Several interrelated diagnostic issues are associated with use of the ICD-10 and DSM- IV as diagnostic standards for cannabis dependence and abuse/harmful use for the development of a screen for cannabis use problems. These include the validity of the cannabis dependence-abuse distinction, its applicability to adolescent cannabis use disorders, and appropriate disposition of a sub-diagnostic group for whom the term “diagnostic orphans” (Hasin & Paykin, 1998) was coined.
Validity of the bi-axial concept: Categorical and dimensional approaches
Statistical and conceptual independence of cannabis dependence from its consequences (abuse) is equivocal. While some researchers report statistical independence (a two- factor solution), others report one latent dimension on which the two constructs were highly correlated (see Swift, 1999). Feingold and Rounsaville (1995a, 1995b) interpreted their results as supporting a quantitative model in which abuse is a mild pro- dromal form of dependence, rather than a qualitatively distinct, but related, entity. Thus, the qualitative (categorical, two constructs) distinction would be more efficiently replaced by a quantitative (one construct, single underlying syndrome) model, in which a pooled set of criteria are used to determine gradations (none, low, mild, moderate, severe) along the problem severity continuum. In this dimensional view, sub-clinical or sub-threshold ratings may indicate an earlier stage in progression from consumption to drug-related problems, while lower dependence severity scores suggest an ‘abuse’ or ‘mild dependence’ diagnosis (Feingold & Rounsaville, 1995a, 1995b). Swift (1999) found evidence that when cannabis dependence symptoms form unidimensional scales, they evince a continuum of severity.
The essence of the dimensional approach is the flexibility of cut-points to suit different populations and purposes (Widiger & Trull, 1991). Distinguishing dependent from non- dependent individuals (i.e., determining caseness) is a matter of degree, with no single arbitrary cut-point universally suitable (Edwards et al., 1981; Streiner & Norman, 1995). Kendell (1975) notes that both the dimensional and categorical models are necessary and complementary. With their predetermined cut-points, categorical
diagnostic techniques are needed to classify individuals for treatment planning and provision. Difficulties arise, however, when continuous dimensions (e.g., psychiatric symptoms) are denoted either ‘present/absent’. Screening and treatment based on fitting “round” dimensions into “square” categories could lead to inappropriate diagnoses and treatment decisions (Goldberg, 2000). An axiom of statistical and psychometric theory is that forcing information based on an underlying dimensional trait into a categorical format sacrifices measurement precision (Jensen, 1995). Although a DSM-IV (and ICD-10) dependence diagnosis overrides an abuse diagnosis (APA, 1994), many drug users qualify for both diagnoses. Hence, recording the number of drug-related symptoms or problems as a way of screening for drug misuse or disorder, measuring severity of dependence or identifying case-ness in drug research “will continue to be a legitimate preference of many in the field” (Sellman, 1994, p. 210; and see Costello, 1992).
Adolescents and DSM-IV/ICD-10 diagnosis
Being neither age nor developmentally specific, DSM-IV and ICD-10 diagnostic frameworks can be challenged with respect to adolescents (Clark, 2004; Deas, Roberts & Grindlinger, 2005). Adolescents generally have shorter drug use histories and more pathological symptom profiles than adults (Bukstein & Kaminer, 1994). Dependence symptoms and medical problems, which may take years to develop, present differently in adolescents (American Academy of Child and Adolescent Psychiatry, 1997; Bailey, Martin, Lynch, & Pollock, 2000; Crowley, MacDonald, Whitmore, & Mikulich, 1998; Harrison, Fulkerson, & Beebe, 1998; Martin, Kaczynski, Maisto, Bukstein, & Moss, 1995; Winters, Latimer, & Stinchfield, 1999). Since many adolescents intentionally initiate out-of-control drug use “to get high or smashed” (Harrison et al., 1998, p. 487), the construct ‘impaired control’ is also problematic (Clark, 2004; Harrison et al., 1998). Nevertheless, teenagers can and do meet formal diagnostic criteria for substance dependence within a year of initial use (Deas et al., 2005; Martin et al., 1995). Moreover, tolerance (Deas et al., 2005) and withdrawal (Crowley et al., 1998; Vandrey, Budney, Kamon, & Stanger, 2005; Wiesbeck et al., 1996) were prevalent among adolescents (12-17 years) who were either cannabis dependent or non-treatment frequent users.
A major limitation is the categorical versus the dimensional nature of DSM-IV and ICD-10 formulations (Deas et al., 2005; Harrison et al., 1998). Given their generally more pathological profile at treatment admission, diagnostic thresholds may be inadequate for adolescents, especially for dependence (Clark, 2004; Dennis et al., 2002b; Mikulich, Hall, Whitmore, & Crowley, 2001; Harrison et al., 1998; Pollock & Martin, 1999; Winters et al., 1999). Similarly, behaviours that elicit legal and social consequences (apropos of an abuse diagnosis) for adolescents as minors may not create problems for adults (Bailey et al., 2000; Hays & Ellickson, 1996). Thus, diagnostic thresholds for both dependence and abuse should be lower for adolescents (Chung et al., 2000; Clark, 2004). However, whether the distinction between abuse and dependence among adolescents is diagnostically meaningful is questionable. The heterogeneity of symptoms endorsed by adolescents in recent studies suggests classificatory criteria may not adequately differentiate abuse from dependence (see Deas et al., 2005; Harrison et al., 1998). Empirical support for the dimensional approach to diagnostic classification based on a problem severity continuum (discussed above) as the most parsimonious model for adolescent drug use disorders is accumulating (Deas et al., 2005; Fulkerson, Harrison, & Beebe, 1999; Harrison et al., 1998; Pollock & Martin, 1999).
In short, “the tip of the iceberg has barely been touched in reaching consensus on the most appropriate criteria for adolescents” (Deas et al., 2005, p. 20). Since using adult measures can present psychometric problems, it cannot be assumed that adult models are directly transferable to adolescents (Chung et al., 2000; Clark, 2004; Leccese & Waldron, 1994; Winters, 2003). Given adolescents’ different developmental stages, patterns of use, family and peer issues, problem recognition and level of self-insight, some consider use of DSM-IV and ICD-10 criteria for adolescent drug disorders dubious (e.g., AACAP, 1997; Dennis et al., 2002b; Fulkerson et al., 1999; Harrison et al., 1998; Weinberg et al., 1998). Albeit, while perhaps not optimal, research generally supports the validity/utility of DSM-IV criteria for adolescent cannabis (and alcohol) use disorders (Bailey et al., 2000; Clark, 2004; Martin et al., 1995; Mikulich et al., 2001; Winters et al., 1999). While much evidence indicates the relevant constructs can be measured reliably and validly (Leccese & Waldron, 1994), cautious use of diagnostic
criteria among adolescents until validated, developmentally-appropriate, adolescent- specific diagnostic criteria are established is recommended (Kaminer, 1994). Meanwhile, the frequent differences between the most common manifestations of drug use, problems, abuse and dependence in adolescents versus adults should be kept in mind (AACAP, 1997; Clark, 2004; Spooner et al., 1996). Data in this research afforded an examination of this important aspect incorporated as an additional empirical question.
“Diagnostic orphans”
A third diagnostic dilemma flows directly from those just discussed. Researchers and clinicians have long recognized that drug use and disorders exist within continua among the population, and that use patterns below the diagnostic threshold may still be associated with substantial morbidity (Drummond, 1992; Edwards, 1992; IOM, 1990a, 1990b). Alcohol research has identified a sub-diagnostic group called “diagnostic orphans” (Hasin & Paykin, 1998, 1999; Pollock & Martin, 1999), defined as users who report one or two dependence symptoms, and therefore do not meet full criteria for dependence, while reporting no abuse symptoms. Compared to other adolescent and adult diagnostic groups, diagnostic orphans were more similar to ‘abuse’ than ‘dependence’ diagnostic groups, or those reporting no problems (Bailey et al., 2000; Hasin & Paykin, 1999; Sarr, Bucholz, & Phelps, 2000). This suggests diagnostic orphans and abuse groups have similar risks of substance-related problems. Diagnostic orphans and those assigned sub-clinical ratings on diagnostic criteria often characterize individuals in earlier, prodromal stages of problem development, providing valuable information beyond traditional categorical ‘present/absent’ symptom classification. Sub-threshold symptoms suggest a trajectory of escalating drug use and related problems.
These concepts extend to cannabis-using adults and adolescents (Deas et al., 2005; Degenhardt, Lynskey, Coffey, & Patton, 2002; Dennis et al., 2002b; Tims et al., 2002; Winters et al., 1999). As in alcohol samples, cannabis diagnostic orphans formed a separate group from dependent groups, evincing similar use patterns and problem profiles, including other illicit drug use, regular tobacco and alcohol use, mental health
problems, to the ‘abuse’ groups. The associated problems and symptom severity reported by diagnostic orphans in adolescent samples clearly indicated need for treatment (Deas et al., 2005; Dennis et al., 2002b; Dennis & McGeary, 1999; Tims et al., 2002). Diagnostic criteria appear inadequate to capture all those having significant problems with their cannabis use (Degenhardt et al., 2002; Tims et al., 2002; Winters et al., 1999). Hence, with incubating problems likely to fall through the ‘diagnostic crack’, cannabis diagnostic orphans and those reporting sub-threshold problem levels are “an important area for assessment in early case identification efforts” that “should alert practitioners to potentially serious problems” (Bailey et al., 2000, p.1800-1801). This at-risk group is an appropriate target for screening/early intervention to attempt to arrest progression of problems to a more advanced stage.