Towards DSM-V: Exploring Diagnostic Thresholds for Alcohol Dependence and Abuse
Towards DSM-V: Exploring Diagnostic Thresholds for Alcohol Dependence and Abuse
Aims: The expected release of the DSM-V in 2012 has renewed the longstanding debate around whether alcohol use disorders are best conceptualized as dimensional or categorical constructs. The current study aimed to validate the current diagnostic thresholds for alcohol dependence and abuse using epidemiological indicators including mental health, disability, psychological distress, functional impairment, service use, suicidality and early age of drinking onset.
Methods: Dichotomous variables were created to allocate a representative sample of Australian adult drinkers (n = 4920) above and below each possible threshold for both disorders. Regression analyses were conducted to assess group differences at each threshold for each epidemiological indicator.
Results: There was some albeit limited support for the current diagnostic threshold of three criteria for alcohol dependence and one criterion for abuse. A number of other cut-offs also showed consistent variation for both disorders.
Conclusions: It is essential to define diagnostic thresholds in a systematic way. The current diagnostic thresholds for alcohol dependence and abuse are adequate but require further validation using a variety of methods and external indicators. Combining these disorders in some way may also prove useful as well as including other potential diagnostic criteria in future research.
There is a longstanding debate concerning whether psychiatric disorders, including alcohol use disorders, should be conceptualized as categorical or dimensional constructs. This debate has been given renewed relevance by the expected release of DSM-V and ICD-11 in 2012 and 2015, respectively. The initial introduction of categorical conceptualizations of mental disorders to DSM-III and ICD-10 has meant better diagnostic agreement, ease of communication among clinicians and researchers, norms for research and teaching of an international system (Kendell and Jablensky, 2003). Public access to these diagnostic systems has also increased understanding and communication for non-clinicians including government agencies, healthcare workers and clients themselves (Kendell and Jablensky, 2003). Although these benefits must be recognized, it is important not to discount the problems and inadequacies of a binary system.
Ongoing oppositions to categorical representations of psychiatric disorders are motivated by the absence of natural boundaries within and between disorders, the fluid nature of disorders over time as well as similar treatments being effective for theoretically distinct entities (Kendell, 1975; Tyrer, 1985; Widiger and Clark, 2000; Melzer et al., 2002). There is growing evidence in support of incorporating both categorical and dimensional approaches to alcohol use disorders in DSM-V and ICD-11 (Hasin et al., 2006b; Helzer et al., 2006, 2007; Muthén, 2006). Dimensions and categories are not mutually exclusive. Muthén and Asparouhov (2006) argue that hybrid models that allow for dimensional and categorical representations of psychiatric disorders simultaneously are essential to understand the inherent nature of such constructs. Similarly, Hasin and Beseler (2009) note that a disorder may potentially take a categorical form under a threshold and be continuous above it or vice versa. Natural boundaries may exist for a disorder that is continuous in nature.
There is already a push towards conceptualizing alcohol use problems along a continuum (Kahler et al., 2004; Krueger et al., 2004; Langenbucher et al., 2004; Kahler and Strong, 2006; Martin et al., 2006; Proudfoot et al., 2006; Saha et al., 2006). However, although a continuum approach is appropriate in a research context and for understanding the phenotypic characteristics of a disorder, in other circumstances, particularly for diagnostic decisions, it may not be as useful. Fundamental to these diagnostic decisions is the way in which diagnostic thresholds are defined.
To date, diagnostic thresholds in psychiatry have been defined relatively arbitrarily based on clinician opinion (Angst and Merikangas, 2001; Vanyukov et al., 2003). Traditionally psychiatric diagnoses have also been defined within clinical samples. However, there is substantial evidence to suggest that examining the utility of diagnostic criteria in general population samples is needed in order to develop rigorous classification systems (Cottler and Grant, 2006; Schuckit and Saunders, 2006; Ruscio, in press; Saunders and Cottler, 2007). Distinctive markers that make it easy to determine the presence of a particular syndrome are lacking in the case of most psychiatric disorders (Kendell, 1989). This has led to the use of epidemiological indicators such as family history, patterns of comorbidity, subjective distress, and occupational and social impairment (Angst and Merikangas, 2001). Consistent differences across such indicators suggest the existence of discrete groups in classification.
There have been a number of attempts to identify discontinuities in the symptomatology of various psychiatric disorders in order to establish valid diagnostic thresholds. An early attempt at such work was carried out by Kendell and Brockington (1980) who proposed that if a genuine natural boundary exists between disorders, then it can be detected by demonstrating a non-linear relationship between the symptoms of that disorder and a second independent variable such as outcome. More recent studies have used this technique to search for discontinuities within specific disorders including depression (Kendler and Gardner, 1998; Angst and Merikangas, 2001; Sakashita et al., 2007). Similarly, Melzer and others (2002) used best-fitting curve analyses to identify whether symptom counts for common mental disorders provide distinct case groups above epidemiological cut-offs. Others have also attempted to search for a boundary between schizoaffective and major mood disorders (Peralta and Cuesta, 2008). There have also been attempts to search for discontinuities in the distribution of alcohol dependence (Hasin et al., 2006b; Hasin and Beseler, 2009) and abuse (Morey et al., 1984; Schuckit et al., 2005) using external correlates such as early age of drinking onset, functional impairment, family history and treatment.
The majority of the work outlined above concludes that there is no more than a linear relationship between the constructs examined. This is highly problematic as it does not provide any clear indication as to where a diagnostic threshold might easily be placed. Whilst others have looked at statistical models with linear terms and discontinuities (Kendler and Gardner, 1998; Sakashita et al., 2007; Hasin and Beseler, 2009), few studies have assessed the relative performance of each diagnostic threshold for a specific disorder.
Specifically, this type of research explores all potential diagnostic thresholds for a given disorder and attempts to locate defensible cut-offs, whilst controlling for the linear association that is so prolific in these types of analyses. The current study addresses this dearth in the literature by exploring each potential diagnostic threshold for DSM-IV alcohol dependence and abuse to assess whether there are differential relationships between various thresholds and external validators of diagnosis. Group differences above and below each potential cut-off on indicators including disability, functional impairment, psychological distress, mental health, service use, suicidality and early age of drinking onset were assessed. Significant group differences at different cut-offs for alcohol dependence and abuse provide the potential for locating defensible thresholds along these continua.
Abstract and Introduction
Abstract
Aims: The expected release of the DSM-V in 2012 has renewed the longstanding debate around whether alcohol use disorders are best conceptualized as dimensional or categorical constructs. The current study aimed to validate the current diagnostic thresholds for alcohol dependence and abuse using epidemiological indicators including mental health, disability, psychological distress, functional impairment, service use, suicidality and early age of drinking onset.
Methods: Dichotomous variables were created to allocate a representative sample of Australian adult drinkers (n = 4920) above and below each possible threshold for both disorders. Regression analyses were conducted to assess group differences at each threshold for each epidemiological indicator.
Results: There was some albeit limited support for the current diagnostic threshold of three criteria for alcohol dependence and one criterion for abuse. A number of other cut-offs also showed consistent variation for both disorders.
Conclusions: It is essential to define diagnostic thresholds in a systematic way. The current diagnostic thresholds for alcohol dependence and abuse are adequate but require further validation using a variety of methods and external indicators. Combining these disorders in some way may also prove useful as well as including other potential diagnostic criteria in future research.
Introduction
There is a longstanding debate concerning whether psychiatric disorders, including alcohol use disorders, should be conceptualized as categorical or dimensional constructs. This debate has been given renewed relevance by the expected release of DSM-V and ICD-11 in 2012 and 2015, respectively. The initial introduction of categorical conceptualizations of mental disorders to DSM-III and ICD-10 has meant better diagnostic agreement, ease of communication among clinicians and researchers, norms for research and teaching of an international system (Kendell and Jablensky, 2003). Public access to these diagnostic systems has also increased understanding and communication for non-clinicians including government agencies, healthcare workers and clients themselves (Kendell and Jablensky, 2003). Although these benefits must be recognized, it is important not to discount the problems and inadequacies of a binary system.
Ongoing oppositions to categorical representations of psychiatric disorders are motivated by the absence of natural boundaries within and between disorders, the fluid nature of disorders over time as well as similar treatments being effective for theoretically distinct entities (Kendell, 1975; Tyrer, 1985; Widiger and Clark, 2000; Melzer et al., 2002). There is growing evidence in support of incorporating both categorical and dimensional approaches to alcohol use disorders in DSM-V and ICD-11 (Hasin et al., 2006b; Helzer et al., 2006, 2007; Muthén, 2006). Dimensions and categories are not mutually exclusive. Muthén and Asparouhov (2006) argue that hybrid models that allow for dimensional and categorical representations of psychiatric disorders simultaneously are essential to understand the inherent nature of such constructs. Similarly, Hasin and Beseler (2009) note that a disorder may potentially take a categorical form under a threshold and be continuous above it or vice versa. Natural boundaries may exist for a disorder that is continuous in nature.
There is already a push towards conceptualizing alcohol use problems along a continuum (Kahler et al., 2004; Krueger et al., 2004; Langenbucher et al., 2004; Kahler and Strong, 2006; Martin et al., 2006; Proudfoot et al., 2006; Saha et al., 2006). However, although a continuum approach is appropriate in a research context and for understanding the phenotypic characteristics of a disorder, in other circumstances, particularly for diagnostic decisions, it may not be as useful. Fundamental to these diagnostic decisions is the way in which diagnostic thresholds are defined.
To date, diagnostic thresholds in psychiatry have been defined relatively arbitrarily based on clinician opinion (Angst and Merikangas, 2001; Vanyukov et al., 2003). Traditionally psychiatric diagnoses have also been defined within clinical samples. However, there is substantial evidence to suggest that examining the utility of diagnostic criteria in general population samples is needed in order to develop rigorous classification systems (Cottler and Grant, 2006; Schuckit and Saunders, 2006; Ruscio, in press; Saunders and Cottler, 2007). Distinctive markers that make it easy to determine the presence of a particular syndrome are lacking in the case of most psychiatric disorders (Kendell, 1989). This has led to the use of epidemiological indicators such as family history, patterns of comorbidity, subjective distress, and occupational and social impairment (Angst and Merikangas, 2001). Consistent differences across such indicators suggest the existence of discrete groups in classification.
There have been a number of attempts to identify discontinuities in the symptomatology of various psychiatric disorders in order to establish valid diagnostic thresholds. An early attempt at such work was carried out by Kendell and Brockington (1980) who proposed that if a genuine natural boundary exists between disorders, then it can be detected by demonstrating a non-linear relationship between the symptoms of that disorder and a second independent variable such as outcome. More recent studies have used this technique to search for discontinuities within specific disorders including depression (Kendler and Gardner, 1998; Angst and Merikangas, 2001; Sakashita et al., 2007). Similarly, Melzer and others (2002) used best-fitting curve analyses to identify whether symptom counts for common mental disorders provide distinct case groups above epidemiological cut-offs. Others have also attempted to search for a boundary between schizoaffective and major mood disorders (Peralta and Cuesta, 2008). There have also been attempts to search for discontinuities in the distribution of alcohol dependence (Hasin et al., 2006b; Hasin and Beseler, 2009) and abuse (Morey et al., 1984; Schuckit et al., 2005) using external correlates such as early age of drinking onset, functional impairment, family history and treatment.
The majority of the work outlined above concludes that there is no more than a linear relationship between the constructs examined. This is highly problematic as it does not provide any clear indication as to where a diagnostic threshold might easily be placed. Whilst others have looked at statistical models with linear terms and discontinuities (Kendler and Gardner, 1998; Sakashita et al., 2007; Hasin and Beseler, 2009), few studies have assessed the relative performance of each diagnostic threshold for a specific disorder.
Specifically, this type of research explores all potential diagnostic thresholds for a given disorder and attempts to locate defensible cut-offs, whilst controlling for the linear association that is so prolific in these types of analyses. The current study addresses this dearth in the literature by exploring each potential diagnostic threshold for DSM-IV alcohol dependence and abuse to assess whether there are differential relationships between various thresholds and external validators of diagnosis. Group differences above and below each potential cut-off on indicators including disability, functional impairment, psychological distress, mental health, service use, suicidality and early age of drinking onset were assessed. Significant group differences at different cut-offs for alcohol dependence and abuse provide the potential for locating defensible thresholds along these continua.
Source...