Featured Canadian Research on Child,Adolescent,and Young Adult Psychiatry

This month’s issue of The CJP contains four articles focused on the theme of child and adolescent psychiatry, and mental health in young adults. The first of these studies is a systematic review by Roseman et al. ¹ on the topic of depression screening in children and adolescents. Screening is a controversial topic in psychiatry, partially because of divergent concepts of what it actually means to screen. According to a 1968 World Health Organization Report, screening is the “presumptive identification of unrecognized disease or defect by the application of tests, examinations, or other procedures which can be applied rapidly.” ² The UK National Screening Committee defines it as: “The systematic application of a test or inquiry, to identify individuals at sufficient risk of a specific disorder to warrant further investigation or direct preventive action, amongst persons who have not sought medical attention on account of symptoms of that disorder.” Clinicians, on the other hand, increasingly use the term in a less formal sense to refer to any type of questions or assessment that that might be included in a routine clinical encounter. Internationally, recommendations for or against screening for depression are inconsistent.

Implementation of screening as a formal program can be costly and burdensome ³ and therefore requires, at minimum, evidence of benefit. While it is perhaps tempting to assume that screening would necessarily produce benefit, there is also the possibility of offsetting harm. False positive screens might divert limited mental health resources away from those with greater needs. Mild or self-limited cases might be preferentially detected, with negligible benefit. What sort of evidence is needed? This is an important question since observational studies of screening are subject to bias. This has been extensively acknowledged in non-psychiatric screening, where outcomes in progressive diseases may include mortality as an endpoint. One notorious example is lead time bias, in which earlier detection creates an artefactual appearance of longer survival even in the absence of any impact on the natural course of the disease. Another issue is length bias, in which screening leads to the detection of more slowly progressing cases, leading to a biased estimate of effectiveness. Avid participants in screening programs often have personal characteristics that affect outcomes, such as being more attentive to their health. This is a potential source of selection bias. Randomized controlled trials (RCTs) protect against such biases. Random assignment helps to ensure that all personal and clinical characteristics (even those that are unknown or unmeasured) are equally distributed between groups being compared. Consequently, Roseman et al. focus their literature search on RCTs of screening for depression in adolescents. However, whereas most systematic reviews are reported in lengthy technical manuscripts, theirs is a brief report. They find no RCTs of depression screening in children and adolescents. Indeed they find no useful non-randomized trials either. There was no high level evidence to review. Of course, an absence of evidence for benefit is not the same as evidence of absence of benefit. Nevertheless, their paper suggests that screening for depression in children and adolescents cannot at this time be considered an evidence-based strategy.

Ferro et al. focus on a young adult population, aged 15-30 sampled in the Canadian Community Health Survey—Mental Health, conducted by Statistics Canada in 2012, ⁴ with the goal of evaluating the possible association of chronic illnesses with suicidal thoughts, plans and attempts in this population. ⁵ They use a multinomial regression model to analyze their data. This model is similar to the more familiar logistic regression model for a binary data, but is capable of handling multiple categories, in this case thoughts, plans and attempts. They report evidence that both chronic conditions and psychiatric disorders are associated with suicidal thoughts, plans and actions. A unique aspect of their study is an examination of moderation or modification of effect, evaluated through the use of interaction terms in their models. They find that mood disorders increase the frequency of suicidal thoughts in young people with and without chronic medical conditions, but to a greater extent in those with chronic conditions. People with both mood disorders and chronic medical conditions have a higher frequency of suicidal thoughts than would be expected based on the occurrence of either condition on its own. Notably, the authors conclude that clinicians should “routinely ask” about suicidal thoughts and behaviour in adolescents with chronic conditions, a recommendation that falls short of formalized screening and which takes the shape of a recommendation for clinical vigilance, consistent with Roseman et al.’s conclusions about depression screening.

In this same issue, Ilies et al. ⁶ report data from a monitoring program for metabolic effects of second generation antipsychotic (SGA) treatment in children and adolescents (mean age 12.8 years). Their goal is to compare metabolic outcomes in monotherapy versus polytherapy. It should be emphasized that their concept of polytherapy includes both switches from one SGA to another and simultaneously taking two or more SGAs. They report no significant difference between these groups, but qualify their results by acknowledging that their analysis may have lacked statistical power to detect a difference. Power is the probability of finding a statistically significant difference between groups when a difference truly exists. A non-significant statistical test may have occurred even if there was is a higher rate of metabolic abnormalities in association with polytherapy in the population studied. In other words, this non-significant result should be interpreted as a failure to reject a null hypothesis (of no difference between the groups compared) rather than as providing evidence that no difference exists. Nevertheless, such results make an important contribution to the literature. Meta-analyses allow power to be gained through pooling of study results, and such studies may ultimately provide more a decisive answer to their question, so it is essential that such studies be published. The Ilies et al. results are also important for another reason: they provide a real world description of some of the risks associated with SGA treatment. Of patients treated for 24 months there was a mean weight gain of 12.8 kg. Increases were also observed in fasting blood glucose and BMI z-score.

While a metabolic monitoring system such as the Second-Generation Antipsychotic Monitoring Program at the Hôtel-Dieu de Lévis is a good source of “real world” data for research it may also serve some of the purposes of a screening program—ideally, the individual data collected would feed back to inform individual clinical decisions. Consistent with the concept of precision medicine, such data can inform group comparisons, as presented by Ilies et al. in this issue, ⁶ but may also be able to contribute to more precise and personalized clinical decisions. The identification of elevated fasting glucose for example, may represent a latent stage of progression towards Type II diabetes. Early detection, coupled with effective intervention, may facilitate prevention and is closer to the classical concept of screening than the depression screening (which tends to focus on increased recognition rather than earlier detection) literature targeted by the Roseman et al. ¹ review.

Vasiliadis et al. ⁷ use administrative data (physician billing records and hospital discharge abstracts) to examine the epidemiology of ADHD in four Canadian provinces from 1999 to 2012. Since their data sources are person-specific and longitudinal they are able to examine both annual period prevalence (proportion of the population assigned the diagnosis during one year) and incidence (new diagnoses assigned annually). In three of the four provinces both incidence and prevalence increased over this interval whereas in Ontario prevalence increased, but not incidence. A common explanation for increasing prevalence in the face of stable incidence is changing duration of a disease. Prevalence (more precisely the point prevalence odds) is the product of the incidence rate and the mean duration of a disease. Increasing persistence of diagnoses of the condition (e.g. into adulthood) could explain this pattern. Connecting this observation to the concept of screening in child and adolescent psychiatry is the counter-intuitive possibility that screening, by leading to earlier detection, can be expected to increase the treated prevalence of a condition by prolonging its apparent duration. Since this study examined treated prevalence, the changes may reflect either actual changes in the epidemiology or, more likely, changing diagnostic practices. In turn, the prevalence estimates reported are lower than those typically reported in population surveys, which may indicate that the treated prevalence estimates pertain to a subset of more severe cases (excluding, for example, cases identified in schools and managed without medical intervention), or it may suggest that survey data overestimate the prevalence. For a complex condition such as ADHD, where diagnostic criteria and thresholds are fluid and even controversial ^{8 –10} no single study of the epidemiology is likely to address all of the uncertainties, however, the aggregation of vast amounts of data by Vasiliadis et al. ⁷ provides a firm reference point, supporting an increasingly precise understanding of the epidemiology of this condition.