The results of this study replicate and extend the current literature and suggest that the KSADS-PL may be used as a reliable tool for the evaluation of DSM-IV psychiatric disorders in preschoolers. As hypothesized, in general the KSADS-PL showed good psychometric properties for the screening and diagnosis of DSM-IV disorders in this population of children. Moreover, as reviewed by Egger and Angold (2006)
, the lifetime prevalence of any DSM psychiatric disorders and specific disorders (e.g., anxiety, mood, and behavioral disorders) found in our sample are comparable to the rates reported in non-psychiatric settings.
Before discussing the findings in more detail, the limitations of this study should be considered. First, since part of the sample consisted of children of parents selected for a high risk study and most of the sample was Caucasian, the generalizability of these findings is limited. However, as noted above, the rates of psychiatric disorders ascertained through the KSADS were comparable to those of other studies in the field (Egger & Angold, 2006
; Gadow et al., 2001
; Keenan et al., 1997
; Lavigne et al., 1996
; Roberts et al., 1998
). Second, the relatively small number of children with certain definite lifetime DSM diagnoses (e.g., mood disorders) limited the range of disorders that were analyzed. Thus, there is a need to evaluate the usefulness of the KSADS for preschoolers using larger community and clinically-referred samples. Third, no direct observational data was available. Also, since a small proportion of children attended kindergarten or other preschool services, we did not obtain information from teachers or other non-family sources. Fourth, since it is very difficult to administer the KSADS and the PAPA because both interviews are lengthy, the comparison between the KSADS and the PAPA was based on a small sample of cases. Furthermore, the KSADS and PAPA were administered on separate occasions that were an average of 8 months apart. The primary period of the PAPA is the last 3 months, whereas the KSADS assessed lifetime occurrence. However, despite these issues the rate of agreement between the two instruments was good. Finally, pervasive developmental disorders were not included because the KSADS-PL does not include these conditions.
For definite DSM-IV lifetime psychiatric disorders, the K-SADS showed good overall convergent and discriminant validity and good inter-rater reliability. Divergent validity was strong for anxiety disorders. Discriminant validity was excellent. In fact, preschool children who were diagnosed with ODD, ADHD, ANX, MOOD, or ELIM had significantly higher scores than preschoolers without these disorders on the ECI-4. Inter-rater kappa coefficients for all definite diagnoses were high. Also, though preliminary, a preschooler’s intake diagnosis appears to be predictive of a current diagnosis at two-year follow-up assessment.
In the same manner, results were very similar when looking at the positive KSADS screens. With the exception of MOOD, all convergent CBCL correlations were significant. As previously noted, 86% of those who screened positive for a mood disorder screened positive for a behavioral disorder—which explains the highest correlations found between any positive mood screen and the aggressive and externalizing subscales. As with the definite disorders, divergent validity results were strongest for anxiety disorder positive screens. Additionally, ODD and emotional positive screens showed good divergent validity and statistically separated from the attention domain. Discriminant validity was significant for all positive screens. Inter-rater reliability of the KSADS-PL was very good (Kappas for all individual positive screens and for positive screens overall were ≥.70).
Despite the fact that children were interviewed with the PAPA an average of 8 months after the KSADS administration and methodological differences between these two instruments, the results obtained with both instruments were comparable. These results are not surprising because, with some exception, both instruments share similar content. The few discrepancies suggest that the differences between the PAPA and the KSADS may reflect different degrees of severity rather than a global disconnect. Also, other factors may account by the differences between these two instruments such as time effects, inherent differences between semi-structured and structured instruments, and the fact that diagnoses of the PAPA are generated by a computer algorithm whereas the final diagnoses of the KSADS are generated by a clinical consensus with a child psychiatrist. However, given the very small numbers, the observed differences may be accounted for by error indicating the need for further studies with larger samples.
The detection of emerging psychiatric symptomatology in preschoolers is extremely important in that it helps to define the earliest manifestations and boundaries of phenotypes, provides an opportunity for early intervention, and may reduce the risk and severity of later disorders (DelCarmen-Wiggins & Carter, 2001
; Egger & Angold, 2004
; Luby et al., 2004b
). Though much work remains in examining the appropriateness of DSM-IV diagnostic criteria for this age group (Egger & Angold, 2006
; Luby et al., 2003
; Sprafkin et al., 2002
), including assessing psychopathology in the context of normal preschool developmental stages (Carter et al., 2004
; Gadow & Sprafkin, 2000
; Keenan & Wakschlag, 2002
), temperamental variation (Egger & Angold, 2006
), and the delineation between typical and atypical behaviors in an age group marked by rapid developmental changes, results of this pilot and other studies suggest that the KSADS may be useful to screen and diagnose preschool children through parent interviews. The KSADS-PL may be considered for studies that assess children from ages 2 to 18 years, and in particular those that study clinical samples. While the parent remains the sole reporter for children under age 6, once children reach 6 years old, they—along with their parents—can be directly interviewed about their psychiatric symptomatology, using the same instrument. However, it is important to emphasize that given the limitations noted above, the results of this study need to be considered preliminary and further evaluation of the usefulness of the KSADS-PL, the K-MRS and DEP section of the KSADS-P for preschoolers is warranted. Furthermore, some of the prompts of the KSADS may require further modification for easy administration for this age group. Finally, it is crucial to train others to make empirically determined “developmentally appropriate” decisions in coding the KSADS.
Future studies need to take into the account any changes in the validity of DSM-IV criteria for preschoolers (Carter et al., 2004
; Keenan and Wakschlag, 2004
; Luby et al., 2003
; Scheeringa et al, 2001
). Also, studies should compare whether diagnoses in preschoolers generated through computer algorithms are comparable to diagnoses generated though clinical consensus. In addition, multiple sources of information such as teachers and other family members, psychiatric family history and other familial factors, life events, parent-child interaction observation, are needed in order to provide a comprehensive assessment as well as to uncover possible predictors and clinical correlates (Carter et al., 2004