Male sex has also been identified as a risk factor for early onset of drinking (Grucza et al., 2008b; Keyes et al., 2010; Schuckit et al., 1998) and the development of alcohol use disorders (Hasin et al., 2007; Keyes et al., 2008; Rice et al., 2003). Analyses of more than 6,000 respondents between 15 and 44 years of age in the National Comorbidity Survey found that while the number of men and women alcohol drinkers was similar, there were almost twice as many cases of alcohol dependence among men (Wagner and Anthony, 2007). Further, the same study found that the cumulative probability of alcohol dependence ten years after initiating alcohol use was approximately three times higher for men. Lopez-Quintero et al. (in press) found that men were more likely than women to transition from alcohol use to dependence.

Research on the progression from first alcohol use to the development of alcohol use disorders has shown a sex difference, with women moving faster than men from the initiation of drinking to the onset of first alcohol-related problems, alcohol dependence and treatment entry (Hernandez-Avila et al., 2004; Piazza et al., 1989; Randall et al., 1999; Schuckit et al., 1998). While this “telescoping” effect has been demonstrated in several samples, it seems to be less common in younger age cohorts (Johnson et al., 2005; Randall et al., 1999). However, the majority of studies examining telescoping were conducted in treatment samples, limiting the generalizability of findings. A recent study from a general population sample did not find that women progressed from drinking onset to alcohol dependence faster than men; in fact, it found the opposite, with men demonstrating a faster transition (Keyes et al., 2010).

Despite at least two decades of research on telescoping, there is a paucity of data on race/ethnicity differences in the progression from drinking onset to development of alcohol dependence. One study of participants recruited from substance abuse treatment programs included in the National Drug and Alcoholism Treatment Unit Survey (NDATUS) database found that while Blacks were older than Whites and Hispanics at onset of first regular alcohol use, both Blacks and Hispanics progressed from first regular use to intoxication faster than Whites (Johnson et al., 2005). However, this study did not examine time to development of alcohol dependence. Less is known about whether differences in progression vary when examining race/ethnicity subgroups by sex.

First, similar to other recent studies conducted in general population samples, we found no evidence for a telescoping effect for women (Keyes et al., 2010; Lopez-Quintero et al. in press; Wagner and Anthony, 2007). While White, Black, and Hispanic women in our study were older than men of the same race/ethnicity at onset of alcohol use, they did not progress to dependence in a shorter time period than men. These results are in concert with a recent study by Keyes and colleagues (2010) that combined data from both the NESARC and the National Longitudinal Alcohol Epidemiologic Survey, which was conducted ten years earlier. In that study, not only was there an absence of a telescoping effect for women, but men progressed to dependence at a faster rate than women. In our sample, there were no sex differences in time to progression to alcohol dependence for Whites, Blacks or Hispanics. One potential explanation for their finding of a faster transition to dependence for men is the fact that the Keyes study did not include alcohol consumption as a covariate. In our study, the adjusted model excluding the consumption variable (not shown) found that White men progressed to dependence faster than White women only; however, with the addition of alcohol consumption as a covariate there was no difference. Thus, the greater consumption typical for men may account for sex differences in progression to alcohol dependence for Whites at least.

Our lack of a telescoping effect for women is in contrast to the many earlier studies (Hernandez-Avila et al., 2004; Johnson et al., 2005; Piazza et al., 1989; Randall et al., 1999), which were conducted in treatment samples and thus less able to generalize to the general population. By analyzing data from a population-based sample, consisting of individuals with and without alcohol dependence, and by removing the potential selection biases of treatment populations, our findings are expected to represent the course from first alcohol use to dependence in the general population of the United States.

It is also possible that our results may be reflective of secular trends in rates of alcohol use and use disorders such that while telescoping may have at one time existed in the general population it is no longer evident. Several recent studies have demonstrated an increase in the prevalence of both alcohol use and alcohol dependence by birth cohort, with higher rates in younger cohorts compared to older cohorts (Grucza et al., 2008a; Keyes et al., 2008; Keyes et al., 2010; Rice et al., 2003). This increase is most prominent in women and has resulted in a progressive decrease in the odds ratio of dependence between men and women (Grucza et al., 2008a; Keyes et al., 2008). Further, younger cohorts have an earlier age of drinking onset than older cohorts, but the net decrease in age at drinking initiation is twice as large for women (Grucza et al., 2008b). Thus, it is possible that the absence of a telescoping effect may be reflective of the restriction of our analyses to a younger sample (age < 45 years old). However, as discussed earlier, restricting our sample to younger participants was important to decrease both the potential for recall bias and differential alcohol-related mortality (Keyes et al., 2010; Wagner and Anthony, 2007).

Second, we found notable differences by race/ethnicity. In general, White men and women had a younger mean age at onset of drinking, younger mean age of onset of alcohol dependence, and faster progression from drinking initiation to dependence than Black and Hispanic men and women, respectively. These findings are consistent with other studies showing increased odds of ever using alcohol, earlier age of regular alcohol use, and higher prevalence of alcohol dependence in Whites when compared to Blacks or Hispanics (Hasin et al., 2007; Johnson et al., 2005; Kalaydjian et al., 2009). However, few studies have explored race/ethnicity differences in progression from drinking initiation to development of alcohol dependence. In a recent study, using both Waves of the NESARC, Lopez-Quintero and colleagues (in press) found that White alcohol users were more likely to transition to alcohol dependence than Black users but found no significant difference in the hazard of transitioning to dependence between Whites and Hispanics. Our study examines race/ethnicity differences in transition from use to dependence in a general population sample and builds on the existing literature by examining these differences, stratifying race/ethnicity groups by sex. While our results for Blacks are consistent with the Lopez-Quintero study, our findings for Hispanics differ. These differences may be explained by differences in sampling strategy, including the restriction of their sample to persons who participated in both Waves 1 and 2 and the restriction of our sample to Wave 1 participants younger than 45 years of age. Additionally, our studies used different covariates in the adjusted models.

The etiology of these race/ethnicity differences is not well characterized and additional research is needed to identify the characteristics elevating the risks of earlier drinking initiation and faster transition to dependence for Whites or those attenuating the risks for other groups. Some studies suggest that these racial and ethnic disparities may be partially explained by differences in social and cultural norms. A study by Caetano and Clark (1999) found that both Black and Hispanics reported more conservative alcohol norms and attitudes when compared to Whites. Additionally, multiple studies have found that religiosity is higher among Blacks than Whites (Brown et al., 2001; Donahue and Benson, 1995; Neff and Hoppe, 1993; Taylor et al., 1999). Religiosity has consistently been shown to have an inverse relationship with alcohol use, with those reporting higher levels of religiosity being less likely to initiate alcohol use, consuming fewer drinks per occasion, and reporting fewer alcohol-related problems (Benda, 1997; Brown et al., 2001; Donahue and Benson, 1995; Patock-Peckham et al., 1998). For Hispanics in the United States, acculturation has been identified as a risk factor. A greater degree of acculturation has been associated with increased risk of drinking, alcohol-related problems, and alcohol dependence (Black and Markides, 1993; Caetano et al., 2009; Gil et al., 2000). In addition to sociocultural factors, at least one biological factor has been identified. Up to one quarter of persons of African descent have a variant of the alcohol dehydrogenase enzyme, ADH1B*3, which has been found to be protective against alcohol dependence and alcohol-related birth defects (Scott and Taylor, 2007). This variant of the enzyme is generally not found in Whites.

Our findings must be interpreted with regard to several limitations. First, all data collected was via self-report, which can raise concerns about the validity of the results. In particular, study participants may be reporting on events that happened many years earlier, which introduces the possibility of recall bias. Recall bias threatens the validity of inferences drawn from observational studies, and is an especially vexing problem in cross-sectional studies, when the exact timing of events under study may be unknown. We attempted to mitigate this risk by restricting our sample to persons less than 45 years of age (Keyes et al., 2010; Wagner and Anthony, 2007). While this strategy may have reduced the potential for recall bias, our results are less generalizable to older populations and may not have exhausted the threat of recall bias entirely. Second, reporting bias may also be a factor particularly when participants are asked to report on potentially stigmatizing behaviors such as underage drinking or excessive alcohol use. However, there is no reason to suspect that recall or reporting bias varies by race or ethnicity and thus should not influence major findings on racial or ethnic differences. Further, there is some evidence for the reliability and stability of retrospectively recalled history of substance involvement and age of first use (Koenig et al., 2009; Labouvie et al., 1997; Prause et al., 2007). Third, the coding scheme of the NESARC does not allow for an exploration of potential differences in persons of mixed race, and this must be considered when interpreting these results. Fourth, due to the limited number of participants of other races, we were unable to examine relationships by sex for such racial subgroups as Asian or Native Americans. Finally, other important variables were not measured or could not be included in the models, thus leaving room for some unaccounted rival explanations (e.g., religiosity, family and/or cultural norms about alcohol use, and neighborhood characteristics, among others). The present study may help motivate new studies that can address these alternative explanations.