This month’s crop of papers provides the expected diverse and interesting array of studies on important questions in child psychology and psychiatry. Several commonalities stand out. There is, for instance, attention to familiar conditions, including autism, conduct disorder, and learning and cognitive processing difficulties; in addition, there are familiar kinds of exposures, such as maltreatment, early stress, and broad psychosocial risk. In order to gain leverage in testing hypotheses, investigators have employed familiar scientific techniques, such as family and sibling designs, longitudinal follow-up designs, and treatment designs. Finally, investigators harvest their findings using tools that are familiar to most readers, including careful observation, imaging, and secondary data analysis. However familiar the procedures are for data gathering and analyzing, each paper must stimulate the field to make an appreciable step forward in understanding the basic phenomena or appreciating the limits of models that were hitherto presumed to be adequate. In the new format, the authors state how it is that their results do just that (see “Key points” at the conclusion of the text); readers are encouraged to consider if they agree, disagree, or can embellish that section.

Each month, a joint editor has the opportunity to offer a commentary on the papers included in the issue or raise an issue of general concern to those in the child mental health field. I will take my turn to focus on three points raised in this month’s papers. The first is the matter of how well research instruments can be implemented in practice. Honda et al. provide some encouraging data on screening for autism in 18 month-olds using the YACHT-18. In addition to demonstrating a solid example of the science of screening, the investigators’ study is notable for being embedded within a public health context. That means that their methodology was embraced as relevant and manageable to a wide array of practitioners. (The process by which the adoption of the measure occurred is not covered in the paper, but might offer some valuable lessons to those who worry about how widely evidence-based assessment methods are used in typical clinical and community settings.) Honda et al. note that successful application of the instrument depends on “the experience and keen perception” of the public health nurses. One might infer that the measure makes ready use of the skills and interests of the public health nurses, and that probably contributed to the success of the project. Other real-life features of the noteworthy study were that it appreciated that distinction between assessment and intervention is not usually a simple one in practice, and that optimal detection of those needing services is an iterative, refining process.

In summary, the Honda et al. study is valuable not only for documenting that a particular screening measure for autism has good sensitivity and specificity, but also for demonstrating that a research instrument can be integrated into a larger public health context. There are many examples of how that does not occur; there are, for instance, strikingly few examples of evidence-based assessments of parenting quality that are used in the many settings in which they are needed, from the family court, to the child mental health clinic, to the social services department. The science of detection in practical settings has a long way to go (see Tough et al., 2007); more effort is needed to demonstrate that “research-grade” measures can be incorporated into familiar non-research settings.

The second topic concerns the fast and captivating pace of research on attachment and genetics. Spangler et al. provide valuable data to metabolize alongside that reported by many investigators in recent years (see, e.g., Barry et al., 2008; Gervai et al., 2007). Notable features of their study were careful observational assessment – a hallmark of good attachment research – and a consideration of multiple genetic candidates. The study was retro-fitted to test molecular genetic hypotheses. Notably, probably none of those so far reporting on this issue were originally designed to test genetic hypotheses. That makes the nearly consistent set of findings so far reported even more notable, and may illustrate the point that careful observational measurement and a focus on gene-environment interplay may be a way of circumventing the need for an abundance of subjects implied in traditional molecular genetics studies (that requirement, of course, imperils the collection of careful observational data). As to the latter point, it may not be a surprise that few if any of these studies with n’s ranging from 50-150 report main effects of a particular genetic candidate; that raises the counterintuitive possibility that these small studies may be better powered to detect interactions than main effects. In any event, we would do well to remember that there is a good deal of developmental biology that needs to be sorted out before we can really begin to interpret these associations.

One issue that has so far eluded attachment-informed studies of gene-environment interaction is the particular nature of the interaction. Statistics textbooks offer several scenarios of non-parallel lines that would yield a statistical interaction; nearly the full complement has been reported or presented in the case of gene-environment interactions involving attachment. That has led to a range of competing hypotheses about vulnerability, susceptibility, and suppression or exacerbation of genetic risk by caregiving context. It may be that this is where the small samples in these studies is most limiting. Samples contributing to this debate have been restricted to low/normal ambient risk or, less commonly, restricted to high risk. It may be that understanding the nature of the interaction – assuming that it is robust – will require much larger samples that index genetic risk across the wide continuum of caretaking causality. Sample pooling across sites may be a short-cut to achieving that.

The third topic is the helpful encroachment of child psychology and psychiatry into unfamiliar territory. Gatti et al. offer a good case for how developmental science can inform practices affecting young people in the case of juvenile justice. The findings, that more intensive juvenile justice intervention predicts more adult criminality independent of earlier delinquency, may be surprising and perhaps unwelcome to those in the juvenile justice system. It is hard to judge the degree of penetration of knowledge about child psychology and psychiatry into the juvenile justice system, but it is probably fair to say that it is rather less than systematic. There are lots of good reasons for this, such as the difficulty of orchestrating a sound study (for an exception see McNeil & Binder, 2007). Nonetheless, the juvenile justice and social service systems rely on developmental science as much as any child mental health clinic, and so should be a common setting for research to appear in this journal.


  1. Top of page
  2. References
  • Barry, R.A., Kochanska, G., & Philibert, R.A. (2008). G x E interaction in the organization of attachment: mothers’ responsiveness as a moderator of children’s genotypes. Journal of Child Psychology and Psychiatry, 49, 13131320.
  • Gervai, J., Novak, A., Lakatos, K., Toth, I., Davis, I., Ronai, Z., Nemoda, Z., Sasvari-Szekely, M., Bureau, J.F., Bronfman, E., & Lyons-Ruth, K. (2007). Infant genotype may moderate sensitivity to maternal affective communications: Attachment disorganization, quality of care, and the DRD4 polymorphism. Social Neuroscience, 2, 307319.
  • McNeil, D.E., & Binder, R.L. (2007). Effectiveness of a mental health court in reducing criminal recidivism and violence. American Journal of Psychiatry, 164, 13951403.
  • Tough, S.C., Siever, J.E., Leew, S., Johnston, D.W., Benzies, K., & Clark, D. (2008). Maternal mental health predicts risk of developmental problems at 3 years of age: Follow up of a community based trial. BMC Pregnancy and Childbirth, 8, 16.