Reliability and validity of the codependent questionnaire. Alcoholism Treatment Quarterly 14 (1): 85-95.
Rogosch, F., Chassin, L., & Sher, K. 1990. Personality variables as mediators and moderators of family history risk for alcoholism: Conceptual and methodological issues. Journal of Studies on Alcohol 51 (4): 310-318.
Rosenqvist, P. 1991. AA, Al-Anon and gender. Contemporary Drug Problems: 687-704.
Russell, M. 1990. "Prevalence of alcoholism among children of alcoholics." In Windle, M., & Searles, J.S. (Eds.), Children of Alcoholics: Critical Perspectives (pp. 9 - 38). New York: Guildford Press.
Shatkin, B. 1990. Health care utilization of children of alcoholics. Ph.D. dissertation, Brandeis University.
Spann, L. & Fischer, J.L. 1990. Identifying co-dependency. The Counselor 8:27.
Substance Abuse and Mental Health Services Administration. 1993. National drug and alcoholism treatment unit survey (NDATUS): 1991 main findings report (DHHS Publication No. (SNM) 93-2007). Washington, DC: U.S. Government Printing Office.
Swigonski, M. 1994. The logic of feminist standpoint theory and social work research. Social Work 39: 387-393.
Thorpe, M. 1997. Mental health counselors' diagnoses of female clients on the codependency construct as a function of attitudes toward women. Ph.D. dissertation: University of New Orleans.
Vannicelli, M. 1984a. Treatment outcome of alcoholic women: The state of the art in relation to sex bias and expectancy effects. In Wilsnack, S. & Beckman, L. (Eds.), Alcohol problems in women: Antecedents, consequences, and intervention (pp. 369-412). New York: Guilford.
Vannicelli, M. 1984b. "Treatment outcome of alcoholic women: The state of the art." Bulletin of the Society of Psychologists in Addictive Behaviors 3(3): 155-162.
Vogt. T. 1992. Aging, stress, and illness: Psychobiological linkages. In M. Ory & R. Ables (Eds.), Aging, Health, and Behavior (pp. 207-236). Thousand Oaks, CA: Sage.
Walitzer, K., & Connors, G. 1997. "Gender and treatment of alcohol-related problems." In Wilsnack, R. and Wilsnack, C. (Eds.), Gender and alcohol: Individual and social perspectives (pp. 445-461). Piscataway, N.J.: Rutgers Center of Alcohol Studies.
Weisner, C., & Schmidt, L. 1992. Gender disparities in treatment for alcohol problems. Journal of the American Medical Association 268: 1872-1876.
Wells, M., Glickauf-Hughes, C., & Bruss, K. 1998. The relationship of co-dependency to enduring personality characteristics. Journal of College Student Psychotherapy 12 (3): 25-38.
Wilsnack, S., Klassen, A., Schur, B., & Wilsnack, R. 1991. Predicting onset and chronicity of women's problem drinking: A five-year longitudinal analysis. American Journal of Public Health 81: 305-318.
Windle, M. 1997. Concepts and issues in COA research. Alcohol Health & Research World, 21(3), 185 - 191.
Woodside, M., Coughey, K., & Cohen, R. 1993. Medical costs of children of alcoholics: Pay now or pay later. Journal of Substance Abuse 5: 281-287.
Yoshimura, E. 1991. Illness risk and resistance as a function of codependency, differentiation of self, hardiness, and stress. Ph.D. Dissertation, Arizona State University.
Young, N.K. 1997. Effects of alcohol and other drugs on children. Journal of Psychoactive Drugs 29 (1): 23-42.
1. 1Note. This investigation was funded in part by a faculty research grant awarded to the first author by the College of Social Sciences and Public Affairs, Boise State University.