Telescoping and Gender Differences in Alcohol Dependence: New Evidence From Two National Surveys

In this study we sought to examine evidence for gender differences in the course of alcohol dependence—the telescoping effect—in the general population and evidence for cohort effects on gender differences. We highlight three major findings from these analyses.

First, we found that the full telescoping effect does not appear in the general population. Based on analyses of survival curves as well as mean time to event, we found that women had a later age at initiation of alcohol use (consistent with telescoping) but had neither a shorter time to dependence once alcohol use was initiated nor a shorter time to first treatment after onset of alcohol dependence. In fact, men had a shorter time to dependence in every birth cohort. We examined evidence for these outcomes in five different birth cohorts spanning 40 years and found little evidence of a telescoping effect in any of these cohorts. These results stand in contrast to studies of clinical populations (1–6) and suggest that the telescoping effect disappears when all individuals at risk are considered.

Second, variation by birth cohort in the course of alcohol dependence is evident in these data. Survival analysis showed that later cohorts were more likely to initiate alcohol use (with the exception of the youngest cohort). Decreases in alcohol use initiation from cohort 2 to cohort 1 are consistent with per capita alcohol consumption data in the United States more generally (22), although factors that mediate the decrease (e.g., an increase in the minimum drinking age) remain unclear. Furthermore, the rate of transition to dependence given initiation of alcohol use is increasing in each successive cohort. These findings extend recent work indicating cohort effects in the prevalence of alcohol dependence (12–15), documenting cohort effects in the timing of landmark events, including time to alcohol use initiation and time to alcohol dependence. Third, we found that cohort effects operate differently between genders in the overall hazard of alcohol use and dependence: increases in rates of alcohol use initiation and alcohol dependence between earlier and later birth cohorts were more pronounced in women than in men. While men remain more likely to initiate alcohol use and to develop alcohol dependence compared to women, the gender difference appears to be diminishing in more recent cohorts. Furthermore, survival curves and the direction of gender-specific hazard ratios by cohort suggest that the gender difference in time from use to dependence is actually increasing in more recent cohorts. Men had a shorter time from first drink to dependence compared to both women in more recent cohorts and men in earlier cohorts. This finding is in direct conflict with the telescoping hypothesis, which suggests that women have a faster progression from first drink to dependence.

Gender convergence in the overall hazard of alcohol dependence in recent birth cohorts is consistent with the possibility that earlier gender differences in alcohol problems stemmed from protective social factors (e.g., drinking norms sanctioning female drinking) that have diminished as overall gender norms have shifted (1, 4, 5, 15). While we know of no formal tests of this possibility, research provides evidence consistent with a role for changing norms. For example, studies have documented increases in the acceptability of women drinking and getting drunk (23) and increased exposure to alcohol advertising among adolescent women (24). Furthermore, the World Health Organization World Mental Health Surveys (15) indicate that diminished gender differences in alcohol disorders are most pronounced in countries where traditional gender roles (e.g., women in the workforce) have concurrently changed. Research formally testing mediators of the cohort-by-gender effect is necessary to advance our understanding of the role of changing social factors in the development of alcohol consumption and alcohol disorders. While cohort effects on gender differences in the overall hazard appear to be converging in more recent birth cohorts, the gender difference in time from first use to onset of dependence is becoming wider, with the average time from first use to dependence faster in men than in women, especially in younger cohorts. Further research assessing potential mediators of both time to dependence and overall hazard of dependence is warranted.

We found limited evidence for a cohort effect on time from dependence to treatment and no evidence of a gender-by-cohort or gender-by-cohort-by-time interaction on this outcome. It is likely that forces other than gender norms are important in the delivery of mental health services (such as access through health insurance), which could account for the discrepant findings on this outcome. However, continued monitoring of gender differences in treatment initiation is necessary to observe long-term effects of both the earlier age at initiation of use and the shorter time from use to dependence in more recent cohorts of women. Monitoring is especially important in light of evidence indicating that women with alcohol use disorders have more medical problems compared to men (including higher overall mortality [25], cancer of the liver and breast [26, 27], myocardial infarction [28], and neurological damage [29, 30]).

Our study had several limitations. First, the older participants in these surveys were reporting on the timing of landmark events that may have occurred several years before the survey interview. Retrospective reporting of alcohol history is imperfect (31, 32). However, our research aims could not be addressed without using such data, because existing longitudinal studies have neither the long-term follow-up nor the potential variation across many birth cohorts to observe these effects. In a sensitivity analysis designed to detect potential recall bias, we tested whether the prevalence of alcohol dependence that began at age 25 or later differed between members of any given birth cohort by whether they were interviewed in the NLAES or in the NESARC, when the birth cohort was 10 years older. A lower prevalence of dependence beginning at age 25 or older in members of a given cohort interviewed in the NESARC would suggest the presence of recall bias. In fact, no such differences in prevalence were found, which suggests a limited impact of recall bias on these analyses. Second, while this analysis included over 50,000 individuals, with at least 4,600 in each cohort, treatment utilization is relatively rare in the general population, with 24.6% of alcohol-dependent individuals in these data ever utilizing any type of alcohol treatment (consistent with national estimates [16]). Thus, while these data provide the most comprehensive resource for general population estimates, our study did not have sufficient power to detect small effects regarding cohort differences in treatment. Third, while the NLAES and the NESARC are comparable in sampling frame, study design, and measurement, unmeasured confounding factors between surveys could have introduced bias into the results. To address this possibility, we controlled for the survey in all analyses. Additionally, in post hoc analyses, we examined outcomes by birth cohort and gender in each survey. All findings were similar in direction and magnitude to those presented here, which suggests that minor study design differences did not confound our results.

A greater clinical focus on screening and treatment for alcohol use disorders among young women is warranted given this and other accumulating evidence (12, 13, 15) on the changing course of alcohol use disorders in women. As women increasingly engage in alcohol use earlier in life and exhibit drinking patterns and disorders at ages similar to men, screening for alcohol use disorders among younger women who present for primary care or treatment for non-alcohol-related psychiatric disorders appears to be warranted as an increasingly important component of overall evaluation. In addition, our evidence suggesting that the time from first use to dependence is shorter in men than in women, especially in later birth cohorts, emphasizes the importance of continued alcohol use disorders assessment and intervention in men and the need for additional investigation of this apparent acceleration of the course of alcohol dependence in younger men. Evidence-based guidelines from the National Institute on Alcohol Abuse and Alcoholism for brief alcohol assessment and intervention provide useful information on such screening for clinicians who do not specialize in the treatment of alcohol use disorders (33).

A complete telescoping effect is not apparent in this large general population sample. While women do initiate alcohol use and develop dependence at a later age than men, evidence for a compressed course after drinking initiation is limited. Furthermore, the assumption that alcohol dependence is predominantly a male psychiatric disorder may need to be shifted, as women are increasingly using alcohol and becoming dependent at younger ages. Continued investigation of cohort effects on gender differences across psychiatric outcomes may indicate areas for targeted intervention efforts and assist in the development of a better understanding of the etiology of psychiatric disorders.

The authors thank Wei-Yann Tsai, Mary Beth Terry, and Margaret Keyes for comments on previous drafts of the manuscript.

Supported in part by National Institute on Alcoholism and Alcohol Abuse grant K05 AA014223 to Dr. Hasin; National Institute on Drug Abuse grants RO1 DA018652 to Dr. Hasin, R21 DA020667 and R03 DA023434 to Dr. Martins, and K02 DA023200 and R01 DA018606 to Dr. Blanco; a fellowship from the National Institute on Drug Abuse (F31-DA026689) to Ms. Keyes; and support from the New York State Psychiatric Institute.