In line with the self-medication hypothesis, emotional disorders robustly predicted the onset of SUDs, whereas SUDs predicted the onset of emotional disorders in only one case. This is the first prospective study to our knowledge examining the role of both ADs and UMDs in the subsequent onset of SUDs. Emotional disorders predicted the onset of alcohol use disorders but not other SUDs. The heterogeneous group of drugs represented in non-alcohol SUDs may have obscured effects that would be observed for certain drugs but not others. As expected in a non-clinical adolescent sample, there were too few onsets of non-alcohol SUDs other than cannabis use disorder to examine whether emotional disorders predicted non-alcohol SUDs for specific drugs.
In contrast, the presence of SUDs generally did not confer risk for emotional disorders. Thus, these findings show stronger support for the self-medication hypothesis than the substance-induced enhancement theory. This finding conflicts with some previous research in MDD/SUD comorbidity (e.g., Fergusson et al., 2009
). There may be something unique about those who develop SUDs by late adolescence, as this is earlier than the typical age of onset for SUDs. Findings may have differed with an older sample.
In contrast to other studies (e.g., Falk et al., 2008
), SUDs did not predict the onset of panic disorder, even though cannabis use disorders, which have previously been shown to be associated with panic disorder onset (Kushner et al., 1990
), were well-represented in this sample. However, other drugs (e.g., stimulants) that may confer risk for panic disorder were not well-represented in this sample. Also, the low incidence of panic disorder over the follow-up period may have resulted in insufficient power to detect this effect. Future research should use longer follow-up periods to assess individuals into the mid-20s when panic disorder is more likely to onset (Kessler, Berglund et al., 2005
Additionally, PTSD was associated with increased risk of SUDs. Research investigating the directionality of PTSD-SUDs comorbidity has been mixed. Our finding adds to the literature demonstrating that PTSD precedes SUDs (Cottler et al., 1992
; Breslau et al., 2003
) and suggests that those with PTSD may misuse substances to reduce symptoms, but does not support the idea that those with SUDs are at greater risk for PTSD (see Chilcoat & Breslau, 1998
). Future research should explore this comorbidity longitudinally with older samples to evaluate whether age of onset of the first disorder (i.e., either SUD or PTSD) moderates this relation.
In addition, social anxiety disorder was associated with increased risk of AUDs, but AUDs did not predict social anxiety disorder, thus providing support for a unidirectional association. The social anxiety and alcohol association is well-documented in the literature on anxiety disorder/SUD comorbidity (e.g., Buckner et al., 2008
; Kushner et al., 1990
; Conger, 1956
). Individuals with social anxiety disorder may begin misusing alcohol as a safety behavior to reduce short-term anxiety during social gatherings in which alcohol is present, thereby maintaining anxiety by preventing the gathering of information to disconfirm beliefs that a negative outcome would occur in social situations if alcohol is not consumed. These individuals may develop an AUD after frequent use during social gatherings and develop beliefs that they need the alcohol in order to cope with social situations.
Although SUDs generally did not predict emotional disorder onset, AUDs predicted OCD onset. Little research has explored SUD/OCD comorbidity. Possibly, these individuals self-medicated sub-threshold OC symptoms (Fatseas et al., 2010
). Indeed, examination of the data revealed that over half of the participants with baseline AUDs who developed OCD over the follow-up had OC symptoms at baseline. These results should be interpreted cautiously, because there were few onsets of OCD over follow-up and results could be driven by a few outliers, as indicated by the large 95% confidence interval.
Although important conclusions can be drawn, the study did have limitations. The mean ages in this sample by the end of follow-up were in the early 20s, which is earlier than the mean age of onset for SUDs, panic disorder, and PTSD (Kessler et al., 2005b
; Compton et al., 2000
). Thus, although this sample represents an ideal age for examining first onsets of most emotional disorders, it is not ideal for identifying all emotional disorders or for SUD onsets. Future research should examine this question with older ages and longer follow-up periods. Also, consistent with epidemiological findings, onsets for some ADs had low frequencies, which may have resulted in low power to detect some effects.
The current data point to the importance of early detection of emotional disorders for identifying those who may be at risk for the development of SUDs later in adolescence or young adulthood. School counselors should assess for emotional disorders and be attuned to the increased risk this poses for later SUDs. SUD prevention programs should be developed for those with emotional disorders, and emotional disorder prevention programs should assess whether these programs also reduce risk for SUDs. Finally, treatment of emotional disorders in the context of SUD treatment may not only have a positive impact on mood and anxiety symptoms but on SUD outcomes as well. Thus, typical care for SUDs should include identifying and treating comorbid emotional disorders.
- A sample of 627 adolescents were assessed at baseline and over a four-year period for the presence of clinically significant unipolar mood disorders, anxiety disorders, and substance use disorders (SUDs).
- Anxiety and unipolar mood disorders conferred significant risk for SUD onset.
- Social anxiety disorder and PTSD in particular predicted onset of SUDs.
- SUDs in general were not associated with later onset of anxiety or unipolar mood disorders, but alcohol use disorders predicted the onset of obsessive compulsive disorder.