Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Sep 1.
Published in final edited form as: Int J Eat Disord. 2021 Jul 14;54(9):1632–1640. doi: 10.1002/eat.23580

Examining the Significance of Age of Onset in Persons with Lifetime Anorexia Nervosa: Comparing Child, Adolescent, and Emerging Adult Onsets in Nationally Representative U.S. Study

Carlos M Grilo 1, Tomoko Udo 2
PMCID: PMC8416938  NIHMSID: NIHMS1721787  PMID: 34263464

Abstract

Objective:

This study compared sociodemographic and clinical profiles of adult patients with lifetime DSM-5-defined anorexia nervosa (AN) categorized by age-of-onset using data from U.S. national sample of adults.

Method:

Study included 216 participants from Third National Epidemiological Survey Alcohol and Related Conditions (NESARC-III) who met criteria for lifetime AN based on structured diagnostic interviews (AUDADIS-5) with age-of-onset prior to age 25. Of the 216 participants, 30 were categorized as child-onset (<15 years old), 104 adolescent-onset (15–18 years of age), and 82 “emerging-adult” (19–24 years of age); the three groups were compared on their clinical characteristics.

Results:

Among participants with lifetime diagnoses of AN with onsets earlier than 25 years, adjusted prevalence rates for the three groups were: 11.8% (SE=2.04; child-onset), 39.6% (SE=2.69; adolescent-onset), and 48.6% (SE=2.67; emerging-adult). Child-onset group reported more frequent adverse childhood experiences (ACEs), lowest BMI, longest episode-duration, was least likely to attend college, and had highest rate of lifetime psychiatric comorbidity. Child-onset group had earliest age of help-seeking and were most likely to have been hospitalized. Group differences persisted in analyses adjusting for sociodemographic characteristics and duration of AN episode.

Discussion:

Our findings, based on a nationally representative sample of U.S. adults with lifetime diagnoses of AN, suggest that those with child-onset had more severe AN, greater life difficulties, and greater lifetime psychiatric comorbidity. Findings emphasize the importance of earlier recognition and rapid referral to effective treatments.

Keywords: anorexia nervosa, eating disorders, children, adolescents, emerging adults, comorbidity, BMI, adverse child experiences, treatment

INTRODUCTION

Anorexia nervosa (AN), a serious and often chronic eating disorder with high morbidity and mortality, develops most frequently in adolescence with studies generally reporting peak incidence at roughly 15–19 years (Hudson, Hiripi, Pope, & Kessler, 2007; Mohler-Kuo, Schnyder, Dermota, Wei, & Milos, 2016; Silen et al., 2019; Udo & Grilo, 2018; see Hoek, 2016). Epidemiological studies limited to adolescents report earlier ages of AN onsets with interquartile ranges of approximately 11 to 13 years (Swanson, Crow, Le Grange, Swendsen, & Merikangis, 2011). Some surveillance and registry data suggest trends towards increasing cases of AN in children (Madden, Morris, Zurynski, Kohn, & Elliot, 2006; Nicholls, Lynn, & Viner, 2011; Pinhas, Morris, Crosby, & Katzman, 2011; Steinhausen & Jensen, 2015).

Relatively little is known about the clinical features and natural course of early or child-onset AN (Herpertz-Dahlmann & Dahmen, 2019). Some data from registry studies suggest that early onset cases tend to be particularly severe (Madden et al., 2006) whereas population-based data of eating-disorder behaviors suggest that duration, rather than age-of-onset, is associated with greater severity and impairment (Wong & Hay, 2020). To date, the few studies that have directly compared child-onset to adolescent-onset AN cases have produced mixed findings with regard to severity and outcomes (Herpertz-Dahlmann., et al., 2018; Jaite et al., 2019; Kwok, Kwok, Lee, & Tan, 2019; Neubauer et al., 2014; Peebles, Wilson, & Lock, 2006; Wentz et al., 2009; see Herpertz-Dahlmann & Dahmen, 2019). For example, Herpertz-Dahlmann & Dahmen (2019), in their recent review, noted that several studies found that child-onset AN trended towards poorer physical and mental health outcomes whereas the recent analysis of child-onset versus adolescent-onset AN from a German web-based registry for inpatient hospitalizations (Jaite et al., 2019) reported a more favorable clinical profile amongst those with child-onset. Herpertz-Dahlmann & Dahmen (2019) concluded that there is an urgent need for further research to improve our understanding of the clinical characteristics of children and adolescents who develop AN as this may inform improved recognition, treatment formulation, and outcomes.

Nearly all of the available research directly comparing child-onset and adolescent-onset AN comes from treatment-seeking clinical samples mostly from inpatient and hospital registry settings (Herpertz-Dahlmann & Dahmen, 2019). This contrasts with the available literature for adolescents and young adults which includes growing numbers of studies with large representative samples throughout the world that have contributed data on natural course and outcome (e.g., Keski-Rahkonen et al., 2007; Lewinsohn, Striegel-Moore, & Seeley, 2000; Nagl et al., 2016; Wade, Bergin, Tiggemann, Bulik, & Fairburn, 2006). Thus, the few available data on early child-onset cases of AN come from patients who either sought or necessitated the most intensive available treatments. It is well-established, yet frequently forgotten, that representativeness of findings from treatment-seeking clinical samples may be limited for several reasons. Community cases might be less severe and have better prognosis (Keski-Rahkonen et al., 2007; Wade et al., 2006) and often do not seek treatment for their eating disorder and especially not intensive residential treatments (Mohler-Kuo et al., 2016; Coffino, Udo, & Grilo, 2019). For example, in a U.S. nationally representative sample, only 7.3% of patients with lifetime AN reported having been hospitalized (Coffino et al., 2019).

Representativeness of findings from treatment-seeking clinical samples of AN may be further limited by other confounds which could, in turn, also limit the utility of the data for informing recognition and intervention efforts. Many factors influence treatment-seeking behaviors including failure to seek or amount of delay in seeking help (Wang, Berglund, Olfson, Pincus, Wells, & Kessler, 2005) and this might be especially complicated in the case of AN which is well-known to be ego-syntonic, associated with secrecy and denial, often not recognized by generalist healthcare providers, and infrequently obtain specialist-providers (Hart et al., 2011; Udo & Grilo, 2019). Treatment-seeking is further characterized by various disparities (e.g., racial/ethnic, sex, economic); these have been documented for eating disorders (Coffino et al., 2019; Marques et al., 2011; Sonneville & Lipson, 2018) and other psychiatric conditions (Laliberte et al., 2020; Wang et al., 2005). Representativeness of clinical samples is also further confounded due “Berkson’s bias” (mathematical bias that a person with two or more disorders can seek treatment for either disorder) because of a related, yet distinct, “clinical bias” whereby different comorbidities influence whether treatment is sought and, if so, for which disorder (du Fort, Newman, & Bland, 1993; Laliberte et al., 2020). Indeed, a recent review of barriers and facilitators towards help-seeking for eating disorders identified the presence of other psychiatric problems as a prominent factor (Ali et al., 2017).

This study aimed to provide broad description and comparison of patients with lifetime DSM-5-defined anorexia nervosa (AN) categorized by age-of-onset using data from national sample of U.S. adults. Specific aims were to compare sociodemographic (sex, ethnicity/race, education, income, martial/partner status), AN-related clinical profiles (lowest weight, AN duration, AN-related psychosocial impairment) and treatment-seeking behaviors, adverse childhood experiences (ACE), and lifetime psychiatric comorbidity in persons with histories of child-onset versus adolescent onset AN. To provide further context, a third comparison group comprising “emerging-adult”-onset of AN was used. The emerging-adult-onset AN group was selected given the importance of this developmental stage in the life course (Wood et al., 2018) and emerging research on initial onsets in emerging adults (Potterton, Austin, Allen, Lawrence, & Schmidt, 2020) and continuities of eating disorders from adolescence (Lewinsohn, Striegel-Moore, & Seeley, 2000; Nagl et al., 2016; Waszezuk, Waaktaar, Eley, & Torgersen, 2019). This analysis with a nationally representative sample was intended to address some of the potential confounds in treatment-seeking clinical biases in the available literature regarding “child-onset” AN relative to later onsets (Herpertz-Dahlman & Dahmen, 2019). We hypothesized that child-onset AN would have a more negative clinical profile and greater difficulties across multiple life and functioning areas than later onsets (adolescent-onset and emerging-adult onset) of AN.

METHOD

Sample

The National Epidemiologic Survey on Alcohol and Related Conditions–III (NESARC-III) was designed to assess the prevalence and correlates of alcohol and other drug use disorders in non-institutionalized U.S adults 18 years and older, The NESARC-III included a total of 36,309 respondents who completed computer-assisted face-to-face personal interviews between April 2012 and June 2013 (for details on sampling procedure, see Grant et al., 2014; Grant et al., 2016). The NESARC-III employed multi-stage probability sampling. Counties or groups of contiguous counties were primary sampling units, groups of Census-defined blocks were secondary sampling units, and households within secondary sampling units were tertiary sampling units. Within each household, eligible adults were randomly selected, but Hispanic, Black, and Asian household members were oversampled (i.e., two respondents from households with more than four eligible minority members), relative to White household members. Each field interviewer completed initial structured home study and in-person training prior to interviewing respondents and received on-going supervision to prevent drift.

NESARC-III was approved by the NIH IRB and respondents provided oral informed-consent which was electronically recorded (Grant et al., 2014). The study was determined to be exempt from the IRB oversight by from SUNY-Albany for use of existing de-identified data.

Assessments

Diagnostic Assessment of DSM-5 Eating Disorders and Psychiatric Disorders.

The NIAAA Alcohol Use Disorder and Associated Disabilities Interview Schedule-5 (AUDADIS-5; Grant et al., 2011) is structured diagnostic interview that assessed a range of DSM-5-defined psychiatric disorders and their criteria, including AN, BN, and BED. As detailed in Udo & Grilo (2018), after close inspection of the NESARC-III ED diagnostic and criteria variables we created DSM-5-based ED categories for analysis, rather than utilizing the diagnosis variables provided by NESARC-III. We previously published the ED diagnostic codes, including all the AUDADIS-5 items used in NESARC-III used to generate DSM-5-defined ED diagnoses, in Supplemental Tables available on-line (Udo & Grilo, 2018; Udo & Grilo, 2019). The AUDADIS-5 also asked questions about the age of onset which was defined when respondents first began to experience the specific symptoms for each formal ED,. For AN, the age-of-onset item (N17Q9) was “About how old were you when you FIRST weighed less than (i.e., 85% of expected weight) and had SOME of the other experiences (i.e., items from N17Q1 – N17Q6E, reflecting restriction of food intake and various body image concerns and related behaviors) you mentioned at the same time?” Furthermore, the interview asked age at the most recent episode (item N17Q12R for AN), and the duration of the most recent episode (item N17Q14R for AN), which, along with age of onset, were used to calculate duration of AN.

The reliability and validity of the AUDADIS-5 have been reported mood disorders (major depressive episodes, persistent depression, and bipolar I), anxiety disorders (specific phobia, social phobia, panic disorders, agoraphobia, and generalized anxiety disorder), posttraumatic stress disorder (PTSD), substance use disorder (alcohol use disorder, drug use disorder, and nicotine use disorder), personality disorders (antisocial, borderline, and schizotypal), and conduct disorder (Grant et al., 2015; Hasin, Greenstein, et al., 2015; Hasin, Shmulewitz, et al., 2015), but these studies did not include EDs. As detailed in Udo & Grilo (2018), after close inspection of the NESARC-III AN diagnostic and criteria variables we created DSM-5-based ED categories for analysis, rather than utilizing the AN diagnosis variables provided by NESARC-III. We used the original diagnostic variables for other psychiatric disorders as coded by NESARC-III.

Sociodemographic Characteristics.

Respondents provided sociodemographic information including age, sex, ethnicity/race (non-Hispanic White, non-Hispanic Black, Hispanic, and Other [non-Hispanic Asian/Pacific Islander, and non-Hispanic American-Indian/Alaska Native], and Hispanic [any race]), education (categorized as less than H.S., H.S./GED, at least some college), and income (categorized as <$25,000, $25,000–39,999, $40,000–69,999, ≤$70,000).

Persistence and Impairment of Anorexia Nervosa.

We calculated years with AN episode based on current age and age for most recent episode. 12-month persistence of AN was defined as the proportion of those with 12-month diagnosis among those with the lifetime diagnosis; this definition follows previous conventions (Hudson et al., 2007; Udo & Grilo, 2018). The AUDADIS-5 also assessed for impairment in social function due to EDs, including: (1) interference with normal daily activities, (2) serious problems getting along with others, and (3) serious problems fulfilling responsibilities. Self-reported height and weight were used to calculate current BMI.

Treatment and help-seeking.

Respondents were asked questions regarding help-seeking for each specific ED. For AN, the help-seeking questions focused specifically the low weight and associated behaviors as leading to seek help from a very broad range of possible sources. These questions included: (1) Talked to any kind of counselor, therapist, doctor, psychologist or any person like that to get help for your low weight; (2) Went to a self-help or support group, use a hotline or visit an internet chat room; (3) Were you a patient in any kind of hospital overnight or longer; (4) Went to an emergency room; (5) Were prescribed any medicines or drugs; and (6) Went to Overeaters Anonymous or other 12-step group. Respondents were also asked the age when they first sought any help for each low weight-related problem.

Adverse Childhood Experiences (ACEs).

Respondents self-reported five types of childhood maltreatment (physical neglect, emotional neglect, physical abuse, emotional abuse, and sexual abuse) by parents or caregivers, and 13 other adverse events that occurred before 18 years old. Following the methods of previous studies (Udo, 2019; Udo & Grilo, 2016), report of experiencing any form of childhood maltreatment or other adverse events was coded as having a positive history (see, Udo, 2019, for the operationalization of ACEs).

Categorization of Anorexia Nervosa Age-of-Onset Groups

This study included 216 individuals who participated in NESARC-III, met criteria for lifetime DSM-5 AN diagnosis, and reported age of AN onset below 25 years old. The additional N=60 individuals with lifetime AN in NESARC-III who reported age of onset 25 or older were excluded from the present analyses because of this study’s focus on AN onset through “emerging adulthood.” The 216 participants were categorized into three age-of-onset groups: “child-onset” (less than 15 years of age), “adolescent-onset” (15–18 years of age), and “emerging-adult” (19–24 years of age).

Statistical Analysis

Weighted means, frequencies, and cross-tabulations were computed for sociodemographic characteristics and clinical profiles (current and lowest BMI, duration of AN episode, age of first seeking help, 12-month persistence, psychosocial impairment, types of help sought, and other lifetime psychiatric disorders) for the three age-of-onset AN groups. Sociodemographic characteristics were then compared, using Rao-Scott Chi-Square test (for categorical variables) and analysis of variance (ANOVA; for continuous variables). Analysis of covariance were used to compare current and lowest BMI, duration of AN episode, age of first seeking help between three groups, and significant chi-square tests for clinical variables were also followed with multiple logistic regression, with including current age, sex, education, race, income, marital status, and duration of AN episodes as covariates. We, however, did not perform multivariate analyses for ACEs given the presumed temporal relationship between timing of exposure to ACEs and onset of AN symptoms. Significant omnibus chi-square tests were probed by comparing cells (Sharpe, 2015; Marascuilo & Serlin, 1988) to identify significant differences between ED groups. Significant ANOVA, ANCOVA, and multiple regression analyses results were probed by Tukey-Kramer post-hoc analysis.

RESULTS

Among participants with lifetime diagnoses of AN with age of onset prior to 25 years, 11.8% (SE=2.04, N=30) were categorized as child-onset, 39.6% (SE=2.69, N=104) as adolescent-onset, and 48.6% (SE=2.67, N=82) as emerging-adult.

Table 1 summarizes sociodemographic characteristics across the three age-of-onset groups. At the time of participation in NESARC-III, the emerging-adult group was significantly older than both the child-onset and adolescent-onset groups. The child-onset group was significantly less likely to attend college than the two other groups and was more likely to be single (not married/partnered) than the emerging adult group. The three age-of-onset groups did not differ significantly in the distribution of sex, ethnicity/race, or income levels.

Table 1.

Sociodemographic Characteristics Across the Age-of-Onset Anorexia Nervosa Groups

Child-onset (n = 30) Adolescent-onset (n = 104) Emerging-Adult (n = 82) Statistical Test p-value
Male (%) 16.3 (8.81) 7.1 (2.70) 7.9 (2.05) χ2(3)=3.25 0.2
Mean age (years) 38.9 (3.24) 40.1 (0.29) 42.3 a, b (0.40) F(2, 80)=353.65 <.0001
Race (%)
Non-Hispanic White 96.1 (1.16) 76.6 (5.27) 76.7 (4.35)
Non-Hispanic Black 0.9 (0.92) 2.7 (1.30) 3.3 (1.42)
Hispanic 3.0 (0.58) 8.5 (2.24) 10.2 (2.13)
Other 0.0 (0.00) 12.2 (5.14) 9.7 (3.35)
Education (%) χ2 (4)=9.88 0.04
Less than high school 16.8 (6.95) 8.2 (2.54) 1.8 (1.44)
High school/GED 20.9 (10.07) 11.6 (2.36) 21.3 (3.90)
Some college 62.2 (10.33) 80.2 (3.52) 76.9 (4.06)
Income (%) χ2 (6)=11.00 0.09
Less than $25,000 22.5 (9.06) 17.8 (3.40) 13.3 (3.88)
$25,000–39,999 8.1 (2.75) 13.5 (3.37) 17.2 (4.40)
$40,000–69,999 45.5 (9.10) 21.4 (4.96) 27.7 (4.55)
More than $70,000 23.9 (8.43) 47.3 (5.42) 41.8 (4.50)
Marital status (%) χ2 (6)=11.11 0.02
Never married 41.8 (9.87) 22.7 (3.75) 15.7a (4.19)
Married/living as if married 39.8 (10.22) 59.5 (4.51) 57.9 (4.88)
Divorced/separated/widowed 18.3 (5.54) 17.8 (3.49) 26.4 (4.09)
Open in a new tab

Notes. All analyses were adjusted for the NESARC-III complex survey design. Numbers in parentheses are standard errors and the sample size.

a

= significantly different from the child-onset group at p < .05

b

= significantly different from the adolescent-onset group at p < .05.

Clinical profiles by age at AN onset

Table 2 summarizes clinical profiles across the three age-of-onset groups. Current BMI was significantly different between all three groups with the emerging-adult group reporting the lowest current BMI, followed by the adolescent-onset group. The lowest BMI was also significantly different among three groups; it was lowest in the child-onset group, followed by the emerging-adult group. The emerging-adult group reported significantly fewer years AN episode duration (14.1+1.9) than the adolescent-onset (12.3+0.3) and emerging-adult (9.7+0.5) groups which did not differ from each other. There were no significant differences in rates of impairment across the three age-of-onset groups.

Table 2.

Clinical Characteristics Across the Age-of-Onset Anorexia Nervosa Groups

Child-Onset (n = 30) Adolescent-Onset (n = 104) Emerging-Adult (n = 82) Statistical Test p-values
Clinical profiles
Mean BMI
Current1 27.4 (0.21) 26.4 a (0.32) 24.1 a,b (0.38) F(2, 80)=92.06 <.0001
Lowest1 16.7 (0.24) 17.3 a (0.02) 17.1 a,b (0.10) F(2, 80)=184.34 <.0001
Mean years with episode 14.1 (1.88) 12.3 (0.25) 9.7 a,b (0.53) F(2, 80)=262.37 <.0001
12-month persistence1 (%) 10.8 (8.57) 4.5 (2.20) 6.3 (2.90) χ2 (2)=1.11 0.57
Impairment (%)
Interfere with normal daily activities 29.1 (8.03) 18.9 (4.86) 16.5 (3.70) χ2 (2)=2.02 0.36
Serious problems getting along with others 25.2 (7.63) 19.0 (4.59) 21.0 (4.50) χ2 (2)=0.47 0.79
Serious problems fulfilling responsibilities 23.9 (7.74) 14.4 (4.84) 12.3 (3.32) χ2 (2)=1.56 0.46
Any Form 32.4 (8.36) 28.4 (4.78) 26.1 (5.02) χ2 (2)=0.39 0.82
Treatment seeking
Age of first seeking help1 17.0 (0.13) 17.1 a (0.40) 24.5 a,b (0.15) F(1, 45)= 47009.5 <.0001
 Median (IQR)3 12.9 (10.7, 15.74) 16.4 (15.6, 17.4) a 22.9 (19.9, 26.6) a,b <.0001
Type of help ever sought (%)
Counselor/psychologist 38.0 (9.54) 29.0 (4.07) 23.6 (3.84) χ2(2)=2.35 0.31
Self-help or support group 2.7 (2.68) 9.2 (2.43) 12.1 (3.13) χ2 (2)=3.46 0.18
Hospitalization 19.4 (6.54) 4.1 (1.83) 11.1 b (3.46) χ2 (2)=7.77 0.02
Emergency room 10.0 (6.42) 7.1 (4.75) 0.3 (0.27) χ2(2)=4.45 0.11
Medication 14.5 (6.95) 4.4 (1.85) 10.0 (3.25) χ2 (2)=4.62 0.10
12-step groups 11.0 (2.57) 2.8 (0.93) 6.5 (1.88) χ2 (2)=9.39 <.01
Any help 40.1 (9.58) 33.7 (4.97) 28.0 (4.35) χ2 (2)=1.38 0.50
Adverse childhood experiences
Any ACEs (%) 66.8 (9.21) 45.4 (5.16) 59.2 (4.15) χ2 (2)=5.96 0.05
Number of ACEs (%) χ2 (6)=18.13 <.01
None 16.3 (6.34) 27.5 (4.73) 22.0 (3.27)
1 13.1 (4.99) 27.9 (4.51) 18.9 (4.35)
2 or 3 15.0 (4.24) 22.7 (5.24) 22.1 (3.61)
4 or more 55.6 (9.32) 21.9 a (4.11) 37.1 b (4.49)
Other psychiatric disorders
Mood disorders 69.2 (10.20) 55.1 (4.44) 45.7 (4.14) χ2 (2)=4.61 0.10
Anxiety disorders 55.4 (10.06) 47.0 (5.12) 27.1 a,b (4.71) χ2 (2)=9.53 <.01
PTSD 47.9 (9.25) 21.1 a (5.17) 12.6 a (3.91) χ2 (2)=14.48 <.01
Substance use disorders 86.7 (5.49) 58.9 a (5.03) 54.2 a (4.99) χ2(2)=13.01 <.01
Personality disorders 55.5 (9.89) 30.1 a (5.15) 30.9 (4.45) χ2 (2)=7.11 0.03
Conduct disorder 15.3 (8.67) 8.4 (2.74) 9.4 (2.76) χ2 (2)=1.01 0.6
Binge-eating disorder 17.1 (6.60) 7.8 (2.35) 8.7 (3.71) χ2 (2)=2.35 0.31
Bulimia nervosa 17.1 (6.60) 4.5 (1.98) 5.4 (2.65) χ2 (2)=6.39 0.04
Open in a new tab

Notes. BMI = body mass index; PTSD = posttraumatic stress disorder. All analyses were adjusted for NESARC-III complex survey design. 12-month persistence defined as proportion of those with 12-month diagnosis among those with the lifetime diagnosis. Mood disorders included major depressive episodes, persistent depression, and bipolar I; Anxiety disorders included specific phobia, social phobia, panic disorders, agoraphobia, and generalized anxiety disorder; substance use disorders included alcohol use disorder, drug use disorder, and nicotine use disorder; Personality disorders included antisocial, borderline, and schizotypal.

1

= analysis was adjusted for sociodemographic characteristics and duration of AN episode

2

= comparison was made based on log-transformed data;

a

= significantly different from the child-onset group at p < .05 based on comparison of cell or Tukey-Kramer post-hoc comparison

3

= Medians were compared using the Kruskal-Wallis Test with Dwass, Steel, Critchlow-Fligner post-hoc test

b

= significantly different from the adolescent-onset group at p < .05 based on comparison of cell or Tukey-Kramer post-hoc comparison.

Age of first help-seeking (log-transformed for skewness and kurtosis) for low weight was significantly different among three groups, with the child-onset reporting youngest age, followed by the adolescent-onset groups. In terms of specific forms of help-seeking, the child-onset group reported significantly higher rate of hospitalization (19.4%) due to low weight than the adolescent-onset group (4.1%). Adjusting for sociodemographic characteristics and duration of AN episode, odds of reporting hospitalization remained significantly greater for the child-onset group, relative to the adolescent-onset group (AOR = 5.66, 95% CI = 1.28–24.96). The child-onset group also reported higher rates of 12-step groups than the adolescent-onset group; the post-hoc analysis of the two groups did not reach significance).

Rate of reporting four or more ACEs was significantly lower in the adolescent-onset group than the two other groups. Prevalence of lifetime anxiety disorders was significantly higher in the child-onset (55.4%) and the adolescent-onset (47.0%) groups than in the emerging-adult group (27.1%). Adjusting for sociodemographic characteristics and duration of AN episode, odds of lifetime anxiety disorders were significantly greater in the child-onset and adolescent-onset groups, relative to the emerging-adult group (AOR = 4.14, 95% CI = 1.40–12.26, and AOR = 2.48, 95% CI = 1.20–5.09, respectively). Prevalence of PTSD (47.9%) and SUD (86.7%) were significantly higher in the child-onset group than the two older age-of-onset groups (21.1%, 12.6% and 58.9%, 54.2%, respectively). Prevalence of personality disorders was significantly higher in the child-onset than in the adolescent-onset group. Prevalence of bulimia nervosa, but not binge-eating disorder, differed significantly across age onset groups; although lifetime comorbidity with bulimia nervosa was higher in the child-onset relative to the other age-onset groups, post-hoc comparisons revealed no statistically significant differences. When adjusting for sociodemographic characteristics and duration of AN episode, however, odds of lifetime PTSD, SUD, and personality disorders were not significantly different between three groups.

DISCUSSION

Using data from a nationally representative sample of U.S. adults, this study compared clinical profiles of patients with lifetime DSM-5-defined AN who were categorized as child-onset, adolescent-onset, and emerging-adult onset based on their age of onset of AN. When restricted to the 216 patients with onset of AN prior to age 25, adjusted prevalence rates were 11.8%, 39.6%, and 48.6%, respectively for child-, adolescent-, and emerging-adult groups. Our findings suggest that those with child-onset had more severe AN (lower BMI and longer episode duration), greater life difficulties (more adverse childhood experiences and lower educational attainment), and greater lifetime psychiatric comorbidity. Group differences persisted even after adjusting statistically for sociodemographic characteristics and duration of AN episode. Our findings extend the prior literature that focused primarily on comparing child- versus adolescent-onsets of AN by including an emerging-adult group for additional contrast and particularly by taking a broader more comprehensive lifetime perspective of multiple domains beyond the AN profile, including sociodemographic features, adverse childhood experiences, help-seeking and treatment histories, and psychiatric comorbidities. We discuss our findings below with a view towards stimulating future research.

The age of onset findings for AN in this U.S. sample is similar to previous national samples. For example, Hudson et al (2007), in the DSM-IV-defined study in the U.S., reported a mean age of onset of 18.9 (SE=0.8) and median age of 18 (IQR 16–22), and Mohler-Kuo et al (2016), in their DSM-5-defined study in Switzerland, reported a mean age of onset of 17.8 (SE=0.35) and median age of 17 (IQR 16–19.5) whereas our study’s (Udo & Grilo, 2018) mean age of onset was 19.3 (SE=0.06) and median age of 17.4 (IQR 15.2–20.5). With this context in mind, we note the prevalence of “emerging-adult” onset is also consistent with other recent epidemiological research (Nagl et at., 2016). These findings highlight the need for greater research attention devoted to onsets of AN during this developmental era (Potterton et al., 2020) in addition to continued work on understanding continuities from adolescence to adulthood (Lewinsohn et al., 2000; Nagl et al., 2016; Waszezuk et al., 2019). Indeed, this developmental period involving transition towards independence might represent a vulnerable period in terms of decreased recognition by self and others of the development of AN. Indeed, our analyses revealed especially low rates of help-seeking reported by respondents with emerging-adult onsets of AN.

Collectively, our broad characterization and comparison of the three age-of-onset groups of persons with AN converge in suggesting that those with child-onset AN are more likely to have a more severe course of AN (lower BMI and longer episode) and greater life difficulties. This finding persisted even after adjusting for duration of AN episode and sociodemographic variables. The greater severity of AN course in those with child-onset of AN in this national sample is generally consistent with a number of reports with treatment-seeking clinical samples indicating earlier onsets trended towards poorer physical and mental health outcomes (see Herpertz-Dahlmann & Dahmen, 2019). Notably, a report based on data from a German Registry of inpatient hospitalizations (Jaite et al., 2019) suggested a more favorable clinical profile amongst those with child-onset versus adolescent-onset. The reasons for these mixed findings are uncertain; they might reflect possible positive effects of effective intensive treatment delivered quickly and before a longer duration of the illness sets in (Eisler, Dare, Russell, Szmukler, le Grange, & Dodge, 1997).

In terms of life difficulties, child-onset cases reported more adverse childhood experiences, were more likely to be single (un-partnered) and had significantly lower educational attainment. Whereas the adolescent-onset and emerging-adult-onset groups attended college at high rates (80.2% and 76.7%, respectively), which is consistent with high educational achievement reported in a large Danish Registry Study (Dalsgaard et al., 2020), only 62.2% of the child-onset group attended college. The child-onset group and significantly greater lifetime psychiatric comorbidity than the adolescent-onset and emerging-onset groups. In particular, the child-onset group had significantly elevated lifetime prevalence rates of SUD (86.7%), anxiety disorders (55.4%), and PTSD (47.9%). Greater lifetime anxiety disorder comorbidity, but not other psychiatric comorbidities, was observed after adjusting for duration of AN episode and sociodemographic characteristics. Overall, these findings, which overall are consistent with the known complexity and chronicity of AN (Mitchell & Peterson, 2020) and its striking mortality rates (Arcelus, Mitchell, Wales, & Nielsen, 2011), highlight the heightened psychosocial difficulties and life challenges experienced by those with child-onset forms of AN.

A particularly striking finding is the low rates of treatment or help-seeking in this national sample of persons with onsets of AN prior to age 25. Only 40.1% of those with child-onset and 33.7% of those with adolescent onset reported seeking any form of help for their low-weight-related issues. While the rate is even lower for those who were emerging adults when they developed AN, that is perhaps less surprising given the well-known ego-syntonic nature of AN and associated denial. The finding that only 40.1% of child-onset cases resulted in any help-seeking highlights the continued challenges in the early recognition of this serious and often debilitating and lethal disorder (Arcelus et al., 2011) across the entire spectrum of persons who interact with youth. AN continues to be frequently “missed” by patents, caregivers, teachers, coaches, pediatricians, and healthcare workers. Continued research on understanding barriers and facilitators (Ali et al. 2017; Ali et al., 2020; Regan et al., 2017) as well as marked disparities across different groups (Marques et al., 2011; Sonneville et al., 2018) has potential to inform improved recognition by both lay and professional communities. While AN is generally regarded as one of the most difficult to treatment disorders (Mitchell & Peterson, 2020), shorter delays in seeking treatment (Austin et al., 2021) and early intervention (Eisler et al., 1997; Fukutomi et al., 2020; Treasure & Russell, 2011) are associated with better outcomes. Early identification and early referral to “evidence-based” specific interventions seems critical (Kaye & Bulik, 2021).

Our findings should be considered within the context of the methods with their strengths and limitations. Strengths of the study include the use of a large epidemiological data set with a representative sample of 36,309 U.S. adults and broad coverage of important life domains (ACEs, education, several areas of functioning), treatment-seeking behaviors, and lifetime psychiatric disorders. We previously summarized challenges faced by large-scale epidemiological studies and various pros and cons associated with different assessment methods (Udo & Grilo, 2019) and we briefly highlight here several potential limitations as context for our findings and to inform future study designs. The AUDADIS-5, a respondent-based structured interview, does not involve the amount of clinician judgement as some investigator-based semi-structured interviews that might allow for more careful and detailed understanding of respondents’ clinical histories. The AUDADIS-5 was administered by lay interviewers, not by clinicians; the use of experienced mental-health surveyors and standardized training, however, might offset this limitation to some degree. The AUDADIS-5 has been validated for most psychiatric disorders (Grant et al., 2015; Hasin et al., 2015) but has not been evaluated for reliability and validity of ED diagnosis. The skip-out rules and the lack of specific details in questions within the ED sections of the AUDADIS-5 (many but not all available interviews use skip-out rules) might have resulted in possible underestimation of prevalence rates due to missing data and precluded us from investigating other childhood restrictive eating disorders (e.g., avoidant/restrictive food intake disorder; ARFID). The cross-sectional nature of our analyses cannot speak to causality and we emphasize that retrospective recall of many behaviors and events and when they occurred over the lifetime is challenging and might be inaccurate, biased, or influenced by the passage of time and intervening events. The assessment of AN is further complicated, in some instances, by its ego-syntonic nature, and in many instances with associated secrecy, denial, and resistance to treatment-seeking. In these regards, it is possible that the anonymous and the purely research focus of NESARC-III coupled with its complete separation from any treatment agenda might have facilitated honest disclosures about the sensitive topics.

In conclusion, our findings, based on a nationally representative sample of U.S. adults with lifetime diagnoses of AN, suggest that those with child-onset had more severe AN, greater life difficulties, and greater lifetime psychiatric comorbidity than those with adolescent- or emerging-adult onsets and that this was not merely attributable to longer duration of episode. Our findings highlight the low rates of help-seeking in persons with AN and this is striking given the high levels of associated difficulties. Findings emphasize the importance of earlier recognition and rapid referral to effective treatments.

Funding:

The article was prepared using a limited access dataset obtained from the National Institute on Alcohol Abuse and Alcoholism. Dr. Grilo was supported, in part, by grants from the National Institutes of Health (R01 DK114075, R01 DK112771, R01 DK49587). This article does not reflect the opinions or views of the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute on Alcohol Abuse and Alcoholism, or the United States Government.

Footnotes

Disclosures: Drs. Grilo and Udo report no conflicts of interest. Although Dr. Grilo reports no relevant direct or indirect conflicts of interest with respect to this study, he reports during the past 12 months having received honoraria for CME-related lectures and royalties from academic books published by Guilford Press and Taylor & Francis Publishers.

Data Availability Statement:

The data that support the findings of this study are available from the National Institute on Alcohol Abuse and Alcoholism (NIAAA). Restrictions apply to the availability of these data, which were used under license for this study. Data are available (with permission of NIAAA): https://www.niaaa.nih.gov/research/nesarc-iii/nesarc-iii-data-access.

REFERENCES

  1. Ali K, Farrer L, Fassnacht DB, Gulliver A, Bauer S & Griffiths KM (2017). Perceived barriers and facilitators towards help-seeking for eating disorders: a systematic review. International Journal of Eating Disorders, 50, 9–21. [DOI] [PubMed] [Google Scholar]
  2. Ali K, Fassnacht DB, Farrer L, Rieger E, Feldhege J, Moessner M, Griffiths KM, & Bauer S (2020). What prevents young adults from seeking help? Barriers towards help-seeking for eating disorder symptomatology. International Journal of Eating Disorders, 53, 894–906. [DOI] [PubMed] [Google Scholar]
  3. Arcelus J, Mitchell AJ, Wales J, & Nielsen S (2011). Mortality rates in patients with anorexia nervosa and other eating disorders: a meta-analysis of 36 studies. Archives of General Psychiatry, 68, 724–731. [DOI] [PubMed] [Google Scholar]
  4. Austin A, Flynn M, Richards K, Hodsoll J, Duarte TA, Robinson P, Kelly J, & Schmidt U (2021). Duration of untreated eating disorder and relationship to outcomes: a systematic review of the literature. European Eating Disorders Review, 29, 329–345. [DOI] [PubMed] [Google Scholar]
  5. Coffino JA, Udo T, & Grilo CM (2019). Rates of help-seeking in US adults with lifetime DSM-5 eating disorders: prevalence across diagnoses and differences by sex and ethncity/race. Mayo Clinic Proceedings, 94, 1415–1426. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. du Fort GG, Newman SC, & Bland RC (1993). Psychiatric comorbidity and treatment seeking. Sources of selection bias in the study of clinical populations. Journal of Nervous and Mental Disease, 181, 467–74. [PubMed] [Google Scholar]
  7. Dalsgaard S, McGrath J, Ostergaard SD, Wray NR, Pedersen C, Mortensen P, & Peterson L (2020). Association of mental disorder in childhood and adolescence with subsequent educational achievement. JAMA Psychiatry, 77, 797–805. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Eisler I, Dare C, Russell GFM, Szmukler GI, le Grange D, & Dodge E (1997). Family and individual therapy in anorexia nervosa: a five-year follow-up. Archives of General Psychiatry, 54, 1025–1030. [DOI] [PubMed] [Google Scholar]
  9. Fukutomi A, Austin A, McClelland J, Brown A, Glennon D, Mountford V, Grant N, Allen K, & Schmidt U (2020). First episode rapid intervention for eating disorders: a two-year follow-up. Early Intervention in Psychiatry, 14, 137–141. [DOI] [PubMed] [Google Scholar]
  10. Grant BF, Chu A, Sigman RAM, Kali J, Sugawara Y, Jiao R, . . . Goldstein R (2014). Sources and Accuracy Statement: National Epidemiologic Survey on Alcohol and Related Conditions-III (NESARC-III). Rockville, MD: Retrieved from http://www.niaaa.nih.gov/sites/default/files/NESARC_Final_Report_FINAL_1_8_15.pdf [Google Scholar]
  11. Grant BF, Goldstein RB, Chou SP, Saha TD, Ruan WJ, Huang B, . . . Hasin DS (2011). The Alcohol Use Disorder and Associated Disabilities Interview Schedule-Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition Version (AUDADIS-5) Rockville, MD. [Google Scholar]
  12. Grant BF, Goldstein RB, Smith SM, Jung J, Zhang H, Chou SP, . . . Hasin DS (2015). The Alcohol Use Disorder and Associated Disabilities Interview Schedule-5 (AUDADIS-5): Reliability of substance use and psychiatric disorder modules in a general population sample. Drug and Alcohol Dependence, 148, 27–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Grant BF, Saha TD, Ruan WJ, Goldstein RB, Chou SP, Jung J, . . . Hasin DS (2016). Epidemiology of DSM-5 drug use disorder: results from the National Epidemiologic Survey Alcohol and Related Conditions-III. JAMA Psychiatry, 73, 39–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hart LM, Granillo MT, Jorm AF, & Paxton SJ (2011). Unmet need for treatment in the eating disorders: a systematic review of eating disorder specific treatment seeking among community cases. Clinical Psychology Review, 31, 727–735. [DOI] [PubMed] [Google Scholar]
  15. Hasin DS, Greenstein E, Aivadyan C, Stohl M, Aharonovich E, Saha T, . . . Grant BF (2015). The Alcohol Use Disorder and Associated Disabilities Interview Schedule-5 (AUDADIS-5): Procedural validity of substance use disorders modules through clinical re-appraisal in a general population sample. Drug and Alcohol Dependence, 148, 40–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hasin DS, Shmulewitz D, Stohl M, Greenstein E, Aivadyan C, Morita K, . . . Zhang H (2015). Procedural validity of the AUDADIS-5 depression, anxiety and post-traumatic stress disorder modules: Substance abusers and others in the general population. Drug and Alcohol Dependence, 152, 246–256. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Herpertz-Dahlmann B, Dempfle A, Egberts KM, Kappel V, Konrad K, Vloet JA, & Buhren K (2018). Outcome of childhood anorexia nervosa – the results of a five- to ten-year follow-up study. International Journal of Eating Disorders, 51, 295–304. [DOI] [PubMed] [Google Scholar]
  18. Herpertz-Dahlmann B, & Dahmen B (2019). Children in need – diagnostics, epidemiology, treatment and outcome of early onset anorexia nervosa. Nutrients, 11, 1932. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hoek HW (2016). Review of the worldwide epidemiology of eating disorders. Current Opinion in Psychiatry, 29, 336–339. [DOI] [PubMed] [Google Scholar]
  20. Jaite C, Buhren K, Dahmen B, Demple A, Becker K, Correll CU, et al. (2019). Clinical characteristics of inpatients with childhood vs. adolescent anorexia nervosa. Nutrients, 11, 2593. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Keski-Rahkonen A, Hoek HW, Susser ES, Linna ML, Sihvola E, Raevuori A, Bulik CM, Kaprio J, & Rissanen A (2007). Epidemiology and course of anorexia nervosa in the community. American Journal of Psychiatry, 164, 1259–1265. [DOI] [PubMed] [Google Scholar]
  22. Kwok C, Kwok V, Lee HY, & Tan SM (2020). Clinical and socio-demographic features in childhood vs. adolescent-onset anorexia nervosa in an Asian population. Eating and Weight Disorders, 25, 821–826. [DOI] [PubMed] [Google Scholar]
  23. Laliberte V, Giguere CE, Potvin S, Lesage A, & Signature Consortium. (2020). Berkeson’s bias in biobank sampling in a specialised mental health care setiing: a comparative cross-sectional study. BMJ Open, 10, e035088. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lewinsohn PM, Striegel-Moore R, & Seeley JR (2000). Epidemiology and natural course of eating disorders in young women from adolescence to young adulthood. Journal of the American Academy of Child and Adolescent Psychiatry, 39, 1284–1292. [DOI] [PubMed] [Google Scholar]
  25. Madden S, Morris A, Zurynski YA, Kohn M, & Elliot EJ (2009). Burden of eating disorders in 5-to-13-year-old children in Australia. Medical Journal of Australia, 190, 410–414. [DOI] [PubMed] [Google Scholar]
  26. Marascuilo LA, Serlin RC (1988). Statistical methods for the social and behavioral sciences. New York: W. H. Freeman. [Google Scholar]
  27. Marques L, Alegria M, Becker AE, et al. (2011). Comparative prevalence, correlates of impairment, and service utilization for eating disorders across US ethnic groups: implications for reducing ethnic disparities in health care access for eating disorders. International Journal of Eating Disorders, 44, 412–420. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Mitchell JE, & Peterson CB (2020). Anorexia nervosa. New England Journal of Medicine, 382, 1343–1351. [DOI] [PubMed] [Google Scholar]
  29. Mohler-Kuo M, Schnyder U, Dermota P, Wei W, & Milos G (2016). The prevalence, correlates, and help-seeking of eating disorders in Switerland. Psychological Medicine, 46, 2749–2758. [DOI] [PubMed] [Google Scholar]
  30. Nagl M, Jacobi C, Paul M, Beesdo-Baum K, Hofler M, Lieb R, & Wittchen H-U. (2016). Prevalence, incidence, and natural course of anorexia nervosa and bulimia nervosa among adolescents and yound adults. European Child Adolescent Psychiatry, 25, 903–918. [DOI] [PubMed] [Google Scholar]
  31. Neubauer K, Weigel A, Daubmann A, Wendt H, Rossi M, Lowe B & Gumz A (2014). Paths to first treatment and duration of untreated illness in anorexia nervosa: are there differences according to age of onset. European Eating Disorders Review, 22, 292–298. [DOI] [PubMed] [Google Scholar]
  32. Nicholls DE, Lynn R, & Viner RM (2011). Childhood eating disorders: British national surveillance study. British Journal of Psychiatry, 198, 295–301. [DOI] [PubMed] [Google Scholar]
  33. Peebles R; Wilson JL, Lock JD (2006). How do children with eating disorders differ from adolescents with eating disorders at initial evaluation? Journal Adolescent Health, 39, 800–805. [DOI] [PubMed] [Google Scholar]
  34. Pinhas L, Morris A, Crosby RD, & Katzman DK (2011). Incidence and age-specific presentation of restrictive eating disorders in children. Archives Pediatrics Adolescent Medicine, 165, 895–899. [DOI] [PubMed] [Google Scholar]
  35. Potterton R, Austin A, Allen K, Lawrence V, & Schmidt U (2020). “I’m not a teenager, I’m 22. Why can’t I snap out of it?”: a qualitative exploration of seeking help for a first-episode eating disorder during emerging adulthood. Journal of Eating Disorders, 8, 46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Regan P, Cachelin FM, & Minnick AM (2017). Initial treatment seeking from professional health care providers for eating disorders: a review and synthesis of potential barriers to and facilitators of “first” contact.” International Journal of Eating Disorders, 50, 190–209. [DOI] [PubMed] [Google Scholar]
  37. Sharpe D (2015). Your chi-square test is statistically significant: Now what? Practical Assessment, Research & Evaluation, 20(8), 1–10. [Google Scholar]
  38. Silen Y, Sipila PN, Raevuori A, Mustelin L, Marttunen M, Kaprio J, & Keski-Rahkonen A (2020). DSM-5 eating disorders among adolescents and young adults in Finland: a public health concern. International Journal of Eating Disorders, 53, 790–801. [DOI] [PubMed] [Google Scholar]
  39. Sonneville KR, & Lipson SK (2018). Disparities in eating disorder diagnosis and treatment according to weight status, race/ethnicity, socioeconomic background, and sex among college students. International Journal of Eating Disorders, 51, 518–526. [DOI] [PubMed] [Google Scholar]
  40. Steinhausen H-C., & Jensen CM (2015). Time trends in lifetime incidence rates of first-time diagnosed anorexia nervosa and bulimia nervosa across 16 years in a Danish nationwide psychiatric registry study. International Journal of Eating Disorders, 48, 845–850. [DOI] [PubMed] [Google Scholar]
  41. Swanson SA, Crow SJ, Le Grange D, Swendsen J, & Merikangas KR (2011). Prevalence and correlates of eating disorders in adolescents: results from the national comorbidity survey replication adolescent supplement. Archives of General Psychiatry, 68, 714–723. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Treasure J, & Russell G (2011). The case for early intervention in anorexia nervosa: theoretical exploration of maintaining factors. British Journal of Psychiatry, 199, 5–7. [DOI] [PubMed] [Google Scholar]
  43. Udo T (2019). Chronic medical conditions in U.S. adults with incarceration history. Health Psychology, 38, 217–225. [DOI] [PubMed] [Google Scholar]
  44. Udo T, & Grilo CM (2016). Perceived weight discrimination, childhood maltreatment, and weight gain in U.S. adults with overweight/obesity. Obesity, 24, 1366–1372. [DOI] [PubMed] [Google Scholar]
  45. Udo T, & Grilo CM (2018). Prevalence and correlates of DSM-5-defined eating disorders in a nationally representative sample of U.S. adults. Biological Psychiatry, 84, 345–354. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Udo T, & Grilo CM (2019). Reply to: Insight may limit identification of eating disorders. Biological Psychiatry, 85, e57–e58. [DOI] [PubMed] [Google Scholar]
  47. Wade TD, Bergin JL, Tiggemann M, Bulik CM, & Fairburn CG (2006). Prevalence and long-term course of lifetime eating disorders in an adult Australian twin cohort. Australian and New Zealand Journal of Psychiatry, 40, 121–128. [DOI] [PubMed] [Google Scholar]
  48. Wang PS, Berglund P, Olfson M, Pincus HA, Wells KB, & Kessler RC (2005). Failure and delay in initial treatment contact after first onset of mental disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62, 603–613. [DOI] [PubMed] [Google Scholar]
  49. Waszczuk MA, Waaktaar T, Eley TC, & Torgersen S (2019). Etiological influences on continuity and co-occurrence of eating disorders symptoms across adolescence and emerging adulthood. International Journal of Eating Disorders, 52, 554–563. [DOI] [PubMed] [Google Scholar]
  50. Wentz E, Gillberg IC, Anckarsater H, Gillberg C, & Rastam M (2009). Adolescent-onset Anorexia nervosa: 18-year outcome. British Journal of Psychiatry, 194, 168–174. [DOI] [PubMed] [Google Scholar]
  51. Wong MNT, & Hay P (2020). Exploring associations between age of onset and quality of life of people with eating disorder behaviours and weight/shape overvaluation: a general population study. Australasian Psychiatry, 28, 660–663. [DOI] [PubMed] [Google Scholar]
  52. Wood D, Crapnell T, Lau L, Bennett A, Lotstein D, Ferris M, & Kuo A (2018). Emerging adulthood as a critical stage in the life course. In Halfon N (ed), Handbook of life course health development (pages 123–143). [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The data that support the findings of this study are available from the National Institute on Alcohol Abuse and Alcoholism (NIAAA). Restrictions apply to the availability of these data, which were used under license for this study. Data are available (with permission of NIAAA): https://www.niaaa.nih.gov/research/nesarc-iii/nesarc-iii-data-access.

RESOURCES