Recently, the Substance Use Disorder Work group of the DSM-5 announced the inclusion of “craving” in the diagnostic criteria for all substance use disorders despite its lack of empirical support from the very analyses conducted by that Workgroup. In addition, no detailed literature review supports the decision to make “craving” a core symptom of Substance Use Disorder syndromes. Addiction psychiatrists should understand the nature of “craving” as it is addressed in the scientific literature, and the evidence of its importance to the field.
The research literature suggests that “craving” per se is a problematic psychological construct for two very fundamental reasons. First, “craving” is not a well-defined construct.1 Some researchers distinguish between “craving“ as anticipation of pleasure from drug use2 versus the “urge” or “desire” to intentionally use a drug3 or use the term interchangeably. At its core, the definition is a subjective state of wanting or desire to use.4 Some have linked craving to abstinence or withdrawal phenomena. Others specify a type of “craving” associated with the desire to use alcohol(Drug information on alcohol), nicotine(Drug information on nicotine), or another drug in response to acute presentation of stimuli associated with that substance’s use. Much of this latter work has its origins in classical conditioning.5 This latter type of cue-induced craving is also referred to as “cue reactivity” in the literature, and is typically induced in a laboratory setting and responses are measured with either self-report or physiological methodology such as heart rate, galvanic skin responses, eye tracking, EEG, ERP, MRI, and PET.
However, across the literature, the results of these studies have not been consistent. For example, Niaura and colleagues6 reviewed studies of measured physiological responses of alcoholics to alcohol-related stimuli. One study showed a decrease in heart rate, two studies showed an increase in heart rate, and three other showed no significant heart-rate effects of exposure to alcohol stimuli. In addition, Drobes and Tiffany7 found that smokers exposed to smoking cures showed only modest physiological responses, but exhibited large changes in self-reports of craving. In follow-up work, Carter and Tiffany1 conducted a meta-analysis of cue-induced craving across studies of alcoholics, smokers, cocaine addicts, and heroin addicts. They found that alcoholics had less than half the effect size for self-reported craving than did other chemically dependent groups, suggesting heterogeneity of cue-induced craving among different types of substance use disorders.
The second and perhaps greater concern is that measures of craving and alcohol and drug use behaviors are not always strongly associated. There is little evidence that this “cue-specific” form of craving as measured in the laboratory has clinical predictive utility. That is, there is only modest evidence that this type of craving predicts clinical relapse.8 Cue exposure undoubtedly influences more than just craving. It also influences emotional, cognitive, and motivational domains. To date, only the study of Rohsenow and colleagues9 has demonstrated that cocaine craving predicts treatment outcomes. Interestingly, in their study they found that the urge to use in relapse situations was unrelated to self-reported cocaine-use cues or with negative mood states. Perkins8 has noted that very little evidence supports an association between the magnitude of cue-induced craving to smoke and duration of abstinence after a quit attempt, with the exception of a small study using the nicotine patch.10 Alcohol research has also failed to consistently show a relationship between self-reported craving and treatment outcome. For example, the NIAAA supported COMBINE trial failed to demonstrate an association between a craving measure and treatment outcomes.11 Although Oslin and colleagues12 demonstrated that craving could be incorporated into a typology with several other variables that predicted treatment outcome for inpatient VA alcoholics. Drummond and associates13 suggested four possible explanations for the failure self-reported craving to predict relapse:
1. Craving and relapse are unrelated.
2. Craving predicts relapse, but our measures of craving are inadequate
3. Craving predicts relapse, but only under certain yet to be determined conditions.
4. Self-reported craving does not predict relapse by itself, but other factors covary with self-reported craving and do predict relapse.
In summary, while “craving” represents a fertile area for ongoing research in the substance use disorders, the lack of specificity of the construct and the inconsistencies in its capacity to predict relapse or treatment outcome suggests that it is not yet ready for inclusion in the DSM-5 diagnostic criteria. For me, what is most amazing is that the Substance Use Disorder Work group has recommended the construct’s inclusion in our diagnostic criteria despite the lack of a scientific or empirical rationale. Opinions go only so far, and if addiction psychiatry is to have a scientifically rigorous diagnostic system, it needs to rely on scientific methods to establish the ground rules. However, it is clear that much more research remains to be done in this particular scientific arena to bring “craving” into better focus.