Personality disorders, alcohol use, and alcohol misuse.

Social Science & Medicine 120 (2014) 286e300

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

Personality disorders, alcohol use, and alcohol misuse Johanna Catherine Maclean a, *, Michael T. French b a

Department of Economics, Temple University, USA Health Economics Research Group, Department of Sociology, Department of Economics, and Department of Public Health Sciences, University of Miami, USA

b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 11 March 2014 Received in revised form 10 September 2014 Accepted 16 September 2014 Available online 18 September 2014

Personality disorders (PDs) are psychiatric conditions that manifest early in life from a mixture of genetics and environment, are highly persistent, and lead to substantial dysfunction for the affected individual and those with whom s/he interacts. In this study we offer new information on the associations between PDs and alcohol use/misuse. Specifically, we consider all 10 PDs recognized by the American Psychiatric Association; carefully address important sources of bias in our regression models; and study heterogeneity across PDs, drinking pattern, and gender. To investigate the relationships between PDs and alcohol consumption we analyze data from the 2004/2005 National Epidemiological Survey of Alcohol and Related Conditions (N ¼ 34,653). We construct measures of any drinking, drinking quantity, and patterns of misuse that could lead to significant social costs including drinking to intoxication, driving after drinking, drinking during the day, and alcohol abuse/dependence. Results show that persons with PDs are significantly more likely to use and misuse alcohol, although associations vary across gender. Moreover, antisocial, borderline, histrionic, and narcissistic PDs display the strongest links with alcohol use and misuse, and the relationships are strongest among the heaviest drinkers. These findings have important public health implications and underscore the potential social costs associated with mental health conditions. © 2014 Elsevier Ltd. All rights reserved.

Keywords: United states Personality disorders Alcohol use and misuse Mental health Social costs

1. Introduction This study examines associations between personality disorders (PDs), alcohol use, and alcohol misuse. To this end, we analyze data from the 2004/2005 National Epidemiological Survey of Alcohol and Related Conditions (NESARC), a large and nationally representative data set specifically designed to study alcohol use/misuse and mental health conditions such as PDs. The present study considers all 10 PDs recognized by the American Psychiatric Association (APA); addresses potential sources of bias in regression models; and explores heterogeneity across PDs, drinking patterns, and gender. PDs are a class of psychiatric disorders that lead to diminished social functioning and impose substantial costs on both the person with a PD and those with whom s/he interacts. As defined by the American Psychiatric Association's (APA) Diagnostic and Statistical Manual of Mental Disorders (DSM) (2000), PDs are ‘‘pervasive, inflexible and enduring patterns of inner experiences and behavior

* Corresponding author. E-mail address: [email protected] (J.C. Maclean). http://dx.doi.org/10.1016/j.socscimed.2014.09.029 0277-9536/© 2014 Elsevier Ltd. All rights reserved.

that can lead to clinically significant distress or impairment in social, occupational, or other areas of functioning.” The psychiatric literature attributes the development of PDs to a confluence of genetics and early childhood environment (American Psychiatric Association, 2000; Yudofsky, 2005). Because PDs manifest early in life (childhood or early adolescence) and are exceedingly difficult to treat (American Psychiatric Association, 2000; Yudofsky, 2005), they are considered lifetime conditions, unlike episodic mental health disorders (e.g., depression). Put differently, once an individual is diagnosed with PD, s/he is expected to suffer from this disorder for the rest of her life. It is widely reported that alcohol misuse causes negative externalities for society. Due to market imperfections, drinkers who misuse alcohol do not fully internalize the cost of their actions and thereby potentially impose costs on others through motor vehicle accidents (Lovenheim and Steefel, 2011), increased use of publically-provided health care and addiction treatment services (Balsa et al., 2009; Levit et al., 2008), suicide attempts (Chatterji et al., 2004), domestic abuse (Markowitz and Grossman, 2000), crime (Carpenter, 2007), and reduced productivity in the labor market (Terza, 2002). As a result, the direct medical, crime, and labor market costs of alcohol misuse in the U.S. amount to

J.C. Maclean, M.T. French / Social Science & Medicine 120 (2014) 286e300

approximately $259 billion annually (Bouchery et al., 2011). The full burden of alcohol misuse is probably much higher as the negative consequences also impact friends, family members, and coworkers. For example, Balsa (2008) finds that parental alcohol misuse often leads to poor labor market outcomes for their children. Thus, understanding determinants of alcohol misuse, and utilizing this information to minimize associated social costs, could lead to substantial welfare gains for both current and future generations. 2. Methods To study relationships between PDs and alcohol use/misuse, we analyze data from the National Epidemiological Survey of Alcohol and Related Conditions (NESARC) and model measures of past year alcohol use (any drinking, number of drinks conditional on any drinking) and misuse (weekly drinking to intoxication, any driving after drinking, weekly daytime drinking, alcohol abuse/dependence) as a function of PDs. We utilize detailed information in the NESARC to study the importance of meeting diagnostic criteria for any PD as well as each of the PDs recognized by the APA. We first describe the etiology of PDs and discuss how these disorders are potentially related to alcohol use and misuse. Next, we introduce our conceptual framework, data, and empirical models. 2.1. Background on personality disorders To be diagnosed with a PD, an individual must exhibit ‘‘an enduring pattern of inner experience and behavior that deviates markedly from the expectations of the individual's culture” (APA, 2000). The pattern must be inflexible and pervasive across a broad range of personal and social situations; must lead to clinically significant impairment in social, occupational, or other important areas; must be stable and of long duration, with onset traceable back to adolescence or at least early adulthood; and cannot be a consequence of another medical condition. The DSM divides PDs into three clusters. Cluster A includes paranoid, schizoid, and schizotypal PDs. People with Cluster A disorders are often viewed as odd; speak, think, and act in strange ways; and have difficulty relating to others. Cluster B includes antisocial, borderline, histrionic, and narcissistic PDs. People with Cluster B disorders tend to act in dramatic, hostile, and erratic fashions; have difficulty with impulsive behavior; and violate social norms. Cluster C includes avoidant, dependent, and obsessivecompulsive PDs. People with Cluster C disorders are regularly anxious, fearful, and afraid of social interactions and of feeling out of control. Appendix Table A offers a summary of PD traits. Based on their defining characteristics, PDs could be related to alcohol use and misuse, and the relationships likely differ across PD types. For example, borderline PD is associated with impulsivity and reckless behaviors while a defining feature of narcissistic PD is a lack of empathy for others and extreme self-interest. Relatedly, those who suffer from antisocial PD feel no remorse for their actions. Persons affected by histrionic PD have problems delaying gratification. These features may lead to alcohol use and perhaps misuse as a coping or self-medicating mechanism. Alternatively, some PDs may protect against alcohol misuse. For example, those who suffer from schizoid and schizotypal PDs shun activities that require personal interactions while those who suffer from paranoid PD are deeply distrustful of others. Persons affected by these conditions may avoid social situations where alcohol use is common (e.g., bars). It is beyond the scope of this study to fully articulate and explore all pathways through which PDs could conceptually be associated with alcohol use/misuse, but this overview suggests

287

potentially strong, yet complex, relationships between PDs and alcohol use/misuse. 2.2. Conceptual framework Almond and Currie (2011) propose a health production function that forms the conceptual foundation of our research. This model permits health investmentsdboth health harming and promotingdduring childhood to have a sustained impact on adult health outcomes. In combination with the psychiatric literature, the AC model provides a useful framework within which to understand the impact of PDs on health in general and alcohol use/misuse by extension (Maclean et al., 2014). In the AC model, an individual's lifespan is divided into two periods: 1) childhood and 2) post childhood. This structure is formalized through a linear health production function:

h ¼ A½gI1 þ ð1 gÞI2

(1)

where h is health in adulthood (i.e., post childhood), I1 is health investments made in childhood, I2 is health investments made post childhood, A is a shift parameter, and g is a share parameter, which ranges from zero to one and captures the relative weights of I1 and I2. If g s 0.5 both the level and the timing of health investments are important for adult health. For example, if g > 0.5, then health investments that occur in childhood yield higher returns than health investments occurring later in life. If Ag > 1, adult health (h) is affected more than proportionally by childhood health investments (I1). Thus, this framework allows health investments received in childhood to have a sustained impact on adult health. Because PDs develop early in life as a result of genetics and early childhood environment, in the AC framework they can be viewed as a form of negative health investment, which can persistently impact health and health behaviors into adulthood. 2.3. Data The National Epidemiological Survey of Alcohol and Related Conditions (NESARC) is a nationally representative survey conducted by the U.S. Bureau of the Census for the National Institute on Alcohol Abuse and Alcoholism (NIAAA). These data are widely utilized to study mental health and substance use (Grant et al., 2004; Hasin et al., 2011, 2007). The NESARC data are deidentified, contain no personal identifying information, and are thus exempt from human subjects review. Interested readers can consult with Grant et al. (2003) for more details on the data set. We utilize Wave II of the NESARC (fielded in 2004/2005) supplemented with Wave I (fielded in 2001/2002) data for selected PDs that were not re-measured in Wave II. Subjects were interviewed face-to-face through computer-assisted personal interviewing and 34,653 individuals of the 43,093 original respondents completed Wave II. Respondents were age 20 years and older at Wave II. The NESARC is particularly well-suited for our research as it is specifically designed to measure alcohol misuse and psychiatric disorders including PDs, contains a rich set of personal characteristics, and includes all ten PDs recognized by the APA. We exclude respondents younger than 21 years because they are unable to legally consume alcohol and respondents above 64 years because they are entering retirement, which alters alcohol consumption decisions due to health challenges and changes in social networks (Moos et al., 2005). After further excluding respondents with missing data on certain analysis variables (detailed in a later section), our sample includes 11,497 men and 15,199 women.

288


2.4. Outcome variables We examine several measures of past year alcohol use and misuse using Wave II data. First, we construct an indicator for consuming 12 or more drinks during the past year as well as a continuous measure for the conditional number of drinks (conditional on consuming at least 1 drink in the past year). These measures capture common patterns of alcohol use that are not typically associated with negative externalities. Because we take the natural logarithm of the number of conditional drinks to address skewness, coefficient estimates have an approximate percentage change interpretation. We next construct a set of dummy variables to capture alcohol use patterns that are more likely to lead to negative externalities. These indicators (coded one if the respondent reports the alcohol use pattern, zero otherwise) include weekly drinking to selfreported intoxication, any driving after drinking three or more drinks, weekly drinking before 3 pm, and meeting the DSM criteria for alcohol abuse and/or dependence. Previous studies have used comparable drinking measures in the NESARC (Baldwin and Marcus, 2014; Davalos et al., 2012; French et al., 2010; Keyes et al., 2011; Popovici and French, 2013; Slopen et al., 2011). 2.5. Personality disorders The explanatory variables of primary interest are PDs. NESARC administrators utilized the Alcohol Use Disorder and Associated Disabilities Schedule (AUDADIS) DSM-IV to classify respondents as meeting criteria for ten PDs recognized by the APA. The validity of the AUDADIS is well documented (Ruan et al., 2008) and this instrument is commonly utilized to diagnose psychiatric disorders, including PDs, in survey data (Blanco et al., 2013; Hasin et al., 2011, 2007). NESARC respondents used laptop computers to answer a series of questions on lifetime behaviors. NIAAA epidemiologists later applied the AUDADIS algorithm to the completed surveys and determined whether a respondent met criteria for each specific PD. To receive a particular disorder diagnosis, NESARC respondents must have endorsed a requisite number of symptoms pertaining to the given PD (e.g., at least four of the seven criteria for avoidant PD), with a least one symptom causing social and/or occupational dysfunction. Seven PDs are measured in Wave I (antisocial, avoidant, dependent, obsessive-compulsive, paranoid, schizoid, histrionic) and four are measured in Wave II (antisocial, schizotypal, narcissistic, borderline). To examine all ten PDs, our measures of avoidant, dependent, obsessive-compulsive, paranoid, schizoid, and histrionic PDs are generated using data from Wave I while our measures of schizotypal, narcissistic, and borderline PDs are generated using data from Wave II. We assume that if a PD was present in Wave I it was also present in Wave II. This assumption is consistent with the persistent nature of PDs. Antisocial is the only PD assessed in both Waves and we code respondents as meeting the antisocial PD criteria if NESARC administrators classify them with this disorder in either Wave. The correlation between antisocial PD in Waves I and II is 97%. We construct two different types of PD measures for our empirical models. First, we define a dichotomous measure for any PD, coded one if the respondent meets the criteria for any of the ten PDs measured in the NESARC, and zero otherwise. Second, we include unique indicators for each of the ten (non-mutually exclusive) PDs. Analysis of individual PDs can shed light on how the relationships between PDs and alcohol use/misuse vary across disorders. In May 2013 the APA released the updated DSM-5. Although NESARC administrators utilized the DSM-IV to classify PDs, all 10

PDs that we study are included in DSM-5. Moreover, their definitions are largely unchanged. We confirmed the symmetry between PD measures based on DSM-IV and DSM-5 in conversations with Dr. David Kupfer, a clinical psychologist and Chair of the DSM-5 Taskforce (4/23/2013). In addition, the DSM-5 combined the alcohol abuse and dependence diagnoses into one diagnosis. However, this change does not affect our study as we already combine alcohol abuse and/or dependence into one outcome. 2.6. Other control variables We control for a detailed set of alcohol use/misuse predictors in all regressions. These variables are constructed using Wave II data. We include age, race/ethnicity, birth outside the U.S., household income, marital status, presence of a child under age 18 in the household, education, employment, tobacco use, illicit drug use (the NESARC collects information on 10 illicit drugs, and we code a respondent as using illicit drugs if they report use of any of these substances in the past year), and physical health (measured by the SF-12 physical component score [PCS]; higher scores indicate better health). Because the NESARC contains substantially more detailed background and mental health information than is typically found in social science surveys, we are able to include proxies for genetic predisposition for alcohol misuse (before age 18 a parent or caregiver with an alcohol problem), family background (before age 18 a parent or caregiver incarcerated, a parent or caregiver hospitalized for a mental illness, family received money from government assistance programs, lived with at least one biological parent, exposed to domestic violence as measured by a father hitting a mother), and experiencing abuse by a parent or caregiver (before age 18 verbal, physical, or sexual abuse). We also include lifetime episodic mental health disorders: mood disorders (depressive, manic, hypomanic, dysthymic disorder) and anxiety disorders (panic, agoraphobia, social phobia, generalized anxiety). 2.7. Empirical models To estimate the associations between PDs and alcohol use/ misuse, we estimate the following regression model:

Ai ¼ b0 þ PDi b1 þ Xi b2 þ EXi b3 þ εi

(2)

where Ai is a measure of alcohol use or misuse, PDi represents any PD or unique indicators for each of the ten specific PDs, Xi is a vector of common individual characteristics, and EXi is a vector of individual characteristics that are typically not included in social science surveys. The b's are parameters to estimate and εi is a random error term. We employ a probit model when alcohol use or misuse is dichotomous and ordinary least squares (OLS) regression when estimating the logarithm of the number of drinks (conditional on at least 1 drink). In probit models, we report average marginal effects. We apply NESARC survey weights and cluster standard errors around the primary sampling unit. As noted earlier, we estimate models separately by gender given different alcohol use/misuse patterns between men and women (Naimi et al., 2003) and different risk for PDs, and types of PDs, across these groups (Golomb et al., 1995; Paris, 2004). In unreported analyses available from the corresponding author, we interact the PD variables with an indicator for male gender and test the statistical significance of the interaction terms in pooled regression models that include both men and women. The majority of pooled regressions indicate that gender-specific models are warranted. Moreover, our gender-specific findings are easier to compare to the existing literature as estimating gender-specific


models is common in social science analyses of alcohol use/misuse (Balsa and French, 2010; Davalos et al., 2012; Maclean, 2014; Mullahy and Sindelar, 1996; Popovici and French, 2013). Nevertheless, we also report results based on the pooled sample of men and women in Appendix Table B (with an added dummy variable for male gender). Comparisons with the core findings (reported later) suggest that the pooled sample generates estimates that blend the male- and female-specific estimates, thus masking gender-specific heterogeneity. To explore heterogeneity in the relationships between PDs and alcohol use across the drinking distribution, we apply unconditional quantile regression or UQR (Firpo et al., 2009). UQR allows consistent estimates of treatment effects at virtually any quantile of the unconditional distribution (quantiles are points taken at regular intervals from the cumulative distribution function of a random variable) and may uncover heterogeneity in relationships between PDs and a continuous measure of alcohol use that are masked by OLS (Manning et al., 1995). To examine statistical significance, we rely on parametric bootstrapped standard errors with 400 repetitions. In extensions to the core analyses, we assess the robustness of our findings. First, Trull and colleagues critique the NESARC PD classification algorithm and develop more restrictive diagnostic criteria to classify PDs using the NESARC data (Trull et al., 2010). Both the NESARC and Trull algorithms require the respondent to endorse the requisite number of symptoms for the specific disorder (e.g., at least four of the seven symptoms for avoidant PD). The difference is that in the Trull algorithm all symptoms (not just one as in the NESARC algorithm) must cause social or occupational dysfunction. Thus, the NESARC algorithm will code more persons with a PD than those designated by the Trull algorithm. As a sensitivity check we implement the Trull algorithm to classify PDs and re-estimate all models. Second, we re-classify our ten PDs into the three clusters recognized by the APA (Clusters A, B, and C). Third, a common concern with longitudinal data such as the NESARC is non-random attrition. Respondents who attrite between Waves I and II may be inherently different from respondents who complete both Waves. Such non-random attrition can lead to biased estimates. To assess this potential data issue, we compare the Wave I PD status of Wave II attritors and completers. Lastly, we consider potential bias in our estimates from reverse causality and over-controlling. In particular, we limit our sample based on age to address reverse causality among those respondents who may not fully develop a PD until later in life, and we utilize alternative covariate sets in our regression models and compare estimated associations. 3. Results 3.1. Summary statistics Table 1 presents summary statistics for each gender-specific sample, stratified by PD status. For brevity, we combine parental alcohol problems, mental illness hospitalizations, and incarceration; verbal, physical, and sexual abuse; and lifetime mood and anxiety disorders into collective groups. Among men, 76% report any drinking in the past year and the mean number of conditional drinks consumed during this period is 440. 4.9%, 21%, and 4.5% of men report weekly drinking to intoxication, any driving after drinking 3 or more drinks, and weekly drinking before 3 pm. 17% of men meet the DSM criteria for alcohol abuse and/or dependence. Among women, 67% report any drinking in the past year and the conditional mean number of drinks consumed during this period is 214. 1.5%, 7.2%, and 1.1% report weekly drinking to intoxication, any driving after drinking 3 or more drinks, and weekly drinking before

289

3 pm, while 6.6% meet the DSM criteria for alcohol abuse/dependence. Again, gender-specific differences in these summary statistics highlight the importance of running separate models for men and women. 25% of men and 22% of women in our sample meet the DSM criteria for any PD. Obsessive-compulsive is the most common PD among both men and women in our sample. All proportions and means for the alcohol use/misuse measures are higher among those with a PD than without. For example, 6.3% of men with a PD report weekly drinking to intoxication while just 3.9% of those without a PD report this behavior. Individuals who suffer from a PD are less likely to hold a college degree, less likely to be married, and less likely to work full time. Moreover, they are more likely to smoke and use illicit drugs, have lower physical functioning, have higher genetic predisposition for alcohol misuse, report higher levels of family dysfunction before age 18, and are more likely to suffer from mood and anxiety disorders. In unreported analyses available from the corresponding author, we conduct c2 - (binary variables) or t(continuous variables) tests to determine the statistical significance of the differences between respondents with and without PDs. For the majority of variables, the differences are statistically different from zero at the p 0.05 level or better. 3.2. Regression results As reported earlier, individuals with PDs are more likely to exhibit alcohol misuse than persons without PDs and are less advantaged in terms of socioeconomic status. Tables 2 (men) and 3 (women) report selected regression results from the alcohol use/misuse regressions. In both tables, the top panel pertains to specifications that include a measure of any PD and the bottom panel applies to models with indicators for the ten unique PDs. Appendix Tables C (men) and D (women) present the full set of estimates for models that include a single dummy variable for any PD. Among men, we find evidence that suffering from a PD is associated with a higher risk for some, but not all, patterns of alcohol use/misuse. Male respondents who meet criteria for any PD are 2.1 and 4.7 percentage points (10% and 27.6%) more likely to drive after drinking 3 or more drinks and to receive a diagnosis for alcohol abuse and/or dependence (the coefficient in the driving after drinking 3 or more drinks regression approaches statistical significance with p 0.10). Although results also suggest that men with a PD are more likely to report weekly drinking to intoxication and weekly drinking before 3 pm, and consume more alcohol during the past year, these estimates are not statistically significant. The estimated marginal effect for consuming 12 or more drinks in the past year is negative, small in size, and non-significant. In disaggregated models that include the ten specific PDs, antisocial, borderline, and narcissistic PDs are associated with a higher probability of alcohol use and misuse. For example, meeting the criteria for antisocial PD (recall that a defining feature of this disorder is a disregard for the safely of oneself and others) is associated with a 17.6% increase in the conditional number of drinks consumed in the past year and a 2.3 and 1.3 percentage point (46.9% and 28.9%) increase in the probability of weekly drinking to intoxication and weekly drinking before 3 pm (the latter coefficient estimate approaches statistical significance with p 0.10). Narcissistic PD (which is defined by lack of empathy for others and extreme self-interest) is associated with an increased probability of any driving after drinking 3 or more drinks, weekly drinking before 3 pm (approaching statistical significance with p 0.10), and meeting the DSM criteria for alcohol abuse and/or dependence (5.1, 1.2, and 4.7 percentage points or 24.3%, 26.7%, and 27.6%). Although usually not statistically significant, avoidant, paranoid, schizoid,

290


Table 1 Summary statistics. Men Full sample Past year alcohol use Drink 12þ drinks Number of drinks (conditional sample) Weekly drinking to intoxication Any driving after drinking 3þ drinks Weekly drinking before 3 pm Abuse and/or dependence Personality disorders Any PD Paranoid Schizoid Schizotypal Antisocial Borderline Histrionic Narcissistic Avoidant Dependent Obsessive-compulsive Demographics Age 20e34 Age 35e50 Age 51e64 White African American Other non-White race Asian Hispanic Born outside the U.S. Less than high school High school Some college College Married or living as married Widowed/divorced/separated Never married Any children in the household Work full time Work part time Unemployed Not in labor force Household income < $20,000 Household income $20,000e59,999 Household income $60,000e$99,999 Household income $100,000 Risky behaviors and health status Tobacco use Illicit drug use Physical component score (PCS) Family background/childhood abuse Parent problem drinker, hospitalized for mental illness, or incarcerated Parent verbal, physical, or sexual abuse Father physically abuse mother Family received public assistance Live with at least 1 biological parent Lifetime mental health disorders Mood or anxiety disorder N

Women Any PD

No PD

Full sample

Any PD

No PD

75.54 439.90 4.88 20.76 4.51 16.55

75.69 613.26 7.38 24.41 6.32 25.39

75.49 382.90 4.05 19.55 3.91 13.61

67.07 214.31 1.55 7.20 1.11 6.58

69.48 266.39 2.27 9.16 1.64 10.79

66.39 199.35 1.34 6.64 0.95 5.37

24.95 4.10 3.20 4.74 6.57 6.16 2.07 8.34 2.13 0.34 8.40

100.00 16.43 12.83 18.98 26.32 24.71 8.28 33.41 8.55 1.36 33.69

0 0 0 0 0 0 0 0 0 0 0

22.24 5.64 3.43 4.04 2.35 7.12 2.03 5.43 3.14 0.60 8.79

100.00 25.36 15.41 18.17 10.57 32.02 9.12 24.40 14.10 2.70 39.53

0 0 0 0 0 0 0 0 0 0 0

31.82 38.43 29.75 69.02 10.53 2.13 4.71 13.62 15.51 12.24 26.26 32.09 29.41 66.14 11.04 22.81 36.79 76.09 7.98 3.98 11.94 12.43 40.83 27.15 19.59

37.60 37.13 25.27 67.73 12.09 3.58 3.12 13.49 11.20 14.64 27.55 35.30 22.50 58.44 14.44 27.11 34.82 69.44 9.82 6.23 14.52 17.41 44.02 24.10 14.47

29.90 38.86 31.23 69.45 10.01 1.65 5.24 13.66 16.94 11.44 25.84 31.02 31.70 68.70 9.91 21.38 37.45 78.31 7.37 3.24 11.08 10.78 39.77 28.16 21.29

31.09 38.57 30.34 67.97 12.88 2.31 4.53 12.30 14.11 10.54 25.45 35.18 28.83 64.76 17.09 18.15 43.19 52.88 17.78 3.78 25.56 16.86 43.25 24.05 15.83

36.96 38.07 24.97 63.90 17.41 3.23 3.52 11.95 11.35 12.96 25.47 39.72 21.86 55.76 21.95 22.30 44.03 50.91 16.20 5.24 27.65 22.71 45.17 20.83 11.30

29.40 38.72 31.88 69.13 11.59 2.05 4.82 12.40 14.90 9.85 25.44 33.88 30.83 67.34 15.70 16.96 42.95 53.45 18.23 3.36 24.96 15.19 42.71 24.97 17.13

25.89 12.56 52.2

34.04 22.53 50.8

23.19 9.24 52.7

21.64 7.87 51.6

31.63 14.91 49.5

18.78 5.85 52.2

25.50

36.99

21.69

29.62

41.83

26.13

41.48 8.79 13.77 97.91

58.57 14.08 20.88 97.38

35.80 7.03 11.41 98.08

40.09 11.08 15.58 97.74

60.43 19.36 23.01 96.66

34.27 8.72 13.46 98.05

33.93 11,497

63.9 2974

23.96 8523

51.04 15,199

82.02 3646

42.18 11,553

Notes: NESARC survey weights are applied. Percents are reported for binary variables and means are reported for continuous variables.

schizotypal, and obsessive-compulsive PDs are associated with a lower risk of alcohol use and misuse, while dependent PD is associated with an increased risk. The findings for women are presented in Table 3. Meeting the criteria for any PD is associated with a higher risk of alcohol abuse and/or dependence (0.9 percentage points or 13.6%; approaching statistical significance with p 0.10), but it is not significantly

associated with other measures of alcohol use or misuse. Similar to the findings for males, these aggregated PD regressions mask a substantial degree of heterogeneity across specific PD types. When separating any PD into the ten specific PDs recognized by the APA we find that antisocial, borderline, narcissistic, and dependent PDs are associated with an increased risk of alcohol use and misuse. Unlike men, histrionic PD (defined by gullibility and problems


291

Table 2 Estimated associations between personality disorders and past year alcohol use/misuse for men. 12þ drinks

Log (number of drinks)a

Weekly drinking to intoxication

Driving after 3þ drinks

Weekly drinking before 3 pm

Alcohol abuse/ dependence

Percent/mean Estimated object in each cell: Any PD

75.54% Beta (standard error)

439.90 Beta (standard error)





Percent/mean Estimated object in each cell: Paranoid Schizoid Schizotypal Antisocial Borderline Histrionic Narcissistic Avoidant Dependent Obsessivecompulsive







0.038 0.014 0.010 0.003 0.007 0.030 0.020 0.048 0.027 0.015

0.123 0.197 0.246* 0.176** 0.487*** 0.136 0.064 0.195 0.692* 0.024

0.003 0.000 0.002 0.023*** 0.034*** 0.000 0.002 0.024* 0.044 0.007

0.001 0.041 0.030 0.010 0.028 0.003 0.051*** 0.056 0.021 0.010

0.007 0.002 0.007 0.013* 0.021** 0.002 0.012* 0.023 0.036 0.002

0.005 0.015 0.028 0.014 0.073*** 0.001 0.047*** 0.028 0.004 0.022

N Estimator

0.005 (0.011)

(0.025) (0.025) (0.026) (0.020) (0.020) (0.036) (0.020) (0.035) (0.083) (0.018)

11,497 Probit

0.064 (0.050)

(0.127) (0.139) (0.143) (0.088) (0.103) (0.161) (0.088) (0.218) (0.375) (0.081)

7612 OLS

0.005 (0.005)

(0.011) (0.013) (0.010) (0.007) (0.009) (0.013) (0.008) (0.013) (0.035) (0.007)

11,486 Probit

0.021* (0.011)

(0.026) (0.030) (0.022) (0.018) (0.022) (0.031) (0.016) (0.039) (0.095) (0.018)

11,481 Probit

0.008 (0.006)

(0.011) (0.011) (0.010) (0.008) (0.009) (0.014) (0.007) (0.017) (0.038) (0.009)

11,483 Probit

0.047*** (0.010)

(0.019) (0.021) (0.019) (0.015) (0.017) (0.025) (0.014) (0.033) (0.078) (0.014)

11,497 Probit

Notes: All models apply NESARC survey weights and adjust for personal characteristics, health behaviors and outcomes, family background and childhood abuse, and lifetime mental health disorders. Each cell of the table contains an estimated beta and standard error. Average marginal effects are reported in probit models and beta coefficients are reported in OLS models. Standard errors are clustered around the primary sampling unit and reported in parentheses. ***; **; * ¼ statistically different from zero at the 1%; 5%; 10% level. a Conditional on consuming at least one drink in the past year.

delaying gratification) is a significant predictor of alcohol use/ misuse among women. In addition, the estimated relationships between dependent PD and alcohol misuse are statistically and quantitatively stronger for women compared to men. Paranoid, schizoid, schizotypal, and obsessive-compulsive PDs are negatively related to alcohol use and misuse among women. Our results suggest that PDs are stronger predictors of alcohol use/misuse among women than among men. This finding is not necessarily surprising as previous research indicates that PDs are

stronger predictors for behavioral health outcomes (e.g., body weight, risky sexual behaviors) among women than among men (Maclean et al., 2014, 2013). Future studies should rigorously assess potential mechanisms for these gender differences. 3.3. Unconditional quantile regressions We report UQR coefficients for the 5th through 95th quantiles in Figs. 1 (men) and 2 (women) for models that include an indicator

Table 3 Estimated associations between personality disorders and past year alcohol use/misuse for women. 12þ drinks













Percent/mean Estimated object in each cell: Paranoid Schizoid Schizotypal Antisocial Borderline Histrionic Narcissistic Avoidant Dependent Obsessivecompulsive







0.005 0.067** 0.059** 0.046 0.008 0.092*** 0.029 0.020 0.023 0.007

0.051 0.202 0.001 0.369*** 0.017 0.062 0.030 0.189 0.030 0.107

0.001 0.001 0.002 0.002 0.005 0.016** 0.002 0.004 0.024** 0.009**

0.023* 0.001 0.016 0.010 0.008 0.036** 0.023* 0.012 0.003 0.00

0.004 0.003 0.003 0.005 0.004 0.002 0.004 0.002 0.021*** 0.001

0.004 0.013 0.003 0.004 0.021** 0.048*** 0.000 0.002 0.012 0.010

N Estimator

0.001 (0.011)

15,199 Probit

(0.024) (0.027) (0.025) (0.031) (0.021) (0.034) (0.021) (0.028) (0.063) (0.016)

0.012 (0.053)

7063 OLS

(0.118) (0.151) (0.133) (0.122) (0.105) (0.125) (0.085) (0.165) (0.417) (0.078)

0.002 (0.003)

15,194 Probit

(0.006) (0.005) (0.005) (0.006) (0.005) (0.007) (0.005) (0.007) (0.010) (0.004)

0.002 (0.007)

15,195 Probit

(0.013) (0.014) (0.013) (0.014) (0.011) (0.016) (0.012) (0.014) (0.030) (0.009)

0.002 (0.003)

14,746 Probit

(0.005) (0.005) (0.005) (0.005) (0.004) (0.007) (0.004) (0.006) (0.007) (0.004)

0.009* (0.005)

(0.010) (0.012) (0.011) (0.012) (0.009) (0.013) (0.009) (0.013) (0.024) (0.008)

15,199 Probit


292


Fig. 1. Unconditional quantile regression results for the association between any personality disorder and conditional log(number of drinks) among men.

for any PD. In both figures the X-axis reports quantiles of the drinking distribution and the Y-axis reports the estimated beta at each quantile. For comparison, we also report the OLS estimates discussed earlier. The UQR results suggest that the relationships

between PDs and the number of drinks consumed in the past year vary across the drinking distribution for both men and women. Estimated associations are largest for men (women) who consume beyond the 85th (90th) quantile. Few practically or statistically

Fig. 2. Unconditional quantile regression results for the association between any personality disorder and conditional log(number of drinks) among women.


significant results emerge at lower quantiles of the distribution. Overall, this exercise suggests that the associations between PDs and alcohol misuse are concentrated among the heaviest drinkers. To further explore heterogeneity, we compare demographics of respondents who consume more drinks per month than the 95th quantile of the unconditional drinking distribution with those respondents who consume less (we include non-drinkers in the analysis and construct the 95th quantile separately for men and women). Results are reported in Appendix Table E. In unreported analyses available from the corresponding author, we conduct c2 (binary variables) or t- (continuous variables) tests to determine the statistical significance of the differences between respondents who consume more drinks per month than the 95th quantile of the unconditional drinking distribution with those respondents who consume less. In general, respondents who consume above the 95th quantile report higher levels of alcohol use and misuse; display higher prevalence rates of PDs; are more disadvantaged in terms of racial and ethnic minority status, education, marital status, employment, and income; have worse health behaviors and outcomes; and have more dysfunctional family backgrounds. Most of these differences are quantitatively large and statistically significant.

293

variables in alcohol use/misuse analyses is challenging (French and Popovici, 2011) and our research design requires that we obtain at least 10 such variables to instrument our 10 PDs. Because PDs emerge early in life from some combination of genetics and environment, finding 10 instruments that are both powerful and excludable from our estimating equationdEquation (2)dis a daunting task. Relying on weak and/or invalid IVs can lead to more severe bias than the bias in OLS estimates one is attempting to correct (Murray, 2006). For these reasons, we chose not to employ IV estimation and instead note this as a limitation of our study. Failure to adequately address omitted variables and reverse causality bias potentially inflates our estimates of the associations between PDs and alcohol use/misuse. Although we control for a large set of covariates in our regression models, some of these variables may themselves be influenced by PDs and, if so, our estimates may suffer from over-controlling bias (Angrist and Pischke, 2008). In unreported analyses available from the corresponding author, we re-estimate our models controlling only for a set of pre-determined variables (i.e., age, race/ ethnicity, birth outside the U.S., genetic pre-disposition to alcohol misuse, family background). Results are consistent in sign and significance with those from our core models, but, as expected, somewhat larger in magnitude.

3.4. Robustness checks 4. Discussion As noted earlier in the manuscript, Trull et al. (2010) develope an alternative, more restrictive, algorithm to classify PDs among NESARC respondents. As a first robustness check, we re-estimate our core models utilizing this algorithm. Selected results are reported in Appendix Tables F (men) and G (women). The prevalence of any PD using the Trull algorithm is 9% among men and 7% among women, which is considerably lower than the PD prevalence estimates using NESARC criteria (25% and 22%). Nevertheless, the estimated associations are highly consistent in sign and significance with the findings based on the more inclusive PD definitions. The DSM divides PDs into three clusters: A (paranoid, schizoid, schizotypal), B (antisocial, borderline, histrionic, narcissistic), and C (avoidant, dependent, obsessive-compulsive). We re-estimate our models utilizing these three groups instead of the single inclusive measure. Results are reported in Appendix Table H. The estimates here suggest that Cluster B disorders largely drive our main findings in that this cluster leads to higher estimated alcohol use and misuse effect sizes in all regressions. We next consider non-random attrition between Waves I and II of the NESARC. Specifically, in unreported analyses (results available on request from the corresponding author), we compare the Wave I PD status of Wave II attritors and completers based on the seven PDs that are measured at Wave I. In all cases, no statistically significant differences are present in Wave I PD diagnoses among Wave II attritors and completers. Although PDs manifest in childhood and adolescence, in some cases disorders may not fully develop until young adulthood. If a disorder emerges sufficiently late in young adulthood, reverse causality could be a concern for some younger members of our sample. To address this issue, we restrict the sample to ages 30e64 years and re-estimate our models (estimates available upon request from the corresponding author). Again, the results are consistent with the core findings. While this test cannot fully address potential bias from reverse causality, it is one approach to potentially reduce such bias given the psychiatric understanding of PD development. Ideally, we would like to utilize instrumental variables to address potential bias from both reverse causality and omitted variables. (A randomized control trial is arguably the gold standard in estimating causal effects, but is both unethical and infeasible in this setting.) However, locating powerful and valid instrumental

This study analyzes the relationships between PDs and several measures of alcohol use/misuse. We find strong evidence that PDs are associated with each measure of alcohol use/misuse, although substantial heterogeneity is present by gender and PD type, which is consistent with defining features of the particular PDs. In particular, the estimated associations between PDs and alcohol use/ misuse are generally stronger among women. Moreover, the magnitude of the estimated associations indicates that the findings are practically, as well as statistically, significant. Lastly, the UQR results reveal particularly large associations at the highest quantiles of the alcohol use distribution. Although we are not the first to study the relationships between PDs and alcohol use/misuse (Di Pierro et al., 2014; Hasin et al., 2007; Kerridge et al., 2014; Kienast et al., 2014; Stinson et al., 2008), we make several important contributions to the literature. First, we use insight gained from economic models of health production to motivate the importance of PDs for adult health outcomes (Heckman, 2007). Second, we carefully control for important background variables and comorbidities that may influence both the risk for PDs and alcohol use/misuse, and thus are able to minimize concerns regarding omitted variables. Third, we compare findings across a broad set of alcohol use and misuse measures to determine if PDs are associated with social drinking (which does not necessarily impose external costs on society) and alcohol misuse (which almost always generates external costs). Comparing estimates across alcohol use patterns can shed light on whether PDs are associated with the type of alcohol consumption that represents a public health concern. Fourth, we consider how the relationship between PDs and alcohol use varies across the drinking distribution using unconditional quantile regressions (Firpo et al., 2009). As noted by Manning et al. (1995), failure to consider such heterogeneity can miss important policy relevant information. Fifth, to address controversy within the psychiatric literature on how best to classify PD status in a survey setting, we consider multiple PD classification algorithms (Trull et al., 2010). Sixth, we estimate models separately by gender given stark differences between men and women in their risks for both alcohol use/misuse and PDs. The majority of existing research pools these two groups which, as we show, can muddle relationships between

294


PDs and alcohol use/misuse. Lastly, while the majority of previous studies consider only a subset of PDs, and often in convenience samples, we include all 10 PDs recognized by the APA via a large and nationally representative survey. The associations we estimate here have a causal interpretation if two critical assumptions are met: 1) no reverse causality is present from alcohol use/misuse to PDs and 2) no important omitted variables are significantly associated with both the diagnosis of PDs and alcohol use/misuse. Even if these conditions are not strictly met, we believe our unique data and rigorous research design are better able to address these two potential sources of bias than all previous studies. Specifically, the etiology of PDs (i.e., PDs manifest early in life and are persistent) minimizes reverse causality concerns. To this point, a sensitivity analysis shows that the results are robust to excluding those members of the sample whose PDs may not have fully developed at the time of the NESARC survey. Furthermore, because the NESARC is specifically designed to study alcohol use/misuse, we are able to include a rich set of important predictors in our regression models, and thus reduce the possibility of omitted variables bias. Despite these redeeming features, our study still has several limitations. First, because our alcohol use/misuse and PD measures are based on survey data, we cannot rule out the possibility that these variables are measured with error. Second, our PD prevalence rates fall in the upper range of prevalence rates derived from community-based samples of adults (Torgersen et al., 2001). Thus, we may be overstating PD prevalence. Third, we do not address non-random survey or item non-response in the NESARC. Individuals who are different in unobservable ways may decide not to respond to the NESARC survey at all or decline particular items in the survey. Lastly, even in a survey as detailed as the NESARC, we are unable to control for all important variables, so some residual omitted variable bias is possible. The estimated associations for PDs and alcohol abuse and/or dependence are particularly policy relevant because the latter conditions may lead to costly addiction treatment and health care utilization, as well as externalities at home and in the workplace. Unlike primary medical care, federal, state, and local governments cover the majority of substance abuse treatment services in the U.S., thus placing a direct financial burden on the American taxpayer. The most recent data suggest that the costs of such treatment will reach $239 billion by 2014 (Bouchery et al., 2011). Another area with possible costly externalities is traffic safety. Our findings for driving after drinking 3 or more drinks suggest that individuals with a PD may negatively impact those with whom they share roadways. These actions could lead to serious accidents and subsequent property damage, injuries, and fatalities. Alcohol misuse is also associated with lowered productivity in the workplace, which leads to costs for employers, coworkers, and consumers. Our measure of weekly drinking before 3 pm may, albeit imperfectly,

proxy for drinking while on the job for some individuals. If this is the case, our findings suggest that PDs, by increasing alcohol consumption during working hours, may lower worker productivity. Lastly, weekly drinking to intoxication is a risk factor for accidents, violence, crime, and excessive use of healthcare services (Centers for Disease Control and Prevention, 2013). In conclusion, this study provides new information on the associations between mental health problems, as measured by PDs, and alcohol use/misuse. We find evidence that personality disorders are strongly associated with an increased risk for alcohol use and misuse. Moreover, the measures of misuse examined here likely lead to costly externalities for society. Given the enduring nature of PDs, health professionals and policymakers should increase their awareness of these conditions and then try to minimize the associated externalities. Recent research indicates that health care professionals are not adequately prepared to address the needs of their patients who suffer from PDs (Kienast et al., 2014), suggesting that they may benefit from more information and better training on PD diagnosis and treatment options. For example, health professional organizations could offer continuing education classes which outline PD diagnoses, costs to the affected persons and those with whom s/he interacts, and methods to modify treatment to meet the needs of patients with PDs. Employers may wish to incorporate aspects of PDs into their Employee Assistance Programs (EAPs). Because persons suffering from PDs are likely to impose challenges for coworkers (Ettner et al., 2011), EAPs could offer guidance to employees without PDs who must collaborate with affected persons (e.g., conflict resolution skills tailored to persons affected by PDs). However, a key shortcoming of the current scientific knowledge base on PDs is the sparse amount of evidence on effective treatment options for most PDs (Duggan et al., 2007; Verheul and Herbirjnk, 2007). One notable exception is borderline PD (Lieb et al., 2010). While this study does not evaluate a particular intervention targeting PDs and cannot offer specific guidance on which interventions are most promising, it does provide additional motivation to invest more resources in developing and evaluating PD interventions (ideally via randomized control trials). Moreover, when it becomes available, this information should be disseminated widely and rapidly to mitigate the internal and external costs associated with PDs. Acknowledgments We thank Carmen Martinez for editorial assistance. All errors are our own. Appendix

Appendix Table A Brief descriptions of DSM personality disorders. Disorder Cluster A Paranoid

Schizoid Schizotypal

Individuals with this disorder:

Source

“Are highly suspicious of other people. As a result, people with this condition severely limit their social lives. They often feel that they are in danger, and look for evidence to support their suspicions. People with this disorder have trouble seeing that their distrustfulness is out of proportion to their environment.” “Are primarily characterized by a very limited range of emotion, both in expression of and experiencing. Persons with this disorder are indifferent to social relationships and display flattened affect.” “May be very disturbed. Their odd behavior may look like that of people with schizophrenia. For example, they may also have unusual preoccupations and fears, such as fears of being monitored by government agencies. More commonly, however, people with schizotypal personality disorder behave oddly and have unusual beliefs (such as aliens). They cling to these beliefs so strongly that it prevents them from having relationships. People with schizotypal personality disorder feel upset by their difficulty in forming and keeping close relationships.”

A.D.A.M. Medical Encyclopedia (2012) PsyWeb.com (2012) A.D.A.M. Medical Encyclopedia (2012)


295

Appendix Table A (continued ) Disorder Cluster B Antisocial

Narcissistic

Borderline

Histrionic

Cluster C Obsessivecompulsive Avoidant

Dependent

Individuals with this disorder:

Source

“Are characterized by a lack of regard for the moral or legal standards in the local culture. There is a marked inability to get along with others or abide by societal rules. Individuals with this disorder are sometimes called psychopaths or sociopaths.” “Are characterized by behavior or a fantasy of grandiosity, a lack of empathy and a need to be admired by others. Narcissistic personality has a pathological unrealistic or inflated sense of self-importance, has an inability to see the viewpoints of others, and is hypersensitive to the opinions of others.” “Are often uncertain about their identity. As a result, their interests and values may change rapidly. People with BPD also tend to see things in terms of extremes. Their views of other people may change quickly. These suddenly shifting feelings often lead to intense and unstable relationships. Other symptoms of BPD include: fear of being abandoned; feelings of emptiness and boredom; frequent displays of inappropriate anger; impulsivity with money, substance abuse, sexual relationships, binge eating, or shoplifting; intolerance of being alone; and repeated crises and acts of self-injury”. “Is primarily characterized by exaggerated displays of emotion in everyday behavior. Emotions are expressed with extreme and often inappropriate exaggeration. Persons with this disorder are prone to sudden and rapidly shifting emotional expression.”

PsyWeb.com (2012)

PsyWeb.com (2012)

A.D.A.M. Medical Encyclopedia (2012)

PsyWeb.com (2012)

“Is characterized by perfectionism and inflexibility. A person with obsessive-compulsive disorder becomes preoccupied with uncontrollable patterns of thought and action. Obsessive-compulsive symptoms may cause extreme distress and interfere with a person's occupational and social functioning.” “Can't stop thinking about their own shortcomings. They form relationships with other people only if they believe they will not be rejected. Loss and rejection are so painful that these people will choose to be lonely rather than risk trying to connect with others.” “Is primarily characterized by an extreme need for other people, to a point where the person is unable to make any decisions or take an independent stand on their own. There is a fear of separation, clinging, and submissive behavior. People with dependent personality disorder have a marked lack of decisiveness, self-confidence, and are selfdenigrating.”

PsyWeb.com (2012)

A.D.A.M. Medical Encyclopedia (2012) PsyWeb.com (2012)

Appendix Table B Estimated associations between personality disorders and past year alcohol use/misuse for pooled sample of males and females. 12þ drinks













Percent/mean Estimated object in each cell: Paranoid Schizoid Schizotypal Antisocial Borderline Histrionic Narcissistic Avoidant Dependent Obsessive-compulsive







0.022 0.042** 0.035* 0.002 0.014 0.059** 0.017 0.025 0.017 0.003

0.062 0.210** 0.132 0.237*** 0.249*** 0.018 0.040 0.193 0.333 0.061

0.002 0.001 0.000 0.014*** 0.019*** 0.012* 0.001 0.012 0.038** 0.008*

0.015 0.017 0.023* 0.006 0.008 0.028 0.036*** 0.028 0.009 0.008

0.006 0.001 0.005 0.008* 0.012*** 0.001 0.008* 0.010 0.033** 0.000

0.001 0.015 0.015 0.007 0.045*** 0.034** 0.024*** 0.012 0.008 0.005

N Estimator

26,696 Probit

0.005 (0.008)

(0.017) (0.019) (0.019) (0.016) (0.015) (0.026) (0.014) (0.021) (0.051) (0.012)

0.034 (0.035)

(0.086) (0.100) (0.104) (0.071) (0.078) (0.101) (0.067) (0.140) (0.304) (0.058)

14,675 OLS

0.004 (0.003)

(0.006) (0.007) (0.006) (0.004) (0.005) (0.007) (0.005) (0.008) (0.015) (0.004)

26,680 Probit

0.011* (0.007)

(0.014) (0.017) (0.013) (0.011) (0.012) (0.018) (0.010) (0.017) (0.044) (0.010)

26,676 Probit

0.005 (0.003)

0.027*** (0.005)

(0.006) (0.006) (0.005) (0.005) (0.004) (0.008) (0.004) (0.008) (0.015) (0.005)

26,676 Probit

(0.010) (0.012) (0.011) (0.009) (0.009) (0.014) (0.008) (0.017) (0.034) (0.008)

26,696 Probit

Notes: All models apply NESARC survey weights and adjust for male gender, personal characteristics, health behaviors and outcomes, family background and childhood abuse, and lifetime mental health disorders. Each cell of the table contains an estimated beta and standard error. Average marginal effects are reported in probit models and beta coefficients are reported in OLS models. Standard errors are clustered around the primary sampling unit and reported in parentheses. ***; **; * ¼ statistically different from zero at the 1%; 5%; 10% level. a Conditional on consuming at least one drink in the past year. Appendix Table C Estimated associations between personality disorders and past year alcohol use/misuse for men. 12þ drinks




Weekly drink before 3 pm









0.005 (0.011)

0.064 (0.050)

0.005 (0.005)

0.021* (0.011)

0.008 (0.006)

Percent/mean Estimated object in each cell: Age 35e49 Age 50e64






0.019*** (0.006) 0.034*** (0.008)

0.043*** (0.012) 0.064*** (0.014)

0.068*** (0.013) 0.086*** (0.016)

0.010 (0.062) 0.126* (0.068)

0.024*** (0.006) 0.027*** (0.007)

0.047*** (0.010) 16.55% Beta (standard error) 0.042*** (0.010) 0.081*** (0.013) (continued on next page)

296


Appendix Table C (continued )

African American Other non-White race Asian Hispanic Born outside the U.S. High school Some college College Widowed/divorced/ separated Never married Number of children in the household Work part time Unemployed Not in labor force Household income $20,000e$59,999 Household income $60,000e$99,999 Household income $100,000 Tobacco use Illicit drug use Physical component score (PCS) Parent problem drinker Parent hospitalized for mental illness Parent incarcerated Verbal abuse by parent Physical abuse by parent Sexual abuse by parent Father physically abused mother Family received public assistance Live with at least 1 biological parent Lifetime mood disorder Lifetime anxiety disorder N Estimator

12þ drinks






0.063*** 0.002 0.086*** 0.019 0.039** 0.024 0.061*** 0.118*** 0.037***

0.063 0.222 0.180 0.010 0.191** 0.150* 0.325*** 0.314*** 0.323***

0.003 0.009 0.004 0.008 0.021** 0.015** 0.017** 0.024*** 0.011*

0.067*** 0.020 0.065** 0.014 0.084*** 0.050*** 0.051*** 0.061*** 0.080***

0.027*** 0.002 0.019 0.015* 0.022*** 0.012 0.016** 0.029*** 0.004

0.037*** 0.010 0.052** 0.002 0.060*** 0.013 0.002 0.012 0.073***

(0.013) (0.039) (0.026) (0.017) (0.016) (0.017) (0.017) (0.019) (0.014)

0.004 (0.015) 0.006 (0.012) 0.014 0.029 0.058*** 0.022

(0.017) (0.025) (0.015) (0.016)

(0.069) (0.151) (0.128) (0.070) (0.076) (0.087) (0.090) (0.095) (0.066)

0.230*** (0.061) 0.111** (0.051) 0.066 0.195 0.001 0.067

(0.075) (0.124) (0.072) (0.079)

(0.006) (0.016) (0.013) (0.008) (0.009) (0.008) (0.007) (0.008) (0.006)

0.015** (0.006) 0.023*** (0.006) 0.004 0.008 0.011 0.005

(0.009) (0.009) (0.007) (0.007)

(0.014) (0.034) (0.028) (0.017) (0.016) (0.018) (0.016) (0.019) (0.013)

0.031** (0.013) 0.027** (0.012) 0.018 0.025 0.077*** 0.050***

(0.017) (0.024) (0.017) (0.016)

(0.005) (0.017) (0.017) (0.007) (0.007) (0.007) (0.007) (0.009) (0.006)

0.002 (0.006) 0.013** (0.006) 0.006 0.017** 0.005 0.012*

(0.008) (0.008) (0.008) (0.006)

(0.011) (0.029) (0.025) (0.016) (0.015) (0.015) (0.014) (0.015) (0.012)

0.042*** (0.012) 0.017* (0.010) 0.002 0.001 0.049*** 0.020

(0.015) (0.018) (0.014) (0.014)

0.070*** (0.017)

0.115 (0.085)

0.009 (0.008)

0.087*** (0.017)

0.013 (0.008)

0.031**(0.015)

0.139*** (0.019)

0.219** (0.095)

0.008 (0.009)

0.130*** (0.019)

0.024** (0.010)

0.050*** (0.018)

0.103*** (0.012) 0.169*** (0.018) 0.003*** (0.001)

0.418*** (0.053) 0.636*** (0.059) 0.001 (0.003)

0.032*** (0.005) 0.048*** (0.006) 0.001** (0.000)

0.055** (0.010) 0.137*** (0.012) 0.003*** (0.001)

0.032*** (0.005) 0.039*** (0.006) 0.000* (0.000)

0.063*** (0.009) 0.161*** (0.009) 0.001 (0.001)

0.039*** (0.012) 0.002 (0.019)

0.042 (0.054) 0.143 (0.104)

0.009 (0.006) 0.003 (0.009)

0.005 (0.011) 0.013 (0.018)

0.001 (0.005) 0.001 (0.009)

0.010 (0.010) 0.005 (0.017)

0.032 0.002 0.064*** 0.057* 0.017

(0.023) (0.012) (0.012) (0.032) (0.017)

0.147 0.032 0.041 0.182 0.063

(0.103) (0.056) (0.050) (0.158) (0.081)

0.008 0.001 0.007 0.012 0.007

(0.010) (0.006) (0.006) (0.015) (0.009)

0.014 0.002 0.045*** 0.088*** 0.009

(0.019) (0.012) (0.010) (0.033) (0.017)

0.004 0.002 0.001 0.034* 0.003

(0.011) (0.006) (0.007) (0.018) (0.008)

0.008 0.024** 0.025*** 0.028 0.006

(0.017) (0.011) (0.010) (0.029) (0.015)

0.020 (0.014)

0.075 (0.072)

0.005 (0.006)

0.004 (0.013)

0.011 (0.007)

0.011 (0.011)

0.014 (0.031)

0.114 (0.164)

0.027** (0.012)

0.019 (0.031)

0.026** (0.013)

0.025 (0.024)

0.045*** (0.012) 0.017 (0.012)

0.016 (0.055) 0.016 (0.051)

0.004 (0.005) 0.005 (0.006)

0.005 (0.012) 0.003 (0.011)

0.000 (0.005) 0.002 (0.006)

0.006 (0.010) 0.008 (0.010)

11,497 Probit

7612 OLS

11,486 Probit

11,481 Probit

11,483 Probit

11,497 Probit

Notes: All models apply NESARC survey weights. Each cell of the table contains an estimated beta and standard error. Average marginal effects are reported in probit models and beta coefficients are reported in OLS models. Standard errors are clustered around the primary sampling unit and reported in parentheses. ***; **; * ¼ statistically different from zero at the 1%; 5%; 10% level. a Conditional on consuming at least one drink in the past year.

Appendix Table D Estimated associations between personality disorders and past year alcohol use/misuse for women. 12þ drinks






Percent/mean 67.07% 214.31 1.55% 7.20% 1.11% 6.58% Estimated object in each cell Beta (standard error) Beta (standard error) Beta (standard error) Beta (standard error) Beta (standard error) Beta (standard error) Any PD 0.001 (0.011) 0.012 (0.053) 0.002 (0.003) 0.002 (0.007) 0.002 (0.003) 0.009* (0.005) Percent/mean 67.07% 214.31 1.55% Estimated object in each cell Beta (standard error) Beta (standard error) Beta (standard error) Age 35e49 0.051*** (0.011) 0.175*** (0.050) 0.001 (0.003) Age 50e64 0.096*** (0.015) 0.104 (0.063) 0.016*** (0.004) African American 0.148*** (0.012) 0.053 (0.064) 0.003 (0.003) Other non-White race 0.129*** (0.032) 0.124 (0.226) 0.005 (0.009) Asian 0.199*** (0.031) 0.255 (0.169) 0.009 (0.009) Hispanic 0.053*** (0.014) 0.192*** (0.068) 0.006 (0.004) Born outside the U.S. 0.103*** (0.015) 0.070 (0.086) 0.002 (0.005) High school 0.078*** (0.016) 0.072 (0.103) 0.002 (0.004) Some college 0.145*** (0.017) 0.005 (0.097) 0.002 (0.004) College 0.182*** (0.018) 0.036 (0.098) 0.001 (0.005) Widowed/divorced/ 0.050*** (0.012) 0.112* (0.058) 0.007** (0.003) separated

7.20% Beta (standard error) 0.016*** (0.006) 0.061*** (0.008) 0.033*** (0.007) 0.032 (0.021) 0.014 (0.018) 0.017** (0.008) 0.050*** (0.011) 0.030*** (0.011) 0.042*** (0.011) 0.040*** (0.011) 0.029*** (0.007)

1.11% Beta (standard error) 0.010*** (0.003) 0.005 (0.003) 0.005* (0.002) 0.008 (0.007) –b 0.014*** (0.004) 0.008** (0.004) 0.005 (0.003) 0.007* (0.004) 0.007* (0.004) 0.001 (0.003)

6.58% Beta (standard error) 0.010* (0.006) 0.044*** (0.007) 0.004 (0.006) 0.010 (0.022) 0.028 (0.023) 0.004 (0.008) 0.039*** (0.010) 0.019** (0.009) 0.033*** (0.009) 0.034*** (0.010) 0.019*** (0.006)


297

Appendix Table D (continued ) 12þ drinks Never married Number of children in the household Work part time Unemployed Not in labor force Household income $20,000e$59,999 Household income $60,000e$99,999 Household income $100,000 Tobacco use Illicit drug use Physical component score (PCS) Parent problem drinker Parent hospitalized for mental illness Parent incarcerated Verbal abuse by parent Physical abuse by parent Sexual abuse by parent Father physically abused mother Family received public assistance Live with at least 1 biological parent Lifetime mood disorder Lifetime anxiety disorder N Estimator

0.087*** (0.014) 0.005 (0.011) 0.016 0.020 0.093*** 0.053***






0.223*** (0.055) 0.316*** (0.045)

0.012*** (0.003) 0.009*** (0.003)

0.036*** (0.007) 0.021*** (0.005)

0.031*** (0.006) 0.020*** (0.005)

(0.007) (0.012) (0.007) (0.007)

0.000 (0.003) 0.007*** (0.003) 0.000 (0.003) 0.010** (0.005) 0.000 (0.003) 0.001 (0.003)

0.049 0.192 0.068 0.138**

(0.012) (0.022) (0.010) (0.014)

(0.053) (0.118) (0.056) (0.066)

0.001 0.001 0.000 0.006*

(0.003) (0.005) (0.003) (0.003)

0.015** 0.007 0.040*** 0.018***

0.004 0.006 0.017*** 0.006

(0.006) (0.012) (0.006) (0.007)

0.107*** (0.017)

0.041 (0.074)

0.002 (0.004)

0.034*** (0.008)

0.002 (0.004)

0.001 (0.008)

0.216*** (0.019)

0.193** (0.082)

0.008* (0.004)

0.041*** (0.011)

0.001 (0.004)

0.010 (0.010)

0.133*** (0.013) 0.155*** (0.020) 0.003*** (0.000)

0.402*** (0.049) 0.616*** (0.067) 0.009*** (0.003)

0.012*** (0.003) 0.017*** (0.003) 0.000 (0.000)

0.051*** (0.005) 0.075*** (0.007) 0.001*** (0.000)

0.009*** (0.002) 0.012*** (0.003) 0.000* (0.000)

0.046*** (0.006) 0.084*** (0.006) 0.001*** (0.000)

0.007 (0.011) 0.026 (0.019)

0.029 (0.050) 0.059 (0.086)

0.000 (0.003) 0.003 (0.004)

0.010* (0.006) 0.004 (0.010)

0.001 (0.003) 0.002 (0.004)

0.012** (0.006) 0.001 (0.008)

0.005 0.000 0.002 0.001 0.001

0.010 0.008 0.011* 0.003 0.014*

0.005 0.046*** 0.016 0.007 0.014

0.084 0.026 0.039 0.111 0.133*

(0.019) (0.013) (0.012) (0.020) (0.016)

(0.078) (0.054) (0.056) (0.101) (0.077)

0.005 0.001 0.005 0.001 0.012***

(0.005) (0.003) (0.003) (0.004) (0.003)

0.010 0.004 0.021*** 0.026*** 0.020**

(0.009) (0.006) (0.006) (0.010) (0.008)

(0.004) (0.002) (0.003) (0.003) (0.003)

(0.009) (0.006) (0.006) (0.009) (0.008)

0.010 (0.013)

0.008 (0.059)

0.005* (0.003)

0.011* (0.007)

0.002 (0.003)

0.001 (0.006)

0.027 (0.026)

0.126 (0.145)

0.003 (0.008)

0.035** (0.015)

0.009 (0.008)

0.009 (0.013)

0.018* (0.010) 0.001 (0.010)

0.053 (0.048) 0.004 (0.047)

0.001 (0.003) 0.001 (0.003)

0.006 (0.006) 0.007 (0.006)

0.004* (0.002) 0.001 (0.002)

0.015*** (0.005) 0.012** (0.005)

15,199 Probit

7063 OLS

15,194 Probit

15,195 Probit

14,746 Probit

15,199 Probit

Notes: All models apply NESARC survey weights. Standard errors are clustered around the primary sampling unit and reported in parentheses. Average marginal effects are reported in probit models and beta coefficients are reported in OLS models. ***; **; * ¼ statistically different from zero at the 1%; 5%; 10% level. a Conditional on consuming at least one drink in the past year. b Variable drops out of model due to small sample sizes.

Appendix Table E Summary statistics for respondents who consume greater than and less than the 95th quantile of the drinking distribution. Men

Past year alcohol use Drink 12þ drinks Number of drinks (conditional sample) Weekly drinking to intoxication Any driving after drinking 3þ drinks Weekly drinking before 3 pm Abuse and/or dependence Personality disorders Any PD Paranoid Schizoid Schizotypal Antisocial Borderline Histrionic Narcissistic Avoidant Dependent Obsessive-compulsive Demographics Age 20e34 Age 35e50 Age 51e64 White African American

Women

Number of drinks > 95th quantile

Number of drinks 95th quantile



100 2475.0 41.2 52.6 36.1 69.4

74.3 176.7 2.97 19.1 2.85 13.8

100 1072.3 16.8 36.6 12.2 42.5

65.5 57.1 0.83 5.81 0.58 4.87

43.1 9.06 5.31 8.84 15.7 18.2 5.11 15.5 3.45 1.43 11.8

24.0 3.84 3.09 4.52 6.08 5.53 1.91 7.96 2.06 0.28 8.23

30.0 8.05 3.82 7.16 5.87 12.4 3.36 7.63 3.64 0.98 9.45

21.9 5.53 3.41 3.89 2.18 6.87 1.97 5.32 3.11 0.58 8.76

38.2 36.8 25.0 67.7 13.5

31.5 38.5 30.0 69.1 10.4

30.3 39.7 30.0 74.8 11.6

31.1 38.5 30.4 67.6 12.9 (continued on next page)

298


Appendix Table E (continued ) Men

Women

Number of drinks > 95th quantile Other non-White race Asian Hispanic Born outside the U.S. Less than high school High school Some college College Married or living as married Widowed/divorced/separated Never married Any children in the household Work full time Work part time Unemployed Not in labor force Household income < $20,000 Household income $20,000e59,999 Household income $60,000e$99,999 Household income $100,000 Risky behaviors and health status Tobacco use Illicit drug use Physical component score (PCS) Family background/childhood abuse Parent problem drinker, hospitalized for mental illness, or incarcerated Parent verbal, physical, or sexual abuse Father physically abuse mother Family received public assistance Live with at least 1 biological parent Lifetime mental health disorders Mood or anxiety disorder N




3.07 2.01 13.7 9.31 23.9 32.2 30.8 13.0 48.3 16.3 35.4 27.5 66.3 9.29 10.2 14.3 18.5 50.4 20.8 10.4

2.08 4.85 13.6 15.8 11.6 26.0 32.2 30.3 67.1 10.8 22.1 37.3 76.6 7.91 3.66 11.8 12.1 40.3 27.5 20.1

4.70 2.44 6.52 7.31 7.41 26.0 39.8 26.9 50.2 23.1 26.7 27.6 54.9 15.5 5.36 24.3 21.3 38.2 21.2 19.3

2.20 4.63 12.6 14.4 10.7 25.4 35.0 28.9 65.5 16.8 17.7 43.9 52.8 17.9 3.70 25.6 16.7 43.5 24.2 15.7

57.2 37.1 50.8

24.2 11.3 52.3

45.5 25.4 52.3

20.5 7.04 51.6

31.8

25.2

36.8

29.3

46.4 10.4 20.8 96.4

41.2 8.70 13.4 98.0

49.0 16.8 20.1 97.5

39.7 10.8 15.4 97.8

44.3

33.4

58.7

50.7

576

10,921

632

14,567

Notes: NESARC survey weights are applied. The 95th quantile of the drinking distribution is estimated separately for men and women. Perecents are reported for binary variables and means are presented for continuous variables.

Appendix Table F Estimated associations between personality disorders and past year alcohol use/misuse for men using the Trull et al. (2010) algorithm to classify PDs. 12þ drinks






Percent/mean 75.54% 439.90 4.88% 20.76% 4.51% 16.55% Estimated object in each cell: Beta (standard error) Beta (standard error) Beta (standard error) Beta (standard error) Beta (standard error) Beta (standard error) Any PD 0.021 (0.016) 0.133* (0.080) 0.020*** (0.007) 0.004 (0.017) 0.010 (0.007) 0.024* (0.013) Percent/mean 75.54% Estimated object in each cell: Beta (standard error) Paranoid 0.014 (0.041) Schizoid 0.046 (0.059) Schizotypal 0.025 (0.054) Antisocial 0.006 (0.025) Borderline 0.042 (0.031) Histrionic 0.039 (0.097) Narcissistic 0.033 (0.041) Avoidant 0.061 (0.050) Dependent 0.016 (0.100) Obsessive-compulsive 0.047 (0.036)

439.90 4.88% Beta (standard error) Beta (standard error) 0.055 (0.171) 0.006 (0.013) 0.606* (0.335) 0.043 (0.033) 0.244 (0.368) 0.018 (0.021) 0.200* (0.110) 0.022** (0.009) 0.419*** (0.150) 0.031** (0.012) 0.064 (0.360) 0.037 (0.033) 0.021 (0.215) 0.001 (0.014) 0.380 (0.269) 0.008 (0.022) 0.764* (0.426) 0.024 (0.049) 0.012 (0.149) 0.001 (0.012)

20.76% Beta (standard error) 0.067* (0.037) 0.073 (0.056) 0.055 (0.057) 0.002 (0.022) 0.063** (0.032) 0.064 (0.073) 0.045 (0.036) 0.141** (0.061) 0.027 (0.099) 0.091** (0.036)

4.51% 16.55% Beta (standard error) Beta (standard error) 0.010 (0.012) 0.015 (0.025) 0.002 (0.022) 0.041 (0.044) 0.013 (0.019) 0.021 (0.040) 0.008 (0.010) 0.015 (0.019) 0.014 (0.014) 0.097*** (0.022) 0.013 (0.028) 0.006 (0.064) 0.013 (0.014) 0.028 (0.030) 0.002 (0.025) 0.047 (0.038) 0.028 (0.043) 0.069 (0.081) 0.003 (0.014) 0.036 (0.030)

N Estimator

7612 OLS

11,481 Probit

11,483 Probit

11,497 Probit

11,486 Probit

11,497 Probit



299

Appendix Table G Estimated associations between personality disorders and past year alcohol use/misuse for women using the Trull et al. (2010) algorithm to classify PDs. 12þ drinks













Percent/mean Estimated object in each cell: Paranoid Schizoid Schizotypal Antisocial Borderline Histrionic Narcissistic Avoidant Dependent Obsessive-compulsive







0.046 0.064 0.067 0.061 0.051** 0.034 0.016 0.088** 0.127 0.017

0.187 0.127 0.445 0.301 0.128 0.124 0.241 0.082 0.233 0.122

0.002 0.012 0.005 0.014** 0.009 0.033* 0.002 0.004 0.023* 0.006

0.011 0.022 0.009 0.005 0.012 0.004 0.002 0.011 0.029 0.003

0.002 0.024** 0.003 0.009 0.002 0.017 0.007 0.001 0.031*** 0.006

0.016 0.014 0.018 0.006 0.003 0.033 0.017 0.007 0.001 0.004

N Estimator

15,199 Probit

0.008 (0.017)

(0.034) (0.060) (0.063) (0.042) (0.026) (0.095) (0.050) (0.040) (0.082) (0.030)

0.117 (0.081)

(0.184) (0.338) (0.337) (0.183) (0.134) (0.411) (0.248) (0.198) (0.426) (0.161)

7063 OLS

0.006 (0.004)

(0.006) (0.012) (0.008) (0.007) (0.006) (0.018) (0.008) (0.007) (0.012) (0.006)

15,194 Probit

0.013 (0.009)

(0.016) (0.035) (0.039) (0.019) (0.015) (0.030) (0.024) (0.019) (0.039) (0.018)

15,195 Probit

0.000 (0.003)

(0.005) (0.012) (0.008) (0.006) (0.005) (0.015) (0.007) (0.007) (0.011) (0.005)

14,746 Probit

0.008 (0.007)

(0.014) (0.027) (0.029) (0.015) (0.013) (0.033) (0.018) (0.016) (0.029) (0.015)

15,199 Probit


Appendix Table H Estimated associations between personality disorders and past year alcohol use/misuse for men and women using PD clusters to classify PDs 12þ drinks












0.026 (0.018) 0.013 (0.014) 0.000 (0.016)

0.131 (0.090) 0.147** (0.062) 0.007 (0.082)

0.004 (0.008) 0.021*** (0.006) 0.009 (0.007)

0.028* (0.017) 0.043*** (0.013) 0.020 (0.018)

0.012 (0.008) 0.019*** (0.005) 0.001 (0.008)

0.017 (0.015) 0.059*** (0.011) 0.015 (0.014)

N Estimator Women Percent/mean Estimated object in each cell: Cluster A Cluster B Cluster C

11,497 Probit

7612 OLS

11,486 Probit

11,481 Probit

11,483 Probit

11,497 Probit







0.063*** (0.017) 0.042*** (0.014) 0.006 (0.015)

0.022 (0.088) 0.056 (0.066) 0.127* (0.073)

0.002 (0.004) 0.007* (0.004) 0.003 (0.003)

0.019* (0.011) 0.016* (0.008) 0.008 (0.009)

0.004 (0.004) 0.006** (0.003) 0.003 (0.003)

0.003 (0.009) 0.023*** (0.007) 0.003 (0.007)

N Estimator

15,199 Probit

7063 OLS

15,194 Probit

15,195 Probit

14,746 Probit

15,199 Probit

Men Percent/mean Estimated object in each cell: Cluster A Cluster B Cluster C

Notes: All models apply NESARC survey weights and adjust for personal characteristics, health behaviors and outcomes, family background and childhood abuse, and lifetime mental health disorders. Each cell of the table contains an estimated beta and standard error. Average marginal effects are reported in probit models and beta coefficients are reported in OLS models. Standard errors are clustered around the primary sampling unit and reported in parentheses. Cluster A includes paranoid, schizoid, and schizotypal PDs. Cluster B includes antisocial, borderline, histrionic, and narcissistic PDs. Cluster C includes avoidant, dependent, and obsessive-compulsive PDs. ***; **; * ¼ statistically different from zero at the 1%; 5%; 10% level. a Conditional on consuming at least one drink in the past year.

References A.D.A.M. Medical Encyclopedia, 2012. Personality disorders. Access date: September 17, 2014. http://www.ncbi.nlm.nih.gov/pubmedhealth/PMH0001935/. Almond, D., Currie, J., 2011. Human capital development before age five. In: Ashenfelter, O., Card, D. (Eds.), Handbook of Labor Economics. Elsevier, pp. 1315e1486. American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental Health Disorders, Text Revision (DSM-IV-TR). American Psychiatric Association, Washington, DC.

Angrist, J.D., Pischke, J.S., 2008. Mostly Harmless Econometrics: an Empiricalist's Companion. Princeton University Press, Princeton, NJ. Baldwin, M.L., Marcus, S.C., 2014. The impact of mental health and substance-use disorders on employment transitions. Health Econ. 23, 332e344. Balsa, A.I., 2008. Parental problem-drinking and adult children's labor market outcomes. J. Hum. Resour. 43, 454e486. Balsa, A.I., French, M.T., 2010. Alcohol use and the labor market in Uruguay. Health Econ. 19, 833e854. Balsa, A.I., French, M.T., Maclean, J.C., Norton, E.C., 2009. From pubs to scrubs: alcohol misuse and health care use. Health Serv. Res. 44, 1480e1503.

300


Blanco, C., Krueger, R.F., Hasin, D.S., Li, S., Wang, S., Kerridge, B.T., et al., 2013. Mapping common psychiatric disorders: structure and predictive validity in the national epidemiologic survey on alcohol and related conditions. JAMA Psychiatry 70, 199e207. Bouchery, E.E., Harwood, H.J., Sacks, J.J., Simon, C.J., Brewer, R.D., 2011. Economic costs of excessive alcohol consumption in the U.S., 2006. Am. J. Prev. Med. 41, 516e524. Carpenter, C., 2007. Heavy alcohol use and crime: evidence from underage drunkdriving laws. J. Law Econ. 50, 539e557. Centers for Disease Control and Prevention, 2013. Alcohol and Public Health: Frequently Asked Questions. Centers for Disease Control and Prevention, Atlanta, GA. Chatterji, P., Dave, D., Kaestner, R., Markowitz, S., 2004. Alcohol abuse and suicide attempts among youth. Econ. Hum. Biol. 2, 159e180. Davalos, M.E., Fang, H., French, M.T., 2012. Easing the pain of an economic downturn: macroeconomic conditions and excessive alcohol consumption. Health Econ. 21, 1318e1335. Di Pierro, R., Preti, E., Vurro, N., Madeddu, F. Dimensions of personality structure among patients with substance use disorders and co-occurring personality disorders: a comparison with psychiatric outpatients and healthy controls. Compr. Psychiatry 2014. Duggan, C., Huband, N., Smailagic, N., Ferriter, M., Adams, C., 2007. The use of psychological treatments for people with personality disorder: a systematic review of randomized controlled trials. Personal. Ment. Health 1, 95e125. Ettner, S.L., Maclean, J.C., French, M.T., 2011. Does having a dysfunctional personality hurt your career? Axis II personality disorders and labor market outcomes. Ind. Relat. A J. Econ. Soc. 50, 149e173. Firpo, S., Fortin, N.M., Lemieux, T., 2009. Unconditional quantile regressions. Econometrica 77, 953e973. French, M.T., Maclean, J.C., Sindelar, J.L., Fang, H., 2010. The morning after: alcohol misuse and employment problems. Appl. Econ. 43, 2705e2720. French, M.T., Popovici, I., 2011. That instrument is lousy! In search of agreement when using instrumental variables estimation in substance use research. Health Econ. 20, 127e146. Golomb, M., Fava, M., Abraham, M., Rosenbaum, J.F., 1995. Gender differences in personality disorders. Am. J. Psychiatry 152, 579e582. Grant, B., Hasin, D.S., Chou, S., Stinson, F.S., Dawson, D.A., 2004. Nicotine dependence and psychiatric disorders in the United States: results from the National Epidemiologic Survey on alcohol and related conditions. Arch. Gen. Psychiatry 61, 1107e1115. Grant, B., Kaplan, K., Shephard, J., T., M., 2003. Source and Accuracy Statement for Wave 1 of the 2001e2002 National Epidemiologic Survey on Alcohol and Related Conditions. National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD. Hasin, D., Fenton, M.C., Skodol, A., Krueger, R.F., Keyes, K., Geier, T., et al., 2011. Personality disorders and the 3-year course of alcohol, drug, and nicotine use disorders. Arch. Gen. Psychiatry 68, 1158e1167. Hasin, D., Stinson, F., Ogburn, E., Grant, B.F., 2007. Prevalence, correlates, disability, and comorbidity of DSM-IV alcohol abuse and dependence in the United States: results from the National Epidemiologic Survey on alcohol and related conditions. Arch. Gen. Psychiatry 64, 830e842. Heckman, J.J., 2007. The economics, technology, and neuroscience of human capability formation. Proc. Natl. Acad. Sci. U. S. A. 104, 13250e13255. Kerridge, B.T., Saha, T.D., Hasin, D.S., 2014. DSM-IV antisocial personality disorder and conduct disorder: evidence for taxonic structures among individuals with and without substance use disorders in the general population. J. Stud. Alcohol Drugs 75, 496. Keyes, K.M., Barnes, D.M., Bates, L.M., 2011. Stress, coping, and depression: testing a new hypothesis in a prospectively studied general population sample of U.S.born Whites and Blacks. Soc. Sci. Med. 72, 650e659.

Kienast, T., Stoffers, J., Bermpohl, F., Lieb, K., 2014. Borderline personality disorder € and comorbid addiction: epidemiology and treatment. Dtsch. Arztebl. Int. 111, 280. Levit, K.R., Kassed, C.A., Coffey, R.M., Mark, T.L., Stranges, E.M., Buck, J.A., et al., 2008. Future funding for mental health and substance abuse: increasing burdens for the public sector. Health Aff. Millwood 27, w513ew522. €llm, B., Rücker, G., Timmer, A., Stoffers, J.M., 2010. Pharmacotherapy for Lieb, K., Vo borderline personality disorder: Cochrane systematic review of randomised trials. Br. J. Psychiatry 196, 4e12. Lovenheim, M., Steefel, D., 2011. Do blue laws save lives? The effect of sunday alcohol sales bans on fatal vehicle accidents. J. Policy Anal. Manag. 30, 798e820. Maclean, J.C., 2014. The lasting effects of leaving school in an economic downturn on alcohol use. Ind. Labor Relat. Rev. (Forthcoming). Maclean, J.C., Xu, H., French, M.T., Ettner, S.L., 2014. Personality disorders and body weight. Econ. Hum. Biol. 12, 153e171. Maclean, J.C., Xu, H.Y., French, M.T., Ettner, S.L., 2013. Mental health and risky sexual behaviors: evidence from DSM-IV axis II disorders. J. Ment. Health Policy Econ. 16, 187e208. Manning, W.G., Blumberg, L., Moulton, L.H., 1995. The demand for alcohol e the differential response to price. J. Health Econ. 14, 123e148. Markowitz, S., Grossman, M., 2000. The effects of beer taxes on physical child abuse. J. Health Econ. 19, 271e282. Moos, R.H., Brennan, P.L., Schutte, K.K., Moos, B.S., 2005. Older adults' health and changes in late-life drinking patterns. Aging Ment. Health 9, 49e59. Mullahy, J., Sindelar, J.L., 1996. Employment, unemployment, and problem drinking. J. Health Econ. 15, 409e434. Murray, M.P., 2006. Avoiding invalid instruments and coping with weak instruments. J. Econ. Perspect. 20, 111e132. Naimi, T.S., Brewer, R.D., Mokdad, A., Denny, C., Serdula, M.K., Marks, J.S., 2003. Binge drinking among us adults. JAMA 289, 70e75. Paris, J., 2004. Gender differences in personality traits and disorders. Curr. Psychiatry Rep. 6, 71e74. Popovici, I., French, M.T., 2013. Does unemployment lead to greater alcohol consumption? Ind. Relat. A J. Econ. Soc. 52, 444e466. PsyWeb.com, 2012. DSM Axis II e Personality Disorders and Intellectual disabilities. Access date: November 4, 2012. http://www.psyweb.com/DSM_IV/jsp/Axis_II.jsp. Ruan, W.J., Goldstein, R.B., Chou, S.P., Smith, S.M., Saha, T.D., Pickering, R.P., et al., 2008. The alcohol use disorder and associated disabilities interview schedule-IV (AUDADIS-IV): reliability of new psychiatric diagnostic modules and risk factors in a general population sample. Drug Alcohol Depend. 92, 27e36. Slopen, N., Williams, D.R., Fitzmaurice, G.M., Gilman, S.E., 2011. Sex, stressful life events, and adult onset depression and alcohol dependence: are men and women equally vulnerable? Soc. Sci. Med. 73, 615e622. Stinson, F.S., Dawson, D.A., Goldstein, R.B., Chou, S.P., Huang, B., Smith, S.M., et al., 2008. Prevalence, correlates, disability, and comorbidity of DSM-IV narcissistic personality disorder: results from the wave 2 National Epidemiologic Survey on alcohol and related conditions. J. Clin. Psychiatry 69, 1033. Terza, J.V., 2002. Alcohol abuse and employment: a second look. J. Appl. Econ. 17, 393e404. Torgersen, S., Kringlen, E., Cramer, V., 2001. The prevalence of personality disorders in a community sample. Arch. Gen. Psychiatry 58, 590e596. Trull, T.J., Jahng, S., Tomko, R.L., Wood, P.K., Sher, K.J., 2010. Revised NESARC personality disorder diagnoses: gender, prevalence, and comorbidity with substance dependence disorders. J. Pers. Disord. 24, 412e426. Verheul, R., Herbirjnk, M., 2007. The efficacy of various modalities of psychotherapy for personality disorders: a systematic review of the evidence and clinical recommendations. Int. Rev. Psychiatry 19, 25e38. Yudofsky, S.C., 2005. Fatal Flaws: Navigating Destructive Relationships with People with Disorders of Personality and Character. American Psychiatric Publishing, Inc, Washington, DC.

Electronic Interventions for Alcohol Misuse and Alcohol Use Disorders: A Systematic Review.

Alcohol misuse disorders and multiple sclerosis risk.

Alcohol use disorders.

Eating disorders and alcohol misuse: features of an addiction spectrum.

Medications for alcohol use disorders.

Medications for alcohol use disorders.

Cochlear implantation and alcohol misuse.

Multiple sclerosis and alcohol misuse.

Screening for alcohol use disorders in HIV patients using the modified Alcohol Use Disorders Identification Test.

Alcohol use disorders: tracts, twins, and trajectories.

[Diagnosis and management of alcohol-use disorders].

Brief interventions for alcohol misuse.

Alcohol use and personality relationships in U.S. and Polish adolescents.

Alcohol misuse: assessment, treatment and aftercare.

Cigarette Smoking and Risk of Alcohol Use Relapse Among Adults in Recovery from Alcohol Use Disorders.

Medications for alcohol use disorders--reply.

Diagnosis of alcohol use disorders in schizophrenia.

Impulsive Personality and Alcohol Use: Bidirectional Relations Over One Year.

Baclofen-induced edema in alcohol use disorders.

Multiple sclerosis and alcohol misuse--reply.

Breastfeeding and maternal alcohol use: Prevalence and effects on child outcomes and fetal alcohol spectrum disorders.

Readability of self-report alcohol misuse measures.

The developmental psychopathology of alcohol use and alcohol disorders: research achievements and future directions.

Neuropeptide S receptor gene variant and environment: contribution to alcohol use disorders and alcohol consumption.