The participant numbers in the NIAAA multi-site COMBINE Study (N=1383) offered a unique opportunity to examine the study’s main results for women separately from men. For the present analyses, we performed a focused examination of selected measures obtained from the majority of the COMBINE study’s women (i.e., n=378 women who received study medication out of a total of 428 women in the trial). Men are typically overepresented in alcohol-dependent clinical trials, and so most available treatment study samples are much smaller and usually do not permit a fair analysis of whether women respond to the investigated treatment. In addition, the present analyses generate information on gender differences in response to medications, like naltrexone, that treat alcohol dependence -- a relatively unexplored area, similar to gender response in alcohol-dependent patients taking selective serotonin reuptake inhibitors (SSRIs) (Pettinati, Dundon & Lipkin, 2004
The present study confirmed the primary findings from the original COMBINE Study, i.e., a significant naltrexone by CBI interaction, for women, separately from men, on two primary outcomes (percent days abstinent; time to first heavy drinking day), and three secondary outcome measures (good clinical response, percent heavy drinking days, and craving for alcohol). Essentially, alcohol-dependent women, like alcohol-dependent men, who received medical management plus naltrexone (no CBI specialty therapy), or, medical management and CBI (no naltrexone), had a better treatment response than those taking placebo or any of the other individual or combination treatments that were investigated in the COMBINE study. For example, we found in the present analyses that acamprosate was no more effective than placebo in women, as well as in men – similar to the results for the overall sample in the parent trial. Treatment effects, while modest, were as large (on some measures larger) in women compared to men. The size of the female subsample in the present study provides confidence that alcohol-dependent women are treatment responders to naltrexone or CBI, in the context of medical management, and there are no detectable treatment-response differences between women and men. The finding also suggests that prior studies of naltrexone in women may have been negative due to inadequate sample sizes
In-trial alcohol craving and changes were also comparable between men and women – only naltrexone affected craving (similar to the parent trial results). However, in the overall analysis of in-trial abstinence, there was a main effect of gender where women, compared to men, reported fewer days of in-trial abstinence, or more days of drinking. This finding is most obvious when comparing the women and men not taking active medication (i.e., receiving placebo tablets), and least obvious when comparing the women and men in specialty therapy (CBI). Because abstinence is the only outcome measure where we found that women responded more poorly than men, this result requires replication, and it would be interesting to investigate this further in future studies. Furthermore, the combined condition, CBI plus naltrexone, had intermediate effect depending on outcome considered for reasons that are not understood at this time. Future studies might consider examining whether outcomes vary as a function of gender and treatment condition depending on whether the endpoint measured is abstinence, reduction in heavy drinking days, or some measure of good clinical response.
In addition to demonstrating that in the COMBINE study, women responded to alcohol treatment like the men, we found evidence of many similarities between these women versus men on pre-treatment as well as in-trial measures. In general, women and men had comparable demographics, pretreatment clinical characteristics, and rates of retention and medication adherence. With regard to the few significant differences between men and women, women had more years of education than men, generally considered a good prognostic indicator (Greenfield et al., 2003
). In addition, women reported a later age of onset of alcohol dependence by an average of approximately 3 years than men, and they reported drinking significantly fewer drinks per drinking day than did men consistent with previous literature on this topic [Randall et al, 1999]. In spite of alcohol severity similar to men, women in this study were less likely to have received alcohol treatment in the past. Studies of community samples have shown that women are less likely than men to receive any treatment for their alcohol problems over their lifetime (Dawson, 1996). Some previous studies with clinical samples have indicated that fewer women with alcohol dependence entered treatment than men (Schober & Annis, 1996; Weisner, Greenfield, & Room, 1995), while a more recent study indicated similar rates of treatment entry for both men and women (Weisner, Mertens, Tam, & Moore, 2001). Barriers to obtaining treatment for alcohol dependence for women remain and are well documented (Greenfield et al., 2007
). In the COMBINE study sample, more women than men were obtaining their first alcohol treatment through participation in a clinical research study. It is unclear if this may be due to women’s perceiving lower stigma for entering a research trial for treatment of alcohol dependence, or experiencing fewer barriers (e.g., cost of treatment, inadequate or lack of insurance, among others) to entering the RCT than other clinical treatment options. Once enrolled, women had comparable levels of attendance in MM and medication adherence and somewhat higher rates of completing CBI compared to men.
The present study has several limitations, in addition to those reported by Anton et al, 2003. Most notably, individuals taking antidepressants, anxiolytics and other psychotropic medications were excluded from the COMBINE parent trial; and women, compared to men, are more likely to be taking these medications because of co-occurring depression, anxiety, etc. (Grant et al, 2004
). It is debatable whether the finding of many similarities between women and men on pre-treatment and in-trial measures, including outcome results, might have been different if we had included individuals with co-occurring illnesses or those who were also taking psychotropic medications. The literature has been inconsistent with regard to whether co-occurring psychiatric disorders are associated with poorer, better, or even similar treatment outcomes compared to alcohol-dependent individuals without co-occurring disorders (see review by Pettinati & Plebani, 2009
). Thus, it is difficult to infer whether women with co-occurring anxiety or depressive disorders and alcohol dependence would have had a differential response to naltrexone or CBI from alcohol dependent men with or without these disorders. Individuals with co-occurring drug dependence were also excluded from the COMBINE trial. In this case, poorer outcomes have been consistently associated with co-occurring drug dependence and abuse (Heil et al., 2001
; Pettinati et al., 2008
). Another limitation is that clinical trials tend to include very traditional pre-treatment and outcome measures, which might not be gender sensitive, leading to finding no gender differences (e.g., no measures were included of childrearing responsibilities). Finally, our analysis focused on the 16 week- treatment period. It is conceivable that gender differences might emerge later given that response to cognitive behavioral treatment may sometimes occur at a time more distal to treatment, the so-called sleeper effect (Carroll, 1994
In summary, the purpose of the present study was to focus on the treatment implications of the large, multicenter NIAAA-sponsored COMBINE Study results for women with alcohol dependence, given that this study provides a large subgroup of alcohol-dependent women where results are more likely to be replicable. This gender-focused analysis found that, despite modest effect sizes, the findings from the original COMBINE study were supported for women as well as for men. In particular, the alcohol-dependent women responded to naltrexone with medical management, similar to the alcohol-dependent men, on a wide range of outcome measures. These results suggest that clinicians can feel reasonably comfortable prescribing naltrexone for alcohol dependence in both men and women. Also, as discovered in the parent trial, the effects of naltrexone were most apparent in the context of a medical model of counseling that can be delivered in primary health care settings (COMBINE’s Medical Management). In this study, it is also notable that fewer women than men reported receiving any alcohol treatment prior to entry into the COMBINE study. Of note, women tend to go to primary health care more frequently than to specialty substance abuse programs for treatment (Weisner & Schmidt, 1992
), and so the benefit we confirm for women of the naltrexone and medical management combination has practical implications for treating alcohol-dependent women.