Correlates of alcohol abuse/dependence in early-onset alcohol-using women

Abstract

Background

Early-onset alcohol use is associated with increased vulnerability to subsequent alcohol abuse and dependence. However, not all early-onset alcohol users develop alcohol use disorders (AUDs). Using a sample of young women from the U.S., we identify correlates that contribute to a greater likelihood of AUDs in early-onset alcohol users.

Methods

Using interview and questionnaire data on participants of the Missouri Adolescent Female Twin Study (MOAFTS), we examine whether measures from domains including socio-demographic, pubertal development, religiosity, educational achievement, adverse life events, internalizing disorders, externalizing disorders and family history and discipline were associated with development of AUDs in 1,158 women who had their first drink of alcohol prior to age 16.

Results

Early-onset drinkers were 3.6 times more likely to meet criteria for AUDs than later onset drinkers. While univariate analyses revealed that a host of correlates were associated with likelihood of AUDs in early-onset drinkers, multivariate analyses suggested that, even after accounting for a particularly early age of onset of drinking, those with a history of physical abuse, co-twin alcohol problems, conduct disorder, regular smoking, older peers and peer substance use were considerably more likely to meet criteria for AUDs than early onset drinkers without a lifetime history of these correlates.

Conclusion

The progression from first drink to AUDs is complex, and while early age at first drink is a potent risk factor, other aspects of psychopathology, family history, conduct problems and peer affiliations can exacerbate or alleviate the risk of AUDs in these young female drinkers.

Introduction

There is consistent evidence that early onset alcohol use is associated with increased risks for the subsequent development of a range of alcohol related problems and alcohol dependence (1–8). For example, Hingson et al (6) reported a steady increase in the prevalence of lifetime alcohol dependence with earlier age of onset of alcohol use: 47% of those reporting onset before age 14 met lifetime criteria for alcohol dependence, compared with only 9% of those initiating alcohol use after the age of 21. While similar associations have been replicated across a range of samples and measures of alcohol-related harms including heavy and prolonged drinking (9–12) and risky behavior while intoxicated (13–18) controversy remains regarding the processes underlying these associations, with the literature supporting the role of correlated genetic and environmental influences that both encourage the early onset of alcohol use and heighten risks for the development of alcohol related problems (19).

Irrespective of the mechanisms underlying such associations it is important to note that many of those who report early onset use do not progress to alcohol use disorders (AUDs). For example, although Hingson et al (6) reported a strong and consistent association between age of onset of alcohol use and risks of lifetime alcohol dependence, 53% of those reporting onset before age 14 (the youngest age of onset studied) did not, in fact, meet criteria for alcohol dependence. Therefore, while early onset alcohol use is an important marker for the development of AUDs it is by no means the case that all – or even a majority – of those commencing alcohol use at a young age progress to AUDs, raising the important issue of what distinguishes between early onset alcohol users who do versus do not subsequently develop AUDs.

In the current paper we address this issue by examining the factors that are associated with alcohol use disorders (abuse/dependence) in those who use alcohol at an early age in a sample of adolescent female twins studied as part of an ongoing longitudinal study.

Materials and Methods

Sample

Data for this study are drawn from the Missouri Adolescent Female Twin Study (MOAFTS). MOAFTS consists of a cohort of female same-sex twin pairs born between July 1^st 1975 – June 30^th 1985 who were identified from birth records (20). Twins were eligible to participate if both members of the twin pair had survived past infancy, were not adopted at birth and if their biological parents were residents of the state at the time of their birth. Further details regarding sample recruitment and characteristics of this first wave of interview data, are given elsewhere (21–22). The current study uses data on 2,856 women who participated in the baseline interview (12–23 years of age) and in the five-year follow-up interview (18–29 years of age - 82.8% of women interviewed at baseline) An additional 109 women who participated in the two-year test-retest study but not in the five-year follow-up were also included. These subjects had lifetime reports of AUDs, however their inclusion (or exclusion) was not found to have noticeable impact on substantive findings. Of the final sample of 2,965 twins, 11.8% of these twins were of African-American ethnicity, the remainder being of European-American descent. 2,582 (87.1%) of the twins reported a lifetime history of alcohol use at any interview assessment.

Measures

Age of onset of Alcohol Use

Twins were asked to recall how old they were when they consumed their first full drink of alcohol. For those twins who had initiated alcohol use before the baseline interview, we used their age of onset reported at baseline. For twins who initiated their alcohol use between the baseline and the five-year or test-retest follow-up, their response from the follow-up waves was used.

Age of onset of alcohol use ranged between 2–25 years, with a median of 16 years. Thus, in order to investigate the action of risk and protective influences on alcohol abuse/dependence on early onset drinkers, we only retained 1,158 women who reported consuming their first full drink of alcohol prior to age 16.

Alcohol use disorders (AUDs)

DSM-IV criteria were used to define alcohol abuse (endorsement of 1 or more symptoms) and dependence (endorsement of 3 or more symptoms). A dichotomous measure was used to define AUDs (either abuse or dependence) with those who had consumed one full drink of alcohol but not meeting criteria for either being coded as ‘0’. 21.3% of alcohol-using twins met criteria for a lifetime history of alcohol abuse/dependence (15.2% abuse and 13.6% dependence).

Correlates of AUDs

We also included a range of potential correlates of AUDs among those reporting early onset of alcohol use. These were selected on the basis of a review of the literature on risk and protective factors for alcohol abuse/dependence and on their availability within the data set and are described in Table 1.Where age at onset was available for the covariate, it was used to ensure that onset was prior to 16 years.

TABLE 1.

Univariate association between correlates and lifetime history of DSM-IV alcohol abuse/dependence in 1,158 young women who reported drinking their first alcohol drinking prior to age 16.

	DSM-IV alcohol abuse/dependence
	Present (N=381)	Absent (N=777)	Univariate Odds-ratio [95% C.I.]	Source
Socio-demographic
African-American ethnicity	5.3	9.9	0.5 [0.3–1.2]	B
Pubertal Development
Onset of Menses   - Prior to age 13	50.0	49.8	1.0 [0.8–1.3]	I2
Self-reported early maturation   - Matured physically earlier than most girls	24.9	20.5	1.3 [0.9–1.8]	B
Religiosity
Self reported importance of God/church   - Bottom quartile of sum score on 4 items assessing self-reported importance of religion [not important, somewhat important and very important]	36.6	34.1	1.1 [0.8–1.5]	Q1
Educational attainment
Self-reported low grades in the past year    - Got mostly C’s and D’s	10.0	8.0	1.3 [0.8–2.0]	B
Adverse Life Events
Any one traumatic event    - Experiencing a traumatic event prior to age 16	35.2	27.2	1.4 [1.1–1.9]*	I2
Child Sexual Abuse   - Forced sexual contact prior to age 16	10.7	7.6	1.5 [0.9–2.3]	I2
Physical Abuse   - Physically hurt or injured on purpose by an adult prior to age 16	10.2	4.0	2.7 [1.6–4.5]*$	I2
Suicidality
Suicidal Ideation prior to age 16	31.2	21.4	1.7 [1.3–2.2]*	B, SF & I2
Suicide Attempt prior to age 16	10.8	4.6	2.5 [1.6–4.0]*$	B, SF & I2
Internalizing Disorders
Depressed, irritable mood or loss of interest prior to age 16	48.6	41.2	1.3 [1.1–1.7]*	B, SF & I2
Panic attack   - Reporting 4 or more panic-related symptoms prior to age 16	19.2	12.3	1.7 [1.2–2.4]*$	I2
Externalizing Disorders
Very early onset of alcohol use   - Onset prior to age 12 vs 12–15	18.4	14.9	1.3 [0.9–1.8]	B, SF & I2
DSM-IV Conduct Disorder	13.1	3.9	3.8 [2.4–6.1]*$	B, SF & I2
Cannabis Use    - Lifetime use with initiation prior at age 16	43.0	26.0	2.2 [1.7–2.8]*	SF, I2
Peer Substance Involvement   - Top quartile of sum score of 9 self-reported items on how many peers used alcohol, cigarettes and illicit drugs [None – All]	69.4	45.5	2.7 [2.0–3.6]*$	Q1
Early sexual debut   - Consensual intercourse prior to age 16	43.7	29.8	1.8 [1.4–2.4]*	I2
Regular smoker (<16y)   - Smoked 100 or more cigarettes or smoked 21–99 cigarettes but as frequently as two days a week for 3 weeks or longer	37.3	20.3	2.3 [1.8–3.1]*$	B, SF & I2
Self Reported Older Friends   - Having more older than younger/same-aged peers	30.5	17.3	2.1 [1.6–2.8]*$	I1
Family History & Discipline
Low Parental Monitoring   - Top quartile of sum score of 10 items assessing parental monitoring [Never – Always]	44.0	31.5	1.7 [1.3–2.2]*$	Q2
Respondent report of co-twin’s alcohol problem	21.7	8.8	2.9 [2.0–4.1]*$	I2
Respondent report of either parent having an alcohol problem	43.4	31.4	1.7 [1.3–2.2]*$	I2

Legend to Table 1: B=baseline interview; SF=subset of 1370 twins at 2 year follow-up; Q1=first mailed questionnaire; I2=five-year follow-up interview; Q2=second mailed questionnaire.

^*=statistically significant at p-value of 0.05 or less.

^$=variables that remained significant for each category in multivariate models, with stepwise selection (entry p-value=0.1, stay p-value=0.05).

Analytic Strategy

To examine potential correlates for the development of AUDs among early onset alcohol users, we selected only those reporting onset before age 16 (N=1,158) and conducted a series of univariate and multivariate logistic regression analyses comparing the associations between each of the correlates and likelihood of AUDs in this subset of early-onset users. First, multivariate models that utilized measures in each category were independently tested, while controlling for ethnicity and very early alcohol use. Selection of variables from each category was conducted using a stepwise approach with an entry p-value of 0.10 and a ‘stay’ p-value of 0.05. Those variables that survived multivariate testing (p-value of 0.05 for less) within each category were then combined in a second multivariate model, also controlling for ethnicity and very early alcohol use, and stepwise selection was used to retain significant correlates of AUDs in these early onset drinkers.

Results

The Association between Early onset Alcohol use and Alcohol use disorders

The median age for initiation of alcohol use was 16 years [mean of 15.8, SD=2.6 years]. Of those who reported using alcohol for the first time prior to age 16, 32.9% subsequently met criteria for AUDs – this was in contrast to 12% of later-onset users who developed AUDs. Thus, early-onset users were 3.6 [95% C.I. 2.9–4.4] times more likely than later-onset users to subsequently develop AUDs. As shown in Figure 1, those initiating use of alcohol between the ages of 12–15 years were most (rates of AUDs as high as 36–43%) likely to meet criteria for lifetime AUDs, which reinforces our selection of onset of use prior to 16 years for this study.

FIGURE 1.

Percentage of women (N=2,577) meeting criteria for DSM-IV alcohol abuse/dependence classified by the age at which they consumed their first alcoholic drink. The percentages are not cumulative across the x-axis and are standardized to each age-band. For instance, 33% of those initiating alcohol use at age 11 years subsequently met criteria for DSM-IV alcohol abuse/dependence – likewise, 8% of those with an age at first drink of 18 years subsequently met criteria for alcohol abuse/dependence,

Correlates of lifetime alcohol abuse/ dependence in early onset alcohol users

Univariate logistic regression analyses revealed that a host of factors were correlated with an increased likelihood of lifetime AUDs in early-onset alcohol users [Table 1]. Early-onset users with AUDs were more likely to report experiencing trauma and physical abuse, suicidal ideation and attempt as well as a history of depressed mood and panic attacks. All aspects of externalizing problems, particularly conduct disorder, were strongly associated with risk for AUDs in early-onset users as was low parental monitoring and a family history of alcohol problems.

In the first series of multivariate analyses, which were conducted with variables from each category, several variables remained statistically significant at p < 0.05. These included: Physical Abuse (Adverse Life Events), Suicidal attempt (Suicidality), Panic attacks (Internalizing Disorders), Conduct Disorder, Regular smoking, Peer substance use and having older friends (Externalizing Disorders) and all variables comprising the Family History and Discipline category (low parental monitoring, respondent report of co-twin and of parents having an alcohol problem). When these were entered into a second multivariate model [Table 2], stepwise selection retained respondent report of co-twin having an alcohol problem, physical abuse and all four externalizing indices (conduct disorder, regular smoking, peer substance use and older peers) as significant covariates of AUDs in early onset users, even while controlling for very early onset of alcohol use. The most significant contribution to the total R² (19%) was from peer substance use (8%) followed by self-reported co-twin alcohol problems (4%).

TABLE 2.

Results from the final multivariate model examining the association between covariates and DSM-IV alcohol abuse/dependence in twin women.

	Odds-ratio [95% C.I.]	Order of selection (R2**)
Adverse Life Events
Physical abuse	2.49 [1.46–5.35]*	#8 (0.19)
Suicidality
Suicide Attempt prior to age 16	1.41 [0.70–2.85]	-
Internalizing Disorders
Panic attack	1.26 [0.78–2.03]	-
Externalizing Disorders
DSM-IV Conduct Disorder	2.29 [1.16–4.56]*	#5 (0.18)
Peer substance involvement	2.13 [1.13–2.98]*	#1 (0.08)#
Regular Smoking	1.52 [1.04–2.23]*	#4 (0.16)
Self-reported older friends	1.88 [1.27–2.78]*	#3 (0.14)
Family History & Discipline
Low Parental Monitoring	1.29 [0.91–1.81]	-
Respondent report of co-twin’s alcohol problem	1.19 [0.85–1.87]	-
Respondent report of either parent having an alcohol problem	2.14 [1.36–3.38]*	#2 (.12)

NOTE: Adjusted for very early onset and African-American ancestry.

^*=statistically significant at p-value of 0.05 or less in the multivariate model.

^**Max-rescaled Rsq upon addition of each variable in a stepwise manner. The Rsq accumulates with the addition of each covariate.

^#The 1^st variable selected by the stepwise model is peer substance involvement – the Rsq (8%) associated with this model includes very early onset and African-American ethnicity – however, these covariates alone contribute to less than 1% of the variance in AUDs.

Discussion

In these analyses we have explored the factors that distinguish between early onset alcohol users who do or do not develop AUDs. In confirmation of a large number of previous studies (3, 4, 6) we found that earlier age of onset of alcohol use was strongly associated with later risks of dependence (OR = 3.6). Nonetheless, not all those who report early onset of alcohol use developed AUDs: among those reporting initiating alcohol use before age 16 (44.9% of the sample), the majority (67.1%) did not develop an AUD. Among these early onset alcohol users, factors distinguishing those who developed AUDs from those who did not included: family history of alcohol problems, physical abuse and a host of externalizing problems.

Some limitations are noteworthy. First, this is a sample of women and results may not generalize to other populations. Second, some recall bias may have influenced self-reported age at onset – we elected to use the first reported age at initiation. This only influenced the responses of 89 subjects who reported a lower age of initiation at the follow-up interview compared to baseline, with a majority of the discrepancies attributable to a 1–2 year difference in reports. Third, even though we only included data on correlates with onsets prior to age 16, no causal conclusions may be drawn from these analyses as it is possible that the onsets of some correlates were subsequent to the onset of alcohol use (e.g. onset of alcohol use at age 12 and onset of cannabis use at age 14). Onset of all correlates, however, preceded onset of AUDs. Finally, to some extent, duration of alcohol exposure may also have contributed to higher rates of AUDs, even in early-onset users. However, this is likely to have smaller influence in the subset selected for this study (i.e. as all subjects are early-onset users).

This was a study of ‘risk’ factors. It may be argued that multiple covariates tested here, such as low parental monitoring, low religiosity and low educational attainment could be reverse-coded to reflect protective influences, and as such more emphasis of protective influences is needed. Future studies may wish to consider aspects of academic achievement and prosocial activities (e.g. volunteering) (23–25) as well as positive home environments as potential influences that ameliorate the risks associated with early onset alcohol use. Stability in romantic relationships has also been found, in young adults, to be associated with reduced substance involvement (26).

Familial history of alcohol problems, physical abuse and externalizing problems, including the clear and important role of peers, predicted the development of AUDs in these early onset users. Coupled with previous findings that the effects of risk are likely to be cumulative, these results argue for targeted approaches that focus on those most at risk by virtue of exposure to a range of risk factors (e.g. individuals exposed to adverse neighborhoods). This will be challenging – while prevention efforts targeted at communities may reduce access to deviant peer cohorts, peer selection is partially genetically driven (27). This phenomenon, whereby individuals select environments that are correlated with their own genetic predisposition, implies that arming adolescents and young adults with better coping strategies may serve as a more efficient intervention strategy than attempt the modification of their social milieus.