Exploring the Association between Lifetime Prevalence of Mental Illness and Transition from Substance Use to Substance Use Disorders: Results from the National Epidemiologic Survey of Alcohol and Related Conditions (NESARC)
Address correspondence to Dr. Lev-Ran, Centre for Addiction and Mental Health, 33 Russell St., Room 2035, Toronto, ON, M5S2S1 Canada. E-mail: email@example.com.
Background and Objectives
The association between substance use disorders (SUDs) and mental illness (MI) has been well established. Previous studies reporting this association in various clinical populations have not taken into account former substance use. This may be important as increased prevalence of substance use among individuals with MI may partially explain the strong association between SUDs and MI.
In this study we included only individuals with previous substance use and explored the association between lifetime diagnosis of MI and transition from substance use to SUDs. Analyses were conducted across six different categories of substances (alcohol, nicotine, cannabis, cocaine, hallucinogens, inhalants) based on a large representative US sample, the National Epidemiologic Survey of Alcohol and Related Conditions (NESARC, n = 43,093).
Lifetime diagnoses of any MI, and particularly personality disorders and psychotic disorders, were found to be associated with higher prevalence of transition from substance use to SUDs across most categories of substances. This association was particularly strong for nicotine (adjusted OR = 2.95 (2.72–3.20)).
Conclusions and Scientific Significance
This cross-sectional study expands on previous research by highlighting the association between lifetime diagnosis of any MI and increased rates of transition from substance use to SUDs across a range of substances. Longitudinal studies exploring temporal effects of this association are further needed. (Am J Addict 2013;22:93-98)
The high prevalence of co-morbid substance use disorders (SUDs) and mental illness (MI) have been well established in both clinical and population-based studies.[1-4] The Epidemiological Catchment Area (ECA) study reported that among individuals with MI, 29% had a co-morbid SUD (compared with 13% of individuals without a MI). Findings from the US National Comorbidity Survey (NCS) report a significantly higher prevalence of SUDs among individuals with MI compared to the general population; among individuals with any MI, 51% were reported as having a co-morbid SUD, and the odds of having a SUD were more than twice as high among individuals with a MI compared to those without any MI.
Previous studies on the prevalence and co-morbidity of SUDs and MI in various clinical populations have not taken into account prevalence of former substance use. This may be important as the increased prevalence of substance use among individuals with MI may partially explain the increased rates of SUDs among individuals with MI. This is in addition to various neurobiological and environmental factors formerly proposed. In addition, epidemiological studies which have explored transition from substance use to SUDs have focused largely on the transition from use to dependence[6-8] without regarding the harmful potential of substance abuse. This is despite reports showing that in fact, some abuse criteria (eg, role interference and legal trouble) may be more predictive than some dependence criteria (eg, consuming in larger amounts and for longer than intended) when assessing patterns and outcomes of substance use.[9, 10] These criticisms have led to the idea that the DSM-V should merge the two disorders (ie, dependence, abuse) into one broad category,[11, 12] reflecting repeated use, loss of control over use of a substance and dysfunction. A single SUD criteria set is currently under consideration for the fifth edition of the DSM, and classifications of alcohol use disorders and substance use disorders are commonly used in epidemiological studies addressing measures relating to both abuse and dependence.[14, 15]
In order to address these methodological considerations, in this study we included only individuals with lifetime prevalence of substance use, and explored the association of lifetime diagnosis of MI and transition from substance use to SUDs using a single diagnostic category. This was done across different categories of substances based on a large representative US sample.
We analyzed cross-sectional data from a population-based national representative sample, the National Epidemiological Survey of Alcohol and Related Conditions (NESARC) study (Wave 1, 2001–2002) conducted by the National Institute on Alcohol Abuse and Alcoholism (NIAAA). The interview was developed to advance measurement of substance use and additional psychiatric disorders in large-scale surveys. Face-to-face interviews were conducted with 43,093 adults (response rate, 81%), aged 18 years and older from the civilian non-institutionalized population residing in the United States, including the District of Columbia, Hawaii, and Alaska. The NESARC sample was weighted to adjust for probabilities of selection of a sample housing unit or housing unit equivalent, the non-response at the household and person levels, the selection of one person per household, and the oversampling of African Americans, Hispanics, and young adults (ages 18–24). The weighted data were post-stratified and adjusted to match the target population based on the 2000 decennial census in terms of region, age, sex, race, and ethnicity. Details regarding sampling, purpose and weighting have been previously published. Characteristics of interviewers, training, and field quality control has been described elsewhere.
Transition from Substance Use to SUD
The Alcohol Use Disorder and Associated Disabilities Interview Schedule—DSM-IV Version (AUDADIS-IV) was used to assess lifetime SUDs. The AUDADIS-IV has been reported to have excellent reliability and validity in the United States[18, 20-22] and internationally.[23, 24] It includes an extensive list of symptom questions that separately operationalizes DSM-IV criteria for abuse and dependence on different categories of substances. The substances included in this study were alcohol, nicotine, cannabis, cocaine (including crack cocaine), hallucinogens, and inhalants. Lifetime exposure to alcohol was defined as at least 1 drink of any kind of alcohol (not counting small tastes or sips). Lifetime exposure to nicotine was defined as smoking at least 100 cigarettes. Lifetime exposure to the remaining substances was defined as any previous exposure to the specific substance.
Lifetime prevalence of a SUD was defined as lifetime substance abuse or dependence. Rates of transition from substance use to a SUD were calculated as the prevalence of lifetime SUD among individuals with lifetime exposure to a substance.
The psychiatric disorders included in this study were mood, anxiety, psychotic, and personality disorders. Wave 1 of the NESARC assessed for mood, anxiety, and personality disorders based on the DSM-IV criteria for these disorders, and assessed for psychotic disorders based on self-report. Mood disorders included major depressive disorder (MDD), disthymia, bipolar I, and bipolar II disorders. Anxiety disorders included panic disorder with or without agoraphobia, generalized anxiety disorder (GAD), social anxiety disorder, and specific phobia. Personality disorders included in the NESARC were schizoid, paranoid, histrionic, anti-social, avoidant, dependent, and obsessive-compulsive personality disorders. All DSM-IV diagnosis included in this study were lifetime diagnosis.
Lifetime diagnosis of a psychotic disorder was based on the question “Did a doctor or other health professional ever diagnose you with schizophrenia or psychotic illness or episode?” This method has been used in previous studies and results echo accepted prevalence rates of schizophrenia in the North American population.[25, 26]
Cross-tabulations and chi-square analyses were used to derive and compare prevalence estimates of specific lifetime SUDs among lifetime users of each of the following categories of substances: alcohol, nicotine, cannabis, cocaine, hallucinogens, and inhalants. As direct comparisons were not possible across categories, rates of transition from substance use to SUD for different substances were compared based on non-overlapping 95% confidence intervals.
Unadjusted and adjusted odds ratios (ORs) were derived from bivariate and multivariate logistic regression analyses to examine the associations between transition from substance use to SUD of a specific category of substances and any lifetime diagnosis of MI. All adjusted ORs were controlled for sex, age, household income, education, region, urbanicity, and race.
All results were based on weighted data. To accurately estimate variances taking the NESARC sample design components into account, all analyses were conducted with Software for Survey Data Analysis (SUDAAN) Version 10, a software program that uses Taylor series linearization to make adjustments for the NESARC's sample design characteristics. Odds ratios and 95% confidence intervals (CI) are reported.
The lifetime prevalence of any MI (mood, anxiety, psychotic, or personality disorder) in the NESARC sample was 33.7% (n = 14,472). Among respondents in the NESARC, 66.2% (n = 28,621) had no lifetime MI. The lifetime prevalence of any mood disorder in the NESARC sample was 19.5% and of any anxiety disorder was 17.2%. Among respondents in the NESARC, the lifetime prevalence of at least one personality disorder was 14.8% and lifetime prevalence of a psychotic disorder was .78%.
The rate of transition from substance use to SUD was higher for individuals with a lifetime diagnosis of MI than for individuals without any MI. This was true for cocaine (OR = 1.83), cannabis (OR = 1.91), nicotine (OR = 3.24), alcohol (OR = 2.2), and hallucinogens (OR = 2.14) compared to controls. These differences retained significance after adjusting for sociodemographic factors (Table 1).
Table 1. Comparison and association of rates of transition from substance use to substance use disorder (SUD) in individuals with and without a lifetime diagnosis of any mental illnessa in the NESARC 2001–2002
|Cocaine||37.08 (383)||1.86||51.93 (776)||1.66||b||1.83 (1.51–2.23)||b||1.71 (1.40–2.08)||b|
|Cannabis||33.08 (1,272)||.99||48.60 (2,025)||1.05||b||1.91 (1.70–2.15)||b||2.05 (1.82–2.31)||b|
|Nicotine||28.01 (2,812)||.66||55.73 (3,918)||.77||b||3.24 (3.00–3.49)||b||2.95 (2.72–3.20)||b|
|Alcohol||29.90 (6,077)||.77||48.45 (5,766)||.77||b||2.2 (2.08–2.34)||b||2.53 (2.37–2.70)||b|
|Hallucinogens||20.14 (163)||1.52||35.02 (460)||1.69||b||2.14 (1.69–2.71)||b||2.14 (1.67–2.73)||b|
|Inhalants||14.21 (29)||2.82||21.81 (109)||2.41||.04||1.68 (1.00–2.84)||.05||1.61 (.92–2.81)||.10|
The relative ranking across substances of rates of transition from substance use to SUD was not different among the psychiatric disorders or between the psychiatric disorders and controls for all substances except nicotine. The relative ranking of rates of transition from nicotine use to nicotine use disorder was higher among respondents with a lifetime diagnosis of any MI compared to respondents with no lifetime diagnosis of MI. Whereas among individuals without any lifetime diagnosis of MI, the relative ranking of nicotine (based on rates of transition from use to SUD) was lower than that of cocaine and cannabis and in the range of alcohol, among individuals with a lifetime diagnosis of MI, the relative ranking of nicotine was highest among the substances studied. This was true for individuals with mood (60.4%, SE = .94), anxiety (58.5%, SE = 1.02), personality (62.8%, SE = 1.12), and psychotic (65.1%, SE = 3.90) disorders.
Among the various MI included in this study, lifetime diagnosis of psychotic disorders and personality disorders were particularly associated with high rates of transition from substance use to SUD, and particularly for alcohol and cannabis. Rates of transition from cannabis use to cannabis use disorder were 54.4% (SE = 1.39) for individuals with a lifetime diagnosis of personality disorders and 63.34% (SE = 5.4) among individuals with a lifetime diagnosis of psychotic disorders (compared to 49.7% (SE = 1.23) and 47.5% (SE = 1.43) among individuals with a lifetime diagnosis of mood and anxiety disorders, respectively). Rates of transition from alcohol use to alcohol use disorder were 56.4% (SE = 1.07) for individuals with a lifetime diagnosis of personality disorders and 58.8% (SE = 3.45) for individuals with a lifetime diagnosis of psychotic disorders, compared to 49.7% (SE = .89) among individuals with a lifetime diagnosis of mood disorders and 48.7% (SE = .91) among individuals with a lifetime diagnosis of an anxiety disorders. For five out of the six categories of substances (cocaine, cannabis, nicotine, alcohol, and hallucinogens), an association was found between lifetime diagnosis of any mood, anxiety, psychotic, and personality disorders and rates of transition from substance use to SUD. These associations retained significance after adjusting for sociodemographic factors (Table 2).
Table 2. Association between rates of transition from substance use to substance use disorder (SUD) and lifetime diagnosis of mood, anxiety, personality, and psychotic disorders in the NESARC 2001–2002
|Cocaine||1.96 (1.57–2.45)||c||1.86 (1.48–2.34)||c||1.81 (1.44–2.28)||c||1.72 (1.35–2.19)||c|
|Cannabis||2.00 (1.76–2.28)||c||2.20 (1.92–2.52)||c||1.83 (1.58–2.12)||c||2.02 (1.74–2.34)||c|
|Nicotine||3.93 (3.59–4.30)||c||3.45 (3.13–3.80)||c||3.63 (3.30–3.99)||c||3.31 (3.00–3.65)||c|
|Alcohol||2.32 (2.15–2.50)||c||2.76 (2.54–2.99)||c||2.23 (2.07–2.40)||c||2.74 (2.52–2.98)||c|
|Hallucinogens||2.46 (1.89–3.21)||c||2.42 (1.85–3.17)||c||2.09 (1.59–2.75)||c||2.19 (1.64–2.93)||c|
|Inhalants||1.69 (.98–2.90)||.06||1.69 (.91–3.16)||.10||1.54 (.84–2.82)||.16||1.39 (.67–2.89)||.37|
|Cocaine||2.16 (1.73–2.70)||c||1.92 (1.52–2.42)||c||3.45 (1.82–6.52)||b||2.69 (1.43–5.04)||.003|
|Cannabis||2.41 (2.10–2.78)||c||2.36 (2.05–2.71)||c||3.49 (2.19–5.57)||c||3.46 (2.28–5.24)||c|
|Nicotine||4.35 (3.91–4.85)||c||3.81 (3.39–4.29)||c||4.80 (3.39–6.78)||c||4.38 (3.11–6.17)||c|
|Alcohol||3.04 (2.80–3.29)||c||3.26 (2.99–3.55)||c||3.35 (2.53–4.43)||c||3.97 (2.95–5.32)||c|
|Hallucinogens||2.47 (1.91–3.21)||c||2.53 (1.90–3.36)||c||5.95 (2.99–11.83)||c||5.66 (2.86–11.19)||c|
|Inhalants||1.95 (1.16–3.30)||.01||1.79 (.98–3.27)||.06||2.67 (1.08–6.58)||.03||1.53 (.59–3.97)||.38|
This study found that a lifetime diagnosis of any MI was associated with higher rates of transition from substance use to SUD compared to individuals without any lifetime diagnosis of MI. This expands on previous findings as this study focuses only on individuals with lifetime exposure to substances and allows to explore the relative rates of transition from substance use to SUD (among categories of substances) based on a large representative sample.
The relative ranking of transition from substance use to SUD among clusters of substances was similar for individuals with and without a lifetime diagnosis of any MI and across the various psychiatric disorders for all substances except nicotine. Among individuals without any lifetime diagnosis of MI, the rate of transition from use to SUD was lower for nicotine than the rate of transition found for cocaine and cannabis and in the same range as the rate of transition for alcohol. Among individuals with a lifetime diagnosis of any and all MI studied, the rates of transition from use to SUD for nicotine was highest among the substances studied. Several explanations have been formerly proposed for the strong association of nicotine smoking and nicotine dependence and various MI. First, these include intrinsic factors (eg, shared genes) that predispose individuals with psychiatric disorders to maintaining smoking behaviors after initial exposure. Second, it has been proposed that nicotine specifically modulates several neurotransmitter systems thought to be involved in the pathogenesis of various psychiatric disorders. Third, individuals with MI may be particularly inclined to maintaining nicotine intake in order to reduce psychotropic drug side-effects and self-medicate psychiatric symptoms.[29-32] Fourth, various social and environmental factors may contribute to maintaining smoking behavior (ie, stressful environments) particularly among individuals with MI after initial exposure. Our results highlight that individuals with a lifetime diagnosis of MI have higher rates of maintaining smoking after initially being exposed to nicotine. It should be noted that rates of transition from nicotine use to nicotine use disorder based on the NESARC may be overestimated in our study, as this calculation is made based on exposure criteria of 100 or more cigarettes to define exposure. Though the 100+ criteria for cigarette smoking is widely used in epidemiological studies of smoking, this underestimates lifetime exposure in adult populations, thus affecting our calculations of rates of transition from nicotine use to nicotine use disorders. Nevertheless, as we compared rates of transition from substance use to SUD among individuals with and without a lifetime diagnosis of MI, this affects calculations across all different groups.
Among individuals with a lifetime diagnosis of MI, strong associations with rates of transition from substance use to SUD were found particularly for individuals with a lifetime diagnosis of psychotic disorders and personality disorders. This is in line with previous cross-sectional data from the ECA study reporting a significantly higher prevalence of SUDs among individuals with schizophrenia (47%) and antisocial personality disorder (84%) compared to individuals with mood (32%) and anxiety (24%) disorders.
The etiologic mechanisms underlying the association of MI and high rates of transition from substance use and SUD, and particularly the strong association of psychotic disorders, are not yet clear and various factors seem to play a significant role in this association. These include genetic, neurobiological, and environmental factors which may contribute to maintaining substance use after initial exposure and increasing the risk for developing SUDs. For example, among individuals with schizophrenia, it has been proposed that better premorbid social activity is associated with increased rates of transition from substance use to SUD. In addition, it has been proposed that individuals with schizophrenia initiate substance use at an earlier age, increasing the risk of developing a future SUD. These factors and others may contribute to the increased rate of transition from substance use to SUD among individuals with psychotic disorders.
Various factors may contribute to the particularly strong association between lifetime diagnosis of personality disorders and increased rates of transition from substance use to SUD. Though the NESARC sample included personality disorders from different diagnostic clusters, by definition these are all pervasive disorders which include enduring patterns of behavior that affect individuals' functioning. This pervasive nature of personality disorders has been associated with considerable emotional disability and psychosocial difficulties, which may affect initiation of substance use, patterns of substance use, and consequently development of a SUD. Though there is a strong association between personality disorders and mood and anxiety disorders, there is also a high degree of diagnostic overlap between individuals with personality disorders which has long been recognized and recent reports on the NESARC sample show a high degree of comorbidity of personality disorders within clusters and between clusters. This may contribute to the strong association observed with rates of transition from use to SUDs; for instance, individuals with concurrent cluster B personality disorders (ie, antisocial personality disorder and histrionic personality disorder) share a tendency towards impulsive behavior which may affect both initiation of substance use and patterns of substance use contributing to the development of SUDs. It should be noted, however, that several researchers have argued against the categorical models of personality disorders currently used in the DSM-IV[42-44] and proposed that dimensional models of personality are more directly related to general personality functioning. This may clearly affect our understanding of the association between personality disorders and transition from substance use to SUD.
To summarize, we emphasize the following major results:
- There is a strong association between lifetime diagnosis of any MI and rates of transition from substance use to SUD. Though this was true for all MI, it was particularly true for personality disorders and psychotic disorders.
- Among individuals with a lifetime diagnosis of any MI, the relative rate of transition from use to SUD for nicotine (compared to other substances) is significantly higher compared to individuals without any lifetime diagnosis of MI.
As is common in most large-scale epidemiological surveys, this study has several limitations. First, information was based on self-report, allowing for recall bias. Second, as the data was cross-sectional, it is not possible to determine the temporal relationship between exposure to the substance and onset of the MI. Third, because the NESARC sample included only civilian households and quarters populations, information on individuals in prisons is missing. Since the prevalence of SUDs has been reported to be high in this population, this may affect calculations. Fourth, data does not include adolescents, a particularly vulnerable population for developing SUDs after exposure to substances.
Despite these limitations, the NESARC constitutes the largest survey to date to include information on exposure to different substances and SUDs, as well as the diagnoses of major MI. This allows for exploring a range of substances and the relative ranking of rates of transition from substance use to SUDs among individuals with a lifetime diagnosis of various MI. This can help guide clinicians in identifying patients at higher risk of developing a SUD after initial exposure to a substance and assist in the organization of prevention services. Longitudinal studies exploring temporal effects of the association between SUDs and MI are further needed.
Declaration of Interest
The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this paper.