The course of mental distress
The level of mental distress at follow-up was as high as that at admission (2.01 ± 0.63 vs. 2.08 ± 0.61, respectively, P = 0.219), and far above the level in the general population. In a population study from Norway [31] using the HSCL-25, the mean score was 1.33 (SD:SE = 0.004). The stability of mental distress over six years is supported by some of the literature [13–16], and shows that, on a group level among substance-dependent subjects who seek treatment, the level of mental distress will not go away by itself or with time. The level of mental distress in our sample was also high at both baseline and follow-up, even when compared with other follow-up studies of SUD patients [17, 39]. We also found more Axis I disorders at baseline than did other studies, like ours, that used CIDI [4, 40, 41].
In the different subgroups, the most important finding was that mental distress substantially decreased in the "abstinent" group, especially among women, whereas it remained stable and high in the "relapse" group. The mean score at follow up among the "abstinent" women (1.41 ± 0.37) was similar to the level of mental distress in the general population study [31], where the mean level for women was 1.37 (SD:SE = 0.006). Among "abstinent" men (1.69 ± 0.66) mental distress was higher than for men in the population study1.28 (SD:SE = 0.006). Therefore, our group of abstinent women seems to have obtained a distribution of mental distress that is very similar to that of women in the general population, whereas the abstinent group of men still has a somewhat higher level of distress than that of men in the general population.
Our findings may also indicate that mental distress in women is more strongly associated with ongoing addictive behaviour than it is among men. In a 16-month follow-up study, Tomasson and Vaglum found a significant relationship between duration of sobriety and lower psychiatric distress, but this association was less clear for women [23]. In our multivariate analyses, sex was not an independent predictor of the level of mental distress at T2.
In all, it is important to remember that all studies, including our own, show that eliminating the addictive behaviour does not always lead to a clinically significant reduction in mental distress. Consequently, Axis I and II disorders, and not only the SUD, should be addressed in treatment programmes.
Among single Axis I disorders, only patients with simple phobias and PTSD had a significant decrease in mental distress. In our study, as in others [42], simple phobias were not as severe as many of the other Axis I disorders and are therefore possibly easier to eradicate. The decrease in mental distress we observed in patients with PTSD is in contrast to a review of SUD patients with PTSD [43] that reported poor outcomes on mental distress and that the negative effect of a comorbid PTSD diagnosis was greater than the effects of other comorbid psychiatric disorders. However, the generalizability of the studies included in that review can be discussed with regard to both sampling (male SUD patients in VA treatment centres) and methodology (lack of structured clinical interview). When controlling for baseline mental distress only a few of the differences in table 2 and 3 remained significant, primarily the variables with the greatest differences in HSCL-25 score at T2. Much of the impact of the various Axis I and II disorders on HSCL-25 at T2 was due to the HSCL-25 level at index admission.
Predictors of mental distress at follow-up
The number of Axis I disorders was significant and independently related to the level of mental distress when we controlled for age, sex, Axis II disorders, and substance-use variables at both T1 and T2. This is in accordance with the results of several studies with shorter follow-up interval [20–24]. Only patients with ≥ 4 Axis I disorders at baseline reported a significant decrease in their levels of mental distress from baseline to follow up. Nevertheless, the level of mental distress was still very high in this subgroup. Grella et al. [6], Melberg et al. [39], and Liskow et al. [44] also found the greatest reduction in psychiatric severity in those groups who were highest at baseline. This may be due to the fact that patients seek help when their mental distress is at a maximum.
In the multivariate analyses, anxiety disorders (GAD, SAD, sum of anxiety disorders) together with somatization disorders were significant and independently related to a high level of mental distress at follow-up. Haver and Gjestad [45] also found phobic anxiety to be an important predictor of outcome in female alcohol-dependent patients, partly due to the influence phobic anxiety had on depression. In general, anxiety disorders are more stable than affective disorders [46], and they should always be the focus of therapeutic interventions among SUD patients. Compared with the other Axis I disorders that had a significant and independent impact on mental distress at follow-up, social anxiety disorder was a more important predictor of mental distress. There is a high degree of impairment associated with SAD and higher risk of persistence compared with that of other anxiety disorders [46, 47]. Patients with SAD in our index sample experienced more comorbid mental disorders than did patients without SAD [27], and most patients with SAD in our sample exhibited generalized SAD, a subtype that causes more impairment than the non-generalized type [48].
No single affective disorder had an impact on the level of mental distress at follow-up, but because the total number of Axis I disorders had such an impact, affective disorders may have made a more non-specific contribution. The literature concerning SUD patients with depression confirms our finding that psychiatric symptoms may improve when sobriety occurs. Hatsukami and Pickens [49] and Driessen et al. [50] also reported that the rate and severity of depressive symptoms among sober alcohol-dependent patients were similar to those of the general population. However, those studies were short-term follow-up studies. In a 10-year follow-up study of patients with major depressive disorders with and without alcohol dependence, there was a twofold greater likelihood of improvement in the major depressive disorders in patients without a current alcohol disorder compared with that of patients with an active alcohol disorder [51]. In a five-year follow-up study of alcohol-dependent patients with major depression, remission of alcoholism strongly and significantly increased the chances of remission of depression [52]. Hodgins et al. followed alcohol-dependent patients for three years and found that among patients with good drinking outcomes, 26% suffered depression compared with 60% of those with poor drinking outcomes [18].
In the multivariate analyses none of the Axis I disorders or number of Axis I disorders was a predictor of mental distress at T2 when controlling for HSCL-25 at T1. Much of the impact of Axis I disorders was due to shared variance with HSCL-25. HSCL-25 is a dimensional measure of anxiety and depressive symptoms whereas diagnosing Axis I disorders is a categorical way of measuring mostly anxiety disorders and affective disorders, i.e. closely related phenomena. Thus, there was a high correlation between Axis I disorders and HSCL-25, not least between a tally of the number of Axis I disorders and HSCL-25 scores (r = 0.61).
Both number of personality disorders and some specific disorders, mostly of the anxious and dramatic types, were independently related to a relatively poor outcome concerning mental distress, when we also controlled for sex, age, number of Axis I disorders, and substance use both at baseline and at follow-up. Those with a personality disorder continued to have a high level of mental distress from T1 to T2. The persistence of mental distress in patients with many Axis II disorders is consistent with a recent review showing that Axis II disorders are often associated with poor outcomes for psychiatric problems among SUD patients [25]. This underlines once more the importance of considering the personality disorders of SUD patients. When entering HSCL-25 at T1 into the analysis, the number of PD's still remained a significant predictor of mental distress at T2. This underscores the long-term impact of personality disorders on mental distress. Number of disorders may be a more robust measure than each specific disorder. Among the specific Axis II disorders only passive-aggressive still remained significant.
Our findings in this study demonstrate the importance of anxiety disorders for the level of mental distress at the six-year follow-up, whereas another study of the same follow-up sample [53] has shown that major depression and agoraphobia are significant and independent predictors of a harmful use of substances in the year prior to follow up. Together, these findings underscore the need to offer substance-dependent patients treatment that focuses not only on their substance abuse, but also on their anxiety, affective, and personality disorders.
On bivariate analysis, there was no significant difference in the mental distress of alcohol-dependent and poly-substance-dependent patients. However, in the multivariate analysis, the main substance of abuse (alcohol = 1) became a significant predictor of a high level of mental distress at T2 when we controlled for age or age at onset of SUD. Alcohol-dependent patients were older than poly-substance-dependent patients and they were older at the onset of SUD. In patients with an onset of SUD before 18 years, there was a significantly higher level of mental distress at T2 among alcohol-dependent patients than among poly-substance-dependent patients (2.39 ± 0.43 vs. 2.06 ± 0.43, respectively, P = 0.051). Even if poly-substance-dependent patients have more Axis I and II disorders [2], the level of mental distress in purely alcohol-dependent patients should not be underestimated.
The subgroup with an early age at onset of SUD had a significantly higher level of mental distress at follow-up (when we controlled for all other variables) compared with that of the late-onset group. At baseline the EO group were younger (33.1 vs. 42.9 years, P = < 0.001), more frequently poly-substance dependent (72% vs. 21%, P = < 0.001), had more often only primary school (62% vs. 35%, P = 0.002) and more often any Axis I disorder (97% vs. 84%, P = 0.016). The high level of mental distress in the EO group is in accordance with the finding of Babor et al., who reported higher average distress in type B alcoholics (with early onset of alcoholism) than that in type A alcoholics (with late-onset of alcoholism), in both one-year and three-year follow-up studies [54]. It seems that patients with an early onset of SUD are more seriously disturbed and less integrated into society.
In contrast to the majority of studies, we did find a direct, independent association between substance-use variables at admission and the level of mental distress at follow-up when we controlled for sex, age, Axis I and II variables, and for substance use during the year preceding follow-up. It is also important to note that, when we controlled for substance use at follow-up (and in the previous year), both Axis I and II disorders and baseline SUD variables remained independent and significant predictors of mental distress at follow-up. This shows that the impact of Axis I and II variables on the course of mental distress does not totally act through the substance use in the follow-up period, but that each of the variables had an independent impact on the mental distress outcome. This implies that one cannot expect total improvement in the other, potentially very disturbing, mental disorders, by limiting treatment to the addictive behaviour. The substance abuse field should involve the diagnosis and treatment of Axis I and II disorders as disorders independent of SUD.
Limitations
A major strength of this study is that the sample is a consecutive sample, with heterogeneity of clients with regard to the number of facilities involved, the numbers of inpatients and outpatients, and the main substance of abuse. We have also used both a well-established structured interview (CIDI) and self-report instruments (MCMI, HSCL-25). The relatively long follow-up period (75 months) is also a strength, minimizing the "baseline effect" [22].
The representativeness of the sample is always a problem in clinical studies. The follow-up sample comprised only 20% of the patients that fulfilled the original inclusion criteria (137/690); 48% of those were included at index admission (137/287); and 54% of those from the index admission sample that were still living six years later (137/254). Our baseline sample was only somewhat skewed towards having fewer young poly-substance-dependent subjects compared with a national sample [26], but age has not been an important variable in this part of our study. In this follow-up study, we found no significant differences between dropouts and participants, either at baseline or in the follow-up sample, in terms of any of the central substance use or psychopathology variables. Nevertheless, a proportion of patients with high mental distress may have found it difficult to participate in the follow-up study, and we may have underestimated the level of mental distress at follow-up.
Although we assessed the most relevant substance use and psychopathology variables at baseline and at follow-up, other factors concerning substance use, such as important life events, treatment received, and social stability factors, may have influenced the outcome during the follow-up period. Furthermore, we were unable to examine either the stability of substance use or temporal changes in substance use. To evaluate mental distress at follow-up, a new CIDI interview would have been ideal, but due to a lack of resources, we had to use well-established self-report instruments with well-demonstrated sensitivity and specificity in measuring mental distress and in identifying harmful use of substances.
Another weakness of this study is that we did not have any confirmatory information regarding either the patients' use of substances nor mental health variables, but had to rely on the patients' willingness to report the truth. However, many recent studies have supported the validity of self-reports of substance use [55].
In all, although the external validity of the level of mental distress after six years may be unclear, we believe that the final sample we have followed up is well suited to identifying important predictors of mental distress among substance-abusing patients.