Journal of Anxiety Disorders 31 (2015) 108–113
Contents lists available at ScienceDirect
Journal of Anxiety Disorders
Anxiety disorders and first alcohol use in the general population. Findings from a nationally representative sample Louise Birrell ∗ , Nicola C. Newton, Maree Teesson, Zoe Tonks, Tim Slade NHMRC Centre of Research Excellence in Mental Health and Substance Use (CREMS), National Drug and Alcohol Research Centre (NDARC), University of New South Wales, Sydney 2052, NSW, Australia
a r t i c l e
i n f o
Article history: Received 9 December 2014 Received in revised form 19 February 2015 Accepted 24 February 2015 Available online 6 March 2015 Keywords: Anxiety Substance use Epidemiology Alcohol drinking Age of onset Adolescent Age factors Age of first drink
a b s t r a c t Objective: To examine how early onset anxiety disorders are related to age of first alcohol use in a general population sample. Method: Discrete time survival analysis was used to model the odds of first alcohol use among those with, vs without, early onset anxiety disorders. Data came from the 2007 Australian National Survey of Mental Health and Wellbeing. Results: After adjusting for the effects of family history of alcohol/drug use, sex, age cohort and education, people who experienced an early onset anxiety disorder had a 27% increased odds of first alcohol use in any given year, when compared to those with no anxiety disorder. This effect was particularly strong for transitions to first alcohol use that occurred after the age of 13 years. Conclusions: Early onset anxiety disorders significantly predict first alcohol use in the general population and this relationship appears to be related to change over time. These results point to the need for developmentally appropriate and integrated prevention programs that target anxiety and alcohol use together. © 2015 Elsevier Ltd. All rights reserved.
1. Introduction The most common mental health and substance use disorders in the general population are anxiety (Kessler et al., 2012a) and alcohol use disorders (Teesson et al., 2010). Epidemiological estimates in community samples approximate that between 20% (McEvoy, Grove, & Slade, 2011) and 31% (Kessler et al., 2007a) of the population in Western countries will suffer from an anxiety disorder during their lifetime, and between 19% (Kessler et al., 2005) and 22% (Teesson et al., 2010) will meet criteria for an alcohol use disorder. These two classes of disorders are also highly comorbid (McEvoy et al., 2011; Teesson et al., 2005). Individuals with an alcohol use disorder are 2.6 times more likely to have a comorbid anxiety disorder during their lifetime, than those without an alcohol use disorder (Teesson et al., 2010). People with comorbid mental health and substance use problems are known to be a particularly high risk group that are more difficult to treat, place a greater burden on service providers and have a more severe illness course (de Graaf, Bijl, ten Have, Beekman, & Vollebergh, 2004; Teesson & Proudfoot, 2003).
∗ Corresponding author. Tel.: +61 2 9385 0333. E-mail addresses: [email protected], [email protected] (L. Birrell), [email protected] (N.C. Newton), [email protected] (M. Teesson), [email protected] (Z. Tonks), [email protected] (T. Slade). http://dx.doi.org/10.1016/j.janxdis.2015.02.008 0887-6185/© 2015 Elsevier Ltd. All rights reserved.
To help understand this comorbidity it is useful to look at its origins and examine how anxiety disorders relate to age of first alcohol use. Age of first alcohol use is an important outcome to examine in understanding the etiology and temporal sequencing of anxiety and alcohol use comorbidity. There is robust evidence demonstrating an association between an early age of first alcohol use and later alcohol-related problems and alcohol use disorders (Dawson, Goldstein, Chou, Ruan, & Grant, 2008; DeWit, Adlaf, Offord, & Ogborne, 2000; Fergusson, Horwood, & Lynskey, 1994; Grant & Dawson, 1997; Grant, Stinson, & Harford, 2001; Gruber, DiClements, Anderson, & Lodico, 1996; Hawkins et al., 1997; Liang & Chikritzhs, 2011). In a large US study, Grant et al. (2001) followed over 5000 young people aged between 14 and 21 years old for 12 years and found that the odds of alcohol dependence was decreased by 9% for each year that first alcohol use was delayed. Debate exists about whether evidence such as this demonstrates a causal relationship between age of first alcohol use and later alcohol use problems (Deutsch et al., 2013; Irons, Iacono, & McGue, 2014) or not (Geels et al., 2013; McGue, Iacono, Legrand, Malone, & Elkins, 2001; Prescott & Kendler, 1999), and as yet there is no definitive consensus about the mechanisms underlying these relationships. A number of key variables are known to play a significant role in the relationship between age of first alcohol use and later alcohol problems, including psychopathology and anxiety disorders (Pang, Farrahi, Glazier, Sussman, & Leventhal, 2014;
L. Birrell et al. / Journal of Anxiety Disorders 31 (2015) 108–113
Sartor, Lynskey, Heath, Jacob, & True, 2007). However, the majority of studies examining the role of anxiety disorders and alcohol use focus on regular alcohol use and alcohol use disorders as the key outcomes of interest (Behrendt et al., 2011). Particular attention has been given to the relationship between anxiety disorders and an increased speed of transition from first alcohol use to alcohol dependence (Behrendt et al., 2011; Lopez-Quintero et al., 2011). Fewer studies focus on first alcohol use as the primary outcome, and even fewer examine the temporal sequencing of anxiety disorders in relation to first alcohol use. This study will focus on first alcohol use as the key outcome. This is warranted, as evidence suggests that risk factors play different roles in the transition to regular alcohol use and alcohol use disorders, than in the transition to first alcohol use (Sartor et al., 2007). Furthermore, previous studies have not differentiated between the individual types of anxiety disorders, which may show differential relationships to alcohol use. The current study will address these gaps. The temporal sequencing of first alcohol use and anxiety disorders centers around the developmental period of adolescence. In most Western countries, adolescence is a time when the majority of individuals consume their first alcoholic drink. It is also the time when a large number of the anxiety disorders have their onset (Falk, Yi, & Hilton, 2008), making this a particular developmental period of interest to examine the comorbidity of anxiety and alcohol use. More specifically, anxiety disorders that have an early onset have been linked to reports of greater anxiety severity, poor treatment response, more significant functional impairment and early depression (Dalrymple & Zimmerman, 2011; Rosellini, Rutter, Bourgeois, Emmert-Aronson, & Brown, 2013). Adolescence is also a distinct time of neuro-development, with key brain structures involved in emotion regulation and behavioral planning (such as the frontal lobes, temporal lobe, amygdala and hippocampus) undergoing dynamic change (Clark, Thatcher, & Tapert, 2008). This leaves adolescents particularly susceptible to risky behaviors, poor decision making and lack of emotion regulation. Furthermore, alcohol is known to inhibit behavioral inhibition and is likely to compound this lack of executive brain function. A recent review of brain imaging studies on the effect of alcohol consumption on the adolescent brain concluded that adolescents who frequently use alcohol show a different pattern of brain structure and function, compared to those who have not yet tried alcohol (Feldstein Ewing, Sakhardande, & Blakemore, 2014). Together this evidence highlights the need for further research around alcohol use at this critical life stage. To understand why anxiety and alcohol use problems occur together it is important to explore their temporal sequencing over both; (a) the lifespan, and (b) across the different stages of alcohol use. This study focuses on first use of alcohol as a logical starting point in the etiology of anxiety and alcohol disorder comorbidity. Specifically, the present study aimed to examine the temporal sequencing of early onset anxiety disorders in relation to age of first alcohol use in a general population sample. It was hypothesized that those who experienced an early onset anxiety disorder would be at increased odds of initiating drinking in any given year. This study will be the first to examine how early onset anxiety disorders relate to first use of alcohol use, using a nationally representative general population sample.
2. Methods and materials 2.1. Sample Data came from the 2007 Australian National Survey of Mental Health and Wellbeing (NSMHWB). The NSMHWB was conducted in 2007 and is a nationally representative household survey including
109
8841 Australians aged 16–85 years old. Participants were randomly selected from a stratified, multistage area, probability sample and data were weighted according to the inverse probability of being selected. The survey achieved a final response rate of 60%, with a non-response follow-up study showing that mis-estimation at the aggregate level is predicted to be small (Slade, Johnston, Browne, Andrews, & Whiteford, 2009). Interviews were conducted by the Australian Bureau of Statistics (ABS), governed by Australian National Legislation that mandates strict provisions for the ethical conduct of research. Further details of the survey and its participants have been extensively reported elsewhere (Slade et al., 2009). 2.2. Measures 2.2.1. Assessment of anxiety disorders A modified version of the World Mental Health Composite Diagnostic Interview (WMH—CIDI 3.0; Kessler, 2004) was used to determine if respondents met criteria for a DSM-IV anxiety disorder during their lifetime. Individual anxiety disorders assessed included; generalized anxiety disorder (GAD), obsessivecompulsive disorder (OCD), agoraphobia, panic disorder, social phobia, and post-traumatic stress disorder (PTSD). Age of onset for each disorder was determined by the self-reported age that respondents first started to experience symptoms, or had the first episode of a disorder. When examining anxiety disorders as an overall class of disorders, age of onset was taken to be the earliest onset age from two, or more, individual disorders. Early onset of a disorder was defined as onset before the lower inter-quartile range for each disorder, thereby capturing cases that developed in the first 25% of the age of onset range for each disorder. The inter-quartile range and median age of onset for anxiety disorders in the Australian general population have been reported elsewhere (McEvoy et al., 2011). 2.2.2. Assessment of first use of alcohol First use of alcohol was measured by the question, “how old were you the very first time you ever drank an alcoholic beverage—including either beer, wine or spirits?”. Age of onset responses in cross-sectional surveys like the NSMHWB rely on retrospective recall and are subject to retrospective recall bias (Simon & VonKorff, 1995). To attenuate this problem the current survey utilized innovative strategies developed in the WMH CIDI, specifically aimed at increasing the accuracy of age of onset reports (Knäuper, Cannell, Schwarz, Bruce, & Kessler, 1999). For example, participants were asked if they could recall their exact age of onset (rather than asking “what was your age?”). If they could not remember their exact age they were asked to estimate their age using key developmental milestones to anchor their thinking (e.g. was it before you started school?). These strategies have been shown to both (a) increase the test-retest consistency and (b) substantially increase the plausibility of age of onset distributions, when compared to standard age of onset questions (Knäuper et al., 1999). 2.2.3. Covariates Family history of alcohol or drug problems, sex, birth cohort and education are known or were hypothesized to be related to the onset of drinking (Dawson, 2000; Degenhardt, Lynskey, & Hally, 2000). These factors were included in all survival models as covariates. Family history of alcohol or drug problems was defined as having one or more close relative/s with a history of problems with alcohol or drug use. Birth cohort was based on year of birth in four categories; 1991–1978, 1977–1968, 1967–1958, and 1957 or before. Education was coded as the highest year of secondary school completed from the following six categories; Year 12 (at least 12 years of education), Year 11 (at least 11
Survey procedures in STATA version 13 were used for all analyses. Prevalence rates were weighted to correspond to population estimates. Ninety five percent confidence intervals were obtained by, and standard errors produced via, the delete-a-group jack-knife variance procedure. This technique takes into account the complex sampling design of the NSMHWB. Kaplan–Meier survival estimates were used to compute median ages of onset, by sex (Homer & Lemeshow, 1989). Discrete-time survival analysis was used to model time to first use of alcohol. Respondents who had never had a full alcoholic drink were censored at their age of interview. Each analysis modeled the conditional probability of transitioning from never having consumed a full alcoholic beverage to first alcohol use (defined as consuming at least one standard drink—10 g of alcohol) in any given year. The survival models assessed the influence of early onset anxiety disorders in predicting this transition. Respondents whose first drink occurred prior to the onset of their anxiety disorder were excluded from the analysis. This resulted in the following sample sizes; any anxiety disorder N = 7544, GAD N = 8222, OCD N = 8619, agoraphobia N = 8742, panic N = 8570, social phobia N = 8256, and PTSD N = 8351. The effect of time in all survival models was estimated by including linear, quadratic and cubic terms. Preliminary modeling ensured that models met the assumption of proportionality of hazards. Where this assumption was not met, models included interaction terms between covariates and time. All survival models controlled for family history of alcohol or drug problems, sex, age cohort, and education. Hierarchy rules were not applied.
3. Results 3.1. Prevalence of alcohol use and early onset anxiety disorders Ninety four percent of the Australian population had consumed at least one alcoholic beverage in their lifetime (95% CI = 92.9–94.4). The median age of first alcohol use in the population was 16 years (IQR: 14–18). On examining these figures separately for males and females, ninety six percent of males (95% CI = 95.4–97.0) and ninety one percent of females (95% CI = 89.7–92.3) had consumed at least one alcoholic drink, with the mean age of first alcohol use 16 and 17 years, respectively. Data on the lifetime prevalence of early onset anxiety disorders for males and females in the Australian population, as well as the corresponding median age of onset of first alcohol use is shown in Table 1. Overall, 6.7% of the Australian population had an early onset anxiety disorder, with individual early onset anxiety disorder prevalence rates ranging from 0.5% for Agoraphobia to 2.1% for Social Phobia. Prevalence estimates for females were consistently higher than for males across all disorders. The median age of first alcohol use, given an early onset anxiety disorder, was 15 years for males, 16 years for females and 16 years overall. For individual anxiety disorders this ranged from 14 years, for males with early onset agoraphobia, to 17 years for both males and females with early onset panic disorder. Fig. 1 displays the Kaplan–Meier failure curves for cumulative lifetime probability of first alcohol use stratified by early onset anxiety disorder vs no anxiety disorder.
1.00 0.75 0.50
2.3. Analysis
Kaplan-Meier failure estimates - first drink by anxiety
0.25
years of education), Year 10 (at least 10 years of education), Year 9 (at least 9 years of education), Year 8 or below (8 or less years of education), or never attended school (0 years of education).
Proportion who have had a full drink
L. Birrell et al. / Journal of Anxiety Disorders 31 (2015) 108–113
0.00
110
0
5
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 Time No Anxiety Disorder
Early Onset Anxiety Disorder
Fig. 1. Cumulative lifetime probability of first alcohol use for persons with an early onset anxiety disorder vs no anxiety disorder.
3.2. Predictors of the onset of first alcohol use Results from the final discrete-time survival analysis model are presented in Table 2. After adjusting for the important confounders, people who experienced an early onset anxiety disorder (N = 528) had a 27% increased odds of first alcohol use in any given year, when compared to those with no anxiety disorder (OR: 1.27, 95% CI 1.08–1.51, p < 0.01). On examining individual anxiety disorders, only early onset agoraphobia remained a significant predictor of first alcohol use (OR: 1.60, 95% CI 1.04–2.46, p < 0.05), with PTSD trending towards a positive predictor (OR: 1.23, 95% CI 0.98–1.53, p = 0.07). Interactions between time and each early onset disorder were tested. Significant time interactions in the following models led us to further explore the relationship between first alcohol use and; (i) early onset anxiety disorders overall, (ii) early onset GAD, (iii) early onset panic, and (iv) early onset PTSD. This exploratory analysis established that the relationship between early onset anxiety and first alcohol use changed around the ages of 13–14 years. For transitions to first alcohol use that occurred before the age of 14 years the presence of an early onset anxiety disorder was not significantly related to the odds of starting to drink (OR: 0.80, 95% CI 0.59–1.07, p = 0.13), while after the age of 13 years the presence of an early onset anxiety disorder significantly increased the odds of starting to drink (OR: 1.50, 95% CI 1.20–1.88, p < 0.001). For individual anxiety disorders the odds ratios before age 14 were non-significant but all displayed a negative trend. The effect was consistently reversed after the age of 13 years, with significant positive associations demonstrated for PTSD (OR: 1.54, 95% CI 1.13–2.10, p ≤ 0.01). However, this follow-up analysis needs to be interpreted with caution given the number who transitioned to drinking before age 14 is low. 4. Discussion This study was the first to examine the relationship and temporal sequencing of anxiety disorders, focusing specifically on early onset anxiety disorders, and age of first alcohol use. Overall, the median age of first alcohol use in the population was 16 years. When looking at the data descriptively, those people with early onset agoraphobia and PTSD had their first drink slightly earlier than the rest of the population, while those with early onset panic disorder had their first drink slightly later. Age of first alcohol use was the same, or slightly older, in females compared to males. On testing the significance of these observed differences, the presence
L. Birrell et al. / Journal of Anxiety Disorders 31 (2015) 108–113
111
Table 1 Lifetime prevalence of early onset anxiety disorders, also occurring before age of first drink, for males and females in the Australian population and median age of first alcohol use† . Early onset disorder Males
Anxiety disorder* GAD OCD* Agoraphobia** Panic Social phobia PTSD
Females
Total
N
Weighted prevalence (%)
95% CI
Median age first alcohol use
N
Weighted Prevalence (%)
185 37 29 9 15 63 50
2.7 0.5 0.5 0.2 0.4 0.9 0.7
2.2–3.4 0.3–0.9 0.3–0.8
Contents lists available at ScienceDirect
Journal of Anxiety Disorders
Anxiety disorders and first alcohol use in the general population. Findings from a nationally representative sample Louise Birrell ∗ , Nicola C. Newton, Maree Teesson, Zoe Tonks, Tim Slade NHMRC Centre of Research Excellence in Mental Health and Substance Use (CREMS), National Drug and Alcohol Research Centre (NDARC), University of New South Wales, Sydney 2052, NSW, Australia
a r t i c l e
i n f o
Article history: Received 9 December 2014 Received in revised form 19 February 2015 Accepted 24 February 2015 Available online 6 March 2015 Keywords: Anxiety Substance use Epidemiology Alcohol drinking Age of onset Adolescent Age factors Age of first drink
a b s t r a c t Objective: To examine how early onset anxiety disorders are related to age of first alcohol use in a general population sample. Method: Discrete time survival analysis was used to model the odds of first alcohol use among those with, vs without, early onset anxiety disorders. Data came from the 2007 Australian National Survey of Mental Health and Wellbeing. Results: After adjusting for the effects of family history of alcohol/drug use, sex, age cohort and education, people who experienced an early onset anxiety disorder had a 27% increased odds of first alcohol use in any given year, when compared to those with no anxiety disorder. This effect was particularly strong for transitions to first alcohol use that occurred after the age of 13 years. Conclusions: Early onset anxiety disorders significantly predict first alcohol use in the general population and this relationship appears to be related to change over time. These results point to the need for developmentally appropriate and integrated prevention programs that target anxiety and alcohol use together. © 2015 Elsevier Ltd. All rights reserved.
1. Introduction The most common mental health and substance use disorders in the general population are anxiety (Kessler et al., 2012a) and alcohol use disorders (Teesson et al., 2010). Epidemiological estimates in community samples approximate that between 20% (McEvoy, Grove, & Slade, 2011) and 31% (Kessler et al., 2007a) of the population in Western countries will suffer from an anxiety disorder during their lifetime, and between 19% (Kessler et al., 2005) and 22% (Teesson et al., 2010) will meet criteria for an alcohol use disorder. These two classes of disorders are also highly comorbid (McEvoy et al., 2011; Teesson et al., 2005). Individuals with an alcohol use disorder are 2.6 times more likely to have a comorbid anxiety disorder during their lifetime, than those without an alcohol use disorder (Teesson et al., 2010). People with comorbid mental health and substance use problems are known to be a particularly high risk group that are more difficult to treat, place a greater burden on service providers and have a more severe illness course (de Graaf, Bijl, ten Have, Beekman, & Vollebergh, 2004; Teesson & Proudfoot, 2003).
∗ Corresponding author. Tel.: +61 2 9385 0333. E-mail addresses: [email protected], [email protected] (L. Birrell), [email protected] (N.C. Newton), [email protected] (M. Teesson), [email protected] (Z. Tonks), [email protected] (T. Slade). http://dx.doi.org/10.1016/j.janxdis.2015.02.008 0887-6185/© 2015 Elsevier Ltd. All rights reserved.
To help understand this comorbidity it is useful to look at its origins and examine how anxiety disorders relate to age of first alcohol use. Age of first alcohol use is an important outcome to examine in understanding the etiology and temporal sequencing of anxiety and alcohol use comorbidity. There is robust evidence demonstrating an association between an early age of first alcohol use and later alcohol-related problems and alcohol use disorders (Dawson, Goldstein, Chou, Ruan, & Grant, 2008; DeWit, Adlaf, Offord, & Ogborne, 2000; Fergusson, Horwood, & Lynskey, 1994; Grant & Dawson, 1997; Grant, Stinson, & Harford, 2001; Gruber, DiClements, Anderson, & Lodico, 1996; Hawkins et al., 1997; Liang & Chikritzhs, 2011). In a large US study, Grant et al. (2001) followed over 5000 young people aged between 14 and 21 years old for 12 years and found that the odds of alcohol dependence was decreased by 9% for each year that first alcohol use was delayed. Debate exists about whether evidence such as this demonstrates a causal relationship between age of first alcohol use and later alcohol use problems (Deutsch et al., 2013; Irons, Iacono, & McGue, 2014) or not (Geels et al., 2013; McGue, Iacono, Legrand, Malone, & Elkins, 2001; Prescott & Kendler, 1999), and as yet there is no definitive consensus about the mechanisms underlying these relationships. A number of key variables are known to play a significant role in the relationship between age of first alcohol use and later alcohol problems, including psychopathology and anxiety disorders (Pang, Farrahi, Glazier, Sussman, & Leventhal, 2014;
L. Birrell et al. / Journal of Anxiety Disorders 31 (2015) 108–113
Sartor, Lynskey, Heath, Jacob, & True, 2007). However, the majority of studies examining the role of anxiety disorders and alcohol use focus on regular alcohol use and alcohol use disorders as the key outcomes of interest (Behrendt et al., 2011). Particular attention has been given to the relationship between anxiety disorders and an increased speed of transition from first alcohol use to alcohol dependence (Behrendt et al., 2011; Lopez-Quintero et al., 2011). Fewer studies focus on first alcohol use as the primary outcome, and even fewer examine the temporal sequencing of anxiety disorders in relation to first alcohol use. This study will focus on first alcohol use as the key outcome. This is warranted, as evidence suggests that risk factors play different roles in the transition to regular alcohol use and alcohol use disorders, than in the transition to first alcohol use (Sartor et al., 2007). Furthermore, previous studies have not differentiated between the individual types of anxiety disorders, which may show differential relationships to alcohol use. The current study will address these gaps. The temporal sequencing of first alcohol use and anxiety disorders centers around the developmental period of adolescence. In most Western countries, adolescence is a time when the majority of individuals consume their first alcoholic drink. It is also the time when a large number of the anxiety disorders have their onset (Falk, Yi, & Hilton, 2008), making this a particular developmental period of interest to examine the comorbidity of anxiety and alcohol use. More specifically, anxiety disorders that have an early onset have been linked to reports of greater anxiety severity, poor treatment response, more significant functional impairment and early depression (Dalrymple & Zimmerman, 2011; Rosellini, Rutter, Bourgeois, Emmert-Aronson, & Brown, 2013). Adolescence is also a distinct time of neuro-development, with key brain structures involved in emotion regulation and behavioral planning (such as the frontal lobes, temporal lobe, amygdala and hippocampus) undergoing dynamic change (Clark, Thatcher, & Tapert, 2008). This leaves adolescents particularly susceptible to risky behaviors, poor decision making and lack of emotion regulation. Furthermore, alcohol is known to inhibit behavioral inhibition and is likely to compound this lack of executive brain function. A recent review of brain imaging studies on the effect of alcohol consumption on the adolescent brain concluded that adolescents who frequently use alcohol show a different pattern of brain structure and function, compared to those who have not yet tried alcohol (Feldstein Ewing, Sakhardande, & Blakemore, 2014). Together this evidence highlights the need for further research around alcohol use at this critical life stage. To understand why anxiety and alcohol use problems occur together it is important to explore their temporal sequencing over both; (a) the lifespan, and (b) across the different stages of alcohol use. This study focuses on first use of alcohol as a logical starting point in the etiology of anxiety and alcohol disorder comorbidity. Specifically, the present study aimed to examine the temporal sequencing of early onset anxiety disorders in relation to age of first alcohol use in a general population sample. It was hypothesized that those who experienced an early onset anxiety disorder would be at increased odds of initiating drinking in any given year. This study will be the first to examine how early onset anxiety disorders relate to first use of alcohol use, using a nationally representative general population sample.
2. Methods and materials 2.1. Sample Data came from the 2007 Australian National Survey of Mental Health and Wellbeing (NSMHWB). The NSMHWB was conducted in 2007 and is a nationally representative household survey including
109
8841 Australians aged 16–85 years old. Participants were randomly selected from a stratified, multistage area, probability sample and data were weighted according to the inverse probability of being selected. The survey achieved a final response rate of 60%, with a non-response follow-up study showing that mis-estimation at the aggregate level is predicted to be small (Slade, Johnston, Browne, Andrews, & Whiteford, 2009). Interviews were conducted by the Australian Bureau of Statistics (ABS), governed by Australian National Legislation that mandates strict provisions for the ethical conduct of research. Further details of the survey and its participants have been extensively reported elsewhere (Slade et al., 2009). 2.2. Measures 2.2.1. Assessment of anxiety disorders A modified version of the World Mental Health Composite Diagnostic Interview (WMH—CIDI 3.0; Kessler, 2004) was used to determine if respondents met criteria for a DSM-IV anxiety disorder during their lifetime. Individual anxiety disorders assessed included; generalized anxiety disorder (GAD), obsessivecompulsive disorder (OCD), agoraphobia, panic disorder, social phobia, and post-traumatic stress disorder (PTSD). Age of onset for each disorder was determined by the self-reported age that respondents first started to experience symptoms, or had the first episode of a disorder. When examining anxiety disorders as an overall class of disorders, age of onset was taken to be the earliest onset age from two, or more, individual disorders. Early onset of a disorder was defined as onset before the lower inter-quartile range for each disorder, thereby capturing cases that developed in the first 25% of the age of onset range for each disorder. The inter-quartile range and median age of onset for anxiety disorders in the Australian general population have been reported elsewhere (McEvoy et al., 2011). 2.2.2. Assessment of first use of alcohol First use of alcohol was measured by the question, “how old were you the very first time you ever drank an alcoholic beverage—including either beer, wine or spirits?”. Age of onset responses in cross-sectional surveys like the NSMHWB rely on retrospective recall and are subject to retrospective recall bias (Simon & VonKorff, 1995). To attenuate this problem the current survey utilized innovative strategies developed in the WMH CIDI, specifically aimed at increasing the accuracy of age of onset reports (Knäuper, Cannell, Schwarz, Bruce, & Kessler, 1999). For example, participants were asked if they could recall their exact age of onset (rather than asking “what was your age?”). If they could not remember their exact age they were asked to estimate their age using key developmental milestones to anchor their thinking (e.g. was it before you started school?). These strategies have been shown to both (a) increase the test-retest consistency and (b) substantially increase the plausibility of age of onset distributions, when compared to standard age of onset questions (Knäuper et al., 1999). 2.2.3. Covariates Family history of alcohol or drug problems, sex, birth cohort and education are known or were hypothesized to be related to the onset of drinking (Dawson, 2000; Degenhardt, Lynskey, & Hally, 2000). These factors were included in all survival models as covariates. Family history of alcohol or drug problems was defined as having one or more close relative/s with a history of problems with alcohol or drug use. Birth cohort was based on year of birth in four categories; 1991–1978, 1977–1968, 1967–1958, and 1957 or before. Education was coded as the highest year of secondary school completed from the following six categories; Year 12 (at least 12 years of education), Year 11 (at least 11
Survey procedures in STATA version 13 were used for all analyses. Prevalence rates were weighted to correspond to population estimates. Ninety five percent confidence intervals were obtained by, and standard errors produced via, the delete-a-group jack-knife variance procedure. This technique takes into account the complex sampling design of the NSMHWB. Kaplan–Meier survival estimates were used to compute median ages of onset, by sex (Homer & Lemeshow, 1989). Discrete-time survival analysis was used to model time to first use of alcohol. Respondents who had never had a full alcoholic drink were censored at their age of interview. Each analysis modeled the conditional probability of transitioning from never having consumed a full alcoholic beverage to first alcohol use (defined as consuming at least one standard drink—10 g of alcohol) in any given year. The survival models assessed the influence of early onset anxiety disorders in predicting this transition. Respondents whose first drink occurred prior to the onset of their anxiety disorder were excluded from the analysis. This resulted in the following sample sizes; any anxiety disorder N = 7544, GAD N = 8222, OCD N = 8619, agoraphobia N = 8742, panic N = 8570, social phobia N = 8256, and PTSD N = 8351. The effect of time in all survival models was estimated by including linear, quadratic and cubic terms. Preliminary modeling ensured that models met the assumption of proportionality of hazards. Where this assumption was not met, models included interaction terms between covariates and time. All survival models controlled for family history of alcohol or drug problems, sex, age cohort, and education. Hierarchy rules were not applied.
3. Results 3.1. Prevalence of alcohol use and early onset anxiety disorders Ninety four percent of the Australian population had consumed at least one alcoholic beverage in their lifetime (95% CI = 92.9–94.4). The median age of first alcohol use in the population was 16 years (IQR: 14–18). On examining these figures separately for males and females, ninety six percent of males (95% CI = 95.4–97.0) and ninety one percent of females (95% CI = 89.7–92.3) had consumed at least one alcoholic drink, with the mean age of first alcohol use 16 and 17 years, respectively. Data on the lifetime prevalence of early onset anxiety disorders for males and females in the Australian population, as well as the corresponding median age of onset of first alcohol use is shown in Table 1. Overall, 6.7% of the Australian population had an early onset anxiety disorder, with individual early onset anxiety disorder prevalence rates ranging from 0.5% for Agoraphobia to 2.1% for Social Phobia. Prevalence estimates for females were consistently higher than for males across all disorders. The median age of first alcohol use, given an early onset anxiety disorder, was 15 years for males, 16 years for females and 16 years overall. For individual anxiety disorders this ranged from 14 years, for males with early onset agoraphobia, to 17 years for both males and females with early onset panic disorder. Fig. 1 displays the Kaplan–Meier failure curves for cumulative lifetime probability of first alcohol use stratified by early onset anxiety disorder vs no anxiety disorder.
1.00 0.75 0.50
2.3. Analysis
Kaplan-Meier failure estimates - first drink by anxiety
0.25
years of education), Year 10 (at least 10 years of education), Year 9 (at least 9 years of education), Year 8 or below (8 or less years of education), or never attended school (0 years of education).
Proportion who have had a full drink
L. Birrell et al. / Journal of Anxiety Disorders 31 (2015) 108–113
0.00
110
0
5
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 Time No Anxiety Disorder
Early Onset Anxiety Disorder
Fig. 1. Cumulative lifetime probability of first alcohol use for persons with an early onset anxiety disorder vs no anxiety disorder.
3.2. Predictors of the onset of first alcohol use Results from the final discrete-time survival analysis model are presented in Table 2. After adjusting for the important confounders, people who experienced an early onset anxiety disorder (N = 528) had a 27% increased odds of first alcohol use in any given year, when compared to those with no anxiety disorder (OR: 1.27, 95% CI 1.08–1.51, p < 0.01). On examining individual anxiety disorders, only early onset agoraphobia remained a significant predictor of first alcohol use (OR: 1.60, 95% CI 1.04–2.46, p < 0.05), with PTSD trending towards a positive predictor (OR: 1.23, 95% CI 0.98–1.53, p = 0.07). Interactions between time and each early onset disorder were tested. Significant time interactions in the following models led us to further explore the relationship between first alcohol use and; (i) early onset anxiety disorders overall, (ii) early onset GAD, (iii) early onset panic, and (iv) early onset PTSD. This exploratory analysis established that the relationship between early onset anxiety and first alcohol use changed around the ages of 13–14 years. For transitions to first alcohol use that occurred before the age of 14 years the presence of an early onset anxiety disorder was not significantly related to the odds of starting to drink (OR: 0.80, 95% CI 0.59–1.07, p = 0.13), while after the age of 13 years the presence of an early onset anxiety disorder significantly increased the odds of starting to drink (OR: 1.50, 95% CI 1.20–1.88, p < 0.001). For individual anxiety disorders the odds ratios before age 14 were non-significant but all displayed a negative trend. The effect was consistently reversed after the age of 13 years, with significant positive associations demonstrated for PTSD (OR: 1.54, 95% CI 1.13–2.10, p ≤ 0.01). However, this follow-up analysis needs to be interpreted with caution given the number who transitioned to drinking before age 14 is low. 4. Discussion This study was the first to examine the relationship and temporal sequencing of anxiety disorders, focusing specifically on early onset anxiety disorders, and age of first alcohol use. Overall, the median age of first alcohol use in the population was 16 years. When looking at the data descriptively, those people with early onset agoraphobia and PTSD had their first drink slightly earlier than the rest of the population, while those with early onset panic disorder had their first drink slightly later. Age of first alcohol use was the same, or slightly older, in females compared to males. On testing the significance of these observed differences, the presence
L. Birrell et al. / Journal of Anxiety Disorders 31 (2015) 108–113
111
Table 1 Lifetime prevalence of early onset anxiety disorders, also occurring before age of first drink, for males and females in the Australian population and median age of first alcohol use† . Early onset disorder Males
Anxiety disorder* GAD OCD* Agoraphobia** Panic Social phobia PTSD
Females
Total
N
Weighted prevalence (%)
95% CI
Median age first alcohol use
N
Weighted Prevalence (%)
185 37 29 9 15 63 50
2.7 0.5 0.5 0.2 0.4 0.9 0.7
2.2–3.4 0.3–0.9 0.3–0.8
