This study finds that substance use in the past 30 days varies with past year unmet need for mental health care and mental health care use in ways predicted by the self-medication hypothesis. However, the nature of the effects varied across the three substance categories. In no category did we observe increased substance use associated with unmet need and decreased substance use associated with mental health care use. Use of illicit drugs other than marijuana increased with unmet need, but was not reduced with mental health care use. Heavy alcohol use was not associated with increased unmet need for mental health care, but was reduced with mental health care use. By contrast, marijuana use did not appear associated with either unmet need or mental health care use.
Notable are differences in our findings across the substance categories. Stronger effects of mental health care use on alcohol may suggest that mental health treatment may be more effective in treating the underlying symptoms that prompt alcohol use. Alternatively, mental health patients may be better screened for alcohol use than for other types of drug use. Likewise, providers may be more forthcoming with advice about the exacerbating effects of alcohol than they are about other forms of substances.
At the same time, the lack of relationship between unmet need for mental health care and heavy alcohol use is somewhat surprising given the frequent observation that alcohol use is common among those with untreated mental health conditions. This lack of relationship suggests that people are not recognizing their mental health symptoms as treatable or, perhaps, attributing symptoms to alcohol use rather than mental health problems.
It is interesting to contrast studies of the relationship between the use of “unconventional” treatments (e.g., homeopathic remedies, herbs) and the use of professionally sanctioned services. Unlike the negative association predicted and observed in this study, previous research has generally found a positive correlation between use of unconventional and conventional medical services (Eisenberg et al. 2001
; Druss and Rosenheck 1999
; 2001) and a positive (Kessler et al. 2001
) or insignificant (Unutzer et al. 2000
) correlation in the case of mental health care. Several factors help explain the differences in observed correlations: (1) patients may perceive positive benefits from unconventional treatments, while the negative effects of substance use are well known and (2) unlike the use of unconventional treatments, substance use is more heavily stigmatized than alternative medications and is, in many cases, illegal.
These findings suggest that timely and appropriate mental health treatment may prevent the development of substance-use disorders. However, the implications of our findings for clinical policy depend on an improved understanding of the sources of perceived unmet need. Unfortunately, sample size constraints limited our ability to explore the effect of interactions between unmet need and the use of mental health care use on substance use. In our sample, among individuals with unmet need, 48 percent received no treatment in the past year. Their reasons for not receiving treatment may be similar to those given by National Comorbidity Survey respondents with serious mental illness who perceived a need for treatment, but received no treatment in the past 12 months (Kessler et al. 2001
): wanted to solve problem on own (72.1 percent); thought problem would get better by itself (60.6 percent); perceived lack of effectiveness (45.4 percent); had financial barriers (45.6 percent); were unsure about where to go for help (40.8 percent); and thought it would take too much time or be inconvenient (27.7 percent).
Although we know of no published study that examines the determinants of unmet need among those who have received services in the past 12 months (in our sample, 52 percent of those with unmet need), we can speculate. Some might have felt treatment was not effective or not acceptable; others may have been stigmatized by neighbors or coworkers, and therefore exited treatment; some might have experienced partial improvement, but financial barriers precluded further treatment; and others might have found treatment to be inconvenient.
The results of this study should be interpreted in light of several limitations. First, while NHSDA/NSDUH offers several advantages over other data sources, it is not possible to unambiguously determine whether the mental health use and need occurred prior to reported substance use and if so, whether need and use of mental health care can meaningfully influence current substance use in the relatively short span of 12 months. Second, the surveys lack a measure of chronic physical pain. While respondents over the age of 65 years were excluded from the analysis and the multivariate analyses contain an indicator of self-reported fair or poor health, these steps are probably insufficient to fully control for the simultaneous effects of pain on mental health and substance use (Rosenblum et al. 2003
). However, the inability to fully control for higher use of both substances and mental health care among those in chronic pain would bias the results away from finding evidence of self-medication.
Third, it is reasonable to expect that a subgroup of respondents do not report unmet need for mental health care because their use of substances has satisfactorily reduced symptoms at the time of the interview. However, this type of behavior would reduce observed differences in substance use by unmet need, making inferences about self-medication conservative.
Fourth, although we attempt to exclude the subpopulation for whom established substance-use disorders contribute to mental health problems, it is unlikely that we have been fully successful and the effect on our inferences is different depending on the measure. To the extent that these individuals remain in the sample, psychological and physiological effects of dependency not treated by mental health care providers would serve to reduce negative differences in substance use by mental health care use, resulting in conservative inferences about self-medicating behavior. The effect on unmet need is less clear. To the extent that individuals with substance-induced mental health problems perceive mental health care as less accessible or less effective, then our findings will overstate the positive association between mental health care use and unmet need for mental health care.
Finally, one alternative explanation for the positive relationship between unmet need and illicit drug use is the so-called “Wrath of Grapes” phenomenon observed in clinical interviews (Frances 1997
). Here, patients use mental health problems as a post hoc justification for substance use. However, it is not clear whether this occurs among nondependent users like the ones in our sample or in the context of survey administration.
Our study highlights the potentially valuable role of research on perceived mental health care need and service use on the underlying causes of and treatments for co-occurring disorders. While not a causal test, our study represents a first attempt to inform the self-medication hypothesis from a health services perspective. The positive relationship between unmet need and illicit drug use and the negative relationship between mental health care use and heavy alcohol use among those without substance dependency is consistent with the behavior predicted under the self-medication hypothesis and suggests that mental health treatment may prevent the development of substance-use disorders.