Early-adult outcome of child and adolescent mental disorders as evidenced by a national-based case register survey

Abstract Background Mental disorders show varying degrees of continuity from childhood to adulthood. This study addresses the relationship of child and adolescent mental disorders to early adult psychiatric morbidity. Methods From a population at risk of 830,819 children and adolescents aged 6-16 years, we selected all those (n = 6043) who were enrolled for the first time in the Danish Psychiatric Register with an ICD-10 F00-99 diagnosis in 1995-1997, and identified any mental disorder for which they received treatment up to 2009. Results Neurodevelopmental and conduct disorders were the principal diagnostic groups at 6-16 years and exhibited a characteristic male preponderance; while affective, eating, neurotic, stress-related and adjustment disorders were more common in girls. Over a mean follow-up period of 10.1 years, 1666 (27.6%) cases, mean age 23.4 years, were referred for treatment to mental health services, and they had a markedly higher risk than the general population (RR 5.1; 95% CI 4.9-5.4). Affective, eating, neurodevelopmental, obsessive-compulsive and psychotic disorders had the strongest continuity. Heterotypic transitions were observed for affective, eating, neurodevelopmental, personality and substance use disorders. Conclusions These findings suggest that individuals with psychiatric antecedents in childhood and adolescence had a high risk of being referred for treatment in early adulthood, and many mental disorders for which they required treatment revealed both homotypic and heterotypic continuity.


Introduction
Major depressive disorder (MDD) is a frequent psychiatric condition with disabling symptoms leading to severe functional impairments and is a major contributor to the overall burden of disease [1]. Despite advances in psychopharmacology and the high number of available pharmacological antidepressant options for MDD, about 30% of patients fail to respond to first-line treatments at adequate dose and duration [2]. After a treatment response failure, international guidelines recommend switching medication to another pharmacological class of antidepressant but symptoms remain in 50% of patients after a response failure to a first line treatment [2], justifying the development of novel therapeutic approaches.
More recently, repetitive transcranial magnetic stimulation (rTMS) has been proposed as a potential therapeutic option after response failure to at least one antidepressant medication [3,4]. The first studies assessing antidepressant effects of rTMS used high frequency (HF) rTMS applied over the left dorsolateral prefrontal cortex (LDLPFC [5]). Other studies have also investigated the clinical efficacy of low frequency (LF) rTMS applied over the right DLPFC (RDLPFC [6]). Comparative studies reported that LF rTMS over the RDLPFC and HF rTMS over the LDLPFC displayed similar beneficial effects on clinical outcome [7,8]. Despite the fact that numerous studies and meta-analyses reported the clinical efficacy of rTMS in TRD and a level A recommendation (definite efficacy) from the European guidelines [4], there is still a lack of studies investigating sociodemographic, clinical and neurobiological features associated with the response status to DLPFC rTMS [9,10]. Regarding sociodemographic and clinical predictors, advanced age, long duration of the depressive episode, and high levels of treatment resistance were described as negative predictors of response to rTMS in several studies [11], but these results remain controversial and non specific [12]. One can hypothesize that a better characterization of sociodemographic and clinical characteristics of responders and non-responders is an important way to refine the place of rTMS in the treatment algorithm for depression in clinical settings.
The Montgomery and Asberg depression rating scale (MADRS [13]) is a clinician-rated scale designed to be sensitive to change, and one of the most commonly used scales to evaluate the severity of depression. The MADRS evaluates depression based on 10 items and each item yields a score of 0 (absent) to 6 (extreme): apparent sadness, reported sadness, inner tension, reduced sleep, reduced appetite, concentration difficulties, lassitude, inability to feel, pessimistic thoughts and suicidal thoughts. Several studies have undertaken a factorial analysis of the MADRS to evaluate the number of independent factors. However, there is no consensus about the number of factor structures of the MADRS, which varies from 1 to 4 [14][15][16][17]. Several factors such as the severity of depression, the duration of the episode, the heterogeneity between treatments, primary diagnosis (uni-versus bi-polar MDD) may explain these discrepancies. Suzuki et al. [16] conducted a principal component analysis of MADRS scores in 132 patients with MDD. They found that a 3-factor model accounted for 61% of the total variance. The factor 1 (defined as ''dysphoria'') included reported sadness, pessimistic thoughts, and suicidal thoughts; the factor 2 (''retardation'') included lassitude, inability to feel, apparent sadness, and concentration difficulties; factor 3 (''vegetative symptoms'') included reduced sleep, reduced appetite, and inner tension. This 3-factor model is the most commonly used model to assess clinical features of responders and non-responders to numerous treatment approaches in patients with MDD. For instance, Okazaki et al. [18] reported that responders to electroconvulsive therapy (ECT) have higher baseline scores at MADRS dysphoria factor than non-responders. They also reported that responders and non-responders only differed by their scores at dysphoria and retardation factors after the end of ECT sessions suggesting that ECT may have small effects on the vegetative factor. In patients receiving transcranial direct current stimulation (tDCS) with the anode over the left DLPFC coupled with the cathode over the right supraorbital region, responders displayed greater improvements in dysphoria and retardation factors than nonresponders after 6 weeks of active tDCS. The authors reported that tDCS had no specific effects on the vegetative factor suggesting that, as seen with ECT, high vegetative symptoms at baseline scores should be a negative predictive factor [19]. In patients with MDD receiving fluvoxamine, a low vegetative symptoms factor baseline score was a good predictor of beneficial clinical response [20]. Thus, this factor model could potentially help differentiate between treatments and be a criterion of choice in the therapeutic switch.
In the present study, we investigated the 3 factor MADRS subscores in responders and non-responders who received 2 to 6 weeks of low frequency rTMS applied over the right DLPFC. We also investigated sociodemographic and clinical differences between these 2 populations.

Material and methods
The present study aim to retrospectively analyzed scores on a 3factor model of MADRS [16] in unipolar patients with TRD receiving 2 to 6 weeks of daily 1 Hz rTMS applied over the right DLPFC. Participants were extracted from a larger sample of 170 participants included in a multicentric 3-arm randomized, sham-controlled trial investigating the effects of low frequency rTMS in combination with venlafaxine [21]. Patients had a diagnosis of major depressive disorder (MDD) according to DSM IV criteria (using MINI 5.0 interview) and a Hamilton Depression Rating Scale (HDRS) score superior to 20. The patients had 2.5 AE 1.8 failed previous treatment (range [1][2][3][4][5][6][7][8][9][10][11][12] and the mean duration of the current episode was 14.1 AE 17.8 weeks. Patients under 18 or with other axis I disorders (except anxiety disorders), substance use disorder (except nicotine), psychotic features, somatic or neurological disorders, failure to respond to venlafaxine during the current depressive episode, or during pregnancy were excluded. Patient who previously benefited from an rTMS course or with rTMS contraindication could not be enrolled in the study. Only data from patients included in the ''active rTMS in combination with placebo venlafaxine'' arm were analyzed in order to specifically investigate characteristics of patients according to their response status to rTMS ( Table 1). Patients were enrolled in the study after a withdrawal phase of 1 to 3 weeks for any antidepressant and anxiolytic medication (only hydroxyzine and cyanemazine were allowed). Patients thus received only rTMS as active antidepressant treatment and were free for any active antidepressant medication. The treatment course was performed using a Magpro Â 100 (Mag2Health, France) or a Magstim-Super rapid (Inomed, France) stimulator system with a 70-mm figureeight coil targeted the right DLPFC which position was determined 6 cm anteriorly to the motor hotspot. The stimulation intensity was set at 120% of the resting motor threshold (RMT) identified by visual inspection. The stimulation session consisted of 6 trains of 1-min duration separated by 30-sec inter-train ''off'' periods at a 1 Hz frequency. Total duration of rTMS session was 8 min 30 s. Treatment course was delivered throughout one daily session on 5 consecutive working days from Monday to Friday for at least 2 weeks (until remission), with a maximum of 6 weeks. Sham procedure was performed using a similar shaped coil with a shield and a synchronized skin stimulation as published in our previous study [21]. The study was approved by a local ethics committee (CPP Sud-Est 6, France, #AU732) and registered (NCT00714090).
The final analyzed sample consisted in 54 patients, 20 males and 34 females, aged from 28 to 79 years old (mean = 54.1 AE standard deviation = 12.2). The mean number of previous treatment failure was 2.2 AE 1.5. There were 3 left-handers, 2 ambidextrous and 49 righthanded patients. The mean severity of depression measured by MADRS scores was 32.5 AE 6.3. Patients were divided into two groups based on their response profile at the end of stimulation sessions. Response was defined as a decrease of at least 50% between baseline MADRS total score and MADRS total score at end point (i.e., after 2 to 6 weeks of rTMS). At endpoint, 29 patients were classified as responders and 25 were non-responders to rTMS.
Our primary outcome was the MADRS sub-scores. The effect of rTMS on the 3 factors of the MADRS (dysphoria, retardation and vegetative symptoms) was analyzed separately using a 2 Â 2 twofactor ANOVA with repeated measures with the group (responders, non-responders) and the time (baseline, endpoint) as conditions. In case of significance of the ANOVA, baseline and endpoint differences between groups were analyzed using 2-tailed Student t tests. Significance was set at P < 0.05.
Regarding secondary outcomes, sociodemographic (i.e., age, gender, handedness, educational level) and clinical features (i.e., number of previous hospitalizations, suicide attempts and previous failed treatments, the duration of the ongoing depressive episode and the smoking status) differences between responders and non-responders were analyzed using 2-tailed Student t tests for quantitative variables and using Fischer' Exact tests for qualitative variables. We applied Bonferroni correction for multiple analysis (Sidak's adjustment for 9 comparisons) and significance level was thus set at P < 0.005.

Whole sample analysis
In the whole sample with 54 patients, we reported a significant decrease of MADRS scores from 32.5 AE 6.3 to 17.7 AE 11.8 (P < 0.0001), corresponding to a 45.3% decrease.

Sociodemographic and clinical characteristics in responders and non-responders
There were no significant differences between responders and non-responders regarding the age of participants, the gender, the handedness, the educational level, the number of previous hospitalization, the number of previous suicide attempts, the number of previous treatment failure and the duration of current depressive episode.
There was a significant difference between groups regarding the smoking status. Namely, we observed a significant difference between the numbers of ex-smokers between groups; there were 8 ex-smokers in the responder group (i.e., 28%) whereas no exsmokers (0%) were observed in the non-responder group (P = 0.005).

Discussion
The aim of this study was to investigate clinical and sociodemographic characteristics of patients according to their response status towards rTMS in a sample of unipolar antidepressant-free TRD patients receiving low frequency rTMS applied over the right DLPFC. We reported that low frequency rTMS has a significant beneficial effect on the 3 factors of the MADRS: dysphoria, retardation and vegetative symptoms. We found that responders had significantly lower baseline scores in the retardation dimension (13.6 AE 2.9) than non-responders to rTMS (15.6 AE 2.9), suggesting that rTMS was not the optimal treatment approach in patients with high retardation symptoms (i.e., lassitude, inability to feel, apparent sadness, and concentration difficulties). Analysis of sociodemographic and clinical characteristics of participants revealed that only the smoking status was different between responders and non-responders. Ex-smokers were more prone to respond to rTMS than current smokers and non-smokers. These results suggest that to be an ex-smoker and to experience relatively few retardation symptoms may be associated with better clinical response to rTMS.
Other studies have investigated the severity of retardation, psychomotor retardation and psychomotor agitation in patients with MDD receiving rTMS. In contrast with our results using low frequency rTMS applied over the right DLPFC, they reported that a high level of retardation was a good predictive outcome in patients with MDD. In a retrospective study, it has been observed that sleep disturbances, high levels of psychomotor retardation and low levels of psychomotor agitation measured by the CORE scale were good predictors of clinical response [22]. These results were however not replicated in a prospective open-label study conducted by the same group of authors, leading to the conclusion that HDRS items 1 (depressed mood) and 2 (feelings of guilt) were negative predictors and that high retardation score was a positive predictor to high frequency rTMS in a sample of drug-free patients with MDD [23]. These results, showing that high retardation symptoms could be a positive predictor of clinical outcome in patients with MDD, were in line with findings in ECT studies. Indeed, numerous studies have observed that ECT was more effective in patients with high psychomotor retardation or even in agitated patients [24]. Moreover, retardation is assumed to be a predictor for response to ECT [25,26]. One can hypothesize that ECT and high frequency rTMS applied over the left DLPFC shared common mechanisms of action in patients with retardation symptoms that are different than those observed with low frequency rTMS applied over the right DLPFC. In that way, Fitzgerald et al. [27] investigated the predictive value of psychomotor agitation, measured by the agitation subscore of the CORE in patients receiving high frequency rTMS over the left DLPFC compared with patients receiving low frequency rTMS over the right DLPFC. They reported that a high CORE agitation score predicted response in patients receiving low frequency rTMS over the right DLPFC group but not in patients receiving high frequency stimulation over the left DLPFC. We suggested that based on retardation/agitation symptoms, patients should be engaged in high or low frequency rTMS protocols. Lower ''retardation'' scores combined with higher agitation scores should be associated with a better response to low frequency rTMS. High retardation scores should be a good predictor of response to high frequency rTMS and ECT.
Another important finding of our study is that other measured clinical and sociodemographic features were not different between responders and non-responders to low frequency rTMS. In our study, the level of treatment resistance measured by the number of previous treatment failure was not different between responders and non-responders. This result is in contradiction with several studies reporting that a low degree of prior treatment resistance is a predictor of positive treatment outcome [11,12,22]. It is however important to note that patients in our sample had relatively low levels of resistance (1.9 AE 0.9 previous treatments failure in the responder group versus 2.5 AE 2.0 in the non-responder group) as compared with previous studies such as the study developed by Brakemeier et al. [23] (4.6 previous treatments failure in the responder group versus 5.3 in the non-responder group). Regarding the duration of the current depressive episode, a shorter duration was reported to be a predictor of positive treatment outcome in several studies [12,28]. The effect seems to be particularly relevant for very long depressive episodes. For instance, Holtzheimer et al. (2004) [28] reported that patients with an episode duration shorter than 4 years had a mean depression score decrease of 52% whereas those with an episode duration longer than 10 years only displays a 6% decrease in depression scores. Experiencing psychotic features was an exclusion criterion in our study; it was thus not possible for us to investigate the relevance of this factor as a positive [29] or negative predictor [30,31] in our sample of patients receiving low frequency rTMS.
Some studies have suggested that certain sociodemographic features might have an influence on positive outcome. Here, we reported no effects of age, educational level, gender and handedness on the beneficial effect of rTMS. These results corroborate previous findings [12] but are in contradiction with others reporting an inverse correlation between age and clinical response in patients with MDD [11,32]. It is important to note that in most studies investigating the effects of clinical and sociodemographic features, patients received high frequency rTMS over the left DLPFC, thus the effects of these parameters in patients receiving low frequency rTMS remain unclear. These two methods are assumed to induce opposing effects on motor cortex excitability and one can hypothesize that predictors of response will allow differentiating responders from non-responders to one or the other method.
Finally, we reported that the smoking status might have an effect on the treatment outcome in patients with MDD. We observed that 100% of the ex-smokers included in the study were responders to low frequency rTMS. This effect could be explained by an interaction between nicotine and mechanisms by which rTMS modulates brain plasticity as well as by the brain state differences between smokers, non-smokers and ex-smokers. In line with the latter, some imaging studies reported that exsmokers have significantly stronger activity in the prefrontal brain structures than smokers and non-smokers [33,34]. Thus, one can hypothesize that the prefrontal hyper-activity in ex-smokers may lead to an increased sensitivity to low frequency rTMS as a priming effect. Other studies have also previously reported an interaction between the smoking status and the therapeutic effect of non invasive brain stimulation techniques. For instance, in hallucinating patients with schizophrenia, it was reported that current smokers were non-responders to transcranial direct current stimulation whereas non-smokers displayed a significant decrease of their hallucinations following brain stimulation protocol [35].
The main limitation of the present study is the retrospective nature of the analyses and the relatively small sample size. For this reason, we could not undertake a statistical approach to determine the predictive value of the retardation score and the smoking status on clinical outcome. Moreover, results were extracted from a previous published sample of patients [21]. The lack of sham control group could constitute a limit for interpretation of the results but reflects reality in clinical settings.

Conclusion
Further prospective rTMS studies with larger sample size and ad hoc statistics are needed to confirm the difference of retardation scores and smoking status between responders and non-responders, and to explore the predictive value of these features towards response to low frequency rTMS in unipolar patients with TRD.

Role of the funding source
The funding sources have no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the manuscript for publication.

Disclosure of interest
The authors declare that they have no competing interest.